The majority of Medicaid beneficiaries receive most or all of their care through managed care models. [1] In 2016, the three Medicaid managed care models most commonly used by states were: (1) comprehensive managed care organizations (CMCs), where states pay comprehensive managed care organizations (MCOs) a risk-based capitation rate to cover a broad set of services that typically include acute, primary, and specialty medical services; (2) primary care case management (PCCM), where primary care practitioners provide a core set of case management services in exchange for an administrative fee, but nearly all services continue to be provided on a fee-for-service basis; and (3) limited-benefit plans, including behavioral health organizations (BHOs), where plans manage a subset of services such as treatment for substance use disorders and inpatient or institutional mental health services.
All states providing services through managed care plans must report enrollment for beneficiaries covered under such plans; however, the accuracy of T-MSIS enrollment reporting varies considerably across both states and plan types. This data quality assessment examines alignment between July data from the T-MSIS Analytic Files (TAF) and the benchmark, the annual Medicaid Managed Care Enrollment and Program Characteristics (MMCEPC) report. [2]
Centers for Medicare & Medicaid Services (CMS). “Medicaid Managed Care Enrollment and Program Characteristics, 2016.” Spring 2018. Available at https://www.medicaid.gov/Medicaid/downloads/2016-medicaid-managed-care-enrollment-report.pdf . Accessed April 10, 2018.
Centers for Medicare & Medicaid Services (CMS). “Medicaid Managed Care Enrollment and Program Characteristics, 2016.” Spring 2018. Available at https://www.medicaid.gov/Medicaid/downloads/2016-medicaid-managed-care-enrollment-report.pdf . Accessed April 10, 2018.
Centers for Medicare & Medicaid Services (CMS). \u201cMedicaid Managed Care Enrollment and Program Characteristics, 2016.\u201d Spring 2018. Available at https://www.medicaid.gov/Medicaid/downloads/2016-medicaid-managed-care-enrollment-report.pdf . Accessed April 10, 2018.
Centers for Medicare & Medicaid Services (CMS). \u201cMedicaid Managed Care Enrollment and Program Characteristics, 2016.\u201d Spring 2018. Available at https://www.medicaid.gov/Medicaid/downloads/2016-medicaid-managed-care-enrollment-report.pdf . Accessed April 10, 2018.
The analysis relies on two main data sources: (1) the TAF annual Demographic and Eligibility (DE) file [3] and (2) the annual MMCEPC report.
We compared how well the TAF aligned with the benchmark by measuring the percent difference between the data sources in each month, then averaged the percent difference over all twelve months of the calendar year. We calculated TAF-based enrollment counts for beneficiaries in managed care by using managed care plan type information found in the TAF DE file. To align TAF counts as closely as possible with the benchmark data, we used information about beneficiary enrollment for July of the given year. We first counted the number of beneficiaries enrolled in CMCs, PCCM entities, and BHOs in July . [4] We then limited the analysis to beneficiaries enrolled in Medicaid. [5] Because the benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, the percent difference is calculated as a percent error or change: 100*(TAF – MMCEPC)/MMCEPC. We categorized a state’s enrollment data into one of four categories according to the level of alignment with the benchmark and the corresponding level of data quality concern (Table 1).
Table 1. Criteria for DQ assessment of enrollment in managed care
| Average monthly percent difference between TAF and MMCEPC enrollment counts | Level of alignment between TAF and MMCEPC enrollment counts | DQ assessment |
|---|---|---|
| x ≤ 10 percent | High | Low concern |
| 10 percent < x ≤ 20 percent | Moderate | Medium concern |
| 20 percent < x ≤ 50 percent | Low | High concern |
| x > 50 percent | Very low | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
To identify beneficiaries enrolled in managed care, we relied on the managed care plan type information (MC_PLAN_TYPE_CD1_07 through MC_PLAN_TYPE_CD16_07) found in the TAF DE file. CMC programs included beneficiaries enrolled in a CMC plan or a health insuring organization (HIO) (managed care plan type 01 or 04); PCCMs included traditional PCCM provider arrangements and enhanced PCCM provider arrangements (managed care plan type 02 or 03); and BHOs included mental health prepaid inpatient health plans (PIHPs) and prepaid ambulatory health plans (PAHPs), substance use disorder (SUD) PIHPs and PAHPs, and mental health and SUD PIHPs and PAHPs (managed care plan types of 08 through 13).
To identify beneficiaries enrolled in Medicaid, we relied on the CHIP code variable for July (CHIP_CD_07) in the TAF DE file, or eligibility group variable for July (ELGBLTY_GRP_CD_07) if the CHIP code was missing. Medicaid beneficiaries are identified using CHIP code 1 and eligibility group code 1-60 or 69-75. For analyses of 2014 through 2017 TAF, we also use CHIP code 4 (Medicaid and S-CHIP) to identify Medicaid beneficiaries, because CHIP code 4 is a valid value for those TAF data years.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
To identify beneficiaries enrolled in managed care, we relied on the managed care plan type information (MC_PLAN_TYPE_CD1_07 through MC_PLAN_TYPE_CD16_07) found in the TAF DE file. CMC programs included beneficiaries enrolled in a CMC plan or a health insuring organization (HIO) (managed care plan type 01 or 04); PCCMs included traditional PCCM provider arrangements and enhanced PCCM provider arrangements (managed care plan type 02 or 03); and BHOs included mental health prepaid inpatient health plans (PIHPs) and prepaid ambulatory health plans (PAHPs), substance use disorder (SUD) PIHPs and PAHPs, and mental health and SUD PIHPs and PAHPs (managed care plan types of 08 through 13).
To identify beneficiaries enrolled in Medicaid, we relied on the CHIP code variable for July (CHIP_CD_07) in the TAF DE file, or eligibility group variable for July (ELGBLTY_GRP_CD_07) if the CHIP code was missing. Medicaid beneficiaries are identified using CHIP code 1 and eligibility group code 1-60 or 69-75. For analyses of 2014 through 2017 TAF, we also use CHIP code 4 (Medicaid and S-CHIP) to identify Medicaid beneficiaries, because CHIP code 4 is a valid value for those TAF data years.
The TAF eligibility files include information about the enrollment of Medicaid and CHIP beneficiaries into different types of managed care models. One of the most common is comprehensive managed care organizations (CMCs), which deliver a broad range of services, including primary care, specialty care, and most other acute services. This analysis examines how well the TAF data on CMC enrollment align with an external benchmark, the Medicaid Managed Care Enrollment and Program Characteristics report.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4021"", ""relatedTopics"": [{""measureId"": 2, ""measureName"": ""Enrollment in PCCM Programs"", ""groupId"": 1, ""groupName"": ""Enrollment Benchmarking"", ""order"": 1}, {""measureId"": 3, ""measureName"": ""Enrollment in BHO Plans"", ""groupId"": 1, ""groupName"": ""Enrollment Benchmarking"", ""order"": 2}]}" 2,"{""measureId"": 2, ""measureName"": ""Enrollment in PCCM Programs"", ""groupId"": 1, ""groupName"": ""Enrollment Benchmarking"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Enroll-PCCM-Programs.pdf"", ""background"": {""content"": ""The majority of Medicaid beneficiaries receive most or all of their care through managed care models. [1] In 2016, the three Medicaid managed care models most commonly used by states were: (1) comprehensive managed care organizations (CMCs), where states pay comprehensive managed care organizations (MCOs) a risk-based capitation rate to cover a broad set of services that typically include acute, primary, and specialty medical services; (2) primary care case management (PCCM), where primary care practitioners provide a core set of case management services in exchange for an administrative fee, but nearly all services continue to be provided on a fee-for-service basis; and (3) limited-benefit plans, including behavioral health organizations (BHOs), where plans manage a subset of services such as treatment for substance use disorders and inpatient or institutional mental health services.
All states providing services through managed care plans must report enrollment for beneficiaries covered under such plans; however, the accuracy of T-MSIS enrollment reporting varies considerably across both states and plan types. This data quality assessment examines alignment between July data from the T-MSIS Analytic Files (TAF) and the benchmark, the annual Medicaid Managed Care Enrollment and Program Characteristics (MMCEPC) report. [2]
Centers for Medicare & Medicaid Services (CMS). “Medicaid Managed Care Enrollment and Program Characteristics, 2016.” Spring 2018. Available at https://www.medicaid.gov/Medicaid/downloads/2016-medicaid-managed-care-enrollment-report.pdf . Accessed April 10, 2018.
Centers for Medicare & Medicaid Services (CMS). “Medicaid Managed Care Enrollment and Program Characteristics, 2016.” Spring 2018. Available at https://www.medicaid.gov/Medicaid/downloads/2016-medicaid-managed-care-enrollment-report.pdf . Accessed April 10, 2018.
Centers for Medicare & Medicaid Services (CMS). \u201cMedicaid Managed Care Enrollment and Program Characteristics, 2016.\u201d Spring 2018. Available at https://www.medicaid.gov/Medicaid/downloads/2016-medicaid-managed-care-enrollment-report.pdf . Accessed April 10, 2018.
Centers for Medicare & Medicaid Services (CMS). \u201cMedicaid Managed Care Enrollment and Program Characteristics, 2016.\u201d Spring 2018. Available at https://www.medicaid.gov/Medicaid/downloads/2016-medicaid-managed-care-enrollment-report.pdf . Accessed April 10, 2018.
The analysis relies on two main data sources: (1) the TAF annual Demographic and Eligibility (DE) file [3] and (2) the annual MMCEPC report.
We compared how well the TAF aligned with the benchmark by measuring the percent difference between the data sources in each month, then averaged the percent difference over all twelve months of the calendar year. We calculated TAF-based enrollment counts for beneficiaries in managed care by using managed care plan type information found in the TAF DE file. To align TAF counts as closely as possible with the benchmark data, we used information about beneficiary enrollment for July of the given year. We first counted the number of beneficiaries enrolled in CMCs, PCCM entities, and BHOs in July . [4] We then limited the analysis to beneficiaries enrolled in Medicaid. [5] Because the benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, the percent difference is calculated as a percent error or change: 100*(TAF – MMCEPC)/MMCEPC. We categorized a state’s enrollment data into one of four categories according to the level of alignment with the benchmark and the corresponding level of data quality concern (Table 1).
Table 1. Criteria for DQ assessment of enrollment in managed care
| Average monthly percent difference between TAF and MMCEPC enrollment counts | Level of alignment between TAF and MMCEPC enrollment counts | DQ assessment |
|---|---|---|
| x ≤ 10 percent | High | Low concern |
| 10 percent < x ≤ 20 percent | Moderate | Medium concern |
| 20 percent < x ≤ 50 percent | Low | High concern |
| x > 50 percent | Very low | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
To identify beneficiaries enrolled in managed care, we relied on the managed care plan type information (MC_PLAN_TYPE_CD1_07 through MC_PLAN_TYPE_CD16_07) found in the TAF DE file. CMC programs included beneficiaries enrolled in a CMC plan or a health insuring organization (HIO) (managed care plan type 01 or 04); PCCMs included traditional PCCM provider arrangements and enhanced PCCM provider arrangements (managed care plan type 02 or 03); and BHOs included mental health prepaid inpatient health plans (PIHPs) and prepaid ambulatory health plans (PAHPs), substance use disorder (SUD) PIHPs and PAHPs, and mental health and SUD PIHPs and PAHPs (managed care plan types of 08 through 13).
To identify beneficiaries enrolled in Medicaid, we relied on the CHIP code variable for July (CHIP_CD_07) in the TAF DE file, or eligibility group variable for July (ELGBLTY_GRP_CD_07) if the CHIP code was missing. Medicaid beneficiaries are identified using CHIP code 1 and eligibility group code 1-60 or 69-75. For analyses of 2014 through 2017 TAF, we also use CHIP code 4 (Medicaid and S-CHIP) to identify Medicaid beneficiaries, because CHIP code 4 is a valid value for those TAF data years.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
To identify beneficiaries enrolled in managed care, we relied on the managed care plan type information (MC_PLAN_TYPE_CD1_07 through MC_PLAN_TYPE_CD16_07) found in the TAF DE file. CMC programs included beneficiaries enrolled in a CMC plan or a health insuring organization (HIO) (managed care plan type 01 or 04); PCCMs included traditional PCCM provider arrangements and enhanced PCCM provider arrangements (managed care plan type 02 or 03); and BHOs included mental health prepaid inpatient health plans (PIHPs) and prepaid ambulatory health plans (PAHPs), substance use disorder (SUD) PIHPs and PAHPs, and mental health and SUD PIHPs and PAHPs (managed care plan types of 08 through 13).
To identify beneficiaries enrolled in Medicaid, we relied on the CHIP code variable for July (CHIP_CD_07) in the TAF DE file, or eligibility group variable for July (ELGBLTY_GRP_CD_07) if the CHIP code was missing. Medicaid beneficiaries are identified using CHIP code 1 and eligibility group code 1-60 or 69-75. For analyses of 2014 through 2017 TAF, we also use CHIP code 4 (Medicaid and S-CHIP) to identify Medicaid beneficiaries, because CHIP code 4 is a valid value for those TAF data years.
The TAF eligibility files include information about the enrollment of Medicaid and CHIP beneficiaries into different types of managed care models. Some states enroll beneficiaries in primary care case management (PCCM) programs, in which a primary care practitioner provides a core set of case management services in exchange for an administrative fee, but all other services continue to be delivered on a fee-for-service basis. This analysis examines how well the TAF data on PCCM enrollment align with an external benchmark, the Medicaid Managed Care Enrollment and Program Characteristics report.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4021"", ""relatedTopics"": [{""measureId"": 1, ""measureName"": ""Enrollment in CMC Plans"", ""groupId"": 1, ""groupName"": ""Enrollment Benchmarking"", ""order"": 0}, {""measureId"": 3, ""measureName"": ""Enrollment in BHO Plans"", ""groupId"": 1, ""groupName"": ""Enrollment Benchmarking"", ""order"": 2}]}" 3,"{""measureId"": 3, ""measureName"": ""Enrollment in BHO Plans"", ""groupId"": 1, ""groupName"": ""Enrollment Benchmarking"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Enroll-BHO-Plans.pdf"", ""background"": {""content"": ""The majority of Medicaid beneficiaries receive most or all of their care through managed care models. [1] In 2016, the three Medicaid managed care models most commonly used by states were: (1) comprehensive managed care organizations (CMCs), where states pay comprehensive managed care organizations (MCOs) a risk-based capitation rate to cover a broad set of services that typically include acute, primary, and specialty medical services; (2) primary care case management (PCCM), where primary care practitioners provide a core set of case management services in exchange for an administrative fee, but nearly all services continue to be provided on a fee-for-service basis; and (3) limited-benefit plans, including behavioral health organizations (BHOs), where plans manage a subset of services such as treatment for substance use disorders and inpatient or institutional mental health services.
All states providing services through managed care plans must report enrollment for beneficiaries covered under such plans; however, the accuracy of T-MSIS enrollment reporting varies considerably across both states and plan types. This data quality assessment examines alignment between July data from the T-MSIS Analytic Files (TAF) and the benchmark, the annual Medicaid Managed Care Enrollment and Program Characteristics (MMCEPC) report. [2]
Centers for Medicare & Medicaid Services (CMS). “Medicaid Managed Care Enrollment and Program Characteristics, 2016.” Spring 2018. Available at https://www.medicaid.gov/Medicaid/downloads/2016-medicaid-managed-care-enrollment-report.pdf . Accessed April 10, 2018.
Centers for Medicare & Medicaid Services (CMS). “Medicaid Managed Care Enrollment and Program Characteristics, 2016.” Spring 2018. Available at https://www.medicaid.gov/Medicaid/downloads/2016-medicaid-managed-care-enrollment-report.pdf . Accessed April 10, 2018.
Centers for Medicare & Medicaid Services (CMS). \u201cMedicaid Managed Care Enrollment and Program Characteristics, 2016.\u201d Spring 2018. Available at https://www.medicaid.gov/Medicaid/downloads/2016-medicaid-managed-care-enrollment-report.pdf . Accessed April 10, 2018.
Centers for Medicare & Medicaid Services (CMS). \u201cMedicaid Managed Care Enrollment and Program Characteristics, 2016.\u201d Spring 2018. Available at https://www.medicaid.gov/Medicaid/downloads/2016-medicaid-managed-care-enrollment-report.pdf . Accessed April 10, 2018.
The analysis relies on two main data sources: (1) the TAF annual Demographic and Eligibility (DE) file [3] and (2) the annual MMCEPC report.
We compared how well the TAF aligned with the benchmark by measuring the percent difference between the data sources in each month, then averaged the percent difference over all twelve months of the calendar year. We calculated TAF-based enrollment counts for beneficiaries in managed care by using managed care plan type information found in the TAF DE file. To align TAF counts as closely as possible with the benchmark data, we used information about beneficiary enrollment for July of the given year. We first counted the number of beneficiaries enrolled in CMCs, PCCM entities, and BHOs in July . [4] We then limited the analysis to beneficiaries enrolled in Medicaid. [5] Because the benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, the percent difference is calculated as a percent error or change: 100*(TAF – MMCEPC)/MMCEPC. We categorized a state’s enrollment data into one of four categories according to the level of alignment with the benchmark and the corresponding level of data quality concern (Table 1).
Table 1. Criteria for DQ assessment of enrollment in managed care
| Average monthly percent difference between TAF and MMCEPC enrollment counts | Level of alignment between TAF and MMCEPC enrollment counts | DQ assessment |
|---|---|---|
| x ≤ 10 percent | High | Low concern |
| 10 percent < x ≤ 20 percent | Moderate | Medium concern |
| 20 percent < x ≤ 50 percent | Low | High concern |
| x > 50 percent | Very low | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
To identify beneficiaries enrolled in managed care, we relied on the managed care plan type information (MC_PLAN_TYPE_CD1_07 through MC_PLAN_TYPE_CD16_07) found in the TAF DE file. CMC programs included beneficiaries enrolled in a CMC plan or a health insuring organization (HIO) (managed care plan type 01 or 04); PCCMs included traditional PCCM provider arrangements and enhanced PCCM provider arrangements (managed care plan type 02 or 03); and BHOs included mental health prepaid inpatient health plans (PIHPs) and prepaid ambulatory health plans (PAHPs), substance use disorder (SUD) PIHPs and PAHPs, and mental health and SUD PIHPs and PAHPs (managed care plan types of 08 through 13).
To identify beneficiaries enrolled in Medicaid, we relied on the CHIP code variable for July (CHIP_CD_07) in the TAF DE file, or eligibility group variable for July (ELGBLTY_GRP_CD_07) if the CHIP code was missing. Medicaid beneficiaries are identified using CHIP code 1 and eligibility group code 1-60 or 69-75. For analyses of 2014 through 2017 TAF, we also use CHIP code 4 (Medicaid and S-CHIP) to identify Medicaid beneficiaries, because CHIP code 4 is a valid value for those TAF data years.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
To identify beneficiaries enrolled in managed care, we relied on the managed care plan type information (MC_PLAN_TYPE_CD1_07 through MC_PLAN_TYPE_CD16_07) found in the TAF DE file. CMC programs included beneficiaries enrolled in a CMC plan or a health insuring organization (HIO) (managed care plan type 01 or 04); PCCMs included traditional PCCM provider arrangements and enhanced PCCM provider arrangements (managed care plan type 02 or 03); and BHOs included mental health prepaid inpatient health plans (PIHPs) and prepaid ambulatory health plans (PAHPs), substance use disorder (SUD) PIHPs and PAHPs, and mental health and SUD PIHPs and PAHPs (managed care plan types of 08 through 13).
To identify beneficiaries enrolled in Medicaid, we relied on the CHIP code variable for July (CHIP_CD_07) in the TAF DE file, or eligibility group variable for July (ELGBLTY_GRP_CD_07) if the CHIP code was missing. Medicaid beneficiaries are identified using CHIP code 1 and eligibility group code 1-60 or 69-75. For analyses of 2014 through 2017 TAF, we also use CHIP code 4 (Medicaid and S-CHIP) to identify Medicaid beneficiaries, because CHIP code 4 is a valid value for those TAF data years.
The TAF eligibility files include information about the enrollment of Medicaid and CHIP beneficiaries into different types of managed care models. Many states enroll beneficiaries in behavioral health organizations (BHOs), which provide mental health and substance use disorder services. This analysis examines how well the TAF data on BHO enrollment align with an external benchmark, the Medicaid Managed Care Enrollment and Program Characteristics report.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4021"", ""relatedTopics"": [{""measureId"": 1, ""measureName"": ""Enrollment in CMC Plans"", ""groupId"": 1, ""groupName"": ""Enrollment Benchmarking"", ""order"": 0}, {""measureId"": 2, ""measureName"": ""Enrollment in PCCM Programs"", ""groupId"": 1, ""groupName"": ""Enrollment Benchmarking"", ""order"": 1}]}" 4,"{""measureId"": 4, ""measureName"": ""M-CHIP and S-CHIP Enrollment"", ""groupId"": 1, ""groupName"": ""Enrollment Benchmarking"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-MCHIP-SCHIP-Enroll.pdf"", ""background"": {""content"": ""Created as part of the Balanced Budget Act of 1997, the Children’s Health Insurance Program (CHIP) provides health care coverage to otherwise uninsured children in low-income families whose income exceeds Medicaid income-eligibility thresholds. States may use CHIP funds to expand their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); create a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopt a combination of both approaches. Because the CHIP population is important for both program administration and policy, many users of the T\u2011MSIS Analytic Files (TAF) data will want to identify CHIP beneficiaries either to study them explicitly or to exclude them from analyses.
There are three relevant data elements in the Annual Demographic and Eligibility (DE) file that can be used to identify different groups of CHIP beneficiaries (Table 1): [1]
Table 1. Potential DE TAF variables for identifying CHIP beneficiaries
| Data element | Use for identifying CHIP beneficiaries |
|---|---|
| CHIP_CD | Identifies beneficiaries in Medicaid (CHIP_CD = 1), M-CHIP (CHIP_CD = 2), and S-CHIP (CHIP_CD = 3). |
| CHIP_ENRLMT_DAYS | Constructed from T-MSIS enrollment date and enrollment type variables (ENROLLMENT-EFF-DATE, ENROLLMENT-END-DATE, and ENROLLMENT-TYPE). |
| ELGBLTY_GRP_CD | Contains the eligibility group applicable to the individual based on the state’s eligibility determination process. When CHIP_CD is missing, ELGBLTY_GRP_CD can be used to indicate CHIP enrollment (ELGBLTY_GRP_CD = 61–68, with 61 used for both M-CHIP and S-CHIP and 62–68 exclusive to S-CHIP) and Medicaid enrollment (ELGBLTY_GRP_CD = 1–60 or 69–76). |
Note: \tThese three data elements are available monthly in the annual DE TAF, with the number of each month appended to the end of the data element name (for instance, CHIP_CD_01 for January, CHIP_CD_02 for February, and so on). For simplicity, we did not list the monthly indicators in this table because we used all months of data. A list of valid values and descriptions of these data elements can be found in the TAF Demographic and Eligibility Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries .
Maintenance assistance status and basis of eligibility, which was constructed from T-MSIS data elements maintenance assistance status and Medicaid basis of eligibility, has been phased out in favor of the new, more detailed eligibility group code; the maintenance assistance status and basis of eligibility code is not recommended for use.
Enrollment type flag groups M-CHIP with Medicaid beneficiaries and therefore cannot be used to identify all CHIP beneficiaries.
Analyses conducted on the 2016 TAF data found that CHIP code is the most reliable data element for counting total enrollment in M-CHIP and S-CHIP across the largest number of states (results not shown). However, in states with high rates of missing data in CHIP code, using eligibility group code may result in more accurate counts of CHIP enrollment. [3] This analysis evaluates whether CHIP code and eligibility group code (when CHIP code is missing) can be used to accurately count total CHIP enrollment in each state. [4]
Two other data elements—Medicaid enrollment days and enrollment type flag—combine Medicaid and M-CHIP beneficiaries. They can be used only in combination with other data elements such as CHIP code and eligibility group code to identify CHIP beneficiaries. Therefore, they are not listed here as potential TAF variables for identifying CHIP beneficiaries.
More information on state CHIP programs is available at https://www.medicaid.gov/chip/state-program-information/index.html and https://www.macpac.gov/subtopic/key-design-features/ .
TAF users should first verify that their state(s) of interest are reporting data for CHIP beneficiaries.
In some states, CHIP code may be used to accurately count total enrollment but does not differentiate well between M-CHIP and S-CHIP enrollees.
Two other data elements\u2014Medicaid enrollment days and enrollment type flag\u2014combine Medicaid and M-CHIP beneficiaries. They can be used only in combination with other data elements such as CHIP code and eligibility group code to identify CHIP beneficiaries. Therefore, they are not listed here as potential TAF variables for identifying CHIP beneficiaries.
More information on state CHIP programs is available at https://www.medicaid.gov/chip/state-program-information/index.html and https://www.macpac.gov/subtopic/key-design-features/ .
TAF users should first verify that their state(s) of interest are reporting data for CHIP beneficiaries.
In some states, CHIP code may be used to accurately count total enrollment but does not differentiate well between M-CHIP and S-CHIP enrollees.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We used the Eligibility and Enrollment Performance Indicator (PI) data as the external benchmark to examine the accuracy of TAF-based enrollment counts for the CHIP population. [5] The PI data include both M-CHIP and S-CHIP beneficiaries as of the last day of each month. Although some states’ PI data contain quality issues prior to 2017 that may affect their accuracy, these data are the best source available for benchmarking the CHIP population because many of the data quality issues are known, and the data provide a consistent benchmark across multiple data quality assessments. [6]
In the TAF, we identified CHIP beneficiaries by selecting DE records where CHIP code [7] was equal to 2 (M-CHIP), 3 (S-CHIP) or 4 (Medicaid and S-CHIP). [8] If the CHIP code was missing, we counted beneficiaries with an eligibility group code that indicated they were eligible for CHIP benefits (eligibility group code of 61–68). In creating the TAF-based counts, we included individuals enrolled at any point (“ever enrolled”) in the month. We compared the PI and TAF-based counts by (1) evaluating the percent difference between TAF-based enrollment counts and the benchmark, averaged across all 12 months, and (2) examining the standard deviation of this measure to assess variation in the difference across months. The average monthly TAF enrollment count is calculated as the sum of the monthly TAF counts divided by 12, and the average monthly PI enrollment count is calculated as the sum of the monthly PI counts divided by 12. [9] Because the benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, the percent difference is calculated as a percent error or change: the difference between the TAF and PI counts divided by the PI count and multiplied by 100. The average monthly percent difference is calculated as the sum of the monthly percent differences, divided by 12. [10] The standard deviation of the monthly percent differences between the TAF-based count and the PI count is calculated as the square root of the sum of the squared differences between the monthly percent differences and the average percent difference, divided by 12.
Table 2 shows the level of concern for the TAF Medicaid enrollment counts based on both the percent difference and the level of alignment between the TAF and the PI enrollment counts. Although we did not assign the level of concern based on the standard deviation, we provide this information in Table 3, and TAF users may want to consider the monthly variability between TAF and the benchmark when determining whether the data are usable for their analysis and whether all months are of similar quality.
Table 2. Criteria for DQ assessment of CHIP enrollment
| Average monthly percent difference between TAF and PI enrollment counts | Level of alignment between TAF and PI enrollment counts | DQ assessment |
|---|---|---|
| x ≤ 5 percent | High | Low concern |
| 5 percent < x ≤ 10 percent | Moderate | Low concern |
| 10 percent < x ≤ 20 percent | Low | Medium concern |
| 20 percent < x ≤ 50 percent | Very low | High concern |
| x > 50 percent | Very low | Unusable |
Methods previously used to assess data quality
Table 3 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 3. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
More information about the PI data set can be found at https://www.medicaid.gov/medicaid/program-information/medicaid-and-chip-enrollment-data/report-highlights/index.html . In some cases, the PI data in the Atlas may not match exactly the PI data publicly available on Medicaid.gov, because our analysis uses a version of the data set that may have been updated more recently.
TAF users should check that CHIP code differentiates well between M-CHIP and S-CHIP beneficiaries in their state(s) of interest. This can be done by cross-checking CHIP code with eligibility group code and CHIP enrollment days.
CHIP code of 4 (individual was both Medicaid eligible and S-CHIP eligible during the same month) is not a valid value in later versions of the T-MSIS data dictionary. However, because some states use the code on a small number of records, we included it in tabulations presented in this analysis.
If a state did not report CHIP PI enrollment counts for all months, the average percent difference is calculated based on the months with data.
The difference between TAF and PI enrollment is based on an average of the monthly differences between these two data sources. As a result, it may not equal the difference between the average annual TAF enrollment and average annual PI enrollment.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
More information about the PI data set can be found at https://www.medicaid.gov/medicaid/program-information/medicaid-and-chip-enrollment-data/report-highlights/index.html . In some cases, the PI data in the Atlas may not match exactly the PI data publicly available on Medicaid.gov, because our analysis uses a version of the data set that may have been updated more recently.
TAF users should check that CHIP code differentiates well between M-CHIP and S-CHIP beneficiaries in their state(s) of interest. This can be done by cross-checking CHIP code with eligibility group code and CHIP enrollment days.
CHIP code of 4 (individual was both Medicaid eligible and S-CHIP eligible during the same month) is not a valid value in later versions of the T-MSIS data dictionary. However, because some states use the code on a small number of records, we included it in tabulations presented in this analysis.
If a state did not report CHIP PI enrollment counts for all months, the average percent difference is calculated based on the months with data.
The difference between TAF and PI enrollment is based on an average of the monthly differences between these two data sources. As a result, it may not equal the difference between the average annual TAF enrollment and average annual PI enrollment.
The TAF eligibility files include information on beneficiaries in both Medicaid and CHIP. CHIP provides coverage to otherwise uninsured children in low-income families whose income exceeds Medicaid eligibility thresholds. States may use CHIP funds to expand their Medicaid program (referred to as M-CHIP), create a separate program (referred to as S-CHIP), or adopt a combination of both approaches. This analysis examines how well the TAF data on total CHIP enrollment align with an external benchmark, the Performance Indicators data set.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4031"", ""relatedTopics"": []}" 5,"{""measureId"": 5, ""measureName"": ""Adult Expansion Enrollment"", ""groupId"": 1, ""groupName"": ""Enrollment Benchmarking"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Adult-Expansion-Enroll.pdf"", ""background"": {""content"": ""Under the Patient Protection and Affordable Care Act of 2010 (the Affordable Care Act), Medicaid eligibility was extended to nearly all adults younger than 65 with an income below 138 percent of the federal poverty level (FPL). [1] , [2] , [3] In June 2012, the Supreme Court ruled in National Federation of Independent Business v. Sebelius that this Medicaid expansion was not mandatory for states. [4] However, by the end of 2016, 31 states plus the District of Columbia and Puerto Rico (hereafter “states”) had used the authority of the Affordable Care Act to provide Medicaid coverage to low-income adults. [5] , [6] Although the majority of these 33 states expanded Medicaid through an amendment to their Medicaid state plan, about one-quarter of them expanded eligibility through an 1115 demonstration waiver. [7]
The Centers for Medicare & Medicaid Services (CMS) requires states to report all Medicaid beneficiaries to a valid eligibility group code in T-MSIS. States that have expanded Medicaid are required to report adult expansion beneficiaries to specific eligibility group codes (codes 72-75). Eligibility group code 72 is designated for newly eligible adult expansion beneficiaries, i.e., those who would not qualify for full Medicaid benefits, benchmark coverage, or benchmark-equivalent coverage under the state’s program rules in place as of December 1, 2009. [8] Eligibility group codes 73, 74, and 75 are designated for beneficiaries not newly eligible, i.e., those who would qualify under such rules. [9] , [10] However, not all Medicaid expansion states are reporting beneficiaries to these codes. In some cases, this may be a symptom of a broader problem in reporting the new eligibility group code in T-MSIS. This data quality assessment examines the accuracy of the adult expansion enrollment counts derived from the T-MSIS Analytic Files (TAF) by comparing these counts to an external benchmark.
Adults entitled to Medicare and those pregnant or eligible for Medicaid under another mandatory eligibility group are generally not eligible for the adult expansion group, which includes individuals below 133 percent of the FPL after applying a 5 percent income disregard, equating to an effective coverage limit of 138 percent of the FPL.
This expansion population is also known as the VIII Group because the eligibility criteria for this group are defined in Section 1902(10)(VIII) of the Social Security Act.
Paradise, Julia. “Moving Medicaid Forward.” San Francisco, CA: The Kaiser Commission on Medicaid and the Uninsured, March 9, 2015. Available at: https://www.kff.org/health-reform/issue-brief/medicaid-moving-forward/ . Accessed December 14, 2018.
Rosenbaum, Sara, and Timothy M. Westmoreland. “The Supreme Court’s Surprising Decision on the Medicaid Expansion; How Will the Federal Government and States Proceed?” Health Affairs, vol. 31, no. 8, August 2012. Available at: https://www.healthaffairs.org/doi/abs/10.1377/hlthaff.2012.0766 . Accessed December 14, 2018.
Kaiser Family Foundation. “Status of State Action on the Medicaid Expansion Decision.” April 9, 2019. Available at: https://www.kff.org/health-reform/state-indicator/state-activity-around-expanding-medicaid-under-the-affordable-care-act/?currentTimeframe=0&sortModel=%7B%22colId%22:%22Location%22,%22sort%22: %22asc%22%7D . Accessed April 15, 2019.
In addition, five states—Idaho, Maine, Nebraska, Utah, and Virginia—adopted the Medicaid expansion after 2016. Maine did so in 2017 via a ballot initiative and implemented the program in 2019. Idaho, Nebraska, and Utah also adopted the adult expansion via ballot initiative in 2018. Idaho is scheduled to implement the program on January 1, 2020, Nebraska is planning to do so on October 1, 2020, and Utah implemented a limited expansion in April 2019. Virginia’s legislature voted to adopt the adult expansion in 2018 and began enrollment on January 1, 2019.
Arizona, Arkansas, Iowa, Indiana, Kentucky, Michigan, Montana, and New Hampshire expanded Medicaid through an 1115 demonstration waiver. Pennsylvania initially implemented its adult expansion as an 1115 demonstration in 2015 but shifted to a state plan amendment a few months later.
A beneficiary is also considered “newly eligible” if he or she would have been eligible but could not have been enrolled for these benefits or this coverage because the applicable Medicaid waiver or demonstration had limited or capped enrollment as of December 1, 2009.
Centers for Medicare & Medicaid Services. “Medicaid and CHIP FAQs: Newly Eligible FMAP and Expansion State FMAP.” February 2013. Available at: http://www.statecoverage.org/files/ACA-FAQ-BHP.pdf . Accessed April 15, 2019.
States that met certain statutory criteria for health benefits coverage, as described in section 1905(z)(3) of the Affordable Care Act, qualified for increased Federal Medical Assistance Percentage (FMAP) rates for a subset of the \""not newly eligible\"" population. These states were designated as 1905(z)(3) expansion states. Eligibility group 73 identifies not newly eligible beneficiaries for non-1905(z)(3) states. Eligibility groups 74 and 75 identify not newly eligible beneficiaries for 1905(z)(3) states, with eligibility group 74 specifying parents, caretakers, or relatives of expansion beneficiaries.
Adults entitled to Medicare and those pregnant or eligible for Medicaid under another mandatory eligibility group are generally not eligible for the adult expansion group, which includes individuals below 133 percent of the FPL after applying a 5 percent income disregard, equating to an effective coverage limit of 138 percent of the FPL.
This expansion population is also known as the VIII Group because the eligibility criteria for this group are defined in Section 1902(10)(VIII) of the Social Security Act.
Paradise, Julia. \u201cMoving Medicaid Forward.\u201d San Francisco, CA: The Kaiser Commission on Medicaid and the Uninsured, March 9, 2015. Available at: https://www.kff.org/health-reform/issue-brief/medicaid-moving-forward/ . Accessed December 14, 2018.
Rosenbaum, Sara, and Timothy M. Westmoreland. \u201cThe Supreme Court\u2019s Surprising Decision on the Medicaid Expansion; How Will the Federal Government and States Proceed?\u201d Health Affairs, vol. 31, no. 8, August 2012. Available at: https://www.healthaffairs.org/doi/abs/10.1377/hlthaff.2012.0766 . Accessed December 14, 2018.
Kaiser Family Foundation. \u201cStatus of State Action on the Medicaid Expansion Decision.\u201d April 9, 2019. Available at: https://www.kff.org/health-reform/state-indicator/state-activity-around-expanding-medicaid-under-the-affordable-care-act/?currentTimeframe=0&sortModel=%7B%22colId%22:%22Location%22,%22sort%22: %22asc%22%7D . Accessed April 15, 2019.
In addition, five states\u2014Idaho, Maine, Nebraska, Utah, and Virginia\u2014adopted the Medicaid expansion after 2016. Maine did so in 2017 via a ballot initiative and implemented the program in 2019. Idaho, Nebraska, and Utah also adopted the adult expansion via ballot initiative in 2018. Idaho is scheduled to implement the program on January 1, 2020, Nebraska is planning to do so on October 1, 2020, and Utah implemented a limited expansion in April 2019. Virginia\u2019s legislature voted to adopt the adult expansion in 2018 and began enrollment on January 1, 2019.
Arizona, Arkansas, Iowa, Indiana, Kentucky, Michigan, Montana, and New Hampshire expanded Medicaid through an 1115 demonstration waiver. Pennsylvania initially implemented its adult expansion as an 1115 demonstration in 2015 but shifted to a state plan amendment a few months later.
A beneficiary is also considered \u201cnewly eligible\u201d if he or she would have been eligible but could not have been enrolled for these benefits or this coverage because the applicable Medicaid waiver or demonstration had limited or capped enrollment as of December 1, 2009.
Centers for Medicare & Medicaid Services. \u201cMedicaid and CHIP FAQs: Newly Eligible FMAP and Expansion State FMAP.\u201d February 2013. Available at: http://www.statecoverage.org/files/ACA-FAQ-BHP.pdf . Accessed April 15, 2019.
States that met certain statutory criteria for health benefits coverage, as described in section 1905(z)(3) of the Affordable Care Act, qualified for increased Federal Medical Assistance Percentage (FMAP) rates for a subset of the \""not newly eligible\"" population. These states were designated as 1905(z)(3) expansion states. Eligibility group 73 identifies not newly eligible beneficiaries for non-1905(z)(3) states. Eligibility groups 74 and 75 identify not newly eligible beneficiaries for 1905(z)(3) states, with eligibility group 74 specifying parents, caretakers, or relatives of expansion beneficiaries.
We used Medicaid Budget and Expenditure System (MBES) enrollment data as the external benchmark to examine the accuracy of the TAF-based enrollment counts for the adult expansion group. [11] We examined adult expansion population enrollment as a whole and for the subset of newly eligible beneficiaries using the TAF annual Demographic & Eligibility (DE) file. The MBES data are submitted by states to CMS on Form CMS-64 and include aggregated monthly counts of total Medicaid enrollment, enrollment for Medicaid beneficiaries in the adult expansion group, and enrollment for the subsets of newly eligible and not newly eligible adult expansion beneficiaries. [12] Substantial differences between the TAF-based counts and MBES counts raise concerns about the validity of a state’s T-MSIS reporting.
To investigate this issue, we obtained a monthly count of beneficiaries with eligibility group codes in the TAF DE file that indicated they were in the adult expansion group (ELGBLTY_GRP_CD_mm = 72, 73, 74, or 75). These codes indicate that individuals were eligible for and enrolled in Medicaid in a given month under one of the adult expansion group categories, in which a value of 72 designates newly eligible beneficiaries, and a value of 73, 74, or 75 designates beneficiaries who are not newly eligible. [13]
We also obtained a monthly count of the VIII Group enrollees and VIII Group newly eligible enrollees reported by states in the MBES data. For the adult expansion population and for the newly eligible subset in each state and each month, we calculated a percent difference between the TAF-based enrollment counts and the benchmark, and then averaged the monthly percent difference across all 12 months of the calendar year. We also examined the standard deviation of this measure to assess the variation in the difference across months. Because the benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: the difference between the TAF and MBES counts divided by the MBES count.
We evaluated the average percent difference against the thresholds listed in Table 1 to categorize the states into levels of concern about data quality according to the alignment between the counts of beneficiaries reported in the TAF DE and the counts in the benchmark. Although we did not assign the level of concern based on the standard deviation, we provide this information in the table, and TAF users may want to consider the monthly variability between TAF and the benchmark when determining whether the data are usable for their analysis.
Table 1. Criteria for DQ assessment of enrollment in adult expansion group
| Average monthly percent difference between TAF and MBES enrollment counts | Level of alignment with between TAF and MBES enrollment counts | DQ assessment |
|---|---|---|
| x ≤ 5 percent | High | Low concern |
| 5 percent < x ≤ 10 percent | Moderate | Low concern |
| 10 percent < x ≤ 20 percent | Low | Medium concern |
| 20 percent < x ≤ 50 percent | Very low | High concern |
| x > 50 percent | Very low | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Centers for Medicare & Medicaid Services. “Medicaid Enrollment Data Collected Through MBES.” November 2018. Available at https://www.medicaid.gov/medicaid/program-information/medicaid-and-chip-enrollment-data/enrollment-mbes/index.html . Accessed December 24, 2018.
In early years of T-MSIS reporting, at least one state (Pennsylvania) reported its adult expansion beneficiaries using eligibility group code 71 (other expansions under 1115 authority). Because other states use this eligibility group code to identify non-expansion populations covered under the 1115 authority, it was not used in this analysis to calculate TAF-based counts of the adult expansion population.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Centers for Medicare & Medicaid Services. \u201cMedicaid Enrollment Data Collected Through MBES.\u201d November 2018. Available at https://www.medicaid.gov/medicaid/program-information/medicaid-and-chip-enrollment-data/enrollment-mbes/index.html . Accessed December 24, 2018.
In early years of T-MSIS reporting, at least one state (Pennsylvania) reported its adult expansion beneficiaries using eligibility group code 71 (other expansions under 1115 authority). Because other states use this eligibility group code to identify non-expansion populations covered under the 1115 authority, it was not used in this analysis to calculate TAF-based counts of the adult expansion population.
Under the Affordable Care Act, some states elected to expand coverage to all low-income adults with an income below 138 percent of the federal poverty level. This analysis examines how well the TAF data on enrollment in the adult expansion group align with an external benchmark, the Medicaid Budget and Expenditure System.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4041"", ""relatedTopics"": [{""measureId"": 6, ""measureName"": ""Newly Eligible Adult Enrollment"", ""groupId"": 1, ""groupName"": ""Enrollment Benchmarking"", ""order"": 1}]}" 6,"{""measureId"": 6, ""measureName"": ""Newly Eligible Adult Enrollment"", ""groupId"": 1, ""groupName"": ""Enrollment Benchmarking"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Newly-Elig-Adult-Enroll.pdf"", ""background"": {""content"": ""Under the Patient Protection and Affordable Care Act of 2010 (the Affordable Care Act), Medicaid eligibility was extended to nearly all adults younger than 65 with an income below 138 percent of the federal poverty level (FPL). [1] , [2] , [3] In June 2012, the Supreme Court ruled in National Federation of Independent Business v. Sebelius that this Medicaid expansion was not mandatory for states. [4] However, by the end of 2016, 31 states plus the District of Columbia and Puerto Rico (hereafter “states”) had used the authority of the Affordable Care Act to provide Medicaid coverage to low-income adults. [5] , [6] Although the majority of these 33 states expanded Medicaid through an amendment to their Medicaid state plan, about one-quarter of them expanded eligibility through an 1115 demonstration waiver. [7]
The Centers for Medicare & Medicaid Services (CMS) requires states to report all Medicaid beneficiaries to a valid eligibility group code in T-MSIS. States that have expanded Medicaid are required to report adult expansion beneficiaries to specific eligibility group codes (codes 72-75). Eligibility group code 72 is designated for newly eligible adult expansion beneficiaries, i.e., those who would not qualify for full Medicaid benefits, benchmark coverage, or benchmark-equivalent coverage under the state’s program rules in place as of December 1, 2009. [8] Eligibility group codes 73, 74, and 75 are designated for beneficiaries not newly eligible, i.e., those who would qualify under such rules. [9] , [10] However, not all Medicaid expansion states are reporting beneficiaries to these codes. In some cases, this may be a symptom of a broader problem in reporting the new eligibility group code in T-MSIS. This data quality assessment examines the accuracy of the adult expansion enrollment counts derived from the T-MSIS Analytic Files (TAF) by comparing these counts to an external benchmark.
Adults entitled to Medicare and those pregnant or eligible for Medicaid under another mandatory eligibility group are generally not eligible for the adult expansion group, which includes individuals below 133 percent of the FPL after applying a 5 percent income disregard, equating to an effective coverage limit of 138 percent of the FPL.
This expansion population is also known as the VIII Group because the eligibility criteria for this group are defined in Section 1902(10)(VIII) of the Social Security Act.
Paradise, Julia. “Moving Medicaid Forward.” San Francisco, CA: The Kaiser Commission on Medicaid and the Uninsured, March 9, 2015. Available at: https://www.kff.org/health-reform/issue-brief/medicaid-moving-forward/ . Accessed December 14, 2018.
Rosenbaum, Sara, and Timothy M. Westmoreland. “The Supreme Court’s Surprising Decision on the Medicaid Expansion; How Will the Federal Government and States Proceed?” Health Affairs, vol. 31, no. 8, August 2012. Available at: https://www.healthaffairs.org/doi/abs/10.1377/hlthaff.2012.0766 . Accessed December 14, 2018.
Kaiser Family Foundation. “Status of State Action on the Medicaid Expansion Decision.” April 9, 2019. Available at: https://www.kff.org/health-reform/state-indicator/state-activity-around-expanding-medicaid-under-the-affordable-care-act/?currentTimeframe=0&sortModel=%7B%22colId%22:%22Location%22,%22sort%22: %22asc%22%7D . Accessed April 15, 2019.
In addition, five states—Idaho, Maine, Nebraska, Utah, and Virginia—adopted the Medicaid expansion after 2016. Maine did so in 2017 via a ballot initiative and implemented the program in 2019. Idaho, Nebraska, and Utah also adopted the adult expansion via ballot initiative in 2018. Idaho is scheduled to implement the program on January 1, 2020, Nebraska is planning to do so on October 1, 2020, and Utah implemented a limited expansion in April 2019. Virginia’s legislature voted to adopt the adult expansion in 2018 and began enrollment on January 1, 2019.
Arizona, Arkansas, Iowa, Indiana, Kentucky, Michigan, Montana, and New Hampshire expanded Medicaid through an 1115 demonstration waiver. Pennsylvania initially implemented its adult expansion as an 1115 demonstration in 2015 but shifted to a state plan amendment a few months later.
A beneficiary is also considered “newly eligible” if he or she would have been eligible but could not have been enrolled for these benefits or this coverage because the applicable Medicaid waiver or demonstration had limited or capped enrollment as of December 1, 2009.
Centers for Medicare & Medicaid Services. “Medicaid and CHIP FAQs: Newly Eligible FMAP and Expansion State FMAP.” February 2013. Available at: http://www.statecoverage.org/files/ACA-FAQ-BHP.pdf . Accessed April 15, 2019.
States that met certain statutory criteria for health benefits coverage, as described in section 1905(z)(3) of the Affordable Care Act, qualified for increased Federal Medical Assistance Percentage (FMAP) rates for a subset of the \""not newly eligible\"" population. These states were designated as 1905(z)(3) expansion states. Eligibility group 73 identifies not newly eligible beneficiaries for non-1905(z)(3) states. Eligibility groups 74 and 75 identify not newly eligible beneficiaries for 1905(z)(3) states, with eligibility group 74 specifying parents, caretakers, or relatives of expansion beneficiaries.
Adults entitled to Medicare and those pregnant or eligible for Medicaid under another mandatory eligibility group are generally not eligible for the adult expansion group, which includes individuals below 133 percent of the FPL after applying a 5 percent income disregard, equating to an effective coverage limit of 138 percent of the FPL.
This expansion population is also known as the VIII Group because the eligibility criteria for this group are defined in Section 1902(10)(VIII) of the Social Security Act.
Paradise, Julia. \u201cMoving Medicaid Forward.\u201d San Francisco, CA: The Kaiser Commission on Medicaid and the Uninsured, March 9, 2015. Available at: https://www.kff.org/health-reform/issue-brief/medicaid-moving-forward/ . Accessed December 14, 2018.
Rosenbaum, Sara, and Timothy M. Westmoreland. \u201cThe Supreme Court\u2019s Surprising Decision on the Medicaid Expansion; How Will the Federal Government and States Proceed?\u201d Health Affairs, vol. 31, no. 8, August 2012. Available at: https://www.healthaffairs.org/doi/abs/10.1377/hlthaff.2012.0766 . Accessed December 14, 2018.
Kaiser Family Foundation. \u201cStatus of State Action on the Medicaid Expansion Decision.\u201d April 9, 2019. Available at: https://www.kff.org/health-reform/state-indicator/state-activity-around-expanding-medicaid-under-the-affordable-care-act/?currentTimeframe=0&sortModel=%7B%22colId%22:%22Location%22,%22sort%22: %22asc%22%7D . Accessed April 15, 2019.
In addition, five states\u2014Idaho, Maine, Nebraska, Utah, and Virginia\u2014adopted the Medicaid expansion after 2016. Maine did so in 2017 via a ballot initiative and implemented the program in 2019. Idaho, Nebraska, and Utah also adopted the adult expansion via ballot initiative in 2018. Idaho is scheduled to implement the program on January 1, 2020, Nebraska is planning to do so on October 1, 2020, and Utah implemented a limited expansion in April 2019. Virginia\u2019s legislature voted to adopt the adult expansion in 2018 and began enrollment on January 1, 2019.
Arizona, Arkansas, Iowa, Indiana, Kentucky, Michigan, Montana, and New Hampshire expanded Medicaid through an 1115 demonstration waiver. Pennsylvania initially implemented its adult expansion as an 1115 demonstration in 2015 but shifted to a state plan amendment a few months later.
A beneficiary is also considered \u201cnewly eligible\u201d if he or she would have been eligible but could not have been enrolled for these benefits or this coverage because the applicable Medicaid waiver or demonstration had limited or capped enrollment as of December 1, 2009.
Centers for Medicare & Medicaid Services. \u201cMedicaid and CHIP FAQs: Newly Eligible FMAP and Expansion State FMAP.\u201d February 2013. Available at: http://www.statecoverage.org/files/ACA-FAQ-BHP.pdf . Accessed April 15, 2019.
States that met certain statutory criteria for health benefits coverage, as described in section 1905(z)(3) of the Affordable Care Act, qualified for increased Federal Medical Assistance Percentage (FMAP) rates for a subset of the \""not newly eligible\"" population. These states were designated as 1905(z)(3) expansion states. Eligibility group 73 identifies not newly eligible beneficiaries for non-1905(z)(3) states. Eligibility groups 74 and 75 identify not newly eligible beneficiaries for 1905(z)(3) states, with eligibility group 74 specifying parents, caretakers, or relatives of expansion beneficiaries.
We used Medicaid Budget and Expenditure System (MBES) enrollment data as the external benchmark to examine the accuracy of the TAF-based enrollment counts for the adult expansion group. [11] We examined adult expansion population enrollment as a whole and for the subset of newly eligible beneficiaries using the TAF annual Demographic & Eligibility (DE) file. The MBES data are submitted by states to CMS on Form CMS-64 and include aggregated monthly counts of total Medicaid enrollment, enrollment for Medicaid beneficiaries in the adult expansion group, and enrollment for the subsets of newly eligible and not newly eligible adult expansion beneficiaries. [12] Substantial differences between the TAF-based counts and MBES counts raise concerns about the validity of a state’s T-MSIS reporting.
To investigate this issue, we obtained a monthly count of beneficiaries with eligibility group codes in the TAF DE file that indicated they were in the adult expansion group (ELGBLTY_GRP_CD_mm = 72, 73, 74, or 75). These codes indicate that individuals were eligible for and enrolled in Medicaid in a given month under one of the adult expansion group categories, in which a value of 72 designates newly eligible beneficiaries, and a value of 73, 74, or 75 designates beneficiaries who are not newly eligible. [13]
We also obtained a monthly count of the VIII Group enrollees and VIII Group newly eligible enrollees reported by states in the MBES data. For the adult expansion population and for the newly eligible subset in each state and each month, we calculated a percent difference between the TAF-based enrollment counts and the benchmark, and then averaged the monthly percent difference across all 12 months of the calendar year. We also examined the standard deviation of this measure to assess the variation in the difference across months. Because the benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: the difference between the TAF and MBES counts divided by the MBES count.
We evaluated the average percent difference against the thresholds listed in Table 1 to categorize the states into levels of concern about data quality according to the alignment between the counts of beneficiaries reported in the TAF DE and the counts in the benchmark. Although we did not assign the level of concern based on the standard deviation, we provide this information in the table, and TAF users may want to consider the monthly variability between TAF and the benchmark when determining whether the data are usable for their analysis.
Table 1. Criteria for DQ assessment of enrollment in adult expansion group
| Average monthly percent difference between TAF and MBES enrollment counts | Level of alignment with between TAF and MBES enrollment counts | DQ assessment |
|---|---|---|
| x ≤ 5 percent | High | Low concern |
| 5 percent < x ≤ 10 percent | Moderate | Low concern |
| 10 percent < x ≤ 20 percent | Low | Medium concern |
| 20 percent < x ≤ 50 percent | Very low | High concern |
| x > 50 percent | Very low | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Centers for Medicare & Medicaid Services. “Medicaid Enrollment Data Collected Through MBES.” November 2018. Available at https://www.medicaid.gov/medicaid/program-information/medicaid-and-chip-enrollment-data/enrollment-mbes/index.html . Accessed December 24, 2018.
In early years of T-MSIS reporting, at least one state (Pennsylvania) reported its adult expansion beneficiaries using eligibility group code 71 (other expansions under 1115 authority). Because other states use this eligibility group code to identify non-expansion populations covered under the 1115 authority, it was not used in this analysis to calculate TAF-based counts of the adult expansion population.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Centers for Medicare & Medicaid Services. \u201cMedicaid Enrollment Data Collected Through MBES.\u201d November 2018. Available at https://www.medicaid.gov/medicaid/program-information/medicaid-and-chip-enrollment-data/enrollment-mbes/index.html . Accessed December 24, 2018.
In early years of T-MSIS reporting, at least one state (Pennsylvania) reported its adult expansion beneficiaries using eligibility group code 71 (other expansions under 1115 authority). Because other states use this eligibility group code to identify non-expansion populations covered under the 1115 authority, it was not used in this analysis to calculate TAF-based counts of the adult expansion population.
Under the Affordable Care Act, some states elected to expand coverage to all low-income adults with an income below 138 percent of the federal poverty level. Individuals in the adult expansion group are considered \""newly eligible\"" if they would not have qualified for Medicaid coverage under the eligibility rules in place as of December 1, 2009. This analysis examines how well the TAF data on newly eligible adult enrollment align with an external benchmark, the Medicaid Budget and Expenditure System.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4041"", ""relatedTopics"": [{""measureId"": 5, ""measureName"": ""Adult Expansion Enrollment"", ""groupId"": 1, ""groupName"": ""Enrollment Benchmarking"", ""order"": 0}]}" 7,"{""measureId"": 7, ""measureName"": ""Total Medicaid and CHIP Enrollment"", ""groupId"": 1, ""groupName"": ""Enrollment Benchmarking"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Total-Enroll.pdf"", ""background"": {""content"": ""The research-ready T\u2011MSIS Analytic Files (TAF), are an enhanced set of data on beneficiaries in Medicaid and the Children’s Health Insurance Program (CHIP), their claims, and the participating managed care plans and providers that serve them. The TAF eligibility files contain beneficiaries enrolled in Medicaid or CHIP at any point during the year, including beneficiaries who qualify for comprehensive medical coverage as well as those who only qualify for limited benefits. [1] Analyses that require a relatively complete record of an individual’s service use often need to be restricted to Medicaid beneficiaries with comprehensive benefits at a minimum.
Five variables in the annual Demographic and Eligibility (DE) TAF can be used to identify Medicaid and CHIP beneficiaries who have comprehensive benefits (Table 1).
Table 1. Potential TAF DE variables for identifying Medicaid and CHIP beneficiaries
| Data element | Use for identifying Medicaid and CHIP beneficiaries |
|---|---|
| CHIP_CD | Identifies individuals in Medicaid (CHIP_CD = 1), Medicaid Expansion CHIP (M-CHIP) (CHIP_CD = 2), and Separate CHIP (S-CHIP) (CHIP_CD = 3). |
| MDCD_ENRLMT_DAYS and CHIP_ENRLMT_DAYS | Built from T-MSIS enrollment date and enrollment type variables (ENROLLMENT-EFF-DATE, ENROLLMENT-END-DATE, and ENROLLMENT-TYPE). The beneficiary will have at least one day of Medicaid or M-CHIP enrollment (MDCD_ENRLMT_DAYS > 0) if he or she had an enrollment span that covered at least one day in the month and the enrollment span was classified as ENROLLMENT-TYPE = 1 (Medicaid or M-CHIP). The beneficiary will have at least one day of S-CHIP enrollment (CHIP_ENRLMT_DAYS > 0) if he or she had an enrollment span that covered at least one day in the month and the enrollment span was classified as S-CHIP (ENROLLMENT-TYPE = 2 [Separate Title XXI CHIP]). |
| ELGBLTY_GRP_CD | Contains the eligibility group applicable to the individual based on the state’s eligibility determination process. Can be used to distinguish S-CHIP and M-CHIP (ELGBLTY_GRP_CD = 61–68) from Medicaid (ELGBLTY_GRP_CD = 1-9, 11-56, 59–60, or 69–75) enrollment. |
| RSTRCTD_BNFTS_CD | Indicates the scope of Medicaid or CHIP benefits to which a beneficiary is entitled during the month. Can be used to distinguish between individuals not eligible for Medicaid or CHIP benefits during the month (value of 0); those enrolled with full or comprehensive benefits (values of 1, 4, 5, 7, A, B, or D); a and those enrolled with limited benefits (values of 2, 3, 6, C, E, or F). Beneficiaries with a restricted benefits code of 4 (restricted benefits for pregnancy-related services) have benefits that meet the Minimum Essential Coverage (MEC) requirements in all states except Arkansas, Idaho, and South Dakota. Beneficiaries with a restricted benefits code of 4 in those three states have limited benefits. |
Note: \tThese five data elements are available monthly in the TAF DE, with the number of each month appended to the end of the data element name (for instance, CHIP_CD_01 for January, CHIP_CD_02 for February, and so on). For simplicity, we did not list the monthly indicators in this table because this analysis used all months of data. A list of valid values and descriptions of these data elements can be found in the TAF Demographic and Eligibility Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries .
Maintenance assistance status and basis of eligibility (MASBOE_CD), which was constructed from T-MSIS data elements MAINTENANCE-ASSISTANCE-STATUS and MEDICAID-BASIS-OF-ELIGIBILITY, has been phased out in favor of the new, more detailed eligibility group code; MASBOE_CD is not recommended for use.
A restricted benefits code value of 1 indicates full-scope Medicaid or CHIP benefits; value 4 indicates that the individual is eligible for Medicaid or CHIP but only entitled to restricted benefits for pregnancy-related services; value 5 indicates that the individual is eligible for Medicaid or CHIP, but for reasons other than alien, dual-eligibility, or pregnancy-related status, is only entitled to restricted benefits that meet the MEC standard; value 7 indicates Medicaid enrollment in an alternative package of benchmark-equivalent Medicaid coverage; value A indicates entitlement to Medicaid benefits under the Psychiatric Residential Treatment Facilities Demonstration Grant; value B indicates entitlement to Medicaid benefits using a Health Opportunity Account; and value D indicates entitlement to Medicaid benefits under a Money Follows the Person rebalancing demonstration.
Analyses conducted on the 2016 TAF data (results not shown) found that using the restricted benefits code alone is the most reliable approach for counting the number of Medicaid and CHIP beneficiaries with comprehensive benefits. This analysis evaluates whether this data element can be used to accurately count enrollment for this group in each state.
States can offer limited Medicaid benefits to individuals based on alien status, dual eligibility, or pregnancy-related status. In addition, some beneficiaries are eligible for family planning or emergency services only. These benefit packages do not meet the Minimum Essential Coverage threshold and are therefore not considered comprehensive benefits.
States can offer limited Medicaid benefits to individuals based on alien status, dual eligibility, or pregnancy-related status. In addition, some beneficiaries are eligible for family planning or emergency services only. These benefit packages do not meet the Minimum Essential Coverage threshold and are therefore not considered comprehensive benefits.
We used Eligibility and Enrollment Performance Indicator (PI) data as the external benchmark to examine the accuracy of TAF-based enrollment counts for the Medicaid and CHIP population with comprehensive benefits. [2] Although some states’ PI data contain quality issues prior to 2017 that may affect their accuracy, these data are the best source available for use as an external benchmark for the Medicaid and CHIP population because many of the data quality issues are known, and the data provide a consistent benchmark across multiple data quality assessments. [3]
In the TAF, we identified Medicaid and CHIP beneficiaries with comprehensive benefits by selecting records where the restricted benefits code was equal to 1, 4, 5, 7, A, B, or D in all states but Arkansas, Idaho and South Dakota, which do not provide comprehensive benefits to Medicaid beneficiaries who are eligible because of pregnancy. In these states, we calculated TAF-based enrollment counts for beneficiaries with comprehensive benefits using a restricted benefits code equal to 1, 5, 7, A, B, or D. [4]
Using the restricted benefits code to create TAF-based counts has the effect of including all individuals enrolled for one or more days (“ever enrolled”) in the month. We compared the PI and TAF-based counts by (1) evaluating the percent difference between TAF-based enrollment counts and the benchmark, averaged across all 12 months; and (2) examining the standard deviation of this measure to assess variation in the difference across months. The average monthly TAF enrollment count is calculated as the sum of the monthly TAF counts divided by 12, and the average monthly PI enrollment count is calculated as the sum of the monthly PI counts divided by 12. Because the benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, the percent difference between the two is calculated as a percent error or change: the difference between the TAF and PI counts divided by the PI count and multiplied by 100. The average monthly percent difference is calculated as the sum of the monthly percent differences, divided by 12. [5] The standard deviation of the monthly percent differences between the TAF-based count and the PI count is calculated as the square root of the sum of the squared differences between the monthly percent differences and the average percent difference, divided by 12.
Table 2 shows the level of concern for the TAF Medicaid enrollment counts based on both the percent difference and the level of alignment between the TAF and the PI enrollment counts. Although we did not assign the level of concern based on the standard deviation, we provide this information in the tables, and TAF users may want to consider the monthly variability between TAF and the benchmark when determining whether the data are usable for their analysis and whether all months are of similar quality.
Table 2. Criteria for DQ assessment of Medicaid and CHIP enrollment
| Average monthly percent difference between TAF and PI enrollment counts | Level of alignment between TAF and PI enrollment counts | DQ assessment |
|---|---|---|
| x ≤ 5 percent | High | Low concern |
| 5 percent < x ≤ 10 percent | Moderate | Low concern |
| 10 percent < x ≤ 20 percent | Low | Medium concern |
| 20 percent < x ≤ 50 percent | Very low | High concern |
| x > 50 percent | Very low | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
More information about the PI data set can be found at https://www.medicaid.gov/medicaid/program-information/medicaid-and-chip-enrollment-data/report-highlights/index.html . In some cases, the PI data in the Atlas may not match exactly the PI data publicly available on Medicaid.gov, because our analysis uses a version of the data set that may have been updated more recently.
As of 2020, the restricted benefits code value of 5 (the individual is eligible for Medicaid or Medicaid-Expansion CHIP but, for reasons other than alien, dual-eligibility, or pregnancy-related status, is entitled to restricted benefits only) should be used only if the coverage meets the MEC standard and a new valid value of E should be used if the coverage does not meet the MEC standard. For years prior to 2020, we did not include the code 5 group for any state because it represented a more heterogenous mix of beneficiaries (some of whom had limited benefits in some states).
The difference between TAF and PI enrollment is based on an average of the monthly differences between these two data sources. As a result, it may not equal the difference between the average annual TAF enrollment and average annual PI enrollment.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
More information about the PI data set can be found at https://www.medicaid.gov/medicaid/program-information/medicaid-and-chip-enrollment-data/report-highlights/index.html . In some cases, the PI data in the Atlas may not match exactly the PI data publicly available on Medicaid.gov, because our analysis uses a version of the data set that may have been updated more recently. \u00a0
As of 2020, the restricted benefits code value of 5 (the individual is eligible for Medicaid or Medicaid-Expansion CHIP but, for reasons other than alien, dual-eligibility, or pregnancy-related status, is entitled to restricted benefits only) should be used only if the coverage meets the MEC standard and a new valid value of E should be used if the coverage does not meet the MEC standard. For years prior to 2020, we did not include the code 5 group for any state because it represented a more heterogenous mix of beneficiaries (some of whom had limited benefits in some states).
The difference between TAF and PI enrollment is based on an average of the monthly differences between these two data sources. As a result, it may not equal the difference between the average annual TAF enrollment and average annual PI enrollment.
The TAF eligibility files include information on beneficiaries in both Medicaid and CHIP. This analysis examines how well the TAF data on the number of Medicaid and CHIP beneficiaries align with an external benchmark, the Performance Indicators data set.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4051"", ""relatedTopics"": []}" 8,"{""measureId"": 8, ""measureName"": ""Medicaid-Only Enrollment"", ""groupId"": 1, ""groupName"": ""Enrollment Benchmarking"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Medicaid-Only-Enroll.pdf"", ""background"": {""content"": ""The research-ready T\u2011MSIS Analytic Files (TAF), are an enhanced set of data on beneficiaries in Medicaid and the Children’s Health Insurance Program (CHIP), their claims, and the participating managed care plans and providers that serve them. The TAF eligibility files include beneficiaries enrolled in Medicaid, Medicaid expansion CHIP (referred to as M-CHIP) and separate CHIP (referred to as S-CHIP). [1] Many data users will want to identify and study only beneficiaries in non-CHIP (Title XIX-funded) Medicaid (and for analyses requiring a relatively complete record of individuals’ service use, particularly those with comprehensive benefits [2] ) and will need to know whether these beneficiaries are completely reported in a state’s TAF data.
Three variables in the annual Demographic and Eligibility (DE) TAF can be used in combination to identify Title XIX Medicaid beneficiaries (which does not include M-CHIP) who have comprehensive benefits (Table 1). These variables are CHIP code (CHIP_CD), eligibility group code (ELGBLTY_GRP_CD), and restricted benefits code (RSTRCTD_BNFTS_CD). The CHIP code (CHIP_CD) is most useful for distinguishing beneficiaries enrolled in a Title XIX Medicaid program, from those enrolled in the Title XXI M-CHIP, and from those enrolled in Title XXI S-CHIP. It is the only variable that can be used to identify the entire Title XXI CHIP population. TAF users cannot identify Title XIX Medicaid enrollment by using the Medicaid enrollment days variable (MDCD_ENRLMT_DAYS) because that variable also includes days during the month in which the beneficiary was enrolled in M-CHIP. While eligibility group code (ELGBLTY_GRP_CD) is most useful for obtaining detailed information on the eligibility group through which a beneficiary is enrolled in Medicaid or CHIP, it cannot be used alone to identify all beneficiaries enrolled in Title XXI CHIP or to separate Title XXI M-CHIP from Title XXI S-CHIP beneficiaries due to overlap in the use of certain eligibility groups.
Table 1. Potential TAF DE variables for identifying Title XIX Medicaid beneficiaries
| Data element | Use for identifying Medicaid and CHIP beneficiaries |
|---|---|
| CHIP_CD | Identifies individuals in Medicaid (CHIP_CD = 1), Medicaid expansion CHIP (M-CHIP) (CHIP_CD = 2), and Separate CHIP (S-CHIP) (CHIP_CD = 3). |
| ELGBLTY_GRP_CD | Contains the eligibility group applicable to the individual based on the state’s eligibility determination process. When CHIP_CD is missing, ELGBLTY_GRP_CD can be used to indicate CHIP enrollment (ELGBLTY_GRP_CD = 61–68, with 61 used for both M-CHIP and S-CHIP and 62–68 exclusive to S-CHIP) and Medicaid enrollment (ELGBLTY_GRP_CD = 1–60 or 69–76). |
| RSTRCTD_BNFTS_CD | Indicates the scope of Medicaid benefits to which a beneficiary is entitled during the month. Can be used to distinguish between individuals not eligible for Medicaid or CHIP benefits during the month (value of 0); those enrolled with full or comprehensive benefits (values of 1, 4, 5, 7, A, B, or D); and those enrolled with limited or partial benefits (values of 2, 3, 6, C, E, or F). Beneficiaries with a restricted benefits code of 4 (restricted benefits for pregnancy-related services) have benefits that meet the Minimum Essential Coverage (MEC) requirements in all states except Arkansas, Idaho, and South Dakota. Beneficiaries with a restricted benefits code of 4 in those three states have limited benefits. |
Note:\tThese three data elements are available monthly in the TAF DE, with the number of each month appended to the end of the data element name (for instance, CHIP_CD_01 for January, CHIP_CD_02 for February, and so on). For simplicity, we did not list the monthly indicators in this brief because we used all months of data. A list of valid values and descriptions of these data elements can be found in the TAF Demographic and Eligibility Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries .
Maintenance assistance status and basis of eligibility (MASBOE_CD), which was constructed from T-MSIS data elements MAINTENANCE-ASSISTANCE-STATUS and MEDICAID-BASIS-OF-ELIGIBILITY, has been phased out in favor of the new, more detailed T-MSIS eligibility group code; MASBOE_CD is not recommended for use.
We also do not include Medicaid enrollment days (MDCD_ENRLMT_DAYS) or CHIP enrollment days (CHIP_ENRLMT_DAYS) because they cannot be used to distinguish the Medicaid-only population funded through Title XIX. The Medicaid enrollment days variable combines Medicaid with M-CHIP; CHIP enrollment days is just for S-CHIP. In addition, the TAF DE equivalent of the T-MSIS enrollment type variable (ENROLLMENT-TYPE) is enrollment type flag (ENRL_TYPE_FLAG), which groups beneficiaries in the same manner as Medicaid and CHIP enrollment days and therefore cannot be used to identify the Medicaid-only population.
A restricted benefits code value of 1 indicates full-scope Medicaid or CHIP benefits; value 4 indicates that the individual, although eligible for Medicaid or CHIP, is entitled to only restricted benefits for pregnancy-related services; value 5 indicates that the individual is eligible for Medicaid or CHIP, but for reasons other than alien, dual-eligibility, or pregnancy-related status, is only entitled to restricted benefits that meet the MEC standard; value 7 indicates Medicaid enrollment in an alternative package of benchmark-equivalent Medicaid coverage; value A indicates entitlement to Medicaid benefits under the Psychiatric Residential Treatment Facilities Demonstration Grant; value B indicates entitlement to Medicaid benefits using a Health Opportunity Account; and value D indicates entitlement to Medicaid benefits under a Money Follows the Person rebalancing demonstration.
Analyses conducted on the 2016 TAF data found that using a combination of all three data elements is the most reliable approach for counting the number of beneficiaries enrolled in Title XIX Medicaid who qualify for comprehensive benefits (results not shown). This analysis evaluates whether these data elements can be used to accurately count enrollment among Title XIX Medicaid beneficiaries with comprehensive benefits in each state.
States may use CHIP funds to expand their Medicaid program (M-CHIP), create a program separate from their existing Medicaid program (S-CHIP), or adopt a combination of both approaches. Medicaid beneficiaries qualify for Title XIX funding, M-CHIP beneficiaries qualify for enhanced Title XXI finding, and S-CHIP beneficiaries qualify for Title XXI funding.
Comprehensive benefits refers to coverage that is comparable to that provided to categorically needy Medicaid beneficiaries and considered Minimum Essential Coverage under the Affordable Care Act. In addition, states can offer restricted Medicaid benefits to individuals on the basis of alien status, dual eligibility, or pregnancy-related status; some beneficiaries are eligible for only limited benefits such as family planning or emergency services.
States may use CHIP funds to expand their Medicaid program (M-CHIP), create a program separate from their existing Medicaid program (S-CHIP), or adopt a combination of both approaches. Medicaid beneficiaries qualify for Title XIX funding, M-CHIP beneficiaries qualify for enhanced Title XXI finding, and S-CHIP beneficiaries qualify for Title XXI funding.
Comprehensive benefits refers to coverage that is comparable to that provided to categorically needy Medicaid beneficiaries and considered Minimum Essential Coverage under the Affordable Care Act. In addition, states can offer restricted Medicaid benefits to individuals on the basis of alien status, dual eligibility, or pregnancy-related status; some beneficiaries are eligible for only limited benefits such as family planning or emergency services.
We used Eligibility and Enrollment Performance Indicator (PI) data as the external benchmark to examine the accuracy of TAF-based enrollment counts for the Medicaid population with comprehensive benefits. [3] Although some states’ PI data contain quality issues prior to 2017 that may affect their accuracy, these data are the best source available for use as an external benchmark for the Medicaid population because many of the data quality issues are known, and the data provide a consistent benchmark across multiple data quality assessments. [4]
To create the TAF-based counts, we included individuals enrolled at any point (“ever enrolled”) in the month. Specifically, we used CHIP code of 1 to identify the Title XIX Medicaid population. If the CHIP code was missing, we counted beneficiaries with an eligibility group code that indicated they were eligible for Medicaid benefits (eligibility group code of 1–60 or 69–76). We then used restricted benefits code of 1, 4, 5, 7, A, B, or D to identify those with comprehensive benefits. We excluded the restricted benefits code 4 group in the three states that do not extend comprehensive benefits to women in the pregnancy group (Arkansas, Idaho, and South Dakota). [5]
We compared the performance of different variables by (1) evaluating the percent difference between TAF-based enrollment counts and the benchmark, averaged across all 12 months, and (2) examining the standard deviation of this measure to assess variation in the difference across months. The average monthly TAF enrollment count is calculated as the sum of the monthly TAF counts divided by 12, and the average monthly PI enrollment count is calculated as the sum of the monthly PI counts divided by 12. Because the benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, the percent difference is calculated as a percent error or change: the difference between the TAF and PI counts divided by the PI count and multiplied by 100. The average monthly percent difference is calculated as the sum of the monthly percent differences, divided by 12. [6] The standard deviation of the monthly percent differences between the TAF-based count and the PI count is calculated as the square root of the sum of the squared differences between the monthly percent differences and the average percent difference, divided by 12.
Table 2 shows the level of concern for the TAF Title XIX Medicaid enrollment counts based on both the percent difference and the level of alignment between the TAF and the PI enrollment counts. Although we did not assign the level of concern based on the standard deviation, we provide this information in the tables, and TAF users may want to consider the monthly variability between TAF and the benchmark when determining whether the data are usable for their analysis and whether all months are of similar quality.
Table 2. Criteria for DQ assessment of Title XIX Medicaid enrollment
| Average monthly percent difference between TAF and PI enrollment counts | Level of alignment between TAF and PI enrollment counts | DQ assessment |
|---|---|---|
| x ≤ 5 percent | High | Low concern |
| 5 percent < x ≤ 10 percent | Moderate | Low concern |
| 10 percent < x ≤ 20 percent | Low | Medium concern |
| 20 percent < x ≤ 50 percent | Very low | High concern |
| x > 50 percent | Very low | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
More information about the PI data set can be found at https://www.medicaid.gov/medicaid/program-information/medicaid-and-chip-enrollment-data/report-highlights/index.html . In some cases, the PI data in the Atlas may not match exactly the PI data publicly available on Medicaid.gov, because our analysis uses a version of the data set that may have been updated more recently.
As of 2020, the restricted benefits code value of 5 (the individual is eligible for Medicaid or Medicaid-Expansion CHIP but, for reasons other than alien, dual-eligibility, or pregnancy-related status, is entitled to restricted benefits only) should be used only if the coverage meets the MEC standard and a new valid value of E should be used if the coverage does not meet the MEC standard. For years prior to 2020, we did not include the code 5 group for any state because it represented a more heterogenous mix of beneficiaries (some of whom had limited benefits in some states).
The difference between TAF and PI enrollment is based on an average of the monthly differences between these two data sources. As a result, it may not equal the difference between the average annual TAF enrollment and average annual PI enrollment.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
More information about the PI data set can be found at https://www.medicaid.gov/medicaid/program-information/medicaid-and-chip-enrollment-data/report-highlights/index.html . In some cases, the PI data in the Atlas may not match exactly the PI data publicly available on Medicaid.gov, because our analysis uses a version of the data set that may have been updated more recently.
As of 2020, the restricted benefits code value of 5 (the individual is eligible for Medicaid or Medicaid-Expansion CHIP but, for reasons other than alien, dual-eligibility, or pregnancy-related status, is entitled to restricted benefits only) should be used only if the coverage meets the MEC standard and a new valid value of E should be used if the coverage does not meet the MEC standard. For years prior to 2020, we did not include the code 5 group for any state because it represented a more heterogenous mix of beneficiaries (some of whom had limited benefits in some states).
The difference between TAF and PI enrollment is based on an average of the monthly differences between these two data sources. As a result, it may not equal the difference between the average annual TAF enrollment and average annual PI enrollment.
The TAF eligibility files include information on beneficiaries in both Medicaid and CHIP. This analysis examines how well the TAF data on the number of total Medicaid beneficiaries align with an external benchmark, the Performance Indicators data set.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4061"", ""relatedTopics"": []}" 9,"{""measureId"": 9, ""measureName"": ""Dually Enrolled in Medicare"", ""groupId"": 1, ""groupName"": ""Enrollment Benchmarking"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Dually-Enrolled.pdf"", ""background"": {""content"": ""Dually eligible beneficiaries are Medicaid beneficiaries also enrolled in Medicare Part A (hospital insurance and other costs) and/or Medicare Part B (medical insurance) and/or a Medicare Savings Program (MSP). [1] , [2] For these beneficiaries, Medicare is the primary payer for services that are jointly covered by both programs. Medicaid covers services that Medicare does not cover as well as Medicare premiums and cost-sharing. [3] All dually eligible beneficiaries qualify for full Medicare benefits, but the level of benefits for which they are eligible under Medicaid can vary, generally depending upon the beneficiary’s income and asset levels, disability, and employment status. Users of the T\u2011MSIS Analytic Files (TAF) data may study dually eligible beneficiaries because they often have complex and costly health care needs. Additionally, some TAF users may need to remove dually eligible beneficiaries from their analyses because the TAF does not capture all of their service use (unless data users themselves link the TAF to Medicare claims). It is therefore important to be able to accurately identify dually eligible beneficiaries in the TAF.
The dually eligible fall into two groups—“partial-benefit dual eligibility” and “full-benefit dual eligibility”—depending upon the level of Medicaid benefits for which an individual is eligible. There are seven types of dual eligibility, four of which are considered partial-benefit and three of which are considered full-benefit. Beneficiaries with partial-benefit dual eligibility are entitled to have Medicaid pay for only some of the expenses they incur under Medicare. These expenses include the premiums for Part A and for Part B, if applicable. Medicaid may also pay for some other cost-sharing amounts owed under Medicare, such as deductibles, coinsurance, and copayments. In addition to the benefits to which beneficiaries with partial-benefit dual eligibility are entitled, beneficiaries with full-benefit dual eligibility are entitled to Medicaid coverage for various health care services that Medicare does not cover, such as most types of long-term services and supports. Table 1 maps the seven eligibility groups to values of the TAF dual-eligible code data element. [4]
Table 1. Dually eligible beneficiaries in the TAF and their benefits
| Dually eligible groups | TAF dual-eligible code | Medicare Part A premiums (when applicable) | Medicare Part B premiums | Coinsurance under Medicare Part A and Part B | Coverage of all Medicaid benefits (full dual eligibility) |
|---|---|---|---|---|---|
| Qualified Medicare Beneficiary (QMB) Only | 01 | X | X | X |
|
| QMB Plus | 02 | X | X | X | X |
| Specified Low-Income Medicare Beneficiaries (SLMB) Only | 03 |
| X |
|
|
| SLMB Plus | 04 |
| X |
| X |
| Qualified Disabled and Working Individual (QDWI) | 05 | X |
|
|
|
| Qualified Individual (QI) | 06 |
| X |
|
|
| Other Full Dual Eligibles | 08 |
|
|
| X |
Note: We did not list or include the dual-eligible code value of 10 (Separate CHIP Eligible is entitled to Medicare) because this assessment focuses only on beneficiaries dually eligible for Medicare. There is also a dual-eligible code of 09 (Eligible is entitled to Medicare—Other [this code is to be used only with specific CMS approval]). This category cannot be partitioned into the partial-benefit or full-benefit dually eligible groups. Because this group does not appear to be reported in the Medicare Modernization Act (MMA) data, beneficiaries in this group are not included in the analyses presented here.
This data quality assessment evaluates the reliability of the TAF-based counts of dually eligible beneficiaries by benchmarking them to an external source, the Medicare-Medicaid (MM) Enrollment Snapshot.
Medicare Shared Savings Programs cover costs such as Part A premiums and Part A and B deductibles, coinsurance, and copayments, depending upon the program.
This assessment does not include CHIP beneficiaries.
This background information is available in the CMS guidance, Reporting Expectations for Dual-Eligible Beneficiaries, which is available in the T-MSIS coding blog: https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=51064 and in the TAF Demographic and Eligibility Codebook available at https://www2.ccwdata.org/web/guest/data-dictionaries .
This data element is available monthly in the TAF DE, with the number of each month appended to the end of the data element name (for instance, DUAL_ELGBL_CD_01 for January, DUAL_ELGBL_CD_02 for February, and so on). For simplicity, we did not list the monthly indicators in this assessment.
Medicare Shared Savings Programs cover costs such as Part A premiums and Part A and B deductibles, coinsurance, and copayments, depending upon the program.
This assessment does not include CHIP beneficiaries.
This background information is available in the CMS guidance, Reporting Expectations for Dual-Eligible Beneficiaries, which is available in the T-MSIS coding blog: https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=51064 and in the TAF Demographic and Eligibility Codebook available at https://www2.ccwdata.org/web/guest/data-dictionaries .
This data element is available monthly in the TAF DE, with the number of each month appended to the end of the data element name (for instance, DUAL_ELGBL_CD_01 for January, DUAL_ELGBL_CD_02 for February, and so on). For simplicity, we did not list the monthly indicators in this assessment.
We compared the number of dually eligible beneficiaries in the TAF [5] to the Medicare-Medicaid Enrollment Snapshot from the Medicare-Medicaid Coordination Office (MMCO). [6] The Medicare-Medicaid Snapshot presents counts of dually eligible beneficiaries by eligibility type at four points in the year (March, June, September, and December). These counts are derived from the Medicare Modernization Act (MMA) data that states submit monthly to CMS. Although the MMA data may have quality issues that might limit their accuracy in some cases, they are considered the best source available for use as an external benchmark for the dually eligible population.
We evaluated how well the dual eligible code (DUAL_ELGBL_CD) captured total enrollment of dually eligible beneficiaries compared with the benchmark. [7] Specifically, we calculated the count of all, partial-benefit, and full-benefit dually eligible beneficiaries in March, June, September, and December using the annual Demographic and Eligibility (DE) TAF. We then compared the TAF counts to the MMA data for those months and calculated the percentage difference in each month. Because the benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, the percent difference is calculated as a percent error or change: the difference between the TAF and MMA counts divided by the MMA count. We assessed the quality of the dual eligible code in TAF based on the average of the percent difference across the four months (Table 2). We also examined the standard deviation of this measure to assess the variation in the difference across the four months. [8]
Table 2. Criteria for DQ assessment of dually eligible enrollment counts
| Average percent difference between TAF counts and MMA counts | Level of alignment | DQ assessment |
|---|---|---|
| x ≤ 5 percent | High | Low concern |
| 5 percent < x ≤ 10 percent | Moderate | Low concern |
| 10 percent < x ≤ 20 percent | Low | Medium concern |
| 20 percent < x ≤ 50 percent | Very low | High concern |
| x > 50 percent | Very low | Unusable |
Although we did not assign the level of concern based on the standard deviation, we provide this information in the table, and TAF users may want to consider the monthly variability between TAF and the benchmark when determining whether the data are usable for their analysis.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
These data are available at https://www.cms.gov/Medicare-Medicaid-Coordination/Medicare-and-Medicaid-Coordination/Medicare-Medicaid-Coordination-Office/Analytics.html .
All dually eligible beneficiaries are identified in the TAF as DUAL_ELGBL_CD = 01, 02, 03, 04, 05, 06, or 08. Full-benefit dually eligible beneficiaries are DUAL_ELGBL_CD = 02, 04, or 08. Partial-benefit dually eligible beneficiaries are DUAL_ELGBL_CD = 01, 03, 05, or 06. We did not include DUAL_ELGBL_CD = 10 (Separate CHIP Eligible is entitled to Medicare) because the benchmarking data include only beneficiaries dually eligible for Medicare.
The difference between the TAF and the MMA enrollment counts is based on an average of the differences between these two data sources across March, June, September, and December. As a result, it may not equal the difference between the average TAF enrollment and the average MMA enrollment.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
These data are available at https://www.cms.gov/Medicare-Medicaid-Coordination/Medicare-and-Medicaid-Coordination/Medicare-Medicaid-Coordination-Office/Analytics.html .
All dually eligible beneficiaries are identified in the TAF as DUAL_ELGBL_CD = 01, 02, 03, 04, 05, 06, or 08. Full-benefit dually eligible beneficiaries are DUAL_ELGBL_CD = 02, 04, or 08. Partial-benefit dually eligible beneficiaries are DUAL_ELGBL_CD = 01, 03, 05, or 06. We did not include DUAL_ELGBL_CD = 10 (Separate CHIP Eligible is entitled to Medicare) because the benchmarking data include only beneficiaries dually eligible for Medicare.
The difference between the TAF and the MMA enrollment counts is based on an average of the differences between these two data sources across March, June, September, and December. As a result, it may not equal the difference between the average TAF enrollment and the average MMA enrollment.
Dually eligible beneficiaries are individuals who are enrolled in both Medicaid and Medicare. This analysis examines how well the TAF data on the dually eligible population align with an external benchmark, the Medicare-Medicaid Enrollment Snapshot.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4071"", ""relatedTopics"": []}" 10,"{""measureId"": 10, ""measureName"": ""Number of Enrollment Spans"", ""groupId"": 2, ""groupName"": ""Enrollment Patterns Over Time"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Number-of-Enroll-Spans.pdf"", ""background"": {""content"": ""Understanding Medicaid and CHIP (Children’s Health Insurance Program) enrollment patterns is critical for assessing continuity of Medicaid and CHIP coverage and beneficiary churn in and out of these programs. T-MSIS, and its research-ready counterpart, the T-MSIS Analytic Files (TAF), contain enrollment records with specific start and end dates for each beneficiary, an enhancement to the prior national Medicaid data system and research files. [1] As long as the dates the states provide are accurate, TAF users can use those dates to examine detailed enrollment patterns.
We used the effective eligibility dates available in the annual Demographic and Eligibility (DE) file to identify problematic enrollment patterns indicative of data quality issues. First, we examined beneficiaries’ enrollment spans, which is a single enrollment episode with a start date and an end date.
Medicaid and CHIP beneficiaries can disenroll and subsequently re-enroll in these programs within a short period of time (a phenomenon often referred to as churn), and these enrollment patterns vary considerably both across and within states due to variation in eligibility requirements and enrollment redeterminations. For example, states with ex parte redeterminations are likely to experience lower rates of churn. However, recent studies found that overall, only a small proportion of beneficiaries are enrolled for fewer than 12 months, and most individuals who disenroll do not end up reenrolling in the program within 12 months. [2] Based on these findings and policy changes as part of the Patient Protection and Affordable Care Act (ACA), [3] we expect the majority of beneficiaries to have continuous enrollment for the full 12 months of a given year. It is therefore uncommon for a beneficiary to have more than two separate enrollment spans in a single year, so we would expect to see only a small percentage of beneficiaries with three or more enrollment spans.
We also examined the gaps between enrollment spans among those with more than one span during the year. In many states, once an individual is determined eligible, coverage begins on the first day of the month of the application and generally ends on the last day of the month. Therefore, gaps of less than a month may represent administrative errors or other data quality issues rather than true disenrollment and reenrollment in Medicaid or CHIP in these states.
In addition, we examined beneficiaries with enrollment spans that suggest simultaneous enrollment in Medicaid and S-CHIP (overlapping enrollment spans). The building of the TAF DE includes business rules that fix some known data quality issues; to the extent possible, all overlapping and contiguous enrollment spans reported in the T-MSIS data are merged and consolidated. [4] As a result, the TAF enrollment dates are much cleaner than those submitted by the states through T-MSIS. However, there are still beneficiaries with overlapping enrollment spans, representing cases in which the enrollment dates on different state-submitted eligibility records erroneously suggest that the beneficiary is enrolled in Medicaid and S-CHIP simultaneously. [5] For instances in which a beneficiary has overlapping enrollment spans, there are likely data quality issues with the enrollment dates and the enrollment type code, which indicates whether someone is enrolled in Medicaid or S-CHIP. When a beneficiary has overlapping enrollment spans, TAF users must conduct additional analyses to determine in which program the beneficiaries were enrolled.
The Medicaid Analytic eXtract (MAX) person summary file, the predecessor of the TAF, provided monthly enrollment flags for whether a beneficiary was enrolled in Medicaid or CHIP on any day in a given month. The Medicaid Statistical Information System was the data source for the MAX files.
Medicaid and CHIP Payment and Access Commission (MACPAC). “An Updated Look at Rates of Churn and Continuous Coverage in Medicaid and CHIP.” Issue Brief Advising Congress on Medicaid and CHIP Policy. October 2021. Available at https://www.macpac.gov/wp-content/uploads/2021/10/An-Updated-Look-at-Rates-of-Churn-and-Continuous-Coverage-in-Medicaid-and-CHIP.pdf . Accessed February 17, 2022.
In 2010, the ACA (P.L. 111-148 as amended) created a federal standard that limits the frequency of Medicaid and CHIP renewals to no more than once every 12 months for beneficiaries whose income is determined based on modified adjusted gross income (MAGI) (i.e. those under age 65 who are not eligible for Medicaid on the basis of a disability).
In the TAF DE, all enrollment records with overlapping dates are merged, with the exception of records with nonmatching enrollment types. Enrollment type indicates whether a beneficiary is enrolled in Medicaid or CHIP. Overlapping enrollment spells that are not merged indicate that a beneficiary is enrolled in both Medicaid and CHIP for the duration of the overlap.
An individual cannot be enrolled in Medicaid and S-CHIP at the same time, with the exception of a mother and her unborn or deemed newborn.
The Medicaid Analytic eXtract (MAX) person summary file, the predecessor of the TAF, provided monthly enrollment flags for whether a beneficiary was enrolled in Medicaid or CHIP on any day in a given month. The Medicaid Statistical Information System was the data source for the MAX files.
Medicaid and CHIP Payment and Access Commission (MACPAC). \u201cAn Updated Look at Rates of Churn and Continuous Coverage in Medicaid and CHIP.\u201d Issue Brief Advising Congress on Medicaid and CHIP Policy. October 2021. Available at https://www.macpac.gov/wp-content/uploads/2021/10/An-Updated-Look-at-Rates-of-Churn-and-Continuous-Coverage-in-Medicaid-and-CHIP.pdf . Accessed February 17, 2022.
In 2010, the ACA (P.L. 111-148 as amended) created a federal standard that limits the frequency of Medicaid and CHIP renewals to no more than once every 12 months for beneficiaries whose income is determined based on modified adjusted gross income (MAGI) (i.e. those under age 65 who are not eligible for Medicaid on the basis of a disability).
In the TAF DE, all enrollment records with overlapping dates are merged, with the exception of records with nonmatching enrollment types. Enrollment type indicates whether a beneficiary is enrolled in Medicaid or CHIP. Overlapping enrollment spells that are not merged indicate that a beneficiary is enrolled in both Medicaid and CHIP for the duration of the overlap.
An individual cannot be enrolled in Medicaid and S-CHIP at the same time, with the exception of a mother and her unborn or deemed newborn.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
Using the TAF DE [6] , we examined enrollment start dates (ENRLMT_EFCTV_CY_DT) and end dates (ENRLMT_END_CY_DT), which capture each enrollment episode for the beneficiary in the year. We used the start and end dates to examine three different types of enrollment patterns: (1) number of enrollment spans, (2) length of enrollment gaps, and (3) overlapping enrollment spans.
We calculated the share of beneficiaries with one deduplicated enrollment span and the share of those with three or more deduplicated enrollment spans during the year. [7] Table 1 shows how these measures were used to assess data quality. To be categorized into the low concern category, both conditions must be met. In contrast, only one of the conditions needs to be met to be categorized as medium concern or having unusable data.
Table 1. Criteria for DQ assessment of the number of enrollment spans
| Percentage of beneficiaries with only one enrollment span | Percentage of beneficiaries with three or more enrollment spans | DQ assessment |
|---|---|---|
| 80 percent ≤ x ≤ 98 percent | x < 1 percent | Low concern |
| 98 percent < x < 99.5 percent or 0% < x < 80% | 1 percent ≤ x ≤ 3 percent | Medium concern |
| x > 99.5 percent or no beneficiaries | x > 3 percent | Unusable |
Length of enrollment gaps
Using the same method to deduplicate enrollment spans as in the first analysis, we calculated the share of beneficiaries with gaps in enrollment of at least one day [8] and the average number of days between subsequent enrollment spans. We classified states into the low concern category if the average length of the enrollment gap was at least 31 days among beneficiaries with at least two spans of enrollment (Table 2). Since a break in enrollment of less than a month is more likely to reflect a data quality issue than true disenrollment and re-enrollment, we classified any state for which the average length of the enrollment gap was between 1 and 31 days into the medium data quality concern category. States where no beneficiaries had two or more enrollment spans, and as a result the average length of an enrollment gap could not be calculated, were also classified into the medium concern category.
Table 2. Criteria for DQ assessment of the average length of enrollment gaps
| Average length of enrollment gap | Percentage of beneficiaries with an enrollment gap | DQ assessment |
|---|---|---|
| x ≥ 31 days | x > 0 percent | Low concern |
| 0 < x < 31 days | x = 0 percent | Medium concern |
Overlapping enrollment spans
We calculated the share of beneficiaries with any overlapping Medicaid and CHIP enrollment spans in the year, defined as enrollment records for the same beneficiary where the start date for one enrollment span began before or was the same as the end date for the previous span, and one record indicated the beneficiary was in Medicaid and the other record indicated CHIP. We did not deduplicate enrollment spans that had the same enrollment start and end dates before this calculation. We also examined the number of days of overlap between records. Table 3 shows the level of concern for overlapping enrollment spans based on the percentage of beneficiaries with overlapping segment spans.
Table 3. Criteria for DQ assessment of overlapping enrollment spans
| Percentage of beneficiaries with overlapping enrollment spans | DQ assessment |
|---|---|
| x ≤ 1 percent | Low concern |
| 1 percent < x ≤ 5 percent | Medium concern |
| x > 5 percent | High concern |
Methods previously used to assess data quality
Table 4 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 4. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Before we counted enrollment spans, we deduplicated enrollment spans that had the same MSIS ID, enrollment start date, and enrollment end date. This deduplication approach considers completely overlapping enrollment spans as a single span and partially overlapping enrollment spans (those that overlap but do not match on both enrollment start and end dates) as two spans.
A beneficiary with more than one enrollment span but zero days between spans is not considered to have an enrollment gap, but rather an overlapping enrollment span.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Before we counted enrollment spans, we deduplicated enrollment spans that had the same MSIS ID, enrollment start date, and enrollment end date. This deduplication approach considers completely overlapping enrollment spans as a single span and partially overlapping enrollment spans (those that overlap but do not match on both enrollment start and end dates) as two spans.
A beneficiary with more than one enrollment span but zero days between spans is not considered to have an enrollment gap, but rather an overlapping enrollment span.
Effective eligibility dates can be used to understand the detailed enrollment patterns of Medicaid and CHIP beneficiaries. Most beneficiaries in a state should have only one distinct enrollment span during the calendar year, with a small proportion having two or more. This analysis examines patterns in the number of enrollment spans in order to identify states that may have a data quality issue with effective eligibility dates.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4091"", ""relatedTopics"": [{""measureId"": 11, ""measureName"": ""Length of Enrollment Gaps"", ""groupId"": 2, ""groupName"": ""Enrollment Patterns Over Time"", ""order"": 1}, {""measureId"": 12, ""measureName"": ""Overlapping Enrollment Spans"", ""groupId"": 2, ""groupName"": ""Enrollment Patterns Over Time"", ""order"": 2}]}" 11,"{""measureId"": 11, ""measureName"": ""Length of Enrollment Gaps"", ""groupId"": 2, ""groupName"": ""Enrollment Patterns Over Time"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Length-of-Enroll-Gaps.pdf"", ""background"": {""content"": ""Understanding Medicaid and CHIP (Children’s Health Insurance Program) enrollment patterns is critical for assessing continuity of Medicaid and CHIP coverage and beneficiary churn in and out of these programs. T-MSIS, and its research-ready counterpart, the T-MSIS Analytic Files (TAF), contain enrollment records with specific start and end dates for each beneficiary, an enhancement to the prior national Medicaid data system and research files. [1] As long as the dates the states provide are accurate, TAF users can use those dates to examine detailed enrollment patterns.
We used the effective eligibility dates available in the annual Demographic and Eligibility (DE) file to identify problematic enrollment patterns indicative of data quality issues. First, we examined beneficiaries’ enrollment spans, which is a single enrollment episode with a start date and an end date.
Medicaid and CHIP beneficiaries can disenroll and subsequently re-enroll in these programs within a short period of time (a phenomenon often referred to as churn), and these enrollment patterns vary considerably both across and within states due to variation in eligibility requirements and enrollment redeterminations. For example, states with ex parte redeterminations are likely to experience lower rates of churn. However, recent studies found that overall, only a small proportion of beneficiaries are enrolled for fewer than 12 months, and most individuals who disenroll do not end up reenrolling in the program within 12 months. [2] Based on these findings and policy changes as part of the Patient Protection and Affordable Care Act (ACA), [3] we expect the majority of beneficiaries to have continuous enrollment for the full 12 months of a given year. It is therefore uncommon for a beneficiary to have more than two separate enrollment spans in a single year, so we would expect to see only a small percentage of beneficiaries with three or more enrollment spans.
We also examined the gaps between enrollment spans among those with more than one span during the year. In many states, once an individual is determined eligible, coverage begins on the first day of the month of the application and generally ends on the last day of the month. Therefore, gaps of less than a month may represent administrative errors or other data quality issues rather than true disenrollment and reenrollment in Medicaid or CHIP in these states.
In addition, we examined beneficiaries with enrollment spans that suggest simultaneous enrollment in Medicaid and S-CHIP (overlapping enrollment spans). The building of the TAF DE includes business rules that fix some known data quality issues; to the extent possible, all overlapping and contiguous enrollment spans reported in the T-MSIS data are merged and consolidated. [4] As a result, the TAF enrollment dates are much cleaner than those submitted by the states through T-MSIS. However, there are still beneficiaries with overlapping enrollment spans, representing cases in which the enrollment dates on different state-submitted eligibility records erroneously suggest that the beneficiary is enrolled in Medicaid and S-CHIP simultaneously. [5] For instances in which a beneficiary has overlapping enrollment spans, there are likely data quality issues with the enrollment dates and the enrollment type code, which indicates whether someone is enrolled in Medicaid or S-CHIP. When a beneficiary has overlapping enrollment spans, TAF users must conduct additional analyses to determine in which program the beneficiaries were enrolled.
The Medicaid Analytic eXtract (MAX) person summary file, the predecessor of the TAF, provided monthly enrollment flags for whether a beneficiary was enrolled in Medicaid or CHIP on any day in a given month. The Medicaid Statistical Information System was the data source for the MAX files.
Medicaid and CHIP Payment and Access Commission (MACPAC). “An Updated Look at Rates of Churn and Continuous Coverage in Medicaid and CHIP.” Issue Brief Advising Congress on Medicaid and CHIP Policy. October 2021. Available at https://www.macpac.gov/wp-content/uploads/2021/10/An-Updated-Look-at-Rates-of-Churn-and-Continuous-Coverage-in-Medicaid-and-CHIP.pdf . Accessed February 17, 2022.
In 2010, the ACA (P.L. 111-148 as amended) created a federal standard that limits the frequency of Medicaid and CHIP renewals to no more than once every 12 months for beneficiaries whose income is determined based on modified adjusted gross income (MAGI) (i.e. those under age 65 who are not eligible for Medicaid on the basis of a disability).
In the TAF DE, all enrollment records with overlapping dates are merged, with the exception of records with nonmatching enrollment types. Enrollment type indicates whether a beneficiary is enrolled in Medicaid or CHIP. Overlapping enrollment spells that are not merged indicate that a beneficiary is enrolled in both Medicaid and CHIP for the duration of the overlap.
An individual cannot be enrolled in Medicaid and S-CHIP at the same time, with the exception of a mother and her unborn or deemed newborn.
The Medicaid Analytic eXtract (MAX) person summary file, the predecessor of the TAF, provided monthly enrollment flags for whether a beneficiary was enrolled in Medicaid or CHIP on any day in a given month. The Medicaid Statistical Information System was the data source for the MAX files.
Medicaid and CHIP Payment and Access Commission (MACPAC). \u201cAn Updated Look at Rates of Churn and Continuous Coverage in Medicaid and CHIP.\u201d Issue Brief Advising Congress on Medicaid and CHIP Policy. October 2021. Available at https://www.macpac.gov/wp-content/uploads/2021/10/An-Updated-Look-at-Rates-of-Churn-and-Continuous-Coverage-in-Medicaid-and-CHIP.pdf . Accessed February 17, 2022.
In 2010, the ACA (P.L. 111-148 as amended) created a federal standard that limits the frequency of Medicaid and CHIP renewals to no more than once every 12 months for beneficiaries whose income is determined based on modified adjusted gross income (MAGI) (i.e. those under age 65 who are not eligible for Medicaid on the basis of a disability).
In the TAF DE, all enrollment records with overlapping dates are merged, with the exception of records with nonmatching enrollment types. Enrollment type indicates whether a beneficiary is enrolled in Medicaid or CHIP. Overlapping enrollment spells that are not merged indicate that a beneficiary is enrolled in both Medicaid and CHIP for the duration of the overlap.
An individual cannot be enrolled in Medicaid and S-CHIP at the same time, with the exception of a mother and her unborn or deemed newborn.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
Using the TAF DE [6] , we examined enrollment start dates (ENRLMT_EFCTV_CY_DT) and end dates (ENRLMT_END_CY_DT), which capture each enrollment episode for the beneficiary in the year. We used the start and end dates to examine three different types of enrollment patterns: (1) number of enrollment spans, (2) length of enrollment gaps, and (3) overlapping enrollment spans.
We calculated the share of beneficiaries with one deduplicated enrollment span and the share of those with three or more deduplicated enrollment spans during the year. [7] Table 1 shows how these measures were used to assess data quality. To be categorized into the low concern category, both conditions must be met. In contrast, only one of the conditions needs to be met to be categorized as medium concern or having unusable data.
Table 1. Criteria for DQ assessment of the number of enrollment spans
| Percentage of beneficiaries with only one enrollment span | Percentage of beneficiaries with three or more enrollment spans | DQ assessment |
|---|---|---|
| 80 percent ≤ x ≤ 98 percent | x < 1 percent | Low concern |
| 98 percent < x < 99.5 percent or 0% < x < 80% | 1 percent ≤ x ≤ 3 percent | Medium concern |
| x > 99.5 percent or no beneficiaries | x > 3 percent | Unusable |
Length of enrollment gaps
Using the same method to deduplicate enrollment spans as in the first analysis, we calculated the share of beneficiaries with gaps in enrollment of at least one day [8] and the average number of days between subsequent enrollment spans. We classified states into the low concern category if the average length of the enrollment gap was at least 31 days among beneficiaries with at least two spans of enrollment (Table 2). Since a break in enrollment of less than a month is more likely to reflect a data quality issue than true disenrollment and re-enrollment, we classified any state for which the average length of the enrollment gap was between 1 and 31 days into the medium data quality concern category. States where no beneficiaries had two or more enrollment spans, and as a result the average length of an enrollment gap could not be calculated, were also classified into the medium concern category.
Table 2. Criteria for DQ assessment of the average length of enrollment gaps
| Average length of enrollment gap | Percentage of beneficiaries with an enrollment gap | DQ assessment |
|---|---|---|
| x ≥ 31 days | x > 0 percent | Low concern |
| 0 < x < 31 days | x = 0 percent | Medium concern |
Overlapping enrollment spans
We calculated the share of beneficiaries with any overlapping Medicaid and CHIP enrollment spans in the year, defined as enrollment records for the same beneficiary where the start date for one enrollment span began before or was the same as the end date for the previous span, and one record indicated the beneficiary was in Medicaid and the other record indicated CHIP. We did not deduplicate enrollment spans that had the same enrollment start and end dates before this calculation. We also examined the number of days of overlap between records. Table 3 shows the level of concern for overlapping enrollment spans based on the percentage of beneficiaries with overlapping segment spans.
Table 3. Criteria for DQ assessment of overlapping enrollment spans
| Percentage of beneficiaries with overlapping enrollment spans | DQ assessment |
|---|---|
| x ≤ 1 percent | Low concern |
| 1 percent < x ≤ 5 percent | Medium concern |
| x > 5 percent | High concern |
Methods previously used to assess data quality
Table 4 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 4. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Before we counted enrollment spans, we deduplicated enrollment spans that had the same MSIS ID, enrollment start date, and enrollment end date. This deduplication approach considers completely overlapping enrollment spans as a single span and partially overlapping enrollment spans (those that overlap but do not match on both enrollment start and end dates) as two spans.
A beneficiary with more than one enrollment span but zero days between spans is not considered to have an enrollment gap, but rather an overlapping enrollment span.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Before we counted enrollment spans, we deduplicated enrollment spans that had the same MSIS ID, enrollment start date, and enrollment end date. This deduplication approach considers completely overlapping enrollment spans as a single span and partially overlapping enrollment spans (those that overlap but do not match on both enrollment start and end dates) as two spans.
A beneficiary with more than one enrollment span but zero days between spans is not considered to have an enrollment gap, but rather an overlapping enrollment span.
Effective eligibility dates can be used to understand the detailed enrollment patterns of Medicaid and CHIP beneficiaries. Among the beneficiaries who disenroll and later re-enroll during the same calendar year, we would expect the enrollment gap to be at least 30 days. This analysis examines the average length of the enrollment gap among beneficiaries who disenroll and re-enroll to identify states that may have a data quality problem with effective eligibility dates.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4091"", ""relatedTopics"": [{""measureId"": 10, ""measureName"": ""Number of Enrollment Spans"", ""groupId"": 2, ""groupName"": ""Enrollment Patterns Over Time"", ""order"": 0}, {""measureId"": 12, ""measureName"": ""Overlapping Enrollment Spans"", ""groupId"": 2, ""groupName"": ""Enrollment Patterns Over Time"", ""order"": 2}]}" 12,"{""measureId"": 12, ""measureName"": ""Overlapping Enrollment Spans"", ""groupId"": 2, ""groupName"": ""Enrollment Patterns Over Time"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Overlap-Enroll-Spans.pdf"", ""background"": {""content"": ""Understanding Medicaid and CHIP (Children’s Health Insurance Program) enrollment patterns is critical for assessing continuity of Medicaid and CHIP coverage and beneficiary churn in and out of these programs. T-MSIS, and its research-ready counterpart, the T-MSIS Analytic Files (TAF), contain enrollment records with specific start and end dates for each beneficiary, an enhancement to the prior national Medicaid data system and research files. [1] As long as the dates the states provide are accurate, TAF users can use those dates to examine detailed enrollment patterns.
We used the effective eligibility dates available in the annual Demographic and Eligibility (DE) file to identify problematic enrollment patterns indicative of data quality issues. First, we examined beneficiaries’ enrollment spans, which is a single enrollment episode with a start date and an end date.
Medicaid and CHIP beneficiaries can disenroll and subsequently re-enroll in these programs within a short period of time (a phenomenon often referred to as churn), and these enrollment patterns vary considerably both across and within states due to variation in eligibility requirements and enrollment redeterminations. For example, states with ex parte redeterminations are likely to experience lower rates of churn. However, recent studies found that overall, only a small proportion of beneficiaries are enrolled for fewer than 12 months, and most individuals who disenroll do not end up reenrolling in the program within 12 months. [2] Based on these findings and policy changes as part of the Patient Protection and Affordable Care Act (ACA), [3] we expect the majority of beneficiaries to have continuous enrollment for the full 12 months of a given year. It is therefore uncommon for a beneficiary to have more than two separate enrollment spans in a single year, so we would expect to see only a small percentage of beneficiaries with three or more enrollment spans.
We also examined the gaps between enrollment spans among those with more than one span during the year. In many states, once an individual is determined eligible, coverage begins on the first day of the month of the application and generally ends on the last day of the month. Therefore, gaps of less than a month may represent administrative errors or other data quality issues rather than true disenrollment and reenrollment in Medicaid or CHIP in these states.
In addition, we examined beneficiaries with enrollment spans that suggest simultaneous enrollment in Medicaid and S-CHIP (overlapping enrollment spans). The building of the TAF DE includes business rules that fix some known data quality issues; to the extent possible, all overlapping and contiguous enrollment spans reported in the T-MSIS data are merged and consolidated. [4] As a result, the TAF enrollment dates are much cleaner than those submitted by the states through T-MSIS. However, there are still beneficiaries with overlapping enrollment spans, representing cases in which the enrollment dates on different state-submitted eligibility records erroneously suggest that the beneficiary is enrolled in Medicaid and S-CHIP simultaneously. [5] For instances in which a beneficiary has overlapping enrollment spans, there are likely data quality issues with the enrollment dates and the enrollment type code, which indicates whether someone is enrolled in Medicaid or S-CHIP. When a beneficiary has overlapping enrollment spans, TAF users must conduct additional analyses to determine in which program the beneficiaries were enrolled.
The Medicaid Analytic eXtract (MAX) person summary file, the predecessor of the TAF, provided monthly enrollment flags for whether a beneficiary was enrolled in Medicaid or CHIP on any day in a given month. The Medicaid Statistical Information System was the data source for the MAX files.
Medicaid and CHIP Payment and Access Commission (MACPAC). “An Updated Look at Rates of Churn and Continuous Coverage in Medicaid and CHIP.” Issue Brief Advising Congress on Medicaid and CHIP Policy. October 2021. Available at https://www.macpac.gov/wp-content/uploads/2021/10/An-Updated-Look-at-Rates-of-Churn-and-Continuous-Coverage-in-Medicaid-and-CHIP.pdf . Accessed February 17, 2022.
In 2010, the ACA (P.L. 111-148 as amended) created a federal standard that limits the frequency of Medicaid and CHIP renewals to no more than once every 12 months for beneficiaries whose income is determined based on modified adjusted gross income (MAGI) (i.e. those under age 65 who are not eligible for Medicaid on the basis of a disability).
In the TAF DE, all enrollment records with overlapping dates are merged, with the exception of records with nonmatching enrollment types. Enrollment type indicates whether a beneficiary is enrolled in Medicaid or CHIP. Overlapping enrollment spells that are not merged indicate that a beneficiary is enrolled in both Medicaid and CHIP for the duration of the overlap.
An individual cannot be enrolled in Medicaid and S-CHIP at the same time, with the exception of a mother and her unborn or deemed newborn.
The Medicaid Analytic eXtract (MAX) person summary file, the predecessor of the TAF, provided monthly enrollment flags for whether a beneficiary was enrolled in Medicaid or CHIP on any day in a given month. The Medicaid Statistical Information System was the data source for the MAX files.
Medicaid and CHIP Payment and Access Commission (MACPAC). \u201cAn Updated Look at Rates of Churn and Continuous Coverage in Medicaid and CHIP.\u201d Issue Brief Advising Congress on Medicaid and CHIP Policy. October 2021. Available at https://www.macpac.gov/wp-content/uploads/2021/10/An-Updated-Look-at-Rates-of-Churn-and-Continuous-Coverage-in-Medicaid-and-CHIP.pdf . Accessed February 17, 2022.
In 2010, the ACA (P.L. 111-148 as amended) created a federal standard that limits the frequency of Medicaid and CHIP renewals to no more than once every 12 months for beneficiaries whose income is determined based on modified adjusted gross income (MAGI) (i.e. those under age 65 who are not eligible for Medicaid on the basis of a disability).
In the TAF DE, all enrollment records with overlapping dates are merged, with the exception of records with nonmatching enrollment types. Enrollment type indicates whether a beneficiary is enrolled in Medicaid or CHIP. Overlapping enrollment spells that are not merged indicate that a beneficiary is enrolled in both Medicaid and CHIP for the duration of the overlap.
An individual cannot be enrolled in Medicaid and S-CHIP at the same time, with the exception of a mother and her unborn or deemed newborn.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
Using the TAF DE [6] , we examined enrollment start dates (ENRLMT_EFCTV_CY_DT) and end dates (ENRLMT_END_CY_DT), which capture each enrollment episode for the beneficiary in the year. We used the start and end dates to examine three different types of enrollment patterns: (1) number of enrollment spans, (2) length of enrollment gaps, and (3) overlapping enrollment spans.
We calculated the share of beneficiaries with one deduplicated enrollment span and the share of those with three or more deduplicated enrollment spans during the year. [7] Table 1 shows how these measures were used to assess data quality. To be categorized into the low concern category, both conditions must be met. In contrast, only one of the conditions needs to be met to be categorized as medium concern or having unusable data.
Table 1. Criteria for DQ assessment of the number of enrollment spans
| Percentage of beneficiaries with only one enrollment span | Percentage of beneficiaries with three or more enrollment spans | DQ assessment |
|---|---|---|
| 80 percent ≤ x ≤ 98 percent | x < 1 percent | Low concern |
| 98 percent < x < 99.5 percent or 0% < x < 80% | 1 percent ≤ x ≤ 3 percent | Medium concern |
| x > 99.5 percent or no beneficiaries | x > 3 percent | Unusable |
Length of enrollment gaps
Using the same method to deduplicate enrollment spans as in the first analysis, we calculated the share of beneficiaries with gaps in enrollment of at least one day [8] and the average number of days between subsequent enrollment spans. We classified states into the low concern category if the average length of the enrollment gap was at least 31 days among beneficiaries with at least two spans of enrollment (Table 2). Since a break in enrollment of less than a month is more likely to reflect a data quality issue than true disenrollment and re-enrollment, we classified any state for which the average length of the enrollment gap was between 1 and 31 days into the medium data quality concern category. States where no beneficiaries had two or more enrollment spans, and as a result the average length of an enrollment gap could not be calculated, were also classified into the medium concern category.
Table 2. Criteria for DQ assessment of the average length of enrollment gaps
| Average length of enrollment gap | Percentage of beneficiaries with an enrollment gap | DQ assessment |
|---|---|---|
| x ≥ 31 days | x > 0 percent | Low concern |
| 0 < x < 31 days | x = 0 percent | Medium concern |
Overlapping enrollment spans
We calculated the share of beneficiaries with any overlapping Medicaid and CHIP enrollment spans in the year, defined as enrollment records for the same beneficiary where the start date for one enrollment span began before or was the same as the end date for the previous span, and one record indicated the beneficiary was in Medicaid and the other record indicated CHIP. We did not deduplicate enrollment spans that had the same enrollment start and end dates before this calculation. We also examined the number of days of overlap between records. Table 3 shows the level of concern for overlapping enrollment spans based on the percentage of beneficiaries with overlapping segment spans.
Table 3. Criteria for DQ assessment of overlapping enrollment spans
| Percentage of beneficiaries with overlapping enrollment spans | DQ assessment |
|---|---|
| x ≤ 1 percent | Low concern |
| 1 percent < x ≤ 5 percent | Medium concern |
| x > 5 percent | High concern |
Methods previously used to assess data quality
Table 4 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 4. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Before we counted enrollment spans, we deduplicated enrollment spans that had the same MSIS ID, enrollment start date, and enrollment end date. This deduplication approach considers completely overlapping enrollment spans as a single span and partially overlapping enrollment spans (those that overlap but do not match on both enrollment start and end dates) as two spans.
A beneficiary with more than one enrollment span but zero days between spans is not considered to have an enrollment gap, but rather an overlapping enrollment span.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Before we counted enrollment spans, we deduplicated enrollment spans that had the same MSIS ID, enrollment start date, and enrollment end date. This deduplication approach considers completely overlapping enrollment spans as a single span and partially overlapping enrollment spans (those that overlap but do not match on both enrollment start and end dates) as two spans.
A beneficiary with more than one enrollment span but zero days between spans is not considered to have an enrollment gap, but rather an overlapping enrollment span.
The TAF eligibility files include information on the program in which a beneficiary is enrolled for each of their enrollment spans. Program enrollment can be Medicaid, including M-CHIP, or S-CHIP. In some eligibility records, enrollment spans overlap, which suggests that a beneficiary was enrolled in both programs at the same time. This analysis examines the extent to which eligibility records have overlapping enrollment spans in order to identify states that may have a data quality problem with effective eligibility dates or with other data elements related to eligibility.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4091"", ""relatedTopics"": [{""measureId"": 10, ""measureName"": ""Number of Enrollment Spans"", ""groupId"": 2, ""groupName"": ""Enrollment Patterns Over Time"", ""order"": 0}, {""measureId"": 11, ""measureName"": ""Length of Enrollment Gaps"", ""groupId"": 2, ""groupName"": ""Enrollment Patterns Over Time"", ""order"": 1}]}" 13,"{""measureId"": 13, ""measureName"": ""Age"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Age.pdf"", ""background"": {""content"": ""The T\u2011MSIS Analytic Files (TAF) are an enhanced set of data on beneficiaries in Medicaid and the Children’s Health Insurance Program (CHIP). These data include select demographic characteristics, which are critical to understanding those whom the programs serve. This data quality assessment examines the completeness and face validity of selected demographic variables in the TAF annual Demographic and Eligibility (DE) file. We focused on four key demographic variables in the DE file that are important for analytic purposes: age group, gender, income, and ZIP code (Table 1). This data quality assessment tabulates the proportion of records in the DE file that have missing information for each of these data elements.
Table 1. Valid values for age group, gender, income, and zip code variables
| Variable | Description | Valid values |
|---|---|---|
| Age group | Beneficiary’s age as of the last month of enrollment in the calendar year, or as of the date of death if within the calendar year, grouped into 10 categories | 1 = Age <1 |
| Gender | Beneficiary’s biological sex | M = Male |
| Income | Family’s income level, grouped into 8 categories | 01 = Individual’s State-defined family income is from 0 to 100% of the FPL |
| Zip code | A beneficiary’s location of residence, or if residence ZIP code is unavailable, the ZIP code corresponding to the beneficiary’s mailing address. | All values other than missing values or any 0-filled, 8-filled, or 9-filled values. |
We used the DE file to calculate the percentage of non-dummy [1] enrollment records that have complete data, which we define as valid, non-missing values. [2] For the age group, gender, and income variables, all non-missing values represent valid values because in the creation of TAF, all invalid values for categorical variables are recoded to null. We therefore considered non-missing values to represent complete data for these variables. [3] We also examined the ZIP code of residence on the enrollment record. For ZIP code, we considered complete data to include all values other than missing values or any 0-filled, 8-filled, or 9-filled values, which also represent missing data. [4]
We assessed data quality for each of these data elements based on the percentage of enrollment records with missing data (Table 2).
Table 2. Criteria for DQ assessment of key demographic variables
| DQ assessment | |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
We also tabulated the distribution of valid values for age group and gender to allow users to check for face validity. Table 3 provides information about how the age group data element was combined into three age categories.
Table 3. Collapsing of age groups into categories
| Age group value in TAF | Description | Assigned age category |
|---|---|---|
| 1 | Age <1 | Age 0–18 |
| 2 | Age 1–5 | Age 0–18 |
| 3 | Age 6–14 | Age 0–18 |
| 4 | Age 15–18 | Age 0–18 |
| 5 | Age 19–20 | Age 19–64 |
| 6 | Age 21–44 | Age 19–64 |
| 7 | Age 45–64 | Age 19–64 |
| 8 | Age 65–74 | Age 65+ |
| 9 | Age 75–84 | Age 65+ |
| 10 | Age 85–125 | Age 65+ |
The distribution of age group and gender is presented by state and Medicaid expansion status of the state. In states that opted to expand Medicaid, the expansion population is relatively large, and the age and gender distribution systematically differs from the traditional Medicaid population. We did not stratify based on other optional coverage groups because the populations are not as large and do not have the same impact on the age and gender distribution for a state’s Medicaid and CHIP population.
We did not analyze income data beyond missing rates because there are no available benchmarks by which to compare the findings in the TAF. Furthermore, because Medicaid policies regarding income eligibility vary by state, checking for consistency across states is not feasible.
As part of the creation of the TAF RIF, dummy records are added to the DE file that represent beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state. These dummy records can be identified using the data element MISG_ELGBLTY_DATA_IND.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Age group is equal to null if the source value from T-MSIS for that variable is missing, unknown, or not on the valid value list or within the range of valid values; gender is equal to null if the source value is missing, unknown, not on the valid value list or within the range of valid values, or equal to U; income is equal to null if the source value is missing, unknown, not on the valid value list or within the range of valid values, or equal to 88 or 99.
ZIP code is equal to null if the source value for residence ZIP code and the mailing address ZIP code are missing, unknown, or invalid. Typically, 0-filled, 8-filled, and 9-filled values indicate missing or unknown values.
As part of the creation of the TAF RIF, dummy records are added to the DE file that represent beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state. These dummy records can be identified using the data element MISG_ELGBLTY_DATA_IND.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Age group is equal to null if the source value from T-MSIS for that variable is missing, unknown, or not on the valid value list or within the range of valid values; gender is equal to null if the source value is missing, unknown, not on the valid value list or within the range of valid values, or equal to U; income is equal to null if the source value is missing, unknown, not on the valid value list or within the range of valid values, or equal to 88 or 99.
ZIP code is equal to null if the source value for residence ZIP code and the mailing address ZIP code are missing, unknown, or invalid. Typically, 0-filled, 8-filled, and 9-filled values indicate missing or unknown values.
The TAF eligibility files include information on select demographic characteristics of beneficiaries enrolled in Medicaid or CHIP. This analysis examines the completeness and distribution of beneficiary age information in the TAF.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4121"", ""relatedTopics"": [{""measureId"": 14, ""measureName"": ""Gender"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""order"": 1}, {""measureId"": 17, ""measureName"": ""Income"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""order"": 2}, {""measureId"": 15, ""measureName"": ""ZIP Code"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""order"": 3}]}" 14,"{""measureId"": 14, ""measureName"": ""Gender"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Gender.pdf"", ""background"": {""content"": ""The T\u2011MSIS Analytic Files (TAF) are an enhanced set of data on beneficiaries in Medicaid and the Children’s Health Insurance Program (CHIP). These data include select demographic characteristics, which are critical to understanding those whom the programs serve. This data quality assessment examines the completeness and face validity of selected demographic variables in the TAF annual Demographic and Eligibility (DE) file. We focused on four key demographic variables in the DE file that are important for analytic purposes: age group, gender, income, and ZIP code (Table 1). This data quality assessment tabulates the proportion of records in the DE file that have missing information for each of these data elements.
Table 1. Valid values for age group, gender, income, and zip code variables
| Variable | Description | Valid values |
|---|---|---|
| Age group | Beneficiary’s age as of the last month of enrollment in the calendar year, or as of the date of death if within the calendar year, grouped into 10 categories | 1 = Age <1 |
| Gender | Beneficiary’s biological sex | M = Male |
| Income | Family’s income level, grouped into 8 categories | 01 = Individual’s State-defined family income is from 0 to 100% of the FPL |
| Zip code | A beneficiary’s location of residence, or if residence ZIP code is unavailable, the ZIP code corresponding to the beneficiary’s mailing address. | All values other than missing values or any 0-filled, 8-filled, or 9-filled values. |
We used the DE file to calculate the percentage of non-dummy [1] enrollment records that have complete data, which we define as valid, non-missing values. [2] For the age group, gender, and income variables, all non-missing values represent valid values because in the creation of TAF, all invalid values for categorical variables are recoded to null. We therefore considered non-missing values to represent complete data for these variables. [3] We also examined the ZIP code of residence on the enrollment record. For ZIP code, we considered complete data to include all values other than missing values or any 0-filled, 8-filled, or 9-filled values, which also represent missing data. [4]
We assessed data quality for each of these data elements based on the percentage of enrollment records with missing data (Table 2).
Table 2. Criteria for DQ assessment of key demographic variables
| DQ assessment | |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
We also tabulated the distribution of valid values for age group and gender to allow users to check for face validity. Table 3 provides information about how the age group data element was combined into three age categories.
Table 3. Collapsing of age groups into categories
| Age group value in TAF | Description | Assigned age category |
|---|---|---|
| 1 | Age <1 | Age 0–18 |
| 2 | Age 1–5 | Age 0–18 |
| 3 | Age 6–14 | Age 0–18 |
| 4 | Age 15–18 | Age 0–18 |
| 5 | Age 19–20 | Age 19–64 |
| 6 | Age 21–44 | Age 19–64 |
| 7 | Age 45–64 | Age 19–64 |
| 8 | Age 65–74 | Age 65+ |
| 9 | Age 75–84 | Age 65+ |
| 10 | Age 85–125 | Age 65+ |
The distribution of age group and gender is presented by state and Medicaid expansion status of the state. In states that opted to expand Medicaid, the expansion population is relatively large, and the age and gender distribution systematically differs from the traditional Medicaid population. We did not stratify based on other optional coverage groups because the populations are not as large and do not have the same impact on the age and gender distribution for a state’s Medicaid and CHIP population.
We did not analyze income data beyond missing rates because there are no available benchmarks by which to compare the findings in the TAF. Furthermore, because Medicaid policies regarding income eligibility vary by state, checking for consistency across states is not feasible.
As part of the creation of the TAF RIF, dummy records are added to the DE file that represent beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state. These dummy records can be identified using the data element MISG_ELGBLTY_DATA_IND.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Age group is equal to null if the source value from T-MSIS for that variable is missing, unknown, or not on the valid value list or within the range of valid values; gender is equal to null if the source value is missing, unknown, not on the valid value list or within the range of valid values, or equal to U; income is equal to null if the source value is missing, unknown, not on the valid value list or within the range of valid values, or equal to 88 or 99.
ZIP code is equal to null if the source value for residence ZIP code and the mailing address ZIP code are missing, unknown, or invalid. Typically, 0-filled, 8-filled, and 9-filled values indicate missing or unknown values.
As part of the creation of the TAF RIF, dummy records are added to the DE file that represent beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state. These dummy records can be identified using the data element MISG_ELGBLTY_DATA_IND.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Age group is equal to null if the source value from T-MSIS for that variable is missing, unknown, or not on the valid value list or within the range of valid values; gender is equal to null if the source value is missing, unknown, not on the valid value list or within the range of valid values, or equal to U; income is equal to null if the source value is missing, unknown, not on the valid value list or within the range of valid values, or equal to 88 or 99.
ZIP code is equal to null if the source value for residence ZIP code and the mailing address ZIP code are missing, unknown, or invalid. Typically, 0-filled, 8-filled, and 9-filled values indicate missing or unknown values.
The TAF eligibility files include information on select demographic characteristics of beneficiaries enrolled in Medicaid or CHIP. This analysis examines the completeness and distribution of information on beneficiary gender in the TAF.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4121"", ""relatedTopics"": [{""measureId"": 13, ""measureName"": ""Age"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""order"": 0}, {""measureId"": 17, ""measureName"": ""Income"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""order"": 2}, {""measureId"": 15, ""measureName"": ""ZIP Code"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""order"": 3}]}" 15,"{""measureId"": 15, ""measureName"": ""ZIP Code"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-ZIP-Cd.pdf"", ""background"": {""content"": ""The T\u2011MSIS Analytic Files (TAF) are an enhanced set of data on beneficiaries in Medicaid and the Children’s Health Insurance Program (CHIP). These data include select demographic characteristics, which are critical to understanding those whom the programs serve. This data quality assessment examines the completeness and face validity of selected demographic variables in the TAF annual Demographic and Eligibility (DE) file. We focused on four key demographic variables in the DE file that are important for analytic purposes: age group, gender, income, and ZIP code (Table 1). This data quality assessment tabulates the proportion of records in the DE file that have missing information for each of these data elements.
Table 1. Valid values for age group, gender, income, and zip code variables
| Variable | Description | Valid values |
|---|---|---|
| Age group | Beneficiary’s age as of the last month of enrollment in the calendar year, or as of the date of death if within the calendar year, grouped into 10 categories | 1 = Age <1 |
| Gender | Beneficiary’s biological sex | M = Male |
| Income | Family’s income level, grouped into 8 categories | 01 = Individual’s State-defined family income is from 0 to 100% of the FPL |
| Zip code | A beneficiary’s location of residence, or if residence ZIP code is unavailable, the ZIP code corresponding to the beneficiary’s mailing address. | All values other than missing values or any 0-filled, 8-filled, or 9-filled values. |
We used the DE file to calculate the percentage of non-dummy [1] enrollment records that have complete data, which we define as valid, non-missing values. [2] For the age group, gender, and income variables, all non-missing values represent valid values because in the creation of TAF, all invalid values for categorical variables are recoded to null. We therefore considered non-missing values to represent complete data for these variables. [3] We also examined the ZIP code of residence on the enrollment record. For ZIP code, we considered complete data to include all values other than missing values or any 0-filled, 8-filled, or 9-filled values, which also represent missing data. [4]
We assessed data quality for each of these data elements based on the percentage of enrollment records with missing data (Table 2).
Table 2. Criteria for DQ assessment of key demographic variables
| DQ assessment | |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
We also tabulated the distribution of valid values for age group and gender to allow users to check for face validity. Table 3 provides information about how the age group data element was combined into three age categories.
Table 3. Collapsing of age groups into categories
| Age group value in TAF | Description | Assigned age category |
|---|---|---|
| 1 | Age <1 | Age 0–18 |
| 2 | Age 1–5 | Age 0–18 |
| 3 | Age 6–14 | Age 0–18 |
| 4 | Age 15–18 | Age 0–18 |
| 5 | Age 19–20 | Age 19–64 |
| 6 | Age 21–44 | Age 19–64 |
| 7 | Age 45–64 | Age 19–64 |
| 8 | Age 65–74 | Age 65+ |
| 9 | Age 75–84 | Age 65+ |
| 10 | Age 85–125 | Age 65+ |
The distribution of age group and gender is presented by state and Medicaid expansion status of the state. In states that opted to expand Medicaid, the expansion population is relatively large, and the age and gender distribution systematically differs from the traditional Medicaid population. We did not stratify based on other optional coverage groups because the populations are not as large and do not have the same impact on the age and gender distribution for a state’s Medicaid and CHIP population.
We did not analyze income data beyond missing rates because there are no available benchmarks by which to compare the findings in the TAF. Furthermore, because Medicaid policies regarding income eligibility vary by state, checking for consistency across states is not feasible.
As part of the creation of the TAF RIF, dummy records are added to the DE file that represent beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state. These dummy records can be identified using the data element MISG_ELGBLTY_DATA_IND.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Age group is equal to null if the source value from T-MSIS for that variable is missing, unknown, or not on the valid value list or within the range of valid values; gender is equal to null if the source value is missing, unknown, not on the valid value list or within the range of valid values, or equal to U; income is equal to null if the source value is missing, unknown, not on the valid value list or within the range of valid values, or equal to 88 or 99.
ZIP code is equal to null if the source value for residence ZIP code and the mailing address ZIP code are missing, unknown, or invalid. Typically, 0-filled, 8-filled, and 9-filled values indicate missing or unknown values.
As part of the creation of the TAF RIF, dummy records are added to the DE file that represent beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state. These dummy records can be identified using the data element MISG_ELGBLTY_DATA_IND.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Age group is equal to null if the source value from T-MSIS for that variable is missing, unknown, or not on the valid value list or within the range of valid values; gender is equal to null if the source value is missing, unknown, not on the valid value list or within the range of valid values, or equal to U; income is equal to null if the source value is missing, unknown, not on the valid value list or within the range of valid values, or equal to 88 or 99.
ZIP code is equal to null if the source value for residence ZIP code and the mailing address ZIP code are missing, unknown, or invalid. Typically, 0-filled, 8-filled, and 9-filled values indicate missing or unknown values.
The TAF eligibility files include information on select demographic characteristics of beneficiaries enrolled in Medicaid or CHIP. This analysis examines the completeness of information on beneficiary ZIP code of residence in TAF.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4121"", ""relatedTopics"": [{""measureId"": 13, ""measureName"": ""Age"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""order"": 0}, {""measureId"": 14, ""measureName"": ""Gender"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""order"": 1}, {""measureId"": 17, ""measureName"": ""Income"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""order"": 2}]}" 16,"{""measureId"": 16, ""measureName"": ""Race and Ethnicity"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Race-Ethnicity.pdf"", ""background"": {""content"": ""The T\u2011MSIS Analytic Files (TAF) are research-optimized data on beneficiaries enrolled in Medicaid and the Children’s Health Insurance Program (CHIP). The Annual Demographic and Eligibility (DE) file contains information on beneficiary demographic characteristics, including race and ethnicity.
States use two separate data elements to submit information on a beneficiary’s race and ethnicity for T-MSIS: the race code, which has 17 categories, and the ethnicity code, which has 7 (Table 1). When eligibility records are created in TAF, these two source data elements are combined into an expanded race/ethnicity code that takes on 20 valid values and a condensed race/ethnicity code with 7 valid values. The expanded race/ethnicity code retains the detail of the source T-MSIS race code for beneficiaries with non-Hispanic or unspecified ethnicity and classifies beneficiaries with Hispanic ethnicity of any race in a single valid value. The condensed race/ethnicity code collapses the 19 non-Hispanic race codes into 6 broader valid values and retains the single valid value for all beneficiaries with Hispanic ethnicity. Both the expanded and condensed race/ethnicity codes are set to null (missing) in TAF when the T-MSIS source variables are missing or unknown, or when the race code is missing and the ethnicity code is either missing or equal to zero (non-Hispanic).
Table 1. Construction of the race/ethnicity codes in TAF
| Data element | T-MSIS race | T-MSIS ethnicity | TAF expanded race/ethnicity | TAF condensed race/ethnicity |
|---|---|---|---|---|
| Step | 1 | 1 | 2 | 3 |
| Valid values | 001 = White 002 = Black or African American 003 = American Indian or Alaskan Native 004 = Asian Indian 005 = Chinese 006 = Filipino 007 = Japanese 008 = Korean 009 = Vietnamese 010 = Other Asian 011 = Asian Unknown 012 = Native Hawaiian 013 = Guamanian or Chamorro 014 = Samoan 015 = Other Pacific Islander 016 = Native Hawaiian or Other Pacific Islander Unknown | 0 = Not of Hispanic, Latino/a, or Spanish origin | 1 = White, non-Hispanic 2 = Black, non-Hispanic 3 = American Indian or Alaska Native, non-Hispanic 4 = Asian Indian, non-Hispanic 5 = Chinese, non-Hispanic 6 = Filipino, non-Hispanic 7 = Japanese, non-Hispanic 8 = Korean, non-Hispanic 9 = Vietnamese, non-Hispanic 10 = Other Asian, non-Hispanic 11 = Asian Unknown, non-Hispanic 12 = Multi-Asian, non-Hispanic 13 = Native Hawaiian, non-Hispanic 14 = Guamanian or Chamorro, non-Hispanic 15 = Samoan, non-Hispanic 16 = Other Pacific Islander, non-Hispanic 17 = Native Hawaiian or Other Pacific Islander Unknown, non-Hispanic 18 = Multi-Islander, non-Hispanic 19 = Multiracial, non-Hispanic 20 = Hispanic, any race | 1 = White, non-Hispanic 2 = Black, non-Hispanic 3 = Asian, non-Hispanic 4 = American Indian and Alaska Native, non-Hispanic 5 = Hawaiian/Pacific Islander 6 = Multiracial, non-Hispanic 7 = Hispanic, all races |
The race and ethnicity codes in the source T-MSIS data are monthly variables. During the construction of the annual TAF DE file, the race/ethnicity code is converted to an annual variable using the “last-best” method. The last-best method selects a variable’s value from the most recent month for which a non-missing valid value exists. If no valid value was available in any month, TAF uses the last-best value from a previous calendar year. Thus, the DE file contains one expanded and one condensed race/ethnicity value per beneficiary in a given year.
Although states are expected to report the information that they receive on both race and ethnicity in T-MSIS, some states may not submit complete information because the data were not collected or technical difficulties arose in reporting. States may not have complete data on race and ethnicity because most do not require beneficiaries to disclose this information on enrollment forms, following the guidance from the Office of Management and Budget that establishes self-identification as the preferred means of obtaining this information. [1] States also vary in how they ask beneficiaries to identify ethnicity on enrollment forms, which may result in incomplete or inconsistent information on ethnicity in TAF for a state. [2]
Missing race and ethnicity information may affect all eligibility records in the DE equally (that is, the missingness may be randomly distributed), or the missing information may be concentrated among certain racial and ethnic groups. As a result, examining the proportion of DE records with missing data is not sufficient for understanding how usable this data element is for analytic purposes. This data quality assessment measures both the level of missing data and the extent to which the distribution of non-missing data aligns with an external benchmark, the American Community Survey (ACS).
The OMB Standards for Maintaining, Collecting, and Presenting Federal Data on Race and Ethnicity were last revised in 1997. The 1997 revisions can be found at https://www.whitehouse.gov/wp-content/uploads/2017/11/Revisions-to-the-Standards-for-the-Classification-of-Federal-Data-on-Race-and-Ethnicity-October30-1997.pdf . See the Census Bureau website for a concise list of OMB race categories: https://www.census.gov/topics/population/race/about.html .
For example, some states provide beneficiaries with options to identify as “Hispanic/Latinx” or “Not Hispanic/Latinx” on enrollment forms, while other states list options for ethnicity without an explicit option to self-identify as “Not Hispanic/Latinx”. Therefore, it may be difficult in the latter scenario to determine whether the beneficiary is non-Hispanic or if their Hispanic status is not reported.
The OMB Standards for Maintaining, Collecting, and Presenting Federal Data on Race and Ethnicity were last revised in 1997. The 1997 revisions can be found at https://www.whitehouse.gov/wp-content/uploads/2017/11/Revisions-to-the-Standards-for-the-Classification-of-Federal-Data-on-Race-and-Ethnicity-October30-1997.pdf . See the Census Bureau website for a concise list of OMB race categories: https://www.census.gov/topics/population/race/about.html .
For example, some states provide beneficiaries with options to identify as \u201cHispanic/Latinx\u201d or \u201cNot Hispanic/Latinx\u201d on enrollment forms, while other states list options for ethnicity without an explicit option to self-identify as \u201cNot Hispanic/Latinx\u201d. Therefore, it may be difficult in the latter scenario to determine whether the beneficiary is non-Hispanic or if their Hispanic status is not reported.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
For this analysis, we examined the condensed race/ethnicity code (RACE_ETHNCTY_FLAG) on non-dummy enrollment records [3] in the TAF DE file. [4] We tabulated the proportion of records that fell into each of the six valid race/ethnicity categories, as well as the proportion with missing values. The TAF production process recodes all invalid combinations of race and ethnicity to a null value in TAF. [5]
To construct the benchmark, we used the ACS 5-year estimates [6] Public Use Microdata Sample (PUMS) [7] for a given year. The ACS data, which are collected annually from a nationally representative random sample of households, contains information on self-reported race, ethnicity, and health insurance coverage. After pulling the ACS microdata from PUMS, we selected all individuals who reported having “Medicaid, Medical Assistance, or any kind of government-assistance plan for those with low incomes or a disability” at the time of the survey. For individuals in that group, we calculated the percentage who were in the six race/ethnicity categories (Table 2).
ACS data is used by many stakeholders, including by federal and state government agencies for policy and program funding activities, and are considered a highly reliable source of demographic data. However, self-reporting of health insurance coverage for the ACS often results in an undercount of the number of Medicaid beneficiaries who appear in administrative data. Therefore, in this data quality assessment, we compare the percentage of Medicaid beneficiaries with non-missing race/ethnicity in each race/ethnicity group in TAF to the comparable distribution in the ACS, rather than comparing the count of individuals in each category.
Table 2. Race and ethnicity categories in TAF and ACS
| Race/ethnicity category | RACE_ETHNCTY_FLAG value in TAF a c | Combination of race and Hispanic variables in the ACS |
|---|---|---|
| White, non-Hispanic | 1 = White, non-Hispanic | White alone, non-Hispanic |
| Black, non-Hispanic | 2 = Black, non-Hispanic | Black or African American alone, non-Hispanic |
| Asian, Native Hawaiian and other Pacific Islander, non-Hispanic b | 3 = Asian, non-Hispanic 5 = Hawaiian/Pacific Islander | Asian alone, non-Hispanic Native Hawaiian and Other Pacific Islander alone, non-Hispanic |
| American Indian and Alaska Native, non-Hispanic | 4 = American Indian and Alaska Native (AIAN), non-Hispanic | American Indian alone, non-Hispanic Alaska Native alone, non-Hispanic American Indian and Alaska Native tribes specified; or American Indian or Alaska native, not specified and no other races, non-Hispanic |
| Multiracial, non-Hispanic | 6 = Multiracial, non-Hispanic | Two or more races, non-Hispanic |
| Hispanic, all races | 7 = Hispanic, all races | Hispanic, all races |
a For the race/ethnicity flag TAF, a “non-Hispanic” value may indicate that (1) the beneficiary is not Hispanic or (2) the beneficiary’s ethnicity or Hispanic status is not reported.
b While Asian and Hawaiian/Pacific Islander groups are combined into one race/ethnicity category for the data quality assessment, we also present data for Asian beneficiaries and Hawaiian/Pacific Islander beneficiaries separately in the table for context.
c Some states report records where the TAF race/ethnicity flag has a value of 8, which represents records for non-Hispanic beneficiaries whose race is not reported. For the purposes of this analysis, we consider these records to have missing race/ethnicity.
We used two criteria to assess each state’s race/ethnicity data: (1) the percentage of enrollment records with missing data and (2) how well the percentage of beneficiaries in each of the six race/ethnicity categories aligned with the ACS (Table 3).
Table 3. Criteria for DQ assessment of race/ethnicity code
Methods previously used to assess data quality
Table 4 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 4. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
We excluded DE records that represent beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state (those with MISG_ELGBLTY_DATA_IND = 1).
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
When both the race and ethnicity codes in the source T-MSIS data are missing, the race/ethnicity flag in TAF will be set to null. Additionally, if the ethnicity code is equal to zero (a valid value indicating non-Hispanic ethnicity) and the race code is missing in the source T-MSIS data, the race/ethnicity flag in TAF will be set to null. However, if the ethnicity code is missing and the race code is non-missing in T-MSIS, then the race/ethnicity code in TAF is set equal to the reported race code in T-MSIS. Additionally, when the race code is missing and the ethnicity code indicates the beneficiary is non-Hispanic the race/ethnicity flag is reported for some states with the value of 8. For this analysis, we consider this value as having a missing race/ethnicity.
ACS 5-year estimates are more reliable and complete than ACS 1-year estimates and the Current Population Survey, as it includes smaller geographic areas and has a larger sample size. For more details, see: https://www.census.gov/programs-surveys/acs/guidance/estimates.html and https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2677056/
We excluded DE records that represent beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state (those with MISG_ELGBLTY_DATA_IND = 1).
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
When both the race and ethnicity codes in the source T-MSIS data are missing, the race/ethnicity flag in TAF will be set to null. Additionally, if the ethnicity code is equal to zero (a valid value indicating non-Hispanic ethnicity) and the race code is missing in the source T-MSIS data, the race/ethnicity flag in TAF will be set to null. However, if the ethnicity code is missing and the race code is non-missing in T-MSIS, then the race/ethnicity code in TAF is set equal to the reported race code in T-MSIS. Additionally, when the race code is missing and the ethnicity code indicates the beneficiary is non-Hispanic the race/ethnicity flag is reported for some states with the value of 8. For this analysis, we consider this value as having a missing race/ethnicity.
ACS 5-year estimates are more reliable and complete than ACS 1-year estimates and the Current Population Survey, as it includes smaller geographic areas and has a larger sample size. For more details, see: https://www.census.gov/programs-surveys/acs/guidance/estimates.html and https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2677056/
The TAF eligibility files include information on select demographic characteristics of beneficiaries enrolled in Medicaid or CHIP. This analysis examines the completeness of race and ethnicity information in the TAF. This analysis also examines how well the TAF data on race and ethnicity align with an external benchmark, the U.S. Census Bureau's American Community Survey.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4171"", ""relatedTopics"": []}" 17,"{""measureId"": 17, ""measureName"": ""Income"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Income.pdf"", ""background"": {""content"": ""The T\u2011MSIS Analytic Files (TAF) are an enhanced set of data on beneficiaries in Medicaid and the Children’s Health Insurance Program (CHIP). These data include select demographic characteristics, which are critical to understanding those whom the programs serve. This data quality assessment examines the completeness and face validity of selected demographic variables in the TAF annual Demographic and Eligibility (DE) file. We focused on four key demographic variables in the DE file that are important for analytic purposes: age group, gender, income, and ZIP code (Table 1). This data quality assessment tabulates the proportion of records in the DE file that have missing information for each of these data elements.
Table 1. Valid values for age group, gender, income, and zip code variables
| Variable | Description | Valid values |
|---|---|---|
| Age group | Beneficiary’s age as of the last month of enrollment in the calendar year, or as of the date of death if within the calendar year, grouped into 10 categories | 1 = Age <1 |
| Gender | Beneficiary’s biological sex | M = Male |
| Income | Family’s income level, grouped into 8 categories | 01 = Individual’s State-defined family income is from 0 to 100% of the FPL |
| Zip code | A beneficiary’s location of residence, or if residence ZIP code is unavailable, the ZIP code corresponding to the beneficiary’s mailing address. | All values other than missing values or any 0-filled, 8-filled, or 9-filled values. |
We used the DE file to calculate the percentage of non-dummy [1] enrollment records that have complete data, which we define as valid, non-missing values. [2] For the age group, gender, and income variables, all non-missing values represent valid values because in the creation of TAF, all invalid values for categorical variables are recoded to null. We therefore considered non-missing values to represent complete data for these variables. [3] We also examined the ZIP code of residence on the enrollment record. For ZIP code, we considered complete data to include all values other than missing values or any 0-filled, 8-filled, or 9-filled values, which also represent missing data. [4]
We assessed data quality for each of these data elements based on the percentage of enrollment records with missing data (Table 2).
Table 2. Criteria for DQ assessment of key demographic variables
| DQ assessment | |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
We also tabulated the distribution of valid values for age group and gender to allow users to check for face validity. Table 3 provides information about how the age group data element was combined into three age categories.
Table 3. Collapsing of age groups into categories
| Age group value in TAF | Description | Assigned age category |
|---|---|---|
| 1 | Age <1 | Age 0–18 |
| 2 | Age 1–5 | Age 0–18 |
| 3 | Age 6–14 | Age 0–18 |
| 4 | Age 15–18 | Age 0–18 |
| 5 | Age 19–20 | Age 19–64 |
| 6 | Age 21–44 | Age 19–64 |
| 7 | Age 45–64 | Age 19–64 |
| 8 | Age 65–74 | Age 65+ |
| 9 | Age 75–84 | Age 65+ |
| 10 | Age 85–125 | Age 65+ |
The distribution of age group and gender is presented by state and Medicaid expansion status of the state. In states that opted to expand Medicaid, the expansion population is relatively large, and the age and gender distribution systematically differs from the traditional Medicaid population. We did not stratify based on other optional coverage groups because the populations are not as large and do not have the same impact on the age and gender distribution for a state’s Medicaid and CHIP population.
We did not analyze income data beyond missing rates because there are no available benchmarks by which to compare the findings in the TAF. Furthermore, because Medicaid policies regarding income eligibility vary by state, checking for consistency across states is not feasible.
As part of the creation of the TAF RIF, dummy records are added to the DE file that represent beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state. These dummy records can be identified using the data element MISG_ELGBLTY_DATA_IND.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Age group is equal to null if the source value from T-MSIS for that variable is missing, unknown, or not on the valid value list or within the range of valid values; gender is equal to null if the source value is missing, unknown, not on the valid value list or within the range of valid values, or equal to U; income is equal to null if the source value is missing, unknown, not on the valid value list or within the range of valid values, or equal to 88 or 99.
ZIP code is equal to null if the source value for residence ZIP code and the mailing address ZIP code are missing, unknown, or invalid. Typically, 0-filled, 8-filled, and 9-filled values indicate missing or unknown values.
As part of the creation of the TAF RIF, dummy records are added to the DE file that represent beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state. These dummy records can be identified using the data element MISG_ELGBLTY_DATA_IND.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Age group is equal to null if the source value from T-MSIS for that variable is missing, unknown, or not on the valid value list or within the range of valid values; gender is equal to null if the source value is missing, unknown, not on the valid value list or within the range of valid values, or equal to U; income is equal to null if the source value is missing, unknown, not on the valid value list or within the range of valid values, or equal to 88 or 99.
ZIP code is equal to null if the source value for residence ZIP code and the mailing address ZIP code are missing, unknown, or invalid. Typically, 0-filled, 8-filled, and 9-filled values indicate missing or unknown values.
The TAF eligibility files include information on select demographic characteristics of beneficiaries enrolled in Medicaid or CHIP. This analysis examines the completeness of information on household income in TAF.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4121"", ""relatedTopics"": [{""measureId"": 13, ""measureName"": ""Age"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""order"": 0}, {""measureId"": 14, ""measureName"": ""Gender"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""order"": 1}, {""measureId"": 15, ""measureName"": ""ZIP Code"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""order"": 3}]}" 18,"{""measureId"": 18, ""measureName"": ""Dual Eligibility Code"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Dual-Eligibility-Cd.pdf"", ""background"": {""content"": ""The T\u2011MSIS Analytic Files (TAF) contain research-ready data on beneficiaries in Medicaid and the Children’s Health Insurance Program (CHIP). These data include select program characteristics, including whether the beneficiary is enrolled in CHIP or in Medicare as a dually-eligible beneficiary, which are critical to understanding the individuals served by the programs.
Created as part of the Balanced Budget Act of 1997, CHIP provides health care coverage to otherwise uninsured children in low-income families whose income exceeds Medicaid income-eligibility thresholds. States may use CHIP funds to expand their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); create a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopt a combination of both approaches. The CHIP code in the annual Demographic and Eligibility (DE) file can be used to identify whether a beneficiary was enrolled in Medicaid, M-CHIP, or S-CHIP during each month of the year.
Dually eligible beneficiaries are Medicaid beneficiaries also enrolled in Medicare Part A (hospital insurance and other costs), Medicare Part B (medical insurance), and/or a Medicare Savings Program (MSP). [1] For these beneficiaries, Medicare is the primary payer for services that are jointly covered by both programs. Medicaid covers services that Medicare does not cover as well as Medicare premiums and cost-sharing. The dually eligible fall into two groups—“partial-benefit dual eligibility” and “full-benefit dual eligibility”—depending upon the level of Medicaid benefits for which they qualify. Beneficiaries with partial-benefit dual eligibility are entitled to have Medicaid pay for only some of the expenses they incur under Medicare. In addition to the benefits to which beneficiaries with partial-benefit dual eligibility are entitled, beneficiaries with full-benefit dual eligibility are entitled to Medicaid coverage for various health care services that Medicare does not cover, such as most types of long-term services and supports. The dual eligible status code in the DE file can be used to identify whether a beneficiary was enrolled in Medicare during each month, and if so, whether they qualified for partial-benefit or full-benefit dual eligibility.
This data quality assessment examines the completeness and face validity of the CHIP and dual status codes in the DE file. [2] Table 1 lists these variables, along with their data element name in the DE file and a brief description.
Table 1. Key program variables in the TAF DE
| Variable | TAF Data Element Name | Description |
|---|---|---|
| CHIP code | CHIP_CD_LTST | A code used to distinguish among Medicaid, Medicaid CHIP expansion, and separate CHIP populations. The “last best” version of this variable represents the most recent non-missing value for the beneficiary in the calendar year. |
| Dual status code | DUAL_ELGBL_CD_LTST | A code that indicates Medicare coverage for individuals entitled to either Part A and/or Part B benefits and eligible for some category of Medicaid benefits. The “last best” version of this variable represents the most recent non-missing value for the beneficiary in the calendar year. |
Note: \tThe latest CHIP code and dual status code in the DE file were each constructed using a “last-best” method for selecting values from the monthly TAF Beneficiary Summary Files (BSF). The “last-best” method selects a variable’s value from the most recent month in the BSF for which a non-missing value exists.
Medicare Shared Savings Programs cover costs such as Part A premiums and Part A and B deductibles, coinsurance, and copayments, depending upon the program.
Assessments of how well the CHIP code and dual status code match enrollment as captured in other data sources can be found in the DQ Atlas single topic displays for M-CHIP and S-CHIP Enrollment and Dually Enrolled in Medicare .
Medicare Shared Savings Programs cover costs such as Part A premiums and Part A and B deductibles, coinsurance, and copayments, depending upon the program.
Assessments of how well the CHIP code and dual status code match enrollment as captured in other data sources can be found in the DQ Atlas single topic displays for M-CHIP and S-CHIP Enrollment and Dually Enrolled in Medicare .
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
For this analysis, we used the TAF DE file to calculate the percentage of non-dummy [3] enrollment records with complete data. [4] For the CHIP code and dual status code variables, all non-missing values represent valid values because all invalid values for categorical variables are re-coded to null. We therefore considered the rate of non-missing values to represent the completeness of data for these variables.
We also examined the distribution of valid values for both codes to check for face validity. For the CHIP code variable, we examined whether the beneficiaries reported in each CHIP code group aligned with the type of CHIP program found in the state. We classified states into three CHIP program types and evaluated whether the state was using the expected CHIP code values for the program, as shown in Table 2.
Table 2. Expected use of CHIP code by program type
| CHIP program type | Description | Expected use of CHIP codes |
|---|---|---|
| M-CHIP | States that use CHIP funds to expand their Medicaid programs are referred to as Medicaid expansion or M-CHIP a | CHIP_CD = 2 only |
| S-CHIP | States that use CHIP funds to create a program separate from their existing Medicaid programs are referred to as separate CHIP or S-CHIP | CHIP_CD = 3 only b |
| Combination | States that cover some children under M-CHIP and some children under S-CHIP States that classify themselves as S-CHIP, but after the Affordable Care Act increased Medicaid income eligibility to 133% of the federal poverty level, some children were transitioned from S-CHIP to Medicaid and continue to qualify for Title XXI CHIP funding (these children should be reported as M-CHIP enrollees) | CHIP_CD = 2 and 3 b |
Note:\tA map of CHIP program types by state can be found at https://www.medicaid.gov/chip/downloads/chip-map.pdf .
a For the 2014-2016 TAF data only, we expect to see some beneficiaries in two M-CHIP states (District of Columbia and Vermont) coded as S-CHIP enrollees (CHIP_CD=3). Both of these states are 2101(f) states, in which the State Plan Amendment assures that through April 2016, separate CHIP coverage will be provided for children ineligible for Medicaid due to the elimination of income disregards under the Affordable Care Act. In 2017 and later years, we expect to see only M-CHIP children (CHIP_CD=2) in these states.
b CHIP_CD = 4 (individual was both Medicaid eligible and S-CHIP eligible during the same month) is not a valid value in later versions of the T-MSIS data dictionary. However, because at least two states are still using the code for a small number of beneficiaries, we allow it as a valid value for S-CHIP and combination programs in this analysis.
For the dual status code variable, we examined whether each state reported at least some beneficiaries in the non-dual, full dual, and partial dual categories. If we find no beneficiaries in any one of these groups in one of the fifty states or the District of Columbia, that would indicate a data quality concern. For Puerto Rico and other U.S. territories, we checked whether at least some beneficiaries were reported in the non-dual and full dual categories, as these areas may not have partial dual programs. We mapped dual code values into the categories of non-dual, full dual, and partial dual as shown in Table 3.
Table 3. Categorization of Medicare-Medicaid dual beneficiaries in the TAF
| DUAL_ELGBL_CD_LTST | Dual-eligibility groups | Category |
|---|---|---|
| 0 | Not a Medicare beneficiary (not a dually eligible beneficiary) | Non-dual |
| 1 | Qualified Medicare beneficiary (QMB) only | Partial dual |
| 2 | QMB plus | Full dual |
| 3 | Specified low-income Medicare beneficiaries (SLMB) only | Partial dual |
| 4 | SLMB plus | Full dual |
| 5 | Qualified disabled and working individual (QDWI) | Partial dual |
| 6 | Qualified individual (QI) | Partial dual |
| 8 | Other dual | Full dual |
| 9 | Eligible is entitled to Medicare—other (this code is to be used only with specific approval from Centers for Medicare & Medicaid Services [CMS]) | Other dual |
| 10 | Separate CHIP eligible is entitled to Medicare | Other dual |
Source: \tAdditional background information is available in “CMS Guidance: Reporting Expectations for Dual-Eligible Beneficiaries, Updated,” which is available in the T-MSIS coding blog https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=51064 and in the annual TAF DE Data Dictionary. DUAL_ELGBL_CD = 09 is classified as “other” dual because states use this code for participation in state-specific programs. Very few states report any beneficiaries in this group. DUAL_ELGBL_CD = 10 is classified as “other” dual because these beneficiaries are dually enrolled in separate CHIP and Medicare, and not in Medicaid and Medicare as is the case for all other dual eligibles.
For each variable, we assessed the quality of the CHIP and dual code information based on the percentage of records with missing values and any unexpected patterns in the data, using the criteria shown in Tables 4 and 5. For the CHIP code analysis, we considered whether the CHIP code values the state reported aligned with the type of CHIP program in the state. For the analysis of the dual status code, we considered whether states reported at least some beneficiaries into the expected categories of non-duals, full duals, and partial duals. After evaluating the extent of missing data and unexpected patterns, we categorized each state in the highest level of concern that applied.
Table 4. Criteria for DQ assessment of CHIP code
| Percentage of records with missing CHIP code values | Alignment of CHIP code values with type of CHIP program in the state (M-CHIP, S-CHIP, or combination) | DQ assessment |
|---|---|---|
| x ≤ 10 percent | Aligned | Low concern a |
| 10 percent < x ≤ 20 percent | Medium concern | |
| 20 percent < x ≤ 50 percent | Not aligned | High concern b |
| x > 50 percent | No beneficiaries reported with an M-CHIP, S-CHIP or combination CHIP code | Unusable b |
a Both criteria must be true for a state to receive the given DQ Assessment.
b One of the two criteria must be true for a state to receive the given DQ Assessment.
Table 5. Criteria for DQ assessment of dual status code
| Percentage of records with missing dual status code values | Number of beneficiaries reported in the expected categories of non-dual, partial dual, and full dual | DQ assessment |
|---|---|---|
| x ≤ 10 percent | At least one beneficiary reported into all expected categories | Low concern a |
| 10 percent < x ≤ 20 percent | Medium concern | |
| x ≤ 10 percent | No beneficiaries reported in at least one of the expected categories | Medium concern a |
| 20 percent < x ≤ 50 percent | High concern | |
| 10 percent < x ≤ 20 percent | No beneficiaries reported in at least one of the expected categories | High concern a |
| x > 50 percent | No beneficiaries reported in the partial or full dual categories | Unusable b |
a Both criteria must be true for a state to receive the given DQ Assessment.
b One of the two criteria must be true for a state to receive the given DQ Assessment.
Methods previously used to assess data quality
Table 6 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 6. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
As part of the creation of the TAF RIF, dummy records are added to the DE file that represent beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state. These dummy records can be identified using the data element MISG_ELGBLTY_DATA_IND.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
As part of the creation of the TAF RIF, dummy records are added to the DE file that represent beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state. These dummy records can be identified using the data element MISG_ELGBLTY_DATA_IND.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
The dual status code in the TAF indicates whether a beneficiary is enrolled in both Medicaid and Medicare, and if so, the level of Medicaid coverage to which the beneficiary is entitled. This analysis examines how often the dual status code is missing on individual eligibility records and whether the distribution of the dual status code within a state displays any unexpected patterns.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4131"", ""relatedTopics"": [{""measureId"": 19, ""measureName"": ""CHIP Code"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""order"": 1}]}" 19,"{""measureId"": 19, ""measureName"": ""CHIP Code"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-CHIP-Cd.pdf"", ""background"": {""content"": ""The T\u2011MSIS Analytic Files (TAF) contain research-ready data on beneficiaries in Medicaid and the Children’s Health Insurance Program (CHIP). These data include select program characteristics, including whether the beneficiary is enrolled in CHIP or in Medicare as a dually-eligible beneficiary, which are critical to understanding the individuals served by the programs.
Created as part of the Balanced Budget Act of 1997, CHIP provides health care coverage to otherwise uninsured children in low-income families whose income exceeds Medicaid income-eligibility thresholds. States may use CHIP funds to expand their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); create a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopt a combination of both approaches. The CHIP code in the annual Demographic and Eligibility (DE) file can be used to identify whether a beneficiary was enrolled in Medicaid, M-CHIP, or S-CHIP during each month of the year.
Dually eligible beneficiaries are Medicaid beneficiaries also enrolled in Medicare Part A (hospital insurance and other costs), Medicare Part B (medical insurance), and/or a Medicare Savings Program (MSP). [1] For these beneficiaries, Medicare is the primary payer for services that are jointly covered by both programs. Medicaid covers services that Medicare does not cover as well as Medicare premiums and cost-sharing. The dually eligible fall into two groups—“partial-benefit dual eligibility” and “full-benefit dual eligibility”—depending upon the level of Medicaid benefits for which they qualify. Beneficiaries with partial-benefit dual eligibility are entitled to have Medicaid pay for only some of the expenses they incur under Medicare. In addition to the benefits to which beneficiaries with partial-benefit dual eligibility are entitled, beneficiaries with full-benefit dual eligibility are entitled to Medicaid coverage for various health care services that Medicare does not cover, such as most types of long-term services and supports. The dual eligible status code in the DE file can be used to identify whether a beneficiary was enrolled in Medicare during each month, and if so, whether they qualified for partial-benefit or full-benefit dual eligibility.
This data quality assessment examines the completeness and face validity of the CHIP and dual status codes in the DE file. [2] Table 1 lists these variables, along with their data element name in the DE file and a brief description.
Table 1. Key program variables in the TAF DE
| Variable | TAF Data Element Name | Description |
|---|---|---|
| CHIP code | CHIP_CD_LTST | A code used to distinguish among Medicaid, Medicaid CHIP expansion, and separate CHIP populations. The “last best” version of this variable represents the most recent non-missing value for the beneficiary in the calendar year. |
| Dual status code | DUAL_ELGBL_CD_LTST | A code that indicates Medicare coverage for individuals entitled to either Part A and/or Part B benefits and eligible for some category of Medicaid benefits. The “last best” version of this variable represents the most recent non-missing value for the beneficiary in the calendar year. |
Note: \tThe latest CHIP code and dual status code in the DE file were each constructed using a “last-best” method for selecting values from the monthly TAF Beneficiary Summary Files (BSF). The “last-best” method selects a variable’s value from the most recent month in the BSF for which a non-missing value exists.
Medicare Shared Savings Programs cover costs such as Part A premiums and Part A and B deductibles, coinsurance, and copayments, depending upon the program.
Assessments of how well the CHIP code and dual status code match enrollment as captured in other data sources can be found in the DQ Atlas single topic displays for M-CHIP and S-CHIP Enrollment and Dually Enrolled in Medicare .
Medicare Shared Savings Programs cover costs such as Part A premiums and Part A and B deductibles, coinsurance, and copayments, depending upon the program.
Assessments of how well the CHIP code and dual status code match enrollment as captured in other data sources can be found in the DQ Atlas single topic displays for M-CHIP and S-CHIP Enrollment and Dually Enrolled in Medicare .
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
For this analysis, we used the TAF DE file to calculate the percentage of non-dummy [3] enrollment records with complete data. [4] For the CHIP code and dual status code variables, all non-missing values represent valid values because all invalid values for categorical variables are re-coded to null. We therefore considered the rate of non-missing values to represent the completeness of data for these variables.
We also examined the distribution of valid values for both codes to check for face validity. For the CHIP code variable, we examined whether the beneficiaries reported in each CHIP code group aligned with the type of CHIP program found in the state. We classified states into three CHIP program types and evaluated whether the state was using the expected CHIP code values for the program, as shown in Table 2.
Table 2. Expected use of CHIP code by program type
| CHIP program type | Description | Expected use of CHIP codes |
|---|---|---|
| M-CHIP | States that use CHIP funds to expand their Medicaid programs are referred to as Medicaid expansion or M-CHIP a | CHIP_CD = 2 only |
| S-CHIP | States that use CHIP funds to create a program separate from their existing Medicaid programs are referred to as separate CHIP or S-CHIP | CHIP_CD = 3 only b |
| Combination | States that cover some children under M-CHIP and some children under S-CHIP States that classify themselves as S-CHIP, but after the Affordable Care Act increased Medicaid income eligibility to 133% of the federal poverty level, some children were transitioned from S-CHIP to Medicaid and continue to qualify for Title XXI CHIP funding (these children should be reported as M-CHIP enrollees) | CHIP_CD = 2 and 3 b |
Note:\tA map of CHIP program types by state can be found at https://www.medicaid.gov/chip/downloads/chip-map.pdf .
a For the 2014-2016 TAF data only, we expect to see some beneficiaries in two M-CHIP states (District of Columbia and Vermont) coded as S-CHIP enrollees (CHIP_CD=3). Both of these states are 2101(f) states, in which the State Plan Amendment assures that through April 2016, separate CHIP coverage will be provided for children ineligible for Medicaid due to the elimination of income disregards under the Affordable Care Act. In 2017 and later years, we expect to see only M-CHIP children (CHIP_CD=2) in these states.
b CHIP_CD = 4 (individual was both Medicaid eligible and S-CHIP eligible during the same month) is not a valid value in later versions of the T-MSIS data dictionary. However, because at least two states are still using the code for a small number of beneficiaries, we allow it as a valid value for S-CHIP and combination programs in this analysis.
For the dual status code variable, we examined whether each state reported at least some beneficiaries in the non-dual, full dual, and partial dual categories. If we find no beneficiaries in any one of these groups in one of the fifty states or the District of Columbia, that would indicate a data quality concern. For Puerto Rico and other U.S. territories, we checked whether at least some beneficiaries were reported in the non-dual and full dual categories, as these areas may not have partial dual programs. We mapped dual code values into the categories of non-dual, full dual, and partial dual as shown in Table 3.
Table 3. Categorization of Medicare-Medicaid dual beneficiaries in the TAF
| DUAL_ELGBL_CD_LTST | Dual-eligibility groups | Category |
|---|---|---|
| 0 | Not a Medicare beneficiary (not a dually eligible beneficiary) | Non-dual |
| 1 | Qualified Medicare beneficiary (QMB) only | Partial dual |
| 2 | QMB plus | Full dual |
| 3 | Specified low-income Medicare beneficiaries (SLMB) only | Partial dual |
| 4 | SLMB plus | Full dual |
| 5 | Qualified disabled and working individual (QDWI) | Partial dual |
| 6 | Qualified individual (QI) | Partial dual |
| 8 | Other dual | Full dual |
| 9 | Eligible is entitled to Medicare—other (this code is to be used only with specific approval from Centers for Medicare & Medicaid Services [CMS]) | Other dual |
| 10 | Separate CHIP eligible is entitled to Medicare | Other dual |
Source: \tAdditional background information is available in “CMS Guidance: Reporting Expectations for Dual-Eligible Beneficiaries, Updated,” which is available in the T-MSIS coding blog https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=51064 and in the annual TAF DE Data Dictionary. DUAL_ELGBL_CD = 09 is classified as “other” dual because states use this code for participation in state-specific programs. Very few states report any beneficiaries in this group. DUAL_ELGBL_CD = 10 is classified as “other” dual because these beneficiaries are dually enrolled in separate CHIP and Medicare, and not in Medicaid and Medicare as is the case for all other dual eligibles.
For each variable, we assessed the quality of the CHIP and dual code information based on the percentage of records with missing values and any unexpected patterns in the data, using the criteria shown in Tables 4 and 5. For the CHIP code analysis, we considered whether the CHIP code values the state reported aligned with the type of CHIP program in the state. For the analysis of the dual status code, we considered whether states reported at least some beneficiaries into the expected categories of non-duals, full duals, and partial duals. After evaluating the extent of missing data and unexpected patterns, we categorized each state in the highest level of concern that applied.
Table 4. Criteria for DQ assessment of CHIP code
| Percentage of records with missing CHIP code values | Alignment of CHIP code values with type of CHIP program in the state (M-CHIP, S-CHIP, or combination) | DQ assessment |
|---|---|---|
| x ≤ 10 percent | Aligned | Low concern a |
| 10 percent < x ≤ 20 percent | Medium concern | |
| 20 percent < x ≤ 50 percent | Not aligned | High concern b |
| x > 50 percent | No beneficiaries reported with an M-CHIP, S-CHIP or combination CHIP code | Unusable b |
a Both criteria must be true for a state to receive the given DQ Assessment.
b One of the two criteria must be true for a state to receive the given DQ Assessment.
Table 5. Criteria for DQ assessment of dual status code
| Percentage of records with missing dual status code values | Number of beneficiaries reported in the expected categories of non-dual, partial dual, and full dual | DQ assessment |
|---|---|---|
| x ≤ 10 percent | At least one beneficiary reported into all expected categories | Low concern a |
| 10 percent < x ≤ 20 percent | Medium concern | |
| x ≤ 10 percent | No beneficiaries reported in at least one of the expected categories | Medium concern a |
| 20 percent < x ≤ 50 percent | High concern | |
| 10 percent < x ≤ 20 percent | No beneficiaries reported in at least one of the expected categories | High concern a |
| x > 50 percent | No beneficiaries reported in the partial or full dual categories | Unusable b |
a Both criteria must be true for a state to receive the given DQ Assessment.
b One of the two criteria must be true for a state to receive the given DQ Assessment.
Methods previously used to assess data quality
Table 6 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 6. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
As part of the creation of the TAF RIF, dummy records are added to the DE file that represent beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state. These dummy records can be identified using the data element MISG_ELGBLTY_DATA_IND.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
As part of the creation of the TAF RIF, dummy records are added to the DE file that represent beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state. These dummy records can be identified using the data element MISG_ELGBLTY_DATA_IND.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
The CHIP code in the TAF can be used to distinguish between the Medicaid-only, Medicaid-expansion CHIP, and separate CHIP populations. This analysis examines eligibility records to assess how often the CHIP code is missing or inconsistent with the CHIP program type operating in a state.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4131"", ""relatedTopics"": [{""measureId"": 18, ""measureName"": ""Dual Eligibility Code"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""order"": 0}]}" 20,"{""measureId"": 20, ""measureName"": ""Eligibility Group Code"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Eligibility-Group-Cd.pdf"", ""background"": {""content"": ""To be eligible for Medicaid, individuals typically must meet financial requirements or be a part of a group that is categorically eligible for coverage (for instance, individuals covered under the Supplemental Security Income [SSI] program). Federal law requires that all state Medicaid programs cover certain groups of individuals, known as mandatory eligibility groups, whereas coverage of other groups is optional. [1] The makeup of these groups has changed over time. Historically, coverage was more limited, focusing on low-income children and their parents, qualified pregnant women, the elderly, and people with disabilities. Several policies have expanded eligibility for both the mandatory and the optional groups. For example, the State Children’s Health Insurance Program (CHIP) under the Balanced Budget Act of 1997 gave states additional funding opportunities and greater flexibility in covering uninsured children in families whose income is too high to qualify for Medicaid. In 2010, the Patient Protection and Affordable Care Act expanded coverage to former foster care children and gave states the option to further expand Medicaid to other low-income adults. [2]
The eligibility group code in the T-MSIS Analytic Files (TAF), the research-ready version of T-MSIS, can be used to identify the basis on which an individual was deemed eligible for Medicaid or CHIP. [3] This data element defines 72 eligibility group values, including 26 mandatory eligibility groups that every state is required by law to cover and 46 optional eligibility groups that may or may not be relevant to beneficiaries in a given state, depending on the state’s Medicaid and CHIP state plan and waiver arrangements. [4] States should assign every Medicaid and CHIP beneficiary to one of the 72 eligibility groups. However, TAF users would not expect the code for every eligibility group to be represented in every state’s data because some of the mandatory eligibility groups represent small populations that are not present in every state, whereas other mandatory eligibility groups represent populations that are large enough to exist in every state. This data quality assessment examines (1) how often eligibility group codes were missing, which suggests that the data were incomplete; and (2) how often we fail to observe enrollment in the large mandatory eligibility groups, which suggests that a state may not be assigning eligibility group codes to enrollment records accurately.
For a full description of the eligibility groups, see the T-MSIS Data Dictionary Appendices, Version 2.4, Appendix F, p. 67, at https://www.medicaid.gov/medicaid/data-systems/downloads/tmsis-data-appendices.docx .
TAF-based counts of the adult expansion population can be found in the DQ Atlas single topic displays for Adult Expansion Enrollment and Newly Eligible Adult Enrollment .
Historically, states reported the basis of eligibility in the legacy MSIS in two fields populated with the Maintenance Assistance Status (MAS) and Basis of Eligibility (BOE) codes. These codes were combined in T-MSIS but are no longer required fields. Although MAS/BOE may continue to be reported, fewer states are reporting this data element over time. In place of MAS and BOE, CMS developed a new coding system for classifying eligibility, known as the Eligibility Group, which is the focus of this assessment.
See the T-MSIS Coding Blog on Reporting Eligibility Group at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=47569 .
For a full description of the eligibility groups, see the T-MSIS Data Dictionary Appendices, Version 2.4, Appendix F, p. 67, at https://www.medicaid.gov/medicaid/data-systems/downloads/tmsis-data-appendices.docx .
TAF-based counts of the adult expansion population can be found in the DQ Atlas single topic displays for Adult Expansion Enrollment and Newly Eligible Adult Enrollment .
Historically, states reported the basis of eligibility in the legacy MSIS in two fields populated with the Maintenance Assistance Status (MAS) and Basis of Eligibility (BOE) codes. These codes were combined in T-MSIS but are no longer required fields. Although MAS/BOE may continue to be reported, fewer states are reporting this data element over time. In place of MAS and BOE, CMS developed a new coding system for classifying eligibility, known as the Eligibility Group, which is the focus of this assessment.
See the T-MSIS Coding Blog on Reporting Eligibility Group at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=47569 .
We conducted two types of analyses using the TAF annual Demographic and Eligibility (DE) file. [5] First, we calculated for each state the percentage of non-dummy [6] enrollment records for Medicaid and CHIP beneficiaries that were missing eligibility group codes for the entire year. [7] Second, we examined whether the state had enrollment in each of the 12 large mandatory eligibility groups. One of the eligibility groups, low-income beneficiaries with disabilities, is captured by one of two eligibility group codes, depending on the state’s 209(b) status, so we examined 13 codes in total. [8] Of all the mandatory eligibility groups, we selected these 12 because they are large enough that we would expect every state to report at least one beneficiary in each group (Table 1). We excluded the remaining mandatory eligibility groups from the second analysis, because they either have limited applicability—for example, individuals who are essential spouses (eligibility group code 14)—or are expected to have relatively low enrollment, such as disabled widows and widowers ineligible for SSI due to early receipt of Social Security (eligibility group code 20).
When a state did not have a beneficiary in one of the 12 large mandatory eligibility groups, we assumed that the state may be incorrectly assigning eligibility group codes in its T-MSIS data. Therefore, we flagged it as a possible data quality issue and counted the total number of groups into which no beneficiary appeared, by state. The greater the number of large mandatory eligibility categories with no beneficiaries, the greater the concern that the state is not correctly mapping eligibility information to the T-MSIS eligibility group codes. However, some individuals are eligible for Medicaid under more than one mandatory eligibility group, and states do not have a consistent way of assigning these individuals to a primary eligibility group. In some cases, it may be the absence of a hierarchy for eligibility group codes that leads to non-reporting for certain mandatory groups.
Table 1. Eligibility group codes for large mandatory Medicaid eligibility groups
| Eligibility group code | Large mandatory Medicaid eligibility group |
|---|---|
| 1 | Parents and other caretaker relatives |
| 5 | Pregnant women |
| 6 | Deemed newborns |
| 7 | Infants and children under age 19 |
| 8 | Children with Title IV-E adoption assistance, foster care, or guardianship care |
| 9 | Former foster care children |
| 11 | Individuals receiving SSI automatically |
| 12 | Individuals receiving SSI through 209(b) provisions |
| 21 | Working disabled under 1619(b) |
| 22 | Disabled adult children |
| 23 | Qualified Medicare beneficiaries |
| 25 | Specified low-income Medicare beneficiaries |
| 26 | Qualifying individuals |
To synthesize the findings from the two analyses, we grouped states into three levels of concern—low, medium, and high—about the quality of eligibility group codes, depending on the percentage of records that were missing an eligibility group code and the count of large mandatory eligibility groups with no enrollment (Table 2). States with particularly high rates of missing data or a large number of mandatory eligibility group codes with no enrollment were classified as unusable.
Table 2. Criteria for DQ assessment of eligibility group codes
| Percentage of beneficiaries missing an eligibility group code | Count of large mandatory eligibility groups with no enrollment | |||
|---|---|---|---|---|
| 0 | 1-2 | 3–6 | 7–12 | |
| x ≤ 5 percent | Low concern | Medium concern | High concern | Unusable |
| 5 percent < x ≤ 10 percent | Medium concern | Medium concern | High concern | Unusable |
| 10 percent < x ≤ 20 percent | High concern | High concern | High concern | Unusable |
| x > 20 percent | Unusable | Unusable | Unusable | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
As part of the creation of the TAF RIF, dummy records are added to the DE file that represent beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state. These dummy records can be identified using the data element MISG_ELGBLTY_DATA_IND.
We used the latest available eligibility group code (ELGBLTY_GRP_CD_LTST) = NULL to define beneficiaries for whom the eligibility group code was missing for the entire year. ELGBLTY_GRP_CD_LTST was constructed in the TAF DE by selecting the value in the most recent month in which an eligibility group code was present. Its value would be NULL if all 12 monthly T-MSIS source values for the eligibility group were missing, unknown, or not on the valid value list. Because the DE file contains one record for every Medicaid and CHIP beneficiary ever enrolled during the year, every beneficiary should have a known eligibility group code. We did not, however, further examine whether the months with a known eligibility group code correspond to the months enrolled or whether the mapping of every known eligibility group code is accurate.
Federal law requires states to cover low-income individuals with disabilities. Most states automatically grant Medicaid to all individuals who receive SSI benefits and assign them to eligibility group code 11. Other states (known as the 209(b) states) use their own eligibility criteria, which are different from the SSI program eligibility criteria, and also use eligibility group code 12 instead of 11. For our analysis, we considered eligibility group codes 11 and 12 as one mandatory coverage category and verified that at least one beneficiary was assigned either code. If neither code was populated, we considered the state to have one mandatory eligibility group with no enrollment.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
As part of the creation of the TAF RIF, dummy records are added to the DE file that represent beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state. These dummy records can be identified using the data element MISG_ELGBLTY_DATA_IND.
We used the latest available eligibility group code (ELGBLTY_GRP_CD_LTST) = NULL to define beneficiaries for whom the eligibility group code was missing for the entire year. ELGBLTY_GRP_CD_LTST was constructed in the TAF DE by selecting the value in the most recent month in which an eligibility group code was present. Its value would be NULL if all 12 monthly T-MSIS source values for the eligibility group were missing, unknown, or not on the valid value list. Because the DE file contains one record for every Medicaid and CHIP beneficiary ever enrolled during the year, every beneficiary should have a known eligibility group code. We did not, however, further examine whether the months with a known eligibility group code correspond to the months enrolled or whether the mapping of every known eligibility group code is accurate.
Federal law requires states to cover low-income individuals with disabilities. Most states automatically grant Medicaid to all individuals who receive SSI benefits and assign them to eligibility group code 11. Other states (known as the 209(b) states) use their own eligibility criteria, which are different from the SSI program eligibility criteria, and also use eligibility group code 12 instead of 11. For our analysis, we considered eligibility group codes 11 and 12 as one mandatory coverage category and verified that at least one beneficiary was assigned either code. If neither code was populated, we considered the state to have one mandatory eligibility group with no enrollment.
The eligibility group code in the TAF can be used to identify the basis on which an individual was deemed eligible for Medicaid or CHIP, including both mandatory and optional eligibility groups. This analysis examines how often the eligibility group code is missing on individual eligibility records and whether states are using all expected codes for certain large, mandatory eligibility groups.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4141"", ""relatedTopics"": []}" 21,"{""measureId"": 21, ""measureName"": ""Restricted Benefits Code"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Restricted-Benefits-Cd.pdf"", ""background"": {""content"": ""The benefits packages available to Medicaid and CHIP beneficiaries vary, depending on beneficiaries’ state of residence and their resources, health conditions, and citizenship status, among other factors. In broad terms, beneficiaries might be entitled to the following: full-scope benefits or all services covered under the Medicaid state plan; comprehensive but not full-scope benefits, which would cover most medical and pharmacy services but not everything in the state plan; or limited benefits, which cover only a small set of services. Users of the T-MSIS Analytic Files (TAF) may wish to study beneficiaries covered by particular benefits packages, or they may wish to remove beneficiaries who do not have full-scope or comprehensive benefits from their analyses because the service use of these beneficiaries cannot be fully evaluated without linking to an external data set.
States have substantial flexibility in enrolling Medicaid beneficiaries into different benefits packages. As a result, there is no national benchmark for the proportion of Medicaid beneficiaries in each state who should qualify for full-scope, comprehensive, or limited benefits. In the majority of states, most beneficiaries qualify for full-scope benefits. However, some states vary significantly from this pattern, which may indicate a data quality issue or may reflect true policy variation in whether and how the states have decided to offer some coverage to beneficiaries who would otherwise not be eligible for Medicaid. For example, some states may be operating large limited-benefit programs covering only family planning or emergency care. Other states are using options such as a Section 1115 demonstration program to provide comprehensive but not full-scope benefits to some beneficiary populations. In contrast, there is less variation in benefit packages under CHIP. Under IRS regulation, all CHIP coverage meets the standard for comprehensive benefits, which means we would not expect any CHIP beneficiaries to be coded as having limited benefits. [1]
The restricted benefits code is the TAF data element that provides standardized information on benefits packages. The purpose of this analysis is to provide TAF users with an assessment of the restricted benefits code in each state, including how often it is missing, coded with an unusable value, or showing an unusual distribution that may indicate the data element is unreliable for analytic purposes. When the restricted benefit code is missing, TAF users may in some cases be able to turn to other enrollment-related data elements in TAF, such as eligibility group code or CHIP code, to impute a beneficiary’s likely benefits package. For example, for a beneficiary with a missing restricted benefits code, a CHIP code that indicates S-CHIP (CHIP_CD = 3) and an eligibility group code that indicates CHIP (ELGBLTY_GRP_CD = 61) implies that the beneficiary likely has either full or comprehensive benefits. The methodological approach in which TAF users deal with missing or unusable values in the restricted benefits code field will depend on the quality and completeness of other variables in their state(s) of interest.
See Mann, Cindy. SHO #14-002. Letter to state health officials and state Medicaid directors, November 7, 2014. Available at https://www.medicaid.gov/federal-policy-guidance/downloads/sho-14-002.pdf .
See Mann, Cindy. SHO #14-002. Letter to state health officials and state Medicaid directors, November 7, 2014. Available at https://www.medicaid.gov/federal-policy-guidance/downloads/sho-14-002.pdf .
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
Using the TAF, [2] we examined the restricted benefits code variable (RSTRCTD_BNFTS_CD_LTST) associated with each non-dummy [3] enrollment record in the annual Demographic and Eligibility (DE) file. We conducted the two analyses described below to evaluate the usability of the restricted benefits code.
Missing and unusable values. We determined the percentage of beneficiaries for whom the restricted benefits code was missing for all months of the year. We separately determined the percentage of beneficiaries who, according to the restricted benefits code, were not enrolled in Medicaid or CHIP during any month in the year (RSTRCTD_BNFTS_CD_LTST = 0). This was a valid value in the T-MSIS Data Dictionary as of 2016 but did not provide any information about the benefits to which a beneficiary is entitled; this value has since been retired.
We assessed the usability of the restricted benefits code in each state based on the percentage of individuals with known benefits, that is, beneficiaries without missing or unusable information in the restricted benefits code field (Table 1).
Table 1. Criteria for DQ assessment of restricted benefits code
| Percentage of beneficiaries with a known benefit type in the restricted benefits code field | DQ assessment |
|---|---|
| x ≥ 95 percent | Low concern |
| 90 percent ≤ x < 95 percent | Medium concern |
| 50 percent ≤ x < 90 percent | High concern |
| x < 50 percent | Unusable |
Distribution of beneficiaries by benefits category. We compared the distribution of beneficiaries in each state who were entitled to full-scope, comprehensive, or limited benefits, as well as those for whom data for the restricted benefits code were missing or unusable. We grouped the values for the restricted benefits code into the benefits categories shown in Table 2 and tabulated the distribution separately for beneficiaries in Medicaid and in Medicaid-Expansion CHIP (M-CHIP), and for beneficiaries in Separate CHIP (S-CHIP). In order to tabulate the distribution of benefits packages for S-CHIP beneficiaries apart from M-CHIP beneficiaries, we used the CHIP code variable. [4] If a state did not report this variable, we used enrollment days to identify Medicaid and M-CHIP beneficiaries (MDCD_ENRLMT_DAYS > 0) and S-CHIP beneficiaries (CHIP_ENRLMT_DAYS > 0).
We present the distribution of Medicaid beneficiaries in each benefits category for TAF users who are interested in this information. Those who are interested in evaluating the reasonableness of the distribution in a given state should do so by using state-specific information about the Medicaid coverage options in that state. In contrast, we would expect S-CHIP beneficiaries to receive only full-scope or comprehensive benefits. Therefore, S-CHIP beneficiaries coded as having limited benefits represent a potential error that reduces the quality of the data.
We excluded states without S-CHIP programs from the tabulation of S-CHIP benefits packages. We excluded states with S-CHIP or combination programs if neither the CHIP code nor enrollment days identified any S-CHIP beneficiaries, indicating major data quality issues.
Table 2. Restricted benefits code values and descriptions, by benefits category
| Benefits category | Values of restricted benefits code | Description of values |
|---|---|---|
| Full-scope | RSTRCTD_BNFTS_CD_LTST = 1, A, B, D | 1: Individual is eligible for Medicaid or CHIP and entitled to the full scope of Medicaid or CHIP benefits. A: Individual is eligible for Medicaid and entitled to benefits under the Psychiatric Residential Treatment Facilities Demonstration Grant Program (PRTF), as enacted by the Deficit Reduction Act of 2005. B: Individual is eligible for Medicaid and entitled to Medicaid benefits using a Health Opportunity Account (HOA) D: Individual is eligible for Medicaid and entitled to benefits under a “Money Follows the Person” (MFP) rebalancing demonstration, as enacted by the Deficit Reduction Act of 2005, to allow states to develop community based long-term care opportunities. |
| Comprehensive | RSTRCTD_BNFTS_CD_LTST = 4 (in states where pregnancy-related services meet the Minimum Essential Coverage standard a ),5, 7 | 4: Individual is eligible for Medicaid or CHIP but only entitled to restricted benefits for pregnancy-related services (including only those states where services meet the Minimum Essential Coverage standard). 5: Individual is eligible for Medicaid or M-CHIP but, for reasons other than alien, dual-eligibility or pregnancy-related status, is only entitled to restricted benefits (e.g., restricted benefits based upon substance abuse, medically needy, or other criteria) that meet the standard for Minimum Essential Coverage. 7: Individual is eligible for Medicaid and entitled to Medicaid benefits under an alternative package of benchmark-equivalent coverage, as enacted by the Deficit Reduction Act of 2005. |
| Limited | RSTRCTD_BNFTS_CD_LTST = 2, 3, 4 (in states where pregnancy-related services do not meet the Minimum Essential Coverage standard a ), 6, E, F | 2: Individual is eligible for Medicaid or M-CHIP, but only entitled to restricted benefits based on alien status. 3: Individual is eligible for Medicaid but only entitled to restricted benefits based on Medicare dual-eligibility status (e.g., QMB, SLMB, QDWI, QI). 4: Individual is eligible for Medicaid or CHIP but only entitled to restricted benefits for pregnancy-related services (including only those states where services do not meet the Minimum Essential Coverage standard). 6: Individual is eligible for Medicaid or M-CHIP but only entitled to restricted benefits for family planning services. E: Individual is eligible for Medicaid or Medicaid-Expansion CHIP, but for reasons other than alien, dual-eligibility, or pregnancy-related status, is only entitled to restricted benefits (e.g., restricted benefits based on substance abuse, medically needy, or other criteria) that do not meet the standard for Minimum Essential Coverage F: Individual is eligible for Medicaid but is only entitled to restricted benefits for medical assistance for COVID-19 diagnostic products and any visit described as a COVID-19 testing-related service for which payment may be made under the State plan during any portion of the public health emergency period, beginning March 18, 2020, as described in Sections 1902(a)(10)(A)(ii)(XXIII), 1902(ss) and clause XVIII in the matter following 1902(a)(10)(G) of the Social Security Act. |
| Unusable | RSTRCTD_BNFTS_CD_LTST = 0 | 0: Individual is not eligible for Medicaid or CHIP during the month. |
| Missing | RSTRCTD_BNFTS_CD_LTST = NULL | NULL: Missing/invalid for all 12 months of the calendar year. |
a Restricted benefits for pregnancy-related services in Arkansas, Idaho, South Dakota, Puerto Rico, and the U.S. Virgin Islands do not meet the Minimum Essential Coverage standard; therefore, beneficiaries in those states with a restricted benefits code value of 4 are considered to have limited Medicaid benefits. In all other states, pregnancy-related Medicaid is considered comprehensive coverage.
b The definition for the restricted benefits code value of 5 changed in 2020 to include only restricted benefits that meet the standard for Minimum Essential Coverage. Prior to the definition change in 2020, the restricted benefits code value of 5 was grouped in the limited benefits category because some states used this code to report limited-benefit groups.
c Restricted benefits code values E and F are newly valid as of 2020.
Methods previously used to assess data quality
Table 3 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 3. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
As part of the creation of the TAF RIF, dummy records are added to the DE file that represent beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state. These dummy records can be identified using the data element MISG_ELGBLTY_DATA_IND.
We classified CHIP_CD_LTST = 1 records as Medicaid beneficiaries, CHIP_CD_LTST = 2 records as M-CHIP beneficiaries, and CHIP_CD_LTST = 3 records as S-CHIP beneficiaries. Since M-CHIP beneficiaries are enrolled into the state Medicaid program and qualify for Medicaid benefits, we included these children in the tabulation of Medicaid enrollees. Information on the completeness of the CHIP code variable can be found in the DQ Atlas single topic display for CHIP Code .
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
As part of the creation of the TAF RIF, dummy records are added to the DE file that represent beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state. These dummy records can be identified using the data element MISG_ELGBLTY_DATA_IND.
We classified CHIP_CD_LTST = 1 records as Medicaid beneficiaries, CHIP_CD_LTST = 2 records as M-CHIP beneficiaries, and CHIP_CD_LTST = 3 records as S-CHIP beneficiaries. Since M-CHIP beneficiaries are enrolled into the state Medicaid program and qualify for Medicaid benefits, we included these children in the tabulation of Medicaid enrollees. Information on the completeness of the CHIP code variable can be found in the DQ Atlas single topic display for CHIP Code .
The restricted benefits code provides information on whether each beneficiary in Medicaid and CHIP is eligible for full-scope, comprehensive, or restricted benefits. This analysis examines how often eligibility records have missing or unusable information in the restricted benefits code. It also provides the distribution of beneficiaries in each state across full-scope, comprehensive, and limited benefit packages in order to identify unusual distributions that may indicate that the data element is unreliable.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4161"", ""relatedTopics"": []}" 22,"{""measureId"": 22, ""measureName"": ""Benchmarking Inpatient Stays - IP"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Benchmarking-IP-Stays.pdf"", ""background"": {""content"": ""Inpatient hospital stays are a critical element when examining health care use, quality, and spending. This data quality assessment examines the reliability of the T-MSIS Analytic Files (TAF) for analyzing inpatient hospital stays among the adult Medicaid population by comparing them to an external benchmark, the Healthcare Cost and Utilization Project (HCUP) data. TAF users should base decisions about using a state’s inpatient data on how the results of their work will be used. However, they may want to be cautious about including states with substantial differences between the numbers of annual inpatient hospital stays reported in the TAF and HCUP data.
"", ""footnotes"": []}, ""methods"": {""content"": ""The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
The analysis relies on three main data sources: (1) the TAF Inpatient (IP) file, (2) the TAF annual Demographic and Eligibility (DE) file, and (3) the HCUP State Inpatient Databases. [1] The TAF IP file contains all inpatient hospital claims for beneficiaries enrolled in either Medicaid or the Children’s Health Insurance Program (CHIP). The HCUP State Inpatient Databases collect inpatient discharge records for all adults regardless of insurance coverage from community hospitals in 47 states and the District of Columbia.
We created an annual file from 12 months of TAF inpatient claims, including both FFS claims and managed care encounters. [2] We excluded states with an unusably low volume of header records in the IP file. For Illinois, we restricted our analysis to the original version of the IP claim, and we excluded all subsequent adjustment records in the state’s TAF data. [3] We then identified unique inpatient hospital stays by rolling up claim records for the same beneficiary if they had (1) the same MSIS identification number and (2) service dates that represent contiguous or overlapping time periods (inpatient services that had a break of more than one day were counted as separate inpatient hospital stays). [4] We then used the beneficiary identification number (MSIS ID) to link the inpatient hospital stays with the beneficiary’s demographic and enrollment information available in the TAF DE for the year.
For all states, we used the TAF to calculate the annual number of inpatient hospital stays among adult beneficiaries. To mirror HCUP as closely as possible, we restricted the TAF output by limiting the data to inpatient hospital stays for adults (Medicaid beneficiaries 19–64 years old) and excluding stays at children’s hospitals (hospital type 07). [5] , [6] We then calculated the percent difference as a percent error: the difference between the TAF and HCUP counts divided by the HCUP count. We organized the data into one of four categories according to their level of alignment with the benchmark and the corresponding level of data quality (Table 1). If a state falls into the high concern or unusable category and TAF inpatient hospital stay counts that were below HCUP counts, this potentially indicates incomplete data in their IP file.
Table 1. Criteria for DQ assessment of TAF inpatient hospital stays in the IP file
| Absolute percent difference between TAF and HCUP data | Alignment category | DQ assessment |
|---|---|---|
| x ≤ 15 percent | High | Low concern |
| 15 percent < x ≤ 30 percent | Moderate | Medium concern |
| 30 percent < x ≤ 50 percent | Low | High concern |
| x > 50 percent | Very low | Unusable |
The methods used to identify inpatient hospital stays in both the TAF and HCUP data differ in several key ways that may result in different counts. First, the TAF claims represent actual billing records paid by Medicaid whereas HCUP provides information on the type of payer the hospital expects will pay. As a result, HCUP may undercount the number of inpatient hospital stays by Medicaid beneficiaries, particularly when a Medicaid managed care plan covers the cost of care. [7] Second, the TAF IP file includes all Medicaid- or CHIP-funded stays at any hospital, whereas the HCUP data do not typically include stays at federal hospitals (such as those operated by the Veterans Administration, Department of Defense, and Indian Health Service) or hospital units within institutions such as prisons. While we attempted to mirror the HCUP universe as closely as possible, data quality problems (such as state errors in the type of claims submitted into the IP file [8] or problems with the hospital type code [9] ) may have prevented us from doing so accurately.
Methods previously used to assess data quality
Table 2 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 2. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We used both header and line claims records in this analysis. Header records summarize a service while the line claims records present the details of the service. We only included claim headers and corresponding lines if the federally assigned service category (FASC) code indicated an inpatient hospital stay. For more information about the FASC code, see the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will over-count service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
A full description of the recommended method for rolling IP records into unique inpatient hospital stays is described in the guide “TAF Technical Documentation: Claims Files,” available on the DQ Atlas Resources page . This analysis follows the recommended method, with one exception. Because HCUP excludes transfers when counting stays, we did not require claims to have the same billing provider when rolling claims into stays and therefore did not count transfers to other acute care hospitals as a new stay.
We limited the data to inpatient hospital stays for adults (19–64 years old) based on the age variable as of their last month of enrollment in the calendar year.
HCUP only includes pregnancy-related inpatient admissions for beneficiaries ages 19–45, whereas our analysis in TAF also includes pregnancy-related stays for beneficiaries over the age of 45. We do not think this difference will substantially impact the results, since women over 45 account for only 0.2 percent of births nationally. See: Martin, Joyce A., Brady E. Hamilton, Michelle J. K. Osterman, Anne K. Driscoll, and Patrick Drake. “Births: Final Data for 2016.” National Vital Statistics Reports, vol. 67, no. 1, January 2018. Available at https://www.cdc.gov/nchs/data/nvsr/nvsr67/nvsr67_01.pdf . Accessed September 5, 2019.
Barrett, M., L. Lopez-Gonzalez, A. Hines, R. Andrews, and J. Jiang. “An Examination of Expected Payer Coding in HCUP Databases.” HCUP Methods Series Report 2014-0. Rockville, MD: Agency for Healthcare Research and Quality, December 17, 2014. Available at https://www.hcup-us.ahrq.gov/reports/methods/2014-03UserGuide.pdf . Accessed September 5, 2019.
For example, states submitting claims for outpatient or rehabilitation facility services into the IP file.
More information can be found in the DQ Atlas single topic display for Hospital Type – IP .
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We used both header and line claims records in this analysis. Header records summarize a service while the line claims records present the details of the service. We only included claim headers and corresponding lines if the federally assigned service category (FASC) code indicated an inpatient hospital stay. For more information about the FASC code, see the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will over-count service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
A full description of the recommended method for rolling IP records into unique inpatient hospital stays is described in the guide \u201cTAF Technical Documentation: Claims Files,\u201d available on the DQ Atlas Resources page . This analysis follows the recommended method, with one exception. Because HCUP excludes transfers when counting stays, we did not require claims to have the same billing provider when rolling claims into stays and therefore did not count transfers to other acute care hospitals as a new stay.
We limited the data to inpatient hospital stays for adults (19\u201364 years old) based on the age variable as of their last month of enrollment in the calendar year.
HCUP only includes pregnancy-related inpatient admissions for beneficiaries ages 19\u201345, whereas our analysis in TAF also includes pregnancy-related stays for beneficiaries over the age of 45. We do not think this difference will substantially impact the results, since women over 45 account for only 0.2 percent of births nationally. See: Martin, Joyce A., Brady E. Hamilton, Michelle J. K. Osterman, Anne K. Driscoll, and Patrick Drake. \u201cBirths: Final Data for 2016.\u201d National Vital Statistics Reports, vol. 67, no. 1, January 2018. Available at https://www.cdc.gov/nchs/data/nvsr/nvsr67/nvsr67_01.pdf . Accessed September 5, 2019.
Barrett, M., L. Lopez-Gonzalez, A. Hines, R. Andrews, and J. Jiang. \u201cAn Examination of Expected Payer Coding in HCUP Databases.\u201d HCUP Methods Series Report 2014-0. Rockville, MD: Agency for Healthcare Research and Quality, December 17, 2014. Available at https://www.hcup-us.ahrq.gov/reports/methods/2014-03UserGuide.pdf . Accessed September 5, 2019.
For example, states submitting claims for outpatient or rehabilitation facility services into the IP file.
More information can be found in the DQ Atlas single topic display for Hospital Type \u2013 IP .
This analysis examines the completeness and reliability of the TAF IP file data by comparing the number of inpatient hospital stays among adults enrolled in Medicaid to an external benchmark from the Healthcare Cost and Utilization Project (HCUP).
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5021"", ""relatedTopics"": []}" 23,"{""measureId"": 23, ""measureName"": ""Type of Bill - IP"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-TOB-IP.pdf"", ""background"": {""content"": ""All medical claims fall into one of two categories: those submitted on an institutional claim form and those submitted on a professional claim form. [1] In general, facilities such as hospitals, nursing facilities, intermediate care facilities for individuals with intellectual or developmental disabilities, rehabilitation facilities, home health agencies, and clinics submit institutional claims. Physicians (both individual and groups), other clinical professionals, free-standing laboratories and outpatient facilities, [2] ambulances, and durable medical equipment suppliers submit professional claims. It is important for users of the T-MSIS Analytic Files (TAF) to be able to distinguish between institutional and professional claims, as the standardized fields in each form, and hence the information available for each type of claim, differ slightly. One important field that is reported only on institutional claims is the type of bill. This field is used to report the type of facility that provides care. Because the type of bill field is used by most payers to determine the payment amount for the claim, it is often well-populated in claims data and is considered a reliable source of information. As a result, it is often the first and easiest data element used to differentiate among key settings and types of institutional care, such as inpatient hospital stays, outpatient hospital visits, or nursing facility care. [3]
This data quality assessment examines the completeness and quality of the type of bill field in the TAF and whether the distribution of values within each medical claim file reflects the types of claims that states are expected to submit.
Institutional claims are often referred to as “UB-04 claims” when submitted in paper form or as “837I claims” when submitted in electronic form. Professional claims are referred to as “CMS-1500 claims” when submitted in paper form or “837P” when submitted in electronic form.
Freestanding facilities are those not owned by a hospital or another institutional provider, such as independent ambulatory surgery centers.
The national provider identifier or provider taxonomy can be used to differentiate among most settings of care, such as nursing facilities versus hospitals, but it requires outside data that can map a large number of potential values to provider type. The revenue code can be used to differentiate types of care, such as inpatient versus outpatient services, but it does not provide information about the type of institution that delivered the care. Type of service is considered less reliable but could be used when type of bill is missing or invalid.
Institutional claims are often referred to as \u201cUB-04 claims\u201d when submitted in paper form or as \u201c837I claims\u201d when submitted in electronic form. Professional claims are referred to as \u201cCMS-1500 claims\u201d when submitted in paper form or \u201c837P\u201d when submitted in electronic form.
Freestanding facilities are those not owned by a hospital or another institutional provider, such as independent ambulatory surgery centers.
The national provider identifier or provider taxonomy can be used to differentiate among most settings of care, such as nursing facilities versus hospitals, but it requires outside data that can map a large number of potential values to provider type. The revenue code can be used to differentiate types of care, such as inpatient versus outpatient services, but it does not provide information about the type of institution that delivered the care. Type of service is considered less reliable but could be used when type of bill is missing or invalid.
Using the TAF, we examined the type of bill field (BILL_TYPE_CD) on header records in the inpatient (IP), long-term care (LT), and other services (OT) files. [4] Since type of bill is not captured on pharmacy claims, we did not examine the pharmacy (RX) file. We included fee-for-service (FFS) claims and managed care encounter records for both Medicaid and CHIP beneficiaries. [5] For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [6]
On institutional claims, the type of bill should always be formatted as a four-digit number that starts with a leading zero. [7] The second and third digits can be used to identify the type of service and facility associated with the claim. The fourth digit provides information about the relationship of the claim to other claims for the same stay; for example, whether the claim covers the entire stay from admission through discharge, or whether it is a continuation claim for a stay that has already been partly billed. [8] For this data quality assessment, we focused only on the second and third digits and allowed any value in the fourth position.
We grouped each of the possible 55 values for the second and third digits in the type of bill into those that are expected or unexpected in each file (Table 1). We also tabulated the extent of missing and invalid values. The IP file should include institutional claims for inpatient hospital services, whereas the LT file should include institutional claims for overnight stays at nursing facilities, intermediate care facilities for individuals with intellectual or developmental disabilities, and residential treatment facilities. The OT file should include a mix of outpatient institutional claims and professional claims from all settings of care. We only expect a type of bill code to be populated on institutional claims and therefore expect that the type of bill code will be missing for a large share of claims in the OT file.
Table 1. Mapping of type of bill values to expected file location
| Value | Description | Expected value in IP file | Expected value in OT file | Expected value in LT file |
|---|---|---|---|---|
| 011x-012x | Inpatient hospital | Yes | ||
| 013x-014x | Outpatient hospital | Yes | ||
| 015x-018x | Hospital intermediate care and swing beds | Yes | ||
| 021x-022x | Nursing facilities - inpatient | Yes | ||
| 023x-024x | Nursing facilities - outpatient | Yes | ||
| 025x-028x | Nursing facilities - intermediate care, swing beds | Yes | ||
| 031x-038x | Home health | Yes | ||
| 041x-042x | Religious nonmedical hospital - inpatient | Yes | ||
| 043x-044x | Religious nonmedical hospital - outpatient | Yes | ||
| 045x-048x | Religious nonmedical hospital - intermediate care, swing beds | Yes | ||
| 061x-068x | Intermediate care facilities | Yes | ||
| 071x-079x | Clinics | Yes | ||
| 081x-084x | Other special facilities | Yes | ||
| 085x | Critical access hospital | Yes | Yes | |
| 086x | Residential facility | Yes | Yes | |
| 089x | Other special facility | Yes | Yes |
In the analysis of the IP and LT files, we assessed level of concern about data based on the percentage of claim headers with expected type of bill values (Table 2). In the data quality assessment of the OT file, we assigned states to either a low or high level of concern about data quality based on a combined percentage of headers with expected or missing type of bill values, because missing values are expected in this file (Table 3).
Table 2. Criteria for DQ assessment of type of bill in the IP and LT files
| Percentage of claim headers with an expected type of bill value | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
Table 3. Criteria for DQ assessment of type of bill in the OT file
| Percentage of claim headers with either an unexpected or invalid type of bill value | Percentage of claim headers missing type of bill value | DQ assessment |
|---|---|---|
| x < 1 percent | x < 99 percent | Low concern a |
| x ≥ 1 percent | x = 100 percent | High concern b |
a Both criteria must be true for a state to receive the given DQ Assessment.
b One of the two criteria must be true for a state to receive the given DQ Assessment.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Claim type code (CLM_TYPE_CD) was used to determine which records to include and exclude. FFS records (claim type 1 or A) and managed care encounters (3 and C) were retained in the analysis. We excluded records with all other claim type values, including capitation payments, service tracking claims, and supplemental payments, none of which are expected to have a valid type of bill value since they are financial transaction records that are not submitted on an institutional claim form. We also excluded the “other” records that the state did not classify as either Medicaid or CHIP payment records; these may represent services that do not qualify for federal matching funds under Title XIX or Title XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
A list of valid values for the second, third, and fourth digits in the type of bill field, and a description of each possible value can be found in the TAF Claims Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries .
There is some variation across states in the coding of the type of bill field. In some states, most or all records had a type of bill value that was three digits long because the leading zero was dropped. We considered these three-digit values to be valid as long as they matched to a valid value once a leading zero was added back in. We did not consider type of bill codes of one or two digits, or three digits with a leading zero (i.e., missing a fourth digit) as valid.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Claim type code (CLM_TYPE_CD) was used to determine which records to include and exclude. FFS records (claim type 1 or A) and managed care encounters (3 and C) were retained in the analysis. We excluded records with all other claim type values, including capitation payments, service tracking claims, and supplemental payments, none of which are expected to have a valid type of bill value since they are financial transaction records that are not submitted on an institutional claim form. We also excluded the \u201cother\u201d records that the state did not classify as either Medicaid or CHIP payment records; these may represent services that do not qualify for federal matching funds under Title XIX or Title XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
A list of valid values for the second, third, and fourth digits in the type of bill field, and a description of each possible value can be found in the TAF Claims Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries .
There is some variation across states in the coding of the type of bill field. In some states, most or all records had a type of bill value that was three digits long because the leading zero was dropped. We considered these three-digit values to be valid as long as they matched to a valid value once a leading zero was added back in. We did not consider type of bill codes of one or two digits, or three digits with a leading zero (i.e., missing a fourth digit) as valid.
The type of bill is a data element present on institutional claims submitted by facilities such as hospitals, nursing facilities, intermediate care facilities, and clinics. It can be used to differentiate between key settings and types of institutional care. This analysis examines how often the type of bill on IP claims is missing or coded with unexpected or invalid values.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5041"", ""relatedTopics"": [{""measureId"": 24, ""measureName"": ""Type of Bill - LT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 1}, {""measureId"": 25, ""measureName"": ""Type of Bill - OT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 2}]}" 24,"{""measureId"": 24, ""measureName"": ""Type of Bill - LT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-TOB-LT.pdf"", ""background"": {""content"": ""All medical claims fall into one of two categories: those submitted on an institutional claim form and those submitted on a professional claim form. [1] In general, facilities such as hospitals, nursing facilities, intermediate care facilities for individuals with intellectual or developmental disabilities, rehabilitation facilities, home health agencies, and clinics submit institutional claims. Physicians (both individual and groups), other clinical professionals, free-standing laboratories and outpatient facilities, [2] ambulances, and durable medical equipment suppliers submit professional claims. It is important for users of the T-MSIS Analytic Files (TAF) to be able to distinguish between institutional and professional claims, as the standardized fields in each form, and hence the information available for each type of claim, differ slightly. One important field that is reported only on institutional claims is the type of bill. This field is used to report the type of facility that provides care. Because the type of bill field is used by most payers to determine the payment amount for the claim, it is often well-populated in claims data and is considered a reliable source of information. As a result, it is often the first and easiest data element used to differentiate among key settings and types of institutional care, such as inpatient hospital stays, outpatient hospital visits, or nursing facility care. [3]
This data quality assessment examines the completeness and quality of the type of bill field in the TAF and whether the distribution of values within each medical claim file reflects the types of claims that states are expected to submit.
Institutional claims are often referred to as “UB-04 claims” when submitted in paper form or as “837I claims” when submitted in electronic form. Professional claims are referred to as “CMS-1500 claims” when submitted in paper form or “837P” when submitted in electronic form.
Freestanding facilities are those not owned by a hospital or another institutional provider, such as independent ambulatory surgery centers.
The national provider identifier or provider taxonomy can be used to differentiate among most settings of care, such as nursing facilities versus hospitals, but it requires outside data that can map a large number of potential values to provider type. The revenue code can be used to differentiate types of care, such as inpatient versus outpatient services, but it does not provide information about the type of institution that delivered the care. Type of service is considered less reliable but could be used when type of bill is missing or invalid.
Institutional claims are often referred to as \u201cUB-04 claims\u201d when submitted in paper form or as \u201c837I claims\u201d when submitted in electronic form. Professional claims are referred to as \u201cCMS-1500 claims\u201d when submitted in paper form or \u201c837P\u201d when submitted in electronic form.
Freestanding facilities are those not owned by a hospital or another institutional provider, such as independent ambulatory surgery centers.
The national provider identifier or provider taxonomy can be used to differentiate among most settings of care, such as nursing facilities versus hospitals, but it requires outside data that can map a large number of potential values to provider type. The revenue code can be used to differentiate types of care, such as inpatient versus outpatient services, but it does not provide information about the type of institution that delivered the care. Type of service is considered less reliable but could be used when type of bill is missing or invalid.
Using the TAF, we examined the type of bill field (BILL_TYPE_CD) on header records in the inpatient (IP), long-term care (LT), and other services (OT) files. [4] Since type of bill is not captured on pharmacy claims, we did not examine the pharmacy (RX) file. We included fee-for-service (FFS) claims and managed care encounter records for both Medicaid and CHIP beneficiaries. [5] For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [6]
On institutional claims, the type of bill should always be formatted as a four-digit number that starts with a leading zero. [7] The second and third digits can be used to identify the type of service and facility associated with the claim. The fourth digit provides information about the relationship of the claim to other claims for the same stay; for example, whether the claim covers the entire stay from admission through discharge, or whether it is a continuation claim for a stay that has already been partly billed. [8] For this data quality assessment, we focused only on the second and third digits and allowed any value in the fourth position.
We grouped each of the possible 55 values for the second and third digits in the type of bill into those that are expected or unexpected in each file (Table 1). We also tabulated the extent of missing and invalid values. The IP file should include institutional claims for inpatient hospital services, whereas the LT file should include institutional claims for overnight stays at nursing facilities, intermediate care facilities for individuals with intellectual or developmental disabilities, and residential treatment facilities. The OT file should include a mix of outpatient institutional claims and professional claims from all settings of care. We only expect a type of bill code to be populated on institutional claims and therefore expect that the type of bill code will be missing for a large share of claims in the OT file.
Table 1. Mapping of type of bill values to expected file location
| Value | Description | Expected value in IP file | Expected value in OT file | Expected value in LT file |
|---|---|---|---|---|
| 011x-012x | Inpatient hospital | Yes | ||
| 013x-014x | Outpatient hospital | Yes | ||
| 015x-018x | Hospital intermediate care and swing beds | Yes | ||
| 021x-022x | Nursing facilities - inpatient | Yes | ||
| 023x-024x | Nursing facilities - outpatient | Yes | ||
| 025x-028x | Nursing facilities - intermediate care, swing beds | Yes | ||
| 031x-038x | Home health | Yes | ||
| 041x-042x | Religious nonmedical hospital - inpatient | Yes | ||
| 043x-044x | Religious nonmedical hospital - outpatient | Yes | ||
| 045x-048x | Religious nonmedical hospital - intermediate care, swing beds | Yes | ||
| 061x-068x | Intermediate care facilities | Yes | ||
| 071x-079x | Clinics | Yes | ||
| 081x-084x | Other special facilities | Yes | ||
| 085x | Critical access hospital | Yes | Yes | |
| 086x | Residential facility | Yes | Yes | |
| 089x | Other special facility | Yes | Yes |
In the analysis of the IP and LT files, we assessed level of concern about data based on the percentage of claim headers with expected type of bill values (Table 2). In the data quality assessment of the OT file, we assigned states to either a low or high level of concern about data quality based on a combined percentage of headers with expected or missing type of bill values, because missing values are expected in this file (Table 3).
Table 2. Criteria for DQ assessment of type of bill in the IP and LT files
| Percentage of claim headers with an expected type of bill value | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
Table 3. Criteria for DQ assessment of type of bill in the OT file
| Percentage of claim headers with either an unexpected or invalid type of bill value | Percentage of claim headers missing type of bill value | DQ assessment |
|---|---|---|
| x < 1 percent | x < 99 percent | Low concern a |
| x ≥ 1 percent | x = 100 percent | High concern b |
a Both criteria must be true for a state to receive the given DQ Assessment.
b One of the two criteria must be true for a state to receive the given DQ Assessment.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Claim type code (CLM_TYPE_CD) was used to determine which records to include and exclude. FFS records (claim type 1 or A) and managed care encounters (3 and C) were retained in the analysis. We excluded records with all other claim type values, including capitation payments, service tracking claims, and supplemental payments, none of which are expected to have a valid type of bill value since they are financial transaction records that are not submitted on an institutional claim form. We also excluded the “other” records that the state did not classify as either Medicaid or CHIP payment records; these may represent services that do not qualify for federal matching funds under Title XIX or Title XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
A list of valid values for the second, third, and fourth digits in the type of bill field, and a description of each possible value can be found in the TAF Claims Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries .
There is some variation across states in the coding of the type of bill field. In some states, most or all records had a type of bill value that was three digits long because the leading zero was dropped. We considered these three-digit values to be valid as long as they matched to a valid value once a leading zero was added back in. We did not consider type of bill codes of one or two digits, or three digits with a leading zero (i.e., missing a fourth digit) as valid.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Claim type code (CLM_TYPE_CD) was used to determine which records to include and exclude. FFS records (claim type 1 or A) and managed care encounters (3 and C) were retained in the analysis. We excluded records with all other claim type values, including capitation payments, service tracking claims, and supplemental payments, none of which are expected to have a valid type of bill value since they are financial transaction records that are not submitted on an institutional claim form. We also excluded the \u201cother\u201d records that the state did not classify as either Medicaid or CHIP payment records; these may represent services that do not qualify for federal matching funds under Title XIX or Title XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
A list of valid values for the second, third, and fourth digits in the type of bill field, and a description of each possible value can be found in the TAF Claims Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries .
There is some variation across states in the coding of the type of bill field. In some states, most or all records had a type of bill value that was three digits long because the leading zero was dropped. We considered these three-digit values to be valid as long as they matched to a valid value once a leading zero was added back in. We did not consider type of bill codes of one or two digits, or three digits with a leading zero (i.e., missing a fourth digit) as valid.
The type of bill is a data element present on institutional claims submitted by facilities such as hospitals, nursing facilities, intermediate care facilities, and clinics. It can be used to differentiate between key settings and types of institutional care. This analysis examines how often the type of bill on LT claims is missing or coded with unexpected or invalid values.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5041"", ""relatedTopics"": [{""measureId"": 23, ""measureName"": ""Type of Bill - IP"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 0}, {""measureId"": 25, ""measureName"": ""Type of Bill - OT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 2}]}" 25,"{""measureId"": 25, ""measureName"": ""Type of Bill - OT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-TOB-OT.pdf"", ""background"": {""content"": ""All medical claims fall into one of two categories: those submitted on an institutional claim form and those submitted on a professional claim form. [1] In general, facilities such as hospitals, nursing facilities, intermediate care facilities for individuals with intellectual or developmental disabilities, rehabilitation facilities, home health agencies, and clinics submit institutional claims. Physicians (both individual and groups), other clinical professionals, free-standing laboratories and outpatient facilities, [2] ambulances, and durable medical equipment suppliers submit professional claims. It is important for users of the T-MSIS Analytic Files (TAF) to be able to distinguish between institutional and professional claims, as the standardized fields in each form, and hence the information available for each type of claim, differ slightly. One important field that is reported only on institutional claims is the type of bill. This field is used to report the type of facility that provides care. Because the type of bill field is used by most payers to determine the payment amount for the claim, it is often well-populated in claims data and is considered a reliable source of information. As a result, it is often the first and easiest data element used to differentiate among key settings and types of institutional care, such as inpatient hospital stays, outpatient hospital visits, or nursing facility care. [3]
This data quality assessment examines the completeness and quality of the type of bill field in the TAF and whether the distribution of values within each medical claim file reflects the types of claims that states are expected to submit.
Institutional claims are often referred to as “UB-04 claims” when submitted in paper form or as “837I claims” when submitted in electronic form. Professional claims are referred to as “CMS-1500 claims” when submitted in paper form or “837P” when submitted in electronic form.
Freestanding facilities are those not owned by a hospital or another institutional provider, such as independent ambulatory surgery centers.
The national provider identifier or provider taxonomy can be used to differentiate among most settings of care, such as nursing facilities versus hospitals, but it requires outside data that can map a large number of potential values to provider type. The revenue code can be used to differentiate types of care, such as inpatient versus outpatient services, but it does not provide information about the type of institution that delivered the care. Type of service is considered less reliable but could be used when type of bill is missing or invalid.
Institutional claims are often referred to as \u201cUB-04 claims\u201d when submitted in paper form or as \u201c837I claims\u201d when submitted in electronic form. Professional claims are referred to as \u201cCMS-1500 claims\u201d when submitted in paper form or \u201c837P\u201d when submitted in electronic form.
Freestanding facilities are those not owned by a hospital or another institutional provider, such as independent ambulatory surgery centers.
The national provider identifier or provider taxonomy can be used to differentiate among most settings of care, such as nursing facilities versus hospitals, but it requires outside data that can map a large number of potential values to provider type. The revenue code can be used to differentiate types of care, such as inpatient versus outpatient services, but it does not provide information about the type of institution that delivered the care. Type of service is considered less reliable but could be used when type of bill is missing or invalid.
Using the TAF, we examined the type of bill field (BILL_TYPE_CD) on header records in the inpatient (IP), long-term care (LT), and other services (OT) files. [4] Since type of bill is not captured on pharmacy claims, we did not examine the pharmacy (RX) file. We included fee-for-service (FFS) claims and managed care encounter records for both Medicaid and CHIP beneficiaries. [5] For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [6]
On institutional claims, the type of bill should always be formatted as a four-digit number that starts with a leading zero. [7] The second and third digits can be used to identify the type of service and facility associated with the claim. The fourth digit provides information about the relationship of the claim to other claims for the same stay; for example, whether the claim covers the entire stay from admission through discharge, or whether it is a continuation claim for a stay that has already been partly billed. [8] For this data quality assessment, we focused only on the second and third digits and allowed any value in the fourth position.
We grouped each of the possible 55 values for the second and third digits in the type of bill into those that are expected or unexpected in each file (Table 1). We also tabulated the extent of missing and invalid values. The IP file should include institutional claims for inpatient hospital services, whereas the LT file should include institutional claims for overnight stays at nursing facilities, intermediate care facilities for individuals with intellectual or developmental disabilities, and residential treatment facilities. The OT file should include a mix of outpatient institutional claims and professional claims from all settings of care. We only expect a type of bill code to be populated on institutional claims and therefore expect that the type of bill code will be missing for a large share of claims in the OT file.
Table 1. Mapping of type of bill values to expected file location
| Value | Description | Expected value in IP file | Expected value in OT file | Expected value in LT file |
|---|---|---|---|---|
| 011x-012x | Inpatient hospital | Yes | ||
| 013x-014x | Outpatient hospital | Yes | ||
| 015x-018x | Hospital intermediate care and swing beds | Yes | ||
| 021x-022x | Nursing facilities - inpatient | Yes | ||
| 023x-024x | Nursing facilities - outpatient | Yes | ||
| 025x-028x | Nursing facilities - intermediate care, swing beds | Yes | ||
| 031x-038x | Home health | Yes | ||
| 041x-042x | Religious nonmedical hospital - inpatient | Yes | ||
| 043x-044x | Religious nonmedical hospital - outpatient | Yes | ||
| 045x-048x | Religious nonmedical hospital - intermediate care, swing beds | Yes | ||
| 061x-068x | Intermediate care facilities | Yes | ||
| 071x-079x | Clinics | Yes | ||
| 081x-084x | Other special facilities | Yes | ||
| 085x | Critical access hospital | Yes | Yes | |
| 086x | Residential facility | Yes | Yes | |
| 089x | Other special facility | Yes | Yes |
In the analysis of the IP and LT files, we assessed level of concern about data based on the percentage of claim headers with expected type of bill values (Table 2). In the data quality assessment of the OT file, we assigned states to either a low or high level of concern about data quality based on a combined percentage of headers with expected or missing type of bill values, because missing values are expected in this file (Table 3).
Table 2. Criteria for DQ assessment of type of bill in the IP and LT files
| Percentage of claim headers with an expected type of bill value | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
Table 3. Criteria for DQ assessment of type of bill in the OT file
| Percentage of claim headers with either an unexpected or invalid type of bill value | Percentage of claim headers missing type of bill value | DQ assessment |
|---|---|---|
| x < 1 percent | x < 99 percent | Low concern a |
| x ≥ 1 percent | x = 100 percent | High concern b |
a Both criteria must be true for a state to receive the given DQ Assessment.
b One of the two criteria must be true for a state to receive the given DQ Assessment.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Claim type code (CLM_TYPE_CD) was used to determine which records to include and exclude. FFS records (claim type 1 or A) and managed care encounters (3 and C) were retained in the analysis. We excluded records with all other claim type values, including capitation payments, service tracking claims, and supplemental payments, none of which are expected to have a valid type of bill value since they are financial transaction records that are not submitted on an institutional claim form. We also excluded the “other” records that the state did not classify as either Medicaid or CHIP payment records; these may represent services that do not qualify for federal matching funds under Title XIX or Title XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
A list of valid values for the second, third, and fourth digits in the type of bill field, and a description of each possible value can be found in the TAF Claims Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries .
There is some variation across states in the coding of the type of bill field. In some states, most or all records had a type of bill value that was three digits long because the leading zero was dropped. We considered these three-digit values to be valid as long as they matched to a valid value once a leading zero was added back in. We did not consider type of bill codes of one or two digits, or three digits with a leading zero (i.e., missing a fourth digit) as valid.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Claim type code (CLM_TYPE_CD) was used to determine which records to include and exclude. FFS records (claim type 1 or A) and managed care encounters (3 and C) were retained in the analysis. We excluded records with all other claim type values, including capitation payments, service tracking claims, and supplemental payments, none of which are expected to have a valid type of bill value since they are financial transaction records that are not submitted on an institutional claim form. We also excluded the \u201cother\u201d records that the state did not classify as either Medicaid or CHIP payment records; these may represent services that do not qualify for federal matching funds under Title XIX or Title XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
A list of valid values for the second, third, and fourth digits in the type of bill field, and a description of each possible value can be found in the TAF Claims Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries .
There is some variation across states in the coding of the type of bill field. In some states, most or all records had a type of bill value that was three digits long because the leading zero was dropped. We considered these three-digit values to be valid as long as they matched to a valid value once a leading zero was added back in. We did not consider type of bill codes of one or two digits, or three digits with a leading zero (i.e., missing a fourth digit) as valid.
The type of bill is a data element present on institutional claims submitted by facilities such as hospitals, nursing facilities, intermediate care facilities, and clinics. It can be used to differentiate between key settings and types of institutional care. Many records in the OT file represent professional claims and are expected to have a missing type of bill code. This analysis examines how often the type of bill on OT claims is coded with unexpected or invalid values.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5041"", ""relatedTopics"": [{""measureId"": 23, ""measureName"": ""Type of Bill - IP"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 0}, {""measureId"": 24, ""measureName"": ""Type of Bill - LT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 1}]}" 26,"{""measureId"": 26, ""measureName"": ""Hospital Type - IP"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Hospital-Type-IP.pdf"", ""background"": {""content"": ""Users of the T-MSIS Analytic Files (TAF) may be interested in the types of hospitals that treat Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries in inpatient settings. Claims for inpatient hospital care are in the monthly inpatient (IP) TAF claims files and include both (1) header records that summarize the claims and (2) line records that detail the services covered by the claim. Each claim header record in the IP file includes a code showing the type of hospital that researchers can use to distinguish broad groups of hospital-based care: inpatient, outpatient, critical access hospital (CAH), swing bed, inpatient psychiatric, Indian Health Service (IHS), children’s hospital, other, or not a hospital.
TAF users can use the type of hospital field to understand which facilities are providing inpatient services to Medicaid and CHIP beneficiaries. Without this field, TAF users who wish to determine hospital type must either (1) use the provider taxonomy codes [1] in the TAF to crosswalk between providers and the Health Care Provider Taxonomy Code Set; [2] (2) link National Provider Identifiers in the TAF to provider taxonomies reported in the National Plan and Provider Enumeration System, which also uses the Health Care Provider Taxonomy Code Set; or (3) use the TAF codes for billing provider type, which include two classifications expected in the IP file (Hospital-General and Indian Health Service Facility). [3] However, the type of hospital field must be reported completely and accurately in the claims records to be used for analysis. States were not given explicit guidance on coding this field, and their construction methods could vary. We expect the large majority of inpatient claims records will reflect visits to inpatient, critical access, and children’s hospitals, and that only a small percentage of each state’s records will reflect use of the other hospital types. In this data quality assessment, we identify states with unusual or unexpected patterns in the type of hospital field.
The billing provider taxonomy code is based on the provider classification self-selected by the provider when applying for a National Provider Identifier. The completeness and quality of data in this field are not covered in this analysis.
Washington Publishing Company. “Health Care Provider Taxonomy Code Set.” 2019. Available at http://www.wpc-edi.com/reference/codelists/healthcare/health-care-provider-taxonomy-code-set/ . Accessed April 12, 2019.
More information can be found in the DQ Atlas single topic display for Billing Provider Type – IP .
The billing provider taxonomy code is based on the provider classification self-selected by the provider when applying for a National Provider Identifier. The completeness and quality of data in this field are not covered in this analysis.
Washington Publishing Company. \u201cHealth Care Provider Taxonomy Code Set.\u201d 2019. Available at http://www.wpc-edi.com/reference/codelists/healthcare/health-care-provider-taxonomy-code-set/ . Accessed April 12, 2019.
More information can be found in the DQ Atlas single topic display for Billing Provider Type \u2013 IP .
Using the TAF, we examined the type of hospital field (HOSP_TYPE_CD) on header records in the IP file. [4] We included fee-for-service claims and managed care encounter records, because information on the hospital type should be available in both. We excluded financial transaction records, supplemental payments, and “other” records that the state did not classify as being covered by either the Medicaid or CHIP program. [5] For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [6]
On IP claims, the type of hospital code should always be formatted as a two-digit number that starts with a leading zero. The second digit identifies the type of hospital associated with the claim. For this analysis, we grouped each of the possible 10 values (including missing) in the type of hospital field into higher-level categories of expected, unexpected, other, and missing (Table 1). We then identified states with unanticipated patterns in coding for this field.
Table 1. Aggregation of type-of-hospital categories
| Type of hospital | Value for type of hospital variable | Type of hospital aggregated category |
|---|---|---|
| Not a hospital | 00 | Unexpected code |
| Inpatient hospital | 01 | Expected code |
| Outpatient hospital | 02 | Unexpected code |
| Critical access hospital | 03 | Expected code |
| Swing bed hospital | 04 | Unexpected code |
| Inpatient psychiatric hospital | 05 | Unexpected code |
| IHS hospital | 06 | Expected code |
| Children’s hospital | 07 | Expected code |
| Other | 08 | Other |
| Missing | . or NULL | Missing |
Note:\tValues not listed in the table were invalid.
We would not expect records from all hospitals to be found in the IP file. We expect claims with an outpatient hospital type to be reported in the other services (OT) file and claims with an inpatient psychiatric hospital type or for beneficiaries receiving nursing facility services furnished by a hospital with swing-bed approval to be reported in the long-term care (LT) file. While CMS guidance in the T-MSIS Data Dictionary instructs states to include services provided in a psychiatric wing of a general hospital in the IP file if the wing is not administratively separate from the hospital, we expect claims for services provided at all other types of inpatient psychiatric facilities to be included in the LT file. [7] In addition, claims with missing, unexpected, or the nonspecific “other” code may not be usable when the analytic work requires identifying specific types of hospital settings.
We assessed the usability of the hospital type code based on the percentage of IP records that had an expected, specific (non-other) type of hospital code for the file (Table 2).
Table 2. Criteria for DQ assessment of hospital type code
| DQ assessment | |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Claim type code (CLM_TYPE_CD) was used to determine which records to include and which to exclude from our analysis. Included records are those with a claim type code indicating fee-for-service claims (values of 1 and A) and managed care encounters (3 and C). We excluded capitated payments (2 and B), supplemental payments (5 and E), and service-tracking claims (4 and D). We also used claim type code to exclude “other” records (values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information about the “other” records can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
The T-MSIS Data Dictionary is available at: https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/index.html .
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Claim type code (CLM_TYPE_CD) was used to determine which records to include and which to exclude from our analysis. Included records are those with a claim type code indicating fee-for-service claims (values of 1 and A) and managed care encounters (3 and C). We excluded capitated payments (2 and B), supplemental payments (5 and E), and service-tracking claims (4 and D). We also used claim type code to exclude \u201cother\u201d records (values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information about the \u201cother\u201d records can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
The T-MSIS Data Dictionary is available at: https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/index.html .
Each header record in the IP file includes a code indicating the type of hospital that submitted the claim, which can be used to understand what types of facilities are providing inpatient services to Medicaid and CHIP beneficiaries. This analysis examines how often the hospital type field contains missing or unexpected information that may indicate a data quality problem.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5051"", ""relatedTopics"": []}" 27,"{""measureId"": 27, ""measureName"": ""Service Users - IP"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Service-Users-IP.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) consist of enrollment and service use for beneficiaries enrolled in Medicaid and the Children’s Health Insurance Program (CHIP). Records of service use, including both fee-for-service (FFS) claims paid by the state Medicaid agency and managed care encounters paid by a Medicaid managed care plan, are included in the TAF inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) files.
Within federal guidelines, states choose which populations and benefit categories that they cover, and as a result, their Medicaid and CHIP programs vary in the characteristics of their covered populations and in their benefit packages. Nevertheless, examining the overall percentage of beneficiaries with a claim in a given year can help identify outlier states that TAF users should examine more closely before starting their analysis. For example, if a state submits incomplete data on FFS claims or managed care encounters, TAF users could underestimate utilization, expenditures, and the prevalence of medical conditions in beneficiaries in that state. If too many beneficiaries in a state had a claim, it could signal that the states either submitted service records into the wrong file or had other data quality issues. TAF users would need to consider these issues when designing their own data quality checks and analyses.
This data quality assessment examines the percentage of beneficiaries in each state with any medical or pharmacy claims in the TAF data, as a way to identify states with potentially incomplete service use data. [1]
More information about the completeness of each state’s TAF data can be found in the DQ Atlas single topic displays for Claims Volume - IP , Claims Volume - LT , Claims Volume - OT , and Claims Volume - RX .
More information about the completeness of each state\u2019s TAF data can be found in the DQ Atlas single topic displays for Claims Volume - IP , Claims Volume - LT , Claims Volume - OT , and Claims Volume - RX .
We used the annual Demographic and Eligibility (DE) TAF and three TAF claim files (IP, OT, and RX) to examine the percentage of beneficiaries who were enrolled in Medicaid or CHIP during any month in the calendar year and had any claims in each file. [2]
We do not examine the percentage of beneficiaries who have service use in the LT file for several reasons. First, states have very different policies on providing, billing for, and recording LT services, which include facility-based long-term care provided by nursing homes, intermediate care facilities for individuals with intellectual or developmental disabilities, services provided in a mental health facility, and independent psychiatric wings of acute care hospitals. Second, although most states emphasize the use of home- and community-based services over institutional long-term care, the success of this strategy varies by state. Third, this analysis excludes dually eligible beneficiaries, which translates to excluding nearly all beneficiaries over the age of 65 and substantially limits the population in each state that we would expect to have an LT claim. As a result, small differences across states in the percentage of non-dually eligible beneficiaries who use long-term care services cannot be interpreted as a data quality issue, especially because the beneficiary population’s need for institutional long-term care could also vary from one state to the next. [3]
To identify beneficiaries enrolled in Medicaid or CHIP for at least one month in the year, we used the CHIP code variable in the annual DE file. [4] We excluded beneficiaries who were dually eligible for Medicaid and Medicare in any month of the year because Medicare is the primary payer for most medical care for these beneficiaries. [5] For each non-dually eligible beneficiary, we then looked for at least one claim in each file (IP, OT, and RX) that linked to the beneficiary’s enrollment record in the annual DE file. We examined this linkage on an annual basis instead of a monthly basis because not all beneficiaries receive care each month of the year. We included both FFS claims and managed care encounter records in this analysis, but excluded capitation payments, supplemental claims, service tracking claims, and “other” claims. [6] We also excluded crossover claims because they were for dually eligible beneficiaries whose services were partially paid by Medicare. [7]
We used header (and not line) records when linking claims to beneficiaries for this analysis. Header records include summary information about the claim as a whole, whereas line records detail the individual goods and services billed as part of the claim. In some instances, data quality issues prevent the accurate linking of header records with their corresponding lines. Header claim records with missing line-level detail are included in TAF, but line records that do not link to a header are excluded. As a result, the findings from this analysis identify states that may be missing entire claims from their claims files, but do not identify states in which claims data are partially incomplete due to missing line records.
We compared the percentage of beneficiaries in every state who had at least one claim in each file to the national median for that file to identify states that might have data quality issues. States may be outliers on this measure because of issues with data quality or completeness in a claims file. Alternatively, some states may be outliers because of data quality issues in the DE file. For instance, some states have known issues with missing or unreliable information in the CHIP code or dual code variables that would prevent accurately measuring the size of the non-dual population. [8] Finally, some states may be outliers because of data quality issues that prevent accurately linking service use records to beneficiary eligibility records.
We classified a state’s IP, OT, or RX file as unusable if the percentage of beneficiaries with service use was less than 10 percent of the national median value for the file, because that pattern suggests the data are incomplete or have major data quality problems (Table 1). We had a high level of concern about a state’s claims data file if the percentage of beneficiaries with service use was either: (1) between 10 and 50 percent of the national median, raising the possibility of an incomplete data file, or (2) more than 200 percent of the national median, raising the possibility that the state was formatting claims incorrectly or including claims records that should have been reported into a different file. Finally, we had a medium level of concern about a state’s claims data file if the percentage of beneficiaries with service use was either (1) between 50 and 75 percent, or (2) between 150 to 200 percent of the national median.
Table 1. Criteria for DQ assessment of service use data in the IP, OT, and RX files
| Beneficiaries who are service users, as a percentage of the national median value | Number of service users | DQ assessment |
|---|---|---|
| x < 10 percent | Too low | Unusable |
| 10 percent ≤ x < 50 percent | Too low | High concern |
| 50 percent ≤ x < 75 percent | Low | Medium concern |
| 75 percent ≤ x < 150 percent | Acceptable | Low concern |
| 150 percent ≤ x ≤ 200 percent | High | Medium concern |
| x > 200 percent | Too high | High concern |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Eligibility for institutional-based long-term care services that would be captured in the LT file is limited to beneficiaries who have been determined to need this level of care. Because this analysis excludes dual eligible beneficiaries, users of long-term care services are most likely to be beneficiaries who qualify for Medicaid on the basis of disability. States have varying rates of beneficiaries who qualify for Medicaid on the basis of disability; consequently, they are likely to have varying rates of non-dually eligible beneficiaries who qualify for LT services.
If it was available, we used the monthly CHIP code (CHIP_CD) to identify beneficiaries enrolled in Medicaid or CHIP that month (CHIP_CD values of 1–4). CHIP code 4 (Medicaid and S-CHIP) is a valid value for 2014 through 2017 TAF. If the CHIP code was missing, we considered the beneficiary as enrolled in the month only if the eligibility group variable (ELGBLTY_GRP_CD) was not missing.
We removed data on dually eligible beneficiaries by removing any person identified in the annual DE for whom any dual status code (DUAL_ELGBL_CD_mm, where mm refers to the month) took on one of the following values that indicate when a Medicaid beneficiary is eligible for Medicare: 01, 02, 03, 04, 05, 06, 08, 09, or 10.
We identified FFS claims and managed care encounters by using claim type code (CLM_TYPE_CD) values of 1, A, 3, and C. We excluded any records that the state classified as capitation payments, supplemental claims, service tracking claims, or “other” claims. “Other” claims represent FFS claims, capitated payments, managed care encounters, service tracking claims, and supplemental payments that states have classified as “other” rather than as Medicaid or CHIP. More information can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
Crossover claims are service use records where Medicare processes and pays the claim first as the primary payer, and Medicaid processes the claim as a secondary payer. Crossover claims are identified using the criteria XOVR_IND = 1.
Using CHIP code to identify beneficiaries enrolled in Medicaid or CHIP provides the closest estimates to external program enrollment benchmarks in the most states. However, some states have missing or unreliable information captured in the CHIP code variable, and TAF-based enrollment counts are not as accurate in those states. Some states also have problems correctly identifying dually eligible beneficiaries using the dual code.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Eligibility for institutional-based long-term care services that would be captured in the LT file is limited to beneficiaries who have been determined to need this level of care. Because this analysis excludes dual eligible beneficiaries, users of long-term care services are most likely to be beneficiaries who qualify for Medicaid on the basis of disability. States have varying rates of beneficiaries who qualify for Medicaid on the basis of disability; consequently, they are likely to have varying rates of non-dually eligible beneficiaries who qualify for LT services.
If it was available, we used the monthly CHIP code (CHIP_CD) to identify beneficiaries enrolled in Medicaid or CHIP that month (CHIP_CD values of 1\u20134). CHIP code 4 (Medicaid and S-CHIP) is a valid value for 2014 through 2017 TAF. If the CHIP code was missing, we considered the beneficiary as enrolled in the month only if the eligibility group variable (ELGBLTY_GRP_CD) was not missing.
We removed data on dually eligible beneficiaries by removing any person identified in the annual DE for whom any dual status code (DUAL_ELGBL_CD_mm, where mm refers to the month) took on one of the following values that indicate when a Medicaid beneficiary is eligible for Medicare: 01, 02, 03, 04, 05, 06, 08, 09, or 10.
We identified FFS claims and managed care encounters by using claim type code (CLM_TYPE_CD) values of 1, A, 3, and C. We excluded any records that the state classified as capitation payments, supplemental claims, service tracking claims, or \u201cother\u201d claims. \u201cOther\u201d claims represent FFS claims, capitated payments, managed care encounters, service tracking claims, and supplemental payments that states have classified as \u201cother\u201d rather than as Medicaid or CHIP. More information can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
Crossover claims are service use records where Medicare processes and pays the claim first as the primary payer, and Medicaid processes the claim as a secondary payer. Crossover claims are identified using the criteria XOVR_IND = 1.
Using CHIP code to identify beneficiaries enrolled in Medicaid or CHIP provides the closest estimates to external program enrollment benchmarks in the most states. However, some states have missing or unreliable information captured in the CHIP code variable, and TAF-based enrollment counts are not as accurate in those states. Some states also have problems correctly identifying dually eligible beneficiaries using the dual code.
Although states differ in the populations and benefits covered in their Medicaid and CHIP programs, examining the overall percentage of beneficiaries with any service use can identify outlier states that may have incomplete claims, encounter, or eligibility data in the TAF. Low rates of service use may also indicate problems in linking service use and eligibility records. This analysis examines the percentage of beneficiaries in each state with an IP record indicating the receipt of inpatient services during the year.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5061"", ""relatedTopics"": [{""measureId"": 28, ""measureName"": ""Service Users - OT"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 1}, {""measureId"": 29, ""measureName"": ""Service Users - RX"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 2}]}" 28,"{""measureId"": 28, ""measureName"": ""Service Users - OT"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Service-Users-OT.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) consist of enrollment and service use for beneficiaries enrolled in Medicaid and the Children’s Health Insurance Program (CHIP). Records of service use, including both fee-for-service (FFS) claims paid by the state Medicaid agency and managed care encounters paid by a Medicaid managed care plan, are included in the TAF inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) files.
Within federal guidelines, states choose which populations and benefit categories that they cover, and as a result, their Medicaid and CHIP programs vary in the characteristics of their covered populations and in their benefit packages. Nevertheless, examining the overall percentage of beneficiaries with a claim in a given year can help identify outlier states that TAF users should examine more closely before starting their analysis. For example, if a state submits incomplete data on FFS claims or managed care encounters, TAF users could underestimate utilization, expenditures, and the prevalence of medical conditions in beneficiaries in that state. If too many beneficiaries in a state had a claim, it could signal that the states either submitted service records into the wrong file or had other data quality issues. TAF users would need to consider these issues when designing their own data quality checks and analyses.
This data quality assessment examines the percentage of beneficiaries in each state with any medical or pharmacy claims in the TAF data, as a way to identify states with potentially incomplete service use data. [1]
More information about the completeness of each state’s TAF data can be found in the DQ Atlas single topic displays for Claims Volume - IP , Claims Volume - LT , Claims Volume - OT , and Claims Volume - RX .
More information about the completeness of each state\u2019s TAF data can be found in the DQ Atlas single topic displays for Claims Volume - IP , Claims Volume - LT , Claims Volume - OT , and Claims Volume - RX .
We used the annual Demographic and Eligibility (DE) TAF and three TAF claim files (IP, OT, and RX) to examine the percentage of beneficiaries who were enrolled in Medicaid or CHIP during any month in the calendar year and had any claims in each file. [2]
We do not examine the percentage of beneficiaries who have service use in the LT file for several reasons. First, states have very different policies on providing, billing for, and recording LT services, which include facility-based long-term care provided by nursing homes, intermediate care facilities for individuals with intellectual or developmental disabilities, services provided in a mental health facility, and independent psychiatric wings of acute care hospitals. Second, although most states emphasize the use of home- and community-based services over institutional long-term care, the success of this strategy varies by state. Third, this analysis excludes dually eligible beneficiaries, which translates to excluding nearly all beneficiaries over the age of 65 and substantially limits the population in each state that we would expect to have an LT claim. As a result, small differences across states in the percentage of non-dually eligible beneficiaries who use long-term care services cannot be interpreted as a data quality issue, especially because the beneficiary population’s need for institutional long-term care could also vary from one state to the next. [3]
To identify beneficiaries enrolled in Medicaid or CHIP for at least one month in the year, we used the CHIP code variable in the annual DE file. [4] We excluded beneficiaries who were dually eligible for Medicaid and Medicare in any month of the year because Medicare is the primary payer for most medical care for these beneficiaries. [5] For each non-dually eligible beneficiary, we then looked for at least one claim in each file (IP, OT, and RX) that linked to the beneficiary’s enrollment record in the annual DE file. We examined this linkage on an annual basis instead of a monthly basis because not all beneficiaries receive care each month of the year. We included both FFS claims and managed care encounter records in this analysis, but excluded capitation payments, supplemental claims, service tracking claims, and “other” claims. [6] We also excluded crossover claims because they were for dually eligible beneficiaries whose services were partially paid by Medicare. [7]
We used header (and not line) records when linking claims to beneficiaries for this analysis. Header records include summary information about the claim as a whole, whereas line records detail the individual goods and services billed as part of the claim. In some instances, data quality issues prevent the accurate linking of header records with their corresponding lines. Header claim records with missing line-level detail are included in TAF, but line records that do not link to a header are excluded. As a result, the findings from this analysis identify states that may be missing entire claims from their claims files, but do not identify states in which claims data are partially incomplete due to missing line records.
We compared the percentage of beneficiaries in every state who had at least one claim in each file to the national median for that file to identify states that might have data quality issues. States may be outliers on this measure because of issues with data quality or completeness in a claims file. Alternatively, some states may be outliers because of data quality issues in the DE file. For instance, some states have known issues with missing or unreliable information in the CHIP code or dual code variables that would prevent accurately measuring the size of the non-dual population. [8] Finally, some states may be outliers because of data quality issues that prevent accurately linking service use records to beneficiary eligibility records.
We classified a state’s IP, OT, or RX file as unusable if the percentage of beneficiaries with service use was less than 10 percent of the national median value for the file, because that pattern suggests the data are incomplete or have major data quality problems (Table 1). We had a high level of concern about a state’s claims data file if the percentage of beneficiaries with service use was either: (1) between 10 and 50 percent of the national median, raising the possibility of an incomplete data file, or (2) more than 200 percent of the national median, raising the possibility that the state was formatting claims incorrectly or including claims records that should have been reported into a different file. Finally, we had a medium level of concern about a state’s claims data file if the percentage of beneficiaries with service use was either (1) between 50 and 75 percent, or (2) between 150 to 200 percent of the national median.
Table 1. Criteria for DQ assessment of service use data in the IP, OT, and RX files
| Beneficiaries who are service users, as a percentage of the national median value | Number of service users | DQ assessment |
|---|---|---|
| x < 10 percent | Too low | Unusable |
| 10 percent ≤ x < 50 percent | Too low | High concern |
| 50 percent ≤ x < 75 percent | Low | Medium concern |
| 75 percent ≤ x < 150 percent | Acceptable | Low concern |
| 150 percent ≤ x ≤ 200 percent | High | Medium concern |
| x > 200 percent | Too high | High concern |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Eligibility for institutional-based long-term care services that would be captured in the LT file is limited to beneficiaries who have been determined to need this level of care. Because this analysis excludes dual eligible beneficiaries, users of long-term care services are most likely to be beneficiaries who qualify for Medicaid on the basis of disability. States have varying rates of beneficiaries who qualify for Medicaid on the basis of disability; consequently, they are likely to have varying rates of non-dually eligible beneficiaries who qualify for LT services.
If it was available, we used the monthly CHIP code (CHIP_CD) to identify beneficiaries enrolled in Medicaid or CHIP that month (CHIP_CD values of 1–4). CHIP code 4 (Medicaid and S-CHIP) is a valid value for 2014 through 2017 TAF. If the CHIP code was missing, we considered the beneficiary as enrolled in the month only if the eligibility group variable (ELGBLTY_GRP_CD) was not missing.
We removed data on dually eligible beneficiaries by removing any person identified in the annual DE for whom any dual status code (DUAL_ELGBL_CD_mm, where mm refers to the month) took on one of the following values that indicate when a Medicaid beneficiary is eligible for Medicare: 01, 02, 03, 04, 05, 06, 08, 09, or 10.
We identified FFS claims and managed care encounters by using claim type code (CLM_TYPE_CD) values of 1, A, 3, and C. We excluded any records that the state classified as capitation payments, supplemental claims, service tracking claims, or “other” claims. “Other” claims represent FFS claims, capitated payments, managed care encounters, service tracking claims, and supplemental payments that states have classified as “other” rather than as Medicaid or CHIP. More information can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
Crossover claims are service use records where Medicare processes and pays the claim first as the primary payer, and Medicaid processes the claim as a secondary payer. Crossover claims are identified using the criteria XOVR_IND = 1.
Using CHIP code to identify beneficiaries enrolled in Medicaid or CHIP provides the closest estimates to external program enrollment benchmarks in the most states. However, some states have missing or unreliable information captured in the CHIP code variable, and TAF-based enrollment counts are not as accurate in those states. Some states also have problems correctly identifying dually eligible beneficiaries using the dual code.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Eligibility for institutional-based long-term care services that would be captured in the LT file is limited to beneficiaries who have been determined to need this level of care. Because this analysis excludes dual eligible beneficiaries, users of long-term care services are most likely to be beneficiaries who qualify for Medicaid on the basis of disability. States have varying rates of beneficiaries who qualify for Medicaid on the basis of disability; consequently, they are likely to have varying rates of non-dually eligible beneficiaries who qualify for LT services.
If it was available, we used the monthly CHIP code (CHIP_CD) to identify beneficiaries enrolled in Medicaid or CHIP that month (CHIP_CD values of 1\u20134). CHIP code 4 (Medicaid and S-CHIP) is a valid value for 2014 through 2017 TAF. If the CHIP code was missing, we considered the beneficiary as enrolled in the month only if the eligibility group variable (ELGBLTY_GRP_CD) was not missing.
We removed data on dually eligible beneficiaries by removing any person identified in the annual DE for whom any dual status code (DUAL_ELGBL_CD_mm, where mm refers to the month) took on one of the following values that indicate when a Medicaid beneficiary is eligible for Medicare: 01, 02, 03, 04, 05, 06, 08, 09, or 10.
We identified FFS claims and managed care encounters by using claim type code (CLM_TYPE_CD) values of 1, A, 3, and C. We excluded any records that the state classified as capitation payments, supplemental claims, service tracking claims, or \u201cother\u201d claims. \u201cOther\u201d claims represent FFS claims, capitated payments, managed care encounters, service tracking claims, and supplemental payments that states have classified as \u201cother\u201d rather than as Medicaid or CHIP. More information can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
Crossover claims are service use records where Medicare processes and pays the claim first as the primary payer, and Medicaid processes the claim as a secondary payer. Crossover claims are identified using the criteria XOVR_IND = 1.
Using CHIP code to identify beneficiaries enrolled in Medicaid or CHIP provides the closest estimates to external program enrollment benchmarks in the most states. However, some states have missing or unreliable information captured in the CHIP code variable, and TAF-based enrollment counts are not as accurate in those states. Some states also have problems correctly identifying dually eligible beneficiaries using the dual code.
Although states differ in the populations and benefits covered in their Medicaid and CHIP programs, examining the overall percentage of beneficiaries with any service use can identify outlier states that may have incomplete claims, encounter, or eligibility data in the TAF. Low rates of service use may also indicate problems in linking service use and eligibility records. This analysis examines the percentage of beneficiaries in each state with an OT record indicating the receipt of ambulatory, physician, or other medical services during the year.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5061"", ""relatedTopics"": [{""measureId"": 27, ""measureName"": ""Service Users - IP"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 0}, {""measureId"": 29, ""measureName"": ""Service Users - RX"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 2}]}" 29,"{""measureId"": 29, ""measureName"": ""Service Users - RX"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Service-Users-RX.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) consist of enrollment and service use for beneficiaries enrolled in Medicaid and the Children’s Health Insurance Program (CHIP). Records of service use, including both fee-for-service (FFS) claims paid by the state Medicaid agency and managed care encounters paid by a Medicaid managed care plan, are included in the TAF inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) files.
Within federal guidelines, states choose which populations and benefit categories that they cover, and as a result, their Medicaid and CHIP programs vary in the characteristics of their covered populations and in their benefit packages. Nevertheless, examining the overall percentage of beneficiaries with a claim in a given year can help identify outlier states that TAF users should examine more closely before starting their analysis. For example, if a state submits incomplete data on FFS claims or managed care encounters, TAF users could underestimate utilization, expenditures, and the prevalence of medical conditions in beneficiaries in that state. If too many beneficiaries in a state had a claim, it could signal that the states either submitted service records into the wrong file or had other data quality issues. TAF users would need to consider these issues when designing their own data quality checks and analyses.
This data quality assessment examines the percentage of beneficiaries in each state with any medical or pharmacy claims in the TAF data, as a way to identify states with potentially incomplete service use data. [1]
More information about the completeness of each state’s TAF data can be found in the DQ Atlas single topic displays for Claims Volume - IP , Claims Volume - LT , Claims Volume - OT , and Claims Volume - RX .
More information about the completeness of each state\u2019s TAF data can be found in the DQ Atlas single topic displays for Claims Volume - IP , Claims Volume - LT , Claims Volume - OT , and Claims Volume - RX .
We used the annual Demographic and Eligibility (DE) TAF and three TAF claim files (IP, OT, and RX) to examine the percentage of beneficiaries who were enrolled in Medicaid or CHIP during any month in the calendar year and had any claims in each file. [2]
We do not examine the percentage of beneficiaries who have service use in the LT file for several reasons. First, states have very different policies on providing, billing for, and recording LT services, which include facility-based long-term care provided by nursing homes, intermediate care facilities for individuals with intellectual or developmental disabilities, services provided in a mental health facility, and independent psychiatric wings of acute care hospitals. Second, although most states emphasize the use of home- and community-based services over institutional long-term care, the success of this strategy varies by state. Third, this analysis excludes dually eligible beneficiaries, which translates to excluding nearly all beneficiaries over the age of 65 and substantially limits the population in each state that we would expect to have an LT claim. As a result, small differences across states in the percentage of non-dually eligible beneficiaries who use long-term care services cannot be interpreted as a data quality issue, especially because the beneficiary population’s need for institutional long-term care could also vary from one state to the next. [3]
To identify beneficiaries enrolled in Medicaid or CHIP for at least one month in the year, we used the CHIP code variable in the annual DE file. [4] We excluded beneficiaries who were dually eligible for Medicaid and Medicare in any month of the year because Medicare is the primary payer for most medical care for these beneficiaries. [5] For each non-dually eligible beneficiary, we then looked for at least one claim in each file (IP, OT, and RX) that linked to the beneficiary’s enrollment record in the annual DE file. We examined this linkage on an annual basis instead of a monthly basis because not all beneficiaries receive care each month of the year. We included both FFS claims and managed care encounter records in this analysis, but excluded capitation payments, supplemental claims, service tracking claims, and “other” claims. [6] We also excluded crossover claims because they were for dually eligible beneficiaries whose services were partially paid by Medicare. [7]
We used header (and not line) records when linking claims to beneficiaries for this analysis. Header records include summary information about the claim as a whole, whereas line records detail the individual goods and services billed as part of the claim. In some instances, data quality issues prevent the accurate linking of header records with their corresponding lines. Header claim records with missing line-level detail are included in TAF, but line records that do not link to a header are excluded. As a result, the findings from this analysis identify states that may be missing entire claims from their claims files, but do not identify states in which claims data are partially incomplete due to missing line records.
We compared the percentage of beneficiaries in every state who had at least one claim in each file to the national median for that file to identify states that might have data quality issues. States may be outliers on this measure because of issues with data quality or completeness in a claims file. Alternatively, some states may be outliers because of data quality issues in the DE file. For instance, some states have known issues with missing or unreliable information in the CHIP code or dual code variables that would prevent accurately measuring the size of the non-dual population. [8] Finally, some states may be outliers because of data quality issues that prevent accurately linking service use records to beneficiary eligibility records.
We classified a state’s IP, OT, or RX file as unusable if the percentage of beneficiaries with service use was less than 10 percent of the national median value for the file, because that pattern suggests the data are incomplete or have major data quality problems (Table 1). We had a high level of concern about a state’s claims data file if the percentage of beneficiaries with service use was either: (1) between 10 and 50 percent of the national median, raising the possibility of an incomplete data file, or (2) more than 200 percent of the national median, raising the possibility that the state was formatting claims incorrectly or including claims records that should have been reported into a different file. Finally, we had a medium level of concern about a state’s claims data file if the percentage of beneficiaries with service use was either (1) between 50 and 75 percent, or (2) between 150 to 200 percent of the national median.
Table 1. Criteria for DQ assessment of service use data in the IP, OT, and RX files
| Beneficiaries who are service users, as a percentage of the national median value | Number of service users | DQ assessment |
|---|---|---|
| x < 10 percent | Too low | Unusable |
| 10 percent ≤ x < 50 percent | Too low | High concern |
| 50 percent ≤ x < 75 percent | Low | Medium concern |
| 75 percent ≤ x < 150 percent | Acceptable | Low concern |
| 150 percent ≤ x ≤ 200 percent | High | Medium concern |
| x > 200 percent | Too high | High concern |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Eligibility for institutional-based long-term care services that would be captured in the LT file is limited to beneficiaries who have been determined to need this level of care. Because this analysis excludes dual eligible beneficiaries, users of long-term care services are most likely to be beneficiaries who qualify for Medicaid on the basis of disability. States have varying rates of beneficiaries who qualify for Medicaid on the basis of disability; consequently, they are likely to have varying rates of non-dually eligible beneficiaries who qualify for LT services.
If it was available, we used the monthly CHIP code (CHIP_CD) to identify beneficiaries enrolled in Medicaid or CHIP that month (CHIP_CD values of 1–4). CHIP code 4 (Medicaid and S-CHIP) is a valid value for 2014 through 2017 TAF. If the CHIP code was missing, we considered the beneficiary as enrolled in the month only if the eligibility group variable (ELGBLTY_GRP_CD) was not missing.
We removed data on dually eligible beneficiaries by removing any person identified in the annual DE for whom any dual status code (DUAL_ELGBL_CD_mm, where mm refers to the month) took on one of the following values that indicate when a Medicaid beneficiary is eligible for Medicare: 01, 02, 03, 04, 05, 06, 08, 09, or 10.
We identified FFS claims and managed care encounters by using claim type code (CLM_TYPE_CD) values of 1, A, 3, and C. We excluded any records that the state classified as capitation payments, supplemental claims, service tracking claims, or “other” claims. “Other” claims represent FFS claims, capitated payments, managed care encounters, service tracking claims, and supplemental payments that states have classified as “other” rather than as Medicaid or CHIP. More information can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
Crossover claims are service use records where Medicare processes and pays the claim first as the primary payer, and Medicaid processes the claim as a secondary payer. Crossover claims are identified using the criteria XOVR_IND = 1.
Using CHIP code to identify beneficiaries enrolled in Medicaid or CHIP provides the closest estimates to external program enrollment benchmarks in the most states. However, some states have missing or unreliable information captured in the CHIP code variable, and TAF-based enrollment counts are not as accurate in those states. Some states also have problems correctly identifying dually eligible beneficiaries using the dual code.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Eligibility for institutional-based long-term care services that would be captured in the LT file is limited to beneficiaries who have been determined to need this level of care. Because this analysis excludes dual eligible beneficiaries, users of long-term care services are most likely to be beneficiaries who qualify for Medicaid on the basis of disability. States have varying rates of beneficiaries who qualify for Medicaid on the basis of disability; consequently, they are likely to have varying rates of non-dually eligible beneficiaries who qualify for LT services.
If it was available, we used the monthly CHIP code (CHIP_CD) to identify beneficiaries enrolled in Medicaid or CHIP that month (CHIP_CD values of 1\u20134). CHIP code 4 (Medicaid and S-CHIP) is a valid value for 2014 through 2017 TAF. If the CHIP code was missing, we considered the beneficiary as enrolled in the month only if the eligibility group variable (ELGBLTY_GRP_CD) was not missing.
We removed data on dually eligible beneficiaries by removing any person identified in the annual DE for whom any dual status code (DUAL_ELGBL_CD_mm, where mm refers to the month) took on one of the following values that indicate when a Medicaid beneficiary is eligible for Medicare: 01, 02, 03, 04, 05, 06, 08, 09, or 10.
We identified FFS claims and managed care encounters by using claim type code (CLM_TYPE_CD) values of 1, A, 3, and C. We excluded any records that the state classified as capitation payments, supplemental claims, service tracking claims, or \u201cother\u201d claims. \u201cOther\u201d claims represent FFS claims, capitated payments, managed care encounters, service tracking claims, and supplemental payments that states have classified as \u201cother\u201d rather than as Medicaid or CHIP. More information can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
Crossover claims are service use records where Medicare processes and pays the claim first as the primary payer, and Medicaid processes the claim as a secondary payer. Crossover claims are identified using the criteria XOVR_IND = 1.
Using CHIP code to identify beneficiaries enrolled in Medicaid or CHIP provides the closest estimates to external program enrollment benchmarks in the most states. However, some states have missing or unreliable information captured in the CHIP code variable, and TAF-based enrollment counts are not as accurate in those states. Some states also have problems correctly identifying dually eligible beneficiaries using the dual code.
Although states differ in the populations and benefits covered in their Medicaid and CHIP programs, examining the overall percentage of beneficiaries with any service use can identify outlier states that may have incomplete claims, encounter, or eligibility data in the TAF. Low rates of service use may also indicate problems in linking service use and eligibility records. This analysis examines the percentage of beneficiaries in each state with an RX record indicating that a prescription was filled during the year.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5061"", ""relatedTopics"": [{""measureId"": 27, ""measureName"": ""Service Users - IP"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 0}, {""measureId"": 28, ""measureName"": ""Service Users - OT"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 1}]}" 30,"{""measureId"": 30, ""measureName"": ""Servicing Provider NPI - OT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Servicing-Prov-NPI-OT.pdf"", ""background"": {""content"": ""Users of the T-MSIS Analytic Files (TAF) may want to identify the providers or categories of providers that deliver services to Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries. A claim in the TAF could include information for up to six providers, depending on the file type. [1] A provider could be a facility, a group of individual practitioners, or an individual practitioner. The provider types most commonly used for claims-based analyses including the billing provider, the servicing provider, the prescribing provider, and the dispensing provider.
The billing provider represents the entity that submits the claim and is reimbursed by the Medicaid or CHIP agency. Information about billing provider is available in all TAF claims files: inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX). The servicing provider, in contrast, represents the individual practitioner who was responsible for or provided direct care to the beneficiary. Information about servicing provider is available in the IP, LT, and OT files. In some cases, the servicing provider is the same as the billing provider. In other cases, the servicing and billing provider differ, such as when a hospital or large group practice bills for services rendered by individual physicians it employs. Because institutional claims in the IP and LT file represent facility costs and are often submitted by the hospitals and skilled nursing facilities where the care occurred, analyses using these claims would typically be interested in only the billing provider. On prescription drug claims, the prescribing provider is the individual practitioner who prescribed a prescription drug to a beneficiary, while the dispensing provider could represent a large pharmacy chain or an individual pharmacy that filled the prescription. Prescribing provider and dispensing provider NPIs are available only in the RX file.
For each provider type, up to two identifiers are available on the claims record: (1) the National Provider Identifier (NPI), which is the unique, 10-digit identification number that the National Plan and Provider Enumeration System (NPPES) assigns to each HIPAA-covered health care provider, and (2) the state-assigned unique identifier used in the state’s Medicaid Management Information System. [2] These identifiers can be used to obtain additional information about the provider, such as the provider’s name and address, but only if states completely and accurately report the provider identifiers on the T-MSIS claims records that are used to construct the TAF. Data users may be particularly interested in the availability of NPI on claims records, because every NPI can be linked to information about the provider’s characteristics in the publicly available NPPES NPI registry.
This data quality assessment examines the extent to which four key NPIs—the Billing Provider NPI, the Servicing Provider NPI, the Prescribing Provider NPI, and the Dispensing Provider NPI—are available on fee-for-service (FFS) claims and managed care encounters in the TAF. Provider NPIs should be available on most claims and encounters, with the exception of certain records in the OT file that represent claims submitted by “atypical” providers. [3]
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
Centers for Medicare & Medicaid Services (CMS). “CMS Guidance: Reporting Provider Identifiers in T-MSIS.” Baltimore, MD: CMS, January 2019. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=50507 . Accessed January 31, 2019.
Atypical providers do not provide “health care,” as the term is defined in HIPAA 45 C.F.R. 160.103. Among atypical providers that are reimbursed by Medicaid programs are those that offer taxi services, home and vehicle modifications, and respite services (CMS 2006). An atypical provider is not eligible to receive an NPI; therefore, we did not expect the provider NPI field to be populated on claims for these services.
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
Centers for Medicare & Medicaid Services (CMS). \u201cCMS Guidance: Reporting Provider Identifiers in T-MSIS.\u201d Baltimore, MD: CMS, January 2019. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=50507 . Accessed January 31, 2019.
Atypical providers do not provide \u201chealth care,\u201d as the term is defined in HIPAA 45 C.F.R. 160.103. Among atypical providers that are reimbursed by Medicaid programs are those that offer taxi services, home and vehicle modifications, and respite services (CMS 2006). An atypical provider is not eligible to receive an NPI; therefore, we did not expect the provider NPI field to be populated on claims for these services.
Using the 2016 TAF, [4] we examined the extent to which billing, servicing, prescribing, and dispensing provider NPIs are available in the IP, LT, OT, and RX claims files (Table 1). We examined this set of provider NPIs because they are the most relevant to the analyses we expect TAF users to conduct. Billing Provider NPI is available in all claims files and is relevant to studies for all types of care. Therefore, we assessed this provider NPI separately for each claims file. Although Servicing Provider NPIs are available in the IP and LT files, most researchers analyzing IP and LT claims will be interested primarily in the Billing Provider NPI, which identifies the facility that provided care to the beneficiary (for example, the hospital or nursing facility). In contrast, the OT file contains certain claims (for example, professional claims and facility claims from federally qualified health centers and rural health clinics) for which there may be a compelling reason to identify the servicing providers on the claim lines as well as the billing provider on the claim header. For this reason, in this brief we examine the rate of missing Servicing Provider NPIs in the OT file but not in the IP or LT files. The prescribing provider and the dispensing provider NPI are only available in the RX file.
Table 1. Provider NPIs examined in this brief
| Provider identifier | Claims file | TAF field name |
|---|---|---|
| Billing Provider NPI | IP, LT, OT, and RX header | BLG_PRVDR_NPI_NUM |
| Servicing Provider NPI | OT line | SRVCNG_PRVDR_NPI_NUM |
| Prescribing Provider NPI | RX header | SRVCNG_PRVDR_NPI_NUM |
| Dispensing Provider NPI | RX header | DSPNSNG_PD_PRVDR_NPI_NUM |
We included both Medicaid and CHIP FFS claims and managed care encounter records in our analysis. [5] For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [6]
For each state, we calculated the percentage of records that were missing each NPI. We considered any field with a null value to be missing. We also classified as missing those cases in which the entire length of the field was completely filled with repeated “0,” “8,” or “9” values. Otherwise, we did not assess the validity of the NPI information by confirming the NPI value existed in the NPPES NPI registry.
We grouped states into categories of low concern, medium concern, and high concern about the usability of their data, depending on the percentage of records that were missing each NPI in each claims file (Table 2).
Table 2. Criteria for DQ assessment of Provider NPI
| DQ assessment | |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We implemented all claims exclusions by using the claim type code (CLM_TYPE_CD). We excluded capitation payments, supplemental payments, and service tracking payments. We also excluded records with a claim type of “other,” which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Titles XIX or XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data”, on ResDAC.org.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We implemented all claims exclusions by using the claim type code (CLM_TYPE_CD). We excluded capitation payments, supplemental payments, and service tracking payments. We also excluded records with a claim type of \u201cother,\u201d which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Titles XIX or XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data\u201d, on ResDAC.org.
The servicing provider represents the individual practitioner who was responsible for or provided direct care to the beneficiary. There are two identifiers available on OT line records for the servicing provider: the National Provider Identifier (NPI) and the state-assigned unique identifier used in the state's claims processing system. This analysis examines the extent to which the servicing provider's NPI is available on claims and encounter records in the OT file.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5071"", ""relatedTopics"": [{""measureId"": 33, ""measureName"": ""Billing Provider NPI - IP"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 0}, {""measureId"": 34, ""measureName"": ""Billing Provider NPI - LT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 1}, {""measureId"": 35, ""measureName"": ""Billing Provider NPI - OT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 2}, {""measureId"": 36, ""measureName"": ""Billing Provider NPI - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 3}, {""measureId"": 31, ""measureName"": ""Prescribing Provider NPI - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 5}, {""measureId"": 32, ""measureName"": ""Dispensing Provider NPI - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 6}]}" 31,"{""measureId"": 31, ""measureName"": ""Prescribing Provider NPI - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Prescribing-Prov-NPI-RX.pdf"", ""background"": {""content"": ""Users of the T-MSIS Analytic Files (TAF) may want to identify the providers or categories of providers that deliver services to Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries. A claim in the TAF could include information for up to six providers, depending on the file type. [1] A provider could be a facility, a group of individual practitioners, or an individual practitioner. The provider types most commonly used for claims-based analyses including the billing provider, the servicing provider, the prescribing provider, and the dispensing provider.
The billing provider represents the entity that submits the claim and is reimbursed by the Medicaid or CHIP agency. Information about billing provider is available in all TAF claims files: inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX). The servicing provider, in contrast, represents the individual practitioner who was responsible for or provided direct care to the beneficiary. Information about servicing provider is available in the IP, LT, and OT files. In some cases, the servicing provider is the same as the billing provider. In other cases, the servicing and billing provider differ, such as when a hospital or large group practice bills for services rendered by individual physicians it employs. Because institutional claims in the IP and LT file represent facility costs and are often submitted by the hospitals and skilled nursing facilities where the care occurred, analyses using these claims would typically be interested in only the billing provider. On prescription drug claims, the prescribing provider is the individual practitioner who prescribed a prescription drug to a beneficiary, while the dispensing provider could represent a large pharmacy chain or an individual pharmacy that filled the prescription. Prescribing provider and dispensing provider NPIs are available only in the RX file.
For each provider type, up to two identifiers are available on the claims record: (1) the National Provider Identifier (NPI), which is the unique, 10-digit identification number that the National Plan and Provider Enumeration System (NPPES) assigns to each HIPAA-covered health care provider, and (2) the state-assigned unique identifier used in the state’s Medicaid Management Information System. [2] These identifiers can be used to obtain additional information about the provider, such as the provider’s name and address, but only if states completely and accurately report the provider identifiers on the T-MSIS claims records that are used to construct the TAF. Data users may be particularly interested in the availability of NPI on claims records, because every NPI can be linked to information about the provider’s characteristics in the publicly available NPPES NPI registry.
This data quality assessment examines the extent to which four key NPIs—the Billing Provider NPI, the Servicing Provider NPI, the Prescribing Provider NPI, and the Dispensing Provider NPI—are available on fee-for-service (FFS) claims and managed care encounters in the TAF. Provider NPIs should be available on most claims and encounters, with the exception of certain records in the OT file that represent claims submitted by “atypical” providers. [3]
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
Centers for Medicare & Medicaid Services (CMS). “CMS Guidance: Reporting Provider Identifiers in T-MSIS.” Baltimore, MD: CMS, January 2019. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=50507 . Accessed January 31, 2019.
Atypical providers do not provide “health care,” as the term is defined in HIPAA 45 C.F.R. 160.103. Among atypical providers that are reimbursed by Medicaid programs are those that offer taxi services, home and vehicle modifications, and respite services (CMS 2006). An atypical provider is not eligible to receive an NPI; therefore, we did not expect the provider NPI field to be populated on claims for these services.
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
Centers for Medicare & Medicaid Services (CMS). \u201cCMS Guidance: Reporting Provider Identifiers in T-MSIS.\u201d Baltimore, MD: CMS, January 2019. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=50507 . Accessed January 31, 2019.
Atypical providers do not provide \u201chealth care,\u201d as the term is defined in HIPAA 45 C.F.R. 160.103. Among atypical providers that are reimbursed by Medicaid programs are those that offer taxi services, home and vehicle modifications, and respite services (CMS 2006). An atypical provider is not eligible to receive an NPI; therefore, we did not expect the provider NPI field to be populated on claims for these services.
Using the 2016 TAF, [4] we examined the extent to which billing, servicing, prescribing, and dispensing provider NPIs are available in the IP, LT, OT, and RX claims files (Table 1). We examined this set of provider NPIs because they are the most relevant to the analyses we expect TAF users to conduct. Billing Provider NPI is available in all claims files and is relevant to studies for all types of care. Therefore, we assessed this provider NPI separately for each claims file. Although Servicing Provider NPIs are available in the IP and LT files, most researchers analyzing IP and LT claims will be interested primarily in the Billing Provider NPI, which identifies the facility that provided care to the beneficiary (for example, the hospital or nursing facility). In contrast, the OT file contains certain claims (for example, professional claims and facility claims from federally qualified health centers and rural health clinics) for which there may be a compelling reason to identify the servicing providers on the claim lines as well as the billing provider on the claim header. For this reason, in this brief we examine the rate of missing Servicing Provider NPIs in the OT file but not in the IP or LT files. The prescribing provider and the dispensing provider NPI are only available in the RX file.
Table 1. Provider NPIs examined in this brief
| Provider identifier | Claims file | TAF field name |
|---|---|---|
| Billing Provider NPI | IP, LT, OT, and RX header | BLG_PRVDR_NPI_NUM |
| Servicing Provider NPI | OT line | SRVCNG_PRVDR_NPI_NUM |
| Prescribing Provider NPI | RX header | SRVCNG_PRVDR_NPI_NUM |
| Dispensing Provider NPI | RX header | DSPNSNG_PD_PRVDR_NPI_NUM |
We included both Medicaid and CHIP FFS claims and managed care encounter records in our analysis. [5] For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [6]
For each state, we calculated the percentage of records that were missing each NPI. We considered any field with a null value to be missing. We also classified as missing those cases in which the entire length of the field was completely filled with repeated “0,” “8,” or “9” values. Otherwise, we did not assess the validity of the NPI information by confirming the NPI value existed in the NPPES NPI registry.
We grouped states into categories of low concern, medium concern, and high concern about the usability of their data, depending on the percentage of records that were missing each NPI in each claims file (Table 2).
Table 2. Criteria for DQ assessment of Provider NPI
| DQ assessment | |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We implemented all claims exclusions by using the claim type code (CLM_TYPE_CD). We excluded capitation payments, supplemental payments, and service tracking payments. We also excluded records with a claim type of “other,” which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Titles XIX or XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data”, on ResDAC.org.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We implemented all claims exclusions by using the claim type code (CLM_TYPE_CD). We excluded capitation payments, supplemental payments, and service tracking payments. We also excluded records with a claim type of \u201cother,\u201d which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Titles XIX or XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data\u201d, on ResDAC.org.
The prescribing provider represents the individual practitioner who prescribed a drug to a beneficiary. There are two identifiers available on RX header records for the prescribing provider: the National Provider Identifier (NPI) and the state-assigned unique identifier used in the state's claims processing system. This analysis examines the extent to which the prescribing provider's NPI is available on claims and encounter records in the RX file.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5071"", ""relatedTopics"": [{""measureId"": 33, ""measureName"": ""Billing Provider NPI - IP"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 0}, {""measureId"": 34, ""measureName"": ""Billing Provider NPI - LT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 1}, {""measureId"": 35, ""measureName"": ""Billing Provider NPI - OT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 2}, {""measureId"": 36, ""measureName"": ""Billing Provider NPI - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 3}, {""measureId"": 30, ""measureName"": ""Servicing Provider NPI - OT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 4}, {""measureId"": 32, ""measureName"": ""Dispensing Provider NPI - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 6}]}" 32,"{""measureId"": 32, ""measureName"": ""Dispensing Provider NPI - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Dispensing-Prov-NPI-RX.pdf"", ""background"": {""content"": ""Users of the T-MSIS Analytic Files (TAF) may want to identify the providers or categories of providers that deliver services to Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries. A claim in the TAF could include information for up to six providers, depending on the file type. [1] A provider could be a facility, a group of individual practitioners, or an individual practitioner. The provider types most commonly used for claims-based analyses including the billing provider, the servicing provider, the prescribing provider, and the dispensing provider.
The billing provider represents the entity that submits the claim and is reimbursed by the Medicaid or CHIP agency. Information about billing provider is available in all TAF claims files: inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX). The servicing provider, in contrast, represents the individual practitioner who was responsible for or provided direct care to the beneficiary. Information about servicing provider is available in the IP, LT, and OT files. In some cases, the servicing provider is the same as the billing provider. In other cases, the servicing and billing provider differ, such as when a hospital or large group practice bills for services rendered by individual physicians it employs. Because institutional claims in the IP and LT file represent facility costs and are often submitted by the hospitals and skilled nursing facilities where the care occurred, analyses using these claims would typically be interested in only the billing provider. On prescription drug claims, the prescribing provider is the individual practitioner who prescribed a prescription drug to a beneficiary, while the dispensing provider could represent a large pharmacy chain or an individual pharmacy that filled the prescription. Prescribing provider and dispensing provider NPIs are available only in the RX file.
For each provider type, up to two identifiers are available on the claims record: (1) the National Provider Identifier (NPI), which is the unique, 10-digit identification number that the National Plan and Provider Enumeration System (NPPES) assigns to each HIPAA-covered health care provider, and (2) the state-assigned unique identifier used in the state’s Medicaid Management Information System. [2] These identifiers can be used to obtain additional information about the provider, such as the provider’s name and address, but only if states completely and accurately report the provider identifiers on the T-MSIS claims records that are used to construct the TAF. Data users may be particularly interested in the availability of NPI on claims records, because every NPI can be linked to information about the provider’s characteristics in the publicly available NPPES NPI registry.
This data quality assessment examines the extent to which four key NPIs—the Billing Provider NPI, the Servicing Provider NPI, the Prescribing Provider NPI, and the Dispensing Provider NPI—are available on fee-for-service (FFS) claims and managed care encounters in the TAF. Provider NPIs should be available on most claims and encounters, with the exception of certain records in the OT file that represent claims submitted by “atypical” providers. [3]
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
Centers for Medicare & Medicaid Services (CMS). “CMS Guidance: Reporting Provider Identifiers in T-MSIS.” Baltimore, MD: CMS, January 2019. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=50507 . Accessed January 31, 2019.
Atypical providers do not provide “health care,” as the term is defined in HIPAA 45 C.F.R. 160.103. Among atypical providers that are reimbursed by Medicaid programs are those that offer taxi services, home and vehicle modifications, and respite services (CMS 2006). An atypical provider is not eligible to receive an NPI; therefore, we did not expect the provider NPI field to be populated on claims for these services.
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
Centers for Medicare & Medicaid Services (CMS). \u201cCMS Guidance: Reporting Provider Identifiers in T-MSIS.\u201d Baltimore, MD: CMS, January 2019. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=50507 . Accessed January 31, 2019.
Atypical providers do not provide \u201chealth care,\u201d as the term is defined in HIPAA 45 C.F.R. 160.103. Among atypical providers that are reimbursed by Medicaid programs are those that offer taxi services, home and vehicle modifications, and respite services (CMS 2006). An atypical provider is not eligible to receive an NPI; therefore, we did not expect the provider NPI field to be populated on claims for these services.
Using the 2016 TAF, [4] we examined the extent to which billing, servicing, prescribing, and dispensing provider NPIs are available in the IP, LT, OT, and RX claims files (Table 1). We examined this set of provider NPIs because they are the most relevant to the analyses we expect TAF users to conduct. Billing Provider NPI is available in all claims files and is relevant to studies for all types of care. Therefore, we assessed this provider NPI separately for each claims file. Although Servicing Provider NPIs are available in the IP and LT files, most researchers analyzing IP and LT claims will be interested primarily in the Billing Provider NPI, which identifies the facility that provided care to the beneficiary (for example, the hospital or nursing facility). In contrast, the OT file contains certain claims (for example, professional claims and facility claims from federally qualified health centers and rural health clinics) for which there may be a compelling reason to identify the servicing providers on the claim lines as well as the billing provider on the claim header. For this reason, in this brief we examine the rate of missing Servicing Provider NPIs in the OT file but not in the IP or LT files. The prescribing provider and the dispensing provider NPI are only available in the RX file.
Table 1. Provider NPIs examined in this brief
| Provider identifier | Claims file | TAF field name |
|---|---|---|
| Billing Provider NPI | IP, LT, OT, and RX header | BLG_PRVDR_NPI_NUM |
| Servicing Provider NPI | OT line | SRVCNG_PRVDR_NPI_NUM |
| Prescribing Provider NPI | RX header | SRVCNG_PRVDR_NPI_NUM |
| Dispensing Provider NPI | RX header | DSPNSNG_PD_PRVDR_NPI_NUM |
We included both Medicaid and CHIP FFS claims and managed care encounter records in our analysis. [5] For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [6]
For each state, we calculated the percentage of records that were missing each NPI. We considered any field with a null value to be missing. We also classified as missing those cases in which the entire length of the field was completely filled with repeated “0,” “8,” or “9” values. Otherwise, we did not assess the validity of the NPI information by confirming the NPI value existed in the NPPES NPI registry.
We grouped states into categories of low concern, medium concern, and high concern about the usability of their data, depending on the percentage of records that were missing each NPI in each claims file (Table 2).
Table 2. Criteria for DQ assessment of Provider NPI
| DQ assessment | |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We implemented all claims exclusions by using the claim type code (CLM_TYPE_CD). We excluded capitation payments, supplemental payments, and service tracking payments. We also excluded records with a claim type of “other,” which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Titles XIX or XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data”, on ResDAC.org.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We implemented all claims exclusions by using the claim type code (CLM_TYPE_CD). We excluded capitation payments, supplemental payments, and service tracking payments. We also excluded records with a claim type of \u201cother,\u201d which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Titles XIX or XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data\u201d, on ResDAC.org.
The dispensing provider represents the pharmacy chain or an individual pharmacy that filled a beneficiary's prescription. There are two identifiers available on RX header records for the dispensing provider: the National Provider Identifier (NPI) and the state-assigned unique identifier used in the state's claims processing system. This analysis examines the extent to which the dispensing provider's NPI is available on claims and encounter records in the RX file.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5071"", ""relatedTopics"": [{""measureId"": 33, ""measureName"": ""Billing Provider NPI - IP"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 0}, {""measureId"": 34, ""measureName"": ""Billing Provider NPI - LT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 1}, {""measureId"": 35, ""measureName"": ""Billing Provider NPI - OT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 2}, {""measureId"": 36, ""measureName"": ""Billing Provider NPI - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 3}, {""measureId"": 30, ""measureName"": ""Servicing Provider NPI - OT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 4}, {""measureId"": 31, ""measureName"": ""Prescribing Provider NPI - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 5}]}" 33,"{""measureId"": 33, ""measureName"": ""Billing Provider NPI - IP"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Billing-Prov-NPI-IP.pdf"", ""background"": {""content"": ""Users of the T-MSIS Analytic Files (TAF) may want to identify the providers or categories of providers that deliver services to Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries. A claim in the TAF could include information for up to six providers, depending on the file type. [1] A provider could be a facility, a group of individual practitioners, or an individual practitioner. The provider types most commonly used for claims-based analyses including the billing provider, the servicing provider, the prescribing provider, and the dispensing provider.
The billing provider represents the entity that submits the claim and is reimbursed by the Medicaid or CHIP agency. Information about billing provider is available in all TAF claims files: inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX). The servicing provider, in contrast, represents the individual practitioner who was responsible for or provided direct care to the beneficiary. Information about servicing provider is available in the IP, LT, and OT files. In some cases, the servicing provider is the same as the billing provider. In other cases, the servicing and billing provider differ, such as when a hospital or large group practice bills for services rendered by individual physicians it employs. Because institutional claims in the IP and LT file represent facility costs and are often submitted by the hospitals and skilled nursing facilities where the care occurred, analyses using these claims would typically be interested in only the billing provider. On prescription drug claims, the prescribing provider is the individual practitioner who prescribed a prescription drug to a beneficiary, while the dispensing provider could represent a large pharmacy chain or an individual pharmacy that filled the prescription. Prescribing provider and dispensing provider NPIs are available only in the RX file.
For each provider type, up to two identifiers are available on the claims record: (1) the National Provider Identifier (NPI), which is the unique, 10-digit identification number that the National Plan and Provider Enumeration System (NPPES) assigns to each HIPAA-covered health care provider, and (2) the state-assigned unique identifier used in the state’s Medicaid Management Information System. [2] These identifiers can be used to obtain additional information about the provider, such as the provider’s name and address, but only if states completely and accurately report the provider identifiers on the T-MSIS claims records that are used to construct the TAF. Data users may be particularly interested in the availability of NPI on claims records, because every NPI can be linked to information about the provider’s characteristics in the publicly available NPPES NPI registry.
This data quality assessment examines the extent to which four key NPIs—the Billing Provider NPI, the Servicing Provider NPI, the Prescribing Provider NPI, and the Dispensing Provider NPI—are available on fee-for-service (FFS) claims and managed care encounters in the TAF. Provider NPIs should be available on most claims and encounters, with the exception of certain records in the OT file that represent claims submitted by “atypical” providers. [3]
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
Centers for Medicare & Medicaid Services (CMS). “CMS Guidance: Reporting Provider Identifiers in T-MSIS.” Baltimore, MD: CMS, January 2019. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=50507 . Accessed January 31, 2019.
Atypical providers do not provide “health care,” as the term is defined in HIPAA 45 C.F.R. 160.103. Among atypical providers that are reimbursed by Medicaid programs are those that offer taxi services, home and vehicle modifications, and respite services (CMS 2006). An atypical provider is not eligible to receive an NPI; therefore, we did not expect the provider NPI field to be populated on claims for these services.
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
Centers for Medicare & Medicaid Services (CMS). \u201cCMS Guidance: Reporting Provider Identifiers in T-MSIS.\u201d Baltimore, MD: CMS, January 2019. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=50507 . Accessed January 31, 2019.
Atypical providers do not provide \u201chealth care,\u201d as the term is defined in HIPAA 45 C.F.R. 160.103. Among atypical providers that are reimbursed by Medicaid programs are those that offer taxi services, home and vehicle modifications, and respite services (CMS 2006). An atypical provider is not eligible to receive an NPI; therefore, we did not expect the provider NPI field to be populated on claims for these services.
Using the 2016 TAF, [4] we examined the extent to which billing, servicing, prescribing, and dispensing provider NPIs are available in the IP, LT, OT, and RX claims files (Table 1). We examined this set of provider NPIs because they are the most relevant to the analyses we expect TAF users to conduct. Billing Provider NPI is available in all claims files and is relevant to studies for all types of care. Therefore, we assessed this provider NPI separately for each claims file. Although Servicing Provider NPIs are available in the IP and LT files, most researchers analyzing IP and LT claims will be interested primarily in the Billing Provider NPI, which identifies the facility that provided care to the beneficiary (for example, the hospital or nursing facility). In contrast, the OT file contains certain claims (for example, professional claims and facility claims from federally qualified health centers and rural health clinics) for which there may be a compelling reason to identify the servicing providers on the claim lines as well as the billing provider on the claim header. For this reason, in this brief we examine the rate of missing Servicing Provider NPIs in the OT file but not in the IP or LT files. The prescribing provider and the dispensing provider NPI are only available in the RX file.
Table 1. Provider NPIs examined in this brief
| Provider identifier | Claims file | TAF field name |
|---|---|---|
| Billing Provider NPI | IP, LT, OT, and RX header | BLG_PRVDR_NPI_NUM |
| Servicing Provider NPI | OT line | SRVCNG_PRVDR_NPI_NUM |
| Prescribing Provider NPI | RX header | SRVCNG_PRVDR_NPI_NUM |
| Dispensing Provider NPI | RX header | DSPNSNG_PD_PRVDR_NPI_NUM |
We included both Medicaid and CHIP FFS claims and managed care encounter records in our analysis. [5] For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [6]
For each state, we calculated the percentage of records that were missing each NPI. We considered any field with a null value to be missing. We also classified as missing those cases in which the entire length of the field was completely filled with repeated “0,” “8,” or “9” values. Otherwise, we did not assess the validity of the NPI information by confirming the NPI value existed in the NPPES NPI registry.
We grouped states into categories of low concern, medium concern, and high concern about the usability of their data, depending on the percentage of records that were missing each NPI in each claims file (Table 2).
Table 2. Criteria for DQ assessment of Provider NPI
| DQ assessment | |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We implemented all claims exclusions by using the claim type code (CLM_TYPE_CD). We excluded capitation payments, supplemental payments, and service tracking payments. We also excluded records with a claim type of “other,” which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Titles XIX or XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data”, on ResDAC.org.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We implemented all claims exclusions by using the claim type code (CLM_TYPE_CD). We excluded capitation payments, supplemental payments, and service tracking payments. We also excluded records with a claim type of \u201cother,\u201d which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Titles XIX or XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data\u201d, on ResDAC.org.
The billing provider represents the entity that submits a claim and is reimbursed by the Medicaid or CHIP agency. There are two identifiers available on IP header records for the billing provider: the National Provider Identifier (NPI) and the state-assigned unique identifier used in the state's claims processing system. This analysis examines the extent to which the billing provider's NPI is available on claims and encounter records in the IP file.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5071"", ""relatedTopics"": [{""measureId"": 34, ""measureName"": ""Billing Provider NPI - LT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 1}, {""measureId"": 35, ""measureName"": ""Billing Provider NPI - OT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 2}, {""measureId"": 36, ""measureName"": ""Billing Provider NPI - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 3}, {""measureId"": 30, ""measureName"": ""Servicing Provider NPI - OT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 4}, {""measureId"": 31, ""measureName"": ""Prescribing Provider NPI - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 5}, {""measureId"": 32, ""measureName"": ""Dispensing Provider NPI - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 6}]}" 34,"{""measureId"": 34, ""measureName"": ""Billing Provider NPI - LT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Billing-Prov-NPI-LT.pdf"", ""background"": {""content"": ""Users of the T-MSIS Analytic Files (TAF) may want to identify the providers or categories of providers that deliver services to Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries. A claim in the TAF could include information for up to six providers, depending on the file type. [1] A provider could be a facility, a group of individual practitioners, or an individual practitioner. The provider types most commonly used for claims-based analyses including the billing provider, the servicing provider, the prescribing provider, and the dispensing provider.
The billing provider represents the entity that submits the claim and is reimbursed by the Medicaid or CHIP agency. Information about billing provider is available in all TAF claims files: inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX). The servicing provider, in contrast, represents the individual practitioner who was responsible for or provided direct care to the beneficiary. Information about servicing provider is available in the IP, LT, and OT files. In some cases, the servicing provider is the same as the billing provider. In other cases, the servicing and billing provider differ, such as when a hospital or large group practice bills for services rendered by individual physicians it employs. Because institutional claims in the IP and LT file represent facility costs and are often submitted by the hospitals and skilled nursing facilities where the care occurred, analyses using these claims would typically be interested in only the billing provider. On prescription drug claims, the prescribing provider is the individual practitioner who prescribed a prescription drug to a beneficiary, while the dispensing provider could represent a large pharmacy chain or an individual pharmacy that filled the prescription. Prescribing provider and dispensing provider NPIs are available only in the RX file.
For each provider type, up to two identifiers are available on the claims record: (1) the National Provider Identifier (NPI), which is the unique, 10-digit identification number that the National Plan and Provider Enumeration System (NPPES) assigns to each HIPAA-covered health care provider, and (2) the state-assigned unique identifier used in the state’s Medicaid Management Information System. [2] These identifiers can be used to obtain additional information about the provider, such as the provider’s name and address, but only if states completely and accurately report the provider identifiers on the T-MSIS claims records that are used to construct the TAF. Data users may be particularly interested in the availability of NPI on claims records, because every NPI can be linked to information about the provider’s characteristics in the publicly available NPPES NPI registry.
This data quality assessment examines the extent to which four key NPIs—the Billing Provider NPI, the Servicing Provider NPI, the Prescribing Provider NPI, and the Dispensing Provider NPI—are available on fee-for-service (FFS) claims and managed care encounters in the TAF. Provider NPIs should be available on most claims and encounters, with the exception of certain records in the OT file that represent claims submitted by “atypical” providers. [3]
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
Centers for Medicare & Medicaid Services (CMS). “CMS Guidance: Reporting Provider Identifiers in T-MSIS.” Baltimore, MD: CMS, January 2019. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=50507 . Accessed January 31, 2019.
Atypical providers do not provide “health care,” as the term is defined in HIPAA 45 C.F.R. 160.103. Among atypical providers that are reimbursed by Medicaid programs are those that offer taxi services, home and vehicle modifications, and respite services (CMS 2006). An atypical provider is not eligible to receive an NPI; therefore, we did not expect the provider NPI field to be populated on claims for these services.
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
Centers for Medicare & Medicaid Services (CMS). \u201cCMS Guidance: Reporting Provider Identifiers in T-MSIS.\u201d Baltimore, MD: CMS, January 2019. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=50507 . Accessed January 31, 2019.
Atypical providers do not provide \u201chealth care,\u201d as the term is defined in HIPAA 45 C.F.R. 160.103. Among atypical providers that are reimbursed by Medicaid programs are those that offer taxi services, home and vehicle modifications, and respite services (CMS 2006). An atypical provider is not eligible to receive an NPI; therefore, we did not expect the provider NPI field to be populated on claims for these services.
Using the 2016 TAF, [4] we examined the extent to which billing, servicing, prescribing, and dispensing provider NPIs are available in the IP, LT, OT, and RX claims files (Table 1). We examined this set of provider NPIs because they are the most relevant to the analyses we expect TAF users to conduct. Billing Provider NPI is available in all claims files and is relevant to studies for all types of care. Therefore, we assessed this provider NPI separately for each claims file. Although Servicing Provider NPIs are available in the IP and LT files, most researchers analyzing IP and LT claims will be interested primarily in the Billing Provider NPI, which identifies the facility that provided care to the beneficiary (for example, the hospital or nursing facility). In contrast, the OT file contains certain claims (for example, professional claims and facility claims from federally qualified health centers and rural health clinics) for which there may be a compelling reason to identify the servicing providers on the claim lines as well as the billing provider on the claim header. For this reason, in this brief we examine the rate of missing Servicing Provider NPIs in the OT file but not in the IP or LT files. The prescribing provider and the dispensing provider NPI are only available in the RX file.
Table 1. Provider NPIs examined in this brief
| Provider identifier | Claims file | TAF field name |
|---|---|---|
| Billing Provider NPI | IP, LT, OT, and RX header | BLG_PRVDR_NPI_NUM |
| Servicing Provider NPI | OT line | SRVCNG_PRVDR_NPI_NUM |
| Prescribing Provider NPI | RX header | SRVCNG_PRVDR_NPI_NUM |
| Dispensing Provider NPI | RX header | DSPNSNG_PD_PRVDR_NPI_NUM |
We included both Medicaid and CHIP FFS claims and managed care encounter records in our analysis. [5] For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [6]
For each state, we calculated the percentage of records that were missing each NPI. We considered any field with a null value to be missing. We also classified as missing those cases in which the entire length of the field was completely filled with repeated “0,” “8,” or “9” values. Otherwise, we did not assess the validity of the NPI information by confirming the NPI value existed in the NPPES NPI registry.
We grouped states into categories of low concern, medium concern, and high concern about the usability of their data, depending on the percentage of records that were missing each NPI in each claims file (Table 2).
Table 2. Criteria for DQ assessment of Provider NPI
| DQ assessment | |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We implemented all claims exclusions by using the claim type code (CLM_TYPE_CD). We excluded capitation payments, supplemental payments, and service tracking payments. We also excluded records with a claim type of “other,” which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Titles XIX or XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data”, on ResDAC.org.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We implemented all claims exclusions by using the claim type code (CLM_TYPE_CD). We excluded capitation payments, supplemental payments, and service tracking payments. We also excluded records with a claim type of \u201cother,\u201d which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Titles XIX or XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data\u201d, on ResDAC.org.
The billing provider represents the entity that submits a claim and is reimbursed by the Medicaid or CHIP agency. There are two identifiers available on LT header records for the billing provider: the National Provider Identifier (NPI) and the state-assigned unique identifier used in the state's claims processing system. This analysis examines the extent to which the billing provider's NPI is available on claims and encounter records in the LT file.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5071"", ""relatedTopics"": [{""measureId"": 33, ""measureName"": ""Billing Provider NPI - IP"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 0}, {""measureId"": 35, ""measureName"": ""Billing Provider NPI - OT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 2}, {""measureId"": 36, ""measureName"": ""Billing Provider NPI - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 3}, {""measureId"": 30, ""measureName"": ""Servicing Provider NPI - OT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 4}, {""measureId"": 31, ""measureName"": ""Prescribing Provider NPI - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 5}, {""measureId"": 32, ""measureName"": ""Dispensing Provider NPI - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 6}]}" 35,"{""measureId"": 35, ""measureName"": ""Billing Provider NPI - OT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Billing-Prov-NPI-OT.pdf"", ""background"": {""content"": ""Users of the T-MSIS Analytic Files (TAF) may want to identify the providers or categories of providers that deliver services to Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries. A claim in the TAF could include information for up to six providers, depending on the file type. [1] A provider could be a facility, a group of individual practitioners, or an individual practitioner. The provider types most commonly used for claims-based analyses including the billing provider, the servicing provider, the prescribing provider, and the dispensing provider.
The billing provider represents the entity that submits the claim and is reimbursed by the Medicaid or CHIP agency. Information about billing provider is available in all TAF claims files: inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX). The servicing provider, in contrast, represents the individual practitioner who was responsible for or provided direct care to the beneficiary. Information about servicing provider is available in the IP, LT, and OT files. In some cases, the servicing provider is the same as the billing provider. In other cases, the servicing and billing provider differ, such as when a hospital or large group practice bills for services rendered by individual physicians it employs. Because institutional claims in the IP and LT file represent facility costs and are often submitted by the hospitals and skilled nursing facilities where the care occurred, analyses using these claims would typically be interested in only the billing provider. On prescription drug claims, the prescribing provider is the individual practitioner who prescribed a prescription drug to a beneficiary, while the dispensing provider could represent a large pharmacy chain or an individual pharmacy that filled the prescription. Prescribing provider and dispensing provider NPIs are available only in the RX file.
For each provider type, up to two identifiers are available on the claims record: (1) the National Provider Identifier (NPI), which is the unique, 10-digit identification number that the National Plan and Provider Enumeration System (NPPES) assigns to each HIPAA-covered health care provider, and (2) the state-assigned unique identifier used in the state’s Medicaid Management Information System. [2] These identifiers can be used to obtain additional information about the provider, such as the provider’s name and address, but only if states completely and accurately report the provider identifiers on the T-MSIS claims records that are used to construct the TAF. Data users may be particularly interested in the availability of NPI on claims records, because every NPI can be linked to information about the provider’s characteristics in the publicly available NPPES NPI registry.
This data quality assessment examines the extent to which four key NPIs—the Billing Provider NPI, the Servicing Provider NPI, the Prescribing Provider NPI, and the Dispensing Provider NPI—are available on fee-for-service (FFS) claims and managed care encounters in the TAF. Provider NPIs should be available on most claims and encounters, with the exception of certain records in the OT file that represent claims submitted by “atypical” providers. [3]
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
Centers for Medicare & Medicaid Services (CMS). “CMS Guidance: Reporting Provider Identifiers in T-MSIS.” Baltimore, MD: CMS, January 2019. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=50507 . Accessed January 31, 2019.
Atypical providers do not provide “health care,” as the term is defined in HIPAA 45 C.F.R. 160.103. Among atypical providers that are reimbursed by Medicaid programs are those that offer taxi services, home and vehicle modifications, and respite services (CMS 2006). An atypical provider is not eligible to receive an NPI; therefore, we did not expect the provider NPI field to be populated on claims for these services.
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
Centers for Medicare & Medicaid Services (CMS). \u201cCMS Guidance: Reporting Provider Identifiers in T-MSIS.\u201d Baltimore, MD: CMS, January 2019. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=50507 . Accessed January 31, 2019.
Atypical providers do not provide \u201chealth care,\u201d as the term is defined in HIPAA 45 C.F.R. 160.103. Among atypical providers that are reimbursed by Medicaid programs are those that offer taxi services, home and vehicle modifications, and respite services (CMS 2006). An atypical provider is not eligible to receive an NPI; therefore, we did not expect the provider NPI field to be populated on claims for these services.
Using the 2016 TAF, [4] we examined the extent to which billing, servicing, prescribing, and dispensing provider NPIs are available in the IP, LT, OT, and RX claims files (Table 1). We examined this set of provider NPIs because they are the most relevant to the analyses we expect TAF users to conduct. Billing Provider NPI is available in all claims files and is relevant to studies for all types of care. Therefore, we assessed this provider NPI separately for each claims file. Although Servicing Provider NPIs are available in the IP and LT files, most researchers analyzing IP and LT claims will be interested primarily in the Billing Provider NPI, which identifies the facility that provided care to the beneficiary (for example, the hospital or nursing facility). In contrast, the OT file contains certain claims (for example, professional claims and facility claims from federally qualified health centers and rural health clinics) for which there may be a compelling reason to identify the servicing providers on the claim lines as well as the billing provider on the claim header. For this reason, in this brief we examine the rate of missing Servicing Provider NPIs in the OT file but not in the IP or LT files. The prescribing provider and the dispensing provider NPI are only available in the RX file.
Table 1. Provider NPIs examined in this brief
| Provider identifier | Claims file | TAF field name |
|---|---|---|
| Billing Provider NPI | IP, LT, OT, and RX header | BLG_PRVDR_NPI_NUM |
| Servicing Provider NPI | OT line | SRVCNG_PRVDR_NPI_NUM |
| Prescribing Provider NPI | RX header | SRVCNG_PRVDR_NPI_NUM |
| Dispensing Provider NPI | RX header | DSPNSNG_PD_PRVDR_NPI_NUM |
We included both Medicaid and CHIP FFS claims and managed care encounter records in our analysis. [5] For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [6]
For each state, we calculated the percentage of records that were missing each NPI. We considered any field with a null value to be missing. We also classified as missing those cases in which the entire length of the field was completely filled with repeated “0,” “8,” or “9” values. Otherwise, we did not assess the validity of the NPI information by confirming the NPI value existed in the NPPES NPI registry.
We grouped states into categories of low concern, medium concern, and high concern about the usability of their data, depending on the percentage of records that were missing each NPI in each claims file (Table 2).
Table 2. Criteria for DQ assessment of Provider NPI
| DQ assessment | |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We implemented all claims exclusions by using the claim type code (CLM_TYPE_CD). We excluded capitation payments, supplemental payments, and service tracking payments. We also excluded records with a claim type of “other,” which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Titles XIX or XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data”, on ResDAC.org.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We implemented all claims exclusions by using the claim type code (CLM_TYPE_CD). We excluded capitation payments, supplemental payments, and service tracking payments. We also excluded records with a claim type of \u201cother,\u201d which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Titles XIX or XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data\u201d, on ResDAC.org.
The billing provider represents the entity that submits a claim and is reimbursed by the Medicaid or CHIP agency. There are two identifiers available on OT header records for the billing provider: the National Provider Identifier (NPI) and the state-assigned unique identifier used in the state's claims processing system. This analysis examines the extent to which the billing provider's NPI is available on claims and encounter records in the OT file.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5071"", ""relatedTopics"": [{""measureId"": 33, ""measureName"": ""Billing Provider NPI - IP"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 0}, {""measureId"": 34, ""measureName"": ""Billing Provider NPI - LT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 1}, {""measureId"": 36, ""measureName"": ""Billing Provider NPI - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 3}, {""measureId"": 30, ""measureName"": ""Servicing Provider NPI - OT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 4}, {""measureId"": 31, ""measureName"": ""Prescribing Provider NPI - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 5}, {""measureId"": 32, ""measureName"": ""Dispensing Provider NPI - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 6}]}" 36,"{""measureId"": 36, ""measureName"": ""Billing Provider NPI - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Billing-Prov-NPI-RX.pdf"", ""background"": {""content"": ""Users of the T-MSIS Analytic Files (TAF) may want to identify the providers or categories of providers that deliver services to Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries. A claim in the TAF could include information for up to six providers, depending on the file type. [1] A provider could be a facility, a group of individual practitioners, or an individual practitioner. The provider types most commonly used for claims-based analyses including the billing provider, the servicing provider, the prescribing provider, and the dispensing provider.
The billing provider represents the entity that submits the claim and is reimbursed by the Medicaid or CHIP agency. Information about billing provider is available in all TAF claims files: inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX). The servicing provider, in contrast, represents the individual practitioner who was responsible for or provided direct care to the beneficiary. Information about servicing provider is available in the IP, LT, and OT files. In some cases, the servicing provider is the same as the billing provider. In other cases, the servicing and billing provider differ, such as when a hospital or large group practice bills for services rendered by individual physicians it employs. Because institutional claims in the IP and LT file represent facility costs and are often submitted by the hospitals and skilled nursing facilities where the care occurred, analyses using these claims would typically be interested in only the billing provider. On prescription drug claims, the prescribing provider is the individual practitioner who prescribed a prescription drug to a beneficiary, while the dispensing provider could represent a large pharmacy chain or an individual pharmacy that filled the prescription. Prescribing provider and dispensing provider NPIs are available only in the RX file.
For each provider type, up to two identifiers are available on the claims record: (1) the National Provider Identifier (NPI), which is the unique, 10-digit identification number that the National Plan and Provider Enumeration System (NPPES) assigns to each HIPAA-covered health care provider, and (2) the state-assigned unique identifier used in the state’s Medicaid Management Information System. [2] These identifiers can be used to obtain additional information about the provider, such as the provider’s name and address, but only if states completely and accurately report the provider identifiers on the T-MSIS claims records that are used to construct the TAF. Data users may be particularly interested in the availability of NPI on claims records, because every NPI can be linked to information about the provider’s characteristics in the publicly available NPPES NPI registry.
This data quality assessment examines the extent to which four key NPIs—the Billing Provider NPI, the Servicing Provider NPI, the Prescribing Provider NPI, and the Dispensing Provider NPI—are available on fee-for-service (FFS) claims and managed care encounters in the TAF. Provider NPIs should be available on most claims and encounters, with the exception of certain records in the OT file that represent claims submitted by “atypical” providers. [3]
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
Centers for Medicare & Medicaid Services (CMS). “CMS Guidance: Reporting Provider Identifiers in T-MSIS.” Baltimore, MD: CMS, January 2019. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=50507 . Accessed January 31, 2019.
Atypical providers do not provide “health care,” as the term is defined in HIPAA 45 C.F.R. 160.103. Among atypical providers that are reimbursed by Medicaid programs are those that offer taxi services, home and vehicle modifications, and respite services (CMS 2006). An atypical provider is not eligible to receive an NPI; therefore, we did not expect the provider NPI field to be populated on claims for these services.
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
Centers for Medicare & Medicaid Services (CMS). \u201cCMS Guidance: Reporting Provider Identifiers in T-MSIS.\u201d Baltimore, MD: CMS, January 2019. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=50507 . Accessed January 31, 2019.
Atypical providers do not provide \u201chealth care,\u201d as the term is defined in HIPAA 45 C.F.R. 160.103. Among atypical providers that are reimbursed by Medicaid programs are those that offer taxi services, home and vehicle modifications, and respite services (CMS 2006). An atypical provider is not eligible to receive an NPI; therefore, we did not expect the provider NPI field to be populated on claims for these services.
Using the 2016 TAF, [4] we examined the extent to which billing, servicing, prescribing, and dispensing provider NPIs are available in the IP, LT, OT, and RX claims files (Table 1). We examined this set of provider NPIs because they are the most relevant to the analyses we expect TAF users to conduct. Billing Provider NPI is available in all claims files and is relevant to studies for all types of care. Therefore, we assessed this provider NPI separately for each claims file. Although Servicing Provider NPIs are available in the IP and LT files, most researchers analyzing IP and LT claims will be interested primarily in the Billing Provider NPI, which identifies the facility that provided care to the beneficiary (for example, the hospital or nursing facility). In contrast, the OT file contains certain claims (for example, professional claims and facility claims from federally qualified health centers and rural health clinics) for which there may be a compelling reason to identify the servicing providers on the claim lines as well as the billing provider on the claim header. For this reason, in this brief we examine the rate of missing Servicing Provider NPIs in the OT file but not in the IP or LT files. The prescribing provider and the dispensing provider NPI are only available in the RX file.
Table 1. Provider NPIs examined in this brief
| Provider identifier | Claims file | TAF field name |
|---|---|---|
| Billing Provider NPI | IP, LT, OT, and RX header | BLG_PRVDR_NPI_NUM |
| Servicing Provider NPI | OT line | SRVCNG_PRVDR_NPI_NUM |
| Prescribing Provider NPI | RX header | SRVCNG_PRVDR_NPI_NUM |
| Dispensing Provider NPI | RX header | DSPNSNG_PD_PRVDR_NPI_NUM |
We included both Medicaid and CHIP FFS claims and managed care encounter records in our analysis. [5] For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [6]
For each state, we calculated the percentage of records that were missing each NPI. We considered any field with a null value to be missing. We also classified as missing those cases in which the entire length of the field was completely filled with repeated “0,” “8,” or “9” values. Otherwise, we did not assess the validity of the NPI information by confirming the NPI value existed in the NPPES NPI registry.
We grouped states into categories of low concern, medium concern, and high concern about the usability of their data, depending on the percentage of records that were missing each NPI in each claims file (Table 2).
Table 2. Criteria for DQ assessment of Provider NPI
| DQ assessment | |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We implemented all claims exclusions by using the claim type code (CLM_TYPE_CD). We excluded capitation payments, supplemental payments, and service tracking payments. We also excluded records with a claim type of “other,” which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Titles XIX or XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data”, on ResDAC.org.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We implemented all claims exclusions by using the claim type code (CLM_TYPE_CD). We excluded capitation payments, supplemental payments, and service tracking payments. We also excluded records with a claim type of \u201cother,\u201d which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Titles XIX or XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data\u201d, on ResDAC.org.
The billing provider represents the entity that submits a claim and is reimbursed by the Medicaid or CHIP agency. There are two identifiers available on RX header records for the billing provider: the National Provider Identifier (NPI) and the state-assigned unique identifier used in the state's claims processing system. This analysis examines the extent to which the billing provider NPI is available on claims and encounter records in the RX file.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5071"", ""relatedTopics"": [{""measureId"": 33, ""measureName"": ""Billing Provider NPI - IP"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 0}, {""measureId"": 34, ""measureName"": ""Billing Provider NPI - LT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 1}, {""measureId"": 35, ""measureName"": ""Billing Provider NPI - OT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 2}, {""measureId"": 30, ""measureName"": ""Servicing Provider NPI - OT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 4}, {""measureId"": 31, ""measureName"": ""Prescribing Provider NPI - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 5}, {""measureId"": 32, ""measureName"": ""Dispensing Provider NPI - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 6}]}" 37,"{""measureId"": 37, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - IP"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Billing-Prov-Type-IP.pdf"", ""background"": {""content"": ""Users of the T-MSIS Analytic Files (TAF) may want to identify the providers or categories of providers that deliver services to Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries. A claim in the TAF could include information for up to six providers, depending on the file type. [1] A provider could be a facility, a group of individual practitioners, or an individual practitioner. The provider types most often used for claims-based analyses include the billing, servicing, prescribing, and dispensing providers.
The billing provider represents the entity that submits the claim and is reimbursed by the Medicaid or CHIP agency. Information about the billing provider is available in all TAF claims files: inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX).
The servicing provider, in contrast, represents the individual practitioner who was responsible for or provided direct care to the beneficiary (that is, the rendering provider). Information about servicing provider is available in the IP, LT, and OT files. In some cases, the servicing provider is the same as the billing provider. In other cases, the servicing and billing provider differ, such as when a hospital or large group practice bills for services but an individual physician employed by the hospital or group provides the direct care and is the serving provider. Because institutional claims in the IP and LT files represent facility costs and are often submitted by the hospitals and skilled nursing facilities where the care occurred, analyses using these claims would typically only use the billing provider information.
For prescription drug claims, the billing provider is the pharmacy where the prescription was filled, the prescribing provider is the individual practitioner who prescribed a prescription drug to a beneficiary, and the dispensing provider is the pharmacist who filled the prescription or was responsible for overseeing the filling of the prescription. Because the billing provider represents the entity that submits a claim and is reimbursed by the Medicaid or CHIP agency, analyses using prescription drug claims typically focus on the billing provider information.
Data elements related to the billing provider can offer insight into the characteristics of providers who receive payment for Medicaid- and CHIP-funded services. Likewise, data elements related to the servicing provider can offer insight into the characteristics of the individual practitioner who was responsible for or provided direct care to a Medicaid or CHIP beneficiary. There are multiple systems available in TAF claim records for classifying providers, including the following:
Each of these classification systems may be best suited to different types of analyses. Although states are encouraged to populate all these fields on all claims, only one classification type is required. In practice, some states only submit information related to some of the classification systems, which may require TAF users to adjust their methodology across states. If the preferred data element has high rates of missingness, TAF users may be able to link the claims record to the provider’s record in the Annual Provider File (APR) to obtain the needed information. [2]
This data quality assessment examines whether any information on billing or servicing provider characteristics is available in claims records, and which type (taxonomy, specialty, or provider type). This information can help users design their analysis by selecting the fields populated in each state or identify states for which it may be necessary to link to the APR to obtain complete information about providers, rather than rely on claims alone.
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
States are required to submit information about all providers eligible to provide Medicaid- and CHIP-funded services; these records are captured in the APR. It is possible that missing information in the claim about provider taxonomy, specialty, or type could be imputed by linking to the provider’s record in the APR and obtaining the information from that file. In addition, the APR includes information on how states classify providers by “authorized category of service,” a classification scheme that can be used to identify certain non-medical provider categories not captured in the other classification systems, such as transportation or personal care service providers. For more information on the completeness of these fields in each state’s APR file, see the DQ Atlas single-topic displays for Group and Individual Providers - Classification Types and Facilities - Classification Types .
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
States are required to submit information about all providers eligible to provide Medicaid- and CHIP-funded services; these records are captured in the APR. It is possible that missing information in the claim about provider taxonomy, specialty, or type could be imputed by linking to the provider\u2019s record in the APR and obtaining the information from that file. In addition, the APR includes information on how states classify providers by \u201cauthorized category of service,\u201d a classification scheme that can be used to identify certain non-medical provider categories not captured in the other classification systems, such as transportation or personal care service providers. For more information on the completeness of these fields in each state\u2019s APR file, see the DQ Atlas single-topic displays for Group and Individual Providers - Classification Types and Facilities - Classification Types .
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We examined the values for provider type, specialty, and taxonomy codes for the providers most examined in analyses for the TAF IP, LT, OT, and RX files (Table 1). [3] We included both fee-for-service (FFS) claims and managed care encounter records, which represent claims paid by managed care organizations and should generally follow the same reporting standards for billing and servicing providers. We excluded financial transaction records, supplemental payments, and “other” records that the state did not classify as being covered by either the Medicaid or CHIP programs. [4] We also excluded states from the analysis of each file if the number of header records in the TAF was low enough to be unusable. [5] For Illinois, we restricted our analysis to the original version of the claim and excluded all subsequent adjustment records in the state’s TAF data. [6]
Table 1. Crosswalk between file type, data element description, and TAF variable name
| File type | Data element description | Variable name |
|---|---|---|
| IP, LT, OT | Billing provider | BLG_PRVDR_TYPE_CD |
| IP, LT, OT, RX | Billing provider specialty code | BLG_PRVDR_SPCLTY_CD |
| IP, LT, OT, RX | Billing provider state-reported taxonomy code | BLG_PRVDR_TXNMY_CD |
| IP, LT, OT | Billing provider NPPES primary taxonomy code | BLG_PRVDR_NPPES_TXNMY_CD |
| OT | Servicing provider type code | SRVCNG_PRVDR_TYPE_CD |
| OT | Servicing provider specialty code | SRVCNG_PRVDR_SPCLTY_CD |
| OT | Servicing provider state-reported taxonomy code | SRVCNG_PRVDR_TXNMY_CD |
| OT | Servicing provider NPPES primary taxonomy code | SRVCNG_PRVDR_NPPES_TXNMY_CD |
The provider type code includes 57 different valid values, covering both facilities and professionals. As shown in Table 2, we grouped these values into those that are expected, unexpected, and unusable for billing providers for each of the three file types in which this data element is present (IP, LT, OT) and for servicing providers in the OT file.
In the IP file, we would expect to see only general hospitals and Indian Health Service facilities as billing providers. In the LT file, we would expect to see nursing facilities, intermediate care facilities for individuals with intellectual or developmental disabilities, and psychiatric facilities as billing providers, as well as general hospitals in selected cases. [7] In the OT file, we would expect to see a broad range of both facilities and professionals billing for and directly providing services.
For all three files, we examined the “all other” code separately from the expected and unexpected categories. In some instances, use of the non-specific “all other” code may reflect the possibility that the provider is not represented in the list of valid provider type codes, which is the case for certain types of home- and community-based services (HCBS) and non-emergency medical transport providers. Although this code is valid and not considered an unexpected value in any file, it does not provide any information about the billing or servicing provider and thus renders the data element unusable for analysis.
Table 2. Classification of provider type codes, by file and provider field
| File and provider field | Expected codes | Unusable codes (“all other” value) | Unexpected codes |
|---|---|---|---|
| IP—billing provider | 42, 51 | 57 | All other values |
| LT—billing provider | 42, 43, 44, 45 | 57 | All other values |
| OT—billing and servicing provider | 01−56 | 57 | None |
Source: A full list of values is available in the TAF Claims Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries . The provider type code values are identical across all three files.
The provider specialty code includes 115 different valid values, covering both facilities and professionals. As shown in Table 3, we grouped these values into those that are expected and unexpected for each of the file types and provider fields. In the IP file, we would expect to see only general hospitals as billing providers. In the LT file, we would expect to see nursing facilities, intermediate care facilities for individuals with intellectual or developmental disabilities, and psychiatric facilities as billing providers, as well as general hospitals in selected cases. In the OT file, we would expect to see a broad range of both facilities and professionals billing for and directly providing services. In the RX file, we would expect to see pharmacies, medical supply companies, department stores, and grocery stores as billing providers. For all files and provider fields, we examined the “all other” codes (billing provider specialty code 87—All Other Suppliers) separately from the expected and unexpected categories. Although this code is valid and not considered an unexpected value in any file, it does not provide any information about the provider and thus renders the data element unusable for analysis.
Table 3. Classification of provider specialty codes, by file and provider field
| File and provider field | Expected codes | Unusable codes (“all other” value) | Unexpected codes |
|---|---|---|---|
| IP—billing provider | A0 | 87 | All other values except 88 and 99, which are treated as missing |
| LT—billing provider | A0, A1, A2, A3, B4 | 87 | All other values except 88 and 99, which are treated as missing |
| OT—billing and servicing provider | 1−86, 89−98, A0−B5 | 87 | All other values except 88 and 99, which are treated as missing |
| RX—billing provider | 51–54, 58, 73, A5, A6, A9, B1, B3, B4 | 87 | All other values except 88 and 99, which are treated as missing |
Source: A full list of values is available in the TAF Claims Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries . The provider specialty code values are identical across all claims files.
The state-reported billing provider taxonomy code is a 10-character alphanumeric code that identifies the provider’s area of specialization. As shown in Table 4, we grouped these values into those that are expected and unexpected for each of the file types. In the IP file, we would expect to see inpatient hospitals as billing providers. In the LT files we would expect to see nursing facilities, intermediate care facility services for individuals with intellectual disabilities, mental health facility services, and independent (free-standing) psychiatric wings of acute care hospitals as billing providers. In the OT file, we would expect to see a broad range of both facilities and professionals billing for and directly providing services. In the RX file, we would expect to see pharmacies and other suppliers as billing providers.
In addition to the state-reported provider taxonomy codes, IP, OT and LT files produced in 2022 and later years also include a constructed taxonomy variable for billing and servicing providers, the NPPES primary taxonomy code (*_NPPES_TXNMY_CD), that represents the provider’s primary taxonomy information as reported in the National Plan and Provider Enumeration System (NPPES) based on the provider’s National Provider Identifier. We used the same criteria as the state-reported taxonomy codes to determine expected and unexpected values for each file type.
Table 4. Classification of provider taxonomy codes, by file and provider field
| File and provider field | Expected codes | Unexpected codes |
|---|---|---|
| IP—billing provider | First 2 digits of taxonomy code are 27 or 28 | All other values |
| LT—billing provider | Taxonomy code begins with 283Q, 283X, 282E, 31, 32, 385H, or 281P | All other values |
| OT—billing and servicing provider | All valid taxonomy codes | None |
| RX—billing provider | First 2 digits of taxonomy code are 33, first 3 digits are 183, or 302R00000X [8] | All other values |
Source: The health care provider taxonomy code set can be found at http://www.wpc-edi.com/reference/codelists/healthcare/health-care-provider-taxonomy-code-set/ . The provider taxonomy code values are identical across all claims files.
We grouped states into categories of low, medium, and high concern about the usability of their data, based on the percentage of header records (or for servicing providers in the OT file, claim lines) that had an expected provider type, specialty, state-reported taxonomy code, or NPPES primary taxonomy code for the file and provider fields (Table 5).
Because users of the OT file may be interested in professional or institutional (facility) claims, we conducted the DQ assessment separately for each of these claim types. Table 6 shows the methodology used to classify each claim header as a professional or institutional claim. Any claim that did not meet at least one of the criteria for a professional or institutional claim was considered unclassified and is not included in this analysis. Note that the servicing provider analyses were conducted at the claim line level, but the claims were classified as professional or institutional claims at the claim header level.
Table 5. Criteria for DQ assessment of provider type, specialty, and taxonomy code
| Percentage of records with an expected provider type, specialty, state-reported taxonomy code, or NPPES primary taxonomy code | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
Table 6. Classification of OT claim headers as professional or facility claims
| Claim category | Identification rules |
|---|---|
| Professional | The record must meet at least one of the following criteria:
|
| Institutional | The record must meet at least one of the following criteria:
|
Methods previously used to assess data quality
Table 7 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 7. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
|
|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We used claim type code (CLM_TYPE_CD) to determine which records to include and exclude in our analysis. Records included are those with a claim type code indicating FFS claims (values of 1 and A) and managed care encounters (3 and C). We excluded capitated payments (2 and B), supplemental payments (5 and E), and service-tracking claims (4 and D). We also used claim type code to exclude “other” records (values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information on supplemental payments and other claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
More information about the completeness of each state’s TAF data can be found in the DQ Atlas single topic displays for Claims Volume—IP , Claims Volume—LT , Claims Volume—OT , and Claims Volume—RX .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This approach means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only those records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
General hospitals may provide sub-acute care through “swing beds” or other non-acute units within the hospital, and we would expect these claims to be found in the LT file rather than the IP file.
At least one state (Florida) allows Health Maintenance Organizations (HMOs) to dispense pharmaceuticals to enrollees. Therefore, HMOs are included as an “expected” taxonomy for RX billing providers.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We used claim type code (CLM_TYPE_CD) to determine which records to include and exclude in our analysis. Records included are those with a claim type code indicating FFS claims (values of 1 and A) and managed care encounters (3 and C). We excluded capitated payments (2 and B), supplemental payments (5 and E), and service-tracking claims (4 and D). We also used claim type code to exclude \u201cother\u201d records (values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information on supplemental payments and other claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
More information about the completeness of each state\u2019s TAF data can be found in the DQ Atlas single topic displays for Claims Volume\u2014IP , Claims Volume\u2014LT , Claims Volume\u2014OT , and Claims Volume\u2014RX .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This approach means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only those records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
General hospitals may provide sub-acute care through \u201cswing beds\u201d or other non-acute units within the hospital, and we would expect these claims to be found in the LT file rather than the IP file.
At least one state (Florida) allows Health Maintenance Organizations (HMOs) to dispense pharmaceuticals to enrollees. Therefore, HMOs are included as an \u201cexpected\u201d taxonomy for RX billing providers.
The billing provider type, specialty, and taxonomy codes allow TAF users to examine the characteristics of providers who bill and receive payments for Medicaid- and CHIP-funded services. This analysis examines the extent to which records in the IP file have a billing provider type, specialty, or taxonomy missing or coded with unexpected or unusable values.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5081"", ""relatedTopics"": [{""measureId"": 38, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - LT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 1}, {""measureId"": 104, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - OT Professional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 2}, {""measureId"": 105, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - OT Institutional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 3}, {""measureId"": 103, ""measureName"": ""Billing Provider Specialty and Taxonomy - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 4}, {""measureId"": 106, ""measureName"": ""Servicing Provider Type, Specialty, and Taxonomy - OT Professional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 5}, {""measureId"": 107, ""measureName"": ""Servicing Provider Type, Specialty, and Taxonomy - OT Institutional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 6}]}" 38,"{""measureId"": 38, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - LT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Billing-Prov-Type-LT.pdf"", ""background"": {""content"": ""Users of the T-MSIS Analytic Files (TAF) may want to identify the providers or categories of providers that deliver services to Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries. A claim in the TAF could include information for up to six providers, depending on the file type. [1] A provider could be a facility, a group of individual practitioners, or an individual practitioner. The provider types most often used for claims-based analyses include the billing, servicing, prescribing, and dispensing providers.
The billing provider represents the entity that submits the claim and is reimbursed by the Medicaid or CHIP agency. Information about the billing provider is available in all TAF claims files: inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX).
The servicing provider, in contrast, represents the individual practitioner who was responsible for or provided direct care to the beneficiary (that is, the rendering provider). Information about servicing provider is available in the IP, LT, and OT files. In some cases, the servicing provider is the same as the billing provider. In other cases, the servicing and billing provider differ, such as when a hospital or large group practice bills for services but an individual physician employed by the hospital or group provides the direct care and is the serving provider. Because institutional claims in the IP and LT files represent facility costs and are often submitted by the hospitals and skilled nursing facilities where the care occurred, analyses using these claims would typically only use the billing provider information.
For prescription drug claims, the billing provider is the pharmacy where the prescription was filled, the prescribing provider is the individual practitioner who prescribed a prescription drug to a beneficiary, and the dispensing provider is the pharmacist who filled the prescription or was responsible for overseeing the filling of the prescription. Because the billing provider represents the entity that submits a claim and is reimbursed by the Medicaid or CHIP agency, analyses using prescription drug claims typically focus on the billing provider information.
Data elements related to the billing provider can offer insight into the characteristics of providers who receive payment for Medicaid- and CHIP-funded services. Likewise, data elements related to the servicing provider can offer insight into the characteristics of the individual practitioner who was responsible for or provided direct care to a Medicaid or CHIP beneficiary. There are multiple systems available in TAF claim records for classifying providers, including the following:
Each of these classification systems may be best suited to different types of analyses. Although states are encouraged to populate all these fields on all claims, only one classification type is required. In practice, some states only submit information related to some of the classification systems, which may require TAF users to adjust their methodology across states. If the preferred data element has high rates of missingness, TAF users may be able to link the claims record to the provider’s record in the Annual Provider File (APR) to obtain the needed information. [2]
This data quality assessment examines whether any information on billing or servicing provider characteristics is available in claims records, and which type (taxonomy, specialty, or provider type). This information can help users design their analysis by selecting the fields populated in each state or identify states for which it may be necessary to link to the APR to obtain complete information about providers, rather than rely on claims alone.
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
States are required to submit information about all providers eligible to provide Medicaid- and CHIP-funded services; these records are captured in the APR. It is possible that missing information in the claim about provider taxonomy, specialty, or type could be imputed by linking to the provider’s record in the APR and obtaining the information from that file. In addition, the APR includes information on how states classify providers by “authorized category of service,” a classification scheme that can be used to identify certain non-medical provider categories not captured in the other classification systems, such as transportation or personal care service providers. For more information on the completeness of these fields in each state’s APR file, see the DQ Atlas single-topic displays for Group and Individual Providers - Classification Types and Facilities - Classification Types .
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
States are required to submit information about all providers eligible to provide Medicaid- and CHIP-funded services; these records are captured in the APR. It is possible that missing information in the claim about provider taxonomy, specialty, or type could be imputed by linking to the provider\u2019s record in the APR and obtaining the information from that file. In addition, the APR includes information on how states classify providers by \u201cauthorized category of service,\u201d a classification scheme that can be used to identify certain non-medical provider categories not captured in the other classification systems, such as transportation or personal care service providers. For more information on the completeness of these fields in each state\u2019s APR file, see the DQ Atlas single-topic displays for Group and Individual Providers - Classification Types and Facilities - Classification Types .
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We examined the values for provider type, specialty, and taxonomy codes for the providers most examined in analyses for the TAF IP, LT, OT, and RX files (Table 1). [3] We included both fee-for-service (FFS) claims and managed care encounter records, which represent claims paid by managed care organizations and should generally follow the same reporting standards for billing and servicing providers. We excluded financial transaction records, supplemental payments, and “other” records that the state did not classify as being covered by either the Medicaid or CHIP programs. [4] We also excluded states from the analysis of each file if the number of header records in the TAF was low enough to be unusable. [5] For Illinois, we restricted our analysis to the original version of the claim and excluded all subsequent adjustment records in the state’s TAF data. [6]
Table 1. Crosswalk between file type, data element description, and TAF variable name
| File type | Data element description | Variable name |
|---|---|---|
| IP, LT, OT | Billing provider | BLG_PRVDR_TYPE_CD |
| IP, LT, OT, RX | Billing provider specialty code | BLG_PRVDR_SPCLTY_CD |
| IP, LT, OT, RX | Billing provider state-reported taxonomy code | BLG_PRVDR_TXNMY_CD |
| IP, LT, OT | Billing provider NPPES primary taxonomy code | BLG_PRVDR_NPPES_TXNMY_CD |
| OT | Servicing provider type code | SRVCNG_PRVDR_TYPE_CD |
| OT | Servicing provider specialty code | SRVCNG_PRVDR_SPCLTY_CD |
| OT | Servicing provider state-reported taxonomy code | SRVCNG_PRVDR_TXNMY_CD |
| OT | Servicing provider NPPES primary taxonomy code | SRVCNG_PRVDR_NPPES_TXNMY_CD |
The provider type code includes 57 different valid values, covering both facilities and professionals. As shown in Table 2, we grouped these values into those that are expected, unexpected, and unusable for billing providers for each of the three file types in which this data element is present (IP, LT, OT) and for servicing providers in the OT file.
In the IP file, we would expect to see only general hospitals and Indian Health Service facilities as billing providers. In the LT file, we would expect to see nursing facilities, intermediate care facilities for individuals with intellectual or developmental disabilities, and psychiatric facilities as billing providers, as well as general hospitals in selected cases. [7] In the OT file, we would expect to see a broad range of both facilities and professionals billing for and directly providing services.
For all three files, we examined the “all other” code separately from the expected and unexpected categories. In some instances, use of the non-specific “all other” code may reflect the possibility that the provider is not represented in the list of valid provider type codes, which is the case for certain types of home- and community-based services (HCBS) and non-emergency medical transport providers. Although this code is valid and not considered an unexpected value in any file, it does not provide any information about the billing or servicing provider and thus renders the data element unusable for analysis.
Table 2. Classification of provider type codes, by file and provider field
| File and provider field | Expected codes | Unusable codes (“all other” value) | Unexpected codes |
|---|---|---|---|
| IP—billing provider | 42, 51 | 57 | All other values |
| LT—billing provider | 42, 43, 44, 45 | 57 | All other values |
| OT—billing and servicing provider | 01−56 | 57 | None |
Source: A full list of values is available in the TAF Claims Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries . The provider type code values are identical across all three files.
The provider specialty code includes 115 different valid values, covering both facilities and professionals. As shown in Table 3, we grouped these values into those that are expected and unexpected for each of the file types and provider fields. In the IP file, we would expect to see only general hospitals as billing providers. In the LT file, we would expect to see nursing facilities, intermediate care facilities for individuals with intellectual or developmental disabilities, and psychiatric facilities as billing providers, as well as general hospitals in selected cases. In the OT file, we would expect to see a broad range of both facilities and professionals billing for and directly providing services. In the RX file, we would expect to see pharmacies, medical supply companies, department stores, and grocery stores as billing providers. For all files and provider fields, we examined the “all other” codes (billing provider specialty code 87—All Other Suppliers) separately from the expected and unexpected categories. Although this code is valid and not considered an unexpected value in any file, it does not provide any information about the provider and thus renders the data element unusable for analysis.
Table 3. Classification of provider specialty codes, by file and provider field
| File and provider field | Expected codes | Unusable codes (“all other” value) | Unexpected codes |
|---|---|---|---|
| IP—billing provider | A0 | 87 | All other values except 88 and 99, which are treated as missing |
| LT—billing provider | A0, A1, A2, A3, B4 | 87 | All other values except 88 and 99, which are treated as missing |
| OT—billing and servicing provider | 1−86, 89−98, A0−B5 | 87 | All other values except 88 and 99, which are treated as missing |
| RX—billing provider | 51–54, 58, 73, A5, A6, A9, B1, B3, B4 | 87 | All other values except 88 and 99, which are treated as missing |
Source: A full list of values is available in the TAF Claims Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries . The provider specialty code values are identical across all claims files.
The state-reported billing provider taxonomy code is a 10-character alphanumeric code that identifies the provider’s area of specialization. As shown in Table 4, we grouped these values into those that are expected and unexpected for each of the file types. In the IP file, we would expect to see inpatient hospitals as billing providers. In the LT files we would expect to see nursing facilities, intermediate care facility services for individuals with intellectual disabilities, mental health facility services, and independent (free-standing) psychiatric wings of acute care hospitals as billing providers. In the OT file, we would expect to see a broad range of both facilities and professionals billing for and directly providing services. In the RX file, we would expect to see pharmacies and other suppliers as billing providers.
In addition to the state-reported provider taxonomy codes, IP, OT and LT files produced in 2022 and later years also include a constructed taxonomy variable for billing and servicing providers, the NPPES primary taxonomy code (*_NPPES_TXNMY_CD), that represents the provider’s primary taxonomy information as reported in the National Plan and Provider Enumeration System (NPPES) based on the provider’s National Provider Identifier. We used the same criteria as the state-reported taxonomy codes to determine expected and unexpected values for each file type.
Table 4. Classification of provider taxonomy codes, by file and provider field
| File and provider field | Expected codes | Unexpected codes |
|---|---|---|
| IP—billing provider | First 2 digits of taxonomy code are 27 or 28 | All other values |
| LT—billing provider | Taxonomy code begins with 283Q, 283X, 282E, 31, 32, 385H, or 281P | All other values |
| OT—billing and servicing provider | All valid taxonomy codes | None |
| RX—billing provider | First 2 digits of taxonomy code are 33, first 3 digits are 183, or 302R00000X [8] | All other values |
Source: The health care provider taxonomy code set can be found at http://www.wpc-edi.com/reference/codelists/healthcare/health-care-provider-taxonomy-code-set/ . The provider taxonomy code values are identical across all claims files.
We grouped states into categories of low, medium, and high concern about the usability of their data, based on the percentage of header records (or for servicing providers in the OT file, claim lines) that had an expected provider type, specialty, state-reported taxonomy code, or NPPES primary taxonomy code for the file and provider fields (Table 5).
Because users of the OT file may be interested in professional or institutional (facility) claims, we conducted the DQ assessment separately for each of these claim types. Table 6 shows the methodology used to classify each claim header as a professional or institutional claim. Any claim that did not meet at least one of the criteria for a professional or institutional claim was considered unclassified and is not included in this analysis. Note that the servicing provider analyses were conducted at the claim line level, but the claims were classified as professional or institutional claims at the claim header level.
Table 5. Criteria for DQ assessment of provider type, specialty, and taxonomy code
| Percentage of records with an expected provider type, specialty, state-reported taxonomy code, or NPPES primary taxonomy code | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
Table 6. Classification of OT claim headers as professional or facility claims
| Claim category | Identification rules |
|---|---|
| Professional | The record must meet at least one of the following criteria:
|
| Institutional | The record must meet at least one of the following criteria:
|
Methods previously used to assess data quality
Table 7 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 7. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
|
|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We used claim type code (CLM_TYPE_CD) to determine which records to include and exclude in our analysis. Records included are those with a claim type code indicating FFS claims (values of 1 and A) and managed care encounters (3 and C). We excluded capitated payments (2 and B), supplemental payments (5 and E), and service-tracking claims (4 and D). We also used claim type code to exclude “other” records (values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information on supplemental payments and other claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
More information about the completeness of each state’s TAF data can be found in the DQ Atlas single topic displays for Claims Volume—IP , Claims Volume—LT , Claims Volume—OT , and Claims Volume—RX .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This approach means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only those records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
General hospitals may provide sub-acute care through “swing beds” or other non-acute units within the hospital, and we would expect these claims to be found in the LT file rather than the IP file.
At least one state (Florida) allows Health Maintenance Organizations (HMOs) to dispense pharmaceuticals to enrollees. Therefore, HMOs are included as an “expected” taxonomy for RX billing providers.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We used claim type code (CLM_TYPE_CD) to determine which records to include and exclude in our analysis. Records included are those with a claim type code indicating FFS claims (values of 1 and A) and managed care encounters (3 and C). We excluded capitated payments (2 and B), supplemental payments (5 and E), and service-tracking claims (4 and D). We also used claim type code to exclude \u201cother\u201d records (values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information on supplemental payments and other claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
More information about the completeness of each state\u2019s TAF data can be found in the DQ Atlas single topic displays for Claims Volume\u2014IP , Claims Volume\u2014LT , Claims Volume\u2014OT , and Claims Volume\u2014RX .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This approach means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only those records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
General hospitals may provide sub-acute care through \u201cswing beds\u201d or other non-acute units within the hospital, and we would expect these claims to be found in the LT file rather than the IP file.
At least one state (Florida) allows Health Maintenance Organizations (HMOs) to dispense pharmaceuticals to enrollees. Therefore, HMOs are included as an \u201cexpected\u201d taxonomy for RX billing providers.
The billing provider type, specialty, and taxonomy codes allow TAF users to examine the characteristics of providers who bill and receive payments for Medicaid- and CHIP-funded services. This analysis examines the extent to which records in the LT file have a billing provider type, specialty, or taxonomy missing or coded with unexpected or unusable values.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5081"", ""relatedTopics"": [{""measureId"": 37, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - IP"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 0}, {""measureId"": 104, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - OT Professional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 2}, {""measureId"": 105, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - OT Institutional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 3}, {""measureId"": 103, ""measureName"": ""Billing Provider Specialty and Taxonomy - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 4}, {""measureId"": 106, ""measureName"": ""Servicing Provider Type, Specialty, and Taxonomy - OT Professional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 5}, {""measureId"": 107, ""measureName"": ""Servicing Provider Type, Specialty, and Taxonomy - OT Institutional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 6}]}" 40,"{""measureId"": 40, ""measureName"": ""Generic Indicator - RX"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Generic-Indicator-RX.pdf"", ""background"": {""content"": ""Prescription drug costs contribute significantly to national health care spending and are a fast-growing expenditure for Medicaid. [1] Even after prescription drug rebates from drug manufacturers are accounted for, state Medicaid programs collectively spent $31.7 billion on prescription drug fills in 2016. [2] As one strategy to help control prescription drug costs, the Centers for Medicare & Medicaid Services has encouraged states and Medicaid managed care plans to substitute generic for branded drugs for prescriptions whenever possible.
When new drugs are developed and approved by the Food and Drug Administration (FDA), a patent protects the manufacturer from competition for a period of time, during which it can freely market and sell a “branded” drug. After the patent expires, generic drugs—which are chemically identical to branded drugs in ingredients, concentration, safety, and dosage—can be marketed and sold by other manufacturers. Even after a patent expires, branded drugs continue to be sold under their original brand name. However, generic drugs tend to be significantly less expensive, on average, by about 85 percent relative to the branded equivalent. [3] Most state Medicaid programs have created a preferred list of drugs to encourage providers to prescribe the less expensive generic drugs when available. [4] The majority of prescription drug fills for Medicaid beneficiaries are for generic drugs; in fiscal year 2017, they accounted for 83 percent of national Medicaid drug prescription fills, whereas branded drugs accounted for 17 percent. [5]
Instead of classifying drugs as branded or generic, some analyses classify drugs as “single-source”―those marketed and sold by only a single manufacturer―or “multi-source”―those with the same active ingredients that are marketed and sold by more than one manufacturer. During the period of patent protection, all branded drugs are single source. After patent protection ends, a branded drug may continue to be marketed and sold under its brand name, often at a higher price than generics. These drugs are then classified as innovator multisource drugs. Non-branded generic drugs are classified as non-innovator multi-source drugs to differentiate them from the chemically identical branded drugs that continue to be sold after patent protection ends. [6]
Researchers and program administrators who want to understand the extent to which state Medicaid programs are covering branded versus generic prescription drugs can do so by analyzing the generic indicator in the T-MSIS Analytic Files (TAF). The generic indicator is a data element that comes directly from the T-MSIS data reported by states, and is used to classify prescriptions into three categories: branded drugs, generic drugs, and non-drugs. [7] This data quality assessment examines the distribution of missing values and unexpected patterns in valid values reported in the generic indicator field in the TAF.
Medicaid and CHIP Payment and Access Commission. “Medicaid Spending for Prescription Drugs.” January 2016. Available at https://www.macpac.gov/wp-content/uploads/2016/01/Medicaid-Spending-for-Prescription-Drugs.pdf . Accessed March 10, 2019.
Young, Katherine, and Rachel Garfield. “Snapshots of Recent State Initiatives in Medicaid Prescription Drug Cost Control.” Menlo Park, CA: Kaiser Family Foundation, 2018. Available at http://files.kff.org/attachment/Issue-Brief-Snapshots-of-Recent-State-Initiatives-in-Medicaid-Prescription-Drug-Cost-Control . Accessed March 10, 2019.
Food and Drug Administration. “Generic Drug Facts.” Silver Spring, MD: FDA, 2018. Available at https://www.fda.gov/Drugs/ResourcesForYou/Consumers/BuyingUsingMedicineSafely/GenericDrugs/ucm167991.htm . Accessed March 10, 2019.
Young, Katherine, and Rachel Garfield. “Snapshots of Recent State Initiatives in Medicaid Prescription Drug Cost Control.” Menlo Park, CA: Kaiser Family Foundation, 2018. Available at http://files.kff.org/attachment/Issue-Brief-Snapshots-of-Recent-State-Initiatives-in-Medicaid-Prescription-Drug-Cost-Control . Accessed March 10, 2019.
Medicaid and CHIP Payment and Access Commission. “Medicaid Drug Spending Trends.” February 2019. Available at https://www.macpac.gov/wp-content/uploads/2019/02/Medicaid-Drug-Spending-Trends.pdf. Accessed May 21, 2019.
Medicaid and CHIP Payment and Access Commission. “Medicaid Payment for Outpatient Prescription Drugs.” May 2018. Available at https://www.macpac.gov/wp-content/uploads/2015/09/Medicaid-Payment-for-Outpatient-Prescription-Drugs.pdf . Accessed March 10, 2019.
Previous versions of the T-MSIS data dictionary allowed “single-source” and “multi-source” as valid values for the generic indicator. However, these two values were discontinued with the T-MSIS Data Dictionary version 2.1 in November 2017. As a result, the 2016 and 2017 TAF may still have these values, but states are expected to stop reporting them over time.
Medicaid and CHIP Payment and Access Commission. \u201cMedicaid Spending for Prescription Drugs.\u201d January 2016. Available at https://www.macpac.gov/wp-content/uploads/2016/01/Medicaid-Spending-for-Prescription-Drugs.pdf . Accessed March 10, 2019.
Young, Katherine, and Rachel Garfield. \u201cSnapshots of Recent State Initiatives in Medicaid Prescription Drug Cost Control.\u201d Menlo Park, CA: Kaiser Family Foundation, 2018. Available at http://files.kff.org/attachment/Issue-Brief-Snapshots-of-Recent-State-Initiatives-in-Medicaid-Prescription-Drug-Cost-Control . Accessed March 10, 2019.
Food and Drug Administration. \u201cGeneric Drug Facts.\u201d Silver Spring, MD: FDA, 2018. Available at https://www.fda.gov/Drugs/ResourcesForYou/Consumers/BuyingUsingMedicineSafely/GenericDrugs/ucm167991.htm . Accessed March 10, 2019.
Young, Katherine, and Rachel Garfield. \u201cSnapshots of Recent State Initiatives in Medicaid Prescription Drug Cost Control.\u201d Menlo Park, CA: Kaiser Family Foundation, 2018. Available at http://files.kff.org/attachment/Issue-Brief-Snapshots-of-Recent-State-Initiatives-in-Medicaid-Prescription-Drug-Cost-Control . Accessed March 10, 2019.
Medicaid and CHIP Payment and Access Commission. \u201cMedicaid Drug Spending Trends.\u201d February 2019. Available at https://www.macpac.gov/wp-content/uploads/2019/02/Medicaid-Drug-Spending-Trends.pdf. Accessed May 21, 2019.
Medicaid and CHIP Payment and Access Commission. \u201cMedicaid Payment for Outpatient Prescription Drugs.\u201d May 2018. Available at https://www.macpac.gov/wp-content/uploads/2015/09/Medicaid-Payment-for-Outpatient-Prescription-Drugs.pdf . Accessed March 10, 2019.
Previous versions of the T-MSIS data dictionary allowed \u201csingle-source\u201d and \u201cmulti-source\u201d as valid values for the generic indicator. However, these two values were discontinued with the T-MSIS Data Dictionary version 2.1 in November 2017. As a result, the 2016 and 2017 TAF may still have these values, but states are expected to stop reporting them over time.
Using the TAF, we examined the generic indicator (BRND_GNRC_IND) available on line-level records in the TAF RX file. [8] We included fee-for-service (FFS) claims and managed care encounter records for both Medicaid and CHIP beneficiaries in the analysis. [9] For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [10]
In the TAF, the generic indicator has five valid values: generic, branded, multisource, single-source, and non-drug. Because most branded drugs are single-source, and most generic drugs are multisource, we grouped generic and multisource drugs into one category, and branded and single-source drugs into a separate category for the purposes of this analysis. We considered the non-drug value as a third valid category, although we expected a relatively small proportion of all claims submitted to the RX file to fall into this category. Common products that states may classify into the non-drug category include inhalers, over-the-counter (nonprescription) drugs, and medical supplies such as blood glucose test strips.
For each state, we examined the “missingness” of the generic indicator as well as the distribution of claims across the generic, branded, and non-drug categories. We expected a relatively small proportion (less than 10 percent) of RX claims to be for non-drug products. Additionally, consistent with other sources of Medicaid prescription drug data, we expected that the majority of RX claims would be for generic drug fills. [11] We assessed the quality of the generic indicator based on the criteria in Table 1.
Table 1. Criteria for DQ assessment of the generic indicator
| Missing and non-drug | Branded | Generic | DQ assessment |
|---|---|---|---|
| x ≤ 10 percent | 0 percent < x < 50 percent | x ≥ 50 percent | Low concern a |
| 10 percent < x ≤ 20 percent | 0 percent < x < 50 percent | x ≥ 50 percent | Medium concern a |
| x > 20 percent | x = 0 percent or x ≥ 50 percent | x = 0 percent or x = 100 percent | High concern b |
a All criteria must be true for a state to receive the given DQ Assessment.
b At least one of the three criteria must be true for a state to receive the given DQ Assessment.
It was outside the scope of this data quality assessment to evaluate whether states correctly classified each RX record as branded or generic based on the National Drug Code. However, TAF users who wish to examine generic and branded drug fills in states in which we have concerns about the quality of the generic indicator may be able to do so by using drug grouper software that can classify each drug as generic or branded based on the National Drug Code.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Claim type code (CLM_TYPE_CD) was used to determine which records to include and exclude. Through this method, we retained FFS records (claim type 1 or A) and managed care encounters (claim type 3 or C). We excluded records that had all other claim-type values, including capitation payments, service-tracking claims, and supplemental payments, none of which were expected to have a valid generic indicator value because they are financial transaction records that are not submitted on an institutional claim form. We also excluded the “other” records that the state did not classify as either Medicaid or CHIP payment records.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
Medicaid and CHIP Payment and Access Commission. “Medicaid Payment for Outpatient Prescription Drugs.” May 2018b. Available at https://www.macpac.gov/wp-content/uploads/2015/09/Medicaid-Payment-for-Outpatient-Prescription-Drugs.pdf . Accessed March 10, 2019.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Claim type code (CLM_TYPE_CD) was used to determine which records to include and exclude. Through this method, we retained FFS records (claim type 1 or A) and managed care encounters (claim type 3 or C). We excluded records that had all other claim-type values, including capitation payments, service-tracking claims, and supplemental payments, none of which were expected to have a valid generic indicator value because they are financial transaction records that are not submitted on an institutional claim form. We also excluded the \u201cother\u201d records that the state did not classify as either Medicaid or CHIP payment records.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
Medicaid and CHIP Payment and Access Commission. \u201cMedicaid Payment for Outpatient Prescription Drugs.\u201d May 2018b. Available at https://www.macpac.gov/wp-content/uploads/2015/09/Medicaid-Payment-for-Outpatient-Prescription-Drugs.pdf . Accessed March 10, 2019.
The generic indicator in the RX file can be used to differentiate between fills for branded prescription drugs, fills for generic prescription drugs, and fills for non-drug products. This analysis examines how often the generic indicator is missing and whether the distribution of values indicates a likely data quality problem.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5091"", ""relatedTopics"": []}" 41,"{""measureId"": 41, ""measureName"": ""Admission Date - IP"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Admission-Date-IP.pdf"", ""background"": {""content"": ""Users of the T-MSIS Analytic Files (TAF) may need to identify stays in inpatient and long-term care facilities, including when those stays begin and end. This information is crucial in evaluating health outcomes such as lengths of stays or 30-day all-cause readmission rates. To identify stays, date fields are essential. In the TAF, claims records in the inpatient (IP) and long-term care (LT) files include both admission and discharge dates and the start and end dates for individual services provided during the stay. [1] Admission and discharge dates that are correctly coded are particularly important when a long stay generates multiple claims before a patient is discharged. These two dates allow TAF users to link all of the claims associated with a single stay and to correctly calculate the number and length of each inpatient and long-term care stay.
Although states are required to report admission dates on all claims for overnight facility stays, discharge dates are expected to be missing from IP and LT claims in some cases. [2] For instance, states allow providers to engage in “interim” or “split” billing, which means that a facility can generate a claim that covers only a portion of the time that a beneficiary stays in that facility. Interim claims can be identified in administrative data by examining the patient status code to determine whether the beneficiary was still a patient on the service end date of the claim. We would expect the discharge date to be missing from interim claims because the beneficiary has not been discharged; the discharge date may also not be known when the claim is reported. As a result, it is valid for some IP claims and most LT claims to be missing discharge dates. [3]
In addition, some states may be reporting invalid admission and discharge dates due to reporting errors (such as submitting the adjudication date in the discharge date field). This can result in the admission date occurring after the discharge date for some records, which impacts calculations of inpatient length of stay.
This data quality assessment examines the extent to which admission and discharge dates are available and valid in the TAF IP and LT claims files. We also evaluate whether there are differences between the TAF claims files (IP or LT) or claim types (fee-for-service [FFS] claims or managed care encounters).
The records in each monthly claims TAF are selected on the basis of certain dates reported on the claims. For the IP file, claims are included in the TAF if the discharge date on the claim header occurred during the month and year of the TAF; if the discharge date is missing, then the most recent service end date from the claim lines is used. For the LT file, claims are included in the TAF if the service end date on the claim header occurred during the month and year of the TAF.
Some states may not require nursing facilities to report admission and/or discharge dates on claims if the dates are tracked separately through the federally required Preadmission Screening and Resident Review process. In these cases, the data may be missing from the TAF claims files.
Discharge dates are not usually directly reported by providers on the claims that they submit to states or managed care plans. Rather, the states or managed care plans conditionally map the service end date reported by the provider to the discharge date field in T-MSIS if other information on the claim (for example, the discharge hour or the patient status code) indicates that the patient was discharged. Errors in this conditional mapping logic may result in discharge dates being missing or erroneous even when the patient was discharged.
The records in each monthly claims TAF are selected on the basis of certain dates reported on the claims. For the IP file, claims are included in the TAF if the discharge date on the claim header occurred during the month and year of the TAF; if the discharge date is missing, then the most recent service end date from the claim lines is used. For the LT file, claims are included in the TAF if the service end date on the claim header occurred during the month and year of the TAF.
Some states may not require nursing facilities to report admission and/or discharge dates on claims if the dates are tracked separately through the federally required Preadmission Screening and Resident Review process. In these cases, the data may be missing from the TAF claims files.
Discharge dates are not usually directly reported by providers on the claims that they submit to states or managed care plans. Rather, the states or managed care plans conditionally map the service end date reported by the provider to the discharge date field in T-MSIS if other information on the claim (for example, the discharge hour or the patient status code) indicates that the patient was discharged. Errors in this conditional mapping logic may result in discharge dates being missing or erroneous even when the patient was discharged.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
This analysis is based on the TAF. [4] We assessed the extent to which admission and discharge dates are available and valid on FFS claims and managed care encounter records in the IP and LT claims files. [5] We excluded states from the analysis if they had an unusably low volume of claims in the IP or LT files. For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [6]
Admission date
To assess the admission date field, we calculated the percentage of header records that either were missing an admission date (ADMSN_DT) or had an invalid admission date for each claims file. We defined an admission date as invalid if it was greater than its discharge date (DSCHRG_DT) on the same claim, which indicates admission date may have been incorrectly reported. Then, we grouped states into categories of data quality concern, depending on the percentage of records that were missing the field (Table 1).
Table 1. Criteria for DQ assessment of admission dates
| Percentage of claims with missing or invalid admission dates | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
Discharge date
To assess the discharge date field, we calculated the percentage of header records with an unexpectedly missing or invalid discharge date for each claims file and—for the LT file only—the percentage of header records that were missing the discharge date, regardless of whether we expected the date to be missing. Then, we grouped states into categories of data quality concern, depending on those measures (Table 2). We defined a discharge date as invalid if the date occurred before the admission date on the same claim.
We considered the discharge date to be unexpectedly missing when it was missing from a claim on which the patient status code (PTNT_STUS_CD) indicated that the beneficiary was in fact discharged. To identify the number of records on which the discharge date was unexpectedly missing, we took the number of records with a missing discharge date and removed the records on which the patient status code was equal to 30 (still a patient). These subtracted records represent cases in which we did not expect a discharge date to be available because the beneficiary remained a patient instead of being discharged as of the service end date on the claim. The remaining count represented records with an unexpectedly missing discharge date and include cases in which the beneficiary was discharged or transferred to another setting, left the facility against medical advice, or died. [7]
We calculated the overall percentage of header records that were missing the discharge date (without checking the patient status code) only for the LT file because LT claims often represent weekly, biweekly, or monthly interim bills for beneficiaries’ extended stays in long-term care facilities. We would expect most of these beneficiaries to remain in these facilities after each claim is submitted; therefore, a discharge date should not be available on the claim. A very low percentage of missing discharge dates on LT claims would suggest that the dates were not the actual dates on which beneficiaries were discharged. Therefore, either a high percentage of LT records with unexpectedly missing or invalid discharge dates or a very low percentage of LT records with missing discharge dates, regardless of if they are expected, indicate a data quality problem for states. We classified states in which less than 2 percent of LT claims were missing a discharge date as having unusable data.
Table 2. Criteria for DQ assessment of discharge dates
| Percentage of claims with unexpectedly missing or invalid discharge dates | Percentage of claims with missing discharge dates (LT file only) | DQ assessment |
|---|---|---|
| x ≤ 10 percent | x ≥ 2 percent | Low concern |
| 10 percent < x ≤ 20 percent | x ≥ 2 percent | Medium concern |
| 20 percent < x ≤ 50 percent | x ≥ 2 percent | High concern |
| x > 50 percent | x < 2 percent | Unusable |
We did not investigate whether the reported admission and discharge dates were reasonable compared with other dates reported on claims (e.g., service, procedure, or adjudication dates), as doing so requires determining the expected length of stay given the diagnoses and services delivered. This complex undertaking is beyond the scope of this analysis.
Methods previously used to assess data quality
Table 3 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods.
Table 3. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We included all claims with a claim type code for either FFS claims or managed care encounter records (CLM_TYPE_CD equal to 1, 3, A, or C). We did not include capitation payments, supplemental payments, or service tracking payments. We also did not include any records with a claim type of “other,” which are records that the state did not classify as Medicaid or CHIP.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
The count also includes claims with a patient status code of 09 (admitted as an inpatient to this hospital), which occurs only on a very small percentage of states’ claims and tends to be reported on claims on which the discharge dates are not missing.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We included all claims with a claim type code for either FFS claims or managed care encounter records (CLM_TYPE_CD equal to 1, 3, A, or C). We did not include capitation payments, supplemental payments, or service tracking payments. We also did not include any records with a claim type of \u201cother,\u201d which are records that the state did not classify as Medicaid or CHIP.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
The count also includes claims with a patient status code of 09 (admitted as an inpatient to this hospital), which occurs only on a very small percentage of states\u2019 claims and tends to be reported on claims on which the discharge dates are not missing.
Admission and discharge dates indicate when inpatient stays start and end. This analysis examines the extent to which admission dates are missing or invalid from the IP file.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5101"", ""relatedTopics"": [{""measureId"": 42, ""measureName"": ""Admission Date - LT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 1}, {""measureId"": 43, ""measureName"": ""Discharge Date - IP"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 2}, {""measureId"": 44, ""measureName"": ""Discharge Date - LT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 3}]}" 42,"{""measureId"": 42, ""measureName"": ""Admission Date - LT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Admission-Date-LT.pdf"", ""background"": {""content"": ""Users of the T-MSIS Analytic Files (TAF) may need to identify stays in inpatient and long-term care facilities, including when those stays begin and end. This information is crucial in evaluating health outcomes such as lengths of stays or 30-day all-cause readmission rates. To identify stays, date fields are essential. In the TAF, claims records in the inpatient (IP) and long-term care (LT) files include both admission and discharge dates and the start and end dates for individual services provided during the stay. [1] Admission and discharge dates that are correctly coded are particularly important when a long stay generates multiple claims before a patient is discharged. These two dates allow TAF users to link all of the claims associated with a single stay and to correctly calculate the number and length of each inpatient and long-term care stay.
Although states are required to report admission dates on all claims for overnight facility stays, discharge dates are expected to be missing from IP and LT claims in some cases. [2] For instance, states allow providers to engage in “interim” or “split” billing, which means that a facility can generate a claim that covers only a portion of the time that a beneficiary stays in that facility. Interim claims can be identified in administrative data by examining the patient status code to determine whether the beneficiary was still a patient on the service end date of the claim. We would expect the discharge date to be missing from interim claims because the beneficiary has not been discharged; the discharge date may also not be known when the claim is reported. As a result, it is valid for some IP claims and most LT claims to be missing discharge dates. [3]
In addition, some states may be reporting invalid admission and discharge dates due to reporting errors (such as submitting the adjudication date in the discharge date field). This can result in the admission date occurring after the discharge date for some records, which impacts calculations of inpatient length of stay.
This data quality assessment examines the extent to which admission and discharge dates are available and valid in the TAF IP and LT claims files. We also evaluate whether there are differences between the TAF claims files (IP or LT) or claim types (fee-for-service [FFS] claims or managed care encounters).
The records in each monthly claims TAF are selected on the basis of certain dates reported on the claims. For the IP file, claims are included in the TAF if the discharge date on the claim header occurred during the month and year of the TAF; if the discharge date is missing, then the most recent service end date from the claim lines is used. For the LT file, claims are included in the TAF if the service end date on the claim header occurred during the month and year of the TAF.
Some states may not require nursing facilities to report admission and/or discharge dates on claims if the dates are tracked separately through the federally required Preadmission Screening and Resident Review process. In these cases, the data may be missing from the TAF claims files.
Discharge dates are not usually directly reported by providers on the claims that they submit to states or managed care plans. Rather, the states or managed care plans conditionally map the service end date reported by the provider to the discharge date field in T-MSIS if other information on the claim (for example, the discharge hour or the patient status code) indicates that the patient was discharged. Errors in this conditional mapping logic may result in discharge dates being missing or erroneous even when the patient was discharged.
The records in each monthly claims TAF are selected on the basis of certain dates reported on the claims. For the IP file, claims are included in the TAF if the discharge date on the claim header occurred during the month and year of the TAF; if the discharge date is missing, then the most recent service end date from the claim lines is used. For the LT file, claims are included in the TAF if the service end date on the claim header occurred during the month and year of the TAF.
Some states may not require nursing facilities to report admission and/or discharge dates on claims if the dates are tracked separately through the federally required Preadmission Screening and Resident Review process. In these cases, the data may be missing from the TAF claims files.
Discharge dates are not usually directly reported by providers on the claims that they submit to states or managed care plans. Rather, the states or managed care plans conditionally map the service end date reported by the provider to the discharge date field in T-MSIS if other information on the claim (for example, the discharge hour or the patient status code) indicates that the patient was discharged. Errors in this conditional mapping logic may result in discharge dates being missing or erroneous even when the patient was discharged.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
This analysis is based on the TAF. [4] We assessed the extent to which admission and discharge dates are available and valid on FFS claims and managed care encounter records in the IP and LT claims files. [5] We excluded states from the analysis if they had an unusably low volume of claims in the IP or LT files. For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [6]
Admission date
To assess the admission date field, we calculated the percentage of header records that either were missing an admission date (ADMSN_DT) or had an invalid admission date for each claims file. We defined an admission date as invalid if it was greater than its discharge date (DSCHRG_DT) on the same claim, which indicates admission date may have been incorrectly reported. Then, we grouped states into categories of data quality concern, depending on the percentage of records that were missing the field (Table 1).
Table 1. Criteria for DQ assessment of admission dates
| Percentage of claims with missing or invalid admission dates | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
Discharge date
To assess the discharge date field, we calculated the percentage of header records with an unexpectedly missing or invalid discharge date for each claims file and—for the LT file only—the percentage of header records that were missing the discharge date, regardless of whether we expected the date to be missing. Then, we grouped states into categories of data quality concern, depending on those measures (Table 2). We defined a discharge date as invalid if the date occurred before the admission date on the same claim.
We considered the discharge date to be unexpectedly missing when it was missing from a claim on which the patient status code (PTNT_STUS_CD) indicated that the beneficiary was in fact discharged. To identify the number of records on which the discharge date was unexpectedly missing, we took the number of records with a missing discharge date and removed the records on which the patient status code was equal to 30 (still a patient). These subtracted records represent cases in which we did not expect a discharge date to be available because the beneficiary remained a patient instead of being discharged as of the service end date on the claim. The remaining count represented records with an unexpectedly missing discharge date and include cases in which the beneficiary was discharged or transferred to another setting, left the facility against medical advice, or died. [7]
We calculated the overall percentage of header records that were missing the discharge date (without checking the patient status code) only for the LT file because LT claims often represent weekly, biweekly, or monthly interim bills for beneficiaries’ extended stays in long-term care facilities. We would expect most of these beneficiaries to remain in these facilities after each claim is submitted; therefore, a discharge date should not be available on the claim. A very low percentage of missing discharge dates on LT claims would suggest that the dates were not the actual dates on which beneficiaries were discharged. Therefore, either a high percentage of LT records with unexpectedly missing or invalid discharge dates or a very low percentage of LT records with missing discharge dates, regardless of if they are expected, indicate a data quality problem for states. We classified states in which less than 2 percent of LT claims were missing a discharge date as having unusable data.
Table 2. Criteria for DQ assessment of discharge dates
| Percentage of claims with unexpectedly missing or invalid discharge dates | Percentage of claims with missing discharge dates (LT file only) | DQ assessment |
|---|---|---|
| x ≤ 10 percent | x ≥ 2 percent | Low concern |
| 10 percent < x ≤ 20 percent | x ≥ 2 percent | Medium concern |
| 20 percent < x ≤ 50 percent | x ≥ 2 percent | High concern |
| x > 50 percent | x < 2 percent | Unusable |
We did not investigate whether the reported admission and discharge dates were reasonable compared with other dates reported on claims (e.g., service, procedure, or adjudication dates), as doing so requires determining the expected length of stay given the diagnoses and services delivered. This complex undertaking is beyond the scope of this analysis.
Methods previously used to assess data quality
Table 3 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods.
Table 3. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We included all claims with a claim type code for either FFS claims or managed care encounter records (CLM_TYPE_CD equal to 1, 3, A, or C). We did not include capitation payments, supplemental payments, or service tracking payments. We also did not include any records with a claim type of “other,” which are records that the state did not classify as Medicaid or CHIP.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
The count also includes claims with a patient status code of 09 (admitted as an inpatient to this hospital), which occurs only on a very small percentage of states’ claims and tends to be reported on claims on which the discharge dates are not missing.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We included all claims with a claim type code for either FFS claims or managed care encounter records (CLM_TYPE_CD equal to 1, 3, A, or C). We did not include capitation payments, supplemental payments, or service tracking payments. We also did not include any records with a claim type of \u201cother,\u201d which are records that the state did not classify as Medicaid or CHIP.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
The count also includes claims with a patient status code of 09 (admitted as an inpatient to this hospital), which occurs only on a very small percentage of states\u2019 claims and tends to be reported on claims on which the discharge dates are not missing.
Admission and discharge dates indicate when institutional long-term care stays start and end. This analysis examines the extent to which admission dates are missing or invalid from the LT file.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5101"", ""relatedTopics"": [{""measureId"": 41, ""measureName"": ""Admission Date - IP"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 0}, {""measureId"": 43, ""measureName"": ""Discharge Date - IP"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 2}, {""measureId"": 44, ""measureName"": ""Discharge Date - LT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 3}]}" 43,"{""measureId"": 43, ""measureName"": ""Discharge Date - IP"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Discharge-Date-IP.pdf"", ""background"": {""content"": ""Users of the T-MSIS Analytic Files (TAF) may need to identify stays in inpatient and long-term care facilities, including when those stays begin and end. This information is crucial in evaluating health outcomes such as lengths of stays or 30-day all-cause readmission rates. To identify stays, date fields are essential. In the TAF, claims records in the inpatient (IP) and long-term care (LT) files include both admission and discharge dates and the start and end dates for individual services provided during the stay. [1] Admission and discharge dates that are correctly coded are particularly important when a long stay generates multiple claims before a patient is discharged. These two dates allow TAF users to link all of the claims associated with a single stay and to correctly calculate the number and length of each inpatient and long-term care stay.
Although states are required to report admission dates on all claims for overnight facility stays, discharge dates are expected to be missing from IP and LT claims in some cases. [2] For instance, states allow providers to engage in “interim” or “split” billing, which means that a facility can generate a claim that covers only a portion of the time that a beneficiary stays in that facility. Interim claims can be identified in administrative data by examining the patient status code to determine whether the beneficiary was still a patient on the service end date of the claim. We would expect the discharge date to be missing from interim claims because the beneficiary has not been discharged; the discharge date may also not be known when the claim is reported. As a result, it is valid for some IP claims and most LT claims to be missing discharge dates. [3]
In addition, some states may be reporting invalid admission and discharge dates due to reporting errors (such as submitting the adjudication date in the discharge date field). This can result in the admission date occurring after the discharge date for some records, which impacts calculations of inpatient length of stay.
This data quality assessment examines the extent to which admission and discharge dates are available and valid in the TAF IP and LT claims files. We also evaluate whether there are differences between the TAF claims files (IP or LT) or claim types (fee-for-service [FFS] claims or managed care encounters).
The records in each monthly claims TAF are selected on the basis of certain dates reported on the claims. For the IP file, claims are included in the TAF if the discharge date on the claim header occurred during the month and year of the TAF; if the discharge date is missing, then the most recent service end date from the claim lines is used. For the LT file, claims are included in the TAF if the service end date on the claim header occurred during the month and year of the TAF.
Some states may not require nursing facilities to report admission and/or discharge dates on claims if the dates are tracked separately through the federally required Preadmission Screening and Resident Review process. In these cases, the data may be missing from the TAF claims files.
Discharge dates are not usually directly reported by providers on the claims that they submit to states or managed care plans. Rather, the states or managed care plans conditionally map the service end date reported by the provider to the discharge date field in T-MSIS if other information on the claim (for example, the discharge hour or the patient status code) indicates that the patient was discharged. Errors in this conditional mapping logic may result in discharge dates being missing or erroneous even when the patient was discharged.
The records in each monthly claims TAF are selected on the basis of certain dates reported on the claims. For the IP file, claims are included in the TAF if the discharge date on the claim header occurred during the month and year of the TAF; if the discharge date is missing, then the most recent service end date from the claim lines is used. For the LT file, claims are included in the TAF if the service end date on the claim header occurred during the month and year of the TAF.
Some states may not require nursing facilities to report admission and/or discharge dates on claims if the dates are tracked separately through the federally required Preadmission Screening and Resident Review process. In these cases, the data may be missing from the TAF claims files.
Discharge dates are not usually directly reported by providers on the claims that they submit to states or managed care plans. Rather, the states or managed care plans conditionally map the service end date reported by the provider to the discharge date field in T-MSIS if other information on the claim (for example, the discharge hour or the patient status code) indicates that the patient was discharged. Errors in this conditional mapping logic may result in discharge dates being missing or erroneous even when the patient was discharged.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
This analysis is based on the TAF. [4] We assessed the extent to which admission and discharge dates are available and valid on FFS claims and managed care encounter records in the IP and LT claims files. [5] We excluded states from the analysis if they had an unusably low volume of claims in the IP or LT files. For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [6]
Admission date
To assess the admission date field, we calculated the percentage of header records that either were missing an admission date (ADMSN_DT) or had an invalid admission date for each claims file. We defined an admission date as invalid if it was greater than its discharge date (DSCHRG_DT) on the same claim, which indicates admission date may have been incorrectly reported. Then, we grouped states into categories of data quality concern, depending on the percentage of records that were missing the field (Table 1).
Table 1. Criteria for DQ assessment of admission dates
| Percentage of claims with missing or invalid admission dates | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
Discharge date
To assess the discharge date field, we calculated the percentage of header records with an unexpectedly missing or invalid discharge date for each claims file and—for the LT file only—the percentage of header records that were missing the discharge date, regardless of whether we expected the date to be missing. Then, we grouped states into categories of data quality concern, depending on those measures (Table 2). We defined a discharge date as invalid if the date occurred before the admission date on the same claim.
We considered the discharge date to be unexpectedly missing when it was missing from a claim on which the patient status code (PTNT_STUS_CD) indicated that the beneficiary was in fact discharged. To identify the number of records on which the discharge date was unexpectedly missing, we took the number of records with a missing discharge date and removed the records on which the patient status code was equal to 30 (still a patient). These subtracted records represent cases in which we did not expect a discharge date to be available because the beneficiary remained a patient instead of being discharged as of the service end date on the claim. The remaining count represented records with an unexpectedly missing discharge date and include cases in which the beneficiary was discharged or transferred to another setting, left the facility against medical advice, or died. [7]
We calculated the overall percentage of header records that were missing the discharge date (without checking the patient status code) only for the LT file because LT claims often represent weekly, biweekly, or monthly interim bills for beneficiaries’ extended stays in long-term care facilities. We would expect most of these beneficiaries to remain in these facilities after each claim is submitted; therefore, a discharge date should not be available on the claim. A very low percentage of missing discharge dates on LT claims would suggest that the dates were not the actual dates on which beneficiaries were discharged. Therefore, either a high percentage of LT records with unexpectedly missing or invalid discharge dates or a very low percentage of LT records with missing discharge dates, regardless of if they are expected, indicate a data quality problem for states. We classified states in which less than 2 percent of LT claims were missing a discharge date as having unusable data.
Table 2. Criteria for DQ assessment of discharge dates
| Percentage of claims with unexpectedly missing or invalid discharge dates | Percentage of claims with missing discharge dates (LT file only) | DQ assessment |
|---|---|---|
| x ≤ 10 percent | x ≥ 2 percent | Low concern |
| 10 percent < x ≤ 20 percent | x ≥ 2 percent | Medium concern |
| 20 percent < x ≤ 50 percent | x ≥ 2 percent | High concern |
| x > 50 percent | x < 2 percent | Unusable |
We did not investigate whether the reported admission and discharge dates were reasonable compared with other dates reported on claims (e.g., service, procedure, or adjudication dates), as doing so requires determining the expected length of stay given the diagnoses and services delivered. This complex undertaking is beyond the scope of this analysis.
Methods previously used to assess data quality
Table 3 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods.
Table 3. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We included all claims with a claim type code for either FFS claims or managed care encounter records (CLM_TYPE_CD equal to 1, 3, A, or C). We did not include capitation payments, supplemental payments, or service tracking payments. We also did not include any records with a claim type of “other,” which are records that the state did not classify as Medicaid or CHIP.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
The count also includes claims with a patient status code of 09 (admitted as an inpatient to this hospital), which occurs only on a very small percentage of states’ claims and tends to be reported on claims on which the discharge dates are not missing.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We included all claims with a claim type code for either FFS claims or managed care encounter records (CLM_TYPE_CD equal to 1, 3, A, or C). We did not include capitation payments, supplemental payments, or service tracking payments. We also did not include any records with a claim type of \u201cother,\u201d which are records that the state did not classify as Medicaid or CHIP.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
The count also includes claims with a patient status code of 09 (admitted as an inpatient to this hospital), which occurs only on a very small percentage of states\u2019 claims and tends to be reported on claims on which the discharge dates are not missing.
Admission and discharge dates indicate when inpatient stays start and end. This analysis examines the extent to which discharge dates are unexpectedly missing or invalid from the IP file.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5101"", ""relatedTopics"": [{""measureId"": 41, ""measureName"": ""Admission Date - IP"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 0}, {""measureId"": 42, ""measureName"": ""Admission Date - LT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 1}, {""measureId"": 44, ""measureName"": ""Discharge Date - LT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 3}]}" 44,"{""measureId"": 44, ""measureName"": ""Discharge Date - LT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Discharge-Date-LT.pdf"", ""background"": {""content"": ""Users of the T-MSIS Analytic Files (TAF) may need to identify stays in inpatient and long-term care facilities, including when those stays begin and end. This information is crucial in evaluating health outcomes such as lengths of stays or 30-day all-cause readmission rates. To identify stays, date fields are essential. In the TAF, claims records in the inpatient (IP) and long-term care (LT) files include both admission and discharge dates and the start and end dates for individual services provided during the stay. [1] Admission and discharge dates that are correctly coded are particularly important when a long stay generates multiple claims before a patient is discharged. These two dates allow TAF users to link all of the claims associated with a single stay and to correctly calculate the number and length of each inpatient and long-term care stay.
Although states are required to report admission dates on all claims for overnight facility stays, discharge dates are expected to be missing from IP and LT claims in some cases. [2] For instance, states allow providers to engage in “interim” or “split” billing, which means that a facility can generate a claim that covers only a portion of the time that a beneficiary stays in that facility. Interim claims can be identified in administrative data by examining the patient status code to determine whether the beneficiary was still a patient on the service end date of the claim. We would expect the discharge date to be missing from interim claims because the beneficiary has not been discharged; the discharge date may also not be known when the claim is reported. As a result, it is valid for some IP claims and most LT claims to be missing discharge dates. [3]
In addition, some states may be reporting invalid admission and discharge dates due to reporting errors (such as submitting the adjudication date in the discharge date field). This can result in the admission date occurring after the discharge date for some records, which impacts calculations of inpatient length of stay.
This data quality assessment examines the extent to which admission and discharge dates are available and valid in the TAF IP and LT claims files. We also evaluate whether there are differences between the TAF claims files (IP or LT) or claim types (fee-for-service [FFS] claims or managed care encounters).
The records in each monthly claims TAF are selected on the basis of certain dates reported on the claims. For the IP file, claims are included in the TAF if the discharge date on the claim header occurred during the month and year of the TAF; if the discharge date is missing, then the most recent service end date from the claim lines is used. For the LT file, claims are included in the TAF if the service end date on the claim header occurred during the month and year of the TAF.
Some states may not require nursing facilities to report admission and/or discharge dates on claims if the dates are tracked separately through the federally required Preadmission Screening and Resident Review process. In these cases, the data may be missing from the TAF claims files.
Discharge dates are not usually directly reported by providers on the claims that they submit to states or managed care plans. Rather, the states or managed care plans conditionally map the service end date reported by the provider to the discharge date field in T-MSIS if other information on the claim (for example, the discharge hour or the patient status code) indicates that the patient was discharged. Errors in this conditional mapping logic may result in discharge dates being missing or erroneous even when the patient was discharged.
The records in each monthly claims TAF are selected on the basis of certain dates reported on the claims. For the IP file, claims are included in the TAF if the discharge date on the claim header occurred during the month and year of the TAF; if the discharge date is missing, then the most recent service end date from the claim lines is used. For the LT file, claims are included in the TAF if the service end date on the claim header occurred during the month and year of the TAF.
Some states may not require nursing facilities to report admission and/or discharge dates on claims if the dates are tracked separately through the federally required Preadmission Screening and Resident Review process. In these cases, the data may be missing from the TAF claims files.
Discharge dates are not usually directly reported by providers on the claims that they submit to states or managed care plans. Rather, the states or managed care plans conditionally map the service end date reported by the provider to the discharge date field in T-MSIS if other information on the claim (for example, the discharge hour or the patient status code) indicates that the patient was discharged. Errors in this conditional mapping logic may result in discharge dates being missing or erroneous even when the patient was discharged.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
This analysis is based on the TAF. [4] We assessed the extent to which admission and discharge dates are available and valid on FFS claims and managed care encounter records in the IP and LT claims files. [5] We excluded states from the analysis if they had an unusably low volume of claims in the IP or LT files. For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [6]
Admission date
To assess the admission date field, we calculated the percentage of header records that either were missing an admission date (ADMSN_DT) or had an invalid admission date for each claims file. We defined an admission date as invalid if it was greater than its discharge date (DSCHRG_DT) on the same claim, which indicates admission date may have been incorrectly reported. Then, we grouped states into categories of data quality concern, depending on the percentage of records that were missing the field (Table 1).
Table 1. Criteria for DQ assessment of admission dates
| Percentage of claims with missing or invalid admission dates | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
Discharge date
To assess the discharge date field, we calculated the percentage of header records with an unexpectedly missing or invalid discharge date for each claims file and—for the LT file only—the percentage of header records that were missing the discharge date, regardless of whether we expected the date to be missing. Then, we grouped states into categories of data quality concern, depending on those measures (Table 2). We defined a discharge date as invalid if the date occurred before the admission date on the same claim.
We considered the discharge date to be unexpectedly missing when it was missing from a claim on which the patient status code (PTNT_STUS_CD) indicated that the beneficiary was in fact discharged. To identify the number of records on which the discharge date was unexpectedly missing, we took the number of records with a missing discharge date and removed the records on which the patient status code was equal to 30 (still a patient). These subtracted records represent cases in which we did not expect a discharge date to be available because the beneficiary remained a patient instead of being discharged as of the service end date on the claim. The remaining count represented records with an unexpectedly missing discharge date and include cases in which the beneficiary was discharged or transferred to another setting, left the facility against medical advice, or died. [7]
We calculated the overall percentage of header records that were missing the discharge date (without checking the patient status code) only for the LT file because LT claims often represent weekly, biweekly, or monthly interim bills for beneficiaries’ extended stays in long-term care facilities. We would expect most of these beneficiaries to remain in these facilities after each claim is submitted; therefore, a discharge date should not be available on the claim. A very low percentage of missing discharge dates on LT claims would suggest that the dates were not the actual dates on which beneficiaries were discharged. Therefore, either a high percentage of LT records with unexpectedly missing or invalid discharge dates or a very low percentage of LT records with missing discharge dates, regardless of if they are expected, indicate a data quality problem for states. We classified states in which less than 2 percent of LT claims were missing a discharge date as having unusable data.
Table 2. Criteria for DQ assessment of discharge dates
| Percentage of claims with unexpectedly missing or invalid discharge dates | Percentage of claims with missing discharge dates (LT file only) | DQ assessment |
|---|---|---|
| x ≤ 10 percent | x ≥ 2 percent | Low concern |
| 10 percent < x ≤ 20 percent | x ≥ 2 percent | Medium concern |
| 20 percent < x ≤ 50 percent | x ≥ 2 percent | High concern |
| x > 50 percent | x < 2 percent | Unusable |
We did not investigate whether the reported admission and discharge dates were reasonable compared with other dates reported on claims (e.g., service, procedure, or adjudication dates), as doing so requires determining the expected length of stay given the diagnoses and services delivered. This complex undertaking is beyond the scope of this analysis.
Methods previously used to assess data quality
Table 3 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods.
Table 3. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We included all claims with a claim type code for either FFS claims or managed care encounter records (CLM_TYPE_CD equal to 1, 3, A, or C). We did not include capitation payments, supplemental payments, or service tracking payments. We also did not include any records with a claim type of “other,” which are records that the state did not classify as Medicaid or CHIP.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
The count also includes claims with a patient status code of 09 (admitted as an inpatient to this hospital), which occurs only on a very small percentage of states’ claims and tends to be reported on claims on which the discharge dates are not missing.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We included all claims with a claim type code for either FFS claims or managed care encounter records (CLM_TYPE_CD equal to 1, 3, A, or C). We did not include capitation payments, supplemental payments, or service tracking payments. We also did not include any records with a claim type of \u201cother,\u201d which are records that the state did not classify as Medicaid or CHIP.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
The count also includes claims with a patient status code of 09 (admitted as an inpatient to this hospital), which occurs only on a very small percentage of states\u2019 claims and tends to be reported on claims on which the discharge dates are not missing.
Admission and discharge dates indicate when institutional long-term care stays start and end. This analysis examines the extent to which discharge dates are unexpectedly missing or invalid from the LT file.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5101"", ""relatedTopics"": [{""measureId"": 41, ""measureName"": ""Admission Date - IP"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 0}, {""measureId"": 42, ""measureName"": ""Admission Date - LT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 1}, {""measureId"": 43, ""measureName"": ""Discharge Date - IP"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 2}]}" 45,"{""measureId"": 45, ""measureName"": ""Claims Volume - IP"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Claims-Volume-IP.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) consist of data on enrollment and service use for beneficiaries in Medicaid and in the Children’s Health Insurance Program (CHIP). Records of service use, including both fee-for-service (FFS) claims and managed care encounters, are included in the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) files. These claims files are structured so that each record is represented by one header record and one or more line-level records that link to the header. [1] Header records include summary information about the claim as a whole, whereas line records include detailed information about the individual goods and services billed as part of the claim. TAF users must link the header- and line-level records together to get all the information for a service use record.
Since states can choose the optional populations and benefit categories they cover, their Medicaid and CHIP programs vary accordingly to the characteristics of their covered populations, their benefit packages, and their average service use per covered beneficiary. However, examining the volume of service use records adjusted for the number of beneficiaries in a state can identify outlier states that TAF users should examine more closely before beginning their analytic work. An unusually low volume of records may occur, for example, if (1) a state submits incomplete data on FFS claims or managed care encounters or (2) missing or erroneous data elements result in some or all of a claim not being included in the TAF. [2] In these cases, TAF users may underestimate utilization, expenditures, and the prevalence of medical conditions among beneficiaries in a state. An unusually high volume of service use records may indicate problems in how a state formatted or submitted its data.
In contrast to the IP, OT, and RX files, overall claim volume in the LT file can vary substantially without suggesting a data quality issue. Stays in long-term care facilities can last several months or years and generate multiple claims. In some states, institutional long-term care facility claims span a week, whereas in other states they can span up to four weeks. In addition, most states emphasize the use of home and community-based services (HCBS) over institutional long-term care, and success with this strategy varies by state. In most cases, HCBS claims should be reported in the OT file, so states that have most successfully substituted HCBS for institutional long-term care services would be expected to have a lower volume of service use records in their LT file than other states. The availability of institutional long-term care services in each state may also play a role in the overall claims volume. For example, 34 states have certificate-of-need laws, which were designed to limit growth in the number of long-term care facilities. [3]
This data quality assessment examines the volume of service use records in each state for Medicaid and CHIP combined to identify states with potentially incomplete or incorrectly formatted data. [4]
The previous version of the research-ready Medicaid administrative data files, the Medicaid Analytic eXtract (MAX), was structured to include one record per stay in the IP file, one record per claim in the LT and RX files, and one record per claim line in the OT file.
Only final action header records with a known service date and their associated line records are included in the TAF. If data elements related to the service date are missing on a record that a state submits to T-MSIS, the header record and all associated lines would not be included in the TAF. Other exclusions include header and line records not identified by the final action algorithm as representing the final version of the claim and line records that cannot be matched to a header record.
Rahman, M., O. Galarraga, J.S. Zinn, D.C. Grabowski, and V. Mor. “The Impact of Certificate-of-Need Laws on Nursing Home and Home Health Care Expenditures.” Medical Care Research and Review, vol. 73, no. 1, 2015, pp. 85–105.
More information about the completeness of each state’s TAF data can be found in the DQ Atlas single topic displays for Service Users - IP , Service Users - OT , and Service Users - RX .
The previous version of the research-ready Medicaid administrative data files, the Medicaid Analytic eXtract (MAX), was structured to include one record per stay in the IP file, one record per claim in the LT and RX files, and one record per claim line in the OT file.
Only final action header records with a known service date and their associated line records are included in the TAF. If data elements related to the service date are missing on a record that a state submits to T-MSIS, the header record and all associated lines would not be included in the TAF. Other exclusions include header and line records not identified by the final action algorithm as representing the final version of the claim and line records that cannot be matched to a header record.
Rahman, M., O. Galarraga, J.S. Zinn, D.C. Grabowski, and V. Mor. \u201cThe Impact of Certificate-of-Need Laws on Nursing Home and Home Health Care Expenditures.\u201d Medical Care Research and Review, vol. 73, no. 1, 2015, pp. 85\u2013105.
More information about the completeness of each state\u2019s TAF data can be found in the DQ Atlas single topic displays for Service Users - IP , Service Users - OT , and Service Users - RX .
Using the TAF, [5] we tabulated the number of header and non-denied [6] line records in the IP, LT, OT, and RX files. We identified FFS claims and managed care encounters by selecting records with claim type code values of 1, A, 3, and C. We excluded records with other claim type codes, which represent capitation payments, supplemental claims, service tracking claims, or “other” non-program claims. [7] For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [8]
We then calculated the three measures below to understand potential issues with the completeness or quality of the TAF service use data. All three measures identify potential issues with the volume of claims in the TAF. However, for the two measures that are adjusted by number of enrollment months, outliers could be driven by either incomplete claims or incomplete enrollment data. That is, a higher concern level could indicate a data quality problem with either the numerator or the denominator of the metric. [9]
Total volume of header records
For the first measure, we calculated the number of header records per 1,000 enrolled months for each file. [10] As an assessment of the overall completeness of claims and encounter data, this measure identifies states with an unusually low or an unusually high volume of header records compared with other states.
We adjusted the expected volume of service use records according to the size of each state’s Medicaid and CHIP programs combined. We did not require a header record to link to an enrollment record to be included in the analysis (that is, we calculated the numerator and the denominator independent of one another). However, we excluded from the analysis the header records—and their associated line records—that had a missing or invalid beneficiary identifier (MSIS ID). We used the total number of months of Medicaid or CHIP enrollment in 2016 as the denominator for examining the IP, OT, and RX files. For the LT file, we limited the denominator to the number of enrollment months for beneficiaries ages 65 and older. [11]
Total volume of line records
For the second measure, we calculated the number of non-denied line records per 1,000 enrolled months for each file. Line-level volume that appears low relative to the size of a state’s Medicaid and CHIP population can be a sign of incomplete detail on the individual goods and services billed as part of the claim. An unusually high volume may indicate a problem in how the state formatted or submitted its service use records. We used the same denominator that we used for the measure of the total volume of header records: total months of enrollment for the IP, OT, and RX files, and months of enrollment for beneficiaries ages 65 and older for the LT file.
Average number of lines per header
For the third measure, we calculated the average number of non-denied line records per header record in each file. Each header record should have one or more associated line records, and header records with no line records indicate a data quality concern. This measure can identify states in which the header data are complete, but some of the associated line-record data are incomplete. We included all header records in a state’s TAF claims files when we tabulated the header record volume for this measure, including headers that did not link to any line records. Line records that cannot be matched to a header record are not included in the TAF and are therefore not included in this measure.
Overall data quality assessment
To identify states with probable data completeness or quality problems in the IP, OT, and RX files, we compared the three measures to the national median for each measure calculated across all states and territories with available TAF data. We first assessed data quality for each measure based on the extent to which these measures deviated from the national median in the IP, OT, and RX files (Table 1). Then, the overall data assessment for each file was assigned based on the measure with the highest data quality concern. For example, states in which any of the three measures fell below 10 percent of the national median were deemed to have incomplete data that are unusable for analysis.
Table 1. Criteria for DQ assessment of the volume of claims in the IP, OT, and RX files
| Total header volume and total line volume as a percentage of the national median | Average number of line records per header as a percentage of the national median | Record volume | DQ assessment |
|---|---|---|---|
| x < 10 percent | x < 10 percent | Very low | Unusable |
| 10 percent ≤ x < 50 percent | 10 percent ≤ x < 50 percent | Low | High concern |
| 50 percent ≤ x < 75 percent | 50 percent ≤ x ≤ 200 percent | Moderately low | Medium concern |
| 75 percent ≤ x ≤ 150 percent | 50 percent ≤ x ≤ 200 percent | As expected | Low concern |
| 150 percent < x ≤ 200 percent | 50 percent ≤ x ≤ 200 percent | Moderately high | Medium concern |
| x > 200 percent | x > 200 percent | High | High concern |
Note:\tThese criteria apply only to the analysis of the IP, OT, and RX files. We evaluated the LT file solely on the average number of line records per header, given the substantial variation between states in long-term care benefits and utilization patterns.
For the LT file, we calculated the total volume of header and line records adjusted for the number of enrolled months for beneficiaries ages 65 and older, but because we expect high variability across states in long-term care use, we did not base the data quality assessment for the LT file on these measures. The limited data quality assessment we conducted for the LT file is based only on the third measure, average number of lines per header. Using this measure, we classified states into the high-concern or unusable data categories only if the number of lines per header was unusually low, suggesting missing line-level data (Table 2).
Table 2. Criteria for DQ assessment of FFS claims volume in the LT file
| Total header volume and total line volume as a percentage of the national median | Average number of line records per header as a percentage of the national median | Record volume | DQ assessment |
|---|---|---|---|
| Not evaluated | Less than 1 line per header AND ≥ 10 percent of the national median | Low | High concern |
| Not evaluated | < 10 percent of the national median | Very low | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Fully denied claims (also referred to as “denied headers”) are completely excluded from the TAF even when states submit these records in T-MSIS. Thus, all header records in the TAF represent non-denied headers. However, partly denied claims are in the TAF, including the header and both the paid and denied claim lines.
More information on other non-program claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, LT, OT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
More information on states with data quality concerns in their enrollment data can be found in the DQ Atlas single topic display for Total Medicaid and CHIP Enrollment .
We counted enrolled months for each state by tabulating the number of months on each record in the TAF annual Demographic and Eligibility file that had a CHIP code (CHIP_CD) value of 1 (enrolled in Medicaid during the month), 2 (enrolled in Medicaid-expansion CHIP during the month), or 3 (enrolled in Separate CHIP during the month). For analyses of 2014 through 2017 TAF, the enrolled months calculation also includes records with CHIP code value of 4 (enrolled in both Medicaid and S-CHIP during the month), because CHIP code 4 is a valid value for those TAF data years. When the CHIP code was missing, we examined the eligibility group code (ELGBLTY_GRP_CD) and considered it an enrollment month if it was a valid value between 01 and 75.
All Medicaid beneficiaries who have been determined to need institutional long-term care are eligible for these services in Medicaid, including children with disabilities and adults younger than 65. However, for the purpose of this analysis, we limited the denominator to the number of enrolled months for beneficiaries ages 65 and older to exclude the expansion population of non-disabled low-income adults. Because this population makes up a significant proportion of the Medicaid population in states that did expand Medicaid, including them in the denominator is likely to distort the measure of line volume per beneficiary month in the LT file because the expansion population typically does not use LT services.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Fully denied claims (also referred to as \u201cdenied headers\u201d) are completely excluded from the TAF even when states submit these records in T-MSIS. Thus, all header records in the TAF represent non-denied headers. However, partly denied claims are in the TAF, including the header and both the paid and denied claim lines.
More information on other non-program claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, LT, OT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
More information on states with data quality concerns in their enrollment data can be found in the DQ Atlas single topic display for Total Medicaid and CHIP Enrollment .
We counted enrolled months for each state by tabulating the number of months on each record in the TAF annual Demographic and Eligibility file that had a CHIP code (CHIP_CD) value of 1 (enrolled in Medicaid during the month), 2 (enrolled in Medicaid-expansion CHIP during the month), or 3 (enrolled in Separate CHIP during the month). For analyses of 2014 through 2017 TAF, the enrolled months calculation also includes records with CHIP code value of 4 (enrolled in both Medicaid and S-CHIP during the month), because CHIP code 4 is a valid value for those TAF data years. When the CHIP code was missing, we examined the eligibility group code (ELGBLTY_GRP_CD) and considered it an enrollment month if it was a valid value between 01 and 75.
All Medicaid beneficiaries who have been determined to need institutional long-term care are eligible for these services in Medicaid, including children with disabilities and adults younger than 65. However, for the purpose of this analysis, we limited the denominator to the number of enrolled months for beneficiaries ages 65 and older to exclude the expansion population of non-disabled low-income adults. Because this population makes up a significant proportion of the Medicaid population in states that did expand Medicaid, including them in the denominator is likely to distort the measure of line volume per beneficiary month in the LT file because the expansion population typically does not use LT services.
Although states differ in the populations and benefits covered in their Medicaid and CHIP programs, examining the volume of service use records adjusted for program size can identify outlier states that may have incomplete claims, encounter records, or eligibility data in the TAF. This analysis examines the volume of IP header records, the volume of IP line records, and the average number of lines per header.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5111"", ""relatedTopics"": [{""measureId"": 46, ""measureName"": ""Claims Volume - LT"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 1}, {""measureId"": 47, ""measureName"": ""Claims Volume - OT"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 2}, {""measureId"": 48, ""measureName"": ""Claims Volume - RX"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 3}]}" 46,"{""measureId"": 46, ""measureName"": ""Claims Volume - LT"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Claims-Volume-LT.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) consist of data on enrollment and service use for beneficiaries in Medicaid and in the Children’s Health Insurance Program (CHIP). Records of service use, including both fee-for-service (FFS) claims and managed care encounters, are included in the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) files. These claims files are structured so that each record is represented by one header record and one or more line-level records that link to the header. [1] Header records include summary information about the claim as a whole, whereas line records include detailed information about the individual goods and services billed as part of the claim. TAF users must link the header- and line-level records together to get all the information for a service use record.
Since states can choose the optional populations and benefit categories they cover, their Medicaid and CHIP programs vary accordingly to the characteristics of their covered populations, their benefit packages, and their average service use per covered beneficiary. However, examining the volume of service use records adjusted for the number of beneficiaries in a state can identify outlier states that TAF users should examine more closely before beginning their analytic work. An unusually low volume of records may occur, for example, if (1) a state submits incomplete data on FFS claims or managed care encounters or (2) missing or erroneous data elements result in some or all of a claim not being included in the TAF. [2] In these cases, TAF users may underestimate utilization, expenditures, and the prevalence of medical conditions among beneficiaries in a state. An unusually high volume of service use records may indicate problems in how a state formatted or submitted its data.
In contrast to the IP, OT, and RX files, overall claim volume in the LT file can vary substantially without suggesting a data quality issue. Stays in long-term care facilities can last several months or years and generate multiple claims. In some states, institutional long-term care facility claims span a week, whereas in other states they can span up to four weeks. In addition, most states emphasize the use of home and community-based services (HCBS) over institutional long-term care, and success with this strategy varies by state. In most cases, HCBS claims should be reported in the OT file, so states that have most successfully substituted HCBS for institutional long-term care services would be expected to have a lower volume of service use records in their LT file than other states. The availability of institutional long-term care services in each state may also play a role in the overall claims volume. For example, 34 states have certificate-of-need laws, which were designed to limit growth in the number of long-term care facilities. [3]
This data quality assessment examines the volume of service use records in each state for Medicaid and CHIP combined to identify states with potentially incomplete or incorrectly formatted data. [4]
The previous version of the research-ready Medicaid administrative data files, the Medicaid Analytic eXtract (MAX), was structured to include one record per stay in the IP file, one record per claim in the LT and RX files, and one record per claim line in the OT file.
Only final action header records with a known service date and their associated line records are included in the TAF. If data elements related to the service date are missing on a record that a state submits to T-MSIS, the header record and all associated lines would not be included in the TAF. Other exclusions include header and line records not identified by the final action algorithm as representing the final version of the claim and line records that cannot be matched to a header record.
Rahman, M., O. Galarraga, J.S. Zinn, D.C. Grabowski, and V. Mor. “The Impact of Certificate-of-Need Laws on Nursing Home and Home Health Care Expenditures.” Medical Care Research and Review, vol. 73, no. 1, 2015, pp. 85–105.
More information about the completeness of each state’s TAF data can be found in the DQ Atlas single topic displays for Service Users - IP , Service Users - OT , and Service Users - RX .
The previous version of the research-ready Medicaid administrative data files, the Medicaid Analytic eXtract (MAX), was structured to include one record per stay in the IP file, one record per claim in the LT and RX files, and one record per claim line in the OT file.
Only final action header records with a known service date and their associated line records are included in the TAF. If data elements related to the service date are missing on a record that a state submits to T-MSIS, the header record and all associated lines would not be included in the TAF. Other exclusions include header and line records not identified by the final action algorithm as representing the final version of the claim and line records that cannot be matched to a header record.
Rahman, M., O. Galarraga, J.S. Zinn, D.C. Grabowski, and V. Mor. \u201cThe Impact of Certificate-of-Need Laws on Nursing Home and Home Health Care Expenditures.\u201d Medical Care Research and Review, vol. 73, no. 1, 2015, pp. 85\u2013105.
More information about the completeness of each state\u2019s TAF data can be found in the DQ Atlas single topic displays for Service Users - IP , Service Users - OT , and Service Users - RX .
Using the TAF, [5] we tabulated the number of header and non-denied [6] line records in the IP, LT, OT, and RX files. We identified FFS claims and managed care encounters by selecting records with claim type code values of 1, A, 3, and C. We excluded records with other claim type codes, which represent capitation payments, supplemental claims, service tracking claims, or “other” non-program claims. [7] For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [8]
We then calculated the three measures below to understand potential issues with the completeness or quality of the TAF service use data. All three measures identify potential issues with the volume of claims in the TAF. However, for the two measures that are adjusted by number of enrollment months, outliers could be driven by either incomplete claims or incomplete enrollment data. That is, a higher concern level could indicate a data quality problem with either the numerator or the denominator of the metric. [9]
Total volume of header records
For the first measure, we calculated the number of header records per 1,000 enrolled months for each file. [10] As an assessment of the overall completeness of claims and encounter data, this measure identifies states with an unusually low or an unusually high volume of header records compared with other states.
We adjusted the expected volume of service use records according to the size of each state’s Medicaid and CHIP programs combined. We did not require a header record to link to an enrollment record to be included in the analysis (that is, we calculated the numerator and the denominator independent of one another). However, we excluded from the analysis the header records—and their associated line records—that had a missing or invalid beneficiary identifier (MSIS ID). We used the total number of months of Medicaid or CHIP enrollment in 2016 as the denominator for examining the IP, OT, and RX files. For the LT file, we limited the denominator to the number of enrollment months for beneficiaries ages 65 and older. [11]
Total volume of line records
For the second measure, we calculated the number of non-denied line records per 1,000 enrolled months for each file. Line-level volume that appears low relative to the size of a state’s Medicaid and CHIP population can be a sign of incomplete detail on the individual goods and services billed as part of the claim. An unusually high volume may indicate a problem in how the state formatted or submitted its service use records. We used the same denominator that we used for the measure of the total volume of header records: total months of enrollment for the IP, OT, and RX files, and months of enrollment for beneficiaries ages 65 and older for the LT file.
Average number of lines per header
For the third measure, we calculated the average number of non-denied line records per header record in each file. Each header record should have one or more associated line records, and header records with no line records indicate a data quality concern. This measure can identify states in which the header data are complete, but some of the associated line-record data are incomplete. We included all header records in a state’s TAF claims files when we tabulated the header record volume for this measure, including headers that did not link to any line records. Line records that cannot be matched to a header record are not included in the TAF and are therefore not included in this measure.
Overall data quality assessment
To identify states with probable data completeness or quality problems in the IP, OT, and RX files, we compared the three measures to the national median for each measure calculated across all states and territories with available TAF data. We first assessed data quality for each measure based on the extent to which these measures deviated from the national median in the IP, OT, and RX files (Table 1). Then, the overall data assessment for each file was assigned based on the measure with the highest data quality concern. For example, states in which any of the three measures fell below 10 percent of the national median were deemed to have incomplete data that are unusable for analysis.
Table 1. Criteria for DQ assessment of the volume of claims in the IP, OT, and RX files
| Total header volume and total line volume as a percentage of the national median | Average number of line records per header as a percentage of the national median | Record volume | DQ assessment |
|---|---|---|---|
| x < 10 percent | x < 10 percent | Very low | Unusable |
| 10 percent ≤ x < 50 percent | 10 percent ≤ x < 50 percent | Low | High concern |
| 50 percent ≤ x < 75 percent | 50 percent ≤ x ≤ 200 percent | Moderately low | Medium concern |
| 75 percent ≤ x ≤ 150 percent | 50 percent ≤ x ≤ 200 percent | As expected | Low concern |
| 150 percent < x ≤ 200 percent | 50 percent ≤ x ≤ 200 percent | Moderately high | Medium concern |
| x > 200 percent | x > 200 percent | High | High concern |
Note:\tThese criteria apply only to the analysis of the IP, OT, and RX files. We evaluated the LT file solely on the average number of line records per header, given the substantial variation between states in long-term care benefits and utilization patterns.
For the LT file, we calculated the total volume of header and line records adjusted for the number of enrolled months for beneficiaries ages 65 and older, but because we expect high variability across states in long-term care use, we did not base the data quality assessment for the LT file on these measures. The limited data quality assessment we conducted for the LT file is based only on the third measure, average number of lines per header. Using this measure, we classified states into the high-concern or unusable data categories only if the number of lines per header was unusually low, suggesting missing line-level data (Table 2).
Table 2. Criteria for DQ assessment of FFS claims volume in the LT file
| Total header volume and total line volume as a percentage of the national median | Average number of line records per header as a percentage of the national median | Record volume | DQ assessment |
|---|---|---|---|
| Not evaluated | Less than 1 line per header AND ≥ 10 percent of the national median | Low | High concern |
| Not evaluated | < 10 percent of the national median | Very low | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Fully denied claims (also referred to as “denied headers”) are completely excluded from the TAF even when states submit these records in T-MSIS. Thus, all header records in the TAF represent non-denied headers. However, partly denied claims are in the TAF, including the header and both the paid and denied claim lines.
More information on other non-program claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, LT, OT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
More information on states with data quality concerns in their enrollment data can be found in the DQ Atlas single topic display for Total Medicaid and CHIP Enrollment .
We counted enrolled months for each state by tabulating the number of months on each record in the TAF annual Demographic and Eligibility file that had a CHIP code (CHIP_CD) value of 1 (enrolled in Medicaid during the month), 2 (enrolled in Medicaid-expansion CHIP during the month), or 3 (enrolled in Separate CHIP during the month). For analyses of 2014 through 2017 TAF, the enrolled months calculation also includes records with CHIP code value of 4 (enrolled in both Medicaid and S-CHIP during the month), because CHIP code 4 is a valid value for those TAF data years. When the CHIP code was missing, we examined the eligibility group code (ELGBLTY_GRP_CD) and considered it an enrollment month if it was a valid value between 01 and 75.
All Medicaid beneficiaries who have been determined to need institutional long-term care are eligible for these services in Medicaid, including children with disabilities and adults younger than 65. However, for the purpose of this analysis, we limited the denominator to the number of enrolled months for beneficiaries ages 65 and older to exclude the expansion population of non-disabled low-income adults. Because this population makes up a significant proportion of the Medicaid population in states that did expand Medicaid, including them in the denominator is likely to distort the measure of line volume per beneficiary month in the LT file because the expansion population typically does not use LT services.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Fully denied claims (also referred to as \u201cdenied headers\u201d) are completely excluded from the TAF even when states submit these records in T-MSIS. Thus, all header records in the TAF represent non-denied headers. However, partly denied claims are in the TAF, including the header and both the paid and denied claim lines.
More information on other non-program claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, LT, OT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
More information on states with data quality concerns in their enrollment data can be found in the DQ Atlas single topic display for Total Medicaid and CHIP Enrollment .
We counted enrolled months for each state by tabulating the number of months on each record in the TAF annual Demographic and Eligibility file that had a CHIP code (CHIP_CD) value of 1 (enrolled in Medicaid during the month), 2 (enrolled in Medicaid-expansion CHIP during the month), or 3 (enrolled in Separate CHIP during the month). For analyses of 2014 through 2017 TAF, the enrolled months calculation also includes records with CHIP code value of 4 (enrolled in both Medicaid and S-CHIP during the month), because CHIP code 4 is a valid value for those TAF data years. When the CHIP code was missing, we examined the eligibility group code (ELGBLTY_GRP_CD) and considered it an enrollment month if it was a valid value between 01 and 75.
All Medicaid beneficiaries who have been determined to need institutional long-term care are eligible for these services in Medicaid, including children with disabilities and adults younger than 65. However, for the purpose of this analysis, we limited the denominator to the number of enrolled months for beneficiaries ages 65 and older to exclude the expansion population of non-disabled low-income adults. Because this population makes up a significant proportion of the Medicaid population in states that did expand Medicaid, including them in the denominator is likely to distort the measure of line volume per beneficiary month in the LT file because the expansion population typically does not use LT services.
Although states differ in the populations and benefits covered in their Medicaid and CHIP programs, examining the volume of service use records adjusted for program size can identify outlier states that may have incomplete claims, encounter records, or eligibility data in the TAF. This analysis examines the average number of lines per header in the LT file to identify states with patterns that suggest missing LT service use data. The analysis also provides information on LT header and line volume in each state, although the information is not used to assess the quality of a state's data because of variation in state policies on providing and billing for institutional long-term care services.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5111"", ""relatedTopics"": [{""measureId"": 45, ""measureName"": ""Claims Volume - IP"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 0}, {""measureId"": 47, ""measureName"": ""Claims Volume - OT"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 2}, {""measureId"": 48, ""measureName"": ""Claims Volume - RX"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 3}]}" 47,"{""measureId"": 47, ""measureName"": ""Claims Volume - OT"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Claims-Volume-OT.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) consist of data on enrollment and service use for beneficiaries in Medicaid and in the Children’s Health Insurance Program (CHIP). Records of service use, including both fee-for-service (FFS) claims and managed care encounters, are included in the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) files. These claims files are structured so that each record is represented by one header record and one or more line-level records that link to the header. [1] Header records include summary information about the claim as a whole, whereas line records include detailed information about the individual goods and services billed as part of the claim. TAF users must link the header- and line-level records together to get all the information for a service use record.
Since states can choose the optional populations and benefit categories they cover, their Medicaid and CHIP programs vary accordingly to the characteristics of their covered populations, their benefit packages, and their average service use per covered beneficiary. However, examining the volume of service use records adjusted for the number of beneficiaries in a state can identify outlier states that TAF users should examine more closely before beginning their analytic work. An unusually low volume of records may occur, for example, if (1) a state submits incomplete data on FFS claims or managed care encounters or (2) missing or erroneous data elements result in some or all of a claim not being included in the TAF. [2] In these cases, TAF users may underestimate utilization, expenditures, and the prevalence of medical conditions among beneficiaries in a state. An unusually high volume of service use records may indicate problems in how a state formatted or submitted its data.
In contrast to the IP, OT, and RX files, overall claim volume in the LT file can vary substantially without suggesting a data quality issue. Stays in long-term care facilities can last several months or years and generate multiple claims. In some states, institutional long-term care facility claims span a week, whereas in other states they can span up to four weeks. In addition, most states emphasize the use of home and community-based services (HCBS) over institutional long-term care, and success with this strategy varies by state. In most cases, HCBS claims should be reported in the OT file, so states that have most successfully substituted HCBS for institutional long-term care services would be expected to have a lower volume of service use records in their LT file than other states. The availability of institutional long-term care services in each state may also play a role in the overall claims volume. For example, 34 states have certificate-of-need laws, which were designed to limit growth in the number of long-term care facilities. [3]
This data quality assessment examines the volume of service use records in each state for Medicaid and CHIP combined to identify states with potentially incomplete or incorrectly formatted data. [4]
The previous version of the research-ready Medicaid administrative data files, the Medicaid Analytic eXtract (MAX), was structured to include one record per stay in the IP file, one record per claim in the LT and RX files, and one record per claim line in the OT file.
Only final action header records with a known service date and their associated line records are included in the TAF. If data elements related to the service date are missing on a record that a state submits to T-MSIS, the header record and all associated lines would not be included in the TAF. Other exclusions include header and line records not identified by the final action algorithm as representing the final version of the claim and line records that cannot be matched to a header record.
Rahman, M., O. Galarraga, J.S. Zinn, D.C. Grabowski, and V. Mor. “The Impact of Certificate-of-Need Laws on Nursing Home and Home Health Care Expenditures.” Medical Care Research and Review, vol. 73, no. 1, 2015, pp. 85–105.
More information about the completeness of each state’s TAF data can be found in the DQ Atlas single topic displays for Service Users - IP , Service Users - OT , and Service Users - RX .
The previous version of the research-ready Medicaid administrative data files, the Medicaid Analytic eXtract (MAX), was structured to include one record per stay in the IP file, one record per claim in the LT and RX files, and one record per claim line in the OT file.
Only final action header records with a known service date and their associated line records are included in the TAF. If data elements related to the service date are missing on a record that a state submits to T-MSIS, the header record and all associated lines would not be included in the TAF. Other exclusions include header and line records not identified by the final action algorithm as representing the final version of the claim and line records that cannot be matched to a header record.
Rahman, M., O. Galarraga, J.S. Zinn, D.C. Grabowski, and V. Mor. \u201cThe Impact of Certificate-of-Need Laws on Nursing Home and Home Health Care Expenditures.\u201d Medical Care Research and Review, vol. 73, no. 1, 2015, pp. 85\u2013105.
More information about the completeness of each state\u2019s TAF data can be found in the DQ Atlas single topic displays for Service Users - IP , Service Users - OT , and Service Users - RX .
Using the TAF, [5] we tabulated the number of header and non-denied [6] line records in the IP, LT, OT, and RX files. We identified FFS claims and managed care encounters by selecting records with claim type code values of 1, A, 3, and C. We excluded records with other claim type codes, which represent capitation payments, supplemental claims, service tracking claims, or “other” non-program claims. [7] For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [8]
We then calculated the three measures below to understand potential issues with the completeness or quality of the TAF service use data. All three measures identify potential issues with the volume of claims in the TAF. However, for the two measures that are adjusted by number of enrollment months, outliers could be driven by either incomplete claims or incomplete enrollment data. That is, a higher concern level could indicate a data quality problem with either the numerator or the denominator of the metric. [9]
Total volume of header records
For the first measure, we calculated the number of header records per 1,000 enrolled months for each file. [10] As an assessment of the overall completeness of claims and encounter data, this measure identifies states with an unusually low or an unusually high volume of header records compared with other states.
We adjusted the expected volume of service use records according to the size of each state’s Medicaid and CHIP programs combined. We did not require a header record to link to an enrollment record to be included in the analysis (that is, we calculated the numerator and the denominator independent of one another). However, we excluded from the analysis the header records—and their associated line records—that had a missing or invalid beneficiary identifier (MSIS ID). We used the total number of months of Medicaid or CHIP enrollment in 2016 as the denominator for examining the IP, OT, and RX files. For the LT file, we limited the denominator to the number of enrollment months for beneficiaries ages 65 and older. [11]
Total volume of line records
For the second measure, we calculated the number of non-denied line records per 1,000 enrolled months for each file. Line-level volume that appears low relative to the size of a state’s Medicaid and CHIP population can be a sign of incomplete detail on the individual goods and services billed as part of the claim. An unusually high volume may indicate a problem in how the state formatted or submitted its service use records. We used the same denominator that we used for the measure of the total volume of header records: total months of enrollment for the IP, OT, and RX files, and months of enrollment for beneficiaries ages 65 and older for the LT file.
Average number of lines per header
For the third measure, we calculated the average number of non-denied line records per header record in each file. Each header record should have one or more associated line records, and header records with no line records indicate a data quality concern. This measure can identify states in which the header data are complete, but some of the associated line-record data are incomplete. We included all header records in a state’s TAF claims files when we tabulated the header record volume for this measure, including headers that did not link to any line records. Line records that cannot be matched to a header record are not included in the TAF and are therefore not included in this measure.
Overall data quality assessment
To identify states with probable data completeness or quality problems in the IP, OT, and RX files, we compared the three measures to the national median for each measure calculated across all states and territories with available TAF data. We first assessed data quality for each measure based on the extent to which these measures deviated from the national median in the IP, OT, and RX files (Table 1). Then, the overall data assessment for each file was assigned based on the measure with the highest data quality concern. For example, states in which any of the three measures fell below 10 percent of the national median were deemed to have incomplete data that are unusable for analysis.
Table 1. Criteria for DQ assessment of the volume of claims in the IP, OT, and RX files
| Total header volume and total line volume as a percentage of the national median | Average number of line records per header as a percentage of the national median | Record volume | DQ assessment |
|---|---|---|---|
| x < 10 percent | x < 10 percent | Very low | Unusable |
| 10 percent ≤ x < 50 percent | 10 percent ≤ x < 50 percent | Low | High concern |
| 50 percent ≤ x < 75 percent | 50 percent ≤ x ≤ 200 percent | Moderately low | Medium concern |
| 75 percent ≤ x ≤ 150 percent | 50 percent ≤ x ≤ 200 percent | As expected | Low concern |
| 150 percent < x ≤ 200 percent | 50 percent ≤ x ≤ 200 percent | Moderately high | Medium concern |
| x > 200 percent | x > 200 percent | High | High concern |
Note:\tThese criteria apply only to the analysis of the IP, OT, and RX files. We evaluated the LT file solely on the average number of line records per header, given the substantial variation between states in long-term care benefits and utilization patterns.
For the LT file, we calculated the total volume of header and line records adjusted for the number of enrolled months for beneficiaries ages 65 and older, but because we expect high variability across states in long-term care use, we did not base the data quality assessment for the LT file on these measures. The limited data quality assessment we conducted for the LT file is based only on the third measure, average number of lines per header. Using this measure, we classified states into the high-concern or unusable data categories only if the number of lines per header was unusually low, suggesting missing line-level data (Table 2).
Table 2. Criteria for DQ assessment of FFS claims volume in the LT file
| Total header volume and total line volume as a percentage of the national median | Average number of line records per header as a percentage of the national median | Record volume | DQ assessment |
|---|---|---|---|
| Not evaluated | Less than 1 line per header AND ≥ 10 percent of the national median | Low | High concern |
| Not evaluated | < 10 percent of the national median | Very low | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Fully denied claims (also referred to as “denied headers”) are completely excluded from the TAF even when states submit these records in T-MSIS. Thus, all header records in the TAF represent non-denied headers. However, partly denied claims are in the TAF, including the header and both the paid and denied claim lines.
More information on other non-program claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, LT, OT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
More information on states with data quality concerns in their enrollment data can be found in the DQ Atlas single topic display for Total Medicaid and CHIP Enrollment .
We counted enrolled months for each state by tabulating the number of months on each record in the TAF annual Demographic and Eligibility file that had a CHIP code (CHIP_CD) value of 1 (enrolled in Medicaid during the month), 2 (enrolled in Medicaid-expansion CHIP during the month), or 3 (enrolled in Separate CHIP during the month). For analyses of 2014 through 2017 TAF, the enrolled months calculation also includes records with CHIP code value of 4 (enrolled in both Medicaid and S-CHIP during the month), because CHIP code 4 is a valid value for those TAF data years. When the CHIP code was missing, we examined the eligibility group code (ELGBLTY_GRP_CD) and considered it an enrollment month if it was a valid value between 01 and 75.
All Medicaid beneficiaries who have been determined to need institutional long-term care are eligible for these services in Medicaid, including children with disabilities and adults younger than 65. However, for the purpose of this analysis, we limited the denominator to the number of enrolled months for beneficiaries ages 65 and older to exclude the expansion population of non-disabled low-income adults. Because this population makes up a significant proportion of the Medicaid population in states that did expand Medicaid, including them in the denominator is likely to distort the measure of line volume per beneficiary month in the LT file because the expansion population typically does not use LT services.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Fully denied claims (also referred to as \u201cdenied headers\u201d) are completely excluded from the TAF even when states submit these records in T-MSIS. Thus, all header records in the TAF represent non-denied headers. However, partly denied claims are in the TAF, including the header and both the paid and denied claim lines.
More information on other non-program claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, LT, OT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
More information on states with data quality concerns in their enrollment data can be found in the DQ Atlas single topic display for Total Medicaid and CHIP Enrollment .
We counted enrolled months for each state by tabulating the number of months on each record in the TAF annual Demographic and Eligibility file that had a CHIP code (CHIP_CD) value of 1 (enrolled in Medicaid during the month), 2 (enrolled in Medicaid-expansion CHIP during the month), or 3 (enrolled in Separate CHIP during the month). For analyses of 2014 through 2017 TAF, the enrolled months calculation also includes records with CHIP code value of 4 (enrolled in both Medicaid and S-CHIP during the month), because CHIP code 4 is a valid value for those TAF data years. When the CHIP code was missing, we examined the eligibility group code (ELGBLTY_GRP_CD) and considered it an enrollment month if it was a valid value between 01 and 75.
All Medicaid beneficiaries who have been determined to need institutional long-term care are eligible for these services in Medicaid, including children with disabilities and adults younger than 65. However, for the purpose of this analysis, we limited the denominator to the number of enrolled months for beneficiaries ages 65 and older to exclude the expansion population of non-disabled low-income adults. Because this population makes up a significant proportion of the Medicaid population in states that did expand Medicaid, including them in the denominator is likely to distort the measure of line volume per beneficiary month in the LT file because the expansion population typically does not use LT services.
Although states differ in the populations and benefits covered in their Medicaid and CHIP programs, examining the volume of service use records adjusted for program size can identify outlier states that may have incomplete claims, encounter records, or eligibility data in the TAF. This analysis examines the volume of OT header records, the volume of OT line records, and the average number of lines per header.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5111"", ""relatedTopics"": [{""measureId"": 45, ""measureName"": ""Claims Volume - IP"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 0}, {""measureId"": 46, ""measureName"": ""Claims Volume - LT"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 1}, {""measureId"": 48, ""measureName"": ""Claims Volume - RX"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 3}]}" 48,"{""measureId"": 48, ""measureName"": ""Claims Volume - RX"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Claims-Volume-RX.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) consist of data on enrollment and service use for beneficiaries in Medicaid and in the Children’s Health Insurance Program (CHIP). Records of service use, including both fee-for-service (FFS) claims and managed care encounters, are included in the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) files. These claims files are structured so that each record is represented by one header record and one or more line-level records that link to the header. [1] Header records include summary information about the claim as a whole, whereas line records include detailed information about the individual goods and services billed as part of the claim. TAF users must link the header- and line-level records together to get all the information for a service use record.
Since states can choose the optional populations and benefit categories they cover, their Medicaid and CHIP programs vary accordingly to the characteristics of their covered populations, their benefit packages, and their average service use per covered beneficiary. However, examining the volume of service use records adjusted for the number of beneficiaries in a state can identify outlier states that TAF users should examine more closely before beginning their analytic work. An unusually low volume of records may occur, for example, if (1) a state submits incomplete data on FFS claims or managed care encounters or (2) missing or erroneous data elements result in some or all of a claim not being included in the TAF. [2] In these cases, TAF users may underestimate utilization, expenditures, and the prevalence of medical conditions among beneficiaries in a state. An unusually high volume of service use records may indicate problems in how a state formatted or submitted its data.
In contrast to the IP, OT, and RX files, overall claim volume in the LT file can vary substantially without suggesting a data quality issue. Stays in long-term care facilities can last several months or years and generate multiple claims. In some states, institutional long-term care facility claims span a week, whereas in other states they can span up to four weeks. In addition, most states emphasize the use of home and community-based services (HCBS) over institutional long-term care, and success with this strategy varies by state. In most cases, HCBS claims should be reported in the OT file, so states that have most successfully substituted HCBS for institutional long-term care services would be expected to have a lower volume of service use records in their LT file than other states. The availability of institutional long-term care services in each state may also play a role in the overall claims volume. For example, 34 states have certificate-of-need laws, which were designed to limit growth in the number of long-term care facilities. [3]
This data quality assessment examines the volume of service use records in each state for Medicaid and CHIP combined to identify states with potentially incomplete or incorrectly formatted data. [4]
The previous version of the research-ready Medicaid administrative data files, the Medicaid Analytic eXtract (MAX), was structured to include one record per stay in the IP file, one record per claim in the LT and RX files, and one record per claim line in the OT file.
Only final action header records with a known service date and their associated line records are included in the TAF. If data elements related to the service date are missing on a record that a state submits to T-MSIS, the header record and all associated lines would not be included in the TAF. Other exclusions include header and line records not identified by the final action algorithm as representing the final version of the claim and line records that cannot be matched to a header record.
Rahman, M., O. Galarraga, J.S. Zinn, D.C. Grabowski, and V. Mor. “The Impact of Certificate-of-Need Laws on Nursing Home and Home Health Care Expenditures.” Medical Care Research and Review, vol. 73, no. 1, 2015, pp. 85–105.
More information about the completeness of each state’s TAF data can be found in the DQ Atlas single topic displays for Service Users - IP , Service Users - OT , and Service Users - RX .
The previous version of the research-ready Medicaid administrative data files, the Medicaid Analytic eXtract (MAX), was structured to include one record per stay in the IP file, one record per claim in the LT and RX files, and one record per claim line in the OT file.
Only final action header records with a known service date and their associated line records are included in the TAF. If data elements related to the service date are missing on a record that a state submits to T-MSIS, the header record and all associated lines would not be included in the TAF. Other exclusions include header and line records not identified by the final action algorithm as representing the final version of the claim and line records that cannot be matched to a header record.
Rahman, M., O. Galarraga, J.S. Zinn, D.C. Grabowski, and V. Mor. \u201cThe Impact of Certificate-of-Need Laws on Nursing Home and Home Health Care Expenditures.\u201d Medical Care Research and Review, vol. 73, no. 1, 2015, pp. 85\u2013105.
More information about the completeness of each state\u2019s TAF data can be found in the DQ Atlas single topic displays for Service Users - IP , Service Users - OT , and Service Users - RX .
Using the TAF, [5] we tabulated the number of header and non-denied [6] line records in the IP, LT, OT, and RX files. We identified FFS claims and managed care encounters by selecting records with claim type code values of 1, A, 3, and C. We excluded records with other claim type codes, which represent capitation payments, supplemental claims, service tracking claims, or “other” non-program claims. [7] For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [8]
We then calculated the three measures below to understand potential issues with the completeness or quality of the TAF service use data. All three measures identify potential issues with the volume of claims in the TAF. However, for the two measures that are adjusted by number of enrollment months, outliers could be driven by either incomplete claims or incomplete enrollment data. That is, a higher concern level could indicate a data quality problem with either the numerator or the denominator of the metric. [9]
Total volume of header records
For the first measure, we calculated the number of header records per 1,000 enrolled months for each file. [10] As an assessment of the overall completeness of claims and encounter data, this measure identifies states with an unusually low or an unusually high volume of header records compared with other states.
We adjusted the expected volume of service use records according to the size of each state’s Medicaid and CHIP programs combined. We did not require a header record to link to an enrollment record to be included in the analysis (that is, we calculated the numerator and the denominator independent of one another). However, we excluded from the analysis the header records—and their associated line records—that had a missing or invalid beneficiary identifier (MSIS ID). We used the total number of months of Medicaid or CHIP enrollment in 2016 as the denominator for examining the IP, OT, and RX files. For the LT file, we limited the denominator to the number of enrollment months for beneficiaries ages 65 and older. [11]
Total volume of line records
For the second measure, we calculated the number of non-denied line records per 1,000 enrolled months for each file. Line-level volume that appears low relative to the size of a state’s Medicaid and CHIP population can be a sign of incomplete detail on the individual goods and services billed as part of the claim. An unusually high volume may indicate a problem in how the state formatted or submitted its service use records. We used the same denominator that we used for the measure of the total volume of header records: total months of enrollment for the IP, OT, and RX files, and months of enrollment for beneficiaries ages 65 and older for the LT file.
Average number of lines per header
For the third measure, we calculated the average number of non-denied line records per header record in each file. Each header record should have one or more associated line records, and header records with no line records indicate a data quality concern. This measure can identify states in which the header data are complete, but some of the associated line-record data are incomplete. We included all header records in a state’s TAF claims files when we tabulated the header record volume for this measure, including headers that did not link to any line records. Line records that cannot be matched to a header record are not included in the TAF and are therefore not included in this measure.
Overall data quality assessment
To identify states with probable data completeness or quality problems in the IP, OT, and RX files, we compared the three measures to the national median for each measure calculated across all states and territories with available TAF data. We first assessed data quality for each measure based on the extent to which these measures deviated from the national median in the IP, OT, and RX files (Table 1). Then, the overall data assessment for each file was assigned based on the measure with the highest data quality concern. For example, states in which any of the three measures fell below 10 percent of the national median were deemed to have incomplete data that are unusable for analysis.
Table 1. Criteria for DQ assessment of the volume of claims in the IP, OT, and RX files
| Total header volume and total line volume as a percentage of the national median | Average number of line records per header as a percentage of the national median | Record volume | DQ assessment |
|---|---|---|---|
| x < 10 percent | x < 10 percent | Very low | Unusable |
| 10 percent ≤ x < 50 percent | 10 percent ≤ x < 50 percent | Low | High concern |
| 50 percent ≤ x < 75 percent | 50 percent ≤ x ≤ 200 percent | Moderately low | Medium concern |
| 75 percent ≤ x ≤ 150 percent | 50 percent ≤ x ≤ 200 percent | As expected | Low concern |
| 150 percent < x ≤ 200 percent | 50 percent ≤ x ≤ 200 percent | Moderately high | Medium concern |
| x > 200 percent | x > 200 percent | High | High concern |
Note:\tThese criteria apply only to the analysis of the IP, OT, and RX files. We evaluated the LT file solely on the average number of line records per header, given the substantial variation between states in long-term care benefits and utilization patterns.
For the LT file, we calculated the total volume of header and line records adjusted for the number of enrolled months for beneficiaries ages 65 and older, but because we expect high variability across states in long-term care use, we did not base the data quality assessment for the LT file on these measures. The limited data quality assessment we conducted for the LT file is based only on the third measure, average number of lines per header. Using this measure, we classified states into the high-concern or unusable data categories only if the number of lines per header was unusually low, suggesting missing line-level data (Table 2).
Table 2. Criteria for DQ assessment of FFS claims volume in the LT file
| Total header volume and total line volume as a percentage of the national median | Average number of line records per header as a percentage of the national median | Record volume | DQ assessment |
|---|---|---|---|
| Not evaluated | Less than 1 line per header AND ≥ 10 percent of the national median | Low | High concern |
| Not evaluated | < 10 percent of the national median | Very low | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Fully denied claims (also referred to as “denied headers”) are completely excluded from the TAF even when states submit these records in T-MSIS. Thus, all header records in the TAF represent non-denied headers. However, partly denied claims are in the TAF, including the header and both the paid and denied claim lines.
More information on other non-program claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, LT, OT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
More information on states with data quality concerns in their enrollment data can be found in the DQ Atlas single topic display for Total Medicaid and CHIP Enrollment .
We counted enrolled months for each state by tabulating the number of months on each record in the TAF annual Demographic and Eligibility file that had a CHIP code (CHIP_CD) value of 1 (enrolled in Medicaid during the month), 2 (enrolled in Medicaid-expansion CHIP during the month), or 3 (enrolled in Separate CHIP during the month). For analyses of 2014 through 2017 TAF, the enrolled months calculation also includes records with CHIP code value of 4 (enrolled in both Medicaid and S-CHIP during the month), because CHIP code 4 is a valid value for those TAF data years. When the CHIP code was missing, we examined the eligibility group code (ELGBLTY_GRP_CD) and considered it an enrollment month if it was a valid value between 01 and 75.
All Medicaid beneficiaries who have been determined to need institutional long-term care are eligible for these services in Medicaid, including children with disabilities and adults younger than 65. However, for the purpose of this analysis, we limited the denominator to the number of enrolled months for beneficiaries ages 65 and older to exclude the expansion population of non-disabled low-income adults. Because this population makes up a significant proportion of the Medicaid population in states that did expand Medicaid, including them in the denominator is likely to distort the measure of line volume per beneficiary month in the LT file because the expansion population typically does not use LT services.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Fully denied claims (also referred to as \u201cdenied headers\u201d) are completely excluded from the TAF even when states submit these records in T-MSIS. Thus, all header records in the TAF represent non-denied headers. However, partly denied claims are in the TAF, including the header and both the paid and denied claim lines.
More information on other non-program claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, LT, OT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
More information on states with data quality concerns in their enrollment data can be found in the DQ Atlas single topic display for Total Medicaid and CHIP Enrollment .
We counted enrolled months for each state by tabulating the number of months on each record in the TAF annual Demographic and Eligibility file that had a CHIP code (CHIP_CD) value of 1 (enrolled in Medicaid during the month), 2 (enrolled in Medicaid-expansion CHIP during the month), or 3 (enrolled in Separate CHIP during the month). For analyses of 2014 through 2017 TAF, the enrolled months calculation also includes records with CHIP code value of 4 (enrolled in both Medicaid and S-CHIP during the month), because CHIP code 4 is a valid value for those TAF data years. When the CHIP code was missing, we examined the eligibility group code (ELGBLTY_GRP_CD) and considered it an enrollment month if it was a valid value between 01 and 75.
All Medicaid beneficiaries who have been determined to need institutional long-term care are eligible for these services in Medicaid, including children with disabilities and adults younger than 65. However, for the purpose of this analysis, we limited the denominator to the number of enrolled months for beneficiaries ages 65 and older to exclude the expansion population of non-disabled low-income adults. Because this population makes up a significant proportion of the Medicaid population in states that did expand Medicaid, including them in the denominator is likely to distort the measure of line volume per beneficiary month in the LT file because the expansion population typically does not use LT services.
Although states differ in the populations and benefits covered in their Medicaid and CHIP programs, examining the volume of service use records adjusted for program size can identify outlier states that may have incomplete claims, encounter records, or eligibility data in TAF. This analysis examines the volume of RX header records, the volume of RX line records, and the average number of lines per header.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5111"", ""relatedTopics"": [{""measureId"": 45, ""measureName"": ""Claims Volume - IP"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 0}, {""measureId"": 46, ""measureName"": ""Claims Volume - LT"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 1}, {""measureId"": 47, ""measureName"": ""Claims Volume - OT"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 2}]}" 49,"{""measureId"": 49, ""measureName"": ""Diagnosis Code - IP"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Diagnosis-Cd-IP.pdf"", ""background"": {""content"": ""Providers submit diagnosis codes on medical claims that can be used to analyze the prevalence of different medical conditions. Both institutional providers (such as hospitals, nursing facilities, and clinics) and professional providers (such as physicians, other clinical professionals, and ambulances) submit claims on standardized forms that allow them to record multiple diagnosis codes. [1] Diagnosis codes are not, however, submitted on pharmacy claims. Before October 1, 2015, providers used diagnosis codes from the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM). For services provided on or after October 1, 2015, providers are expected to use diagnosis codes from the International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM), which has nearly five times as many possible codes at a far greater level of detail than did the ICD-9-CM.
In the T-MSIS Analytic Files (TAF), all diagnosis codes are located on the header record, which contains information that relates to the claim as a whole. The maximum number of diagnosis codes for a single claim in the TAF varies by file. TAF inpatient (IP) records may have up to 12 diagnosis codes per claim, long-term care (LT) records may have up to 5 diagnosis codes per claim, and other services (OT) records may have up to 2 diagnosis codes per claim. [2] Complete and accurate diagnosis code information is essential, as these fields help TAF users to better understand the underlying health conditions and needs of the populations they are studying. [3]
For the TAF, we expect all claims in the IP and LT claims files to have a valid ICD-10-CM diagnosis code in the primary diagnosis code field (DGNS_1_CD). [4] For the OT claims file, we do not expect that all claims will have diagnosis codes. For certain types of services (for example, medical supplies, prosthetic equipment, or non-emergency medical transportation [NEMT] services), states may not require providers to submit diagnosis codes on the claims because the provider billing for these services may not be in the best position to know and record an accurate diagnosis. Many states also do not require providers to submit diagnosis codes on dental claims. However, if the provider includes any diagnosis codes on those claims, CMS instructs states to pass through those diagnosis codes to T-MSIS as reported on the claims form even if the diagnosis codes are inaccurate or invalid. For this reason, TAF users may want to exercise caution in using diagnosis codes on certain types of OT claims even if they appear to be valid ICD-10 codes.
Institutional claims are often referred to as “UB-04 claims” when submitted in paper form or as “837I claims” when submitted in electronic form. Professional claims are referred to as “CMS-1500 claims” when submitted in paper form or “837P” when submitted in electronic form. All of these formats allow providers to include multiple diagnosis codes on a given claim.
In the IP file, diagnosis codes are captured in the fields DGNS_1_CD through DGNS_12_CD. In the LT file, diagnosis codes are captured in the fields DGNS_1_CD through DGNS_5_CD. In the OT file, diagnosis codes are captured in the fields DGNS_1_CD and DGNS_2_CD. In all three claims files, the first diagnosis code field (DGNS_1_CD) contains the primary diagnosis code, and the additional fields contain other diagnosis codes submitted on the claim. The IP and LT files also include a separate field for the admitting diagnosis code (ADMTG_DGNS_CD), which captures the diagnosis code recorded by the provider when a patient is admitted to a facility. If a claim includes more diagnosis codes than are accommodated in T-MSIS, the source data for TAF, for its given file type, then some diagnosis code information may be missing from the T-MSIS and TAF data systems. This is most likely to be an issue in the OT file because states can only report up to two diagnosis codes per claim in their T-MSIS OT files.
Although diagnosis codes typically capture beneficiaries’ medical conditions, diagnosis codes are sometimes used to provide additional information about the services received during a medical encounter. For example, on evaluation and management claims, diagnosis codes can capture the patient’s age or whether he or she required an examination to participate in sports or to begin a new job.
Some states may not require nursing facilities to report diagnosis codes on their claims if the diagnoses are tracked separately through the federally required Preadmission Screening and Resident Review process or within a different clinical information system. In these cases, the data may be missing from the T-MSIS and TAF claims files.
Institutional claims are often referred to as \u201cUB-04 claims\u201d when submitted in paper form or as \u201c837I claims\u201d when submitted in electronic form. Professional claims are referred to as \u201cCMS-1500 claims\u201d when submitted in paper form or \u201c837P\u201d when submitted in electronic form. All of these formats allow providers to include multiple diagnosis codes on a given claim.
In the IP file, diagnosis codes are captured in the fields DGNS_1_CD through DGNS_12_CD. In the LT file, diagnosis codes are captured in the fields DGNS_1_CD through DGNS_5_CD. In the OT file, diagnosis codes are captured in the fields DGNS_1_CD and DGNS_2_CD. In all three claims files, the first diagnosis code field (DGNS_1_CD) contains the primary diagnosis code, and the additional fields contain other diagnosis codes submitted on the claim. The IP and LT files also include a separate field for the admitting diagnosis code (ADMTG_DGNS_CD), which captures the diagnosis code recorded by the provider when a patient is admitted to a facility. If a claim includes more diagnosis codes than are accommodated in T-MSIS, the source data for TAF, for its given file type, then some diagnosis code information may be missing from the T-MSIS and TAF data systems. This is most likely to be an issue in the OT file because states can only report up to two diagnosis codes per claim in their T-MSIS OT files.
Although diagnosis codes typically capture beneficiaries\u2019 medical conditions, diagnosis codes are sometimes used to provide additional information about the services received during a medical encounter. For example, on evaluation and management claims, diagnosis codes can capture the patient\u2019s age or whether he or she required an examination to participate in sports or to begin a new job.
Some states may not require nursing facilities to report diagnosis codes on their claims if the diagnoses are tracked separately through the federally required Preadmission Screening and Resident Review process or within a different clinical information system. In these cases, the data may be missing from the T-MSIS and TAF claims files.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
This analysis is based on the TAF. [5] We assessed the extent to which diagnosis codes are available on fee-for-service (FFS) claims and on managed care encounter records in the IP, LT, and OT claims files. [6] We excluded crossover claims (those for which Medicare is the primary payer, and Medicaid is responsible only for covering the remaining cost-sharing on behalf of dually eligible beneficiaries). We did not examine the pharmacy claims file because diagnosis codes are not reported on pharmacy claims.
We calculated the percentage of header records in each claims file that had a valid ICD-10 diagnosis code in the field for the primary diagnosis code (DGNS_1_CD). [7] We excluded states from the analyses that had a low volume of claims in the TAF, rendering the data unusable for analysis. For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [8] If a valid ICD-10 diagnosis code was not available, we calculated the percentage of records for which the field (1) was missing, (2) had an ICD-9 diagnosis code, [9] or (3) had another non-missing but invalid value. Because we examined claims for services provided after the transition from ICD-9 to ICD-10, we considered only the ICD-10 diagnosis codes to be “valid” values. We classified all ICD-9 codes as “invalid,” although TAF users may be able to use those codes for certain analyses. Invalid codes may reflect difficulties that the states had in extracting or reporting the data for these fields in their T-MSIS submissions. [10] Alternatively, the invalid codes could reflect providers’ errors in entering the diagnosis codes on claims. [11]
We assessed diagnosis code quality in each state based on the percentage of records that had a valid ICD-10 primary diagnosis code (Table 1). [12]
Table 1. Criteria for DQ assessment of diagnosis code
For the OT claims file only, we calculated the measure for just the subset of claims that included outpatient hospital services, physician services, or clinic services. [13] For the IP and LT files, we calculated the measure with all claims. We also calculated the mean number of unique, valid ICD-10 diagnosis codes available per claim for the IP and LT files. [14] Although we did not use this measure to assess diagnosis code data quality, a mean number of diagnosis codes that is substantially lower than that of other states may suggest that a state’s diagnosis code information is incomplete or indicate variation in state billing practices. We did not assess the appropriateness of the available diagnosis codes against other information on the claims (such as procedure codes) because this complex undertaking is beyond the scope of this analysis.
Methods previously used to assess data quality
Table 2 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods.
Table 2. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We included all claims with a claim type code for either FFS claims or managed care encounter records (CLM_TYPE_CD equal to 1, 3, A, or C). We did not include capitation payments, supplemental payments, or service tracking payments. We also did not include any records with a claim type of “other,” which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Title XIX or Title XXI.
The ICD-10-CM code set is updated each year, often during the middle of the year. For this analysis, we considered codes to be valid ICD-10 diagnosis codes if they were included in either ICD-10-CM that was effective during the year.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
A small number of diagnosis codes are both an ICD-9 diagnosis code and an ICD-10 diagnosis code. For the purposes of our analysis, we classified those codes as valid ICD-10 diagnosis codes, not as ICD-9 codes.
In some cases, the invalid code could be a valid ICD-10 or ICD-9 diagnosis code if it had been submitted with an additional first or final character. An example of an invalid code in the IP file is F1023. Although this code is not a valid ICD-9 or ICD-10 diagnosis code, adding a final character to the submitted code creates four possible valid ICD-10 diagnosis codes: F10230 (alcohol dependence with withdrawal, uncomplicated), F10231 (alcohol dependence with withdrawal delirium), F10232 (alcohol dependence with withdrawal with perceptual disturbance), or F10239 (alcohol dependence with withdrawal, unspecified). Similarly, an example of an invalid code in the OT file is 2990. Although this code is not an ICD-9 or ICD-10 diagnosis code, it could form several possible ICD-9 or ICD-10 codes with an additional first or final character.
If the invalid codes are the result of provider errors, and if those errors did not prevent the state or the managed care plan from adjudicating and paying the claim, then it is consistent with T-MSIS coding guidance for states to pass through paid claims with the invalid codes to T-MSIS.
For the OT file, we used the measure calculated with just the subset of claims for outpatient hospital services, physician services, or clinic services to group states into categories of low, medium, and high data quality concern. For the IP and LT files, we used the measures calculated with all claims to assess the states’ data quality.
We included OT claims in this subset if at least one of the claim lines had a federally assigned service category (FASC) equal to 26 (Outpatient hospital), 38 (Physician and all other professional claims), or 27 (Clinic). Claims with these types of services frequently appear to be the focus of in-depth research on service utilization patterns and are expected to have valid ICD-10 diagnosis codes. For more information about the FASC code, see the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
For the IP file, we counted the number of unique, valid ICD-10 diagnosis codes in the fields DGNS_1_CD through DGNS_12_CD for each claim. For the LT file, we counted the number of unique, valid ICD-10 diagnosis codes in the fields DGNS_1_CD through DGNS_5_CD for each claim. We then divided these counts by the total number of IP claims or LT claims, respectively. This accounts for the fact that some states have non-unique diagnosis codes in multiple fields.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We included all claims with a claim type code for either FFS claims or managed care encounter records (CLM_TYPE_CD equal to 1, 3, A, or C). We did not include capitation payments, supplemental payments, or service tracking payments. We also did not include any records with a claim type of \u201cother,\u201d which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Title XIX or Title XXI.
The ICD-10-CM code set is updated each year, often during the middle of the year. For this analysis, we considered codes to be valid ICD-10 diagnosis codes if they were included in either ICD-10-CM that was effective during the year.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
A small number of diagnosis codes are both an ICD-9 diagnosis code and an ICD-10 diagnosis code. For the purposes of our analysis, we classified those codes as valid ICD-10 diagnosis codes, not as ICD-9 codes.
In some cases, the invalid code could be a valid ICD-10 or ICD-9 diagnosis code if it had been submitted with an additional first or final character. An example of an invalid code in the IP file is F1023. Although this code is not a valid ICD-9 or ICD-10 diagnosis code, adding a final character to the submitted code creates four possible valid ICD-10 diagnosis codes: F10230 (alcohol dependence with withdrawal, uncomplicated), F10231 (alcohol dependence with withdrawal delirium), F10232 (alcohol dependence with withdrawal with perceptual disturbance), or F10239 (alcohol dependence with withdrawal, unspecified). Similarly, an example of an invalid code in the OT file is 2990. Although this code is not an ICD-9 or ICD-10 diagnosis code, it could form several possible ICD-9 or ICD-10 codes with an additional first or final character.
If the invalid codes are the result of provider errors, and if those errors did not prevent the state or the managed care plan from adjudicating and paying the claim, then it is consistent with T-MSIS coding guidance for states to pass through paid claims with the invalid codes to T-MSIS.
For the OT file, we used the measure calculated with just the subset of claims for outpatient hospital services, physician services, or clinic services to group states into categories of low, medium, and high data quality concern. For the IP and LT files, we used the measures calculated with all claims to assess the states\u2019 data quality.
We included OT claims in this subset if at least one of the claim lines had a federally assigned service category (FASC) equal to 26 (Outpatient hospital), 38 (Physician and all other professional claims), or 27 (Clinic). Claims with these types of services frequently appear to be the focus of in-depth research on service utilization patterns and are expected to have valid ICD-10 diagnosis codes. For more information about the FASC code, see the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
For the IP file, we counted the number of unique, valid ICD-10 diagnosis codes in the fields DGNS_1_CD through DGNS_12_CD for each claim. For the LT file, we counted the number of unique, valid ICD-10 diagnosis codes in the fields DGNS_1_CD through DGNS_5_CD for each claim. We then divided these counts by the total number of IP claims or LT claims, respectively. This accounts for the fact that some states have non-unique diagnosis codes in multiple fields.
Diagnosis codes on TAF service use records represent the diagnoses submitted by providers on the medical claim. This analysis examines the extent to which IP header records are completely coded with at least one valid diagnosis code. The analysis also provides the average number of unique, valid diagnosis codes on IP claims to identify states with potentially incomplete diagnosis code data.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5131"", ""relatedTopics"": [{""measureId"": 50, ""measureName"": ""Diagnosis Code - LT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 1}, {""measureId"": 51, ""measureName"": ""Diagnosis Code - OT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 2}]}" 50,"{""measureId"": 50, ""measureName"": ""Diagnosis Code - LT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Diagnosis-Cd-LT.pdf"", ""background"": {""content"": ""Providers submit diagnosis codes on medical claims that can be used to analyze the prevalence of different medical conditions. Both institutional providers (such as hospitals, nursing facilities, and clinics) and professional providers (such as physicians, other clinical professionals, and ambulances) submit claims on standardized forms that allow them to record multiple diagnosis codes. [1] Diagnosis codes are not, however, submitted on pharmacy claims. Before October 1, 2015, providers used diagnosis codes from the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM). For services provided on or after October 1, 2015, providers are expected to use diagnosis codes from the International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM), which has nearly five times as many possible codes at a far greater level of detail than did the ICD-9-CM.
In the T-MSIS Analytic Files (TAF), all diagnosis codes are located on the header record, which contains information that relates to the claim as a whole. The maximum number of diagnosis codes for a single claim in the TAF varies by file. TAF inpatient (IP) records may have up to 12 diagnosis codes per claim, long-term care (LT) records may have up to 5 diagnosis codes per claim, and other services (OT) records may have up to 2 diagnosis codes per claim. [2] Complete and accurate diagnosis code information is essential, as these fields help TAF users to better understand the underlying health conditions and needs of the populations they are studying. [3]
For the TAF, we expect all claims in the IP and LT claims files to have a valid ICD-10-CM diagnosis code in the primary diagnosis code field (DGNS_1_CD). [4] For the OT claims file, we do not expect that all claims will have diagnosis codes. For certain types of services (for example, medical supplies, prosthetic equipment, or non-emergency medical transportation [NEMT] services), states may not require providers to submit diagnosis codes on the claims because the provider billing for these services may not be in the best position to know and record an accurate diagnosis. Many states also do not require providers to submit diagnosis codes on dental claims. However, if the provider includes any diagnosis codes on those claims, CMS instructs states to pass through those diagnosis codes to T-MSIS as reported on the claims form even if the diagnosis codes are inaccurate or invalid. For this reason, TAF users may want to exercise caution in using diagnosis codes on certain types of OT claims even if they appear to be valid ICD-10 codes.
Institutional claims are often referred to as “UB-04 claims” when submitted in paper form or as “837I claims” when submitted in electronic form. Professional claims are referred to as “CMS-1500 claims” when submitted in paper form or “837P” when submitted in electronic form. All of these formats allow providers to include multiple diagnosis codes on a given claim.
In the IP file, diagnosis codes are captured in the fields DGNS_1_CD through DGNS_12_CD. In the LT file, diagnosis codes are captured in the fields DGNS_1_CD through DGNS_5_CD. In the OT file, diagnosis codes are captured in the fields DGNS_1_CD and DGNS_2_CD. In all three claims files, the first diagnosis code field (DGNS_1_CD) contains the primary diagnosis code, and the additional fields contain other diagnosis codes submitted on the claim. The IP and LT files also include a separate field for the admitting diagnosis code (ADMTG_DGNS_CD), which captures the diagnosis code recorded by the provider when a patient is admitted to a facility. If a claim includes more diagnosis codes than are accommodated in T-MSIS, the source data for TAF, for its given file type, then some diagnosis code information may be missing from the T-MSIS and TAF data systems. This is most likely to be an issue in the OT file because states can only report up to two diagnosis codes per claim in their T-MSIS OT files.
Although diagnosis codes typically capture beneficiaries’ medical conditions, diagnosis codes are sometimes used to provide additional information about the services received during a medical encounter. For example, on evaluation and management claims, diagnosis codes can capture the patient’s age or whether he or she required an examination to participate in sports or to begin a new job.
Some states may not require nursing facilities to report diagnosis codes on their claims if the diagnoses are tracked separately through the federally required Preadmission Screening and Resident Review process or within a different clinical information system. In these cases, the data may be missing from the T-MSIS and TAF claims files.
Institutional claims are often referred to as \u201cUB-04 claims\u201d when submitted in paper form or as \u201c837I claims\u201d when submitted in electronic form. Professional claims are referred to as \u201cCMS-1500 claims\u201d when submitted in paper form or \u201c837P\u201d when submitted in electronic form. All of these formats allow providers to include multiple diagnosis codes on a given claim.
In the IP file, diagnosis codes are captured in the fields DGNS_1_CD through DGNS_12_CD. In the LT file, diagnosis codes are captured in the fields DGNS_1_CD through DGNS_5_CD. In the OT file, diagnosis codes are captured in the fields DGNS_1_CD and DGNS_2_CD. In all three claims files, the first diagnosis code field (DGNS_1_CD) contains the primary diagnosis code, and the additional fields contain other diagnosis codes submitted on the claim. The IP and LT files also include a separate field for the admitting diagnosis code (ADMTG_DGNS_CD), which captures the diagnosis code recorded by the provider when a patient is admitted to a facility. If a claim includes more diagnosis codes than are accommodated in T-MSIS, the source data for TAF, for its given file type, then some diagnosis code information may be missing from the T-MSIS and TAF data systems. This is most likely to be an issue in the OT file because states can only report up to two diagnosis codes per claim in their T-MSIS OT files.
Although diagnosis codes typically capture beneficiaries\u2019 medical conditions, diagnosis codes are sometimes used to provide additional information about the services received during a medical encounter. For example, on evaluation and management claims, diagnosis codes can capture the patient\u2019s age or whether he or she required an examination to participate in sports or to begin a new job.
Some states may not require nursing facilities to report diagnosis codes on their claims if the diagnoses are tracked separately through the federally required Preadmission Screening and Resident Review process or within a different clinical information system. In these cases, the data may be missing from the T-MSIS and TAF claims files.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
This analysis is based on the TAF. [5] We assessed the extent to which diagnosis codes are available on fee-for-service (FFS) claims and on managed care encounter records in the IP, LT, and OT claims files. [6] We excluded crossover claims (those for which Medicare is the primary payer, and Medicaid is responsible only for covering the remaining cost-sharing on behalf of dually eligible beneficiaries). We did not examine the pharmacy claims file because diagnosis codes are not reported on pharmacy claims.
We calculated the percentage of header records in each claims file that had a valid ICD-10 diagnosis code in the field for the primary diagnosis code (DGNS_1_CD). [7] We excluded states from the analyses that had a low volume of claims in the TAF, rendering the data unusable for analysis. For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [8] If a valid ICD-10 diagnosis code was not available, we calculated the percentage of records for which the field (1) was missing, (2) had an ICD-9 diagnosis code, [9] or (3) had another non-missing but invalid value. Because we examined claims for services provided after the transition from ICD-9 to ICD-10, we considered only the ICD-10 diagnosis codes to be “valid” values. We classified all ICD-9 codes as “invalid,” although TAF users may be able to use those codes for certain analyses. Invalid codes may reflect difficulties that the states had in extracting or reporting the data for these fields in their T-MSIS submissions. [10] Alternatively, the invalid codes could reflect providers’ errors in entering the diagnosis codes on claims. [11]
We assessed diagnosis code quality in each state based on the percentage of records that had a valid ICD-10 primary diagnosis code (Table 1). [12]
Table 1. Criteria for DQ assessment of diagnosis code
For the OT claims file only, we calculated the measure for just the subset of claims that included outpatient hospital services, physician services, or clinic services. [13] For the IP and LT files, we calculated the measure with all claims. We also calculated the mean number of unique, valid ICD-10 diagnosis codes available per claim for the IP and LT files. [14] Although we did not use this measure to assess diagnosis code data quality, a mean number of diagnosis codes that is substantially lower than that of other states may suggest that a state’s diagnosis code information is incomplete or indicate variation in state billing practices. We did not assess the appropriateness of the available diagnosis codes against other information on the claims (such as procedure codes) because this complex undertaking is beyond the scope of this analysis.
Methods previously used to assess data quality
Table 2 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods.
Table 2. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We included all claims with a claim type code for either FFS claims or managed care encounter records (CLM_TYPE_CD equal to 1, 3, A, or C). We did not include capitation payments, supplemental payments, or service tracking payments. We also did not include any records with a claim type of “other,” which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Title XIX or Title XXI.
The ICD-10-CM code set is updated each year, often during the middle of the year. For this analysis, we considered codes to be valid ICD-10 diagnosis codes if they were included in either ICD-10-CM that was effective during the year.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
A small number of diagnosis codes are both an ICD-9 diagnosis code and an ICD-10 diagnosis code. For the purposes of our analysis, we classified those codes as valid ICD-10 diagnosis codes, not as ICD-9 codes.
In some cases, the invalid code could be a valid ICD-10 or ICD-9 diagnosis code if it had been submitted with an additional first or final character. An example of an invalid code in the IP file is F1023. Although this code is not a valid ICD-9 or ICD-10 diagnosis code, adding a final character to the submitted code creates four possible valid ICD-10 diagnosis codes: F10230 (alcohol dependence with withdrawal, uncomplicated), F10231 (alcohol dependence with withdrawal delirium), F10232 (alcohol dependence with withdrawal with perceptual disturbance), or F10239 (alcohol dependence with withdrawal, unspecified). Similarly, an example of an invalid code in the OT file is 2990. Although this code is not an ICD-9 or ICD-10 diagnosis code, it could form several possible ICD-9 or ICD-10 codes with an additional first or final character.
If the invalid codes are the result of provider errors, and if those errors did not prevent the state or the managed care plan from adjudicating and paying the claim, then it is consistent with T-MSIS coding guidance for states to pass through paid claims with the invalid codes to T-MSIS.
For the OT file, we used the measure calculated with just the subset of claims for outpatient hospital services, physician services, or clinic services to group states into categories of low, medium, and high data quality concern. For the IP and LT files, we used the measures calculated with all claims to assess the states’ data quality.
We included OT claims in this subset if at least one of the claim lines had a federally assigned service category (FASC) equal to 26 (Outpatient hospital), 38 (Physician and all other professional claims), or 27 (Clinic). Claims with these types of services frequently appear to be the focus of in-depth research on service utilization patterns and are expected to have valid ICD-10 diagnosis codes. For more information about the FASC code, see the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
For the IP file, we counted the number of unique, valid ICD-10 diagnosis codes in the fields DGNS_1_CD through DGNS_12_CD for each claim. For the LT file, we counted the number of unique, valid ICD-10 diagnosis codes in the fields DGNS_1_CD through DGNS_5_CD for each claim. We then divided these counts by the total number of IP claims or LT claims, respectively. This accounts for the fact that some states have non-unique diagnosis codes in multiple fields.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We included all claims with a claim type code for either FFS claims or managed care encounter records (CLM_TYPE_CD equal to 1, 3, A, or C). We did not include capitation payments, supplemental payments, or service tracking payments. We also did not include any records with a claim type of \u201cother,\u201d which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Title XIX or Title XXI.
The ICD-10-CM code set is updated each year, often during the middle of the year. For this analysis, we considered codes to be valid ICD-10 diagnosis codes if they were included in either ICD-10-CM that was effective during the year.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
A small number of diagnosis codes are both an ICD-9 diagnosis code and an ICD-10 diagnosis code. For the purposes of our analysis, we classified those codes as valid ICD-10 diagnosis codes, not as ICD-9 codes.
In some cases, the invalid code could be a valid ICD-10 or ICD-9 diagnosis code if it had been submitted with an additional first or final character. An example of an invalid code in the IP file is F1023. Although this code is not a valid ICD-9 or ICD-10 diagnosis code, adding a final character to the submitted code creates four possible valid ICD-10 diagnosis codes: F10230 (alcohol dependence with withdrawal, uncomplicated), F10231 (alcohol dependence with withdrawal delirium), F10232 (alcohol dependence with withdrawal with perceptual disturbance), or F10239 (alcohol dependence with withdrawal, unspecified). Similarly, an example of an invalid code in the OT file is 2990. Although this code is not an ICD-9 or ICD-10 diagnosis code, it could form several possible ICD-9 or ICD-10 codes with an additional first or final character.
If the invalid codes are the result of provider errors, and if those errors did not prevent the state or the managed care plan from adjudicating and paying the claim, then it is consistent with T-MSIS coding guidance for states to pass through paid claims with the invalid codes to T-MSIS.
For the OT file, we used the measure calculated with just the subset of claims for outpatient hospital services, physician services, or clinic services to group states into categories of low, medium, and high data quality concern. For the IP and LT files, we used the measures calculated with all claims to assess the states\u2019 data quality.
We included OT claims in this subset if at least one of the claim lines had a federally assigned service category (FASC) equal to 26 (Outpatient hospital), 38 (Physician and all other professional claims), or 27 (Clinic). Claims with these types of services frequently appear to be the focus of in-depth research on service utilization patterns and are expected to have valid ICD-10 diagnosis codes. For more information about the FASC code, see the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
For the IP file, we counted the number of unique, valid ICD-10 diagnosis codes in the fields DGNS_1_CD through DGNS_12_CD for each claim. For the LT file, we counted the number of unique, valid ICD-10 diagnosis codes in the fields DGNS_1_CD through DGNS_5_CD for each claim. We then divided these counts by the total number of IP claims or LT claims, respectively. This accounts for the fact that some states have non-unique diagnosis codes in multiple fields.
Diagnosis codes on TAF service use records represent the diagnoses submitted by providers on the medical claim. This analysis examines the extent to which LT header records are completely coded with at least one valid diagnosis code. The analysis also provides the average number of unique, valid diagnosis codes on LT claims to identify states with potentially incomplete diagnosis code data.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5131"", ""relatedTopics"": [{""measureId"": 49, ""measureName"": ""Diagnosis Code - IP"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 0}, {""measureId"": 51, ""measureName"": ""Diagnosis Code - OT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 2}]}" 51,"{""measureId"": 51, ""measureName"": ""Diagnosis Code - OT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Diagnosis-Cd-OT.pdf"", ""background"": {""content"": ""Providers submit diagnosis codes on medical claims that can be used to analyze the prevalence of different medical conditions. Both institutional providers (such as hospitals, nursing facilities, and clinics) and professional providers (such as physicians, other clinical professionals, and ambulances) submit claims on standardized forms that allow them to record multiple diagnosis codes. [1] Diagnosis codes are not, however, submitted on pharmacy claims. Before October 1, 2015, providers used diagnosis codes from the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM). For services provided on or after October 1, 2015, providers are expected to use diagnosis codes from the International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM), which has nearly five times as many possible codes at a far greater level of detail than did the ICD-9-CM.
In the T-MSIS Analytic Files (TAF), all diagnosis codes are located on the header record, which contains information that relates to the claim as a whole. The maximum number of diagnosis codes for a single claim in the TAF varies by file. TAF inpatient (IP) records may have up to 12 diagnosis codes per claim, long-term care (LT) records may have up to 5 diagnosis codes per claim, and other services (OT) records may have up to 2 diagnosis codes per claim. [2] Complete and accurate diagnosis code information is essential, as these fields help TAF users to better understand the underlying health conditions and needs of the populations they are studying. [3]
For the TAF, we expect all claims in the IP and LT claims files to have a valid ICD-10-CM diagnosis code in the primary diagnosis code field (DGNS_1_CD). [4] For the OT claims file, we do not expect that all claims will have diagnosis codes. For certain types of services (for example, medical supplies, prosthetic equipment, or non-emergency medical transportation [NEMT] services), states may not require providers to submit diagnosis codes on the claims because the provider billing for these services may not be in the best position to know and record an accurate diagnosis. Many states also do not require providers to submit diagnosis codes on dental claims. However, if the provider includes any diagnosis codes on those claims, CMS instructs states to pass through those diagnosis codes to T-MSIS as reported on the claims form even if the diagnosis codes are inaccurate or invalid. For this reason, TAF users may want to exercise caution in using diagnosis codes on certain types of OT claims even if they appear to be valid ICD-10 codes.
Institutional claims are often referred to as “UB-04 claims” when submitted in paper form or as “837I claims” when submitted in electronic form. Professional claims are referred to as “CMS-1500 claims” when submitted in paper form or “837P” when submitted in electronic form. All of these formats allow providers to include multiple diagnosis codes on a given claim.
In the IP file, diagnosis codes are captured in the fields DGNS_1_CD through DGNS_12_CD. In the LT file, diagnosis codes are captured in the fields DGNS_1_CD through DGNS_5_CD. In the OT file, diagnosis codes are captured in the fields DGNS_1_CD and DGNS_2_CD. In all three claims files, the first diagnosis code field (DGNS_1_CD) contains the primary diagnosis code, and the additional fields contain other diagnosis codes submitted on the claim. The IP and LT files also include a separate field for the admitting diagnosis code (ADMTG_DGNS_CD), which captures the diagnosis code recorded by the provider when a patient is admitted to a facility. If a claim includes more diagnosis codes than are accommodated in T-MSIS, the source data for TAF, for its given file type, then some diagnosis code information may be missing from the T-MSIS and TAF data systems. This is most likely to be an issue in the OT file because states can only report up to two diagnosis codes per claim in their T-MSIS OT files.
Although diagnosis codes typically capture beneficiaries’ medical conditions, diagnosis codes are sometimes used to provide additional information about the services received during a medical encounter. For example, on evaluation and management claims, diagnosis codes can capture the patient’s age or whether he or she required an examination to participate in sports or to begin a new job.
Some states may not require nursing facilities to report diagnosis codes on their claims if the diagnoses are tracked separately through the federally required Preadmission Screening and Resident Review process or within a different clinical information system. In these cases, the data may be missing from the T-MSIS and TAF claims files.
Institutional claims are often referred to as \u201cUB-04 claims\u201d when submitted in paper form or as \u201c837I claims\u201d when submitted in electronic form. Professional claims are referred to as \u201cCMS-1500 claims\u201d when submitted in paper form or \u201c837P\u201d when submitted in electronic form. All of these formats allow providers to include multiple diagnosis codes on a given claim.
In the IP file, diagnosis codes are captured in the fields DGNS_1_CD through DGNS_12_CD. In the LT file, diagnosis codes are captured in the fields DGNS_1_CD through DGNS_5_CD. In the OT file, diagnosis codes are captured in the fields DGNS_1_CD and DGNS_2_CD. In all three claims files, the first diagnosis code field (DGNS_1_CD) contains the primary diagnosis code, and the additional fields contain other diagnosis codes submitted on the claim. The IP and LT files also include a separate field for the admitting diagnosis code (ADMTG_DGNS_CD), which captures the diagnosis code recorded by the provider when a patient is admitted to a facility. If a claim includes more diagnosis codes than are accommodated in T-MSIS, the source data for TAF, for its given file type, then some diagnosis code information may be missing from the T-MSIS and TAF data systems. This is most likely to be an issue in the OT file because states can only report up to two diagnosis codes per claim in their T-MSIS OT files.
Although diagnosis codes typically capture beneficiaries\u2019 medical conditions, diagnosis codes are sometimes used to provide additional information about the services received during a medical encounter. For example, on evaluation and management claims, diagnosis codes can capture the patient\u2019s age or whether he or she required an examination to participate in sports or to begin a new job.
Some states may not require nursing facilities to report diagnosis codes on their claims if the diagnoses are tracked separately through the federally required Preadmission Screening and Resident Review process or within a different clinical information system. In these cases, the data may be missing from the T-MSIS and TAF claims files.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
This analysis is based on the TAF. [5] We assessed the extent to which diagnosis codes are available on fee-for-service (FFS) claims and on managed care encounter records in the IP, LT, and OT claims files. [6] We excluded crossover claims (those for which Medicare is the primary payer, and Medicaid is responsible only for covering the remaining cost-sharing on behalf of dually eligible beneficiaries). We did not examine the pharmacy claims file because diagnosis codes are not reported on pharmacy claims.
We calculated the percentage of header records in each claims file that had a valid ICD-10 diagnosis code in the field for the primary diagnosis code (DGNS_1_CD). [7] We excluded states from the analyses that had a low volume of claims in the TAF, rendering the data unusable for analysis. For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [8] If a valid ICD-10 diagnosis code was not available, we calculated the percentage of records for which the field (1) was missing, (2) had an ICD-9 diagnosis code, [9] or (3) had another non-missing but invalid value. Because we examined claims for services provided after the transition from ICD-9 to ICD-10, we considered only the ICD-10 diagnosis codes to be “valid” values. We classified all ICD-9 codes as “invalid,” although TAF users may be able to use those codes for certain analyses. Invalid codes may reflect difficulties that the states had in extracting or reporting the data for these fields in their T-MSIS submissions. [10] Alternatively, the invalid codes could reflect providers’ errors in entering the diagnosis codes on claims. [11]
We assessed diagnosis code quality in each state based on the percentage of records that had a valid ICD-10 primary diagnosis code (Table 1). [12]
Table 1. Criteria for DQ assessment of diagnosis code
For the OT claims file only, we calculated the measure for just the subset of claims that included outpatient hospital services, physician services, or clinic services. [13] For the IP and LT files, we calculated the measure with all claims. We also calculated the mean number of unique, valid ICD-10 diagnosis codes available per claim for the IP and LT files. [14] Although we did not use this measure to assess diagnosis code data quality, a mean number of diagnosis codes that is substantially lower than that of other states may suggest that a state’s diagnosis code information is incomplete or indicate variation in state billing practices. We did not assess the appropriateness of the available diagnosis codes against other information on the claims (such as procedure codes) because this complex undertaking is beyond the scope of this analysis.
Methods previously used to assess data quality
Table 2 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods.
Table 2. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We included all claims with a claim type code for either FFS claims or managed care encounter records (CLM_TYPE_CD equal to 1, 3, A, or C). We did not include capitation payments, supplemental payments, or service tracking payments. We also did not include any records with a claim type of “other,” which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Title XIX or Title XXI.
The ICD-10-CM code set is updated each year, often during the middle of the year. For this analysis, we considered codes to be valid ICD-10 diagnosis codes if they were included in either ICD-10-CM that was effective during the year.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
A small number of diagnosis codes are both an ICD-9 diagnosis code and an ICD-10 diagnosis code. For the purposes of our analysis, we classified those codes as valid ICD-10 diagnosis codes, not as ICD-9 codes.
In some cases, the invalid code could be a valid ICD-10 or ICD-9 diagnosis code if it had been submitted with an additional first or final character. An example of an invalid code in the IP file is F1023. Although this code is not a valid ICD-9 or ICD-10 diagnosis code, adding a final character to the submitted code creates four possible valid ICD-10 diagnosis codes: F10230 (alcohol dependence with withdrawal, uncomplicated), F10231 (alcohol dependence with withdrawal delirium), F10232 (alcohol dependence with withdrawal with perceptual disturbance), or F10239 (alcohol dependence with withdrawal, unspecified). Similarly, an example of an invalid code in the OT file is 2990. Although this code is not an ICD-9 or ICD-10 diagnosis code, it could form several possible ICD-9 or ICD-10 codes with an additional first or final character.
If the invalid codes are the result of provider errors, and if those errors did not prevent the state or the managed care plan from adjudicating and paying the claim, then it is consistent with T-MSIS coding guidance for states to pass through paid claims with the invalid codes to T-MSIS.
For the OT file, we used the measure calculated with just the subset of claims for outpatient hospital services, physician services, or clinic services to group states into categories of low, medium, and high data quality concern. For the IP and LT files, we used the measures calculated with all claims to assess the states’ data quality.
We included OT claims in this subset if at least one of the claim lines had a federally assigned service category (FASC) equal to 26 (Outpatient hospital), 38 (Physician and all other professional claims), or 27 (Clinic). Claims with these types of services frequently appear to be the focus of in-depth research on service utilization patterns and are expected to have valid ICD-10 diagnosis codes. For more information about the FASC code, see the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
For the IP file, we counted the number of unique, valid ICD-10 diagnosis codes in the fields DGNS_1_CD through DGNS_12_CD for each claim. For the LT file, we counted the number of unique, valid ICD-10 diagnosis codes in the fields DGNS_1_CD through DGNS_5_CD for each claim. We then divided these counts by the total number of IP claims or LT claims, respectively. This accounts for the fact that some states have non-unique diagnosis codes in multiple fields.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We included all claims with a claim type code for either FFS claims or managed care encounter records (CLM_TYPE_CD equal to 1, 3, A, or C). We did not include capitation payments, supplemental payments, or service tracking payments. We also did not include any records with a claim type of \u201cother,\u201d which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Title XIX or Title XXI.
The ICD-10-CM code set is updated each year, often during the middle of the year. For this analysis, we considered codes to be valid ICD-10 diagnosis codes if they were included in either ICD-10-CM that was effective during the year.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
A small number of diagnosis codes are both an ICD-9 diagnosis code and an ICD-10 diagnosis code. For the purposes of our analysis, we classified those codes as valid ICD-10 diagnosis codes, not as ICD-9 codes.
In some cases, the invalid code could be a valid ICD-10 or ICD-9 diagnosis code if it had been submitted with an additional first or final character. An example of an invalid code in the IP file is F1023. Although this code is not a valid ICD-9 or ICD-10 diagnosis code, adding a final character to the submitted code creates four possible valid ICD-10 diagnosis codes: F10230 (alcohol dependence with withdrawal, uncomplicated), F10231 (alcohol dependence with withdrawal delirium), F10232 (alcohol dependence with withdrawal with perceptual disturbance), or F10239 (alcohol dependence with withdrawal, unspecified). Similarly, an example of an invalid code in the OT file is 2990. Although this code is not an ICD-9 or ICD-10 diagnosis code, it could form several possible ICD-9 or ICD-10 codes with an additional first or final character.
If the invalid codes are the result of provider errors, and if those errors did not prevent the state or the managed care plan from adjudicating and paying the claim, then it is consistent with T-MSIS coding guidance for states to pass through paid claims with the invalid codes to T-MSIS.
For the OT file, we used the measure calculated with just the subset of claims for outpatient hospital services, physician services, or clinic services to group states into categories of low, medium, and high data quality concern. For the IP and LT files, we used the measures calculated with all claims to assess the states\u2019 data quality.
We included OT claims in this subset if at least one of the claim lines had a federally assigned service category (FASC) equal to 26 (Outpatient hospital), 38 (Physician and all other professional claims), or 27 (Clinic). Claims with these types of services frequently appear to be the focus of in-depth research on service utilization patterns and are expected to have valid ICD-10 diagnosis codes. For more information about the FASC code, see the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
For the IP file, we counted the number of unique, valid ICD-10 diagnosis codes in the fields DGNS_1_CD through DGNS_12_CD for each claim. For the LT file, we counted the number of unique, valid ICD-10 diagnosis codes in the fields DGNS_1_CD through DGNS_5_CD for each claim. We then divided these counts by the total number of IP claims or LT claims, respectively. This accounts for the fact that some states have non-unique diagnosis codes in multiple fields.
Diagnosis codes on TAF service use records represent the diagnoses submitted by providers on the medical claim. This analysis examines the extent to which OT header records are completely coded with at least one valid diagnosis code.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5131"", ""relatedTopics"": [{""measureId"": 49, ""measureName"": ""Diagnosis Code - IP"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 0}, {""measureId"": 50, ""measureName"": ""Diagnosis Code - LT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 1}]}" 52,"{""measureId"": 52, ""measureName"": ""Type of Service - IP"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Type-of-Service-IP.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF), the research-ready version of T-MSIS, include data on service use for all beneficiaries enrolled in Medicaid or in the Children’s Health Insurance Program (CHIP). Service use records in the TAF are contained in the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) files. TAF service use records capture both data that providers enter on claim forms for payment as well as additional fields specific to T\u2011MSIS that state Medicaid agencies are responsible for populating.
One of these additional fields is the type of service code, which states use to classify individual services appearing on a claim into standard categories that map to benefit and provider definitions in the Code of Federal Regulations (CFR). States have been reporting type of service codes for many years as part of their reporting into the original Medicaid Statistical Information System (MSIS). With the transition to T-MSIS, the set of codes was expanded and refined to reflect more current definitions. In some cases, new codes were introduced that represent services or providers covered through the CHIP (not Medicaid) state plan, which means that some services such as well-child visits may have a different type of service code if delivered to an S-CHIP beneficiary versus a Medicaid or M-CHIP beneficiary. In other cases, new codes were introduced that represent finer, more granular services that were bundled together in previous type of service codes under MSIS. For example, a new code for services provided at critical access hospitals was introduced, while the older code for inpatient hospital services remains available.
Since some services may qualify for Medicaid or CHIP coverage under more than one section of the CFR, they may meet the definition of multiple type of service categories in T-MSIS. For example, depending on the type of provider delivering the service, a well-child visit delivered to a Medicaid-enrolled beneficiary might be coded as “Early and periodic screening and diagnosis and treatment (EPSDT) services” (code 010), as “Preventive services,” (code 041), as “Physician services” (code 012), or as “Clinic services” (code 028). The same service delivered to an S-CHIP-enrolled child could be coded as “Well-baby and well-child care services as defined by the State” (code 042), a type of service code that is specific to the CHIP benefit package. Thus, even in the absence of T-MSIS reporting errors, we would expect to observe differences in the type of service code assigned to the same service in different states. TAF users should be very cautious in using the type of service data element to try to systematically identify a specific service across different states without including information from other fields, such as procedure or revenue codes, as it is unlikely that type of service code alone can consistently identify the same set of services across all states.
The T-MSIS data dictionary defines for states the type of service codes that are expected on records submitted into the IP, LT, OT, and RX files. This data quality assessment examines how often the states’ TAF data have a missing or invalid type of service code. The assessment does not examine whether states appear to be consistent in assigning type of service values to similar claims, because some variation is expected and may reflect differences in what program a benefit is covered under (Medicaid or S-CHIP) or differences in the benefit category used to cover the service.
"", ""footnotes"": []}, ""methods"": {""content"": ""We examined the type of service code (TOS_CD) values on all line records in the TAF [1] IP, LT, OT, and RX files. [2] We included all record types from these files, including fee-for-service claims, managed care encounters, capitation payments, service tracking claims, and supplemental payments, because states are expected to populate every line record submitted to T-MSIS with a valid type of service value. [3] We examined each file separately and excluded states from each file-specific analysis if the number of line records in the TAF were unusably low given the size of the state’s Medicaid and CHIP programs. [4] For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [5] We then examined the extent of missing or invalid type of service codes on records in each state.
We classified all line-level records into one of three categories: (1) valid type of service value for the file, (2) invalid type of service value for the file, or (3) missing value. We used the T-MSIS submission guidelines to define what type of service values are valid in each file. [6] Using the criteria shown in Table 1, we classified states as having type of service data that presented a low, medium, or high level of concern about data completeness—or as having sufficiently incomplete data as to be unusable. Our evaluation of whether the type of service code was valid only examined whether the code was allowable for the file under T-MSIS submission guidelines. We did not evaluate whether the code was consistent with other service-specific information on the record, such as procedure code or revenue code.
Table 1. Criteria for DQ assessment of type of service code
| Percentage of line records with missing or invalid type of service code | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
The T-MSIS claims files are structured to include one header record per claim and one or more line records that link to a header record. Header-level records capture data that apply to the entire claim. Line-level records capture data about the specific goods or services provided as part of the overall claim. Since the type of service categories relate to specific services, this data element is captured on line records. It is possible that a single claim would have lines with different type of service values.
We used the claim type code to exclude “other” records (claim type code values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
More information about the completeness of each state’s TAF data can be found in the DQ Atlas single topic displays for Claims Volume - IP , Claims Volume - LT , Claims Volume - OT , and Claims Volume - RX .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
A list of valid values for the type of service code can be found in the TAF Claims Codebook at https://www2.ccwdata.org/web/quest/data-dictionaries .
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
The T-MSIS claims files are structured to include one header record per claim and one or more line records that link to a header record. Header-level records capture data that apply to the entire claim. Line-level records capture data about the specific goods or services provided as part of the overall claim. Since the type of service categories relate to specific services, this data element is captured on line records. It is possible that a single claim would have lines with different type of service values.
We used the claim type code to exclude \u201cother\u201d records (claim type code values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
More information about the completeness of each state\u2019s TAF data can be found in the DQ Atlas single topic displays for Claims Volume - IP , Claims Volume - LT , Claims Volume - OT , and Claims Volume - RX .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
A list of valid values for the type of service code can be found in the TAF Claims Codebook at https://www2.ccwdata.org/web/quest/data-dictionaries .
TAF service use records capture data that providers enter on standard claim forms as well as additional fields specific to T-MSIS that state Medicaid agencies are responsible for populating. One of these additional fields is the type of service code, which states use to classify individual services appearing on a claim into standard categories that map to benefit and provider definitions in the Code of Federal Regulations. This analysis examines how often IP line records have missing or unexpected type of service codes.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5141"", ""relatedTopics"": [{""measureId"": 54, ""measureName"": ""Type of Service - LT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 1}, {""measureId"": 53, ""measureName"": ""Type of Service - OT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 2}, {""measureId"": 55, ""measureName"": ""Type of Service - RX"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 3}]}" 53,"{""measureId"": 53, ""measureName"": ""Type of Service - OT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Type-of-Service-OT.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF), the research-ready version of T-MSIS, include data on service use for all beneficiaries enrolled in Medicaid or in the Children’s Health Insurance Program (CHIP). Service use records in the TAF are contained in the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) files. TAF service use records capture both data that providers enter on claim forms for payment as well as additional fields specific to T\u2011MSIS that state Medicaid agencies are responsible for populating.
One of these additional fields is the type of service code, which states use to classify individual services appearing on a claim into standard categories that map to benefit and provider definitions in the Code of Federal Regulations (CFR). States have been reporting type of service codes for many years as part of their reporting into the original Medicaid Statistical Information System (MSIS). With the transition to T-MSIS, the set of codes was expanded and refined to reflect more current definitions. In some cases, new codes were introduced that represent services or providers covered through the CHIP (not Medicaid) state plan, which means that some services such as well-child visits may have a different type of service code if delivered to an S-CHIP beneficiary versus a Medicaid or M-CHIP beneficiary. In other cases, new codes were introduced that represent finer, more granular services that were bundled together in previous type of service codes under MSIS. For example, a new code for services provided at critical access hospitals was introduced, while the older code for inpatient hospital services remains available.
Since some services may qualify for Medicaid or CHIP coverage under more than one section of the CFR, they may meet the definition of multiple type of service categories in T-MSIS. For example, depending on the type of provider delivering the service, a well-child visit delivered to a Medicaid-enrolled beneficiary might be coded as “Early and periodic screening and diagnosis and treatment (EPSDT) services” (code 010), as “Preventive services,” (code 041), as “Physician services” (code 012), or as “Clinic services” (code 028). The same service delivered to an S-CHIP-enrolled child could be coded as “Well-baby and well-child care services as defined by the State” (code 042), a type of service code that is specific to the CHIP benefit package. Thus, even in the absence of T-MSIS reporting errors, we would expect to observe differences in the type of service code assigned to the same service in different states. TAF users should be very cautious in using the type of service data element to try to systematically identify a specific service across different states without including information from other fields, such as procedure or revenue codes, as it is unlikely that type of service code alone can consistently identify the same set of services across all states.
The T-MSIS data dictionary defines for states the type of service codes that are expected on records submitted into the IP, LT, OT, and RX files. This data quality assessment examines how often the states’ TAF data have a missing or invalid type of service code. The assessment does not examine whether states appear to be consistent in assigning type of service values to similar claims, because some variation is expected and may reflect differences in what program a benefit is covered under (Medicaid or S-CHIP) or differences in the benefit category used to cover the service.
"", ""footnotes"": []}, ""methods"": {""content"": ""We examined the type of service code (TOS_CD) values on all line records in the TAF [1] IP, LT, OT, and RX files. [2] We included all record types from these files, including fee-for-service claims, managed care encounters, capitation payments, service tracking claims, and supplemental payments, because states are expected to populate every line record submitted to T-MSIS with a valid type of service value. [3] We examined each file separately and excluded states from each file-specific analysis if the number of line records in the TAF were unusably low given the size of the state’s Medicaid and CHIP programs. [4] For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [5] We then examined the extent of missing or invalid type of service codes on records in each state.
We classified all line-level records into one of three categories: (1) valid type of service value for the file, (2) invalid type of service value for the file, or (3) missing value. We used the T-MSIS submission guidelines to define what type of service values are valid in each file. [6] Using the criteria shown in Table 1, we classified states as having type of service data that presented a low, medium, or high level of concern about data completeness—or as having sufficiently incomplete data as to be unusable. Our evaluation of whether the type of service code was valid only examined whether the code was allowable for the file under T-MSIS submission guidelines. We did not evaluate whether the code was consistent with other service-specific information on the record, such as procedure code or revenue code.
Table 1. Criteria for DQ assessment of type of service code
| Percentage of line records with missing or invalid type of service code | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
The T-MSIS claims files are structured to include one header record per claim and one or more line records that link to a header record. Header-level records capture data that apply to the entire claim. Line-level records capture data about the specific goods or services provided as part of the overall claim. Since the type of service categories relate to specific services, this data element is captured on line records. It is possible that a single claim would have lines with different type of service values.
We used the claim type code to exclude “other” records (claim type code values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
More information about the completeness of each state’s TAF data can be found in the DQ Atlas single topic displays for Claims Volume - IP , Claims Volume - LT , Claims Volume - OT , and Claims Volume - RX .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
A list of valid values for the type of service code can be found in the TAF Claims Codebook at https://www2.ccwdata.org/web/quest/data-dictionaries .
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
The T-MSIS claims files are structured to include one header record per claim and one or more line records that link to a header record. Header-level records capture data that apply to the entire claim. Line-level records capture data about the specific goods or services provided as part of the overall claim. Since the type of service categories relate to specific services, this data element is captured on line records. It is possible that a single claim would have lines with different type of service values.
We used the claim type code to exclude \u201cother\u201d records (claim type code values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
More information about the completeness of each state\u2019s TAF data can be found in the DQ Atlas single topic displays for Claims Volume - IP , Claims Volume - LT , Claims Volume - OT , and Claims Volume - RX .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
A list of valid values for the type of service code can be found in the TAF Claims Codebook at https://www2.ccwdata.org/web/quest/data-dictionaries .
TAF service use records capture data that providers enter on standard claim forms as well as additional fields specific to T-MSIS that state Medicaid agencies are responsible for populating. One of these additional fields is the type of service code, which states use to classify individual services appearing on a claim into standard categories that map to benefit and provider definitions in the Code of Federal Regulations. This analysis examines how often OT line records have missing or unexpected type of service codes.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5141"", ""relatedTopics"": [{""measureId"": 52, ""measureName"": ""Type of Service - IP"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 0}, {""measureId"": 54, ""measureName"": ""Type of Service - LT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 1}, {""measureId"": 55, ""measureName"": ""Type of Service - RX"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 3}]}" 54,"{""measureId"": 54, ""measureName"": ""Type of Service - LT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Type-of-Service-LT.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF), the research-ready version of T-MSIS, include data on service use for all beneficiaries enrolled in Medicaid or in the Children’s Health Insurance Program (CHIP). Service use records in the TAF are contained in the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) files. TAF service use records capture both data that providers enter on claim forms for payment as well as additional fields specific to T\u2011MSIS that state Medicaid agencies are responsible for populating.
One of these additional fields is the type of service code, which states use to classify individual services appearing on a claim into standard categories that map to benefit and provider definitions in the Code of Federal Regulations (CFR). States have been reporting type of service codes for many years as part of their reporting into the original Medicaid Statistical Information System (MSIS). With the transition to T-MSIS, the set of codes was expanded and refined to reflect more current definitions. In some cases, new codes were introduced that represent services or providers covered through the CHIP (not Medicaid) state plan, which means that some services such as well-child visits may have a different type of service code if delivered to an S-CHIP beneficiary versus a Medicaid or M-CHIP beneficiary. In other cases, new codes were introduced that represent finer, more granular services that were bundled together in previous type of service codes under MSIS. For example, a new code for services provided at critical access hospitals was introduced, while the older code for inpatient hospital services remains available.
Since some services may qualify for Medicaid or CHIP coverage under more than one section of the CFR, they may meet the definition of multiple type of service categories in T-MSIS. For example, depending on the type of provider delivering the service, a well-child visit delivered to a Medicaid-enrolled beneficiary might be coded as “Early and periodic screening and diagnosis and treatment (EPSDT) services” (code 010), as “Preventive services,” (code 041), as “Physician services” (code 012), or as “Clinic services” (code 028). The same service delivered to an S-CHIP-enrolled child could be coded as “Well-baby and well-child care services as defined by the State” (code 042), a type of service code that is specific to the CHIP benefit package. Thus, even in the absence of T-MSIS reporting errors, we would expect to observe differences in the type of service code assigned to the same service in different states. TAF users should be very cautious in using the type of service data element to try to systematically identify a specific service across different states without including information from other fields, such as procedure or revenue codes, as it is unlikely that type of service code alone can consistently identify the same set of services across all states.
The T-MSIS data dictionary defines for states the type of service codes that are expected on records submitted into the IP, LT, OT, and RX files. This data quality assessment examines how often the states’ TAF data have a missing or invalid type of service code. The assessment does not examine whether states appear to be consistent in assigning type of service values to similar claims, because some variation is expected and may reflect differences in what program a benefit is covered under (Medicaid or S-CHIP) or differences in the benefit category used to cover the service.
"", ""footnotes"": []}, ""methods"": {""content"": ""We examined the type of service code (TOS_CD) values on all line records in the TAF [1] IP, LT, OT, and RX files. [2] We included all record types from these files, including fee-for-service claims, managed care encounters, capitation payments, service tracking claims, and supplemental payments, because states are expected to populate every line record submitted to T-MSIS with a valid type of service value. [3] We examined each file separately and excluded states from each file-specific analysis if the number of line records in the TAF were unusably low given the size of the state’s Medicaid and CHIP programs. [4] For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [5] We then examined the extent of missing or invalid type of service codes on records in each state.
We classified all line-level records into one of three categories: (1) valid type of service value for the file, (2) invalid type of service value for the file, or (3) missing value. We used the T-MSIS submission guidelines to define what type of service values are valid in each file. [6] Using the criteria shown in Table 1, we classified states as having type of service data that presented a low, medium, or high level of concern about data completeness—or as having sufficiently incomplete data as to be unusable. Our evaluation of whether the type of service code was valid only examined whether the code was allowable for the file under T-MSIS submission guidelines. We did not evaluate whether the code was consistent with other service-specific information on the record, such as procedure code or revenue code.
Table 1. Criteria for DQ assessment of type of service code
| Percentage of line records with missing or invalid type of service code | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
The T-MSIS claims files are structured to include one header record per claim and one or more line records that link to a header record. Header-level records capture data that apply to the entire claim. Line-level records capture data about the specific goods or services provided as part of the overall claim. Since the type of service categories relate to specific services, this data element is captured on line records. It is possible that a single claim would have lines with different type of service values.
We used the claim type code to exclude “other” records (claim type code values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
More information about the completeness of each state’s TAF data can be found in the DQ Atlas single topic displays for Claims Volume - IP , Claims Volume - LT , Claims Volume - OT , and Claims Volume - RX .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
A list of valid values for the type of service code can be found in the TAF Claims Codebook at https://www2.ccwdata.org/web/quest/data-dictionaries .
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
The T-MSIS claims files are structured to include one header record per claim and one or more line records that link to a header record. Header-level records capture data that apply to the entire claim. Line-level records capture data about the specific goods or services provided as part of the overall claim. Since the type of service categories relate to specific services, this data element is captured on line records. It is possible that a single claim would have lines with different type of service values.
We used the claim type code to exclude \u201cother\u201d records (claim type code values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
More information about the completeness of each state\u2019s TAF data can be found in the DQ Atlas single topic displays for Claims Volume - IP , Claims Volume - LT , Claims Volume - OT , and Claims Volume - RX .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
A list of valid values for the type of service code can be found in the TAF Claims Codebook at https://www2.ccwdata.org/web/quest/data-dictionaries .
TAF service use records capture data that providers enter on standard claim forms as well as additional fields specific to T-MSIS that state Medicaid agencies are responsible for populating. One of these additional fields is the type of service code, which states use to classify individual services appearing on a claim into standard categories that map to benefit and provider definitions in the Code of Federal Regulations. This analysis examines how often LT line records have missing or unexpected type of service codes.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5141"", ""relatedTopics"": [{""measureId"": 52, ""measureName"": ""Type of Service - IP"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 0}, {""measureId"": 53, ""measureName"": ""Type of Service - OT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 2}, {""measureId"": 55, ""measureName"": ""Type of Service - RX"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 3}]}" 55,"{""measureId"": 55, ""measureName"": ""Type of Service - RX"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Type-of-Service-RX.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF), the research-ready version of T-MSIS, include data on service use for all beneficiaries enrolled in Medicaid or in the Children’s Health Insurance Program (CHIP). Service use records in the TAF are contained in the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) files. TAF service use records capture both data that providers enter on claim forms for payment as well as additional fields specific to T\u2011MSIS that state Medicaid agencies are responsible for populating.
One of these additional fields is the type of service code, which states use to classify individual services appearing on a claim into standard categories that map to benefit and provider definitions in the Code of Federal Regulations (CFR). States have been reporting type of service codes for many years as part of their reporting into the original Medicaid Statistical Information System (MSIS). With the transition to T-MSIS, the set of codes was expanded and refined to reflect more current definitions. In some cases, new codes were introduced that represent services or providers covered through the CHIP (not Medicaid) state plan, which means that some services such as well-child visits may have a different type of service code if delivered to an S-CHIP beneficiary versus a Medicaid or M-CHIP beneficiary. In other cases, new codes were introduced that represent finer, more granular services that were bundled together in previous type of service codes under MSIS. For example, a new code for services provided at critical access hospitals was introduced, while the older code for inpatient hospital services remains available.
Since some services may qualify for Medicaid or CHIP coverage under more than one section of the CFR, they may meet the definition of multiple type of service categories in T-MSIS. For example, depending on the type of provider delivering the service, a well-child visit delivered to a Medicaid-enrolled beneficiary might be coded as “Early and periodic screening and diagnosis and treatment (EPSDT) services” (code 010), as “Preventive services,” (code 041), as “Physician services” (code 012), or as “Clinic services” (code 028). The same service delivered to an S-CHIP-enrolled child could be coded as “Well-baby and well-child care services as defined by the State” (code 042), a type of service code that is specific to the CHIP benefit package. Thus, even in the absence of T-MSIS reporting errors, we would expect to observe differences in the type of service code assigned to the same service in different states. TAF users should be very cautious in using the type of service data element to try to systematically identify a specific service across different states without including information from other fields, such as procedure or revenue codes, as it is unlikely that type of service code alone can consistently identify the same set of services across all states.
The T-MSIS data dictionary defines for states the type of service codes that are expected on records submitted into the IP, LT, OT, and RX files. This data quality assessment examines how often the states’ TAF data have a missing or invalid type of service code. The assessment does not examine whether states appear to be consistent in assigning type of service values to similar claims, because some variation is expected and may reflect differences in what program a benefit is covered under (Medicaid or S-CHIP) or differences in the benefit category used to cover the service.
"", ""footnotes"": []}, ""methods"": {""content"": ""We examined the type of service code (TOS_CD) values on all line records in the TAF [1] IP, LT, OT, and RX files. [2] We included all record types from these files, including fee-for-service claims, managed care encounters, capitation payments, service tracking claims, and supplemental payments, because states are expected to populate every line record submitted to T-MSIS with a valid type of service value. [3] We examined each file separately and excluded states from each file-specific analysis if the number of line records in the TAF were unusably low given the size of the state’s Medicaid and CHIP programs. [4] For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [5] We then examined the extent of missing or invalid type of service codes on records in each state.
We classified all line-level records into one of three categories: (1) valid type of service value for the file, (2) invalid type of service value for the file, or (3) missing value. We used the T-MSIS submission guidelines to define what type of service values are valid in each file. [6] Using the criteria shown in Table 1, we classified states as having type of service data that presented a low, medium, or high level of concern about data completeness—or as having sufficiently incomplete data as to be unusable. Our evaluation of whether the type of service code was valid only examined whether the code was allowable for the file under T-MSIS submission guidelines. We did not evaluate whether the code was consistent with other service-specific information on the record, such as procedure code or revenue code.
Table 1. Criteria for DQ assessment of type of service code
| Percentage of line records with missing or invalid type of service code | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
The T-MSIS claims files are structured to include one header record per claim and one or more line records that link to a header record. Header-level records capture data that apply to the entire claim. Line-level records capture data about the specific goods or services provided as part of the overall claim. Since the type of service categories relate to specific services, this data element is captured on line records. It is possible that a single claim would have lines with different type of service values.
We used the claim type code to exclude “other” records (claim type code values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
More information about the completeness of each state’s TAF data can be found in the DQ Atlas single topic displays for Claims Volume - IP , Claims Volume - LT , Claims Volume - OT , and Claims Volume - RX .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
A list of valid values for the type of service code can be found in the TAF Claims Codebook at https://www2.ccwdata.org/web/quest/data-dictionaries .
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
The T-MSIS claims files are structured to include one header record per claim and one or more line records that link to a header record. Header-level records capture data that apply to the entire claim. Line-level records capture data about the specific goods or services provided as part of the overall claim. Since the type of service categories relate to specific services, this data element is captured on line records. It is possible that a single claim would have lines with different type of service values.
We used the claim type code to exclude \u201cother\u201d records (claim type code values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
More information about the completeness of each state\u2019s TAF data can be found in the DQ Atlas single topic displays for Claims Volume - IP , Claims Volume - LT , Claims Volume - OT , and Claims Volume - RX .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
A list of valid values for the type of service code can be found in the TAF Claims Codebook at https://www2.ccwdata.org/web/quest/data-dictionaries .
TAF service use records capture data that providers enter on standard claim forms as well as additional fields specific to T-MSIS that state Medicaid agencies are responsible for populating. One of these additional fields is the type of service code, which states use to classify individual services appearing on a claim into standard categories that map to benefit and provider definitions in the Code of Federal Regulations. This analysis examines how often RX line records have missing or unexpected type of service codes.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5141"", ""relatedTopics"": [{""measureId"": 52, ""measureName"": ""Type of Service - IP"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 0}, {""measureId"": 54, ""measureName"": ""Type of Service - LT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 1}, {""measureId"": 53, ""measureName"": ""Type of Service - OT"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 2}]}" 56,"{""measureId"": 56, ""measureName"": ""CMC Plan Encounters - IP"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-CMC-Plan-Encounters-IP.pdf"", ""background"": {""content"": ""In 2016, almost 68 percent of Medicaid beneficiaries received care through comprehensive managed care (CMC) organizations. [1] States are required to report the services provided to beneficiaries through CMC organizations in their monthly T-MSIS claims records. [2] Complete reporting of CMC encounter data ensures that TAF users can accurately identify all services received by Medicaid and CHIP beneficiaries, regardless of whether those services were delivered through managed care or the fee-for-service system. Service records from managed care plans are known as encounter records and are structured so that each encounter record is represented by one header record and one or more line-level records that link to the header. [3] Header records include summary information about the claim as a whole, whereas line records include detailed information about the individual goods and services billed as part of the claim. Users of the T-MSIS Analytic Files (TAF) must link the header- and line-level records to get all the information for an encounter.
Since states can choose the optional populations and benefit categories they cover, their Medicaid and Children’s Health Insurance Program (CHIP) programs vary accordingly in the characteristics of their covered populations, their benefit packages, and their average service use per covered beneficiary. However, examining the volume of encounter records adjusted for the number of beneficiaries enrolled in CMC plans in a state can identify outlier states that TAF users should examine more closely before beginning their analytic work. An unusually low volume of encounters may occur, for example, if (1) a state submits incomplete data on encounters or (2) missing or erroneous reporting of key data elements result in some or all of a claim not being included in the TAF. [4] In these cases, TAF users may underestimate utilization, expenditures, and the prevalence of medical conditions in beneficiaries who are enrolled in managed care. An unusually high volume of encounter records may indicate problems in how a state formatted or submitted its managed care encounter data. In some states, one or more CMC plans may not be submitting any encounter data at all, resulting in incomplete service use data in TAF.
This data quality assessment examines the volume of encounter records in each state with comprehensive managed care in its Medicaid program to identify states with potentially incomplete or incorrectly formatted data. It also identifies whether any CMC plans have no encounter data in TAF that could be linked to the plan, which would result in incomplete service use information in TAF for beneficiaries enrolled in those plans.
Centers for Medicare & Medicaid Services. “Medicaid Managed Care Enrollment and Program Characteristics, 2016.” Winter 2018. Available at https://www.medicaid.gov/medicaid/managed-care/downloads/enrollment/2017-medicaid-managed-care-enrollment-report.pdf . Accessed August 29, 2019.
Subpart J: Conditions for Federal Financial Participation (FFP). Enrollee encounter data. 42 CFR §438.818 (May 6, 2016). Available at https://www.govinfo.gov/content/pkg/CFR-2018-title42-vol4/xml/CFR-2018-title42-vol4-part438.xml#seqnum438.818 . Accessed May 29, 2020.
The previous version of the research-ready Medicaid administrative data files, the Medicaid Analytic eXtract, was structured to include one record per stay in the inpatient (IP) file, one record per claim in the long-term care (LT) and pharmacy (RX) files, and one record per claim line in the other services (OT) file.
The TAF only includes final action header records with a known service date and their associated line records. If data elements related to the service date are missing on a record that a state submits to T-MSIS, the header record and all associated lines would not be included in the TAF. In addition, TAF excludes header and line records that are not identified by the final action algorithm as representing the final version of the claim and line records that cannot be matched to a header record.
Centers for Medicare & Medicaid Services. \u201cMedicaid Managed Care Enrollment and Program Characteristics, 2016.\u201d Winter 2018. Available at https://www.medicaid.gov/medicaid/managed-care/downloads/enrollment/2017-medicaid-managed-care-enrollment-report.pdf . Accessed August 29, 2019.
Subpart J: Conditions for Federal Financial Participation (FFP). Enrollee encounter data. 42 CFR \u00a7438.818 (May 6, 2016). Available at https://www.govinfo.gov/content/pkg/CFR-2018-title42-vol4/xml/CFR-2018-title42-vol4-part438.xml#seqnum438.818 . Accessed May 29, 2020.
The previous version of the research-ready Medicaid administrative data files, the Medicaid Analytic eXtract, was structured to include one record per stay in the inpatient (IP) file, one record per claim in the long-term care (LT) and pharmacy (RX) files, and one record per claim line in the other services (OT) file.
The TAF only includes final action header records with a known service date and their associated line records. If data elements related to the service date are missing on a record that a state submits to T-MSIS, the header record and all associated lines would not be included in the TAF. In addition, TAF excludes header and line records that are not identified by the final action algorithm as representing the final version of the claim and line records that cannot be matched to a header record.
Using the TAF, [5] we tabulated the number of headers and non-denied [6] line records that were classified as managed care encounters associated with a CMC plan [7] in the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) files. The analysis included managed care encounter records for all CMC plans serving Medicaid and CHIP beneficiaries. We did not analyze states with no CMC program in operation during the year. [8] Among states with a CMC program, we excluded states from the analysis of encounter record volume in the LT and RX files if these services are not covered by their CMC programs and are typically carved out and paid on a fee-for-service basis. For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [9]
We calculated four measures to assess potential issues with the completeness or quality of the CMC encounter data in each state: (1) total volume of encounter header records, (2) total volume of encounter line records, (3) average number of encounter lines per header, and (4) number of CMC plans with no header records. The first two measures are ratios that adjust for the size of the CMC program in the state. As a result, outliers on these measures could be driven by either incomplete managed care encounters or by incomplete or inaccurate information about managed care enrollment in the eligibility data. Further, because encounter records with a missing or invalidly-formatted beneficiary identifier were dropped, a low volume of encounter header or line records could be caused by problems with the beneficiary identifier.
Total volume of encounter header records
For the first measure, we calculated the number of encounter header records per 1,000 enrolled months for each file. [10] This measure identifies states with an unusually low or an unusually high volume of header records compared with other states, while controlling for the size of the CMC program in each state, as an assessment of the overall completeness of the encounter data.
We adjusted the expected volume of service-use records according to the number of Medicaid and CHIP beneficiaries enrolled in CMC plans in the calendar year. We did not require a header record to link to an enrollment record to be included in the analysis (that is, we calculated the numerator and the denominator independent of one another). However, we excluded from the analysis the header records—and their associated line records—that had a missing beneficiary identifier or beneficiary identifier with an invalid format (values that start with “&”).
Total volume of encounter line records
For the second measure, we calculated the number of non-denied encounter line records per 1,000 enrolled months for each file. As with the header volume measure, we excluded lines that link to a header with a beneficiary identifier that was missing or in an invalid format. Line-level volume that appears low relative to the size of a state’s CMC program population can be a sign of incomplete detail on the individual goods and services billed as part of the claim. An unusually high volume may indicate a problem in how the state formatted or submitted its encounter records.
Average number of encounter lines per header
For the third measure, we calculated the average number of non-denied line records per header record for managed care encounters in each file. Each header record should have one or more associated line records, and header records with no line records indicate a data quality concern. This measure can identify states in which the header data are complete, but some of the associated line-record data are incomplete. We included all CMC encounter header records in a state’s TAF claims files when we tabulated the header volume for this measure, including headers that did not link to any line records. Line records that cannot be matched to a header record are not included in the TAF and are therefore not included in this measure.
Number of CMC plans with no header records
For the fourth measure, we tabulated the number of CMC plans that had beneficiaries enrolled in the calendar year but did not have any encounters in each claims file that could be linked to the plan. This pattern suggests that specific CMC plans are not submitting encounter records.
Overall data quality assessment
To identify states with probable data completeness or quality problems, we used the criteria presented in Table 1 for the IP and OT claims files and Table 2 for the LT and RX files. For the IP and OT files, we calculated three data quality assessment measures independently. Then, we assigned the overall level of concern for each file based on the measure with the highest data quality concern. For instance, if a state had at least one measure that was deemed unusable, the overall level of concern was deemed unusable.
Table 1. Criteria for DQ assessment of encounter record volume in the IP and OT files
| Total header volume and total line volume as a percentage of the national median | Average number of line records per header as a percentage of the national median | Number of CMC plans with no encounter header records | DQ assessment |
|---|---|---|---|
| x < 10 percent | x < 10 percent | All CMC plans | Unusable |
| 10 percent ≤ x < 50 percent | 10 percent ≤ x < 50 percent | Greater than half but not all CMC plans | High concern |
| 50 percent ≤ x < 75 percent | 50 percent ≤ x ≤ 200 percent | Greater than one but less than or equal to half of CMC plans | Medium concern |
| 75 percent ≤ x ≤ 150 percent | 50 percent ≤ x ≤ 200 percent | None | Low concern |
| 150 percent < x ≤ 200 percent | 50 percent ≤ x ≤ 200 percent | Greater than one but less than or equal to half of CMC plans | Medium concern |
| x > 200 percent | x > 200 percent | Greater than half but not all CMC plans | High concern |
For the LT and RX files, we did not classify states into the low or medium categories of concern based on the measures of total volume because CMC coverage of long-term care and prescription drug services varies widely by state and plan. [11] We would therefore expect there to be substantial variation in the volume of LT and RX encounter records across states and did not classify states that deviated from the national median as having potential data quality issues. However, we classified states into the high-concern category if the number of lines per header record in their LT or RX file averaged less than one line per header, because this pattern is highly suggestive of missing data rather than true state-by-state variation in service use or policy. We also classified states as having potential data quality issues if some CMC plans in the state were not reporting encounter data. For the LT and RX files, we calculated two data quality assessment measures independently. Then, like the IP and OT assessment, we assigned the overall level of concern for each file based on the measure with the highest data quality concern.
Table 2. Criteria for DQ assessment of encounter record volume in the LT and RX files
| Total header volume and total line volume as a percentage of the national median | Average number of line records per header as a percentage of the national median | Number of CMC plans with no encounter header records | DQ assessment |
|---|---|---|---|
| Not evaluated | More than 1 line per header AND ≥ 10 percent of the national median | None | Low concern |
| Not evaluated | More than 1 line per header AND ≥ 10 percent of the national median | Greater than one but less than or equal to half of CMC plans | Medium concern |
| Not evaluated | Less than 1 line per header AND ≥ 10 percent of the national median | Greater than half but not all of CMC plans | High concern |
| Not evaluated | <10 percent of the national median | All CMC plans | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Fully denied claims (also referred to as “denied headers”) are completely excluded from the TAF even when states submit these records in T-MSIS. Thus, all header records in the TAF represent non-denied headers. However, partly denied claims are in the TAF, including the header and both the paid and denied claim lines.
We identified managed care encounters by using claim type code (CLM_TYPE_CD) values of 3 and C. We limited the analysis to encounters with a managed care plan ID (MC_PLAN_ID) that linked to a managed care plan type code (MC_PLAN_TYPE_CD) of 01 or 04.
States with no CMC program include those with zero enrollment in CMC plans, as well as those where less than one percent of the Medicaid program is enrolled in a CMC.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
We counted enrolled months for each state by tabulating the number of months on each record in the TAF Annual Demographic and Eligibility file that had a managed care plan ID associated with a managed care plan type code (MC_PLAN_TYPE_CD_mm where mm represents a month in the calendar year) value of 01 (Comprehensive MCO) or 04 (Health Insuring Organization).
For instance, some states do not cover prescription drugs through their CMC program, do not require CMC plans to cover any long-term care services, or only require CMC plans to cover hospice care, which can be facility or home-based care. Further, variation in billing practices for stays in long-term care facilities can result in substantial variation in the overall number of claims associated with a given stay.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Fully denied claims (also referred to as \u201cdenied headers\u201d) are completely excluded from the TAF even when states submit these records in T-MSIS. Thus, all header records in the TAF represent non-denied headers. However, partly denied claims are in the TAF, including the header and both the paid and denied claim lines.
We identified managed care encounters by using claim type code (CLM_TYPE_CD) values of 3 and C. We limited the analysis to encounters with a managed care plan ID (MC_PLAN_ID) that linked to a managed care plan type code (MC_PLAN_TYPE_CD) of 01 or 04.
States with no CMC program include those with zero enrollment in CMC plans, as well as those where less than one percent of the Medicaid program is enrolled in a CMC.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
We counted enrolled months for each state by tabulating the number of months on each record in the TAF Annual Demographic and Eligibility file that had a managed care plan ID associated with a managed care plan type code (MC_PLAN_TYPE_CD_mm where mm represents a month in the calendar year) value of 01 (Comprehensive MCO) or 04 (Health Insuring Organization).
For instance, some states do not cover prescription drugs through their CMC program, do not require CMC plans to cover any long-term care services, or only require CMC plans to cover hospice care, which can be facility or home-based care. Further, variation in billing practices for stays in long-term care facilities can result in substantial variation in the overall number of claims associated with a given stay.
States are required to report into T-MSIS the encounter records that reflect services provided to Medicaid and CHIP beneficiaries by managed care organizations. Although comprehensive managed care (CMC) plans differ in the populations and benefits covered, examining the volume of CMC encounter records adjusted for plan enrollment can identify outlier states that may have incomplete data on CMC service use or enrollment in the TAF. This analysis examines the volume of CMC encounter records in the IP file as well as the number of CMC plans that have no IP encounter records.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5161"", ""relatedTopics"": [{""measureId"": 57, ""measureName"": ""CMC Plan Encounters - LT"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 1}, {""measureId"": 58, ""measureName"": ""CMC Plan Encounters - OT"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 2}, {""measureId"": 59, ""measureName"": ""CMC Plan Encounters - RX"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 3}]}" 57,"{""measureId"": 57, ""measureName"": ""CMC Plan Encounters - LT"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-CMC-Plan-Encounters-LT.pdf"", ""background"": {""content"": ""In 2016, almost 68 percent of Medicaid beneficiaries received care through comprehensive managed care (CMC) organizations. [1] States are required to report the services provided to beneficiaries through CMC organizations in their monthly T-MSIS claims records. [2] Complete reporting of CMC encounter data ensures that TAF users can accurately identify all services received by Medicaid and CHIP beneficiaries, regardless of whether those services were delivered through managed care or the fee-for-service system. Service records from managed care plans are known as encounter records and are structured so that each encounter record is represented by one header record and one or more line-level records that link to the header. [3] Header records include summary information about the claim as a whole, whereas line records include detailed information about the individual goods and services billed as part of the claim. Users of the T-MSIS Analytic Files (TAF) must link the header- and line-level records to get all the information for an encounter.
Since states can choose the optional populations and benefit categories they cover, their Medicaid and Children’s Health Insurance Program (CHIP) programs vary accordingly in the characteristics of their covered populations, their benefit packages, and their average service use per covered beneficiary. However, examining the volume of encounter records adjusted for the number of beneficiaries enrolled in CMC plans in a state can identify outlier states that TAF users should examine more closely before beginning their analytic work. An unusually low volume of encounters may occur, for example, if (1) a state submits incomplete data on encounters or (2) missing or erroneous reporting of key data elements result in some or all of a claim not being included in the TAF. [4] In these cases, TAF users may underestimate utilization, expenditures, and the prevalence of medical conditions in beneficiaries who are enrolled in managed care. An unusually high volume of encounter records may indicate problems in how a state formatted or submitted its managed care encounter data. In some states, one or more CMC plans may not be submitting any encounter data at all, resulting in incomplete service use data in TAF.
This data quality assessment examines the volume of encounter records in each state with comprehensive managed care in its Medicaid program to identify states with potentially incomplete or incorrectly formatted data. It also identifies whether any CMC plans have no encounter data in TAF that could be linked to the plan, which would result in incomplete service use information in TAF for beneficiaries enrolled in those plans.
Centers for Medicare & Medicaid Services. “Medicaid Managed Care Enrollment and Program Characteristics, 2016.” Winter 2018. Available at https://www.medicaid.gov/medicaid/managed-care/downloads/enrollment/2017-medicaid-managed-care-enrollment-report.pdf . Accessed August 29, 2019.
Subpart J: Conditions for Federal Financial Participation (FFP). Enrollee encounter data. 42 CFR §438.818 (May 6, 2016). Available at https://www.govinfo.gov/content/pkg/CFR-2018-title42-vol4/xml/CFR-2018-title42-vol4-part438.xml#seqnum438.818 . Accessed May 29, 2020.
The previous version of the research-ready Medicaid administrative data files, the Medicaid Analytic eXtract, was structured to include one record per stay in the inpatient (IP) file, one record per claim in the long-term care (LT) and pharmacy (RX) files, and one record per claim line in the other services (OT) file.
The TAF only includes final action header records with a known service date and their associated line records. If data elements related to the service date are missing on a record that a state submits to T-MSIS, the header record and all associated lines would not be included in the TAF. In addition, TAF excludes header and line records that are not identified by the final action algorithm as representing the final version of the claim and line records that cannot be matched to a header record.
Centers for Medicare & Medicaid Services. \u201cMedicaid Managed Care Enrollment and Program Characteristics, 2016.\u201d Winter 2018. Available at https://www.medicaid.gov/medicaid/managed-care/downloads/enrollment/2017-medicaid-managed-care-enrollment-report.pdf . Accessed August 29, 2019.
Subpart J: Conditions for Federal Financial Participation (FFP). Enrollee encounter data. 42 CFR \u00a7438.818 (May 6, 2016). Available at https://www.govinfo.gov/content/pkg/CFR-2018-title42-vol4/xml/CFR-2018-title42-vol4-part438.xml#seqnum438.818 . Accessed May 29, 2020.
The previous version of the research-ready Medicaid administrative data files, the Medicaid Analytic eXtract, was structured to include one record per stay in the inpatient (IP) file, one record per claim in the long-term care (LT) and pharmacy (RX) files, and one record per claim line in the other services (OT) file.
The TAF only includes final action header records with a known service date and their associated line records. If data elements related to the service date are missing on a record that a state submits to T-MSIS, the header record and all associated lines would not be included in the TAF. In addition, TAF excludes header and line records that are not identified by the final action algorithm as representing the final version of the claim and line records that cannot be matched to a header record.
Using the TAF, [5] we tabulated the number of headers and non-denied [6] line records that were classified as managed care encounters associated with a CMC plan [7] in the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) files. The analysis included managed care encounter records for all CMC plans serving Medicaid and CHIP beneficiaries. We did not analyze states with no CMC program in operation during the year. [8] Among states with a CMC program, we excluded states from the analysis of encounter record volume in the LT and RX files if these services are not covered by their CMC programs and are typically carved out and paid on a fee-for-service basis. For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [9]
We calculated four measures to assess potential issues with the completeness or quality of the CMC encounter data in each state: (1) total volume of encounter header records, (2) total volume of encounter line records, (3) average number of encounter lines per header, and (4) number of CMC plans with no header records. The first two measures are ratios that adjust for the size of the CMC program in the state. As a result, outliers on these measures could be driven by either incomplete managed care encounters or by incomplete or inaccurate information about managed care enrollment in the eligibility data. Further, because encounter records with a missing or invalidly-formatted beneficiary identifier were dropped, a low volume of encounter header or line records could be caused by problems with the beneficiary identifier.
Total volume of encounter header records
For the first measure, we calculated the number of encounter header records per 1,000 enrolled months for each file. [10] This measure identifies states with an unusually low or an unusually high volume of header records compared with other states, while controlling for the size of the CMC program in each state, as an assessment of the overall completeness of the encounter data.
We adjusted the expected volume of service-use records according to the number of Medicaid and CHIP beneficiaries enrolled in CMC plans in the calendar year. We did not require a header record to link to an enrollment record to be included in the analysis (that is, we calculated the numerator and the denominator independent of one another). However, we excluded from the analysis the header records—and their associated line records—that had a missing beneficiary identifier or beneficiary identifier with an invalid format (values that start with “&”).
Total volume of encounter line records
For the second measure, we calculated the number of non-denied encounter line records per 1,000 enrolled months for each file. As with the header volume measure, we excluded lines that link to a header with a beneficiary identifier that was missing or in an invalid format. Line-level volume that appears low relative to the size of a state’s CMC program population can be a sign of incomplete detail on the individual goods and services billed as part of the claim. An unusually high volume may indicate a problem in how the state formatted or submitted its encounter records.
Average number of encounter lines per header
For the third measure, we calculated the average number of non-denied line records per header record for managed care encounters in each file. Each header record should have one or more associated line records, and header records with no line records indicate a data quality concern. This measure can identify states in which the header data are complete, but some of the associated line-record data are incomplete. We included all CMC encounter header records in a state’s TAF claims files when we tabulated the header volume for this measure, including headers that did not link to any line records. Line records that cannot be matched to a header record are not included in the TAF and are therefore not included in this measure.
Number of CMC plans with no header records
For the fourth measure, we tabulated the number of CMC plans that had beneficiaries enrolled in the calendar year but did not have any encounters in each claims file that could be linked to the plan. This pattern suggests that specific CMC plans are not submitting encounter records.
Overall data quality assessment
To identify states with probable data completeness or quality problems, we used the criteria presented in Table 1 for the IP and OT claims files and Table 2 for the LT and RX files. For the IP and OT files, we calculated three data quality assessment measures independently. Then, we assigned the overall level of concern for each file based on the measure with the highest data quality concern. For instance, if a state had at least one measure that was deemed unusable, the overall level of concern was deemed unusable.
Table 1. Criteria for DQ assessment of encounter record volume in the IP and OT files
| Total header volume and total line volume as a percentage of the national median | Average number of line records per header as a percentage of the national median | Number of CMC plans with no encounter header records | DQ assessment |
|---|---|---|---|
| x < 10 percent | x < 10 percent | All CMC plans | Unusable |
| 10 percent ≤ x < 50 percent | 10 percent ≤ x < 50 percent | Greater than half but not all CMC plans | High concern |
| 50 percent ≤ x < 75 percent | 50 percent ≤ x ≤ 200 percent | Greater than one but less than or equal to half of CMC plans | Medium concern |
| 75 percent ≤ x ≤ 150 percent | 50 percent ≤ x ≤ 200 percent | None | Low concern |
| 150 percent < x ≤ 200 percent | 50 percent ≤ x ≤ 200 percent | Greater than one but less than or equal to half of CMC plans | Medium concern |
| x > 200 percent | x > 200 percent | Greater than half but not all CMC plans | High concern |
For the LT and RX files, we did not classify states into the low or medium categories of concern based on the measures of total volume because CMC coverage of long-term care and prescription drug services varies widely by state and plan. [11] We would therefore expect there to be substantial variation in the volume of LT and RX encounter records across states and did not classify states that deviated from the national median as having potential data quality issues. However, we classified states into the high-concern category if the number of lines per header record in their LT or RX file averaged less than one line per header, because this pattern is highly suggestive of missing data rather than true state-by-state variation in service use or policy. We also classified states as having potential data quality issues if some CMC plans in the state were not reporting encounter data. For the LT and RX files, we calculated two data quality assessment measures independently. Then, like the IP and OT assessment, we assigned the overall level of concern for each file based on the measure with the highest data quality concern.
Table 2. Criteria for DQ assessment of encounter record volume in the LT and RX files
| Total header volume and total line volume as a percentage of the national median | Average number of line records per header as a percentage of the national median | Number of CMC plans with no encounter header records | DQ assessment |
|---|---|---|---|
| Not evaluated | More than 1 line per header AND ≥ 10 percent of the national median | None | Low concern |
| Not evaluated | More than 1 line per header AND ≥ 10 percent of the national median | Greater than one but less than or equal to half of CMC plans | Medium concern |
| Not evaluated | Less than 1 line per header AND ≥ 10 percent of the national median | Greater than half but not all of CMC plans | High concern |
| Not evaluated | <10 percent of the national median | All CMC plans | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Fully denied claims (also referred to as “denied headers”) are completely excluded from the TAF even when states submit these records in T-MSIS. Thus, all header records in the TAF represent non-denied headers. However, partly denied claims are in the TAF, including the header and both the paid and denied claim lines.
We identified managed care encounters by using claim type code (CLM_TYPE_CD) values of 3 and C. We limited the analysis to encounters with a managed care plan ID (MC_PLAN_ID) that linked to a managed care plan type code (MC_PLAN_TYPE_CD) of 01 or 04.
States with no CMC program include those with zero enrollment in CMC plans, as well as those where less than one percent of the Medicaid program is enrolled in a CMC.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
We counted enrolled months for each state by tabulating the number of months on each record in the TAF Annual Demographic and Eligibility file that had a managed care plan ID associated with a managed care plan type code (MC_PLAN_TYPE_CD_mm where mm represents a month in the calendar year) value of 01 (Comprehensive MCO) or 04 (Health Insuring Organization).
For instance, some states do not cover prescription drugs through their CMC program, do not require CMC plans to cover any long-term care services, or only require CMC plans to cover hospice care, which can be facility or home-based care. Further, variation in billing practices for stays in long-term care facilities can result in substantial variation in the overall number of claims associated with a given stay.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Fully denied claims (also referred to as \u201cdenied headers\u201d) are completely excluded from the TAF even when states submit these records in T-MSIS. Thus, all header records in the TAF represent non-denied headers. However, partly denied claims are in the TAF, including the header and both the paid and denied claim lines.
We identified managed care encounters by using claim type code (CLM_TYPE_CD) values of 3 and C. We limited the analysis to encounters with a managed care plan ID (MC_PLAN_ID) that linked to a managed care plan type code (MC_PLAN_TYPE_CD) of 01 or 04.
States with no CMC program include those with zero enrollment in CMC plans, as well as those where less than one percent of the Medicaid program is enrolled in a CMC.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
We counted enrolled months for each state by tabulating the number of months on each record in the TAF Annual Demographic and Eligibility file that had a managed care plan ID associated with a managed care plan type code (MC_PLAN_TYPE_CD_mm where mm represents a month in the calendar year) value of 01 (Comprehensive MCO) or 04 (Health Insuring Organization).
For instance, some states do not cover prescription drugs through their CMC program, do not require CMC plans to cover any long-term care services, or only require CMC plans to cover hospice care, which can be facility or home-based care. Further, variation in billing practices for stays in long-term care facilities can result in substantial variation in the overall number of claims associated with a given stay.
States are required to report into T-MSIS the encounter records that reflect services provided to Medicaid and CHIP beneficiaries by managed care organizations. This analysis examines the average number of CMC encounter lines per header in the LT file and the number of CMC plans that have no LT encounter records in order to identify states with incomplete CMC service use data.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5161"", ""relatedTopics"": [{""measureId"": 56, ""measureName"": ""CMC Plan Encounters - IP"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 0}, {""measureId"": 58, ""measureName"": ""CMC Plan Encounters - OT"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 2}, {""measureId"": 59, ""measureName"": ""CMC Plan Encounters - RX"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 3}]}" 58,"{""measureId"": 58, ""measureName"": ""CMC Plan Encounters - OT"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-CMC-Plan-Encounters-OT.pdf"", ""background"": {""content"": ""In 2016, almost 68 percent of Medicaid beneficiaries received care through comprehensive managed care (CMC) organizations. [1] States are required to report the services provided to beneficiaries through CMC organizations in their monthly T-MSIS claims records. [2] Complete reporting of CMC encounter data ensures that TAF users can accurately identify all services received by Medicaid and CHIP beneficiaries, regardless of whether those services were delivered through managed care or the fee-for-service system. Service records from managed care plans are known as encounter records and are structured so that each encounter record is represented by one header record and one or more line-level records that link to the header. [3] Header records include summary information about the claim as a whole, whereas line records include detailed information about the individual goods and services billed as part of the claim. Users of the T-MSIS Analytic Files (TAF) must link the header- and line-level records to get all the information for an encounter.
Since states can choose the optional populations and benefit categories they cover, their Medicaid and Children’s Health Insurance Program (CHIP) programs vary accordingly in the characteristics of their covered populations, their benefit packages, and their average service use per covered beneficiary. However, examining the volume of encounter records adjusted for the number of beneficiaries enrolled in CMC plans in a state can identify outlier states that TAF users should examine more closely before beginning their analytic work. An unusually low volume of encounters may occur, for example, if (1) a state submits incomplete data on encounters or (2) missing or erroneous reporting of key data elements result in some or all of a claim not being included in the TAF. [4] In these cases, TAF users may underestimate utilization, expenditures, and the prevalence of medical conditions in beneficiaries who are enrolled in managed care. An unusually high volume of encounter records may indicate problems in how a state formatted or submitted its managed care encounter data. In some states, one or more CMC plans may not be submitting any encounter data at all, resulting in incomplete service use data in TAF.
This data quality assessment examines the volume of encounter records in each state with comprehensive managed care in its Medicaid program to identify states with potentially incomplete or incorrectly formatted data. It also identifies whether any CMC plans have no encounter data in TAF that could be linked to the plan, which would result in incomplete service use information in TAF for beneficiaries enrolled in those plans.
Centers for Medicare & Medicaid Services. “Medicaid Managed Care Enrollment and Program Characteristics, 2016.” Winter 2018. Available at https://www.medicaid.gov/medicaid/managed-care/downloads/enrollment/2017-medicaid-managed-care-enrollment-report.pdf . Accessed August 29, 2019.
Subpart J: Conditions for Federal Financial Participation (FFP). Enrollee encounter data. 42 CFR §438.818 (May 6, 2016). Available at https://www.govinfo.gov/content/pkg/CFR-2018-title42-vol4/xml/CFR-2018-title42-vol4-part438.xml#seqnum438.818 . Accessed May 29, 2020.
The previous version of the research-ready Medicaid administrative data files, the Medicaid Analytic eXtract, was structured to include one record per stay in the inpatient (IP) file, one record per claim in the long-term care (LT) and pharmacy (RX) files, and one record per claim line in the other services (OT) file.
The TAF only includes final action header records with a known service date and their associated line records. If data elements related to the service date are missing on a record that a state submits to T-MSIS, the header record and all associated lines would not be included in the TAF. In addition, TAF excludes header and line records that are not identified by the final action algorithm as representing the final version of the claim and line records that cannot be matched to a header record.
Centers for Medicare & Medicaid Services. \u201cMedicaid Managed Care Enrollment and Program Characteristics, 2016.\u201d Winter 2018. Available at https://www.medicaid.gov/medicaid/managed-care/downloads/enrollment/2017-medicaid-managed-care-enrollment-report.pdf . Accessed August 29, 2019.
Subpart J: Conditions for Federal Financial Participation (FFP). Enrollee encounter data. 42 CFR \u00a7438.818 (May 6, 2016). Available at https://www.govinfo.gov/content/pkg/CFR-2018-title42-vol4/xml/CFR-2018-title42-vol4-part438.xml#seqnum438.818 . Accessed May 29, 2020.
The previous version of the research-ready Medicaid administrative data files, the Medicaid Analytic eXtract, was structured to include one record per stay in the inpatient (IP) file, one record per claim in the long-term care (LT) and pharmacy (RX) files, and one record per claim line in the other services (OT) file.
The TAF only includes final action header records with a known service date and their associated line records. If data elements related to the service date are missing on a record that a state submits to T-MSIS, the header record and all associated lines would not be included in the TAF. In addition, TAF excludes header and line records that are not identified by the final action algorithm as representing the final version of the claim and line records that cannot be matched to a header record.
Using the TAF, [5] we tabulated the number of headers and non-denied [6] line records that were classified as managed care encounters associated with a CMC plan [7] in the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) files. The analysis included managed care encounter records for all CMC plans serving Medicaid and CHIP beneficiaries. We did not analyze states with no CMC program in operation during the year. [8] Among states with a CMC program, we excluded states from the analysis of encounter record volume in the LT and RX files if these services are not covered by their CMC programs and are typically carved out and paid on a fee-for-service basis. For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [9]
We calculated four measures to assess potential issues with the completeness or quality of the CMC encounter data in each state: (1) total volume of encounter header records, (2) total volume of encounter line records, (3) average number of encounter lines per header, and (4) number of CMC plans with no header records. The first two measures are ratios that adjust for the size of the CMC program in the state. As a result, outliers on these measures could be driven by either incomplete managed care encounters or by incomplete or inaccurate information about managed care enrollment in the eligibility data. Further, because encounter records with a missing or invalidly-formatted beneficiary identifier were dropped, a low volume of encounter header or line records could be caused by problems with the beneficiary identifier.
Total volume of encounter header records
For the first measure, we calculated the number of encounter header records per 1,000 enrolled months for each file. [10] This measure identifies states with an unusually low or an unusually high volume of header records compared with other states, while controlling for the size of the CMC program in each state, as an assessment of the overall completeness of the encounter data.
We adjusted the expected volume of service-use records according to the number of Medicaid and CHIP beneficiaries enrolled in CMC plans in the calendar year. We did not require a header record to link to an enrollment record to be included in the analysis (that is, we calculated the numerator and the denominator independent of one another). However, we excluded from the analysis the header records—and their associated line records—that had a missing beneficiary identifier or beneficiary identifier with an invalid format (values that start with “&”).
Total volume of encounter line records
For the second measure, we calculated the number of non-denied encounter line records per 1,000 enrolled months for each file. As with the header volume measure, we excluded lines that link to a header with a beneficiary identifier that was missing or in an invalid format. Line-level volume that appears low relative to the size of a state’s CMC program population can be a sign of incomplete detail on the individual goods and services billed as part of the claim. An unusually high volume may indicate a problem in how the state formatted or submitted its encounter records.
Average number of encounter lines per header
For the third measure, we calculated the average number of non-denied line records per header record for managed care encounters in each file. Each header record should have one or more associated line records, and header records with no line records indicate a data quality concern. This measure can identify states in which the header data are complete, but some of the associated line-record data are incomplete. We included all CMC encounter header records in a state’s TAF claims files when we tabulated the header volume for this measure, including headers that did not link to any line records. Line records that cannot be matched to a header record are not included in the TAF and are therefore not included in this measure.
Number of CMC plans with no header records
For the fourth measure, we tabulated the number of CMC plans that had beneficiaries enrolled in the calendar year but did not have any encounters in each claims file that could be linked to the plan. This pattern suggests that specific CMC plans are not submitting encounter records.
Overall data quality assessment
To identify states with probable data completeness or quality problems, we used the criteria presented in Table 1 for the IP and OT claims files and Table 2 for the LT and RX files. For the IP and OT files, we calculated three data quality assessment measures independently. Then, we assigned the overall level of concern for each file based on the measure with the highest data quality concern. For instance, if a state had at least one measure that was deemed unusable, the overall level of concern was deemed unusable.
Table 1. Criteria for DQ assessment of encounter record volume in the IP and OT files
| Total header volume and total line volume as a percentage of the national median | Average number of line records per header as a percentage of the national median | Number of CMC plans with no encounter header records | DQ assessment |
|---|---|---|---|
| x < 10 percent | x < 10 percent | All CMC plans | Unusable |
| 10 percent ≤ x < 50 percent | 10 percent ≤ x < 50 percent | Greater than half but not all CMC plans | High concern |
| 50 percent ≤ x < 75 percent | 50 percent ≤ x ≤ 200 percent | Greater than one but less than or equal to half of CMC plans | Medium concern |
| 75 percent ≤ x ≤ 150 percent | 50 percent ≤ x ≤ 200 percent | None | Low concern |
| 150 percent < x ≤ 200 percent | 50 percent ≤ x ≤ 200 percent | Greater than one but less than or equal to half of CMC plans | Medium concern |
| x > 200 percent | x > 200 percent | Greater than half but not all CMC plans | High concern |
For the LT and RX files, we did not classify states into the low or medium categories of concern based on the measures of total volume because CMC coverage of long-term care and prescription drug services varies widely by state and plan. [11] We would therefore expect there to be substantial variation in the volume of LT and RX encounter records across states and did not classify states that deviated from the national median as having potential data quality issues. However, we classified states into the high-concern category if the number of lines per header record in their LT or RX file averaged less than one line per header, because this pattern is highly suggestive of missing data rather than true state-by-state variation in service use or policy. We also classified states as having potential data quality issues if some CMC plans in the state were not reporting encounter data. For the LT and RX files, we calculated two data quality assessment measures independently. Then, like the IP and OT assessment, we assigned the overall level of concern for each file based on the measure with the highest data quality concern.
Table 2. Criteria for DQ assessment of encounter record volume in the LT and RX files
| Total header volume and total line volume as a percentage of the national median | Average number of line records per header as a percentage of the national median | Number of CMC plans with no encounter header records | DQ assessment |
|---|---|---|---|
| Not evaluated | More than 1 line per header AND ≥ 10 percent of the national median | None | Low concern |
| Not evaluated | More than 1 line per header AND ≥ 10 percent of the national median | Greater than one but less than or equal to half of CMC plans | Medium concern |
| Not evaluated | Less than 1 line per header AND ≥ 10 percent of the national median | Greater than half but not all of CMC plans | High concern |
| Not evaluated | <10 percent of the national median | All CMC plans | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Fully denied claims (also referred to as “denied headers”) are completely excluded from the TAF even when states submit these records in T-MSIS. Thus, all header records in the TAF represent non-denied headers. However, partly denied claims are in the TAF, including the header and both the paid and denied claim lines.
We identified managed care encounters by using claim type code (CLM_TYPE_CD) values of 3 and C. We limited the analysis to encounters with a managed care plan ID (MC_PLAN_ID) that linked to a managed care plan type code (MC_PLAN_TYPE_CD) of 01 or 04.
States with no CMC program include those with zero enrollment in CMC plans, as well as those where less than one percent of the Medicaid program is enrolled in a CMC.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
We counted enrolled months for each state by tabulating the number of months on each record in the TAF Annual Demographic and Eligibility file that had a managed care plan ID associated with a managed care plan type code (MC_PLAN_TYPE_CD_mm where mm represents a month in the calendar year) value of 01 (Comprehensive MCO) or 04 (Health Insuring Organization).
For instance, some states do not cover prescription drugs through their CMC program, do not require CMC plans to cover any long-term care services, or only require CMC plans to cover hospice care, which can be facility or home-based care. Further, variation in billing practices for stays in long-term care facilities can result in substantial variation in the overall number of claims associated with a given stay.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Fully denied claims (also referred to as \u201cdenied headers\u201d) are completely excluded from the TAF even when states submit these records in T-MSIS. Thus, all header records in the TAF represent non-denied headers. However, partly denied claims are in the TAF, including the header and both the paid and denied claim lines.
We identified managed care encounters by using claim type code (CLM_TYPE_CD) values of 3 and C. We limited the analysis to encounters with a managed care plan ID (MC_PLAN_ID) that linked to a managed care plan type code (MC_PLAN_TYPE_CD) of 01 or 04.
States with no CMC program include those with zero enrollment in CMC plans, as well as those where less than one percent of the Medicaid program is enrolled in a CMC.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
We counted enrolled months for each state by tabulating the number of months on each record in the TAF Annual Demographic and Eligibility file that had a managed care plan ID associated with a managed care plan type code (MC_PLAN_TYPE_CD_mm where mm represents a month in the calendar year) value of 01 (Comprehensive MCO) or 04 (Health Insuring Organization).
For instance, some states do not cover prescription drugs through their CMC program, do not require CMC plans to cover any long-term care services, or only require CMC plans to cover hospice care, which can be facility or home-based care. Further, variation in billing practices for stays in long-term care facilities can result in substantial variation in the overall number of claims associated with a given stay.
States are required to report into T-MSIS the encounter records that reflect services provided to Medicaid and CHIP beneficiaries by managed care organizations. Although comprehensive managed care (CMC) plans differ in the populations and benefits covered, examining the volume of CMC encounter records adjusted for plan enrollment can identify outlier states that may have incomplete data on comprehensive managed care service use or enrollment in the TAF. This analysis examines the volume of CMC encounter records in the OT file as well as the number of CMC plans that have no OT encounter records.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5161"", ""relatedTopics"": [{""measureId"": 56, ""measureName"": ""CMC Plan Encounters - IP"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 0}, {""measureId"": 57, ""measureName"": ""CMC Plan Encounters - LT"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 1}, {""measureId"": 59, ""measureName"": ""CMC Plan Encounters - RX"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 3}]}" 59,"{""measureId"": 59, ""measureName"": ""CMC Plan Encounters - RX"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-CMC-Plan-Encounters-RX.pdf"", ""background"": {""content"": ""In 2016, almost 68 percent of Medicaid beneficiaries received care through comprehensive managed care (CMC) organizations. [1] States are required to report the services provided to beneficiaries through CMC organizations in their monthly T-MSIS claims records. [2] Complete reporting of CMC encounter data ensures that TAF users can accurately identify all services received by Medicaid and CHIP beneficiaries, regardless of whether those services were delivered through managed care or the fee-for-service system. Service records from managed care plans are known as encounter records and are structured so that each encounter record is represented by one header record and one or more line-level records that link to the header. [3] Header records include summary information about the claim as a whole, whereas line records include detailed information about the individual goods and services billed as part of the claim. Users of the T-MSIS Analytic Files (TAF) must link the header- and line-level records to get all the information for an encounter.
Since states can choose the optional populations and benefit categories they cover, their Medicaid and Children’s Health Insurance Program (CHIP) programs vary accordingly in the characteristics of their covered populations, their benefit packages, and their average service use per covered beneficiary. However, examining the volume of encounter records adjusted for the number of beneficiaries enrolled in CMC plans in a state can identify outlier states that TAF users should examine more closely before beginning their analytic work. An unusually low volume of encounters may occur, for example, if (1) a state submits incomplete data on encounters or (2) missing or erroneous reporting of key data elements result in some or all of a claim not being included in the TAF. [4] In these cases, TAF users may underestimate utilization, expenditures, and the prevalence of medical conditions in beneficiaries who are enrolled in managed care. An unusually high volume of encounter records may indicate problems in how a state formatted or submitted its managed care encounter data. In some states, one or more CMC plans may not be submitting any encounter data at all, resulting in incomplete service use data in TAF.
This data quality assessment examines the volume of encounter records in each state with comprehensive managed care in its Medicaid program to identify states with potentially incomplete or incorrectly formatted data. It also identifies whether any CMC plans have no encounter data in TAF that could be linked to the plan, which would result in incomplete service use information in TAF for beneficiaries enrolled in those plans.
Centers for Medicare & Medicaid Services. “Medicaid Managed Care Enrollment and Program Characteristics, 2016.” Winter 2018. Available at https://www.medicaid.gov/medicaid/managed-care/downloads/enrollment/2017-medicaid-managed-care-enrollment-report.pdf . Accessed August 29, 2019.
Subpart J: Conditions for Federal Financial Participation (FFP). Enrollee encounter data. 42 CFR §438.818 (May 6, 2016). Available at https://www.govinfo.gov/content/pkg/CFR-2018-title42-vol4/xml/CFR-2018-title42-vol4-part438.xml#seqnum438.818 . Accessed May 29, 2020.
The previous version of the research-ready Medicaid administrative data files, the Medicaid Analytic eXtract, was structured to include one record per stay in the inpatient (IP) file, one record per claim in the long-term care (LT) and pharmacy (RX) files, and one record per claim line in the other services (OT) file.
The TAF only includes final action header records with a known service date and their associated line records. If data elements related to the service date are missing on a record that a state submits to T-MSIS, the header record and all associated lines would not be included in the TAF. In addition, TAF excludes header and line records that are not identified by the final action algorithm as representing the final version of the claim and line records that cannot be matched to a header record.
Centers for Medicare & Medicaid Services. \u201cMedicaid Managed Care Enrollment and Program Characteristics, 2016.\u201d Winter 2018. Available at https://www.medicaid.gov/medicaid/managed-care/downloads/enrollment/2017-medicaid-managed-care-enrollment-report.pdf . Accessed August 29, 2019.
Subpart J: Conditions for Federal Financial Participation (FFP). Enrollee encounter data. 42 CFR \u00a7438.818 (May 6, 2016). Available at https://www.govinfo.gov/content/pkg/CFR-2018-title42-vol4/xml/CFR-2018-title42-vol4-part438.xml#seqnum438.818 . Accessed May 29, 2020.
The previous version of the research-ready Medicaid administrative data files, the Medicaid Analytic eXtract, was structured to include one record per stay in the inpatient (IP) file, one record per claim in the long-term care (LT) and pharmacy (RX) files, and one record per claim line in the other services (OT) file.
The TAF only includes final action header records with a known service date and their associated line records. If data elements related to the service date are missing on a record that a state submits to T-MSIS, the header record and all associated lines would not be included in the TAF. In addition, TAF excludes header and line records that are not identified by the final action algorithm as representing the final version of the claim and line records that cannot be matched to a header record.
Using the TAF, [5] we tabulated the number of headers and non-denied [6] line records that were classified as managed care encounters associated with a CMC plan [7] in the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) files. The analysis included managed care encounter records for all CMC plans serving Medicaid and CHIP beneficiaries. We did not analyze states with no CMC program in operation during the year. [8] Among states with a CMC program, we excluded states from the analysis of encounter record volume in the LT and RX files if these services are not covered by their CMC programs and are typically carved out and paid on a fee-for-service basis. For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [9]
We calculated four measures to assess potential issues with the completeness or quality of the CMC encounter data in each state: (1) total volume of encounter header records, (2) total volume of encounter line records, (3) average number of encounter lines per header, and (4) number of CMC plans with no header records. The first two measures are ratios that adjust for the size of the CMC program in the state. As a result, outliers on these measures could be driven by either incomplete managed care encounters or by incomplete or inaccurate information about managed care enrollment in the eligibility data. Further, because encounter records with a missing or invalidly-formatted beneficiary identifier were dropped, a low volume of encounter header or line records could be caused by problems with the beneficiary identifier.
Total volume of encounter header records
For the first measure, we calculated the number of encounter header records per 1,000 enrolled months for each file. [10] This measure identifies states with an unusually low or an unusually high volume of header records compared with other states, while controlling for the size of the CMC program in each state, as an assessment of the overall completeness of the encounter data.
We adjusted the expected volume of service-use records according to the number of Medicaid and CHIP beneficiaries enrolled in CMC plans in the calendar year. We did not require a header record to link to an enrollment record to be included in the analysis (that is, we calculated the numerator and the denominator independent of one another). However, we excluded from the analysis the header records—and their associated line records—that had a missing beneficiary identifier or beneficiary identifier with an invalid format (values that start with “&”).
Total volume of encounter line records
For the second measure, we calculated the number of non-denied encounter line records per 1,000 enrolled months for each file. As with the header volume measure, we excluded lines that link to a header with a beneficiary identifier that was missing or in an invalid format. Line-level volume that appears low relative to the size of a state’s CMC program population can be a sign of incomplete detail on the individual goods and services billed as part of the claim. An unusually high volume may indicate a problem in how the state formatted or submitted its encounter records.
Average number of encounter lines per header
For the third measure, we calculated the average number of non-denied line records per header record for managed care encounters in each file. Each header record should have one or more associated line records, and header records with no line records indicate a data quality concern. This measure can identify states in which the header data are complete, but some of the associated line-record data are incomplete. We included all CMC encounter header records in a state’s TAF claims files when we tabulated the header volume for this measure, including headers that did not link to any line records. Line records that cannot be matched to a header record are not included in the TAF and are therefore not included in this measure.
Number of CMC plans with no header records
For the fourth measure, we tabulated the number of CMC plans that had beneficiaries enrolled in the calendar year but did not have any encounters in each claims file that could be linked to the plan. This pattern suggests that specific CMC plans are not submitting encounter records.
Overall data quality assessment
To identify states with probable data completeness or quality problems, we used the criteria presented in Table 1 for the IP and OT claims files and Table 2 for the LT and RX files. For the IP and OT files, we calculated three data quality assessment measures independently. Then, we assigned the overall level of concern for each file based on the measure with the highest data quality concern. For instance, if a state had at least one measure that was deemed unusable, the overall level of concern was deemed unusable.
Table 1. Criteria for DQ assessment of encounter record volume in the IP and OT files
| Total header volume and total line volume as a percentage of the national median | Average number of line records per header as a percentage of the national median | Number of CMC plans with no encounter header records | DQ assessment |
|---|---|---|---|
| x < 10 percent | x < 10 percent | All CMC plans | Unusable |
| 10 percent ≤ x < 50 percent | 10 percent ≤ x < 50 percent | Greater than half but not all CMC plans | High concern |
| 50 percent ≤ x < 75 percent | 50 percent ≤ x ≤ 200 percent | Greater than one but less than or equal to half of CMC plans | Medium concern |
| 75 percent ≤ x ≤ 150 percent | 50 percent ≤ x ≤ 200 percent | None | Low concern |
| 150 percent < x ≤ 200 percent | 50 percent ≤ x ≤ 200 percent | Greater than one but less than or equal to half of CMC plans | Medium concern |
| x > 200 percent | x > 200 percent | Greater than half but not all CMC plans | High concern |
For the LT and RX files, we did not classify states into the low or medium categories of concern based on the measures of total volume because CMC coverage of long-term care and prescription drug services varies widely by state and plan. [11] We would therefore expect there to be substantial variation in the volume of LT and RX encounter records across states and did not classify states that deviated from the national median as having potential data quality issues. However, we classified states into the high-concern category if the number of lines per header record in their LT or RX file averaged less than one line per header, because this pattern is highly suggestive of missing data rather than true state-by-state variation in service use or policy. We also classified states as having potential data quality issues if some CMC plans in the state were not reporting encounter data. For the LT and RX files, we calculated two data quality assessment measures independently. Then, like the IP and OT assessment, we assigned the overall level of concern for each file based on the measure with the highest data quality concern.
Table 2. Criteria for DQ assessment of encounter record volume in the LT and RX files
| Total header volume and total line volume as a percentage of the national median | Average number of line records per header as a percentage of the national median | Number of CMC plans with no encounter header records | DQ assessment |
|---|---|---|---|
| Not evaluated | More than 1 line per header AND ≥ 10 percent of the national median | None | Low concern |
| Not evaluated | More than 1 line per header AND ≥ 10 percent of the national median | Greater than one but less than or equal to half of CMC plans | Medium concern |
| Not evaluated | Less than 1 line per header AND ≥ 10 percent of the national median | Greater than half but not all of CMC plans | High concern |
| Not evaluated | <10 percent of the national median | All CMC plans | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Fully denied claims (also referred to as “denied headers”) are completely excluded from the TAF even when states submit these records in T-MSIS. Thus, all header records in the TAF represent non-denied headers. However, partly denied claims are in the TAF, including the header and both the paid and denied claim lines.
We identified managed care encounters by using claim type code (CLM_TYPE_CD) values of 3 and C. We limited the analysis to encounters with a managed care plan ID (MC_PLAN_ID) that linked to a managed care plan type code (MC_PLAN_TYPE_CD) of 01 or 04.
States with no CMC program include those with zero enrollment in CMC plans, as well as those where less than one percent of the Medicaid program is enrolled in a CMC.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
We counted enrolled months for each state by tabulating the number of months on each record in the TAF Annual Demographic and Eligibility file that had a managed care plan ID associated with a managed care plan type code (MC_PLAN_TYPE_CD_mm where mm represents a month in the calendar year) value of 01 (Comprehensive MCO) or 04 (Health Insuring Organization).
For instance, some states do not cover prescription drugs through their CMC program, do not require CMC plans to cover any long-term care services, or only require CMC plans to cover hospice care, which can be facility or home-based care. Further, variation in billing practices for stays in long-term care facilities can result in substantial variation in the overall number of claims associated with a given stay.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Fully denied claims (also referred to as \u201cdenied headers\u201d) are completely excluded from the TAF even when states submit these records in T-MSIS. Thus, all header records in the TAF represent non-denied headers. However, partly denied claims are in the TAF, including the header and both the paid and denied claim lines.
We identified managed care encounters by using claim type code (CLM_TYPE_CD) values of 3 and C. We limited the analysis to encounters with a managed care plan ID (MC_PLAN_ID) that linked to a managed care plan type code (MC_PLAN_TYPE_CD) of 01 or 04.
States with no CMC program include those with zero enrollment in CMC plans, as well as those where less than one percent of the Medicaid program is enrolled in a CMC.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
We counted enrolled months for each state by tabulating the number of months on each record in the TAF Annual Demographic and Eligibility file that had a managed care plan ID associated with a managed care plan type code (MC_PLAN_TYPE_CD_mm where mm represents a month in the calendar year) value of 01 (Comprehensive MCO) or 04 (Health Insuring Organization).
For instance, some states do not cover prescription drugs through their CMC program, do not require CMC plans to cover any long-term care services, or only require CMC plans to cover hospice care, which can be facility or home-based care. Further, variation in billing practices for stays in long-term care facilities can result in substantial variation in the overall number of claims associated with a given stay.
States are required to report into T-MSIS the encounter records that reflect services provided to Medicaid and CHIP beneficiaries by managed care organizations. This analysis examines the average number of CMC encounter lines per header in the RX file and the number of CMC plans that have no RX encounter records in order to identify states with incomplete CMC service use data.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5161"", ""relatedTopics"": [{""measureId"": 56, ""measureName"": ""CMC Plan Encounters - IP"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 0}, {""measureId"": 57, ""measureName"": ""CMC Plan Encounters - LT"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 1}, {""measureId"": 58, ""measureName"": ""CMC Plan Encounters - OT"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""order"": 2}]}" 60,"{""measureId"": 60, ""measureName"": ""Supplemental Payments"", ""groupId"": 7, ""groupName"": ""Non-Claim Records"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Supplemental-Pmts.pdf"", ""background"": {""content"": ""States submit several types of service use and financial transaction records into T-MSIS: (1) fee-for-service (FFS) claims, which represent payments made directly to providers for services rendered to Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries; (2) capitated payments, which represent per-person-per-month payments made to managed care plans; (3) managed care encounters, which represent services rendered to beneficiaries covered under a capitation arrangement; (4) service tracking claims, which represent lump sum payments that cannot be attributed to a single beneficiary (for instance, drug rebates or disproportionate share hospital payments); and (5) supplemental payments, which represent additional payments for services provided to a specific beneficiary.
All of these records are available in the T-MSIS Analytic Files (TAF) Research Identifiable Files (RIFs). [1] Any analysis that uses these data should include only the types of records that will help answer the research question at hand. For example, tabulations of total state Medicaid expenditures should include FFS claims, capitated payments, service tracking claims, and supplemental payments, but they should exclude managed care encounters, which do not represent payments made by the state. In contrast, analyses of beneficiary service use should include FFS claims and managed care encounters (when applicable), but they should exclude capitated payments, service tracking claims, and supplemental payments.
TAF users can distinguish FFS claims, capitated payments, managed care encounters, service tracking claims, supplemental payments, and other payments by using the claim type code found on header records that summarize the claim. The claim type code can also be used to differentiate between service use and payments for Medicaid beneficiaries (including M-CHIP) and for separate CHIP beneficiaries. In addition, claim type code includes values for “other” program type records, which may or may not represent services that qualify for federal matching funds under Title XIX or Title XXI. These records are referred to as “other” or “non-program” records and have claim type code equal to U, V, W, X, or Y.
The distribution of claim type in a given state is driven by how that state implemented its Medicaid and CHIP programs as well as the capacity of the state’s Medicaid Management Information System to report each claim type.
This data quality assessment examines the reporting and usability of service tracking claims, supplemental payments and other, non-program claims. [2]
Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
For more information about what types of claims are in TAF and potential data quality issues, see “TAF Technical Guidance: Claims Files” on ResDAC.org.
Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
For more information about what types of claims are in TAF and potential data quality issues, see \u201cTAF Technical Guidance: Claims Files\u201d on ResDAC.org.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We examined service tracking claims, supplemental payments and other, non-program claims using claim type code (CLM_TYPE_CD) on records in the TAF inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims files. [3] Service tracking claims have claim type code values of 4 or D. Supplemental claims have claim type code values of 5 or E. Other, non-program claims have claim type code values of U, V, W, X, or Y. For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [4]
Service tracking claims
The MSIS identification number (MSIS_IDENT_NUM), service tracking type code (SRVC_TRKNG_TYPE_CD), and service tracking payment amount (SRVC_TRKNG_PYMT_AMT) are key data elements for identifying service tracking claims. Because these claims represent lump sum payments that cannot be attributed to a single Medicaid or CHIP beneficiary, the MSIS identification number should begin with an “&” or be missing. In addition, the service tracking type and payment amount needs to be populated for the claim to be useful for analytic purposes. We classified a claim as having a data quality problem if it met at least one of following conditions:
We grouped states into categories of low, medium, and high data quality concern based on the percentage of their records that were problematic according to any of the three conditions.
Table 1. Criteria for DQ assessment of service tracking claims
| Percentage of claims with a problematic value in the MSIS ID, service tracking type, or service tracking payment field | DQ assessment |
|---|---|
| 0 percent ≤ x < 10 percent | Low concern |
| 10 percent ≤ x < 20 percent | Medium concern |
| 20 percent ≤ x < 50 percent | High concern |
| x ≥ 50 percent | Unusable |
Supplemental payment records
The total Medicaid paid amount (TOT_MDCD_PD_AMT) is a key data element for supplemental payments. If there is no information in the payment field, the record cannot be used for research. These records should also have (1) a valid MSIS ID that does not begin with an “&”; (2) a missing service tracking type; and (3) a missing or zero service tracking payment amount. [5] We classified a claim as having a data quality problem if it met any one of the following conditions:
We grouped the states into low, medium, high, and unusable categories of concern based on the percentage of their supplemental payment records that were problematic according to any of the four conditions.
Table 2. Criteria for DQ assessment of supplemental payment records
| Percentage of claims with a problematic value in the total Medicaid paid amount, MSIS ID, service tracking type, or service tracking payment field | DQ assessment |
|---|---|
| 0 percent ≤ x < 10 percent | Low concern |
| 10 percent ≤ x < 20 percent | Medium concern |
| 20 percent ≤ x < 50 percent | High concern |
| x ≥ 50 percent | Unusable |
Other, non-program claims
TAF users should be aware if the data files for a state include records for other, non-program claims so that they can determine whether to include these claims in analyses. These records may represent services and payments that do not qualify for a federal match or may represent services that receive a federal match that the state wants to distinguish for some reason. Because states were not given specific guidance on what claims should be reported as other and because states are not required to justify why they have submitted a claim with the other, non-program code values, it is unclear what these claims represent in every state and whether a state’s use of this claim type changes over time. The lack of detail for these claims may make them unusable for many analytic purposes. We classified the state as high or low concern based on the percentage of all claims for the state that have an other, non-program claim type code (Table 3).
Table 3. Criteria for DQ assessment of other, non-program claims
| Percentage of claims that have an other, non-program claim type code | DQ assessment |
|---|---|
| x < 2 percent | Low concern |
| x ≥ 2 percent | High concern |
Methods previously used to assess data quality
Table 4 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 4. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND=0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
If the fields for service tracking type and service tracking amount are populated or if the MSIS ID is missing or starts with an “&”, then a TAF user cannot determine whether the record is a supplemental payment or a service tracking record.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND=0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
If the fields for service tracking type and service tracking amount are populated or if the MSIS ID is missing or starts with an \u201c&\u201d, then a TAF user cannot determine whether the record is a supplemental payment or a service tracking record.
States submit several types of service use and financial transaction records into T-MSIS. One type of financial transaction is supplemental payments, which represent additional payments beyond the negotiated rate for a service provided to a specific beneficiary. This analysis identifies unexpected coding patterns for supplemental payments that may indicate data quality problems.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5181"", ""relatedTopics"": [{""measureId"": 61, ""measureName"": ""Non-Program (Other) Claims"", ""groupId"": 7, ""groupName"": ""Non-Claim Records"", ""order"": 1}, {""measureId"": 85, ""measureName"": ""Service Tracking Claims"", ""groupId"": 7, ""groupName"": ""Non-Claim Records"", ""order"": 2}]}" 61,"{""measureId"": 61, ""measureName"": ""Non-Program (Other) Claims"", ""groupId"": 7, ""groupName"": ""Non-Claim Records"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-NonProgram-Other-Claims.pdf"", ""background"": {""content"": ""States submit several types of service use and financial transaction records into T-MSIS: (1) fee-for-service (FFS) claims, which represent payments made directly to providers for services rendered to Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries; (2) capitated payments, which represent per-person-per-month payments made to managed care plans; (3) managed care encounters, which represent services rendered to beneficiaries covered under a capitation arrangement; (4) service tracking claims, which represent lump sum payments that cannot be attributed to a single beneficiary (for instance, drug rebates or disproportionate share hospital payments); and (5) supplemental payments, which represent additional payments for services provided to a specific beneficiary.
All of these records are available in the T-MSIS Analytic Files (TAF) Research Identifiable Files (RIFs). [1] Any analysis that uses these data should include only the types of records that will help answer the research question at hand. For example, tabulations of total state Medicaid expenditures should include FFS claims, capitated payments, service tracking claims, and supplemental payments, but they should exclude managed care encounters, which do not represent payments made by the state. In contrast, analyses of beneficiary service use should include FFS claims and managed care encounters (when applicable), but they should exclude capitated payments, service tracking claims, and supplemental payments.
TAF users can distinguish FFS claims, capitated payments, managed care encounters, service tracking claims, supplemental payments, and other payments by using the claim type code found on header records that summarize the claim. The claim type code can also be used to differentiate between service use and payments for Medicaid beneficiaries (including M-CHIP) and for separate CHIP beneficiaries. In addition, claim type code includes values for “other” program type records, which may or may not represent services that qualify for federal matching funds under Title XIX or Title XXI. These records are referred to as “other” or “non-program” records and have claim type code equal to U, V, W, X, or Y.
The distribution of claim type in a given state is driven by how that state implemented its Medicaid and CHIP programs as well as the capacity of the state’s Medicaid Management Information System to report each claim type.
This data quality assessment examines the reporting and usability of service tracking claims, supplemental payments and other, non-program claims. [2]
Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
For more information about what types of claims are in TAF and potential data quality issues, see “TAF Technical Guidance: Claims Files” on ResDAC.org.
Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
For more information about what types of claims are in TAF and potential data quality issues, see \u201cTAF Technical Guidance: Claims Files\u201d on ResDAC.org.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We examined service tracking claims, supplemental payments and other, non-program claims using claim type code (CLM_TYPE_CD) on records in the TAF inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims files. [3] Service tracking claims have claim type code values of 4 or D. Supplemental claims have claim type code values of 5 or E. Other, non-program claims have claim type code values of U, V, W, X, or Y. For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [4]
Service tracking claims
The MSIS identification number (MSIS_IDENT_NUM), service tracking type code (SRVC_TRKNG_TYPE_CD), and service tracking payment amount (SRVC_TRKNG_PYMT_AMT) are key data elements for identifying service tracking claims. Because these claims represent lump sum payments that cannot be attributed to a single Medicaid or CHIP beneficiary, the MSIS identification number should begin with an “&” or be missing. In addition, the service tracking type and payment amount needs to be populated for the claim to be useful for analytic purposes. We classified a claim as having a data quality problem if it met at least one of following conditions:
We grouped states into categories of low, medium, and high data quality concern based on the percentage of their records that were problematic according to any of the three conditions.
Table 1. Criteria for DQ assessment of service tracking claims
| Percentage of claims with a problematic value in the MSIS ID, service tracking type, or service tracking payment field | DQ assessment |
|---|---|
| 0 percent ≤ x < 10 percent | Low concern |
| 10 percent ≤ x < 20 percent | Medium concern |
| 20 percent ≤ x < 50 percent | High concern |
| x ≥ 50 percent | Unusable |
Supplemental payment records
The total Medicaid paid amount (TOT_MDCD_PD_AMT) is a key data element for supplemental payments. If there is no information in the payment field, the record cannot be used for research. These records should also have (1) a valid MSIS ID that does not begin with an “&”; (2) a missing service tracking type; and (3) a missing or zero service tracking payment amount. [5] We classified a claim as having a data quality problem if it met any one of the following conditions:
We grouped the states into low, medium, high, and unusable categories of concern based on the percentage of their supplemental payment records that were problematic according to any of the four conditions.
Table 2. Criteria for DQ assessment of supplemental payment records
| Percentage of claims with a problematic value in the total Medicaid paid amount, MSIS ID, service tracking type, or service tracking payment field | DQ assessment |
|---|---|
| 0 percent ≤ x < 10 percent | Low concern |
| 10 percent ≤ x < 20 percent | Medium concern |
| 20 percent ≤ x < 50 percent | High concern |
| x ≥ 50 percent | Unusable |
Other, non-program claims
TAF users should be aware if the data files for a state include records for other, non-program claims so that they can determine whether to include these claims in analyses. These records may represent services and payments that do not qualify for a federal match or may represent services that receive a federal match that the state wants to distinguish for some reason. Because states were not given specific guidance on what claims should be reported as other and because states are not required to justify why they have submitted a claim with the other, non-program code values, it is unclear what these claims represent in every state and whether a state’s use of this claim type changes over time. The lack of detail for these claims may make them unusable for many analytic purposes. We classified the state as high or low concern based on the percentage of all claims for the state that have an other, non-program claim type code (Table 3).
Table 3. Criteria for DQ assessment of other, non-program claims
| Percentage of claims that have an other, non-program claim type code | DQ assessment |
|---|---|
| x < 2 percent | Low concern |
| x ≥ 2 percent | High concern |
Methods previously used to assess data quality
Table 4 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 4. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND=0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
If the fields for service tracking type and service tracking amount are populated or if the MSIS ID is missing or starts with an “&”, then a TAF user cannot determine whether the record is a supplemental payment or a service tracking record.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND=0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
If the fields for service tracking type and service tracking amount are populated or if the MSIS ID is missing or starts with an \u201c&\u201d, then a TAF user cannot determine whether the record is a supplemental payment or a service tracking record.
States submit several types of service use and financial transaction records into T-MSIS. States use the claim type code data element to classify these records as those processed as Medicaid benefits, as S-CHIP benefits, or as an \""other\"" (non-Medicaid, non-SCHIP) record. This analysis identifies which states have \""other\"" records in their TAF data and examines the number of claims and dollars associated with these records.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5181"", ""relatedTopics"": [{""measureId"": 60, ""measureName"": ""Supplemental Payments"", ""groupId"": 7, ""groupName"": ""Non-Claim Records"", ""order"": 0}, {""measureId"": 85, ""measureName"": ""Service Tracking Claims"", ""groupId"": 7, ""groupName"": ""Non-Claim Records"", ""order"": 2}]}" 62,"{""measureId"": 62, ""measureName"": ""Place of Service"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Place-of-Service.pdf"", ""background"": {""content"": ""Users of the T-MSIS Analytic Files (TAF) who investigate service use in the Medicaid program will often need to determine the setting where services were delivered. The structure of the TAF can be used to identify service setting for many service use records: the inpatient (IP) file only includes records for stays at inpatient hospitals, the long-term care (LT) file only includes records for stays at institutional long-term care facilities such as nursing homes and intermediate care facilities, and the pharmacy (RX) file only includes records for prescription drugs and durable medical equipment filled by pharmacies or durable medical equipment suppliers. However, the largest TAF—the other services (OT) file—includes both institutional and professional claims from across all settings of care. OT records represent claims for outpatient facilities and for professional services provided in inpatient settings, emergency departments, outpatient settings, offices, and home- and community-based settings. [1]
On medical claims, different fields are used on different types of claims to identify the service setting. [2] The type of bill code should be used to determine the service setting on institutional claims, while the place of service code should be used to determine the service setting on professional claims. Both of these data elements are found in the TAF header record. [3] Each service use record should have only one of the fields populated, depending on what type of claim form was used to submit the claim. Sometimes, information in the header record is problematic: the type of bill and place of service codes may both be missing or invalid, or they may both be valid. If this occurs on an institutional claim, revenue center codes in the line records can be used to determine service setting in the absence of the bill type code. Professional claims should have only missing values in the revenue center field. This data quality assessment describes the extent to which TAF users can determine service setting in the OT file for each state.
Not all claims for home- and community-based services are required to include the information necessary to identify the service setting. These claims were therefore excluded from the analysis presented in this brief.
Institutional claims are submitted on an institutional claim form by hospitals, nursing facilities, intermediate care facilities for individuals with intellectual or developmental disabilities, rehabilitation facilities, home health agencies, and clinics. These claims are often referred to as UB-04 claims, when submitted in paper form, or as 837I claims, when submitted in electronic form. Professional claims are submitted on a professional claim form by physicians (both individuals and groups), other clinical professionals, freestanding laboratories and outpatient facilities, ambulances, and durable medical equipment suppliers. They are referred to as CMS-1500 claims, when submitted in paper form, or as 837P, when submitted in electronic form.
A header record summarizes the services that are captured on the claim lines, which provide details on each service covered by the claim.
Not all claims for home- and community-based services are required to include the information necessary to identify the service setting. These claims were therefore excluded from the analysis presented in this brief.
Institutional claims are submitted on an institutional claim form by hospitals, nursing facilities, intermediate care facilities for individuals with intellectual or developmental disabilities, rehabilitation facilities, home health agencies, and clinics. These claims are often referred to as UB-04 claims, when submitted in paper form, or as 837I claims, when submitted in electronic form. Professional claims are submitted on a professional claim form by physicians (both individuals and groups), other clinical professionals, freestanding laboratories and outpatient facilities, ambulances, and durable medical equipment suppliers. They are referred to as CMS-1500 claims, when submitted in paper form, or as 837P, when submitted in electronic form.
A header record summarizes the services that are captured on the claim lines, which provide details on each service covered by the claim.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
Using the TAF, [4] we examined header and line records in the OT file. Records in the IP, LT, and RX files have a known service setting based on file type and thus were not included in the analysis. We included fee-for-service (FFS) claims and managed care encounter records for Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries. [5] States with an unusably low volume of header or line records in the OT file were excluded from the analysis. For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [6] We further excluded claims that were not expected to have a valid type of bill code (BILL_TYPE_CD) or a place of service code (SRVC_PLC_CD), including transportation claims, dental claims, and claims for home- and community-based services (HCBS). [7] Claims for these services are often submitted on nonstandard forms; therefore, it is not straightforward to identify the setting of care.
Our analysis assessed whether the fields for service setting that we would expect to be populated were in fact populated. More specifically, we examined the percentage of claims with either a valid type of bill code [8] or a valid place of service code. [9] We also evaluated unexpected combinations of type of bill and place of service codes—that is, the percentage of claims with both a valid type of bill code and a valid place of service code and the percentage of claims without a valid type of bill code or a valid place of service code. On claims with these unexpected combinations, the service setting can often be identified indirectly by using the revenue center code (REV_CD), [10] when present, using the method outlined in Table 1. If all three key data elements (type of bill, place of service, and revenue center code) are missing, TAF users will be unable to determine the service setting.
Table 1. Identifying the service setting on claims indirectly
| Unexpected pattern | If revenue center code is populated… | If revenue center code is not populated… |
|---|---|---|
| Both type of bill and place of service codes are populated | Assume that the claim is institutional, and use the type of bill to determine the service setting | Assume that the claim is professional, and use the place of service to determine the service setting |
| Neither type of bill nor place of service code is populated | Assume that the claim is institutional, and use the revenue center code to identify service setting | Unable to determine service setting |
We organized states according to the level of concern about their data quality based on the percentage of claims with the expected combination of service setting fields, in which either the type of bill code or place of service code, but not both, were populated (Table 2). The level of data quality concern is low if 80 percent or more of claims were appropriately populated, medium if between 50 percent and 80 percent of claims were appropriately populated, and high if less than 50 percent of claims were appropriately populated. We did not include an “unusable” category because, when necessary, variables other than the type of bill code and place of service code can be used to identify the service setting indirectly.
Table 2. Criteria for DQ assessment of service setting
| Percentage of claims with the expected combination of service setting fields | DQ assessment |
|---|---|
| x ≥ 80 percent | Low concern |
| 50 percent ≤ x < 80 percent | Medium concern |
| x < 50 percent | High concern |
Methods previously used to assess data quality
Table 3 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods.
Table 3. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We used the claim type code (CLM_TYPE_CD) to determine which records to include and exclude. We retained FFS records (claim types 1 and A) and managed care encounters (claim types 3 and C). We excluded records with all other claim type values, including capitation payments, service tracking claims, and supplemental payments—all of which are financial transaction records and thus are not expected to include information that reflects services provided to an individual.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
Using the federally assigned service category variable (FASC), we excluded claims for transportation services; dental services; HCBS; per member per month (PMPM) payments; Disproportionate Share Hospital (DSH) claims; home health; durable medical equipment (including eyeglasses, dentures, and hearing aids); managed care capitation payments; prescription drugs; and other financial transactions. We excluded these claims because they are not typically submitted on standardized claim forms; thus, the setting of care would not be easily identifiable. Although the American Dental Association claim form has a standard field for place of service, some states process dental claims on their own forms, which do not include this field or require it to be populated. For more information about the FASC code, see the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page .
Valid type of bill codes begin with a leading zero and are four digits long. They are listed in the TAF Claims Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries . We considered three-digit values to be valid as long as they matched to a valid value when a leading zero was added. We did not consider type of bill codes that had one or two digits or that had three digits with a leading zero (that is, missing a fourth digit) as valid. We focused only on the second and third digits and allowed any value in the fourth position.
Valid place of service values are listed in Chapter 26 of CMS’s Medicare Claims Processing Manual, which is available at https://www.cms.gov/Regulations-and-Guidance/Guidance/Manuals/Downloads/clm104c26.pdf .
Valid values for revenue center codes are listed in CMS’s Claims Processing Manual for the UB-04, which is available at https://www.cms.gov/Regulations-and-Guidance/Guidance/Transmittals/downloads/r167cp.pdf .
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We used the claim type code (CLM_TYPE_CD) to determine which records to include and exclude. We retained FFS records (claim types 1 and A) and managed care encounters (claim types 3 and C). We excluded records with all other claim type values, including capitation payments, service tracking claims, and supplemental payments\u2014all of which are financial transaction records and thus are not expected to include information that reflects services provided to an individual.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
Using the federally assigned service category variable (FASC), we excluded claims for transportation services; dental services; HCBS; per member per month (PMPM) payments; Disproportionate Share Hospital (DSH) claims; home health; durable medical equipment (including eyeglasses, dentures, and hearing aids); managed care capitation payments; prescription drugs; and other financial transactions. We excluded these claims because they are not typically submitted on standardized claim forms; thus, the setting of care would not be easily identifiable. Although the American Dental Association claim form has a standard field for place of service, some states process dental claims on their own forms, which do not include this field or require it to be populated. For more information about the FASC code, see the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page .
Valid type of bill codes begin with a leading zero and are four digits long. They are listed in the TAF Claims Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries . We considered three-digit values to be valid as long as they matched to a valid value when a leading zero was added. We did not consider type of bill codes that had one or two digits or that had three digits with a leading zero (that is, missing a fourth digit) as valid. We focused only on the second and third digits and allowed any value in the fourth position.
Valid place of service values are listed in Chapter 26 of CMS\u2019s Medicare Claims Processing Manual, which is available at https://www.cms.gov/Regulations-and-Guidance/Guidance/Manuals/Downloads/clm104c26.pdf .
Valid values for revenue center codes are listed in CMS\u2019s Claims Processing Manual for the UB-04, which is available at https://www.cms.gov/Regulations-and-Guidance/Guidance/Transmittals/downloads/r167cp.pdf .
Different fields are used on different types of claim forms to identify the service setting in which care was delivered. These fields include type of bill, place of service, and revenue center. This analysis examines the extent to which TAF users can identify the place of service in the OT file using these three data elements.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5191"", ""relatedTopics"": []}" 63,"{""measureId"": 63, ""measureName"": ""Procedure Codes - OT Professional"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Proc-Cd-OT-Prof.pdf"", ""background"": {""content"": ""Procedure codes are used to document services rendered and to bill for medical procedures provided to a patient. They represent the most detailed and specific information available in administrative claims data about the services delivered to patients. As such, they are critical to research on service utilization and access to care. Procedure codes are required on most—although not all—medical claims, including all professional claims and some institutional claims submitted by hospitals and other facilities. [1] Procedure codes are required on institutional claims only if a direct service, such as a surgical procedure, was provided during the visit or stay at the facility. Many institutional claims document only other charges, such as room and board or the use of equipment or supplies, for which a procedure code would appropriately be absent. States are required to use national procedure codes on claims for Medicaid services delivered to beneficiaries, but some Medicaid programs allow the use of state-specific procedure codes for certain services. [2]
In the T-MSIS Analytic Files (TAF), procedure codes should be in different fields depending on the claims file and type of claim (Table 1). The TAF Other Services (OT) file includes professional and outpatient institutional claims. For professional and institutional claims, procedure codes should be in the procedure code field at the line level. [3] The inpatient (IP) file, consisting only of institutional claims, can include up to six procedure codes per claim header in the procedure code fields.
Table 1. Expected reporting of procedure codes, by file type and type of claim
| File | Type of claim form | Field in which procedure codes should be reported | Expected on all claims? |
|---|---|---|---|
| OT line | Professional | PRCDR_CD | Yes |
| OT line | Institutional | PRCDR_CD | No |
| IP header | Institutional | PRCDR_1_CD – PRCDR_6_CD | No |
This data quality assessment examines the extent to which appropriate fields in the TAF OT and IP files were populated with valid procedure code values.
All medical claims are submitted on either an institutional or a professional claim form, with slightly different information on each form. Institutional claims are submitted by facilities such as hospitals, nursing facilities, intermediate care facilities for individuals with intellectual or development disabilities, rehabilitation facilities, home health agencies, and clinics. These claims are often referred to as “UB-04 claims” when submitted in paper form or as “837I claims” when submitted in electronic form. Professional claims are submitted by physicians (both individual and group practices); other clinical professionals; free-standing laboratories and outpatient facilities; ambulances; and durable medical equipment suppliers. These claims are referred to as “CMS-1500 claims” when submitted in paper form or “837P” when submitted in electronic form.
State Medicaid programs may allow state-specific procedure codes for several reasons. For example, a state may use state-specific codes when its Medicaid program covers a service for which there is no national procedure code. State Medicaid programs often use state-specific codes for home and community-based services or for behavioral health services.
A header record summarizes the services provided that are captured on the claim lines, which provide details on each service covered by the claim.
All medical claims are submitted on either an institutional or a professional claim form, with slightly different information on each form. Institutional claims are submitted by facilities such as hospitals, nursing facilities, intermediate care facilities for individuals with intellectual or development disabilities, rehabilitation facilities, home health agencies, and clinics. These claims are often referred to as \u201cUB-04 claims\u201d when submitted in paper form or as \u201c837I claims\u201d when submitted in electronic form. Professional claims are submitted by physicians (both individual and group practices); other clinical professionals; free-standing laboratories and outpatient facilities; ambulances; and durable medical equipment suppliers. These claims are referred to as \u201cCMS-1500 claims\u201d when submitted in paper form or \u201c837P\u201d when submitted in electronic form.
State Medicaid programs may allow state-specific procedure codes for several reasons. For example, a state may use state-specific codes when its Medicaid program covers a service for which there is no national procedure code. State Medicaid programs often use state-specific codes for home and community-based services or for behavioral health services.
A header record summarizes the services provided that are captured on the claim lines, which provide details on each service covered by the claim.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
Using the TAF, [4] we examined records in the other services (OT) and inpatient (IP) files. We did not examine records in the long-term care or the pharmacy files because these files do not capture procedure codes. The analysis included fee-for-service (FFS) claims and managed care encounter records for Medicaid beneficiaries. [5] States were excluded from the analyses of procedure codes in the IP file if the volume of header records in the IP file was unusably low. States were excluded from the analyses of procedure codes in the OT file if the volume of line records in the OT file was unusably low. For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [6]
For our analysis of the OT file, we first used an algorithm to classify each claim as either professional or institutional. The standardized fields in each claim form, and hence the information available for each type of claim, differ slightly. The algorithm relies on three fields: (1) place of service, which should only be populated on professional claims; (2) type of bill, which should only be populated on institutional claims; and (3) revenue code, which should only be populated on institutional claims. [7]
Professional claims (OT file)
Professional claims in the OT file should always have a non-missing procedure code reported in the procedure code field on all claim lines. To understand whether any states had problems with incomplete procedure code data, we examined the percentage of claim lines that had a missing value in the procedure code field. Next, we examined the percentage of claim lines that had a non-missing but invalid procedure code (that is, a value that did not match to either a national or state-specific code). [8] We classified states into categories of concern about data quality based on the percentage of all professional line records that had a missing or invalid procedure code using the thresholds shown in Table 2:
Table 2. Criteria for DQ assessment of procedure code on professional claims in the OT file
For informational purposes, we also examined the percentage of claim lines with a valid procedure code that took a national code value versus a state-specific code value. Although the presence of state-specific procedure codes does not indicate a data quality concern, TAF users would need to account for these codes in many analyses.
Outpatient institutional claims (OT file)
Institutional claims in the OT file are expected to have a procedure code only in certain circumstances. We recommend that TAF users employ a state-specific approach to using the procedure code field for analyses that include outpatient institutional claims.
To understand the overall rate of outpatient claims with no usable procedure code information, we examined the percentage of claim lines that had missing or invalid procedure codes in the procedure code field. [9] To understand whether states were reporting procedure code information into the correct field, we then compared the percentage of claim lines with (1) missing values and (2) invalid values in the procedure code field.
We then classified states into categories based on the percentage of all institutional line records that had missing or invalid procedure code information in the procedure code field. We considered a state to have unusable data if greater than 90 percent of claims lines had missing or invalid procedure code information (Table 3).
Table 3. Criteria for DQ assessment of procedure code on institutional claims in the OT file
| Percentage of institutional claim lines with missing or invalid procedure code values in the procedure code field | DQ assessment |
|---|---|
| x ≤ 90 percent | Low concern |
| x > 90 percent | Unusable |
Inpatient institutional claims (IP file)
Records in the IP file are only expected to include a procedure code for inpatient stays that involve surgery or other procedures. To understand the completeness of procedure code information in the IP file, we first examined the percentage of claim headers with a missing primary procedure code (PRCDR_1_CD). Next, we examined the percentage of claim headers with any invalid procedure codes (PRCDR_1_CD through PRCDR_6_CD), as these codes are likely unusable for research purposes. In some cases, TAF users may be able to use certain kinds of invalid codes in their analyses. For example, some states may be populating the procedure code with the national ICD-9 procedure codes, [10] which were retired as of October 2015. Some states may also be populating the procedure code with truncated ICD-10 codes, such as the first six digits of a valid seven-digit ICD-10 code. [11] In these cases, TAF users may be able to append the non-specific seventh-digit “Z” to these codes to make them usable for grouper software or for analyses that require valid procedure codes. [12]
We classified states into categories based on two criteria. First, we examined the percentage of headers missing a primary procedure code. We considered a state to have a high data quality concern if no claims headers had a missing procedure code, as this is an unexpected pattern that likely indicates a data quality issue. We considered a state to have unusable data if greater than 90 percent of header records had a missing procedure code, as this indicates the procedure code data are incomplete. We classified all other states based on the percentage of headers that had an invalid procedure code using thresholds shown in Table 4.
Table 4. Criteria for DQ assessment of procedure code in the IP file
| Percentage of IP claim headers with missing procedure code | Percentage of IP claim headers with invalid procedure code | DQ assessment |
|---|---|---|
| 0 percent < x ≤ 90 percent | 0 percent ≤ x ≤ 10 percent | Low concern a |
| 0 percent < x ≤ 90 percent | 10 percent < x ≤ 20 percent | Medium concern a |
| 0 percent < x ≤ 90 percent | 20 percent < x ≤ 30 percent | High concern a |
| 0 percent | Any value | High concern a |
| x > 90 percent | x > 30 percent | Unusable b |
a Both criteria must be true for a state to receive the given DQ Assessment.
b One of the two criteria must be true for a state to receive the given DQ Assessment.
To identify the types of invalid codes states may be using, we also examined the percentage of IP header records with valid ICD-9 procedure codes or with 6-digit procedure codes indicating that the state may be reporting truncated ICD-10 codes. This information is included in the tables for the topic, but was not used as part of the data quality assessment for inpatient claims.
Methods previously used to assess data quality
Table 5 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 5. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
|
|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We used claim type code (CLM_TYPE_CD) to determine which records to include and which to exclude. We included FFS records (claim type 1 or A) and managed care encounters (3 and C). We excluded records with all other claim type values, including capitation payments, service tracking claims, and supplemental payments, none of which we expected to include procedure codes reflecting services provided.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. We restricted our analysis to the original claim from each claim family by using the adjustment indicator (ADJSTMT_IND=0) in order to ensure that utilization counts are correct. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org
To be classified as a professional claim, the record had to meet one of the following criteria: (1) had a valid place of service code (SRVC_PLC_CD) but a missing or invalid revenue code (REV_CD) on all lines associated with the claim; or (2) had a missing or invalid type of bill code, place of service code, and revenue codes but a valid procedure code (PRCDR_CD) on all lines associated with the claim. To be classified as an institutional claim, the record had to meet one of the following criteria: (1) had a valid type of bill code and a missing or invalid place of service code; or (2) had at least one valid revenue code.
For a full list of valid national procedure codes, see https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/PhysicianFeeSched/PFS-Relative-Value-Files.html . If a state uses its own procedure codes, it is required to submit documentation of the valid procedure code value and a description of the code.
For data years prior to 2022, we considered procedure codes reported in the HCPCS rate field in addition to the procedure code field to evaluate outpatient institutional claims. As of 2021, states were instructed to stop using the HCPCS rate field and only use the procedure code field to report procedure code for both professional and institutional claims in the OT file. The HCPCS rate field is no longer available in T-MSIS as of June 2022. For more information on CMS guidance issued to states, please refer to https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/51965
ICD-9 stands for International Statistical Classification of Diseases Clinical Modification, 9th edition.
ICD-10 stands for International Statistical Classification of Diseases Clinical Modification, 10th edition.
A “Z” at the end of an ICD-10-PCS code indicates that the specific qualifier does not apply to the procedure.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We used claim type code (CLM_TYPE_CD) to determine which records to include and which to exclude. We included FFS records (claim type 1 or A) and managed care encounters (3 and C). We excluded records with all other claim type values, including capitation payments, service tracking claims, and supplemental payments, none of which we expected to include procedure codes reflecting services provided.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. We restricted our analysis to the original claim from each claim family by using the adjustment indicator (ADJSTMT_IND=0) in order to ensure that utilization counts are correct. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org
To be classified as a professional claim, the record had to meet one of the following criteria: (1) had a valid place of service code (SRVC_PLC_CD) but a missing or invalid revenue code (REV_CD) on all lines associated with the claim; or (2) had a missing or invalid type of bill code, place of service code, and revenue codes but a valid procedure code (PRCDR_CD) on all lines associated with the claim. To be classified as an institutional claim, the record had to meet one of the following criteria: (1) had a valid type of bill code and a missing or invalid place of service code; or (2) had at least one valid revenue code.
For a full list of valid national procedure codes, see https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/PhysicianFeeSched/PFS-Relative-Value-Files.html . If a state uses its own procedure codes, it is required to submit documentation of the valid procedure code value and a description of the code.
For data years prior to 2022, we considered procedure codes reported in the HCPCS rate field in addition to the procedure code field to evaluate outpatient institutional claims. As of 2021, states were instructed to stop using the HCPCS rate field and only use the procedure code field to report procedure code for both professional and institutional claims in the OT file. The HCPCS rate field is no longer available in T-MSIS as of June 2022. For more information on CMS guidance issued to states, please refer to https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/51965
ICD-9 stands for International Statistical Classification of Diseases Clinical Modification, 9th edition.
ICD-10 stands for International Statistical Classification of Diseases Clinical Modification, 10th edition.
A \u201cZ\u201d at the end of an ICD-10-PCS code indicates that the specific qualifier does not apply to the procedure.
Providers use procedure codes to document services rendered during a medical encounter and to bill for these services. This analysis examines how often the procedure code is missing on professional claims in the OT file and how often the non-missing values on these claims represent valid national or state-specific codes.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5201"", ""relatedTopics"": [{""measureId"": 65, ""measureName"": ""Procedure Codes - IP"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 0}, {""measureId"": 64, ""measureName"": ""Procedure Codes - OT Institutional"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 2}]}" 64,"{""measureId"": 64, ""measureName"": ""Procedure Codes - OT Institutional"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Proc-Cd-OT-Institution.pdf"", ""background"": {""content"": ""Procedure codes are used to document services rendered and to bill for medical procedures provided to a patient. They represent the most detailed and specific information available in administrative claims data about the services delivered to patients. As such, they are critical to research on service utilization and access to care. Procedure codes are required on most—although not all—medical claims, including all professional claims and some institutional claims submitted by hospitals and other facilities. [1] Procedure codes are required on institutional claims only if a direct service, such as a surgical procedure, was provided during the visit or stay at the facility. Many institutional claims document only other charges, such as room and board or the use of equipment or supplies, for which a procedure code would appropriately be absent. States are required to use national procedure codes on claims for Medicaid services delivered to beneficiaries, but some Medicaid programs allow the use of state-specific procedure codes for certain services. [2]
In the T-MSIS Analytic Files (TAF), procedure codes should be in different fields depending on the claims file and type of claim (Table 1). The TAF Other Services (OT) file includes professional and outpatient institutional claims. For professional and institutional claims, procedure codes should be in the procedure code field at the line level. [3] The inpatient (IP) file, consisting only of institutional claims, can include up to six procedure codes per claim header in the procedure code fields.
Table 1. Expected reporting of procedure codes, by file type and type of claim
| File | Type of claim form | Field in which procedure codes should be reported | Expected on all claims? |
|---|---|---|---|
| OT line | Professional | PRCDR_CD | Yes |
| OT line | Institutional | PRCDR_CD | No |
| IP header | Institutional | PRCDR_1_CD – PRCDR_6_CD | No |
This data quality assessment examines the extent to which appropriate fields in the TAF OT and IP files were populated with valid procedure code values.
All medical claims are submitted on either an institutional or a professional claim form, with slightly different information on each form. Institutional claims are submitted by facilities such as hospitals, nursing facilities, intermediate care facilities for individuals with intellectual or development disabilities, rehabilitation facilities, home health agencies, and clinics. These claims are often referred to as “UB-04 claims” when submitted in paper form or as “837I claims” when submitted in electronic form. Professional claims are submitted by physicians (both individual and group practices); other clinical professionals; free-standing laboratories and outpatient facilities; ambulances; and durable medical equipment suppliers. These claims are referred to as “CMS-1500 claims” when submitted in paper form or “837P” when submitted in electronic form.
State Medicaid programs may allow state-specific procedure codes for several reasons. For example, a state may use state-specific codes when its Medicaid program covers a service for which there is no national procedure code. State Medicaid programs often use state-specific codes for home and community-based services or for behavioral health services.
A header record summarizes the services provided that are captured on the claim lines, which provide details on each service covered by the claim.
All medical claims are submitted on either an institutional or a professional claim form, with slightly different information on each form. Institutional claims are submitted by facilities such as hospitals, nursing facilities, intermediate care facilities for individuals with intellectual or development disabilities, rehabilitation facilities, home health agencies, and clinics. These claims are often referred to as \u201cUB-04 claims\u201d when submitted in paper form or as \u201c837I claims\u201d when submitted in electronic form. Professional claims are submitted by physicians (both individual and group practices); other clinical professionals; free-standing laboratories and outpatient facilities; ambulances; and durable medical equipment suppliers. These claims are referred to as \u201cCMS-1500 claims\u201d when submitted in paper form or \u201c837P\u201d when submitted in electronic form.
State Medicaid programs may allow state-specific procedure codes for several reasons. For example, a state may use state-specific codes when its Medicaid program covers a service for which there is no national procedure code. State Medicaid programs often use state-specific codes for home and community-based services or for behavioral health services.
A header record summarizes the services provided that are captured on the claim lines, which provide details on each service covered by the claim.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
Using the TAF, [4] we examined records in the other services (OT) and inpatient (IP) files. We did not examine records in the long-term care or the pharmacy files because these files do not capture procedure codes. The analysis included fee-for-service (FFS) claims and managed care encounter records for Medicaid beneficiaries. [5] States were excluded from the analyses of procedure codes in the IP file if the volume of header records in the IP file was unusably low. States were excluded from the analyses of procedure codes in the OT file if the volume of line records in the OT file was unusably low. For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [6]
For our analysis of the OT file, we first used an algorithm to classify each claim as either professional or institutional. The standardized fields in each claim form, and hence the information available for each type of claim, differ slightly. The algorithm relies on three fields: (1) place of service, which should only be populated on professional claims; (2) type of bill, which should only be populated on institutional claims; and (3) revenue code, which should only be populated on institutional claims. [7]
Professional claims (OT file)
Professional claims in the OT file should always have a non-missing procedure code reported in the procedure code field on all claim lines. To understand whether any states had problems with incomplete procedure code data, we examined the percentage of claim lines that had a missing value in the procedure code field. Next, we examined the percentage of claim lines that had a non-missing but invalid procedure code (that is, a value that did not match to either a national or state-specific code). [8] We classified states into categories of concern about data quality based on the percentage of all professional line records that had a missing or invalid procedure code using the thresholds shown in Table 2:
Table 2. Criteria for DQ assessment of procedure code on professional claims in the OT file
For informational purposes, we also examined the percentage of claim lines with a valid procedure code that took a national code value versus a state-specific code value. Although the presence of state-specific procedure codes does not indicate a data quality concern, TAF users would need to account for these codes in many analyses.
Outpatient institutional claims (OT file)
Institutional claims in the OT file are expected to have a procedure code only in certain circumstances. We recommend that TAF users employ a state-specific approach to using the procedure code field for analyses that include outpatient institutional claims.
To understand the overall rate of outpatient claims with no usable procedure code information, we examined the percentage of claim lines that had missing or invalid procedure codes in the procedure code field. [9] To understand whether states were reporting procedure code information into the correct field, we then compared the percentage of claim lines with (1) missing values and (2) invalid values in the procedure code field.
We then classified states into categories based on the percentage of all institutional line records that had missing or invalid procedure code information in the procedure code field. We considered a state to have unusable data if greater than 90 percent of claims lines had missing or invalid procedure code information (Table 3).
Table 3. Criteria for DQ assessment of procedure code on institutional claims in the OT file
| Percentage of institutional claim lines with missing or invalid procedure code values in the procedure code field | DQ assessment |
|---|---|
| x ≤ 90 percent | Low concern |
| x > 90 percent | Unusable |
Inpatient institutional claims (IP file)
Records in the IP file are only expected to include a procedure code for inpatient stays that involve surgery or other procedures. To understand the completeness of procedure code information in the IP file, we first examined the percentage of claim headers with a missing primary procedure code (PRCDR_1_CD). Next, we examined the percentage of claim headers with any invalid procedure codes (PRCDR_1_CD through PRCDR_6_CD), as these codes are likely unusable for research purposes. In some cases, TAF users may be able to use certain kinds of invalid codes in their analyses. For example, some states may be populating the procedure code with the national ICD-9 procedure codes, [10] which were retired as of October 2015. Some states may also be populating the procedure code with truncated ICD-10 codes, such as the first six digits of a valid seven-digit ICD-10 code. [11] In these cases, TAF users may be able to append the non-specific seventh-digit “Z” to these codes to make them usable for grouper software or for analyses that require valid procedure codes. [12]
We classified states into categories based on two criteria. First, we examined the percentage of headers missing a primary procedure code. We considered a state to have a high data quality concern if no claims headers had a missing procedure code, as this is an unexpected pattern that likely indicates a data quality issue. We considered a state to have unusable data if greater than 90 percent of header records had a missing procedure code, as this indicates the procedure code data are incomplete. We classified all other states based on the percentage of headers that had an invalid procedure code using thresholds shown in Table 4.
Table 4. Criteria for DQ assessment of procedure code in the IP file
| Percentage of IP claim headers with missing procedure code | Percentage of IP claim headers with invalid procedure code | DQ assessment |
|---|---|---|
| 0 percent < x ≤ 90 percent | 0 percent ≤ x ≤ 10 percent | Low concern a |
| 0 percent < x ≤ 90 percent | 10 percent < x ≤ 20 percent | Medium concern a |
| 0 percent < x ≤ 90 percent | 20 percent < x ≤ 30 percent | High concern a |
| 0 percent | Any value | High concern a |
| x > 90 percent | x > 30 percent | Unusable b |
a Both criteria must be true for a state to receive the given DQ Assessment.
b One of the two criteria must be true for a state to receive the given DQ Assessment.
To identify the types of invalid codes states may be using, we also examined the percentage of IP header records with valid ICD-9 procedure codes or with 6-digit procedure codes indicating that the state may be reporting truncated ICD-10 codes. This information is included in the tables for the topic, but was not used as part of the data quality assessment for inpatient claims.
Methods previously used to assess data quality
Table 5 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 5. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
|
|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We used claim type code (CLM_TYPE_CD) to determine which records to include and which to exclude. We included FFS records (claim type 1 or A) and managed care encounters (3 and C). We excluded records with all other claim type values, including capitation payments, service tracking claims, and supplemental payments, none of which we expected to include procedure codes reflecting services provided.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. We restricted our analysis to the original claim from each claim family by using the adjustment indicator (ADJSTMT_IND=0) in order to ensure that utilization counts are correct. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org
To be classified as a professional claim, the record had to meet one of the following criteria: (1) had a valid place of service code (SRVC_PLC_CD) but a missing or invalid revenue code (REV_CD) on all lines associated with the claim; or (2) had a missing or invalid type of bill code, place of service code, and revenue codes but a valid procedure code (PRCDR_CD) on all lines associated with the claim. To be classified as an institutional claim, the record had to meet one of the following criteria: (1) had a valid type of bill code and a missing or invalid place of service code; or (2) had at least one valid revenue code.
For a full list of valid national procedure codes, see https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/PhysicianFeeSched/PFS-Relative-Value-Files.html . If a state uses its own procedure codes, it is required to submit documentation of the valid procedure code value and a description of the code.
For data years prior to 2022, we considered procedure codes reported in the HCPCS rate field in addition to the procedure code field to evaluate outpatient institutional claims. As of 2021, states were instructed to stop using the HCPCS rate field and only use the procedure code field to report procedure code for both professional and institutional claims in the OT file. The HCPCS rate field is no longer available in T-MSIS as of June 2022. For more information on CMS guidance issued to states, please refer to https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/51965
ICD-9 stands for International Statistical Classification of Diseases Clinical Modification, 9th edition.
ICD-10 stands for International Statistical Classification of Diseases Clinical Modification, 10th edition.
A “Z” at the end of an ICD-10-PCS code indicates that the specific qualifier does not apply to the procedure.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We used claim type code (CLM_TYPE_CD) to determine which records to include and which to exclude. We included FFS records (claim type 1 or A) and managed care encounters (3 and C). We excluded records with all other claim type values, including capitation payments, service tracking claims, and supplemental payments, none of which we expected to include procedure codes reflecting services provided.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. We restricted our analysis to the original claim from each claim family by using the adjustment indicator (ADJSTMT_IND=0) in order to ensure that utilization counts are correct. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org
To be classified as a professional claim, the record had to meet one of the following criteria: (1) had a valid place of service code (SRVC_PLC_CD) but a missing or invalid revenue code (REV_CD) on all lines associated with the claim; or (2) had a missing or invalid type of bill code, place of service code, and revenue codes but a valid procedure code (PRCDR_CD) on all lines associated with the claim. To be classified as an institutional claim, the record had to meet one of the following criteria: (1) had a valid type of bill code and a missing or invalid place of service code; or (2) had at least one valid revenue code.
For a full list of valid national procedure codes, see https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/PhysicianFeeSched/PFS-Relative-Value-Files.html . If a state uses its own procedure codes, it is required to submit documentation of the valid procedure code value and a description of the code.
For data years prior to 2022, we considered procedure codes reported in the HCPCS rate field in addition to the procedure code field to evaluate outpatient institutional claims. As of 2021, states were instructed to stop using the HCPCS rate field and only use the procedure code field to report procedure code for both professional and institutional claims in the OT file. The HCPCS rate field is no longer available in T-MSIS as of June 2022. For more information on CMS guidance issued to states, please refer to https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/51965
ICD-9 stands for International Statistical Classification of Diseases Clinical Modification, 9th edition.
ICD-10 stands for International Statistical Classification of Diseases Clinical Modification, 10th edition.
A \u201cZ\u201d at the end of an ICD-10-PCS code indicates that the specific qualifier does not apply to the procedure.
Providers use procedure codes to document services rendered during a medical encounter and to bill for these services. This analysis examines how often the procedure code is missing on institutional claims in the OT file and how often non-missing values on these claims represent valid national or state-specific codes.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5201"", ""relatedTopics"": [{""measureId"": 65, ""measureName"": ""Procedure Codes - IP"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 0}, {""measureId"": 63, ""measureName"": ""Procedure Codes - OT Professional"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 1}]}" 65,"{""measureId"": 65, ""measureName"": ""Procedure Codes - IP"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Proc-Cd-IP.pdf"", ""background"": {""content"": ""Procedure codes are used to document services rendered and to bill for medical procedures provided to a patient. They represent the most detailed and specific information available in administrative claims data about the services delivered to patients. As such, they are critical to research on service utilization and access to care. Procedure codes are required on most—although not all—medical claims, including all professional claims and some institutional claims submitted by hospitals and other facilities. [1] Procedure codes are required on institutional claims only if a direct service, such as a surgical procedure, was provided during the visit or stay at the facility. Many institutional claims document only other charges, such as room and board or the use of equipment or supplies, for which a procedure code would appropriately be absent. States are required to use national procedure codes on claims for Medicaid services delivered to beneficiaries, but some Medicaid programs allow the use of state-specific procedure codes for certain services. [2]
In the T-MSIS Analytic Files (TAF), procedure codes should be in different fields depending on the claims file and type of claim (Table 1). The TAF Other Services (OT) file includes professional and outpatient institutional claims. For professional and institutional claims, procedure codes should be in the procedure code field at the line level. [3] The inpatient (IP) file, consisting only of institutional claims, can include up to six procedure codes per claim header in the procedure code fields.
Table 1. Expected reporting of procedure codes, by file type and type of claim
| File | Type of claim form | Field in which procedure codes should be reported | Expected on all claims? |
|---|---|---|---|
| OT line | Professional | PRCDR_CD | Yes |
| OT line | Institutional | PRCDR_CD | No |
| IP header | Institutional | PRCDR_1_CD – PRCDR_6_CD | No |
This data quality assessment examines the extent to which appropriate fields in the TAF OT and IP files were populated with valid procedure code values.
All medical claims are submitted on either an institutional or a professional claim form, with slightly different information on each form. Institutional claims are submitted by facilities such as hospitals, nursing facilities, intermediate care facilities for individuals with intellectual or development disabilities, rehabilitation facilities, home health agencies, and clinics. These claims are often referred to as “UB-04 claims” when submitted in paper form or as “837I claims” when submitted in electronic form. Professional claims are submitted by physicians (both individual and group practices); other clinical professionals; free-standing laboratories and outpatient facilities; ambulances; and durable medical equipment suppliers. These claims are referred to as “CMS-1500 claims” when submitted in paper form or “837P” when submitted in electronic form.
State Medicaid programs may allow state-specific procedure codes for several reasons. For example, a state may use state-specific codes when its Medicaid program covers a service for which there is no national procedure code. State Medicaid programs often use state-specific codes for home and community-based services or for behavioral health services.
A header record summarizes the services provided that are captured on the claim lines, which provide details on each service covered by the claim.
All medical claims are submitted on either an institutional or a professional claim form, with slightly different information on each form. Institutional claims are submitted by facilities such as hospitals, nursing facilities, intermediate care facilities for individuals with intellectual or development disabilities, rehabilitation facilities, home health agencies, and clinics. These claims are often referred to as \u201cUB-04 claims\u201d when submitted in paper form or as \u201c837I claims\u201d when submitted in electronic form. Professional claims are submitted by physicians (both individual and group practices); other clinical professionals; free-standing laboratories and outpatient facilities; ambulances; and durable medical equipment suppliers. These claims are referred to as \u201cCMS-1500 claims\u201d when submitted in paper form or \u201c837P\u201d when submitted in electronic form.
State Medicaid programs may allow state-specific procedure codes for several reasons. For example, a state may use state-specific codes when its Medicaid program covers a service for which there is no national procedure code. State Medicaid programs often use state-specific codes for home and community-based services or for behavioral health services.
A header record summarizes the services provided that are captured on the claim lines, which provide details on each service covered by the claim.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
Using the TAF, [4] we examined records in the other services (OT) and inpatient (IP) files. We did not examine records in the long-term care or the pharmacy files because these files do not capture procedure codes. The analysis included fee-for-service (FFS) claims and managed care encounter records for Medicaid beneficiaries. [5] States were excluded from the analyses of procedure codes in the IP file if the volume of header records in the IP file was unusably low. States were excluded from the analyses of procedure codes in the OT file if the volume of line records in the OT file was unusably low. For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [6]
For our analysis of the OT file, we first used an algorithm to classify each claim as either professional or institutional. The standardized fields in each claim form, and hence the information available for each type of claim, differ slightly. The algorithm relies on three fields: (1) place of service, which should only be populated on professional claims; (2) type of bill, which should only be populated on institutional claims; and (3) revenue code, which should only be populated on institutional claims. [7]
Professional claims (OT file)
Professional claims in the OT file should always have a non-missing procedure code reported in the procedure code field on all claim lines. To understand whether any states had problems with incomplete procedure code data, we examined the percentage of claim lines that had a missing value in the procedure code field. Next, we examined the percentage of claim lines that had a non-missing but invalid procedure code (that is, a value that did not match to either a national or state-specific code). [8] We classified states into categories of concern about data quality based on the percentage of all professional line records that had a missing or invalid procedure code using the thresholds shown in Table 2:
Table 2. Criteria for DQ assessment of procedure code on professional claims in the OT file
For informational purposes, we also examined the percentage of claim lines with a valid procedure code that took a national code value versus a state-specific code value. Although the presence of state-specific procedure codes does not indicate a data quality concern, TAF users would need to account for these codes in many analyses.
Outpatient institutional claims (OT file)
Institutional claims in the OT file are expected to have a procedure code only in certain circumstances. We recommend that TAF users employ a state-specific approach to using the procedure code field for analyses that include outpatient institutional claims.
To understand the overall rate of outpatient claims with no usable procedure code information, we examined the percentage of claim lines that had missing or invalid procedure codes in the procedure code field. [9] To understand whether states were reporting procedure code information into the correct field, we then compared the percentage of claim lines with (1) missing values and (2) invalid values in the procedure code field.
We then classified states into categories based on the percentage of all institutional line records that had missing or invalid procedure code information in the procedure code field. We considered a state to have unusable data if greater than 90 percent of claims lines had missing or invalid procedure code information (Table 3).
Table 3. Criteria for DQ assessment of procedure code on institutional claims in the OT file
| Percentage of institutional claim lines with missing or invalid procedure code values in the procedure code field | DQ assessment |
|---|---|
| x ≤ 90 percent | Low concern |
| x > 90 percent | Unusable |
Inpatient institutional claims (IP file)
Records in the IP file are only expected to include a procedure code for inpatient stays that involve surgery or other procedures. To understand the completeness of procedure code information in the IP file, we first examined the percentage of claim headers with a missing primary procedure code (PRCDR_1_CD). Next, we examined the percentage of claim headers with any invalid procedure codes (PRCDR_1_CD through PRCDR_6_CD), as these codes are likely unusable for research purposes. In some cases, TAF users may be able to use certain kinds of invalid codes in their analyses. For example, some states may be populating the procedure code with the national ICD-9 procedure codes, [10] which were retired as of October 2015. Some states may also be populating the procedure code with truncated ICD-10 codes, such as the first six digits of a valid seven-digit ICD-10 code. [11] In these cases, TAF users may be able to append the non-specific seventh-digit “Z” to these codes to make them usable for grouper software or for analyses that require valid procedure codes. [12]
We classified states into categories based on two criteria. First, we examined the percentage of headers missing a primary procedure code. We considered a state to have a high data quality concern if no claims headers had a missing procedure code, as this is an unexpected pattern that likely indicates a data quality issue. We considered a state to have unusable data if greater than 90 percent of header records had a missing procedure code, as this indicates the procedure code data are incomplete. We classified all other states based on the percentage of headers that had an invalid procedure code using thresholds shown in Table 4.
Table 4. Criteria for DQ assessment of procedure code in the IP file
| Percentage of IP claim headers with missing procedure code | Percentage of IP claim headers with invalid procedure code | DQ assessment |
|---|---|---|
| 0 percent < x ≤ 90 percent | 0 percent ≤ x ≤ 10 percent | Low concern a |
| 0 percent < x ≤ 90 percent | 10 percent < x ≤ 20 percent | Medium concern a |
| 0 percent < x ≤ 90 percent | 20 percent < x ≤ 30 percent | High concern a |
| 0 percent | Any value | High concern a |
| x > 90 percent | x > 30 percent | Unusable b |
a Both criteria must be true for a state to receive the given DQ Assessment.
b One of the two criteria must be true for a state to receive the given DQ Assessment.
To identify the types of invalid codes states may be using, we also examined the percentage of IP header records with valid ICD-9 procedure codes or with 6-digit procedure codes indicating that the state may be reporting truncated ICD-10 codes. This information is included in the tables for the topic, but was not used as part of the data quality assessment for inpatient claims.
Methods previously used to assess data quality
Table 5 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 5. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
|
|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We used claim type code (CLM_TYPE_CD) to determine which records to include and which to exclude. We included FFS records (claim type 1 or A) and managed care encounters (3 and C). We excluded records with all other claim type values, including capitation payments, service tracking claims, and supplemental payments, none of which we expected to include procedure codes reflecting services provided.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. We restricted our analysis to the original claim from each claim family by using the adjustment indicator (ADJSTMT_IND=0) in order to ensure that utilization counts are correct. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org
To be classified as a professional claim, the record had to meet one of the following criteria: (1) had a valid place of service code (SRVC_PLC_CD) but a missing or invalid revenue code (REV_CD) on all lines associated with the claim; or (2) had a missing or invalid type of bill code, place of service code, and revenue codes but a valid procedure code (PRCDR_CD) on all lines associated with the claim. To be classified as an institutional claim, the record had to meet one of the following criteria: (1) had a valid type of bill code and a missing or invalid place of service code; or (2) had at least one valid revenue code.
For a full list of valid national procedure codes, see https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/PhysicianFeeSched/PFS-Relative-Value-Files.html . If a state uses its own procedure codes, it is required to submit documentation of the valid procedure code value and a description of the code.
For data years prior to 2022, we considered procedure codes reported in the HCPCS rate field in addition to the procedure code field to evaluate outpatient institutional claims. As of 2021, states were instructed to stop using the HCPCS rate field and only use the procedure code field to report procedure code for both professional and institutional claims in the OT file. The HCPCS rate field is no longer available in T-MSIS as of June 2022. For more information on CMS guidance issued to states, please refer to https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/51965
ICD-9 stands for International Statistical Classification of Diseases Clinical Modification, 9th edition.
ICD-10 stands for International Statistical Classification of Diseases Clinical Modification, 10th edition.
A “Z” at the end of an ICD-10-PCS code indicates that the specific qualifier does not apply to the procedure.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We used claim type code (CLM_TYPE_CD) to determine which records to include and which to exclude. We included FFS records (claim type 1 or A) and managed care encounters (3 and C). We excluded records with all other claim type values, including capitation payments, service tracking claims, and supplemental payments, none of which we expected to include procedure codes reflecting services provided.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. We restricted our analysis to the original claim from each claim family by using the adjustment indicator (ADJSTMT_IND=0) in order to ensure that utilization counts are correct. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org
To be classified as a professional claim, the record had to meet one of the following criteria: (1) had a valid place of service code (SRVC_PLC_CD) but a missing or invalid revenue code (REV_CD) on all lines associated with the claim; or (2) had a missing or invalid type of bill code, place of service code, and revenue codes but a valid procedure code (PRCDR_CD) on all lines associated with the claim. To be classified as an institutional claim, the record had to meet one of the following criteria: (1) had a valid type of bill code and a missing or invalid place of service code; or (2) had at least one valid revenue code.
For a full list of valid national procedure codes, see https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/PhysicianFeeSched/PFS-Relative-Value-Files.html . If a state uses its own procedure codes, it is required to submit documentation of the valid procedure code value and a description of the code.
For data years prior to 2022, we considered procedure codes reported in the HCPCS rate field in addition to the procedure code field to evaluate outpatient institutional claims. As of 2021, states were instructed to stop using the HCPCS rate field and only use the procedure code field to report procedure code for both professional and institutional claims in the OT file. The HCPCS rate field is no longer available in T-MSIS as of June 2022. For more information on CMS guidance issued to states, please refer to https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/51965
ICD-9 stands for International Statistical Classification of Diseases Clinical Modification, 9th edition.
ICD-10 stands for International Statistical Classification of Diseases Clinical Modification, 10th edition.
A \u201cZ\u201d at the end of an ICD-10-PCS code indicates that the specific qualifier does not apply to the procedure.
Inpatient procedure codes are used to document surgeries and other procedures that occur during an inpatient stay. This analysis examines how often the primary (first) procedure code is missing on IP records and the percentage of IP records with invalid procedure codes in any of the six procedure code fields.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5201"", ""relatedTopics"": [{""measureId"": 63, ""measureName"": ""Procedure Codes - OT Professional"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 1}, {""measureId"": 64, ""measureName"": ""Procedure Codes - OT Institutional"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 2}]}" 66,"{""measureId"": 66, ""measureName"": ""Missing Payment Data - FFS Claims"", ""groupId"": 8, ""groupName"": ""Payments"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Missing-Pmt-FFS-Claims.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) provide a unique source of expenditure information for all Medicaid and CHIP programs nationally and by state. Although other data sources, such as the Medicaid Budget & Expenditure System files, provide aggregate state-level expenditure information, T-MSIS is the only data source that allows users to analyze payments associated with managed care organizations, individual providers, subgroups of beneficiaries, and/or specific services. The Centers for Medicare & Medicaid Services (CMS) and other TAF users require accurate payment data to calculate total cost of care metrics; to compare costs for specific conditions, services, or populations; and to compare costs across states or different payment systems.
Medicaid and CHIP payments made on behalf of specific beneficiaries are captured in four types of records in the TAF claims files: (1) fee-for-service (FFS) claims, which represent payments to medical providers made directly by the state Medicaid or CHIP agency; (2) capitation payments, which reflect a set per member per month (PMPM) rate paid by the state Medicaid or CHIP agency to a managed care organization (MCO), prepaid health plan (PHP), or primary care provider; (3) managed care encounter records, which reflect payments made by MCOs or PHPs to providers for services rendered to covered beneficiaries; and (4) supplemental payments, which represent payments made in addition to a capitation payment or negotiated rate. [1] Because TAF claims records only include non-void, non-denied final action claims, nearly all of these records should have a positive total Medicaid paid amount. [2] , [3] A high percentage of claims reported with zero, missing, or negative payments may suggest a data quality or completeness issue, which may affect cost estimates based on TAF data.
This data quality assessment examines the extent to which states are reporting FFS claims and managed care encounters with missing or invalid payment data, as well as variation in the completeness and usability of the payment data on FFS claims and managed care encounters by file type. [4]
The TAF also include payment records that are not tied to specific beneficiaries, such as aggregate payments to transportation providers, called service tracking claims. Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
There are a few situations in which FFS claims are not expected to have a positive payment amount, including (1) claims for which Medicare or another liable third party already paid the full Medicaid allowable amount and (2) claims that are not paid on a FFS basis despite being processed at the individual claim level.
Previously, managed care encounters could be reported without positive payment amounts if the MCOs or PHPs had not agreed to provide payment data to the state. CMS expected all states to report provider payment amounts on managed care encounters no later than June 30, 2019, necessitating updates to some states’ managed care contracts. Because managed care payment data are proprietary, only TAF users with approval from CMS will be able to access information on what managed care plans pay providers for services. For more information, see: Reporting Paid and Billed Amounts on Managed Care Encounters in T-MSIS. July 2019. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/52005 .
Payments on managed care encounter records are redacted from the TAF RIF and require special permission to access.
The TAF also include payment records that are not tied to specific beneficiaries, such as aggregate payments to transportation providers, called service tracking claims. Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
There are a few situations in which FFS claims are not expected to have a positive payment amount, including (1) claims for which Medicare or another liable third party already paid the full Medicaid allowable amount and (2) claims that are not paid on a FFS basis despite being processed at the individual claim level.
Previously, managed care encounters could be reported without positive payment amounts if the MCOs or PHPs had not agreed to provide payment data to the state. CMS expected all states to report provider payment amounts on managed care encounters no later than June 30, 2019, necessitating updates to some states\u2019 managed care contracts. Because managed care payment data are proprietary, only TAF users with approval from CMS will be able to access information on what managed care plans pay providers for services. For more information, see: Reporting Paid and Billed Amounts on Managed Care Encounters in T-MSIS. July 2019. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/52005 .
Payments on managed care encounter records are redacted from the TAF RIF and require special permission to access.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We examined the values in the total Medicaid paid amount field (TOT_MDCD_PD_AMT) in the TAF [5] inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) header files. We included both Medicaid and CHIP FFS claims and managed care encounters, but we excluded (1) crossover claims [6] (those for which Medicare is the primary payer, and Medicaid is responsible only for covering the remaining cost-sharing on behalf of dually eligible beneficiaries); (2) capitation payments; (3) supplemental payments; and (4) service tracking payments. [7] , [8] We excluded states if the low volume of claims in the TAF rendered the data unusable for analysis. For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [9]
For each state, we calculated the percentage of records in each file where the total Medicaid paid amount was (1) missing, (2) zero dollars, or (3) any negative dollar amount. We then calculated the percentage of records that had a positive payment value. We grouped states into categories of low concern, medium concern, and high concern about the usability of their data, depending on the percentage of records that had a missing, zero, or negative payment value (Table 1). [10] , [11] The data quality assessment is based on a combination of the four claims file types, but the level of missingness can vary across file types.
Table 1. Criteria for DQ assessment of payment data
| Percentage of claims or encounters with zero, missing or negative payment in the total Medicaid paid amount field | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
Methods previously used to assess data quality
Table 2 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 2. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We classified crossover claims as records on which the crossover claim indicator (XOVR_IND) was equal to 1. If a claim was reported with a missing value in XOVR_IND, we retained it in our analysis.
We identified capitation, supplemental, and service tracking payments by using the claim type code (CLM_TYPE_CD). We also did not include any records with a claim type of “other,” which are records that the state did not classify as Medicaid or CHIP.
We limited this analysis to only FFS claims and managed care encounters because they are the two types of claims that will be most commonly analyzed by TAF users. Although capitation payments and supplemental payments also capture important payment data that can be attributed to individual beneficiaries, reviewing the usability of payment information on those claims was outside the scope of this assessment.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “TAF Technical Guidance: How to Use Illinois Claims Data,” on ResDAC.org.
It is not uncommon for a small percentage of TAF records to have payments equal to zero dollars, but missing or negative payment amounts are much less common. We recommend that TAF users interpret zero, missing, and negative payment amounts in the same manner—all three cases represent payment information that is likely unusable for research purposes.
The relatively generous threshold of 90 percent for a low level of concern should accommodate cases in which a zero payment is valid—for example, claims for which a third party already paid the Medicaid allowable amount.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We classified crossover claims as records on which the crossover claim indicator (XOVR_IND) was equal to 1. If a claim was reported with a missing value in XOVR_IND, we retained it in our analysis.
We identified capitation, supplemental, and service tracking payments by using the claim type code (CLM_TYPE_CD). We also did not include any records with a claim type of \u201cother,\u201d which are records that the state did not classify as Medicaid or CHIP.
We limited this analysis to only FFS claims and managed care encounters because they are the two types of claims that will be most commonly analyzed by TAF users. Although capitation payments and supplemental payments also capture important payment data that can be attributed to individual beneficiaries, reviewing the usability of payment information on those claims was outside the scope of this assessment.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cTAF Technical Guidance: How to Use Illinois Claims Data,\u201d on ResDAC.org.
It is not uncommon for a small percentage of TAF records to have payments equal to zero dollars, but missing or negative payment amounts are much less common. We recommend that TAF users interpret zero, missing, and negative payment amounts in the same manner\u2014all three cases represent payment information that is likely unusable for research purposes.
The relatively generous threshold of 90 percent for a low level of concern should accommodate cases in which a zero payment is valid\u2014for example, claims for which a third party already paid the Medicaid allowable amount.
The payment amount on TAF fee-for-service claims captures direct payments from the state Medicaid or CHIP agency to medical providers for services rendered to beneficiaries. Since the TAF excludes fully denied and voided claims, all FFS claims should have a positive payment amount. This analysis examines the extent to which FFS claims have missing, zero, or negative payment amounts, which are likely to represent data quality issues.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""6011"", ""relatedTopics"": [{""measureId"": 86, ""measureName"": ""Missing Payment Data - Encounters"", ""groupId"": 8, ""groupName"": ""Payments"", ""order"": 1}]}" 67,"{""measureId"": 67, ""measureName"": ""Total FFS Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Total-FFS-Expenditures.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) provide a unique source of expenditure information on states’ Medicaid and CHIP programs. Although other data sources provide aggregate state-level expenditure information, the TAF allow users to analyze payments associated with managed care organizations, individual providers, subgroups of beneficiaries, and specific services. The Centers for Medicare & Medicaid Services (CMS) and other TAF users require accurate payment data to calculate total cost of care metrics; to compare costs for specific conditions, services, or populations; and to compare costs across states or different payment systems.
States may reimburse health care providers for services delivered to Medicaid beneficiaries by paying directly for each covered service on a fee-for-service (FFS) basis or by paying a flat monthly payment per beneficiary for a set of services that is contracted to another entity—such as a managed care plan—which then assumes responsibility for delivering care to the beneficiary. [1] , [2] Fee-for-service (FFS) claims represent direct payments from the state Medicaid or Children’s Health Insurance Program (CHIP) agency to medical providers for services provided to beneficiaries. Most states use a mix of FFS and monthly beneficiary payments to pay for the care provided to Medicaid beneficiaries. In fiscal year 2017, FFS payments accounted for 47 percent of total Medicaid spending nationally on health services. [3]
In order to claim federal matching funds for their medical assistance programs, states report aggregate expenditure data for Medicaid beneficiaries covered under Title XIX of the Social Security Act to CMS through the Medicaid Budget and Expenditure System (MBES). [4] The MBES system is designed to capture information states would report on the Quarterly Medicaid Statement of Expenditures for the Medical Assistance Program Form CMS-64; therefore, the data are commonly referred to as the CMS-64 data. Because of the oversight involved in ensuring the correct federal reimbursement to state Medicaid programs, the CMS-64 data are seen as a highly reliable source of expenditure information. If the expenditures in the TAF do not align with those in the CMS-64, that may suggest TAF data are incomplete or have other data quality errors that render them unrepresentative of actual Medicaid expenditures.
This data quality assessment examines the extent to which FFS expenditures in the TAF data align with the FFS expenditures states report in the CMS-64, overall and for four major categories of service: (1) inpatient hospital services, (2) institutional long-term care services, (3) all other medical services, and (4) prescription drugs. However, benchmarking TAF expenditure data against CMS-64 data is not straightforward. Although the data in both systems are similar, there are some important differences: [5]
Monthly payments that states make on behalf of Medicaid beneficiaries include capitated payments to managed care plans; flat fees paid to a primary care provider for primary care case management plan (PCCM) services; Medicare Part A and Part B premiums for Medicaid beneficiaries who are dually eligible for Medicare; and in some cases, premium assistance for Medicaid beneficiaries with private coverage. More information on monthly beneficiary payment expenditures can be found in the DQ Atlas single topic displays for Total Monthly Beneficiary Payments , CMC Payments , PHP Payments , PCCM Fees , and Premium Assistance Payments .
In addition to FFS claims and monthly beneficiary payments, TAF also includes other types of payments, such as supplemental payment records and service tracking claims. Supplemental payment records represent payments for a specific service made in addition to a capitation payment or negotiated rate. Service tracking claims represent lump-sum payments made for services that cannot be attributed to a specific beneficiary, such as disproportionate share hospital (DSH) payments or payments to providers made under the Upper Payment Limit (UPL) demonstration. Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
Medicaid and CHIP Payment and Access Commission. “EXHIBIT 17. Total Medicaid Benefit Spending by State and Category, FY 2017 (millions).” MACStats: Medicaid and CHIP Data Book December 2018. Washington, DC: MACPAC, December 2018, pp. 52–54. Available at https://www.macpac.gov/wp-content/uploads/2018/12/December-2018-MACStats-Data-Book.pdf . Accessed June 24, 2019.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
Monthly payments that states make on behalf of Medicaid beneficiaries include capitated payments to managed care plans; flat fees paid to a primary care provider for primary care case management plan (PCCM) services; Medicare Part A and Part B premiums for Medicaid beneficiaries who are dually eligible for Medicare; and in some cases, premium assistance for Medicaid beneficiaries with private coverage. More information on monthly beneficiary payment expenditures can be found in the DQ Atlas single topic displays for Total Monthly Beneficiary Payments , CMC Payments , PHP Payments , PCCM Fees , and Premium Assistance Payments .
In addition to FFS claims and monthly beneficiary payments, TAF also includes other types of payments, such as supplemental payment records and service tracking claims. Supplemental payment records represent payments for a specific service made in addition to a capitation payment or negotiated rate. Service tracking claims represent lump-sum payments made for services that cannot be attributed to a specific beneficiary, such as disproportionate share hospital (DSH) payments or payments to providers made under the Upper Payment Limit (UPL) demonstration. Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
Medicaid and CHIP Payment and Access Commission. \u201cEXHIBIT 17. Total Medicaid Benefit Spending by State and Category, FY 2017 (millions).\u201d MACStats: Medicaid and CHIP Data Book December 2018. Washington, DC: MACPAC, December 2018, pp. 52\u201354. Available at https://www.macpac.gov/wp-content/uploads/2018/12/December-2018-MACStats-Data-Book.pdf . Accessed June 24, 2019.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We compared FFS expenditures for Medicaid beneficiaries drawn from twelve months of TAF data [6] to the CMS-64 benchmark for four quarters of the calendar year. [7]
To calculate FFS expenditures in the TAF, we first selected records from the TAF inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims files that represent FFS payments or supplemental wraparound payments, [8] using the claim type code. [9] We did not include TAF records that represent capitation payment records, managed care encounters, and service tracking claims, as well as all payments for S-CHIP beneficiaries, since those expenditures are not contained in the CMS-64 categories we used as the benchmark.
Next, we further restricted the FFS and supplemental payment records to those that matched to an eligibility record in the Demographic and Eligibility (DE) file that indicated that the beneficiary was enrolled in non-CHIP Medicaid in the month that the claim was incurred. [10] This is necessary because the CMS-64 benchmark includes only non-CHIP Medicaid expenditures, and the TAF claim type variable alone cannot distinguish payments made for Medicaid-expansion CHIP enrollees (whose costs are not reported on the CMS-64) from payments made for non-CHIP Medicaid beneficiaries (whose costs are reported on the CMS-64).
Finally, we excluded TAF records that represent DSH payments, [11] supplemental payments made under the UPL demonstration, electronic health record (EHR) payments, drug rebates, and Medicare premium assistance payments. [12] These payments were excluded because they do not represent FFS payments for health services.
Once we had selected the set of TAF records included in the benchmarking analysis, we tabulated expenditures associated with FFS payments in each state using the total Medicaid paid amount on the header record (TOT_MDCD_PD_AMT) for all included records. [13]
We calculated the CMS-64 benchmark for FFS expenditures using four quarterly net expenditures reports from MBES that cover a calendar year. [14] We summed the net expenditures for all the categories of service except those that would not be captured on FFS claims in the TAF, which include capitation payments, premium assistance payments, drug rebates, DSH payments, General Medical Education payments, and other supplemental payments. The categories of service included in the FFS benchmark are shown in Table 1. [15]
Benchmarking by FFS service category
To better understand why total Medicaid expenditures may differ between the TAF and CMS-64, we also examined four major service categories: (1) inpatient hospital, (2) institutional long-term care, (3) all other medical services, and (4) prescription drugs. Because the CMS-64 categories of service do not line up exactly with the organization of the TAF claims files, we used both TAF file types (IP, LT, OT, and RX) and the federally assigned service category (FASC) code and type of service code (TOS_CD) for this mapping, as shown in Table 1. [16] , [17] In some cases, states have known issues with reporting FFS claims into the wrong file, or with assigning a valid type of service code. [18] To present results consistently across states, we did not correct for any known errors in file type or type of service code when tabulating the TAF-based expenditures by category of service.
Table 1. Assigning FFS expenditures to major service categories in the TAF and CMS-64
| Major category | TAF records a | CMS-64 categories |
|---|---|---|
| Inpatient hospital services | All claims with CLM_TYPE_CD=1 and FED_SRVC_CTGRY_CD = 21 (inpatient hospital) except those with TOS_CD values of 058, 084, 090, or those with a TOS_CD value for DSH payment b (123) as the only code on the claim. The excluded codes represent services by religious nonmedical health care institutions, sterilizations, c and critical access hospital services. d Claims with CLM_TYPE_CD=1 in the IP file where FED_SRVC_CTGRY_CD = NULL (missing) and TOS_CD = 001 (inpatient hospital) or NULL (missing). | Category of service 1A. This represents inpatient hospital services. |
| Institutional long-term care services | Claims with CLM_TYPE_CD=1 and FED_SRVC_CTGRY_CD = 22, 23, or 24. These represent payments for nursing facility, intermediate care facility, and all other overnight facility services. Claims with CLM_TYPE_CD=1 from the LT file that did not have claim lines or where FED_SRVC_CTGRY_CD = NULL (missing) and the TOS_CD = 009, 044, 045, 046, 047, 048, 059, or NULL (missing). All records with CLM_TYPE_CD=5 from the LT file that linked to a beneficiary except those with FED_SRVC_CTGRY_CD = 11 or 12, or TOS_CD values of 119, 120, 121, 122, 132, 133, 134, 138, 143, or 144. e The included codes represent wraparound payments for long-term care services. | Categories of service 2A, 3A, 4A, and 4B. These represent payments for mental health facility, nursing facility, and intermediate care facility services. |
| All other medical services | All claims with CLM_TYPE_CD=1 and FED_SRVC_CTGRY_CD = 25, 26, 27, 28, 31, 32, 33, 34, 35, 36, 37, and 38. These represent payments for hospice services, outpatient facility and institutional claims, clinic services, radiology, laboratory, home health services, transportation services, dental services, HCBS services, durable medical equipment and supplies, and physician and all other professional claims. Claims with CLM_TYPE_CD=1 with FED_SRVC_CTGRY_CD = 21 (inpatient hospital) and TOS_CD values of 058, 084, and 090. The included codes represent services by religious nonmedical health care institutions, sterilizations, c and critical access hospital services (inpatient). All records with CLM_TYPE_CD=5 from the IP and OT files that linked to a beneficiary except those with FED_SRVC_CTGRY_CD = 11 or 12, or TOS_CD values of 119, 120, 121, 122, 132 , 133, 134, 138, 143, or 144. e The included codes represent wraparound payments for inpatient and other medical services. Any remaining claims with CLM_TYPE_CD=1 from the LT file that did not indicate institutional care. | Categories of service 2C, 5A, 5C, 5D, 6A, 8, 9A, 10A, 11, 12, 13, 14, 15, 16, 17D, 18D, 19A, 19B, 19C, 19D, 23A, 23B, 24A, 24B, 26, 27, 28, 29A, 30, 31, 32, 33, 34, 34A, 35, 36, 37A, 38, 39, 40, 41, 42, 43, 44, 45, 46B, 47, 48, 49, 69. f These represent payments for Certified Community Behavioral Health Clinics under the Section 1332 demonstration; physician and surgical services; outpatient hospital services; dental services; other practitioners services; clinic services; laboratory and radiological services; home health services; sterilizations; abortions; EPSDT screenings; rural health services; Medicare deductibles and coinsurance; Medicaid coinsurance; home and community-based services; personal care services; case management services; hospice benefits; emergency services for undocumented aliens; Federally-Qualified Health Center services; regular payments for non-emergency medical transportation; physical therapy; occupational therapy; services for speech, hearing, and language; prosthetic devices, dentures, and eyeglasses; diagnostic screening and preventive services; preventive services and vaccines; nurse mid-wife services; emergency hospital services; regular payments for critical access hospital services; nurse practitioner services; school-based services; rehabilitative services (non-school-based); private duty nursing services; freestanding birth center services; health home for enrollees with chronic conditions; tobacco cessation for pregnant women; health home with chronic conditions, opioid use disorder medication-assisted treatment services; COVID-19 vaccines and vaccine administration, qualified community based mobile crisis intervention, health homes for children with medically complex conditions, and other care services. |
| Prescription drugs | Claims with CLM_TYPE_CD=1 from the RX file with FED_SRVC_CTGRY_CD = 41 (Prescription drug). All records with CLM_TYPE_CD=5 from the RX file that linked to a beneficiary except those with FED_SRVC_CTGRY_CD = 11 or 12, or TOS_CD values of 119, 120, 121, 122, 123, 132, 133, 134, 138, 143, 144. e The included codes represent wraparound payments for prescribed drugs. | Category of service 7 and 46. This represents payments for prescribed drugs and drugs for medication-assisted treatment for opioid use disorder. |
a We first exclude TAF records that represent supplemental payments made under the UPL demonstration, Medicare premiums, EHR payments to providers, and drug rebates.
b We include inpatient hospital claims with a TOS CD of 123 if it is not the only TOS CD on the claim. For claims with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the total FFS expenditure.
c We include sterilizations in inpatient hospital expenditures if it is not the only service on an inpatient hospital claim. We include sterilizations in all other medical expenditures if it is the only service on an inpatient hospital claim.
d We exclude critical access hospital services from inpatient hospital expenditures because they are captured in a separate CMS-64 service category―37A Critical Access Hospitals – Regular Payments. We include the expenditures for critical access hospital services that are captured in the TAF and CMS-64 data in the category for all other medical services.
e Type of service codes 138, 143, and 144 are valid starting in 2020.
f CMS-64 category of service 2C is valid starting in 2017. CMS-64 category of service 45 (health home for enrollees with substance abuse disorder) is a valid value starting in 2019. CMS-64 categories of service 46 and 46B (medication-assisted treatment for opioid use disorder) are valid starting in 2020. CMS-64 categories of service 10A and 10B (clinic services), and 47 (COVID-19 vaccines and vaccine administration) are valid starting in 2021. Category of service 10 (clinic services) was separated into categories 10A and 10B starting in 2021 to differentiate regular payments and supplemental payments, respectively. Category of service 29 (non-emergency medical transportation) was separated into categories 29A and 29B starting in 2022 to differentiate regular payments and supplemental payments, respectively. Category of service 37 (critical access hospital services) was separated into categories 37A, 37B, and 37C starting in 2022 to differentiate regular payments, supplemental payments for inpatient services, and supplemental payments for outpatient services, respectively. CMS-64 category of service 48 (qualified community based mobile crisis intervention services) is valid starting in 2022. CMS-64 category of service 49 (health homes for children with medically complex conditions) is valid starting in 2023.
Data quality assessment criteria
We categorized each state’s FFS expenditures in TAF as having high, moderate, low, or very low alignment with the CMS-64 based on the percent difference between the two data sources (Table 2). [19] To help users determine what may be driving differences between the data sources in each state, we also present the alignment for each of the four major service categories (inpatient hospital, long-term care, all other medical services, and prescription drugs).
Table 2. Criteria for DQ assessment of FFS expenditures
| Percent difference between TAF and CMS-64 FFS expenditures | Level of alignment | DQ assessment |
|---|---|---|
| x < 5 percent | High | Low concern |
| 5 percent ≤ x < 20 percent | Moderate | Medium concern |
| 20 percent ≤ x < 50 percent | Low | High concern |
| x ≥ 50 percent or if state has no TAF data | Very low | Unusable |
States with a low absolute percentage difference between the expenditures captured in TAF and the CMS-64 are considered to present a low level of concern about the completeness and quality of their TAF data. However, TAF users should be aware that in states with large Medicaid programs, even a small percentage difference may represent a substantial difference in the dollar amount between the data sources.
There are three reasons other than the quality and completeness of TAF expenditure data that could result in a state’s FFS expenditures in the TAF differing from the CMS-64: (1) structural differences in the data sets, such as organizing the data by service date versus claim payment date; (2) data quality problems in the TAF that are unrelated to expenditures, such as a low volume of claims in the TAF because of missing date values on claims or issues with assigning a final action status to claims; and (3) variation in how states report certain types of FFS expenditures on the CMS-64 that are difficult to reproduce in the TAF data. For instance, a state may make lump-sum reconciliation payments to hospitals as part of their cost-based FFS payments for inpatient services. These payments may be reported in the inpatient hospital base payment category of service in the CMS-64, which is included as part of the FFS benchmark, but report the payments as service tracking claims in T-MSIS, which would not be included in this analysis. When this occurs, we would expect to see a difference between the TAF-based FFS expenditures and the CMS-64 benchmark even if TAF expenditure data are complete and accurate.
Methods previously used to assess data quality
Table 3 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 3. Previously used methods and applicable data years and versions
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
We consider supplemental payments that link to a non-CHIP Medicaid beneficiary (except those with FED_SRVC_CTGRY_CD values 11 or 12, or TOS_CD values of 119, 120, 121, 122, 132, 133, 134, 138, 143, or 144) to be wraparound payments. Type of service codes 138, 143, and 144 are valid starting in 2020. As noted later, we exclude TAF records that represent supplemental payments made under the UPL demonstration and all other lump sum payments.
FFS claims representing Medicaid or M-CHIP benefits have a claim type code value of 1. Supplemental payment claims have a claim type code value of 5.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the CHIP code value was 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-76, we considered the beneficiary to have non-CHIP Medicaid enrollment in that month.
In some states, DSH payment amounts are included on FFS claim headers. For claims with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the total FFS expenditure.
We used type of service code to identify and exclude supplemental payments (132, 133, 134), EHR payments (135) and Medicare premium assistance payments (139, 140, 141, 142). Type of service codes 139, 140, 141, and 142 are valid starting in 2020. We used the federally assigned service category code to identify and exclude DSH payments (13). Drug rebates were identified as records were any line had a type of service code of 131.
In a sensitivity analysis, we found that expenditures based on the TAF header-level payment variable (TOT_MDCD_PD_AMT) alone had higher alignment with CMS-64 expenditures compared to tabulating expenditures using a mix of header- and line-level payments based on the payment level indicator (PYMT_LVL_IND). In part, this may reflect that TAF includes denied line records that are a part of a larger paid claim, and the payment amount on the denied lines may be incorrect.
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
Definitions for each category of service in the CMS-64 can be found on Medicaid.gov: https://www.medicaid.gov/medicaid/downloads/cms-649-base-category-of-services-definition-2-14.pdf
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
Although the TAF includes a data element for states to indicate the CMS-64 category of service (XIX_SRVC_CTGRY_CD) for each claim, this data element has a high rate of missing values in many states and, when it is not missing, it is unclear whether the logic used to assign values is aligned with how states actually classify spending in their CMS-64 reporting.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX .
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: [(TAF – benchmark) / benchmark]*100.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
We consider supplemental payments that link to a non-CHIP Medicaid beneficiary (except those with FED_SRVC_CTGRY_CD values 11 or 12, or TOS_CD values of 119, 120, 121, 122, 132, 133, 134, 138, 143, or 144) to be wraparound payments. Type of service codes 138, 143, and 144 are valid starting in 2020. As noted later, we exclude TAF records that represent supplemental payments made under the UPL demonstration and all other lump sum payments.
FFS claims representing Medicaid or M-CHIP benefits have a claim type code value of 1. Supplemental payment claims have a claim type code value of 5.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the CHIP code value was 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-76, we considered the beneficiary to have non-CHIP Medicaid enrollment in that month.
In some states, DSH payment amounts are included on FFS claim headers. For claims with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the total FFS expenditure.
We used type of service code to identify and exclude supplemental payments (132, 133, 134), EHR payments (135) and Medicare premium assistance payments (139, 140, 141, 142). Type of service codes 139, 140, 141, and 142 are valid starting in 2020. We used the federally assigned service category code to identify and exclude DSH payments (13). Drug rebates were identified as records were any line had a type of service code of 131.
In a sensitivity analysis, we found that expenditures based on the TAF header-level payment variable (TOT_MDCD_PD_AMT) alone had higher alignment with CMS-64 expenditures compared to tabulating expenditures using a mix of header- and line-level payments based on the payment level indicator (PYMT_LVL_IND). In part, this may reflect that TAF includes denied line records that are a part of a larger paid claim, and the payment amount on the denied lines may be incorrect.
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
Definitions for each category of service in the CMS-64 can be found on Medicaid.gov: https://www.medicaid.gov/medicaid/downloads/cms-649-base-category-of-services-definition-2-14.pdf
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
Although the TAF includes a data element for states to indicate the CMS-64 category of service (XIX_SRVC_CTGRY_CD) for each claim, this data element has a high rate of missing values in many states and, when it is not missing, it is unclear whether the logic used to assign values is aligned with how states actually classify spending in their CMS-64 reporting.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX .
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: [(TAF \u2013 benchmark) / benchmark]*100.
Fee-for-service expenditures represent direct payments from the state Medicaid or CHIP agency to medical providers for services rendered to beneficiaries. This analysis examines how well the TAF data on total FFS expenditures on behalf of Medicaid beneficiaries align with an external benchmark, the CMS-64.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""6021"", ""relatedTopics"": [{""measureId"": 68, ""measureName"": ""FFS Inpatient Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 1}, {""measureId"": 69, ""measureName"": ""FFS Long-Term Care Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 2}, {""measureId"": 70, ""measureName"": ""FFS Other Medical Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 3}, {""measureId"": 71, ""measureName"": ""FFS Prescription Drug Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 4}]}" 68,"{""measureId"": 68, ""measureName"": ""FFS Inpatient Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-FFS-IP-Expenditures.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) provide a unique source of expenditure information on states’ Medicaid and CHIP programs. Although other data sources provide aggregate state-level expenditure information, the TAF allow users to analyze payments associated with managed care organizations, individual providers, subgroups of beneficiaries, and specific services. The Centers for Medicare & Medicaid Services (CMS) and other TAF users require accurate payment data to calculate total cost of care metrics; to compare costs for specific conditions, services, or populations; and to compare costs across states or different payment systems.
States may reimburse health care providers for services delivered to Medicaid beneficiaries by paying directly for each covered service on a fee-for-service (FFS) basis or by paying a flat monthly payment per beneficiary for a set of services that is contracted to another entity—such as a managed care plan—which then assumes responsibility for delivering care to the beneficiary. [1] , [2] Fee-for-service (FFS) claims represent direct payments from the state Medicaid or Children’s Health Insurance Program (CHIP) agency to medical providers for services provided to beneficiaries. Most states use a mix of FFS and monthly beneficiary payments to pay for the care provided to Medicaid beneficiaries. In fiscal year 2017, FFS payments accounted for 47 percent of total Medicaid spending nationally on health services. [3]
In order to claim federal matching funds for their medical assistance programs, states report aggregate expenditure data for Medicaid beneficiaries covered under Title XIX of the Social Security Act to CMS through the Medicaid Budget and Expenditure System (MBES). [4] The MBES system is designed to capture information states would report on the Quarterly Medicaid Statement of Expenditures for the Medical Assistance Program Form CMS-64; therefore, the data are commonly referred to as the CMS-64 data. Because of the oversight involved in ensuring the correct federal reimbursement to state Medicaid programs, the CMS-64 data are seen as a highly reliable source of expenditure information. If the expenditures in the TAF do not align with those in the CMS-64, that may suggest TAF data are incomplete or have other data quality errors that render them unrepresentative of actual Medicaid expenditures.
This data quality assessment examines the extent to which FFS expenditures in the TAF data align with the FFS expenditures states report in the CMS-64, overall and for four major categories of service: (1) inpatient hospital services, (2) institutional long-term care services, (3) all other medical services, and (4) prescription drugs. However, benchmarking TAF expenditure data against CMS-64 data is not straightforward. Although the data in both systems are similar, there are some important differences: [5]
Monthly payments that states make on behalf of Medicaid beneficiaries include capitated payments to managed care plans; flat fees paid to a primary care provider for primary care case management plan (PCCM) services; Medicare Part A and Part B premiums for Medicaid beneficiaries who are dually eligible for Medicare; and in some cases, premium assistance for Medicaid beneficiaries with private coverage. More information on monthly beneficiary payment expenditures can be found in the DQ Atlas single topic displays for Total Monthly Beneficiary Payments , CMC Payments , PHP Payments , PCCM Fees , and Premium Assistance Payments .
In addition to FFS claims and monthly beneficiary payments, TAF also includes other types of payments, such as supplemental payment records and service tracking claims. Supplemental payment records represent payments for a specific service made in addition to a capitation payment or negotiated rate. Service tracking claims represent lump-sum payments made for services that cannot be attributed to a specific beneficiary, such as disproportionate share hospital (DSH) payments or payments to providers made under the Upper Payment Limit (UPL) demonstration. Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
Medicaid and CHIP Payment and Access Commission. “EXHIBIT 17. Total Medicaid Benefit Spending by State and Category, FY 2017 (millions).” MACStats: Medicaid and CHIP Data Book December 2018. Washington, DC: MACPAC, December 2018, pp. 52–54. Available at https://www.macpac.gov/wp-content/uploads/2018/12/December-2018-MACStats-Data-Book.pdf . Accessed June 24, 2019.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
Monthly payments that states make on behalf of Medicaid beneficiaries include capitated payments to managed care plans; flat fees paid to a primary care provider for primary care case management plan (PCCM) services; Medicare Part A and Part B premiums for Medicaid beneficiaries who are dually eligible for Medicare; and in some cases, premium assistance for Medicaid beneficiaries with private coverage. More information on monthly beneficiary payment expenditures can be found in the DQ Atlas single topic displays for Total Monthly Beneficiary Payments , CMC Payments , PHP Payments , PCCM Fees , and Premium Assistance Payments .
In addition to FFS claims and monthly beneficiary payments, TAF also includes other types of payments, such as supplemental payment records and service tracking claims. Supplemental payment records represent payments for a specific service made in addition to a capitation payment or negotiated rate. Service tracking claims represent lump-sum payments made for services that cannot be attributed to a specific beneficiary, such as disproportionate share hospital (DSH) payments or payments to providers made under the Upper Payment Limit (UPL) demonstration. Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
Medicaid and CHIP Payment and Access Commission. \u201cEXHIBIT 17. Total Medicaid Benefit Spending by State and Category, FY 2017 (millions).\u201d MACStats: Medicaid and CHIP Data Book December 2018. Washington, DC: MACPAC, December 2018, pp. 52\u201354. Available at https://www.macpac.gov/wp-content/uploads/2018/12/December-2018-MACStats-Data-Book.pdf . Accessed June 24, 2019.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We compared FFS expenditures for Medicaid beneficiaries drawn from twelve months of TAF data [6] to the CMS-64 benchmark for four quarters of the calendar year. [7]
To calculate FFS expenditures in the TAF, we first selected records from the TAF inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims files that represent FFS payments or supplemental wraparound payments, [8] using the claim type code. [9] We did not include TAF records that represent capitation payment records, managed care encounters, and service tracking claims, as well as all payments for S-CHIP beneficiaries, since those expenditures are not contained in the CMS-64 categories we used as the benchmark.
Next, we further restricted the FFS and supplemental payment records to those that matched to an eligibility record in the Demographic and Eligibility (DE) file that indicated that the beneficiary was enrolled in non-CHIP Medicaid in the month that the claim was incurred. [10] This is necessary because the CMS-64 benchmark includes only non-CHIP Medicaid expenditures, and the TAF claim type variable alone cannot distinguish payments made for Medicaid-expansion CHIP enrollees (whose costs are not reported on the CMS-64) from payments made for non-CHIP Medicaid beneficiaries (whose costs are reported on the CMS-64).
Finally, we excluded TAF records that represent DSH payments, [11] supplemental payments made under the UPL demonstration, electronic health record (EHR) payments, drug rebates, and Medicare premium assistance payments. [12] These payments were excluded because they do not represent FFS payments for health services.
Once we had selected the set of TAF records included in the benchmarking analysis, we tabulated expenditures associated with FFS payments in each state using the total Medicaid paid amount on the header record (TOT_MDCD_PD_AMT) for all included records. [13]
We calculated the CMS-64 benchmark for FFS expenditures using four quarterly net expenditures reports from MBES that cover a calendar year. [14] We summed the net expenditures for all the categories of service except those that would not be captured on FFS claims in the TAF, which include capitation payments, premium assistance payments, drug rebates, DSH payments, General Medical Education payments, and other supplemental payments. The categories of service included in the FFS benchmark are shown in Table 1. [15]
Benchmarking by FFS service category
To better understand why total Medicaid expenditures may differ between the TAF and CMS-64, we also examined four major service categories: (1) inpatient hospital, (2) institutional long-term care, (3) all other medical services, and (4) prescription drugs. Because the CMS-64 categories of service do not line up exactly with the organization of the TAF claims files, we used both TAF file types (IP, LT, OT, and RX) and the federally assigned service category (FASC) code and type of service code (TOS_CD) for this mapping, as shown in Table 1. [16] , [17] In some cases, states have known issues with reporting FFS claims into the wrong file, or with assigning a valid type of service code. [18] To present results consistently across states, we did not correct for any known errors in file type or type of service code when tabulating the TAF-based expenditures by category of service.
Table 1. Assigning FFS expenditures to major service categories in the TAF and CMS-64
| Major category | TAF records a | CMS-64 categories |
|---|---|---|
| Inpatient hospital services | All claims with CLM_TYPE_CD=1 and FED_SRVC_CTGRY_CD = 21 (inpatient hospital) except those with TOS_CD values of 058, 084, 090, or those with a TOS_CD value for DSH payment b (123) as the only code on the claim. The excluded codes represent services by religious nonmedical health care institutions, sterilizations, c and critical access hospital services. d Claims with CLM_TYPE_CD=1 in the IP file where FED_SRVC_CTGRY_CD = NULL (missing) and TOS_CD = 001 (inpatient hospital) or NULL (missing). | Category of service 1A. This represents inpatient hospital services. |
| Institutional long-term care services | Claims with CLM_TYPE_CD=1 and FED_SRVC_CTGRY_CD = 22, 23, or 24. These represent payments for nursing facility, intermediate care facility, and all other overnight facility services. Claims with CLM_TYPE_CD=1 from the LT file that did not have claim lines or where FED_SRVC_CTGRY_CD = NULL (missing) and the TOS_CD = 009, 044, 045, 046, 047, 048, 059, or NULL (missing). All records with CLM_TYPE_CD=5 from the LT file that linked to a beneficiary except those with FED_SRVC_CTGRY_CD = 11 or 12, or TOS_CD values of 119, 120, 121, 122, 132, 133, 134, 138, 143, or 144. e The included codes represent wraparound payments for long-term care services. | Categories of service 2A, 3A, 4A, and 4B. These represent payments for mental health facility, nursing facility, and intermediate care facility services. |
| All other medical services | All claims with CLM_TYPE_CD=1 and FED_SRVC_CTGRY_CD = 25, 26, 27, 28, 31, 32, 33, 34, 35, 36, 37, and 38. These represent payments for hospice services, outpatient facility and institutional claims, clinic services, radiology, laboratory, home health services, transportation services, dental services, HCBS services, durable medical equipment and supplies, and physician and all other professional claims. Claims with CLM_TYPE_CD=1 with FED_SRVC_CTGRY_CD = 21 (inpatient hospital) and TOS_CD values of 058, 084, and 090. The included codes represent services by religious nonmedical health care institutions, sterilizations, c and critical access hospital services (inpatient). All records with CLM_TYPE_CD=5 from the IP and OT files that linked to a beneficiary except those with FED_SRVC_CTGRY_CD = 11 or 12, or TOS_CD values of 119, 120, 121, 122, 132 , 133, 134, 138, 143, or 144. e The included codes represent wraparound payments for inpatient and other medical services. Any remaining claims with CLM_TYPE_CD=1 from the LT file that did not indicate institutional care. | Categories of service 2C, 5A, 5C, 5D, 6A, 8, 9A, 10A, 11, 12, 13, 14, 15, 16, 17D, 18D, 19A, 19B, 19C, 19D, 23A, 23B, 24A, 24B, 26, 27, 28, 29A, 30, 31, 32, 33, 34, 34A, 35, 36, 37A, 38, 39, 40, 41, 42, 43, 44, 45, 46B, 47, 48, 49, 69. f These represent payments for Certified Community Behavioral Health Clinics under the Section 1332 demonstration; physician and surgical services; outpatient hospital services; dental services; other practitioners services; clinic services; laboratory and radiological services; home health services; sterilizations; abortions; EPSDT screenings; rural health services; Medicare deductibles and coinsurance; Medicaid coinsurance; home and community-based services; personal care services; case management services; hospice benefits; emergency services for undocumented aliens; Federally-Qualified Health Center services; regular payments for non-emergency medical transportation; physical therapy; occupational therapy; services for speech, hearing, and language; prosthetic devices, dentures, and eyeglasses; diagnostic screening and preventive services; preventive services and vaccines; nurse mid-wife services; emergency hospital services; regular payments for critical access hospital services; nurse practitioner services; school-based services; rehabilitative services (non-school-based); private duty nursing services; freestanding birth center services; health home for enrollees with chronic conditions; tobacco cessation for pregnant women; health home with chronic conditions, opioid use disorder medication-assisted treatment services; COVID-19 vaccines and vaccine administration, qualified community based mobile crisis intervention, health homes for children with medically complex conditions, and other care services. |
| Prescription drugs | Claims with CLM_TYPE_CD=1 from the RX file with FED_SRVC_CTGRY_CD = 41 (Prescription drug). All records with CLM_TYPE_CD=5 from the RX file that linked to a beneficiary except those with FED_SRVC_CTGRY_CD = 11 or 12, or TOS_CD values of 119, 120, 121, 122, 123, 132, 133, 134, 138, 143, 144. e The included codes represent wraparound payments for prescribed drugs. | Category of service 7 and 46. This represents payments for prescribed drugs and drugs for medication-assisted treatment for opioid use disorder. |
a We first exclude TAF records that represent supplemental payments made under the UPL demonstration, Medicare premiums, EHR payments to providers, and drug rebates.
b We include inpatient hospital claims with a TOS CD of 123 if it is not the only TOS CD on the claim. For claims with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the total FFS expenditure.
c We include sterilizations in inpatient hospital expenditures if it is not the only service on an inpatient hospital claim. We include sterilizations in all other medical expenditures if it is the only service on an inpatient hospital claim.
d We exclude critical access hospital services from inpatient hospital expenditures because they are captured in a separate CMS-64 service category―37A Critical Access Hospitals – Regular Payments. We include the expenditures for critical access hospital services that are captured in the TAF and CMS-64 data in the category for all other medical services.
e Type of service codes 138, 143, and 144 are valid starting in 2020.
f CMS-64 category of service 2C is valid starting in 2017. CMS-64 category of service 45 (health home for enrollees with substance abuse disorder) is a valid value starting in 2019. CMS-64 categories of service 46 and 46B (medication-assisted treatment for opioid use disorder) are valid starting in 2020. CMS-64 categories of service 10A and 10B (clinic services), and 47 (COVID-19 vaccines and vaccine administration) are valid starting in 2021. Category of service 10 (clinic services) was separated into categories 10A and 10B starting in 2021 to differentiate regular payments and supplemental payments, respectively. Category of service 29 (non-emergency medical transportation) was separated into categories 29A and 29B starting in 2022 to differentiate regular payments and supplemental payments, respectively. Category of service 37 (critical access hospital services) was separated into categories 37A, 37B, and 37C starting in 2022 to differentiate regular payments, supplemental payments for inpatient services, and supplemental payments for outpatient services, respectively. CMS-64 category of service 48 (qualified community based mobile crisis intervention services) is valid starting in 2022. CMS-64 category of service 49 (health homes for children with medically complex conditions) is valid starting in 2023.
Data quality assessment criteria
We categorized each state’s FFS expenditures in TAF as having high, moderate, low, or very low alignment with the CMS-64 based on the percent difference between the two data sources (Table 2). [19] To help users determine what may be driving differences between the data sources in each state, we also present the alignment for each of the four major service categories (inpatient hospital, long-term care, all other medical services, and prescription drugs).
Table 2. Criteria for DQ assessment of FFS expenditures
| Percent difference between TAF and CMS-64 FFS expenditures | Level of alignment | DQ assessment |
|---|---|---|
| x < 5 percent | High | Low concern |
| 5 percent ≤ x < 20 percent | Moderate | Medium concern |
| 20 percent ≤ x < 50 percent | Low | High concern |
| x ≥ 50 percent or if state has no TAF data | Very low | Unusable |
States with a low absolute percentage difference between the expenditures captured in TAF and the CMS-64 are considered to present a low level of concern about the completeness and quality of their TAF data. However, TAF users should be aware that in states with large Medicaid programs, even a small percentage difference may represent a substantial difference in the dollar amount between the data sources.
There are three reasons other than the quality and completeness of TAF expenditure data that could result in a state’s FFS expenditures in the TAF differing from the CMS-64: (1) structural differences in the data sets, such as organizing the data by service date versus claim payment date; (2) data quality problems in the TAF that are unrelated to expenditures, such as a low volume of claims in the TAF because of missing date values on claims or issues with assigning a final action status to claims; and (3) variation in how states report certain types of FFS expenditures on the CMS-64 that are difficult to reproduce in the TAF data. For instance, a state may make lump-sum reconciliation payments to hospitals as part of their cost-based FFS payments for inpatient services. These payments may be reported in the inpatient hospital base payment category of service in the CMS-64, which is included as part of the FFS benchmark, but report the payments as service tracking claims in T-MSIS, which would not be included in this analysis. When this occurs, we would expect to see a difference between the TAF-based FFS expenditures and the CMS-64 benchmark even if TAF expenditure data are complete and accurate.
Methods previously used to assess data quality
Table 3 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 3. Previously used methods and applicable data years and versions
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
We consider supplemental payments that link to a non-CHIP Medicaid beneficiary (except those with FED_SRVC_CTGRY_CD values 11 or 12, or TOS_CD values of 119, 120, 121, 122, 132, 133, 134, 138, 143, or 144) to be wraparound payments. Type of service codes 138, 143, and 144 are valid starting in 2020. As noted later, we exclude TAF records that represent supplemental payments made under the UPL demonstration and all other lump sum payments.
FFS claims representing Medicaid or M-CHIP benefits have a claim type code value of 1. Supplemental payment claims have a claim type code value of 5.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the CHIP code value was 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-76, we considered the beneficiary to have non-CHIP Medicaid enrollment in that month.
In some states, DSH payment amounts are included on FFS claim headers. For claims with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the total FFS expenditure.
We used type of service code to identify and exclude supplemental payments (132, 133, 134), EHR payments (135) and Medicare premium assistance payments (139, 140, 141, 142). Type of service codes 139, 140, 141, and 142 are valid starting in 2020. We used the federally assigned service category code to identify and exclude DSH payments (13). Drug rebates were identified as records were any line had a type of service code of 131.
In a sensitivity analysis, we found that expenditures based on the TAF header-level payment variable (TOT_MDCD_PD_AMT) alone had higher alignment with CMS-64 expenditures compared to tabulating expenditures using a mix of header- and line-level payments based on the payment level indicator (PYMT_LVL_IND). In part, this may reflect that TAF includes denied line records that are a part of a larger paid claim, and the payment amount on the denied lines may be incorrect.
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
Definitions for each category of service in the CMS-64 can be found on Medicaid.gov: https://www.medicaid.gov/medicaid/downloads/cms-649-base-category-of-services-definition-2-14.pdf
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
Although the TAF includes a data element for states to indicate the CMS-64 category of service (XIX_SRVC_CTGRY_CD) for each claim, this data element has a high rate of missing values in many states and, when it is not missing, it is unclear whether the logic used to assign values is aligned with how states actually classify spending in their CMS-64 reporting.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX .
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: [(TAF – benchmark) / benchmark]*100.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
We consider supplemental payments that link to a non-CHIP Medicaid beneficiary (except those with FED_SRVC_CTGRY_CD values 11 or 12, or TOS_CD values of 119, 120, 121, 122, 132, 133, 134, 138, 143, or 144) to be wraparound payments. Type of service codes 138, 143, and 144 are valid starting in 2020. As noted later, we exclude TAF records that represent supplemental payments made under the UPL demonstration and all other lump sum payments.
FFS claims representing Medicaid or M-CHIP benefits have a claim type code value of 1. Supplemental payment claims have a claim type code value of 5.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the CHIP code value was 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-76, we considered the beneficiary to have non-CHIP Medicaid enrollment in that month.
In some states, DSH payment amounts are included on FFS claim headers. For claims with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the total FFS expenditure.
We used type of service code to identify and exclude supplemental payments (132, 133, 134), EHR payments (135) and Medicare premium assistance payments (139, 140, 141, 142). Type of service codes 139, 140, 141, and 142 are valid starting in 2020. We used the federally assigned service category code to identify and exclude DSH payments (13). Drug rebates were identified as records were any line had a type of service code of 131.
In a sensitivity analysis, we found that expenditures based on the TAF header-level payment variable (TOT_MDCD_PD_AMT) alone had higher alignment with CMS-64 expenditures compared to tabulating expenditures using a mix of header- and line-level payments based on the payment level indicator (PYMT_LVL_IND). In part, this may reflect that TAF includes denied line records that are a part of a larger paid claim, and the payment amount on the denied lines may be incorrect.
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
Definitions for each category of service in the CMS-64 can be found on Medicaid.gov: https://www.medicaid.gov/medicaid/downloads/cms-649-base-category-of-services-definition-2-14.pdf
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
Although the TAF includes a data element for states to indicate the CMS-64 category of service (XIX_SRVC_CTGRY_CD) for each claim, this data element has a high rate of missing values in many states and, when it is not missing, it is unclear whether the logic used to assign values is aligned with how states actually classify spending in their CMS-64 reporting.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX .
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: [(TAF \u2013 benchmark) / benchmark]*100.
Fee-for-service expenditures represent direct payments from the state Medicaid or CHIP agency to medical providers for services rendered to beneficiaries. This analysis examines how well the TAF data on FFS expenditures for payments to hospitals for inpatient services on behalf of Medicaid beneficiaries align with an external benchmark, the CMS-64.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""6021"", ""relatedTopics"": [{""measureId"": 67, ""measureName"": ""Total FFS Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 0}, {""measureId"": 69, ""measureName"": ""FFS Long-Term Care Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 2}, {""measureId"": 70, ""measureName"": ""FFS Other Medical Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 3}, {""measureId"": 71, ""measureName"": ""FFS Prescription Drug Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 4}]}" 69,"{""measureId"": 69, ""measureName"": ""FFS Long-Term Care Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-FFS-LT-Expenditures.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) provide a unique source of expenditure information on states’ Medicaid and CHIP programs. Although other data sources provide aggregate state-level expenditure information, the TAF allow users to analyze payments associated with managed care organizations, individual providers, subgroups of beneficiaries, and specific services. The Centers for Medicare & Medicaid Services (CMS) and other TAF users require accurate payment data to calculate total cost of care metrics; to compare costs for specific conditions, services, or populations; and to compare costs across states or different payment systems.
States may reimburse health care providers for services delivered to Medicaid beneficiaries by paying directly for each covered service on a fee-for-service (FFS) basis or by paying a flat monthly payment per beneficiary for a set of services that is contracted to another entity—such as a managed care plan—which then assumes responsibility for delivering care to the beneficiary. [1] , [2] Fee-for-service (FFS) claims represent direct payments from the state Medicaid or Children’s Health Insurance Program (CHIP) agency to medical providers for services provided to beneficiaries. Most states use a mix of FFS and monthly beneficiary payments to pay for the care provided to Medicaid beneficiaries. In fiscal year 2017, FFS payments accounted for 47 percent of total Medicaid spending nationally on health services. [3]
In order to claim federal matching funds for their medical assistance programs, states report aggregate expenditure data for Medicaid beneficiaries covered under Title XIX of the Social Security Act to CMS through the Medicaid Budget and Expenditure System (MBES). [4] The MBES system is designed to capture information states would report on the Quarterly Medicaid Statement of Expenditures for the Medical Assistance Program Form CMS-64; therefore, the data are commonly referred to as the CMS-64 data. Because of the oversight involved in ensuring the correct federal reimbursement to state Medicaid programs, the CMS-64 data are seen as a highly reliable source of expenditure information. If the expenditures in the TAF do not align with those in the CMS-64, that may suggest TAF data are incomplete or have other data quality errors that render them unrepresentative of actual Medicaid expenditures.
This data quality assessment examines the extent to which FFS expenditures in the TAF data align with the FFS expenditures states report in the CMS-64, overall and for four major categories of service: (1) inpatient hospital services, (2) institutional long-term care services, (3) all other medical services, and (4) prescription drugs. However, benchmarking TAF expenditure data against CMS-64 data is not straightforward. Although the data in both systems are similar, there are some important differences: [5]
Monthly payments that states make on behalf of Medicaid beneficiaries include capitated payments to managed care plans; flat fees paid to a primary care provider for primary care case management plan (PCCM) services; Medicare Part A and Part B premiums for Medicaid beneficiaries who are dually eligible for Medicare; and in some cases, premium assistance for Medicaid beneficiaries with private coverage. More information on monthly beneficiary payment expenditures can be found in the DQ Atlas single topic displays for Total Monthly Beneficiary Payments , CMC Payments , PHP Payments , PCCM Fees , and Premium Assistance Payments .
In addition to FFS claims and monthly beneficiary payments, TAF also includes other types of payments, such as supplemental payment records and service tracking claims. Supplemental payment records represent payments for a specific service made in addition to a capitation payment or negotiated rate. Service tracking claims represent lump-sum payments made for services that cannot be attributed to a specific beneficiary, such as disproportionate share hospital (DSH) payments or payments to providers made under the Upper Payment Limit (UPL) demonstration. Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
Medicaid and CHIP Payment and Access Commission. “EXHIBIT 17. Total Medicaid Benefit Spending by State and Category, FY 2017 (millions).” MACStats: Medicaid and CHIP Data Book December 2018. Washington, DC: MACPAC, December 2018, pp. 52–54. Available at https://www.macpac.gov/wp-content/uploads/2018/12/December-2018-MACStats-Data-Book.pdf . Accessed June 24, 2019.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
Monthly payments that states make on behalf of Medicaid beneficiaries include capitated payments to managed care plans; flat fees paid to a primary care provider for primary care case management plan (PCCM) services; Medicare Part A and Part B premiums for Medicaid beneficiaries who are dually eligible for Medicare; and in some cases, premium assistance for Medicaid beneficiaries with private coverage. More information on monthly beneficiary payment expenditures can be found in the DQ Atlas single topic displays for Total Monthly Beneficiary Payments , CMC Payments , PHP Payments , PCCM Fees , and Premium Assistance Payments .
In addition to FFS claims and monthly beneficiary payments, TAF also includes other types of payments, such as supplemental payment records and service tracking claims. Supplemental payment records represent payments for a specific service made in addition to a capitation payment or negotiated rate. Service tracking claims represent lump-sum payments made for services that cannot be attributed to a specific beneficiary, such as disproportionate share hospital (DSH) payments or payments to providers made under the Upper Payment Limit (UPL) demonstration. Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
Medicaid and CHIP Payment and Access Commission. \u201cEXHIBIT 17. Total Medicaid Benefit Spending by State and Category, FY 2017 (millions).\u201d MACStats: Medicaid and CHIP Data Book December 2018. Washington, DC: MACPAC, December 2018, pp. 52\u201354. Available at https://www.macpac.gov/wp-content/uploads/2018/12/December-2018-MACStats-Data-Book.pdf . Accessed June 24, 2019.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We compared FFS expenditures for Medicaid beneficiaries drawn from twelve months of TAF data [6] to the CMS-64 benchmark for four quarters of the calendar year. [7]
To calculate FFS expenditures in the TAF, we first selected records from the TAF inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims files that represent FFS payments or supplemental wraparound payments, [8] using the claim type code. [9] We did not include TAF records that represent capitation payment records, managed care encounters, and service tracking claims, as well as all payments for S-CHIP beneficiaries, since those expenditures are not contained in the CMS-64 categories we used as the benchmark.
Next, we further restricted the FFS and supplemental payment records to those that matched to an eligibility record in the Demographic and Eligibility (DE) file that indicated that the beneficiary was enrolled in non-CHIP Medicaid in the month that the claim was incurred. [10] This is necessary because the CMS-64 benchmark includes only non-CHIP Medicaid expenditures, and the TAF claim type variable alone cannot distinguish payments made for Medicaid-expansion CHIP enrollees (whose costs are not reported on the CMS-64) from payments made for non-CHIP Medicaid beneficiaries (whose costs are reported on the CMS-64).
Finally, we excluded TAF records that represent DSH payments, [11] supplemental payments made under the UPL demonstration, electronic health record (EHR) payments, drug rebates, and Medicare premium assistance payments. [12] These payments were excluded because they do not represent FFS payments for health services.
Once we had selected the set of TAF records included in the benchmarking analysis, we tabulated expenditures associated with FFS payments in each state using the total Medicaid paid amount on the header record (TOT_MDCD_PD_AMT) for all included records. [13]
We calculated the CMS-64 benchmark for FFS expenditures using four quarterly net expenditures reports from MBES that cover a calendar year. [14] We summed the net expenditures for all the categories of service except those that would not be captured on FFS claims in the TAF, which include capitation payments, premium assistance payments, drug rebates, DSH payments, General Medical Education payments, and other supplemental payments. The categories of service included in the FFS benchmark are shown in Table 1. [15]
Benchmarking by FFS service category
To better understand why total Medicaid expenditures may differ between the TAF and CMS-64, we also examined four major service categories: (1) inpatient hospital, (2) institutional long-term care, (3) all other medical services, and (4) prescription drugs. Because the CMS-64 categories of service do not line up exactly with the organization of the TAF claims files, we used both TAF file types (IP, LT, OT, and RX) and the federally assigned service category (FASC) code and type of service code (TOS_CD) for this mapping, as shown in Table 1. [16] , [17] In some cases, states have known issues with reporting FFS claims into the wrong file, or with assigning a valid type of service code. [18] To present results consistently across states, we did not correct for any known errors in file type or type of service code when tabulating the TAF-based expenditures by category of service.
Table 1. Assigning FFS expenditures to major service categories in the TAF and CMS-64
| Major category | TAF records a | CMS-64 categories |
|---|---|---|
| Inpatient hospital services | All claims with CLM_TYPE_CD=1 and FED_SRVC_CTGRY_CD = 21 (inpatient hospital) except those with TOS_CD values of 058, 084, 090, or those with a TOS_CD value for DSH payment b (123) as the only code on the claim. The excluded codes represent services by religious nonmedical health care institutions, sterilizations, c and critical access hospital services. d Claims with CLM_TYPE_CD=1 in the IP file where FED_SRVC_CTGRY_CD = NULL (missing) and TOS_CD = 001 (inpatient hospital) or NULL (missing). | Category of service 1A. This represents inpatient hospital services. |
| Institutional long-term care services | Claims with CLM_TYPE_CD=1 and FED_SRVC_CTGRY_CD = 22, 23, or 24. These represent payments for nursing facility, intermediate care facility, and all other overnight facility services. Claims with CLM_TYPE_CD=1 from the LT file that did not have claim lines or where FED_SRVC_CTGRY_CD = NULL (missing) and the TOS_CD = 009, 044, 045, 046, 047, 048, 059, or NULL (missing). All records with CLM_TYPE_CD=5 from the LT file that linked to a beneficiary except those with FED_SRVC_CTGRY_CD = 11 or 12, or TOS_CD values of 119, 120, 121, 122, 132, 133, 134, 138, 143, or 144. e The included codes represent wraparound payments for long-term care services. | Categories of service 2A, 3A, 4A, and 4B. These represent payments for mental health facility, nursing facility, and intermediate care facility services. |
| All other medical services | All claims with CLM_TYPE_CD=1 and FED_SRVC_CTGRY_CD = 25, 26, 27, 28, 31, 32, 33, 34, 35, 36, 37, and 38. These represent payments for hospice services, outpatient facility and institutional claims, clinic services, radiology, laboratory, home health services, transportation services, dental services, HCBS services, durable medical equipment and supplies, and physician and all other professional claims. Claims with CLM_TYPE_CD=1 with FED_SRVC_CTGRY_CD = 21 (inpatient hospital) and TOS_CD values of 058, 084, and 090. The included codes represent services by religious nonmedical health care institutions, sterilizations, c and critical access hospital services (inpatient). All records with CLM_TYPE_CD=5 from the IP and OT files that linked to a beneficiary except those with FED_SRVC_CTGRY_CD = 11 or 12, or TOS_CD values of 119, 120, 121, 122, 132 , 133, 134, 138, 143, or 144. e The included codes represent wraparound payments for inpatient and other medical services. Any remaining claims with CLM_TYPE_CD=1 from the LT file that did not indicate institutional care. | Categories of service 2C, 5A, 5C, 5D, 6A, 8, 9A, 10A, 11, 12, 13, 14, 15, 16, 17D, 18D, 19A, 19B, 19C, 19D, 23A, 23B, 24A, 24B, 26, 27, 28, 29A, 30, 31, 32, 33, 34, 34A, 35, 36, 37A, 38, 39, 40, 41, 42, 43, 44, 45, 46B, 47, 48, 49, 69. f These represent payments for Certified Community Behavioral Health Clinics under the Section 1332 demonstration; physician and surgical services; outpatient hospital services; dental services; other practitioners services; clinic services; laboratory and radiological services; home health services; sterilizations; abortions; EPSDT screenings; rural health services; Medicare deductibles and coinsurance; Medicaid coinsurance; home and community-based services; personal care services; case management services; hospice benefits; emergency services for undocumented aliens; Federally-Qualified Health Center services; regular payments for non-emergency medical transportation; physical therapy; occupational therapy; services for speech, hearing, and language; prosthetic devices, dentures, and eyeglasses; diagnostic screening and preventive services; preventive services and vaccines; nurse mid-wife services; emergency hospital services; regular payments for critical access hospital services; nurse practitioner services; school-based services; rehabilitative services (non-school-based); private duty nursing services; freestanding birth center services; health home for enrollees with chronic conditions; tobacco cessation for pregnant women; health home with chronic conditions, opioid use disorder medication-assisted treatment services; COVID-19 vaccines and vaccine administration, qualified community based mobile crisis intervention, health homes for children with medically complex conditions, and other care services. |
| Prescription drugs | Claims with CLM_TYPE_CD=1 from the RX file with FED_SRVC_CTGRY_CD = 41 (Prescription drug). All records with CLM_TYPE_CD=5 from the RX file that linked to a beneficiary except those with FED_SRVC_CTGRY_CD = 11 or 12, or TOS_CD values of 119, 120, 121, 122, 123, 132, 133, 134, 138, 143, 144. e The included codes represent wraparound payments for prescribed drugs. | Category of service 7 and 46. This represents payments for prescribed drugs and drugs for medication-assisted treatment for opioid use disorder. |
a We first exclude TAF records that represent supplemental payments made under the UPL demonstration, Medicare premiums, EHR payments to providers, and drug rebates.
b We include inpatient hospital claims with a TOS CD of 123 if it is not the only TOS CD on the claim. For claims with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the total FFS expenditure.
c We include sterilizations in inpatient hospital expenditures if it is not the only service on an inpatient hospital claim. We include sterilizations in all other medical expenditures if it is the only service on an inpatient hospital claim.
d We exclude critical access hospital services from inpatient hospital expenditures because they are captured in a separate CMS-64 service category―37A Critical Access Hospitals – Regular Payments. We include the expenditures for critical access hospital services that are captured in the TAF and CMS-64 data in the category for all other medical services.
e Type of service codes 138, 143, and 144 are valid starting in 2020.
f CMS-64 category of service 2C is valid starting in 2017. CMS-64 category of service 45 (health home for enrollees with substance abuse disorder) is a valid value starting in 2019. CMS-64 categories of service 46 and 46B (medication-assisted treatment for opioid use disorder) are valid starting in 2020. CMS-64 categories of service 10A and 10B (clinic services), and 47 (COVID-19 vaccines and vaccine administration) are valid starting in 2021. Category of service 10 (clinic services) was separated into categories 10A and 10B starting in 2021 to differentiate regular payments and supplemental payments, respectively. Category of service 29 (non-emergency medical transportation) was separated into categories 29A and 29B starting in 2022 to differentiate regular payments and supplemental payments, respectively. Category of service 37 (critical access hospital services) was separated into categories 37A, 37B, and 37C starting in 2022 to differentiate regular payments, supplemental payments for inpatient services, and supplemental payments for outpatient services, respectively. CMS-64 category of service 48 (qualified community based mobile crisis intervention services) is valid starting in 2022. CMS-64 category of service 49 (health homes for children with medically complex conditions) is valid starting in 2023.
Data quality assessment criteria
We categorized each state’s FFS expenditures in TAF as having high, moderate, low, or very low alignment with the CMS-64 based on the percent difference between the two data sources (Table 2). [19] To help users determine what may be driving differences between the data sources in each state, we also present the alignment for each of the four major service categories (inpatient hospital, long-term care, all other medical services, and prescription drugs).
Table 2. Criteria for DQ assessment of FFS expenditures
| Percent difference between TAF and CMS-64 FFS expenditures | Level of alignment | DQ assessment |
|---|---|---|
| x < 5 percent | High | Low concern |
| 5 percent ≤ x < 20 percent | Moderate | Medium concern |
| 20 percent ≤ x < 50 percent | Low | High concern |
| x ≥ 50 percent or if state has no TAF data | Very low | Unusable |
States with a low absolute percentage difference between the expenditures captured in TAF and the CMS-64 are considered to present a low level of concern about the completeness and quality of their TAF data. However, TAF users should be aware that in states with large Medicaid programs, even a small percentage difference may represent a substantial difference in the dollar amount between the data sources.
There are three reasons other than the quality and completeness of TAF expenditure data that could result in a state’s FFS expenditures in the TAF differing from the CMS-64: (1) structural differences in the data sets, such as organizing the data by service date versus claim payment date; (2) data quality problems in the TAF that are unrelated to expenditures, such as a low volume of claims in the TAF because of missing date values on claims or issues with assigning a final action status to claims; and (3) variation in how states report certain types of FFS expenditures on the CMS-64 that are difficult to reproduce in the TAF data. For instance, a state may make lump-sum reconciliation payments to hospitals as part of their cost-based FFS payments for inpatient services. These payments may be reported in the inpatient hospital base payment category of service in the CMS-64, which is included as part of the FFS benchmark, but report the payments as service tracking claims in T-MSIS, which would not be included in this analysis. When this occurs, we would expect to see a difference between the TAF-based FFS expenditures and the CMS-64 benchmark even if TAF expenditure data are complete and accurate.
Methods previously used to assess data quality
Table 3 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 3. Previously used methods and applicable data years and versions
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
We consider supplemental payments that link to a non-CHIP Medicaid beneficiary (except those with FED_SRVC_CTGRY_CD values 11 or 12, or TOS_CD values of 119, 120, 121, 122, 132, 133, 134, 138, 143, or 144) to be wraparound payments. Type of service codes 138, 143, and 144 are valid starting in 2020. As noted later, we exclude TAF records that represent supplemental payments made under the UPL demonstration and all other lump sum payments.
FFS claims representing Medicaid or M-CHIP benefits have a claim type code value of 1. Supplemental payment claims have a claim type code value of 5.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the CHIP code value was 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-76, we considered the beneficiary to have non-CHIP Medicaid enrollment in that month.
In some states, DSH payment amounts are included on FFS claim headers. For claims with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the total FFS expenditure.
We used type of service code to identify and exclude supplemental payments (132, 133, 134), EHR payments (135) and Medicare premium assistance payments (139, 140, 141, 142). Type of service codes 139, 140, 141, and 142 are valid starting in 2020. We used the federally assigned service category code to identify and exclude DSH payments (13). Drug rebates were identified as records were any line had a type of service code of 131.
In a sensitivity analysis, we found that expenditures based on the TAF header-level payment variable (TOT_MDCD_PD_AMT) alone had higher alignment with CMS-64 expenditures compared to tabulating expenditures using a mix of header- and line-level payments based on the payment level indicator (PYMT_LVL_IND). In part, this may reflect that TAF includes denied line records that are a part of a larger paid claim, and the payment amount on the denied lines may be incorrect.
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
Definitions for each category of service in the CMS-64 can be found on Medicaid.gov: https://www.medicaid.gov/medicaid/downloads/cms-649-base-category-of-services-definition-2-14.pdf
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
Although the TAF includes a data element for states to indicate the CMS-64 category of service (XIX_SRVC_CTGRY_CD) for each claim, this data element has a high rate of missing values in many states and, when it is not missing, it is unclear whether the logic used to assign values is aligned with how states actually classify spending in their CMS-64 reporting.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX .
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: [(TAF – benchmark) / benchmark]*100.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
We consider supplemental payments that link to a non-CHIP Medicaid beneficiary (except those with FED_SRVC_CTGRY_CD values 11 or 12, or TOS_CD values of 119, 120, 121, 122, 132, 133, 134, 138, 143, or 144) to be wraparound payments. Type of service codes 138, 143, and 144 are valid starting in 2020. As noted later, we exclude TAF records that represent supplemental payments made under the UPL demonstration and all other lump sum payments.
FFS claims representing Medicaid or M-CHIP benefits have a claim type code value of 1. Supplemental payment claims have a claim type code value of 5.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the CHIP code value was 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-76, we considered the beneficiary to have non-CHIP Medicaid enrollment in that month.
In some states, DSH payment amounts are included on FFS claim headers. For claims with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the total FFS expenditure.
We used type of service code to identify and exclude supplemental payments (132, 133, 134), EHR payments (135) and Medicare premium assistance payments (139, 140, 141, 142). Type of service codes 139, 140, 141, and 142 are valid starting in 2020. We used the federally assigned service category code to identify and exclude DSH payments (13). Drug rebates were identified as records were any line had a type of service code of 131.
In a sensitivity analysis, we found that expenditures based on the TAF header-level payment variable (TOT_MDCD_PD_AMT) alone had higher alignment with CMS-64 expenditures compared to tabulating expenditures using a mix of header- and line-level payments based on the payment level indicator (PYMT_LVL_IND). In part, this may reflect that TAF includes denied line records that are a part of a larger paid claim, and the payment amount on the denied lines may be incorrect.
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
Definitions for each category of service in the CMS-64 can be found on Medicaid.gov: https://www.medicaid.gov/medicaid/downloads/cms-649-base-category-of-services-definition-2-14.pdf
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
Although the TAF includes a data element for states to indicate the CMS-64 category of service (XIX_SRVC_CTGRY_CD) for each claim, this data element has a high rate of missing values in many states and, when it is not missing, it is unclear whether the logic used to assign values is aligned with how states actually classify spending in their CMS-64 reporting.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX .
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: [(TAF \u2013 benchmark) / benchmark]*100.
Fee-for-service expenditures represent direct payments from the state Medicaid or CHIP agency to medical providers for services rendered to beneficiaries. This analysis examines how well the TAF data on FFS expenditures for institutional long-term care on behalf of Medicaid beneficiaries align with an external benchmark, the CMS-64.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""6021"", ""relatedTopics"": [{""measureId"": 67, ""measureName"": ""Total FFS Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 0}, {""measureId"": 68, ""measureName"": ""FFS Inpatient Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 1}, {""measureId"": 70, ""measureName"": ""FFS Other Medical Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 3}, {""measureId"": 71, ""measureName"": ""FFS Prescription Drug Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 4}]}" 70,"{""measureId"": 70, ""measureName"": ""FFS Other Medical Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-FFS-Other-Expenditures.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) provide a unique source of expenditure information on states’ Medicaid and CHIP programs. Although other data sources provide aggregate state-level expenditure information, the TAF allow users to analyze payments associated with managed care organizations, individual providers, subgroups of beneficiaries, and specific services. The Centers for Medicare & Medicaid Services (CMS) and other TAF users require accurate payment data to calculate total cost of care metrics; to compare costs for specific conditions, services, or populations; and to compare costs across states or different payment systems.
States may reimburse health care providers for services delivered to Medicaid beneficiaries by paying directly for each covered service on a fee-for-service (FFS) basis or by paying a flat monthly payment per beneficiary for a set of services that is contracted to another entity—such as a managed care plan—which then assumes responsibility for delivering care to the beneficiary. [1] , [2] Fee-for-service (FFS) claims represent direct payments from the state Medicaid or Children’s Health Insurance Program (CHIP) agency to medical providers for services provided to beneficiaries. Most states use a mix of FFS and monthly beneficiary payments to pay for the care provided to Medicaid beneficiaries. In fiscal year 2017, FFS payments accounted for 47 percent of total Medicaid spending nationally on health services. [3]
In order to claim federal matching funds for their medical assistance programs, states report aggregate expenditure data for Medicaid beneficiaries covered under Title XIX of the Social Security Act to CMS through the Medicaid Budget and Expenditure System (MBES). [4] The MBES system is designed to capture information states would report on the Quarterly Medicaid Statement of Expenditures for the Medical Assistance Program Form CMS-64; therefore, the data are commonly referred to as the CMS-64 data. Because of the oversight involved in ensuring the correct federal reimbursement to state Medicaid programs, the CMS-64 data are seen as a highly reliable source of expenditure information. If the expenditures in the TAF do not align with those in the CMS-64, that may suggest TAF data are incomplete or have other data quality errors that render them unrepresentative of actual Medicaid expenditures.
This data quality assessment examines the extent to which FFS expenditures in the TAF data align with the FFS expenditures states report in the CMS-64, overall and for four major categories of service: (1) inpatient hospital services, (2) institutional long-term care services, (3) all other medical services, and (4) prescription drugs. However, benchmarking TAF expenditure data against CMS-64 data is not straightforward. Although the data in both systems are similar, there are some important differences: [5]
Monthly payments that states make on behalf of Medicaid beneficiaries include capitated payments to managed care plans; flat fees paid to a primary care provider for primary care case management plan (PCCM) services; Medicare Part A and Part B premiums for Medicaid beneficiaries who are dually eligible for Medicare; and in some cases, premium assistance for Medicaid beneficiaries with private coverage. More information on monthly beneficiary payment expenditures can be found in the DQ Atlas single topic displays for Total Monthly Beneficiary Payments , CMC Payments , PHP Payments , PCCM Fees , and Premium Assistance Payments .
In addition to FFS claims and monthly beneficiary payments, TAF also includes other types of payments, such as supplemental payment records and service tracking claims. Supplemental payment records represent payments for a specific service made in addition to a capitation payment or negotiated rate. Service tracking claims represent lump-sum payments made for services that cannot be attributed to a specific beneficiary, such as disproportionate share hospital (DSH) payments or payments to providers made under the Upper Payment Limit (UPL) demonstration. Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
Medicaid and CHIP Payment and Access Commission. “EXHIBIT 17. Total Medicaid Benefit Spending by State and Category, FY 2017 (millions).” MACStats: Medicaid and CHIP Data Book December 2018. Washington, DC: MACPAC, December 2018, pp. 52–54. Available at https://www.macpac.gov/wp-content/uploads/2018/12/December-2018-MACStats-Data-Book.pdf . Accessed June 24, 2019.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
Monthly payments that states make on behalf of Medicaid beneficiaries include capitated payments to managed care plans; flat fees paid to a primary care provider for primary care case management plan (PCCM) services; Medicare Part A and Part B premiums for Medicaid beneficiaries who are dually eligible for Medicare; and in some cases, premium assistance for Medicaid beneficiaries with private coverage. More information on monthly beneficiary payment expenditures can be found in the DQ Atlas single topic displays for Total Monthly Beneficiary Payments , CMC Payments , PHP Payments , PCCM Fees , and Premium Assistance Payments .
In addition to FFS claims and monthly beneficiary payments, TAF also includes other types of payments, such as supplemental payment records and service tracking claims. Supplemental payment records represent payments for a specific service made in addition to a capitation payment or negotiated rate. Service tracking claims represent lump-sum payments made for services that cannot be attributed to a specific beneficiary, such as disproportionate share hospital (DSH) payments or payments to providers made under the Upper Payment Limit (UPL) demonstration. Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
Medicaid and CHIP Payment and Access Commission. \u201cEXHIBIT 17. Total Medicaid Benefit Spending by State and Category, FY 2017 (millions).\u201d MACStats: Medicaid and CHIP Data Book December 2018. Washington, DC: MACPAC, December 2018, pp. 52\u201354. Available at https://www.macpac.gov/wp-content/uploads/2018/12/December-2018-MACStats-Data-Book.pdf . Accessed June 24, 2019.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We compared FFS expenditures for Medicaid beneficiaries drawn from twelve months of TAF data [6] to the CMS-64 benchmark for four quarters of the calendar year. [7]
To calculate FFS expenditures in the TAF, we first selected records from the TAF inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims files that represent FFS payments or supplemental wraparound payments, [8] using the claim type code. [9] We did not include TAF records that represent capitation payment records, managed care encounters, and service tracking claims, as well as all payments for S-CHIP beneficiaries, since those expenditures are not contained in the CMS-64 categories we used as the benchmark.
Next, we further restricted the FFS and supplemental payment records to those that matched to an eligibility record in the Demographic and Eligibility (DE) file that indicated that the beneficiary was enrolled in non-CHIP Medicaid in the month that the claim was incurred. [10] This is necessary because the CMS-64 benchmark includes only non-CHIP Medicaid expenditures, and the TAF claim type variable alone cannot distinguish payments made for Medicaid-expansion CHIP enrollees (whose costs are not reported on the CMS-64) from payments made for non-CHIP Medicaid beneficiaries (whose costs are reported on the CMS-64).
Finally, we excluded TAF records that represent DSH payments, [11] supplemental payments made under the UPL demonstration, electronic health record (EHR) payments, drug rebates, and Medicare premium assistance payments. [12] These payments were excluded because they do not represent FFS payments for health services.
Once we had selected the set of TAF records included in the benchmarking analysis, we tabulated expenditures associated with FFS payments in each state using the total Medicaid paid amount on the header record (TOT_MDCD_PD_AMT) for all included records. [13]
We calculated the CMS-64 benchmark for FFS expenditures using four quarterly net expenditures reports from MBES that cover a calendar year. [14] We summed the net expenditures for all the categories of service except those that would not be captured on FFS claims in the TAF, which include capitation payments, premium assistance payments, drug rebates, DSH payments, General Medical Education payments, and other supplemental payments. The categories of service included in the FFS benchmark are shown in Table 1. [15]
Benchmarking by FFS service category
To better understand why total Medicaid expenditures may differ between the TAF and CMS-64, we also examined four major service categories: (1) inpatient hospital, (2) institutional long-term care, (3) all other medical services, and (4) prescription drugs. Because the CMS-64 categories of service do not line up exactly with the organization of the TAF claims files, we used both TAF file types (IP, LT, OT, and RX) and the federally assigned service category (FASC) code and type of service code (TOS_CD) for this mapping, as shown in Table 1. [16] , [17] In some cases, states have known issues with reporting FFS claims into the wrong file, or with assigning a valid type of service code. [18] To present results consistently across states, we did not correct for any known errors in file type or type of service code when tabulating the TAF-based expenditures by category of service.
Table 1. Assigning FFS expenditures to major service categories in the TAF and CMS-64
| Major category | TAF records a | CMS-64 categories |
|---|---|---|
| Inpatient hospital services | All claims with CLM_TYPE_CD=1 and FED_SRVC_CTGRY_CD = 21 (inpatient hospital) except those with TOS_CD values of 058, 084, 090, or those with a TOS_CD value for DSH payment b (123) as the only code on the claim. The excluded codes represent services by religious nonmedical health care institutions, sterilizations, c and critical access hospital services. d Claims with CLM_TYPE_CD=1 in the IP file where FED_SRVC_CTGRY_CD = NULL (missing) and TOS_CD = 001 (inpatient hospital) or NULL (missing). | Category of service 1A. This represents inpatient hospital services. |
| Institutional long-term care services | Claims with CLM_TYPE_CD=1 and FED_SRVC_CTGRY_CD = 22, 23, or 24. These represent payments for nursing facility, intermediate care facility, and all other overnight facility services. Claims with CLM_TYPE_CD=1 from the LT file that did not have claim lines or where FED_SRVC_CTGRY_CD = NULL (missing) and the TOS_CD = 009, 044, 045, 046, 047, 048, 059, or NULL (missing). All records with CLM_TYPE_CD=5 from the LT file that linked to a beneficiary except those with FED_SRVC_CTGRY_CD = 11 or 12, or TOS_CD values of 119, 120, 121, 122, 132, 133, 134, 138, 143, or 144. e The included codes represent wraparound payments for long-term care services. | Categories of service 2A, 3A, 4A, and 4B. These represent payments for mental health facility, nursing facility, and intermediate care facility services. |
| All other medical services | All claims with CLM_TYPE_CD=1 and FED_SRVC_CTGRY_CD = 25, 26, 27, 28, 31, 32, 33, 34, 35, 36, 37, and 38. These represent payments for hospice services, outpatient facility and institutional claims, clinic services, radiology, laboratory, home health services, transportation services, dental services, HCBS services, durable medical equipment and supplies, and physician and all other professional claims. Claims with CLM_TYPE_CD=1 with FED_SRVC_CTGRY_CD = 21 (inpatient hospital) and TOS_CD values of 058, 084, and 090. The included codes represent services by religious nonmedical health care institutions, sterilizations, c and critical access hospital services (inpatient). All records with CLM_TYPE_CD=5 from the IP and OT files that linked to a beneficiary except those with FED_SRVC_CTGRY_CD = 11 or 12, or TOS_CD values of 119, 120, 121, 122, 132 , 133, 134, 138, 143, or 144. e The included codes represent wraparound payments for inpatient and other medical services. Any remaining claims with CLM_TYPE_CD=1 from the LT file that did not indicate institutional care. | Categories of service 2C, 5A, 5C, 5D, 6A, 8, 9A, 10A, 11, 12, 13, 14, 15, 16, 17D, 18D, 19A, 19B, 19C, 19D, 23A, 23B, 24A, 24B, 26, 27, 28, 29A, 30, 31, 32, 33, 34, 34A, 35, 36, 37A, 38, 39, 40, 41, 42, 43, 44, 45, 46B, 47, 48, 49, 69. f These represent payments for Certified Community Behavioral Health Clinics under the Section 1332 demonstration; physician and surgical services; outpatient hospital services; dental services; other practitioners services; clinic services; laboratory and radiological services; home health services; sterilizations; abortions; EPSDT screenings; rural health services; Medicare deductibles and coinsurance; Medicaid coinsurance; home and community-based services; personal care services; case management services; hospice benefits; emergency services for undocumented aliens; Federally-Qualified Health Center services; regular payments for non-emergency medical transportation; physical therapy; occupational therapy; services for speech, hearing, and language; prosthetic devices, dentures, and eyeglasses; diagnostic screening and preventive services; preventive services and vaccines; nurse mid-wife services; emergency hospital services; regular payments for critical access hospital services; nurse practitioner services; school-based services; rehabilitative services (non-school-based); private duty nursing services; freestanding birth center services; health home for enrollees with chronic conditions; tobacco cessation for pregnant women; health home with chronic conditions, opioid use disorder medication-assisted treatment services; COVID-19 vaccines and vaccine administration, qualified community based mobile crisis intervention, health homes for children with medically complex conditions, and other care services. |
| Prescription drugs | Claims with CLM_TYPE_CD=1 from the RX file with FED_SRVC_CTGRY_CD = 41 (Prescription drug). All records with CLM_TYPE_CD=5 from the RX file that linked to a beneficiary except those with FED_SRVC_CTGRY_CD = 11 or 12, or TOS_CD values of 119, 120, 121, 122, 123, 132, 133, 134, 138, 143, 144. e The included codes represent wraparound payments for prescribed drugs. | Category of service 7 and 46. This represents payments for prescribed drugs and drugs for medication-assisted treatment for opioid use disorder. |
a We first exclude TAF records that represent supplemental payments made under the UPL demonstration, Medicare premiums, EHR payments to providers, and drug rebates.
b We include inpatient hospital claims with a TOS CD of 123 if it is not the only TOS CD on the claim. For claims with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the total FFS expenditure.
c We include sterilizations in inpatient hospital expenditures if it is not the only service on an inpatient hospital claim. We include sterilizations in all other medical expenditures if it is the only service on an inpatient hospital claim.
d We exclude critical access hospital services from inpatient hospital expenditures because they are captured in a separate CMS-64 service category―37A Critical Access Hospitals – Regular Payments. We include the expenditures for critical access hospital services that are captured in the TAF and CMS-64 data in the category for all other medical services.
e Type of service codes 138, 143, and 144 are valid starting in 2020.
f CMS-64 category of service 2C is valid starting in 2017. CMS-64 category of service 45 (health home for enrollees with substance abuse disorder) is a valid value starting in 2019. CMS-64 categories of service 46 and 46B (medication-assisted treatment for opioid use disorder) are valid starting in 2020. CMS-64 categories of service 10A and 10B (clinic services), and 47 (COVID-19 vaccines and vaccine administration) are valid starting in 2021. Category of service 10 (clinic services) was separated into categories 10A and 10B starting in 2021 to differentiate regular payments and supplemental payments, respectively. Category of service 29 (non-emergency medical transportation) was separated into categories 29A and 29B starting in 2022 to differentiate regular payments and supplemental payments, respectively. Category of service 37 (critical access hospital services) was separated into categories 37A, 37B, and 37C starting in 2022 to differentiate regular payments, supplemental payments for inpatient services, and supplemental payments for outpatient services, respectively. CMS-64 category of service 48 (qualified community based mobile crisis intervention services) is valid starting in 2022. CMS-64 category of service 49 (health homes for children with medically complex conditions) is valid starting in 2023.
Data quality assessment criteria
We categorized each state’s FFS expenditures in TAF as having high, moderate, low, or very low alignment with the CMS-64 based on the percent difference between the two data sources (Table 2). [19] To help users determine what may be driving differences between the data sources in each state, we also present the alignment for each of the four major service categories (inpatient hospital, long-term care, all other medical services, and prescription drugs).
Table 2. Criteria for DQ assessment of FFS expenditures
| Percent difference between TAF and CMS-64 FFS expenditures | Level of alignment | DQ assessment |
|---|---|---|
| x < 5 percent | High | Low concern |
| 5 percent ≤ x < 20 percent | Moderate | Medium concern |
| 20 percent ≤ x < 50 percent | Low | High concern |
| x ≥ 50 percent or if state has no TAF data | Very low | Unusable |
States with a low absolute percentage difference between the expenditures captured in TAF and the CMS-64 are considered to present a low level of concern about the completeness and quality of their TAF data. However, TAF users should be aware that in states with large Medicaid programs, even a small percentage difference may represent a substantial difference in the dollar amount between the data sources.
There are three reasons other than the quality and completeness of TAF expenditure data that could result in a state’s FFS expenditures in the TAF differing from the CMS-64: (1) structural differences in the data sets, such as organizing the data by service date versus claim payment date; (2) data quality problems in the TAF that are unrelated to expenditures, such as a low volume of claims in the TAF because of missing date values on claims or issues with assigning a final action status to claims; and (3) variation in how states report certain types of FFS expenditures on the CMS-64 that are difficult to reproduce in the TAF data. For instance, a state may make lump-sum reconciliation payments to hospitals as part of their cost-based FFS payments for inpatient services. These payments may be reported in the inpatient hospital base payment category of service in the CMS-64, which is included as part of the FFS benchmark, but report the payments as service tracking claims in T-MSIS, which would not be included in this analysis. When this occurs, we would expect to see a difference between the TAF-based FFS expenditures and the CMS-64 benchmark even if TAF expenditure data are complete and accurate.
Methods previously used to assess data quality
Table 3 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 3. Previously used methods and applicable data years and versions
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
We consider supplemental payments that link to a non-CHIP Medicaid beneficiary (except those with FED_SRVC_CTGRY_CD values 11 or 12, or TOS_CD values of 119, 120, 121, 122, 132, 133, 134, 138, 143, or 144) to be wraparound payments. Type of service codes 138, 143, and 144 are valid starting in 2020. As noted later, we exclude TAF records that represent supplemental payments made under the UPL demonstration and all other lump sum payments.
FFS claims representing Medicaid or M-CHIP benefits have a claim type code value of 1. Supplemental payment claims have a claim type code value of 5.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the CHIP code value was 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-76, we considered the beneficiary to have non-CHIP Medicaid enrollment in that month.
In some states, DSH payment amounts are included on FFS claim headers. For claims with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the total FFS expenditure.
We used type of service code to identify and exclude supplemental payments (132, 133, 134), EHR payments (135) and Medicare premium assistance payments (139, 140, 141, 142). Type of service codes 139, 140, 141, and 142 are valid starting in 2020. We used the federally assigned service category code to identify and exclude DSH payments (13). Drug rebates were identified as records were any line had a type of service code of 131.
In a sensitivity analysis, we found that expenditures based on the TAF header-level payment variable (TOT_MDCD_PD_AMT) alone had higher alignment with CMS-64 expenditures compared to tabulating expenditures using a mix of header- and line-level payments based on the payment level indicator (PYMT_LVL_IND). In part, this may reflect that TAF includes denied line records that are a part of a larger paid claim, and the payment amount on the denied lines may be incorrect.
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
Definitions for each category of service in the CMS-64 can be found on Medicaid.gov: https://www.medicaid.gov/medicaid/downloads/cms-649-base-category-of-services-definition-2-14.pdf
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
Although the TAF includes a data element for states to indicate the CMS-64 category of service (XIX_SRVC_CTGRY_CD) for each claim, this data element has a high rate of missing values in many states and, when it is not missing, it is unclear whether the logic used to assign values is aligned with how states actually classify spending in their CMS-64 reporting.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX .
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: [(TAF – benchmark) / benchmark]*100.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
We consider supplemental payments that link to a non-CHIP Medicaid beneficiary (except those with FED_SRVC_CTGRY_CD values 11 or 12, or TOS_CD values of 119, 120, 121, 122, 132, 133, 134, 138, 143, or 144) to be wraparound payments. Type of service codes 138, 143, and 144 are valid starting in 2020. As noted later, we exclude TAF records that represent supplemental payments made under the UPL demonstration and all other lump sum payments.
FFS claims representing Medicaid or M-CHIP benefits have a claim type code value of 1. Supplemental payment claims have a claim type code value of 5.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the CHIP code value was 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-76, we considered the beneficiary to have non-CHIP Medicaid enrollment in that month.
In some states, DSH payment amounts are included on FFS claim headers. For claims with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the total FFS expenditure.
We used type of service code to identify and exclude supplemental payments (132, 133, 134), EHR payments (135) and Medicare premium assistance payments (139, 140, 141, 142). Type of service codes 139, 140, 141, and 142 are valid starting in 2020. We used the federally assigned service category code to identify and exclude DSH payments (13). Drug rebates were identified as records were any line had a type of service code of 131.
In a sensitivity analysis, we found that expenditures based on the TAF header-level payment variable (TOT_MDCD_PD_AMT) alone had higher alignment with CMS-64 expenditures compared to tabulating expenditures using a mix of header- and line-level payments based on the payment level indicator (PYMT_LVL_IND). In part, this may reflect that TAF includes denied line records that are a part of a larger paid claim, and the payment amount on the denied lines may be incorrect.
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
Definitions for each category of service in the CMS-64 can be found on Medicaid.gov: https://www.medicaid.gov/medicaid/downloads/cms-649-base-category-of-services-definition-2-14.pdf
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
Although the TAF includes a data element for states to indicate the CMS-64 category of service (XIX_SRVC_CTGRY_CD) for each claim, this data element has a high rate of missing values in many states and, when it is not missing, it is unclear whether the logic used to assign values is aligned with how states actually classify spending in their CMS-64 reporting.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX .
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: [(TAF \u2013 benchmark) / benchmark]*100.
Fee-for-service expenditures represent direct payments from the state Medicaid or CHIP agency to medical providers for services rendered to beneficiaries. This analysis examines how well the TAF data on FFS expenditures for outpatient facility services and all other professional services on behalf of Medicaid beneficiaries align with an external benchmark, the CMS-64.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""6021"", ""relatedTopics"": [{""measureId"": 67, ""measureName"": ""Total FFS Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 0}, {""measureId"": 68, ""measureName"": ""FFS Inpatient Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 1}, {""measureId"": 69, ""measureName"": ""FFS Long-Term Care Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 2}, {""measureId"": 71, ""measureName"": ""FFS Prescription Drug Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 4}]}" 71,"{""measureId"": 71, ""measureName"": ""FFS Prescription Drug Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-FFS-RX-Expenditures.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) provide a unique source of expenditure information on states’ Medicaid and CHIP programs. Although other data sources provide aggregate state-level expenditure information, the TAF allow users to analyze payments associated with managed care organizations, individual providers, subgroups of beneficiaries, and specific services. The Centers for Medicare & Medicaid Services (CMS) and other TAF users require accurate payment data to calculate total cost of care metrics; to compare costs for specific conditions, services, or populations; and to compare costs across states or different payment systems.
States may reimburse health care providers for services delivered to Medicaid beneficiaries by paying directly for each covered service on a fee-for-service (FFS) basis or by paying a flat monthly payment per beneficiary for a set of services that is contracted to another entity—such as a managed care plan—which then assumes responsibility for delivering care to the beneficiary. [1] , [2] Fee-for-service (FFS) claims represent direct payments from the state Medicaid or Children’s Health Insurance Program (CHIP) agency to medical providers for services provided to beneficiaries. Most states use a mix of FFS and monthly beneficiary payments to pay for the care provided to Medicaid beneficiaries. In fiscal year 2017, FFS payments accounted for 47 percent of total Medicaid spending nationally on health services. [3]
In order to claim federal matching funds for their medical assistance programs, states report aggregate expenditure data for Medicaid beneficiaries covered under Title XIX of the Social Security Act to CMS through the Medicaid Budget and Expenditure System (MBES). [4] The MBES system is designed to capture information states would report on the Quarterly Medicaid Statement of Expenditures for the Medical Assistance Program Form CMS-64; therefore, the data are commonly referred to as the CMS-64 data. Because of the oversight involved in ensuring the correct federal reimbursement to state Medicaid programs, the CMS-64 data are seen as a highly reliable source of expenditure information. If the expenditures in the TAF do not align with those in the CMS-64, that may suggest TAF data are incomplete or have other data quality errors that render them unrepresentative of actual Medicaid expenditures.
This data quality assessment examines the extent to which FFS expenditures in the TAF data align with the FFS expenditures states report in the CMS-64, overall and for four major categories of service: (1) inpatient hospital services, (2) institutional long-term care services, (3) all other medical services, and (4) prescription drugs. However, benchmarking TAF expenditure data against CMS-64 data is not straightforward. Although the data in both systems are similar, there are some important differences: [5]
Monthly payments that states make on behalf of Medicaid beneficiaries include capitated payments to managed care plans; flat fees paid to a primary care provider for primary care case management plan (PCCM) services; Medicare Part A and Part B premiums for Medicaid beneficiaries who are dually eligible for Medicare; and in some cases, premium assistance for Medicaid beneficiaries with private coverage. More information on monthly beneficiary payment expenditures can be found in the DQ Atlas single topic displays for Total Monthly Beneficiary Payments , CMC Payments , PHP Payments , PCCM Fees , and Premium Assistance Payments .
In addition to FFS claims and monthly beneficiary payments, TAF also includes other types of payments, such as supplemental payment records and service tracking claims. Supplemental payment records represent payments for a specific service made in addition to a capitation payment or negotiated rate. Service tracking claims represent lump-sum payments made for services that cannot be attributed to a specific beneficiary, such as disproportionate share hospital (DSH) payments or payments to providers made under the Upper Payment Limit (UPL) demonstration. Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
Medicaid and CHIP Payment and Access Commission. “EXHIBIT 17. Total Medicaid Benefit Spending by State and Category, FY 2017 (millions).” MACStats: Medicaid and CHIP Data Book December 2018. Washington, DC: MACPAC, December 2018, pp. 52–54. Available at https://www.macpac.gov/wp-content/uploads/2018/12/December-2018-MACStats-Data-Book.pdf . Accessed June 24, 2019.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
Monthly payments that states make on behalf of Medicaid beneficiaries include capitated payments to managed care plans; flat fees paid to a primary care provider for primary care case management plan (PCCM) services; Medicare Part A and Part B premiums for Medicaid beneficiaries who are dually eligible for Medicare; and in some cases, premium assistance for Medicaid beneficiaries with private coverage. More information on monthly beneficiary payment expenditures can be found in the DQ Atlas single topic displays for Total Monthly Beneficiary Payments , CMC Payments , PHP Payments , PCCM Fees , and Premium Assistance Payments .
In addition to FFS claims and monthly beneficiary payments, TAF also includes other types of payments, such as supplemental payment records and service tracking claims. Supplemental payment records represent payments for a specific service made in addition to a capitation payment or negotiated rate. Service tracking claims represent lump-sum payments made for services that cannot be attributed to a specific beneficiary, such as disproportionate share hospital (DSH) payments or payments to providers made under the Upper Payment Limit (UPL) demonstration. Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
Medicaid and CHIP Payment and Access Commission. \u201cEXHIBIT 17. Total Medicaid Benefit Spending by State and Category, FY 2017 (millions).\u201d MACStats: Medicaid and CHIP Data Book December 2018. Washington, DC: MACPAC, December 2018, pp. 52\u201354. Available at https://www.macpac.gov/wp-content/uploads/2018/12/December-2018-MACStats-Data-Book.pdf . Accessed June 24, 2019.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We compared FFS expenditures for Medicaid beneficiaries drawn from twelve months of TAF data [6] to the CMS-64 benchmark for four quarters of the calendar year. [7]
To calculate FFS expenditures in the TAF, we first selected records from the TAF inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims files that represent FFS payments or supplemental wraparound payments, [8] using the claim type code. [9] We did not include TAF records that represent capitation payment records, managed care encounters, and service tracking claims, as well as all payments for S-CHIP beneficiaries, since those expenditures are not contained in the CMS-64 categories we used as the benchmark.
Next, we further restricted the FFS and supplemental payment records to those that matched to an eligibility record in the Demographic and Eligibility (DE) file that indicated that the beneficiary was enrolled in non-CHIP Medicaid in the month that the claim was incurred. [10] This is necessary because the CMS-64 benchmark includes only non-CHIP Medicaid expenditures, and the TAF claim type variable alone cannot distinguish payments made for Medicaid-expansion CHIP enrollees (whose costs are not reported on the CMS-64) from payments made for non-CHIP Medicaid beneficiaries (whose costs are reported on the CMS-64).
Finally, we excluded TAF records that represent DSH payments, [11] supplemental payments made under the UPL demonstration, electronic health record (EHR) payments, drug rebates, and Medicare premium assistance payments. [12] These payments were excluded because they do not represent FFS payments for health services.
Once we had selected the set of TAF records included in the benchmarking analysis, we tabulated expenditures associated with FFS payments in each state using the total Medicaid paid amount on the header record (TOT_MDCD_PD_AMT) for all included records. [13]
We calculated the CMS-64 benchmark for FFS expenditures using four quarterly net expenditures reports from MBES that cover a calendar year. [14] We summed the net expenditures for all the categories of service except those that would not be captured on FFS claims in the TAF, which include capitation payments, premium assistance payments, drug rebates, DSH payments, General Medical Education payments, and other supplemental payments. The categories of service included in the FFS benchmark are shown in Table 1. [15]
Benchmarking by FFS service category
To better understand why total Medicaid expenditures may differ between the TAF and CMS-64, we also examined four major service categories: (1) inpatient hospital, (2) institutional long-term care, (3) all other medical services, and (4) prescription drugs. Because the CMS-64 categories of service do not line up exactly with the organization of the TAF claims files, we used both TAF file types (IP, LT, OT, and RX) and the federally assigned service category (FASC) code and type of service code (TOS_CD) for this mapping, as shown in Table 1. [16] , [17] In some cases, states have known issues with reporting FFS claims into the wrong file, or with assigning a valid type of service code. [18] To present results consistently across states, we did not correct for any known errors in file type or type of service code when tabulating the TAF-based expenditures by category of service.
Table 1. Assigning FFS expenditures to major service categories in the TAF and CMS-64
| Major category | TAF records a | CMS-64 categories |
|---|---|---|
| Inpatient hospital services | All claims with CLM_TYPE_CD=1 and FED_SRVC_CTGRY_CD = 21 (inpatient hospital) except those with TOS_CD values of 058, 084, 090, or those with a TOS_CD value for DSH payment b (123) as the only code on the claim. The excluded codes represent services by religious nonmedical health care institutions, sterilizations, c and critical access hospital services. d Claims with CLM_TYPE_CD=1 in the IP file where FED_SRVC_CTGRY_CD = NULL (missing) and TOS_CD = 001 (inpatient hospital) or NULL (missing). | Category of service 1A. This represents inpatient hospital services. |
| Institutional long-term care services | Claims with CLM_TYPE_CD=1 and FED_SRVC_CTGRY_CD = 22, 23, or 24. These represent payments for nursing facility, intermediate care facility, and all other overnight facility services. Claims with CLM_TYPE_CD=1 from the LT file that did not have claim lines or where FED_SRVC_CTGRY_CD = NULL (missing) and the TOS_CD = 009, 044, 045, 046, 047, 048, 059, or NULL (missing). All records with CLM_TYPE_CD=5 from the LT file that linked to a beneficiary except those with FED_SRVC_CTGRY_CD = 11 or 12, or TOS_CD values of 119, 120, 121, 122, 132, 133, 134, 138, 143, or 144. e The included codes represent wraparound payments for long-term care services. | Categories of service 2A, 3A, 4A, and 4B. These represent payments for mental health facility, nursing facility, and intermediate care facility services. |
| All other medical services | All claims with CLM_TYPE_CD=1 and FED_SRVC_CTGRY_CD = 25, 26, 27, 28, 31, 32, 33, 34, 35, 36, 37, and 38. These represent payments for hospice services, outpatient facility and institutional claims, clinic services, radiology, laboratory, home health services, transportation services, dental services, HCBS services, durable medical equipment and supplies, and physician and all other professional claims. Claims with CLM_TYPE_CD=1 with FED_SRVC_CTGRY_CD = 21 (inpatient hospital) and TOS_CD values of 058, 084, and 090. The included codes represent services by religious nonmedical health care institutions, sterilizations, c and critical access hospital services (inpatient). All records with CLM_TYPE_CD=5 from the IP and OT files that linked to a beneficiary except those with FED_SRVC_CTGRY_CD = 11 or 12, or TOS_CD values of 119, 120, 121, 122, 132 , 133, 134, 138, 143, or 144. e The included codes represent wraparound payments for inpatient and other medical services. Any remaining claims with CLM_TYPE_CD=1 from the LT file that did not indicate institutional care. | Categories of service 2C, 5A, 5C, 5D, 6A, 8, 9A, 10A, 11, 12, 13, 14, 15, 16, 17D, 18D, 19A, 19B, 19C, 19D, 23A, 23B, 24A, 24B, 26, 27, 28, 29A, 30, 31, 32, 33, 34, 34A, 35, 36, 37A, 38, 39, 40, 41, 42, 43, 44, 45, 46B, 47, 48, 49, 69. f These represent payments for Certified Community Behavioral Health Clinics under the Section 1332 demonstration; physician and surgical services; outpatient hospital services; dental services; other practitioners services; clinic services; laboratory and radiological services; home health services; sterilizations; abortions; EPSDT screenings; rural health services; Medicare deductibles and coinsurance; Medicaid coinsurance; home and community-based services; personal care services; case management services; hospice benefits; emergency services for undocumented aliens; Federally-Qualified Health Center services; regular payments for non-emergency medical transportation; physical therapy; occupational therapy; services for speech, hearing, and language; prosthetic devices, dentures, and eyeglasses; diagnostic screening and preventive services; preventive services and vaccines; nurse mid-wife services; emergency hospital services; regular payments for critical access hospital services; nurse practitioner services; school-based services; rehabilitative services (non-school-based); private duty nursing services; freestanding birth center services; health home for enrollees with chronic conditions; tobacco cessation for pregnant women; health home with chronic conditions, opioid use disorder medication-assisted treatment services; COVID-19 vaccines and vaccine administration, qualified community based mobile crisis intervention, health homes for children with medically complex conditions, and other care services. |
| Prescription drugs | Claims with CLM_TYPE_CD=1 from the RX file with FED_SRVC_CTGRY_CD = 41 (Prescription drug). All records with CLM_TYPE_CD=5 from the RX file that linked to a beneficiary except those with FED_SRVC_CTGRY_CD = 11 or 12, or TOS_CD values of 119, 120, 121, 122, 123, 132, 133, 134, 138, 143, 144. e The included codes represent wraparound payments for prescribed drugs. | Category of service 7 and 46. This represents payments for prescribed drugs and drugs for medication-assisted treatment for opioid use disorder. |
a We first exclude TAF records that represent supplemental payments made under the UPL demonstration, Medicare premiums, EHR payments to providers, and drug rebates.
b We include inpatient hospital claims with a TOS CD of 123 if it is not the only TOS CD on the claim. For claims with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the total FFS expenditure.
c We include sterilizations in inpatient hospital expenditures if it is not the only service on an inpatient hospital claim. We include sterilizations in all other medical expenditures if it is the only service on an inpatient hospital claim.
d We exclude critical access hospital services from inpatient hospital expenditures because they are captured in a separate CMS-64 service category―37A Critical Access Hospitals – Regular Payments. We include the expenditures for critical access hospital services that are captured in the TAF and CMS-64 data in the category for all other medical services.
e Type of service codes 138, 143, and 144 are valid starting in 2020.
f CMS-64 category of service 2C is valid starting in 2017. CMS-64 category of service 45 (health home for enrollees with substance abuse disorder) is a valid value starting in 2019. CMS-64 categories of service 46 and 46B (medication-assisted treatment for opioid use disorder) are valid starting in 2020. CMS-64 categories of service 10A and 10B (clinic services), and 47 (COVID-19 vaccines and vaccine administration) are valid starting in 2021. Category of service 10 (clinic services) was separated into categories 10A and 10B starting in 2021 to differentiate regular payments and supplemental payments, respectively. Category of service 29 (non-emergency medical transportation) was separated into categories 29A and 29B starting in 2022 to differentiate regular payments and supplemental payments, respectively. Category of service 37 (critical access hospital services) was separated into categories 37A, 37B, and 37C starting in 2022 to differentiate regular payments, supplemental payments for inpatient services, and supplemental payments for outpatient services, respectively. CMS-64 category of service 48 (qualified community based mobile crisis intervention services) is valid starting in 2022. CMS-64 category of service 49 (health homes for children with medically complex conditions) is valid starting in 2023.
Data quality assessment criteria
We categorized each state’s FFS expenditures in TAF as having high, moderate, low, or very low alignment with the CMS-64 based on the percent difference between the two data sources (Table 2). [19] To help users determine what may be driving differences between the data sources in each state, we also present the alignment for each of the four major service categories (inpatient hospital, long-term care, all other medical services, and prescription drugs).
Table 2. Criteria for DQ assessment of FFS expenditures
| Percent difference between TAF and CMS-64 FFS expenditures | Level of alignment | DQ assessment |
|---|---|---|
| x < 5 percent | High | Low concern |
| 5 percent ≤ x < 20 percent | Moderate | Medium concern |
| 20 percent ≤ x < 50 percent | Low | High concern |
| x ≥ 50 percent or if state has no TAF data | Very low | Unusable |
States with a low absolute percentage difference between the expenditures captured in TAF and the CMS-64 are considered to present a low level of concern about the completeness and quality of their TAF data. However, TAF users should be aware that in states with large Medicaid programs, even a small percentage difference may represent a substantial difference in the dollar amount between the data sources.
There are three reasons other than the quality and completeness of TAF expenditure data that could result in a state’s FFS expenditures in the TAF differing from the CMS-64: (1) structural differences in the data sets, such as organizing the data by service date versus claim payment date; (2) data quality problems in the TAF that are unrelated to expenditures, such as a low volume of claims in the TAF because of missing date values on claims or issues with assigning a final action status to claims; and (3) variation in how states report certain types of FFS expenditures on the CMS-64 that are difficult to reproduce in the TAF data. For instance, a state may make lump-sum reconciliation payments to hospitals as part of their cost-based FFS payments for inpatient services. These payments may be reported in the inpatient hospital base payment category of service in the CMS-64, which is included as part of the FFS benchmark, but report the payments as service tracking claims in T-MSIS, which would not be included in this analysis. When this occurs, we would expect to see a difference between the TAF-based FFS expenditures and the CMS-64 benchmark even if TAF expenditure data are complete and accurate.
Methods previously used to assess data quality
Table 3 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 3. Previously used methods and applicable data years and versions
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
We consider supplemental payments that link to a non-CHIP Medicaid beneficiary (except those with FED_SRVC_CTGRY_CD values 11 or 12, or TOS_CD values of 119, 120, 121, 122, 132, 133, 134, 138, 143, or 144) to be wraparound payments. Type of service codes 138, 143, and 144 are valid starting in 2020. As noted later, we exclude TAF records that represent supplemental payments made under the UPL demonstration and all other lump sum payments.
FFS claims representing Medicaid or M-CHIP benefits have a claim type code value of 1. Supplemental payment claims have a claim type code value of 5.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the CHIP code value was 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-76, we considered the beneficiary to have non-CHIP Medicaid enrollment in that month.
In some states, DSH payment amounts are included on FFS claim headers. For claims with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the total FFS expenditure.
We used type of service code to identify and exclude supplemental payments (132, 133, 134), EHR payments (135) and Medicare premium assistance payments (139, 140, 141, 142). Type of service codes 139, 140, 141, and 142 are valid starting in 2020. We used the federally assigned service category code to identify and exclude DSH payments (13). Drug rebates were identified as records were any line had a type of service code of 131.
In a sensitivity analysis, we found that expenditures based on the TAF header-level payment variable (TOT_MDCD_PD_AMT) alone had higher alignment with CMS-64 expenditures compared to tabulating expenditures using a mix of header- and line-level payments based on the payment level indicator (PYMT_LVL_IND). In part, this may reflect that TAF includes denied line records that are a part of a larger paid claim, and the payment amount on the denied lines may be incorrect.
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
Definitions for each category of service in the CMS-64 can be found on Medicaid.gov: https://www.medicaid.gov/medicaid/downloads/cms-649-base-category-of-services-definition-2-14.pdf
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
Although the TAF includes a data element for states to indicate the CMS-64 category of service (XIX_SRVC_CTGRY_CD) for each claim, this data element has a high rate of missing values in many states and, when it is not missing, it is unclear whether the logic used to assign values is aligned with how states actually classify spending in their CMS-64 reporting.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX .
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: [(TAF – benchmark) / benchmark]*100.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
We consider supplemental payments that link to a non-CHIP Medicaid beneficiary (except those with FED_SRVC_CTGRY_CD values 11 or 12, or TOS_CD values of 119, 120, 121, 122, 132, 133, 134, 138, 143, or 144) to be wraparound payments. Type of service codes 138, 143, and 144 are valid starting in 2020. As noted later, we exclude TAF records that represent supplemental payments made under the UPL demonstration and all other lump sum payments.
FFS claims representing Medicaid or M-CHIP benefits have a claim type code value of 1. Supplemental payment claims have a claim type code value of 5.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the CHIP code value was 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-76, we considered the beneficiary to have non-CHIP Medicaid enrollment in that month.
In some states, DSH payment amounts are included on FFS claim headers. For claims with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the total FFS expenditure.
We used type of service code to identify and exclude supplemental payments (132, 133, 134), EHR payments (135) and Medicare premium assistance payments (139, 140, 141, 142). Type of service codes 139, 140, 141, and 142 are valid starting in 2020. We used the federally assigned service category code to identify and exclude DSH payments (13). Drug rebates were identified as records were any line had a type of service code of 131.
In a sensitivity analysis, we found that expenditures based on the TAF header-level payment variable (TOT_MDCD_PD_AMT) alone had higher alignment with CMS-64 expenditures compared to tabulating expenditures using a mix of header- and line-level payments based on the payment level indicator (PYMT_LVL_IND). In part, this may reflect that TAF includes denied line records that are a part of a larger paid claim, and the payment amount on the denied lines may be incorrect.
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
Definitions for each category of service in the CMS-64 can be found on Medicaid.gov: https://www.medicaid.gov/medicaid/downloads/cms-649-base-category-of-services-definition-2-14.pdf
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
Although the TAF includes a data element for states to indicate the CMS-64 category of service (XIX_SRVC_CTGRY_CD) for each claim, this data element has a high rate of missing values in many states and, when it is not missing, it is unclear whether the logic used to assign values is aligned with how states actually classify spending in their CMS-64 reporting.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX .
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: [(TAF \u2013 benchmark) / benchmark]*100.
Fee-for-service expenditures represent direct payments from the state Medicaid or CHIP agency to medical providers for services rendered to beneficiaries. This analysis examines how well the TAF data on FFS expenditures on prescription drugs for Medicaid beneficiaries align with an external benchmark, the CMS-64.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""6021"", ""relatedTopics"": [{""measureId"": 67, ""measureName"": ""Total FFS Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 0}, {""measureId"": 68, ""measureName"": ""FFS Inpatient Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 1}, {""measureId"": 69, ""measureName"": ""FFS Long-Term Care Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 2}, {""measureId"": 70, ""measureName"": ""FFS Other Medical Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 3}]}" 72,"{""measureId"": 72, ""measureName"": ""Total Monthly Beneficiary Payments"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Total-Mthly-Bene-Pmts.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) provide a unique source of expenditure information on states’ Medicaid and CHIP programs. Although other data sources provide aggregate state-level expenditure information, the TAF allow users to analyze payments associated with managed care organizations, individual providers, subgroups of beneficiaries, and specific services. The Centers for Medicare & Medicaid Services (CMS) and other TAF users require accurate payment data to calculate total cost of care metrics; to compare costs for specific conditions, services, or populations; and to compare costs across states or different payment systems.
States may reimburse health care providers for services delivered to Medicaid beneficiaries by paying directly for each covered service on a fee-for-service (FFS) basis or by paying a flat monthly payment per beneficiary for a set of services that is contracted to another entity—such as a managed care plan—which then assumes responsibility for delivering care to the beneficiary. [1] , [2] When the state contracts directly with a managed care plan, the covered services may either be comprehensive and provided by a Medicaid managed care organization (MCO), or they may be narrower sets of inpatient or outpatient services provided by a prepaid health plan (PHP). [3] Payments associated with these types of Medicaid managed care are referred to as capitation payments. States also make other monthly payments on behalf of Medicaid beneficiaries, including a flat fee paid to a primary care provider for primary care case management plan (PCCM) services; Medicare Part A and Part B premiums for Medicaid beneficiaries who are dually eligible for Medicare; and in some cases, premium assistance for Medicaid beneficiaries with private coverage. [4]
Most states use a mix of FFS and monthly beneficiary payments to pay for the care provided to Medicaid beneficiaries. In fiscal year 2017, monthly beneficiary payments accounted for 53 percent of total Medicaid spending nationally on health services. [5] Across most states, the vast majority of these payments were for comprehensive managed care plans, which accounted for nearly half of all Medicaid expenditures. Spending on PHPs, PCCM, and premium assistance accounted for a far smaller share of total Medicaid expenditures, and Medicare premiums paid on behalf of dually eligible beneficiaries accounted for about 3 percent of total Medicaid spending.
In order to claim federal matching funds for their medical assistance programs, states report aggregate expenditure data for Medicaid beneficiaries covered under Title XIX of the Social Security Act to CMS through the Medicaid Budget and Expenditure System (MBES). [6] The MBES system is designed to capture information states would report on the Quarterly Medicaid Statement of Expenditures for the Medical Assistance Program Form CMS-64; therefore, the data are commonly referred to as the CMS-64 data. Because of the oversight involved in ensuring the correct federal reimbursement to state Medicaid programs, the CMS-64 data are seen as a highly reliable source of expenditure information. If the expenditures in the TAF do not align with those in the CMS-64, that may suggest TAF data are incomplete or have other data quality errors that render them unrepresentative of actual Medicaid expenditures.
This data quality assessment examines the extent to which monthly beneficiary payments captured on states’ TAF records align with the total CMS-64 expenditures for monthly beneficiary payments both overall and for certain types of monthly payments: (1) comprehensive managed care (CMC) capitation payments, (2) PHP capitation payments, (3) PCCM flat fees, and (4) premium assistance payments to private plans. However, benchmarking TAF expenditure data against CMS-64 data is not straightforward. Although the data in both systems are similar, there are some important differences: [7]
More information on FFS expenditure data completeness can be found in the DQ Atlas single topic displays for Total FFS Expenditures , FFS Inpatient Expenditures , FFS Long-term Care Expenditures , FFS Other Medical Expenditures , and FFS Prescription Drug Expenditures .
In addition to FFS claims and monthly beneficiary payments, TAF also includes other types of payments, such as supplemental payment records and service tracking claims. Supplemental payment records represent payments for a specific service made in addition to a capitation payment or negotiated rate. Service tracking claims represent lump-sum payments made for services that cannot be attributed to a specific beneficiary, such as disproportionate share hospital (DSH) payments or payments to providers made under the Upper Payment Limit (UPL) demonstration.
PHPs often cover a specific type of service, such as behavioral health care or dental care.
Some Medicaid beneficiaries qualify for employer-based coverage, and if this coverage is less costly than enrolling the beneficiary in traditional Medicaid, states may pay the premium to the employer on behalf of the beneficiary. Some states have used other policy options, such as 1115 waiver demonstrations or the Basic Health Program, to enroll eligible beneficiaries in private plans available through the state or federal Health Insurance Exchange.
Medicaid and CHIP Payment and Access Commission. “EXHIBIT 17. Total Medicaid Benefit Spending by State and Category, FY 2017 (millions).” MACStats: Medicaid and CHIP Data Book December 2018. Washington, DC: MACPAC, December 2018, pp. 52–54. Available at https://www.macpac.gov/wp-content/uploads/2018/12/December-2018-MACStats-Data-Book.pdf . Accessed June 24, 2019.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
More information on FFS expenditure data completeness can be found in the DQ Atlas single topic displays for Total FFS Expenditures , FFS Inpatient Expenditures , FFS Long-term Care Expenditures , FFS Other Medical Expenditures , and FFS Prescription Drug Expenditures .
In addition to FFS claims and monthly beneficiary payments, TAF also includes other types of payments, such as supplemental payment records and service tracking claims. Supplemental payment records represent payments for a specific service made in addition to a capitation payment or negotiated rate. Service tracking claims represent lump-sum payments made for services that cannot be attributed to a specific beneficiary, such as disproportionate share hospital (DSH) payments or payments to providers made under the Upper Payment Limit (UPL) demonstration.
PHPs often cover a specific type of service, such as behavioral health care or dental care.
Some Medicaid beneficiaries qualify for employer-based coverage, and if this coverage is less costly than enrolling the beneficiary in traditional Medicaid, states may pay the premium to the employer on behalf of the beneficiary. Some states have used other policy options, such as 1115 waiver demonstrations or the Basic Health Program, to enroll eligible beneficiaries in private plans available through the state or federal Health Insurance Exchange.
Medicaid and CHIP Payment and Access Commission. \u201cEXHIBIT 17. Total Medicaid Benefit Spending by State and Category, FY 2017 (millions).\u201d MACStats: Medicaid and CHIP Data Book December 2018. Washington, DC: MACPAC, December 2018, pp. 52\u201354. Available at https://www.macpac.gov/wp-content/uploads/2018/12/December-2018-MACStats-Data-Book.pdf . Accessed June 24, 2019.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We compared monthly expenditures for Medicaid beneficiaries drawn from twelve months of TAF data [8] to the CMS-64 benchmark for four quarters of the calendar year.
To calculate monthly beneficiary expenditures in the TAF, we first selected records in the other services (OT) file that had a claim type code (CLM_TYPE_CD) indicating that the record was a capitation payment made on behalf of a Medicaid beneficiary. [9] In TAF, all monthly beneficiary payments should be reported on records with a claim type code of capitation payment; these records include some types of payments that are not typically referred to as capitation such as PCCM monthly fees. Because we found that some states report monthly beneficiary payments by using other claim types, we also included records with claim type code values of 4 (service tracking claim) [10] and 5 (supplemental payments) if they had a federally assigned service category (FASC) code [11] related to monthly beneficiary payments (FED_SRVC_CTGRY_CD values 11 or 12) or at least one line with a type of service code related to monthly beneficiary payments (TOS_CD values of 119, 120, 121, 122, 138, 143, or 144 [12] ).
Among records with claim type 2 or 5, we only kept those that matched to an eligibility record in the Demographic and Eligibility (DE) file that indicated that the beneficiary was enrolled in non-CHIP Medicaid at some point during the calendar year. [13] This is necessary because the CMS-64 benchmark includes only non-CHIP Medicaid expenditures, and the TAF claim type variable alone cannot distinguish payments made for Medicaid-expansion CHIP enrollees (whose costs are not reported on the CMS-64) from payments made for non-CHIP Medicaid beneficiaries (whose costs are reported on the CMS-64). We did not require records with claim type 4 to match to a Medicaid eligibility record because these types of records are lump-sum payments that are not connected to any one beneficiary and should not have a valid beneficiary identifier.
Finally, we excluded TAF records that represent Disproportionate Share Hospital (DSH) payments, supplemental payments made under the Upper Payment Limit (UPL) demonstration, electronic health record (EHR) payments, and drug rebates. [14] These payments were excluded because they do not represent monthly beneficiary payments for health services. Because very few states are reporting Medicare premium payments in T-MSIS, we also excluded Medicare premium payments. [15]
Once we had selected the set of TAF records included in the benchmarking analysis, we tabulated expenditures associated with monthly beneficiary payments in each state using the total Medicaid paid amount on the header record (TOT_MDCD_PD_AMT) for all included records with claim type codes values of 2 or 5. For records with a claim type code value of 4, we used the service tracking payment amount (SRVC_TRKNG_PYMT_AMT) to tabulate expenditures. If the service tracking payment amount was missing, we used the total Medicaid paid amount.
We calculated the CMS-64 benchmark for monthly beneficiary payments using four quarterly net expenditures reports from MBES that cover a calendar year. [16] We summed the net expenditures for the categories of service covering monthly payments for CMCs, PHPs, PCCMs, and Medicaid premium assistance payments, using the category of service codes shown in Table 1. We excluded Medicare premium payments from the benchmark for monthly beneficiary payments, as well as any CMS-64 categories that did not represent monthly beneficiary payments, including services typically paid on a FFS basis, drug rebates, DSH payments, General Medical Education payments, and other supplemental payments.
Benchmarking by type of monthly beneficiary payment
We examined TAF-based expenditures for total monthly beneficiary payments as well as for four specific types of payment: (1) CMC capitation payments, (2) PHP capitation payments, (3) PCCM flat fees, and (4) premium assistance payments to private plans. A small number of records representing “other” monthly beneficiary payments are included in the total expenditure calculation but are not captured in these four specific payment types. [17]
We classified TAF records into each of these four categories if the record had at least one line with a federally assigned service category (FASC) code and a type of service (TOS) code indicating the given payment category, as shown in Table 1. [18] , [19] , [20]
Table 1. TAF records and CMS-64 categories that represent monthly beneficiary payments
| Payment type | TAF records | CMS-64 categories |
|---|---|---|
| Payments to CMC plans | Records with a FED_SRVC_CTGRY_CD value of 11 and TOS_CD values of 119. These represent capitated payments to HMOs, HIOs, or PACE plans. | Categories of service 18A, 18A1, 18A2, 18A3, 18A4, 18A5, 18A6, and 22. a These categories represent Medicaid MCO, Medicaid MCO - Evaluation and Management, Medicaid MCO - Vaccine Codes, Medicaid MCO - Community First Choice, Medicaid MCO - Preventive Services, Medicaid MCO - Certified Community Behavior Health Clinic Payment, Medicaid MCO - Services Subject to Electronic Visit Verification Requirements, and Programs of All-Inclusive Care for the Elderly (PACE). |
| Payments to PHPs | Records with a FED_SRVC_CTGRY_CD value of 11 and TOS_CD values of 122. These represent capitated payments to prepaid health plans (PHPs). | Categories of service 18B1, 18B1a, 18B1b, 18B1c, 18B1d, 18B1e, 18B1f 18B2, 18B2a, 18B2b, 18B2c, 18B2d, 18B2e and 18B2f a . These categories represent Prepaid Ambulatory Health Plans (PAHP), PAHP - Evaluation and Management, PAHP - Vaccine Codes, PAHP - Community First Choice, PAHP - Preventive Services, Medicaid PAHP - Certified Community Behavior Health Clinic Payments, MCO PAHP - Services Subject to Electronic Visit Verification Requirements, Prepaid Inpatient Health Plans (PIHP), PIHP - Evaluation and Management, PIHP - Vaccine Codes, PIHP - Community First Choice, PIHP - Preventive Services, Medicaid PIHP - Certified Community Behavior Health Clinic Payments, and MCO PIHP - Services Subject to Electronic Visit Verification Requirements |
| Payments for premium assistance | Records with a FED_SRVC_CTGRY_CD value of 12 and TOS_CD values of 121. These represent premium payments by Medicaid agencies for private health insurance. | Categories of service 18C and 18E. These categories represent Medicaid - Group Health and Medicaid – Other. |
| Payments for PCCM | Records with a FED_SRVC_CTGRY_CD value of 12 and TOS_CD values of 120. These represent flat fees for primary care case management (PCCM) services. | Category of service 25. This category represents Primary Care Case Management. |
Note: \tSome states use managed care in their 1915(c) Home and Community-Based Service waiver programs. Some of these states may report a portion of their managed care capitation payments related to 1915(c) managed care services under the CMS-64 table category code 19 (Home & Community-Based Services). This category is not captured in our benchmark because most of these waiver services are not delivered through managed care. In T-MSIS, the capitation payment record does not distinguish between the portions of the capitation payment allocated to each CMS-64 category. When this occurs, the TAF data will not align with the CMS-64 benchmark.
a CMS-64 categories of service 18A5 (Medicaid MCO - Certified Community Behavior Health Clinic Payment), 18B1e (Medicaid PAHP - Certified Community Behavior Health Clinic Payments), and 18B2e (Medicaid PIHP - Certified Community Behavior Health Clinic Payments) are valid starting in 2017. CMS-64 categories of service 18A6, 18B1f, and 18B2f (Services Subject to Electronic Visit Verification Requirements for Medicaid MCO, MCO PAHP, and MCO PIHP, respectively) are valid starting in 2021.
Data quality assessment criteria
We categorized each state’s monthly beneficiary payment expenditures in TAF as having high, moderate, low, or very low alignment with the CMS-64 benchmark based on the percent difference between the two data sources (Table 2). [21] States with a low absolute percentage difference between the expenditures captured in TAF and the CMS-64 are considered to present a low level of concern about the completeness and quality of their TAF data. However, TAF users should be aware that in states with large Medicaid programs, even a small percentage difference may represent a substantial difference in the dollar amount between the data sources.
Table 2. Criteria for DQ assessment of monthly beneficiary payments
| Percent difference between the monthly beneficiary payments captured in TAF and the CMS-64 | Level of alignment | DQ assessment |
|---|---|---|
| x < 5 percent | High | Low concern |
| 5 percent ≤ x < 20 percent | Moderate | Medium concern |
| 20 percent ≤ x < 50 percent | Low | High concern |
| x ≥ 50 percent or if state has no TAF data | Very low | Unusable |
There are three reasons other than the quality and completeness of TAF expenditure data that could result in a state’s monthly beneficiary payment expenditures in the TAF differing from the CMS-64: (1) structural differences in the data sets, such as organizing the data by service date versus claim payment date; (2) data quality problems in the TAF that are unrelated to expenditures, such as a low volume of claims in the TAF because of missing date values on claims or issues with assigning a final action status to claims; and (3) variation in how states report certain types of monthly beneficiary payments on the CMS-64 that are difficult to reproduce in the TAF data. [22] For instance, states using managed care in their 1915(c) Home and Community Based Services (HCBS) waiver programs may elect to report those expenditures in the CMS-64 category for HCBS services rather than in the CMS-64 category for managed care expenditures. When this occurs, we would expect to see a difference between the TAF-based monthly beneficiary expenditures and the CMS-64 benchmark even if TAF data on monthly beneficiary payments are complete and accurate.
Methods previously used to assess data quality
Table 3 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 3. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We included in the analysis Medicaid capitation records (identified by CLAIM_TYPE_CD=2). In the early years of T-MSIS, a few states reported expenditures for Medicaid beneficiaries covered under Title XIX on “other” (non-Medicaid, non-CHIP) capitation records (those with CLAIM_TYPE_CD=V). Because it is unclear what those records represent, we exclude them from this analysis.
Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
Type of service codes 138, 143, and 144 are valid starting in 2020.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the beneficiary had at least one month during the calendar year in which the CHIP code value was 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-75, we considered the beneficiary to have non-CHIP Medicaid enrollment during the year.
We used federally assigned service category code to identify and exclude DSH payments (13), and type of service code to identify and exclude supplemental payments (132, 133, 134) and EHR payments (135). Drug rebates were identified as records were any line had a type of service code of 131 or a service tracking type code (SRVC_TRKNG_TYPE_CD) of 01.
Medicare premiums are identified in TAF as records with a TOS code of 139, 140, 141, or 142 (valid values starting in 2020).
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
“Other” monthly beneficiary payments are identified in TAF as (1) records with claim type 2 where no claim line has a TOS code value of 119, 120, 121, 122, 123, 131, 132, or 135, or (2) any remaining records with claim type code 4 or 5 with a FASC code value of 12 or a TOS code value of 138, 143, or 144. TOS codes 138, 143, and 144 are valid starting in 2020. In contrast, all of the CMS-64 categories of service related to monthly beneficiary payments fall into one of the four specific types of monthly beneficiary payments.
Because the TAF type of service code does not provide a direct one-to-one mapping to the CMS-64 categories, we excluded payments for CMS-64 categories that did not map directly to the categories of CMC, PHP, PCCM, and premium assistance from these sub-analyses.
There are multiple variables that could be used to identify spending associated with different types of monthly beneficiary payments. The two that perform the best are the type of service code and the managed care plan type. The latter provides a finer level of detail than type of service code, but the user will have to link the TAF data from the other services (OT) file to the TAF Managed Care Plan file, which may not be available to all users. Another variable that can be used but that did not perform well in our analyses, is the Title XIX category of service code (XIX_SRVC_CTGRY_CD), which provides a direct one-to-one mapping of TAF line records to the CMS-64 categories of service. However, many states do not report this information consistently, and the high rates of missing values limit the use of this code for meaningful analyses of expenditures.
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: [(TAF – benchmark) / benchmark]*100.
For more information about the differences between TAF and CMS-64 expenditure data, see the methodology brief, “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We included in the analysis Medicaid capitation records (identified by CLAIM_TYPE_CD=2). In the early years of T-MSIS, a few states reported expenditures for Medicaid beneficiaries covered under Title XIX on \u201cother\u201d (non-Medicaid, non-CHIP) capitation records (those with CLAIM_TYPE_CD=V). Because it is unclear what those records represent, we exclude them from this analysis.
Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
Type of service codes 138, 143, and 144 are valid starting in 2020.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the beneficiary had at least one month during the calendar year in which the CHIP code value was 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-75, we considered the beneficiary to have non-CHIP Medicaid enrollment during the year.
We used federally assigned service category code to identify and exclude DSH payments (13), and type of service code to identify and exclude supplemental payments (132, 133, 134) and EHR payments (135). Drug rebates were identified as records were any line had a type of service code of 131 or a service tracking type code (SRVC_TRKNG_TYPE_CD) of 01.
Medicare premiums are identified in TAF as records with a TOS code of 139, 140, 141, or 142 (valid values starting in 2020).
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
\u201cOther\u201d monthly beneficiary payments are identified in TAF as (1) records with claim type 2 where no claim line has a TOS code value of 119, 120, 121, 122, 123, 131, 132, or 135, or (2) any remaining records with claim type code 4 or 5 with a FASC code value of 12 or a TOS code value of 138, 143, or 144. TOS codes 138, 143, and 144 are valid starting in 2020. In contrast, all of the CMS-64 categories of service related to monthly beneficiary payments fall into one of the four specific types of monthly beneficiary payments.
Because the TAF type of service code does not provide a direct one-to-one mapping to the CMS-64 categories, we excluded payments for CMS-64 categories that did not map directly to the categories of CMC, PHP, PCCM, and premium assistance from these sub-analyses.
There are multiple variables that could be used to identify spending associated with different types of monthly beneficiary payments. The two that perform the best are the type of service code and the managed care plan type. The latter provides a finer level of detail than type of service code, but the user will have to link the TAF data from the other services (OT) file to the TAF Managed Care Plan file, which may not be available to all users. Another variable that can be used but that did not perform well in our analyses, is the Title XIX category of service code (XIX_SRVC_CTGRY_CD), which provides a direct one-to-one mapping of TAF line records to the CMS-64 categories of service. However, many states do not report this information consistently, and the high rates of missing values limit the use of this code for meaningful analyses of expenditures.
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: [(TAF \u2013 benchmark) / benchmark]*100.
For more information about the differences between TAF and CMS-64 expenditure data, see the methodology brief, \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
The TAF should include all state spending to provide or cover care for Medicaid beneficiaries, including monthly payments to Medicaid managed care plans, to providers for primary care case management services, to Medicare for premium payments on behalf of dually eligible beneficiaries, and to private plans for premium assistance. This analysis examines how well the total expenditures for monthly payments on behalf of Medicaid beneficiaries align with an external benchmark, the CMS-64.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""6031"", ""relatedTopics"": [{""measureId"": 73, ""measureName"": ""CMC Payments"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 1}, {""measureId"": 74, ""measureName"": ""PHP Payments"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 2}, {""measureId"": 75, ""measureName"": ""PCCM Fees"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 3}, {""measureId"": 76, ""measureName"": ""Premium Assistance Payments"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 4}]}" 73,"{""measureId"": 73, ""measureName"": ""CMC Payments"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-CMC-Pmts.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) provide a unique source of expenditure information on states’ Medicaid and CHIP programs. Although other data sources provide aggregate state-level expenditure information, the TAF allow users to analyze payments associated with managed care organizations, individual providers, subgroups of beneficiaries, and specific services. The Centers for Medicare & Medicaid Services (CMS) and other TAF users require accurate payment data to calculate total cost of care metrics; to compare costs for specific conditions, services, or populations; and to compare costs across states or different payment systems.
States may reimburse health care providers for services delivered to Medicaid beneficiaries by paying directly for each covered service on a fee-for-service (FFS) basis or by paying a flat monthly payment per beneficiary for a set of services that is contracted to another entity—such as a managed care plan—which then assumes responsibility for delivering care to the beneficiary. [1] , [2] When the state contracts directly with a managed care plan, the covered services may either be comprehensive and provided by a Medicaid managed care organization (MCO), or they may be narrower sets of inpatient or outpatient services provided by a prepaid health plan (PHP). [3] Payments associated with these types of Medicaid managed care are referred to as capitation payments. States also make other monthly payments on behalf of Medicaid beneficiaries, including a flat fee paid to a primary care provider for primary care case management plan (PCCM) services; Medicare Part A and Part B premiums for Medicaid beneficiaries who are dually eligible for Medicare; and in some cases, premium assistance for Medicaid beneficiaries with private coverage. [4]
Most states use a mix of FFS and monthly beneficiary payments to pay for the care provided to Medicaid beneficiaries. In fiscal year 2017, monthly beneficiary payments accounted for 53 percent of total Medicaid spending nationally on health services. [5] Across most states, the vast majority of these payments were for comprehensive managed care plans, which accounted for nearly half of all Medicaid expenditures. Spending on PHPs, PCCM, and premium assistance accounted for a far smaller share of total Medicaid expenditures, and Medicare premiums paid on behalf of dually eligible beneficiaries accounted for about 3 percent of total Medicaid spending.
In order to claim federal matching funds for their medical assistance programs, states report aggregate expenditure data for Medicaid beneficiaries covered under Title XIX of the Social Security Act to CMS through the Medicaid Budget and Expenditure System (MBES). [6] The MBES system is designed to capture information states would report on the Quarterly Medicaid Statement of Expenditures for the Medical Assistance Program Form CMS-64; therefore, the data are commonly referred to as the CMS-64 data. Because of the oversight involved in ensuring the correct federal reimbursement to state Medicaid programs, the CMS-64 data are seen as a highly reliable source of expenditure information. If the expenditures in the TAF do not align with those in the CMS-64, that may suggest TAF data are incomplete or have other data quality errors that render them unrepresentative of actual Medicaid expenditures.
This data quality assessment examines the extent to which monthly beneficiary payments captured on states’ TAF records align with the total CMS-64 expenditures for monthly beneficiary payments both overall and for certain types of monthly payments: (1) comprehensive managed care (CMC) capitation payments, (2) PHP capitation payments, (3) PCCM flat fees, and (4) premium assistance payments to private plans. However, benchmarking TAF expenditure data against CMS-64 data is not straightforward. Although the data in both systems are similar, there are some important differences: [7]
More information on FFS expenditure data completeness can be found in the DQ Atlas single topic displays for Total FFS Expenditures , FFS Inpatient Expenditures , FFS Long-term Care Expenditures , FFS Other Medical Expenditures , and FFS Prescription Drug Expenditures .
In addition to FFS claims and monthly beneficiary payments, TAF also includes other types of payments, such as supplemental payment records and service tracking claims. Supplemental payment records represent payments for a specific service made in addition to a capitation payment or negotiated rate. Service tracking claims represent lump-sum payments made for services that cannot be attributed to a specific beneficiary, such as disproportionate share hospital (DSH) payments or payments to providers made under the Upper Payment Limit (UPL) demonstration.
PHPs often cover a specific type of service, such as behavioral health care or dental care.
Some Medicaid beneficiaries qualify for employer-based coverage, and if this coverage is less costly than enrolling the beneficiary in traditional Medicaid, states may pay the premium to the employer on behalf of the beneficiary. Some states have used other policy options, such as 1115 waiver demonstrations or the Basic Health Program, to enroll eligible beneficiaries in private plans available through the state or federal Health Insurance Exchange.
Medicaid and CHIP Payment and Access Commission. “EXHIBIT 17. Total Medicaid Benefit Spending by State and Category, FY 2017 (millions).” MACStats: Medicaid and CHIP Data Book December 2018. Washington, DC: MACPAC, December 2018, pp. 52–54. Available at https://www.macpac.gov/wp-content/uploads/2018/12/December-2018-MACStats-Data-Book.pdf . Accessed June 24, 2019.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
More information on FFS expenditure data completeness can be found in the DQ Atlas single topic displays for Total FFS Expenditures , FFS Inpatient Expenditures , FFS Long-term Care Expenditures , FFS Other Medical Expenditures , and FFS Prescription Drug Expenditures .
In addition to FFS claims and monthly beneficiary payments, TAF also includes other types of payments, such as supplemental payment records and service tracking claims. Supplemental payment records represent payments for a specific service made in addition to a capitation payment or negotiated rate. Service tracking claims represent lump-sum payments made for services that cannot be attributed to a specific beneficiary, such as disproportionate share hospital (DSH) payments or payments to providers made under the Upper Payment Limit (UPL) demonstration.
PHPs often cover a specific type of service, such as behavioral health care or dental care.
Some Medicaid beneficiaries qualify for employer-based coverage, and if this coverage is less costly than enrolling the beneficiary in traditional Medicaid, states may pay the premium to the employer on behalf of the beneficiary. Some states have used other policy options, such as 1115 waiver demonstrations or the Basic Health Program, to enroll eligible beneficiaries in private plans available through the state or federal Health Insurance Exchange.
Medicaid and CHIP Payment and Access Commission. \u201cEXHIBIT 17. Total Medicaid Benefit Spending by State and Category, FY 2017 (millions).\u201d MACStats: Medicaid and CHIP Data Book December 2018. Washington, DC: MACPAC, December 2018, pp. 52\u201354. Available at https://www.macpac.gov/wp-content/uploads/2018/12/December-2018-MACStats-Data-Book.pdf . Accessed June 24, 2019.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We compared monthly expenditures for Medicaid beneficiaries drawn from twelve months of TAF data [8] to the CMS-64 benchmark for four quarters of the calendar year.
To calculate monthly beneficiary expenditures in the TAF, we first selected records in the other services (OT) file that had a claim type code (CLM_TYPE_CD) indicating that the record was a capitation payment made on behalf of a Medicaid beneficiary. [9] In TAF, all monthly beneficiary payments should be reported on records with a claim type code of capitation payment; these records include some types of payments that are not typically referred to as capitation such as PCCM monthly fees. Because we found that some states report monthly beneficiary payments by using other claim types, we also included records with claim type code values of 4 (service tracking claim) [10] and 5 (supplemental payments) if they had a federally assigned service category (FASC) code [11] related to monthly beneficiary payments (FED_SRVC_CTGRY_CD values 11 or 12) or at least one line with a type of service code related to monthly beneficiary payments (TOS_CD values of 119, 120, 121, 122, 138, 143, or 144 [12] ).
Among records with claim type 2 or 5, we only kept those that matched to an eligibility record in the Demographic and Eligibility (DE) file that indicated that the beneficiary was enrolled in non-CHIP Medicaid at some point during the calendar year. [13] This is necessary because the CMS-64 benchmark includes only non-CHIP Medicaid expenditures, and the TAF claim type variable alone cannot distinguish payments made for Medicaid-expansion CHIP enrollees (whose costs are not reported on the CMS-64) from payments made for non-CHIP Medicaid beneficiaries (whose costs are reported on the CMS-64). We did not require records with claim type 4 to match to a Medicaid eligibility record because these types of records are lump-sum payments that are not connected to any one beneficiary and should not have a valid beneficiary identifier.
Finally, we excluded TAF records that represent Disproportionate Share Hospital (DSH) payments, supplemental payments made under the Upper Payment Limit (UPL) demonstration, electronic health record (EHR) payments, and drug rebates. [14] These payments were excluded because they do not represent monthly beneficiary payments for health services. Because very few states are reporting Medicare premium payments in T-MSIS, we also excluded Medicare premium payments. [15]
Once we had selected the set of TAF records included in the benchmarking analysis, we tabulated expenditures associated with monthly beneficiary payments in each state using the total Medicaid paid amount on the header record (TOT_MDCD_PD_AMT) for all included records with claim type codes values of 2 or 5. For records with a claim type code value of 4, we used the service tracking payment amount (SRVC_TRKNG_PYMT_AMT) to tabulate expenditures. If the service tracking payment amount was missing, we used the total Medicaid paid amount.
We calculated the CMS-64 benchmark for monthly beneficiary payments using four quarterly net expenditures reports from MBES that cover a calendar year. [16] We summed the net expenditures for the categories of service covering monthly payments for CMCs, PHPs, PCCMs, and Medicaid premium assistance payments, using the category of service codes shown in Table 1. We excluded Medicare premium payments from the benchmark for monthly beneficiary payments, as well as any CMS-64 categories that did not represent monthly beneficiary payments, including services typically paid on a FFS basis, drug rebates, DSH payments, General Medical Education payments, and other supplemental payments.
Benchmarking by type of monthly beneficiary payment
We examined TAF-based expenditures for total monthly beneficiary payments as well as for four specific types of payment: (1) CMC capitation payments, (2) PHP capitation payments, (3) PCCM flat fees, and (4) premium assistance payments to private plans. A small number of records representing “other” monthly beneficiary payments are included in the total expenditure calculation but are not captured in these four specific payment types. [17]
We classified TAF records into each of these four categories if the record had at least one line with a federally assigned service category (FASC) code and a type of service (TOS) code indicating the given payment category, as shown in Table 1. [18] , [19] , [20]
Table 1. TAF records and CMS-64 categories that represent monthly beneficiary payments
| Payment type | TAF records | CMS-64 categories |
|---|---|---|
| Payments to CMC plans | Records with a FED_SRVC_CTGRY_CD value of 11 and TOS_CD values of 119. These represent capitated payments to HMOs, HIOs, or PACE plans. | Categories of service 18A, 18A1, 18A2, 18A3, 18A4, 18A5, 18A6, and 22. a These categories represent Medicaid MCO, Medicaid MCO - Evaluation and Management, Medicaid MCO - Vaccine Codes, Medicaid MCO - Community First Choice, Medicaid MCO - Preventive Services, Medicaid MCO - Certified Community Behavior Health Clinic Payment, Medicaid MCO - Services Subject to Electronic Visit Verification Requirements, and Programs of All-Inclusive Care for the Elderly (PACE). |
| Payments to PHPs | Records with a FED_SRVC_CTGRY_CD value of 11 and TOS_CD values of 122. These represent capitated payments to prepaid health plans (PHPs). | Categories of service 18B1, 18B1a, 18B1b, 18B1c, 18B1d, 18B1e, 18B1f 18B2, 18B2a, 18B2b, 18B2c, 18B2d, 18B2e and 18B2f a . These categories represent Prepaid Ambulatory Health Plans (PAHP), PAHP - Evaluation and Management, PAHP - Vaccine Codes, PAHP - Community First Choice, PAHP - Preventive Services, Medicaid PAHP - Certified Community Behavior Health Clinic Payments, MCO PAHP - Services Subject to Electronic Visit Verification Requirements, Prepaid Inpatient Health Plans (PIHP), PIHP - Evaluation and Management, PIHP - Vaccine Codes, PIHP - Community First Choice, PIHP - Preventive Services, Medicaid PIHP - Certified Community Behavior Health Clinic Payments, and MCO PIHP - Services Subject to Electronic Visit Verification Requirements |
| Payments for premium assistance | Records with a FED_SRVC_CTGRY_CD value of 12 and TOS_CD values of 121. These represent premium payments by Medicaid agencies for private health insurance. | Categories of service 18C and 18E. These categories represent Medicaid - Group Health and Medicaid – Other. |
| Payments for PCCM | Records with a FED_SRVC_CTGRY_CD value of 12 and TOS_CD values of 120. These represent flat fees for primary care case management (PCCM) services. | Category of service 25. This category represents Primary Care Case Management. |
Note: \tSome states use managed care in their 1915(c) Home and Community-Based Service waiver programs. Some of these states may report a portion of their managed care capitation payments related to 1915(c) managed care services under the CMS-64 table category code 19 (Home & Community-Based Services). This category is not captured in our benchmark because most of these waiver services are not delivered through managed care. In T-MSIS, the capitation payment record does not distinguish between the portions of the capitation payment allocated to each CMS-64 category. When this occurs, the TAF data will not align with the CMS-64 benchmark.
a CMS-64 categories of service 18A5 (Medicaid MCO - Certified Community Behavior Health Clinic Payment), 18B1e (Medicaid PAHP - Certified Community Behavior Health Clinic Payments), and 18B2e (Medicaid PIHP - Certified Community Behavior Health Clinic Payments) are valid starting in 2017. CMS-64 categories of service 18A6, 18B1f, and 18B2f (Services Subject to Electronic Visit Verification Requirements for Medicaid MCO, MCO PAHP, and MCO PIHP, respectively) are valid starting in 2021.
Data quality assessment criteria
We categorized each state’s monthly beneficiary payment expenditures in TAF as having high, moderate, low, or very low alignment with the CMS-64 benchmark based on the percent difference between the two data sources (Table 2). [21] States with a low absolute percentage difference between the expenditures captured in TAF and the CMS-64 are considered to present a low level of concern about the completeness and quality of their TAF data. However, TAF users should be aware that in states with large Medicaid programs, even a small percentage difference may represent a substantial difference in the dollar amount between the data sources.
Table 2. Criteria for DQ assessment of monthly beneficiary payments
| Percent difference between the monthly beneficiary payments captured in TAF and the CMS-64 | Level of alignment | DQ assessment |
|---|---|---|
| x < 5 percent | High | Low concern |
| 5 percent ≤ x < 20 percent | Moderate | Medium concern |
| 20 percent ≤ x < 50 percent | Low | High concern |
| x ≥ 50 percent or if state has no TAF data | Very low | Unusable |
There are three reasons other than the quality and completeness of TAF expenditure data that could result in a state’s monthly beneficiary payment expenditures in the TAF differing from the CMS-64: (1) structural differences in the data sets, such as organizing the data by service date versus claim payment date; (2) data quality problems in the TAF that are unrelated to expenditures, such as a low volume of claims in the TAF because of missing date values on claims or issues with assigning a final action status to claims; and (3) variation in how states report certain types of monthly beneficiary payments on the CMS-64 that are difficult to reproduce in the TAF data. [22] For instance, states using managed care in their 1915(c) Home and Community Based Services (HCBS) waiver programs may elect to report those expenditures in the CMS-64 category for HCBS services rather than in the CMS-64 category for managed care expenditures. When this occurs, we would expect to see a difference between the TAF-based monthly beneficiary expenditures and the CMS-64 benchmark even if TAF data on monthly beneficiary payments are complete and accurate.
Methods previously used to assess data quality
Table 3 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 3. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We included in the analysis Medicaid capitation records (identified by CLAIM_TYPE_CD=2). In the early years of T-MSIS, a few states reported expenditures for Medicaid beneficiaries covered under Title XIX on “other” (non-Medicaid, non-CHIP) capitation records (those with CLAIM_TYPE_CD=V). Because it is unclear what those records represent, we exclude them from this analysis.
Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
Type of service codes 138, 143, and 144 are valid starting in 2020.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the beneficiary had at least one month during the calendar year in which the CHIP code value was 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-75, we considered the beneficiary to have non-CHIP Medicaid enrollment during the year.
We used federally assigned service category code to identify and exclude DSH payments (13), and type of service code to identify and exclude supplemental payments (132, 133, 134) and EHR payments (135). Drug rebates were identified as records were any line had a type of service code of 131 or a service tracking type code (SRVC_TRKNG_TYPE_CD) of 01.
Medicare premiums are identified in TAF as records with a TOS code of 139, 140, 141, or 142 (valid values starting in 2020).
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
“Other” monthly beneficiary payments are identified in TAF as (1) records with claim type 2 where no claim line has a TOS code value of 119, 120, 121, 122, 123, 131, 132, or 135, or (2) any remaining records with claim type code 4 or 5 with a FASC code value of 12 or a TOS code value of 138, 143, or 144. TOS codes 138, 143, and 144 are valid starting in 2020. In contrast, all of the CMS-64 categories of service related to monthly beneficiary payments fall into one of the four specific types of monthly beneficiary payments.
Because the TAF type of service code does not provide a direct one-to-one mapping to the CMS-64 categories, we excluded payments for CMS-64 categories that did not map directly to the categories of CMC, PHP, PCCM, and premium assistance from these sub-analyses.
There are multiple variables that could be used to identify spending associated with different types of monthly beneficiary payments. The two that perform the best are the type of service code and the managed care plan type. The latter provides a finer level of detail than type of service code, but the user will have to link the TAF data from the other services (OT) file to the TAF Managed Care Plan file, which may not be available to all users. Another variable that can be used but that did not perform well in our analyses, is the Title XIX category of service code (XIX_SRVC_CTGRY_CD), which provides a direct one-to-one mapping of TAF line records to the CMS-64 categories of service. However, many states do not report this information consistently, and the high rates of missing values limit the use of this code for meaningful analyses of expenditures.
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: [(TAF – benchmark) / benchmark]*100.
For more information about the differences between TAF and CMS-64 expenditure data, see the methodology brief, “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We included in the analysis Medicaid capitation records (identified by CLAIM_TYPE_CD=2). In the early years of T-MSIS, a few states reported expenditures for Medicaid beneficiaries covered under Title XIX on \u201cother\u201d (non-Medicaid, non-CHIP) capitation records (those with CLAIM_TYPE_CD=V). Because it is unclear what those records represent, we exclude them from this analysis.
Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
Type of service codes 138, 143, and 144 are valid starting in 2020.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the beneficiary had at least one month during the calendar year in which the CHIP code value was 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-75, we considered the beneficiary to have non-CHIP Medicaid enrollment during the year.
We used federally assigned service category code to identify and exclude DSH payments (13), and type of service code to identify and exclude supplemental payments (132, 133, 134) and EHR payments (135). Drug rebates were identified as records were any line had a type of service code of 131 or a service tracking type code (SRVC_TRKNG_TYPE_CD) of 01.
Medicare premiums are identified in TAF as records with a TOS code of 139, 140, 141, or 142 (valid values starting in 2020).
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
\u201cOther\u201d monthly beneficiary payments are identified in TAF as (1) records with claim type 2 where no claim line has a TOS code value of 119, 120, 121, 122, 123, 131, 132, or 135, or (2) any remaining records with claim type code 4 or 5 with a FASC code value of 12 or a TOS code value of 138, 143, or 144. TOS codes 138, 143, and 144 are valid starting in 2020. In contrast, all of the CMS-64 categories of service related to monthly beneficiary payments fall into one of the four specific types of monthly beneficiary payments.
Because the TAF type of service code does not provide a direct one-to-one mapping to the CMS-64 categories, we excluded payments for CMS-64 categories that did not map directly to the categories of CMC, PHP, PCCM, and premium assistance from these sub-analyses.
There are multiple variables that could be used to identify spending associated with different types of monthly beneficiary payments. The two that perform the best are the type of service code and the managed care plan type. The latter provides a finer level of detail than type of service code, but the user will have to link the TAF data from the other services (OT) file to the TAF Managed Care Plan file, which may not be available to all users. Another variable that can be used but that did not perform well in our analyses, is the Title XIX category of service code (XIX_SRVC_CTGRY_CD), which provides a direct one-to-one mapping of TAF line records to the CMS-64 categories of service. However, many states do not report this information consistently, and the high rates of missing values limit the use of this code for meaningful analyses of expenditures.
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: [(TAF \u2013 benchmark) / benchmark]*100.
For more information about the differences between TAF and CMS-64 expenditure data, see the methodology brief, \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
The TAF should include all state spending to provide or cover care for Medicaid beneficiaries, including monthly capitation payments to comprehensive managed care (CMC) plans. CMC arrangements are one of the most common types of Medicaid managed care in which the plan is responsible for delivering a broad range of primary, specialty, and acute care services to beneficiaries in exchange for a monthly capitation payment from the state. This analysis examines how well the total expenditures for monthly CMC capitation payments on behalf of Medicaid beneficiaries align with an external benchmark, the CMS-64.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""6031"", ""relatedTopics"": [{""measureId"": 72, ""measureName"": ""Total Monthly Beneficiary Payments"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 0}, {""measureId"": 74, ""measureName"": ""PHP Payments"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 2}, {""measureId"": 75, ""measureName"": ""PCCM Fees"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 3}, {""measureId"": 76, ""measureName"": ""Premium Assistance Payments"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 4}]}" 74,"{""measureId"": 74, ""measureName"": ""PHP Payments"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-PHP-Pmts.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) provide a unique source of expenditure information on states’ Medicaid and CHIP programs. Although other data sources provide aggregate state-level expenditure information, the TAF allow users to analyze payments associated with managed care organizations, individual providers, subgroups of beneficiaries, and specific services. The Centers for Medicare & Medicaid Services (CMS) and other TAF users require accurate payment data to calculate total cost of care metrics; to compare costs for specific conditions, services, or populations; and to compare costs across states or different payment systems.
States may reimburse health care providers for services delivered to Medicaid beneficiaries by paying directly for each covered service on a fee-for-service (FFS) basis or by paying a flat monthly payment per beneficiary for a set of services that is contracted to another entity—such as a managed care plan—which then assumes responsibility for delivering care to the beneficiary. [1] , [2] When the state contracts directly with a managed care plan, the covered services may either be comprehensive and provided by a Medicaid managed care organization (MCO), or they may be narrower sets of inpatient or outpatient services provided by a prepaid health plan (PHP). [3] Payments associated with these types of Medicaid managed care are referred to as capitation payments. States also make other monthly payments on behalf of Medicaid beneficiaries, including a flat fee paid to a primary care provider for primary care case management plan (PCCM) services; Medicare Part A and Part B premiums for Medicaid beneficiaries who are dually eligible for Medicare; and in some cases, premium assistance for Medicaid beneficiaries with private coverage. [4]
Most states use a mix of FFS and monthly beneficiary payments to pay for the care provided to Medicaid beneficiaries. In fiscal year 2017, monthly beneficiary payments accounted for 53 percent of total Medicaid spending nationally on health services. [5] Across most states, the vast majority of these payments were for comprehensive managed care plans, which accounted for nearly half of all Medicaid expenditures. Spending on PHPs, PCCM, and premium assistance accounted for a far smaller share of total Medicaid expenditures, and Medicare premiums paid on behalf of dually eligible beneficiaries accounted for about 3 percent of total Medicaid spending.
In order to claim federal matching funds for their medical assistance programs, states report aggregate expenditure data for Medicaid beneficiaries covered under Title XIX of the Social Security Act to CMS through the Medicaid Budget and Expenditure System (MBES). [6] The MBES system is designed to capture information states would report on the Quarterly Medicaid Statement of Expenditures for the Medical Assistance Program Form CMS-64; therefore, the data are commonly referred to as the CMS-64 data. Because of the oversight involved in ensuring the correct federal reimbursement to state Medicaid programs, the CMS-64 data are seen as a highly reliable source of expenditure information. If the expenditures in the TAF do not align with those in the CMS-64, that may suggest TAF data are incomplete or have other data quality errors that render them unrepresentative of actual Medicaid expenditures.
This data quality assessment examines the extent to which monthly beneficiary payments captured on states’ TAF records align with the total CMS-64 expenditures for monthly beneficiary payments both overall and for certain types of monthly payments: (1) comprehensive managed care (CMC) capitation payments, (2) PHP capitation payments, (3) PCCM flat fees, and (4) premium assistance payments to private plans. However, benchmarking TAF expenditure data against CMS-64 data is not straightforward. Although the data in both systems are similar, there are some important differences: [7]
More information on FFS expenditure data completeness can be found in the DQ Atlas single topic displays for Total FFS Expenditures , FFS Inpatient Expenditures , FFS Long-term Care Expenditures , FFS Other Medical Expenditures , and FFS Prescription Drug Expenditures .
In addition to FFS claims and monthly beneficiary payments, TAF also includes other types of payments, such as supplemental payment records and service tracking claims. Supplemental payment records represent payments for a specific service made in addition to a capitation payment or negotiated rate. Service tracking claims represent lump-sum payments made for services that cannot be attributed to a specific beneficiary, such as disproportionate share hospital (DSH) payments or payments to providers made under the Upper Payment Limit (UPL) demonstration.
PHPs often cover a specific type of service, such as behavioral health care or dental care.
Some Medicaid beneficiaries qualify for employer-based coverage, and if this coverage is less costly than enrolling the beneficiary in traditional Medicaid, states may pay the premium to the employer on behalf of the beneficiary. Some states have used other policy options, such as 1115 waiver demonstrations or the Basic Health Program, to enroll eligible beneficiaries in private plans available through the state or federal Health Insurance Exchange.
Medicaid and CHIP Payment and Access Commission. “EXHIBIT 17. Total Medicaid Benefit Spending by State and Category, FY 2017 (millions).” MACStats: Medicaid and CHIP Data Book December 2018. Washington, DC: MACPAC, December 2018, pp. 52–54. Available at https://www.macpac.gov/wp-content/uploads/2018/12/December-2018-MACStats-Data-Book.pdf . Accessed June 24, 2019.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
More information on FFS expenditure data completeness can be found in the DQ Atlas single topic displays for Total FFS Expenditures , FFS Inpatient Expenditures , FFS Long-term Care Expenditures , FFS Other Medical Expenditures , and FFS Prescription Drug Expenditures .
In addition to FFS claims and monthly beneficiary payments, TAF also includes other types of payments, such as supplemental payment records and service tracking claims. Supplemental payment records represent payments for a specific service made in addition to a capitation payment or negotiated rate. Service tracking claims represent lump-sum payments made for services that cannot be attributed to a specific beneficiary, such as disproportionate share hospital (DSH) payments or payments to providers made under the Upper Payment Limit (UPL) demonstration.
PHPs often cover a specific type of service, such as behavioral health care or dental care.
Some Medicaid beneficiaries qualify for employer-based coverage, and if this coverage is less costly than enrolling the beneficiary in traditional Medicaid, states may pay the premium to the employer on behalf of the beneficiary. Some states have used other policy options, such as 1115 waiver demonstrations or the Basic Health Program, to enroll eligible beneficiaries in private plans available through the state or federal Health Insurance Exchange.
Medicaid and CHIP Payment and Access Commission. \u201cEXHIBIT 17. Total Medicaid Benefit Spending by State and Category, FY 2017 (millions).\u201d MACStats: Medicaid and CHIP Data Book December 2018. Washington, DC: MACPAC, December 2018, pp. 52\u201354. Available at https://www.macpac.gov/wp-content/uploads/2018/12/December-2018-MACStats-Data-Book.pdf . Accessed June 24, 2019.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We compared monthly expenditures for Medicaid beneficiaries drawn from twelve months of TAF data [8] to the CMS-64 benchmark for four quarters of the calendar year.
To calculate monthly beneficiary expenditures in the TAF, we first selected records in the other services (OT) file that had a claim type code (CLM_TYPE_CD) indicating that the record was a capitation payment made on behalf of a Medicaid beneficiary. [9] In TAF, all monthly beneficiary payments should be reported on records with a claim type code of capitation payment; these records include some types of payments that are not typically referred to as capitation such as PCCM monthly fees. Because we found that some states report monthly beneficiary payments by using other claim types, we also included records with claim type code values of 4 (service tracking claim) [10] and 5 (supplemental payments) if they had a federally assigned service category (FASC) code [11] related to monthly beneficiary payments (FED_SRVC_CTGRY_CD values 11 or 12) or at least one line with a type of service code related to monthly beneficiary payments (TOS_CD values of 119, 120, 121, 122, 138, 143, or 144 [12] ).
Among records with claim type 2 or 5, we only kept those that matched to an eligibility record in the Demographic and Eligibility (DE) file that indicated that the beneficiary was enrolled in non-CHIP Medicaid at some point during the calendar year. [13] This is necessary because the CMS-64 benchmark includes only non-CHIP Medicaid expenditures, and the TAF claim type variable alone cannot distinguish payments made for Medicaid-expansion CHIP enrollees (whose costs are not reported on the CMS-64) from payments made for non-CHIP Medicaid beneficiaries (whose costs are reported on the CMS-64). We did not require records with claim type 4 to match to a Medicaid eligibility record because these types of records are lump-sum payments that are not connected to any one beneficiary and should not have a valid beneficiary identifier.
Finally, we excluded TAF records that represent Disproportionate Share Hospital (DSH) payments, supplemental payments made under the Upper Payment Limit (UPL) demonstration, electronic health record (EHR) payments, and drug rebates. [14] These payments were excluded because they do not represent monthly beneficiary payments for health services. Because very few states are reporting Medicare premium payments in T-MSIS, we also excluded Medicare premium payments. [15]
Once we had selected the set of TAF records included in the benchmarking analysis, we tabulated expenditures associated with monthly beneficiary payments in each state using the total Medicaid paid amount on the header record (TOT_MDCD_PD_AMT) for all included records with claim type codes values of 2 or 5. For records with a claim type code value of 4, we used the service tracking payment amount (SRVC_TRKNG_PYMT_AMT) to tabulate expenditures. If the service tracking payment amount was missing, we used the total Medicaid paid amount.
We calculated the CMS-64 benchmark for monthly beneficiary payments using four quarterly net expenditures reports from MBES that cover a calendar year. [16] We summed the net expenditures for the categories of service covering monthly payments for CMCs, PHPs, PCCMs, and Medicaid premium assistance payments, using the category of service codes shown in Table 1. We excluded Medicare premium payments from the benchmark for monthly beneficiary payments, as well as any CMS-64 categories that did not represent monthly beneficiary payments, including services typically paid on a FFS basis, drug rebates, DSH payments, General Medical Education payments, and other supplemental payments.
Benchmarking by type of monthly beneficiary payment
We examined TAF-based expenditures for total monthly beneficiary payments as well as for four specific types of payment: (1) CMC capitation payments, (2) PHP capitation payments, (3) PCCM flat fees, and (4) premium assistance payments to private plans. A small number of records representing “other” monthly beneficiary payments are included in the total expenditure calculation but are not captured in these four specific payment types. [17]
We classified TAF records into each of these four categories if the record had at least one line with a federally assigned service category (FASC) code and a type of service (TOS) code indicating the given payment category, as shown in Table 1. [18] , [19] , [20]
Table 1. TAF records and CMS-64 categories that represent monthly beneficiary payments
| Payment type | TAF records | CMS-64 categories |
|---|---|---|
| Payments to CMC plans | Records with a FED_SRVC_CTGRY_CD value of 11 and TOS_CD values of 119. These represent capitated payments to HMOs, HIOs, or PACE plans. | Categories of service 18A, 18A1, 18A2, 18A3, 18A4, 18A5, 18A6, and 22. a These categories represent Medicaid MCO, Medicaid MCO - Evaluation and Management, Medicaid MCO - Vaccine Codes, Medicaid MCO - Community First Choice, Medicaid MCO - Preventive Services, Medicaid MCO - Certified Community Behavior Health Clinic Payment, Medicaid MCO - Services Subject to Electronic Visit Verification Requirements, and Programs of All-Inclusive Care for the Elderly (PACE). |
| Payments to PHPs | Records with a FED_SRVC_CTGRY_CD value of 11 and TOS_CD values of 122. These represent capitated payments to prepaid health plans (PHPs). | Categories of service 18B1, 18B1a, 18B1b, 18B1c, 18B1d, 18B1e, 18B1f 18B2, 18B2a, 18B2b, 18B2c, 18B2d, 18B2e and 18B2f a . These categories represent Prepaid Ambulatory Health Plans (PAHP), PAHP - Evaluation and Management, PAHP - Vaccine Codes, PAHP - Community First Choice, PAHP - Preventive Services, Medicaid PAHP - Certified Community Behavior Health Clinic Payments, MCO PAHP - Services Subject to Electronic Visit Verification Requirements, Prepaid Inpatient Health Plans (PIHP), PIHP - Evaluation and Management, PIHP - Vaccine Codes, PIHP - Community First Choice, PIHP - Preventive Services, Medicaid PIHP - Certified Community Behavior Health Clinic Payments, and MCO PIHP - Services Subject to Electronic Visit Verification Requirements |
| Payments for premium assistance | Records with a FED_SRVC_CTGRY_CD value of 12 and TOS_CD values of 121. These represent premium payments by Medicaid agencies for private health insurance. | Categories of service 18C and 18E. These categories represent Medicaid - Group Health and Medicaid – Other. |
| Payments for PCCM | Records with a FED_SRVC_CTGRY_CD value of 12 and TOS_CD values of 120. These represent flat fees for primary care case management (PCCM) services. | Category of service 25. This category represents Primary Care Case Management. |
Note: \tSome states use managed care in their 1915(c) Home and Community-Based Service waiver programs. Some of these states may report a portion of their managed care capitation payments related to 1915(c) managed care services under the CMS-64 table category code 19 (Home & Community-Based Services). This category is not captured in our benchmark because most of these waiver services are not delivered through managed care. In T-MSIS, the capitation payment record does not distinguish between the portions of the capitation payment allocated to each CMS-64 category. When this occurs, the TAF data will not align with the CMS-64 benchmark.
a CMS-64 categories of service 18A5 (Medicaid MCO - Certified Community Behavior Health Clinic Payment), 18B1e (Medicaid PAHP - Certified Community Behavior Health Clinic Payments), and 18B2e (Medicaid PIHP - Certified Community Behavior Health Clinic Payments) are valid starting in 2017. CMS-64 categories of service 18A6, 18B1f, and 18B2f (Services Subject to Electronic Visit Verification Requirements for Medicaid MCO, MCO PAHP, and MCO PIHP, respectively) are valid starting in 2021.
Data quality assessment criteria
We categorized each state’s monthly beneficiary payment expenditures in TAF as having high, moderate, low, or very low alignment with the CMS-64 benchmark based on the percent difference between the two data sources (Table 2). [21] States with a low absolute percentage difference between the expenditures captured in TAF and the CMS-64 are considered to present a low level of concern about the completeness and quality of their TAF data. However, TAF users should be aware that in states with large Medicaid programs, even a small percentage difference may represent a substantial difference in the dollar amount between the data sources.
Table 2. Criteria for DQ assessment of monthly beneficiary payments
| Percent difference between the monthly beneficiary payments captured in TAF and the CMS-64 | Level of alignment | DQ assessment |
|---|---|---|
| x < 5 percent | High | Low concern |
| 5 percent ≤ x < 20 percent | Moderate | Medium concern |
| 20 percent ≤ x < 50 percent | Low | High concern |
| x ≥ 50 percent or if state has no TAF data | Very low | Unusable |
There are three reasons other than the quality and completeness of TAF expenditure data that could result in a state’s monthly beneficiary payment expenditures in the TAF differing from the CMS-64: (1) structural differences in the data sets, such as organizing the data by service date versus claim payment date; (2) data quality problems in the TAF that are unrelated to expenditures, such as a low volume of claims in the TAF because of missing date values on claims or issues with assigning a final action status to claims; and (3) variation in how states report certain types of monthly beneficiary payments on the CMS-64 that are difficult to reproduce in the TAF data. [22] For instance, states using managed care in their 1915(c) Home and Community Based Services (HCBS) waiver programs may elect to report those expenditures in the CMS-64 category for HCBS services rather than in the CMS-64 category for managed care expenditures. When this occurs, we would expect to see a difference between the TAF-based monthly beneficiary expenditures and the CMS-64 benchmark even if TAF data on monthly beneficiary payments are complete and accurate.
Methods previously used to assess data quality
Table 3 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 3. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We included in the analysis Medicaid capitation records (identified by CLAIM_TYPE_CD=2). In the early years of T-MSIS, a few states reported expenditures for Medicaid beneficiaries covered under Title XIX on “other” (non-Medicaid, non-CHIP) capitation records (those with CLAIM_TYPE_CD=V). Because it is unclear what those records represent, we exclude them from this analysis.
Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
Type of service codes 138, 143, and 144 are valid starting in 2020.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the beneficiary had at least one month during the calendar year in which the CHIP code value was 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-75, we considered the beneficiary to have non-CHIP Medicaid enrollment during the year.
We used federally assigned service category code to identify and exclude DSH payments (13), and type of service code to identify and exclude supplemental payments (132, 133, 134) and EHR payments (135). Drug rebates were identified as records were any line had a type of service code of 131 or a service tracking type code (SRVC_TRKNG_TYPE_CD) of 01.
Medicare premiums are identified in TAF as records with a TOS code of 139, 140, 141, or 142 (valid values starting in 2020).
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
“Other” monthly beneficiary payments are identified in TAF as (1) records with claim type 2 where no claim line has a TOS code value of 119, 120, 121, 122, 123, 131, 132, or 135, or (2) any remaining records with claim type code 4 or 5 with a FASC code value of 12 or a TOS code value of 138, 143, or 144. TOS codes 138, 143, and 144 are valid starting in 2020. In contrast, all of the CMS-64 categories of service related to monthly beneficiary payments fall into one of the four specific types of monthly beneficiary payments.
Because the TAF type of service code does not provide a direct one-to-one mapping to the CMS-64 categories, we excluded payments for CMS-64 categories that did not map directly to the categories of CMC, PHP, PCCM, and premium assistance from these sub-analyses.
There are multiple variables that could be used to identify spending associated with different types of monthly beneficiary payments. The two that perform the best are the type of service code and the managed care plan type. The latter provides a finer level of detail than type of service code, but the user will have to link the TAF data from the other services (OT) file to the TAF Managed Care Plan file, which may not be available to all users. Another variable that can be used but that did not perform well in our analyses, is the Title XIX category of service code (XIX_SRVC_CTGRY_CD), which provides a direct one-to-one mapping of TAF line records to the CMS-64 categories of service. However, many states do not report this information consistently, and the high rates of missing values limit the use of this code for meaningful analyses of expenditures.
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: [(TAF – benchmark) / benchmark]*100.
For more information about the differences between TAF and CMS-64 expenditure data, see the methodology brief, “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We included in the analysis Medicaid capitation records (identified by CLAIM_TYPE_CD=2). In the early years of T-MSIS, a few states reported expenditures for Medicaid beneficiaries covered under Title XIX on \u201cother\u201d (non-Medicaid, non-CHIP) capitation records (those with CLAIM_TYPE_CD=V). Because it is unclear what those records represent, we exclude them from this analysis.
Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
Type of service codes 138, 143, and 144 are valid starting in 2020.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the beneficiary had at least one month during the calendar year in which the CHIP code value was 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-75, we considered the beneficiary to have non-CHIP Medicaid enrollment during the year.
We used federally assigned service category code to identify and exclude DSH payments (13), and type of service code to identify and exclude supplemental payments (132, 133, 134) and EHR payments (135). Drug rebates were identified as records were any line had a type of service code of 131 or a service tracking type code (SRVC_TRKNG_TYPE_CD) of 01.
Medicare premiums are identified in TAF as records with a TOS code of 139, 140, 141, or 142 (valid values starting in 2020).
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
\u201cOther\u201d monthly beneficiary payments are identified in TAF as (1) records with claim type 2 where no claim line has a TOS code value of 119, 120, 121, 122, 123, 131, 132, or 135, or (2) any remaining records with claim type code 4 or 5 with a FASC code value of 12 or a TOS code value of 138, 143, or 144. TOS codes 138, 143, and 144 are valid starting in 2020. In contrast, all of the CMS-64 categories of service related to monthly beneficiary payments fall into one of the four specific types of monthly beneficiary payments.
Because the TAF type of service code does not provide a direct one-to-one mapping to the CMS-64 categories, we excluded payments for CMS-64 categories that did not map directly to the categories of CMC, PHP, PCCM, and premium assistance from these sub-analyses.
There are multiple variables that could be used to identify spending associated with different types of monthly beneficiary payments. The two that perform the best are the type of service code and the managed care plan type. The latter provides a finer level of detail than type of service code, but the user will have to link the TAF data from the other services (OT) file to the TAF Managed Care Plan file, which may not be available to all users. Another variable that can be used but that did not perform well in our analyses, is the Title XIX category of service code (XIX_SRVC_CTGRY_CD), which provides a direct one-to-one mapping of TAF line records to the CMS-64 categories of service. However, many states do not report this information consistently, and the high rates of missing values limit the use of this code for meaningful analyses of expenditures.
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: [(TAF \u2013 benchmark) / benchmark]*100.
For more information about the differences between TAF and CMS-64 expenditure data, see the methodology brief, \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
The TAF should include all state spending to provide or cover care for Medicaid beneficiaries, including monthly capitation payments to prepaid health plans (PHPs). Under PHP arrangements, the plan is responsible for delivering a narrower, non-comprehensive set of inpatient or outpatient services to beneficiaries in exchange for a monthly capitation payment from the state. This analysis examines how well the total expenditures for monthly PHP capitation payments on behalf of Medicaid beneficiaries align with an external benchmark, the CMS-64.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""6031"", ""relatedTopics"": [{""measureId"": 72, ""measureName"": ""Total Monthly Beneficiary Payments"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 0}, {""measureId"": 73, ""measureName"": ""CMC Payments"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 1}, {""measureId"": 75, ""measureName"": ""PCCM Fees"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 3}, {""measureId"": 76, ""measureName"": ""Premium Assistance Payments"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 4}]}" 75,"{""measureId"": 75, ""measureName"": ""PCCM Fees"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-PCCM-Fees.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) provide a unique source of expenditure information on states’ Medicaid and CHIP programs. Although other data sources provide aggregate state-level expenditure information, the TAF allow users to analyze payments associated with managed care organizations, individual providers, subgroups of beneficiaries, and specific services. The Centers for Medicare & Medicaid Services (CMS) and other TAF users require accurate payment data to calculate total cost of care metrics; to compare costs for specific conditions, services, or populations; and to compare costs across states or different payment systems.
States may reimburse health care providers for services delivered to Medicaid beneficiaries by paying directly for each covered service on a fee-for-service (FFS) basis or by paying a flat monthly payment per beneficiary for a set of services that is contracted to another entity—such as a managed care plan—which then assumes responsibility for delivering care to the beneficiary. [1] , [2] When the state contracts directly with a managed care plan, the covered services may either be comprehensive and provided by a Medicaid managed care organization (MCO), or they may be narrower sets of inpatient or outpatient services provided by a prepaid health plan (PHP). [3] Payments associated with these types of Medicaid managed care are referred to as capitation payments. States also make other monthly payments on behalf of Medicaid beneficiaries, including a flat fee paid to a primary care provider for primary care case management plan (PCCM) services; Medicare Part A and Part B premiums for Medicaid beneficiaries who are dually eligible for Medicare; and in some cases, premium assistance for Medicaid beneficiaries with private coverage. [4]
Most states use a mix of FFS and monthly beneficiary payments to pay for the care provided to Medicaid beneficiaries. In fiscal year 2017, monthly beneficiary payments accounted for 53 percent of total Medicaid spending nationally on health services. [5] Across most states, the vast majority of these payments were for comprehensive managed care plans, which accounted for nearly half of all Medicaid expenditures. Spending on PHPs, PCCM, and premium assistance accounted for a far smaller share of total Medicaid expenditures, and Medicare premiums paid on behalf of dually eligible beneficiaries accounted for about 3 percent of total Medicaid spending.
In order to claim federal matching funds for their medical assistance programs, states report aggregate expenditure data for Medicaid beneficiaries covered under Title XIX of the Social Security Act to CMS through the Medicaid Budget and Expenditure System (MBES). [6] The MBES system is designed to capture information states would report on the Quarterly Medicaid Statement of Expenditures for the Medical Assistance Program Form CMS-64; therefore, the data are commonly referred to as the CMS-64 data. Because of the oversight involved in ensuring the correct federal reimbursement to state Medicaid programs, the CMS-64 data are seen as a highly reliable source of expenditure information. If the expenditures in the TAF do not align with those in the CMS-64, that may suggest TAF data are incomplete or have other data quality errors that render them unrepresentative of actual Medicaid expenditures.
This data quality assessment examines the extent to which monthly beneficiary payments captured on states’ TAF records align with the total CMS-64 expenditures for monthly beneficiary payments both overall and for certain types of monthly payments: (1) comprehensive managed care (CMC) capitation payments, (2) PHP capitation payments, (3) PCCM flat fees, and (4) premium assistance payments to private plans. However, benchmarking TAF expenditure data against CMS-64 data is not straightforward. Although the data in both systems are similar, there are some important differences: [7]
More information on FFS expenditure data completeness can be found in the DQ Atlas single topic displays for Total FFS Expenditures , FFS Inpatient Expenditures , FFS Long-term Care Expenditures , FFS Other Medical Expenditures , and FFS Prescription Drug Expenditures .
In addition to FFS claims and monthly beneficiary payments, TAF also includes other types of payments, such as supplemental payment records and service tracking claims. Supplemental payment records represent payments for a specific service made in addition to a capitation payment or negotiated rate. Service tracking claims represent lump-sum payments made for services that cannot be attributed to a specific beneficiary, such as disproportionate share hospital (DSH) payments or payments to providers made under the Upper Payment Limit (UPL) demonstration.
PHPs often cover a specific type of service, such as behavioral health care or dental care.
Some Medicaid beneficiaries qualify for employer-based coverage, and if this coverage is less costly than enrolling the beneficiary in traditional Medicaid, states may pay the premium to the employer on behalf of the beneficiary. Some states have used other policy options, such as 1115 waiver demonstrations or the Basic Health Program, to enroll eligible beneficiaries in private plans available through the state or federal Health Insurance Exchange.
Medicaid and CHIP Payment and Access Commission. “EXHIBIT 17. Total Medicaid Benefit Spending by State and Category, FY 2017 (millions).” MACStats: Medicaid and CHIP Data Book December 2018. Washington, DC: MACPAC, December 2018, pp. 52–54. Available at https://www.macpac.gov/wp-content/uploads/2018/12/December-2018-MACStats-Data-Book.pdf . Accessed June 24, 2019.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
More information on FFS expenditure data completeness can be found in the DQ Atlas single topic displays for Total FFS Expenditures , FFS Inpatient Expenditures , FFS Long-term Care Expenditures , FFS Other Medical Expenditures , and FFS Prescription Drug Expenditures .
In addition to FFS claims and monthly beneficiary payments, TAF also includes other types of payments, such as supplemental payment records and service tracking claims. Supplemental payment records represent payments for a specific service made in addition to a capitation payment or negotiated rate. Service tracking claims represent lump-sum payments made for services that cannot be attributed to a specific beneficiary, such as disproportionate share hospital (DSH) payments or payments to providers made under the Upper Payment Limit (UPL) demonstration.
PHPs often cover a specific type of service, such as behavioral health care or dental care.
Some Medicaid beneficiaries qualify for employer-based coverage, and if this coverage is less costly than enrolling the beneficiary in traditional Medicaid, states may pay the premium to the employer on behalf of the beneficiary. Some states have used other policy options, such as 1115 waiver demonstrations or the Basic Health Program, to enroll eligible beneficiaries in private plans available through the state or federal Health Insurance Exchange.
Medicaid and CHIP Payment and Access Commission. \u201cEXHIBIT 17. Total Medicaid Benefit Spending by State and Category, FY 2017 (millions).\u201d MACStats: Medicaid and CHIP Data Book December 2018. Washington, DC: MACPAC, December 2018, pp. 52\u201354. Available at https://www.macpac.gov/wp-content/uploads/2018/12/December-2018-MACStats-Data-Book.pdf . Accessed June 24, 2019.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We compared monthly expenditures for Medicaid beneficiaries drawn from twelve months of TAF data [8] to the CMS-64 benchmark for four quarters of the calendar year.
To calculate monthly beneficiary expenditures in the TAF, we first selected records in the other services (OT) file that had a claim type code (CLM_TYPE_CD) indicating that the record was a capitation payment made on behalf of a Medicaid beneficiary. [9] In TAF, all monthly beneficiary payments should be reported on records with a claim type code of capitation payment; these records include some types of payments that are not typically referred to as capitation such as PCCM monthly fees. Because we found that some states report monthly beneficiary payments by using other claim types, we also included records with claim type code values of 4 (service tracking claim) [10] and 5 (supplemental payments) if they had a federally assigned service category (FASC) code [11] related to monthly beneficiary payments (FED_SRVC_CTGRY_CD values 11 or 12) or at least one line with a type of service code related to monthly beneficiary payments (TOS_CD values of 119, 120, 121, 122, 138, 143, or 144 [12] ).
Among records with claim type 2 or 5, we only kept those that matched to an eligibility record in the Demographic and Eligibility (DE) file that indicated that the beneficiary was enrolled in non-CHIP Medicaid at some point during the calendar year. [13] This is necessary because the CMS-64 benchmark includes only non-CHIP Medicaid expenditures, and the TAF claim type variable alone cannot distinguish payments made for Medicaid-expansion CHIP enrollees (whose costs are not reported on the CMS-64) from payments made for non-CHIP Medicaid beneficiaries (whose costs are reported on the CMS-64). We did not require records with claim type 4 to match to a Medicaid eligibility record because these types of records are lump-sum payments that are not connected to any one beneficiary and should not have a valid beneficiary identifier.
Finally, we excluded TAF records that represent Disproportionate Share Hospital (DSH) payments, supplemental payments made under the Upper Payment Limit (UPL) demonstration, electronic health record (EHR) payments, and drug rebates. [14] These payments were excluded because they do not represent monthly beneficiary payments for health services. Because very few states are reporting Medicare premium payments in T-MSIS, we also excluded Medicare premium payments. [15]
Once we had selected the set of TAF records included in the benchmarking analysis, we tabulated expenditures associated with monthly beneficiary payments in each state using the total Medicaid paid amount on the header record (TOT_MDCD_PD_AMT) for all included records with claim type codes values of 2 or 5. For records with a claim type code value of 4, we used the service tracking payment amount (SRVC_TRKNG_PYMT_AMT) to tabulate expenditures. If the service tracking payment amount was missing, we used the total Medicaid paid amount.
We calculated the CMS-64 benchmark for monthly beneficiary payments using four quarterly net expenditures reports from MBES that cover a calendar year. [16] We summed the net expenditures for the categories of service covering monthly payments for CMCs, PHPs, PCCMs, and Medicaid premium assistance payments, using the category of service codes shown in Table 1. We excluded Medicare premium payments from the benchmark for monthly beneficiary payments, as well as any CMS-64 categories that did not represent monthly beneficiary payments, including services typically paid on a FFS basis, drug rebates, DSH payments, General Medical Education payments, and other supplemental payments.
Benchmarking by type of monthly beneficiary payment
We examined TAF-based expenditures for total monthly beneficiary payments as well as for four specific types of payment: (1) CMC capitation payments, (2) PHP capitation payments, (3) PCCM flat fees, and (4) premium assistance payments to private plans. A small number of records representing “other” monthly beneficiary payments are included in the total expenditure calculation but are not captured in these four specific payment types. [17]
We classified TAF records into each of these four categories if the record had at least one line with a federally assigned service category (FASC) code and a type of service (TOS) code indicating the given payment category, as shown in Table 1. [18] , [19] , [20]
Table 1. TAF records and CMS-64 categories that represent monthly beneficiary payments
| Payment type | TAF records | CMS-64 categories |
|---|---|---|
| Payments to CMC plans | Records with a FED_SRVC_CTGRY_CD value of 11 and TOS_CD values of 119. These represent capitated payments to HMOs, HIOs, or PACE plans. | Categories of service 18A, 18A1, 18A2, 18A3, 18A4, 18A5, 18A6, and 22. a These categories represent Medicaid MCO, Medicaid MCO - Evaluation and Management, Medicaid MCO - Vaccine Codes, Medicaid MCO - Community First Choice, Medicaid MCO - Preventive Services, Medicaid MCO - Certified Community Behavior Health Clinic Payment, Medicaid MCO - Services Subject to Electronic Visit Verification Requirements, and Programs of All-Inclusive Care for the Elderly (PACE). |
| Payments to PHPs | Records with a FED_SRVC_CTGRY_CD value of 11 and TOS_CD values of 122. These represent capitated payments to prepaid health plans (PHPs). | Categories of service 18B1, 18B1a, 18B1b, 18B1c, 18B1d, 18B1e, 18B1f 18B2, 18B2a, 18B2b, 18B2c, 18B2d, 18B2e and 18B2f a . These categories represent Prepaid Ambulatory Health Plans (PAHP), PAHP - Evaluation and Management, PAHP - Vaccine Codes, PAHP - Community First Choice, PAHP - Preventive Services, Medicaid PAHP - Certified Community Behavior Health Clinic Payments, MCO PAHP - Services Subject to Electronic Visit Verification Requirements, Prepaid Inpatient Health Plans (PIHP), PIHP - Evaluation and Management, PIHP - Vaccine Codes, PIHP - Community First Choice, PIHP - Preventive Services, Medicaid PIHP - Certified Community Behavior Health Clinic Payments, and MCO PIHP - Services Subject to Electronic Visit Verification Requirements |
| Payments for premium assistance | Records with a FED_SRVC_CTGRY_CD value of 12 and TOS_CD values of 121. These represent premium payments by Medicaid agencies for private health insurance. | Categories of service 18C and 18E. These categories represent Medicaid - Group Health and Medicaid – Other. |
| Payments for PCCM | Records with a FED_SRVC_CTGRY_CD value of 12 and TOS_CD values of 120. These represent flat fees for primary care case management (PCCM) services. | Category of service 25. This category represents Primary Care Case Management. |
Note: \tSome states use managed care in their 1915(c) Home and Community-Based Service waiver programs. Some of these states may report a portion of their managed care capitation payments related to 1915(c) managed care services under the CMS-64 table category code 19 (Home & Community-Based Services). This category is not captured in our benchmark because most of these waiver services are not delivered through managed care. In T-MSIS, the capitation payment record does not distinguish between the portions of the capitation payment allocated to each CMS-64 category. When this occurs, the TAF data will not align with the CMS-64 benchmark.
a CMS-64 categories of service 18A5 (Medicaid MCO - Certified Community Behavior Health Clinic Payment), 18B1e (Medicaid PAHP - Certified Community Behavior Health Clinic Payments), and 18B2e (Medicaid PIHP - Certified Community Behavior Health Clinic Payments) are valid starting in 2017. CMS-64 categories of service 18A6, 18B1f, and 18B2f (Services Subject to Electronic Visit Verification Requirements for Medicaid MCO, MCO PAHP, and MCO PIHP, respectively) are valid starting in 2021.
Data quality assessment criteria
We categorized each state’s monthly beneficiary payment expenditures in TAF as having high, moderate, low, or very low alignment with the CMS-64 benchmark based on the percent difference between the two data sources (Table 2). [21] States with a low absolute percentage difference between the expenditures captured in TAF and the CMS-64 are considered to present a low level of concern about the completeness and quality of their TAF data. However, TAF users should be aware that in states with large Medicaid programs, even a small percentage difference may represent a substantial difference in the dollar amount between the data sources.
Table 2. Criteria for DQ assessment of monthly beneficiary payments
| Percent difference between the monthly beneficiary payments captured in TAF and the CMS-64 | Level of alignment | DQ assessment |
|---|---|---|
| x < 5 percent | High | Low concern |
| 5 percent ≤ x < 20 percent | Moderate | Medium concern |
| 20 percent ≤ x < 50 percent | Low | High concern |
| x ≥ 50 percent or if state has no TAF data | Very low | Unusable |
There are three reasons other than the quality and completeness of TAF expenditure data that could result in a state’s monthly beneficiary payment expenditures in the TAF differing from the CMS-64: (1) structural differences in the data sets, such as organizing the data by service date versus claim payment date; (2) data quality problems in the TAF that are unrelated to expenditures, such as a low volume of claims in the TAF because of missing date values on claims or issues with assigning a final action status to claims; and (3) variation in how states report certain types of monthly beneficiary payments on the CMS-64 that are difficult to reproduce in the TAF data. [22] For instance, states using managed care in their 1915(c) Home and Community Based Services (HCBS) waiver programs may elect to report those expenditures in the CMS-64 category for HCBS services rather than in the CMS-64 category for managed care expenditures. When this occurs, we would expect to see a difference between the TAF-based monthly beneficiary expenditures and the CMS-64 benchmark even if TAF data on monthly beneficiary payments are complete and accurate.
Methods previously used to assess data quality
Table 3 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 3. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We included in the analysis Medicaid capitation records (identified by CLAIM_TYPE_CD=2). In the early years of T-MSIS, a few states reported expenditures for Medicaid beneficiaries covered under Title XIX on “other” (non-Medicaid, non-CHIP) capitation records (those with CLAIM_TYPE_CD=V). Because it is unclear what those records represent, we exclude them from this analysis.
Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
Type of service codes 138, 143, and 144 are valid starting in 2020.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the beneficiary had at least one month during the calendar year in which the CHIP code value was 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-75, we considered the beneficiary to have non-CHIP Medicaid enrollment during the year.
We used federally assigned service category code to identify and exclude DSH payments (13), and type of service code to identify and exclude supplemental payments (132, 133, 134) and EHR payments (135). Drug rebates were identified as records were any line had a type of service code of 131 or a service tracking type code (SRVC_TRKNG_TYPE_CD) of 01.
Medicare premiums are identified in TAF as records with a TOS code of 139, 140, 141, or 142 (valid values starting in 2020).
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
“Other” monthly beneficiary payments are identified in TAF as (1) records with claim type 2 where no claim line has a TOS code value of 119, 120, 121, 122, 123, 131, 132, or 135, or (2) any remaining records with claim type code 4 or 5 with a FASC code value of 12 or a TOS code value of 138, 143, or 144. TOS codes 138, 143, and 144 are valid starting in 2020. In contrast, all of the CMS-64 categories of service related to monthly beneficiary payments fall into one of the four specific types of monthly beneficiary payments.
Because the TAF type of service code does not provide a direct one-to-one mapping to the CMS-64 categories, we excluded payments for CMS-64 categories that did not map directly to the categories of CMC, PHP, PCCM, and premium assistance from these sub-analyses.
There are multiple variables that could be used to identify spending associated with different types of monthly beneficiary payments. The two that perform the best are the type of service code and the managed care plan type. The latter provides a finer level of detail than type of service code, but the user will have to link the TAF data from the other services (OT) file to the TAF Managed Care Plan file, which may not be available to all users. Another variable that can be used but that did not perform well in our analyses, is the Title XIX category of service code (XIX_SRVC_CTGRY_CD), which provides a direct one-to-one mapping of TAF line records to the CMS-64 categories of service. However, many states do not report this information consistently, and the high rates of missing values limit the use of this code for meaningful analyses of expenditures.
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: [(TAF – benchmark) / benchmark]*100.
For more information about the differences between TAF and CMS-64 expenditure data, see the methodology brief, “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We included in the analysis Medicaid capitation records (identified by CLAIM_TYPE_CD=2). In the early years of T-MSIS, a few states reported expenditures for Medicaid beneficiaries covered under Title XIX on \u201cother\u201d (non-Medicaid, non-CHIP) capitation records (those with CLAIM_TYPE_CD=V). Because it is unclear what those records represent, we exclude them from this analysis.
Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
Type of service codes 138, 143, and 144 are valid starting in 2020.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the beneficiary had at least one month during the calendar year in which the CHIP code value was 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-75, we considered the beneficiary to have non-CHIP Medicaid enrollment during the year.
We used federally assigned service category code to identify and exclude DSH payments (13), and type of service code to identify and exclude supplemental payments (132, 133, 134) and EHR payments (135). Drug rebates were identified as records were any line had a type of service code of 131 or a service tracking type code (SRVC_TRKNG_TYPE_CD) of 01.
Medicare premiums are identified in TAF as records with a TOS code of 139, 140, 141, or 142 (valid values starting in 2020).
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
\u201cOther\u201d monthly beneficiary payments are identified in TAF as (1) records with claim type 2 where no claim line has a TOS code value of 119, 120, 121, 122, 123, 131, 132, or 135, or (2) any remaining records with claim type code 4 or 5 with a FASC code value of 12 or a TOS code value of 138, 143, or 144. TOS codes 138, 143, and 144 are valid starting in 2020. In contrast, all of the CMS-64 categories of service related to monthly beneficiary payments fall into one of the four specific types of monthly beneficiary payments.
Because the TAF type of service code does not provide a direct one-to-one mapping to the CMS-64 categories, we excluded payments for CMS-64 categories that did not map directly to the categories of CMC, PHP, PCCM, and premium assistance from these sub-analyses.
There are multiple variables that could be used to identify spending associated with different types of monthly beneficiary payments. The two that perform the best are the type of service code and the managed care plan type. The latter provides a finer level of detail than type of service code, but the user will have to link the TAF data from the other services (OT) file to the TAF Managed Care Plan file, which may not be available to all users. Another variable that can be used but that did not perform well in our analyses, is the Title XIX category of service code (XIX_SRVC_CTGRY_CD), which provides a direct one-to-one mapping of TAF line records to the CMS-64 categories of service. However, many states do not report this information consistently, and the high rates of missing values limit the use of this code for meaningful analyses of expenditures.
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: [(TAF \u2013 benchmark) / benchmark]*100.
For more information about the differences between TAF and CMS-64 expenditure data, see the methodology brief, \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
The TAF should include all state spending to provide or cover care for Medicaid beneficiaries, including monthly fees for primary care case management (PCCM). Under PCCM arrangements, a primary care practitioner provides a core set of case management services in exchange for a flat monthly administrative fee, but all other services continue to be delivered on a fee-for-service basis. This analysis examines how well the total expenditures for monthly PCCM fees on behalf of Medicaid beneficiaries align with an external benchmark, the CMS-64.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""6031"", ""relatedTopics"": [{""measureId"": 72, ""measureName"": ""Total Monthly Beneficiary Payments"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 0}, {""measureId"": 73, ""measureName"": ""CMC Payments"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 1}, {""measureId"": 74, ""measureName"": ""PHP Payments"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 2}, {""measureId"": 76, ""measureName"": ""Premium Assistance Payments"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 4}]}" 76,"{""measureId"": 76, ""measureName"": ""Premium Assistance Payments"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Premium-Assistance-Pmts.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) provide a unique source of expenditure information on states’ Medicaid and CHIP programs. Although other data sources provide aggregate state-level expenditure information, the TAF allow users to analyze payments associated with managed care organizations, individual providers, subgroups of beneficiaries, and specific services. The Centers for Medicare & Medicaid Services (CMS) and other TAF users require accurate payment data to calculate total cost of care metrics; to compare costs for specific conditions, services, or populations; and to compare costs across states or different payment systems.
States may reimburse health care providers for services delivered to Medicaid beneficiaries by paying directly for each covered service on a fee-for-service (FFS) basis or by paying a flat monthly payment per beneficiary for a set of services that is contracted to another entity—such as a managed care plan—which then assumes responsibility for delivering care to the beneficiary. [1] , [2] When the state contracts directly with a managed care plan, the covered services may either be comprehensive and provided by a Medicaid managed care organization (MCO), or they may be narrower sets of inpatient or outpatient services provided by a prepaid health plan (PHP). [3] Payments associated with these types of Medicaid managed care are referred to as capitation payments. States also make other monthly payments on behalf of Medicaid beneficiaries, including a flat fee paid to a primary care provider for primary care case management plan (PCCM) services; Medicare Part A and Part B premiums for Medicaid beneficiaries who are dually eligible for Medicare; and in some cases, premium assistance for Medicaid beneficiaries with private coverage. [4]
Most states use a mix of FFS and monthly beneficiary payments to pay for the care provided to Medicaid beneficiaries. In fiscal year 2017, monthly beneficiary payments accounted for 53 percent of total Medicaid spending nationally on health services. [5] Across most states, the vast majority of these payments were for comprehensive managed care plans, which accounted for nearly half of all Medicaid expenditures. Spending on PHPs, PCCM, and premium assistance accounted for a far smaller share of total Medicaid expenditures, and Medicare premiums paid on behalf of dually eligible beneficiaries accounted for about 3 percent of total Medicaid spending.
In order to claim federal matching funds for their medical assistance programs, states report aggregate expenditure data for Medicaid beneficiaries covered under Title XIX of the Social Security Act to CMS through the Medicaid Budget and Expenditure System (MBES). [6] The MBES system is designed to capture information states would report on the Quarterly Medicaid Statement of Expenditures for the Medical Assistance Program Form CMS-64; therefore, the data are commonly referred to as the CMS-64 data. Because of the oversight involved in ensuring the correct federal reimbursement to state Medicaid programs, the CMS-64 data are seen as a highly reliable source of expenditure information. If the expenditures in the TAF do not align with those in the CMS-64, that may suggest TAF data are incomplete or have other data quality errors that render them unrepresentative of actual Medicaid expenditures.
This data quality assessment examines the extent to which monthly beneficiary payments captured on states’ TAF records align with the total CMS-64 expenditures for monthly beneficiary payments both overall and for certain types of monthly payments: (1) comprehensive managed care (CMC) capitation payments, (2) PHP capitation payments, (3) PCCM flat fees, and (4) premium assistance payments to private plans. However, benchmarking TAF expenditure data against CMS-64 data is not straightforward. Although the data in both systems are similar, there are some important differences: [7]
More information on FFS expenditure data completeness can be found in the DQ Atlas single topic displays for Total FFS Expenditures , FFS Inpatient Expenditures , FFS Long-term Care Expenditures , FFS Other Medical Expenditures , and FFS Prescription Drug Expenditures .
In addition to FFS claims and monthly beneficiary payments, TAF also includes other types of payments, such as supplemental payment records and service tracking claims. Supplemental payment records represent payments for a specific service made in addition to a capitation payment or negotiated rate. Service tracking claims represent lump-sum payments made for services that cannot be attributed to a specific beneficiary, such as disproportionate share hospital (DSH) payments or payments to providers made under the Upper Payment Limit (UPL) demonstration.
PHPs often cover a specific type of service, such as behavioral health care or dental care.
Some Medicaid beneficiaries qualify for employer-based coverage, and if this coverage is less costly than enrolling the beneficiary in traditional Medicaid, states may pay the premium to the employer on behalf of the beneficiary. Some states have used other policy options, such as 1115 waiver demonstrations or the Basic Health Program, to enroll eligible beneficiaries in private plans available through the state or federal Health Insurance Exchange.
Medicaid and CHIP Payment and Access Commission. “EXHIBIT 17. Total Medicaid Benefit Spending by State and Category, FY 2017 (millions).” MACStats: Medicaid and CHIP Data Book December 2018. Washington, DC: MACPAC, December 2018, pp. 52–54. Available at https://www.macpac.gov/wp-content/uploads/2018/12/December-2018-MACStats-Data-Book.pdf . Accessed June 24, 2019.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
More information on FFS expenditure data completeness can be found in the DQ Atlas single topic displays for Total FFS Expenditures , FFS Inpatient Expenditures , FFS Long-term Care Expenditures , FFS Other Medical Expenditures , and FFS Prescription Drug Expenditures .
In addition to FFS claims and monthly beneficiary payments, TAF also includes other types of payments, such as supplemental payment records and service tracking claims. Supplemental payment records represent payments for a specific service made in addition to a capitation payment or negotiated rate. Service tracking claims represent lump-sum payments made for services that cannot be attributed to a specific beneficiary, such as disproportionate share hospital (DSH) payments or payments to providers made under the Upper Payment Limit (UPL) demonstration.
PHPs often cover a specific type of service, such as behavioral health care or dental care.
Some Medicaid beneficiaries qualify for employer-based coverage, and if this coverage is less costly than enrolling the beneficiary in traditional Medicaid, states may pay the premium to the employer on behalf of the beneficiary. Some states have used other policy options, such as 1115 waiver demonstrations or the Basic Health Program, to enroll eligible beneficiaries in private plans available through the state or federal Health Insurance Exchange.
Medicaid and CHIP Payment and Access Commission. \u201cEXHIBIT 17. Total Medicaid Benefit Spending by State and Category, FY 2017 (millions).\u201d MACStats: Medicaid and CHIP Data Book December 2018. Washington, DC: MACPAC, December 2018, pp. 52\u201354. Available at https://www.macpac.gov/wp-content/uploads/2018/12/December-2018-MACStats-Data-Book.pdf . Accessed June 24, 2019.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We compared monthly expenditures for Medicaid beneficiaries drawn from twelve months of TAF data [8] to the CMS-64 benchmark for four quarters of the calendar year.
To calculate monthly beneficiary expenditures in the TAF, we first selected records in the other services (OT) file that had a claim type code (CLM_TYPE_CD) indicating that the record was a capitation payment made on behalf of a Medicaid beneficiary. [9] In TAF, all monthly beneficiary payments should be reported on records with a claim type code of capitation payment; these records include some types of payments that are not typically referred to as capitation such as PCCM monthly fees. Because we found that some states report monthly beneficiary payments by using other claim types, we also included records with claim type code values of 4 (service tracking claim) [10] and 5 (supplemental payments) if they had a federally assigned service category (FASC) code [11] related to monthly beneficiary payments (FED_SRVC_CTGRY_CD values 11 or 12) or at least one line with a type of service code related to monthly beneficiary payments (TOS_CD values of 119, 120, 121, 122, 138, 143, or 144 [12] ).
Among records with claim type 2 or 5, we only kept those that matched to an eligibility record in the Demographic and Eligibility (DE) file that indicated that the beneficiary was enrolled in non-CHIP Medicaid at some point during the calendar year. [13] This is necessary because the CMS-64 benchmark includes only non-CHIP Medicaid expenditures, and the TAF claim type variable alone cannot distinguish payments made for Medicaid-expansion CHIP enrollees (whose costs are not reported on the CMS-64) from payments made for non-CHIP Medicaid beneficiaries (whose costs are reported on the CMS-64). We did not require records with claim type 4 to match to a Medicaid eligibility record because these types of records are lump-sum payments that are not connected to any one beneficiary and should not have a valid beneficiary identifier.
Finally, we excluded TAF records that represent Disproportionate Share Hospital (DSH) payments, supplemental payments made under the Upper Payment Limit (UPL) demonstration, electronic health record (EHR) payments, and drug rebates. [14] These payments were excluded because they do not represent monthly beneficiary payments for health services. Because very few states are reporting Medicare premium payments in T-MSIS, we also excluded Medicare premium payments. [15]
Once we had selected the set of TAF records included in the benchmarking analysis, we tabulated expenditures associated with monthly beneficiary payments in each state using the total Medicaid paid amount on the header record (TOT_MDCD_PD_AMT) for all included records with claim type codes values of 2 or 5. For records with a claim type code value of 4, we used the service tracking payment amount (SRVC_TRKNG_PYMT_AMT) to tabulate expenditures. If the service tracking payment amount was missing, we used the total Medicaid paid amount.
We calculated the CMS-64 benchmark for monthly beneficiary payments using four quarterly net expenditures reports from MBES that cover a calendar year. [16] We summed the net expenditures for the categories of service covering monthly payments for CMCs, PHPs, PCCMs, and Medicaid premium assistance payments, using the category of service codes shown in Table 1. We excluded Medicare premium payments from the benchmark for monthly beneficiary payments, as well as any CMS-64 categories that did not represent monthly beneficiary payments, including services typically paid on a FFS basis, drug rebates, DSH payments, General Medical Education payments, and other supplemental payments.
Benchmarking by type of monthly beneficiary payment
We examined TAF-based expenditures for total monthly beneficiary payments as well as for four specific types of payment: (1) CMC capitation payments, (2) PHP capitation payments, (3) PCCM flat fees, and (4) premium assistance payments to private plans. A small number of records representing “other” monthly beneficiary payments are included in the total expenditure calculation but are not captured in these four specific payment types. [17]
We classified TAF records into each of these four categories if the record had at least one line with a federally assigned service category (FASC) code and a type of service (TOS) code indicating the given payment category, as shown in Table 1. [18] , [19] , [20]
Table 1. TAF records and CMS-64 categories that represent monthly beneficiary payments
| Payment type | TAF records | CMS-64 categories |
|---|---|---|
| Payments to CMC plans | Records with a FED_SRVC_CTGRY_CD value of 11 and TOS_CD values of 119. These represent capitated payments to HMOs, HIOs, or PACE plans. | Categories of service 18A, 18A1, 18A2, 18A3, 18A4, 18A5, 18A6, and 22. a These categories represent Medicaid MCO, Medicaid MCO - Evaluation and Management, Medicaid MCO - Vaccine Codes, Medicaid MCO - Community First Choice, Medicaid MCO - Preventive Services, Medicaid MCO - Certified Community Behavior Health Clinic Payment, Medicaid MCO - Services Subject to Electronic Visit Verification Requirements, and Programs of All-Inclusive Care for the Elderly (PACE). |
| Payments to PHPs | Records with a FED_SRVC_CTGRY_CD value of 11 and TOS_CD values of 122. These represent capitated payments to prepaid health plans (PHPs). | Categories of service 18B1, 18B1a, 18B1b, 18B1c, 18B1d, 18B1e, 18B1f 18B2, 18B2a, 18B2b, 18B2c, 18B2d, 18B2e and 18B2f a . These categories represent Prepaid Ambulatory Health Plans (PAHP), PAHP - Evaluation and Management, PAHP - Vaccine Codes, PAHP - Community First Choice, PAHP - Preventive Services, Medicaid PAHP - Certified Community Behavior Health Clinic Payments, MCO PAHP - Services Subject to Electronic Visit Verification Requirements, Prepaid Inpatient Health Plans (PIHP), PIHP - Evaluation and Management, PIHP - Vaccine Codes, PIHP - Community First Choice, PIHP - Preventive Services, Medicaid PIHP - Certified Community Behavior Health Clinic Payments, and MCO PIHP - Services Subject to Electronic Visit Verification Requirements |
| Payments for premium assistance | Records with a FED_SRVC_CTGRY_CD value of 12 and TOS_CD values of 121. These represent premium payments by Medicaid agencies for private health insurance. | Categories of service 18C and 18E. These categories represent Medicaid - Group Health and Medicaid – Other. |
| Payments for PCCM | Records with a FED_SRVC_CTGRY_CD value of 12 and TOS_CD values of 120. These represent flat fees for primary care case management (PCCM) services. | Category of service 25. This category represents Primary Care Case Management. |
Note: \tSome states use managed care in their 1915(c) Home and Community-Based Service waiver programs. Some of these states may report a portion of their managed care capitation payments related to 1915(c) managed care services under the CMS-64 table category code 19 (Home & Community-Based Services). This category is not captured in our benchmark because most of these waiver services are not delivered through managed care. In T-MSIS, the capitation payment record does not distinguish between the portions of the capitation payment allocated to each CMS-64 category. When this occurs, the TAF data will not align with the CMS-64 benchmark.
a CMS-64 categories of service 18A5 (Medicaid MCO - Certified Community Behavior Health Clinic Payment), 18B1e (Medicaid PAHP - Certified Community Behavior Health Clinic Payments), and 18B2e (Medicaid PIHP - Certified Community Behavior Health Clinic Payments) are valid starting in 2017. CMS-64 categories of service 18A6, 18B1f, and 18B2f (Services Subject to Electronic Visit Verification Requirements for Medicaid MCO, MCO PAHP, and MCO PIHP, respectively) are valid starting in 2021.
Data quality assessment criteria
We categorized each state’s monthly beneficiary payment expenditures in TAF as having high, moderate, low, or very low alignment with the CMS-64 benchmark based on the percent difference between the two data sources (Table 2). [21] States with a low absolute percentage difference between the expenditures captured in TAF and the CMS-64 are considered to present a low level of concern about the completeness and quality of their TAF data. However, TAF users should be aware that in states with large Medicaid programs, even a small percentage difference may represent a substantial difference in the dollar amount between the data sources.
Table 2. Criteria for DQ assessment of monthly beneficiary payments
| Percent difference between the monthly beneficiary payments captured in TAF and the CMS-64 | Level of alignment | DQ assessment |
|---|---|---|
| x < 5 percent | High | Low concern |
| 5 percent ≤ x < 20 percent | Moderate | Medium concern |
| 20 percent ≤ x < 50 percent | Low | High concern |
| x ≥ 50 percent or if state has no TAF data | Very low | Unusable |
There are three reasons other than the quality and completeness of TAF expenditure data that could result in a state’s monthly beneficiary payment expenditures in the TAF differing from the CMS-64: (1) structural differences in the data sets, such as organizing the data by service date versus claim payment date; (2) data quality problems in the TAF that are unrelated to expenditures, such as a low volume of claims in the TAF because of missing date values on claims or issues with assigning a final action status to claims; and (3) variation in how states report certain types of monthly beneficiary payments on the CMS-64 that are difficult to reproduce in the TAF data. [22] For instance, states using managed care in their 1915(c) Home and Community Based Services (HCBS) waiver programs may elect to report those expenditures in the CMS-64 category for HCBS services rather than in the CMS-64 category for managed care expenditures. When this occurs, we would expect to see a difference between the TAF-based monthly beneficiary expenditures and the CMS-64 benchmark even if TAF data on monthly beneficiary payments are complete and accurate.
Methods previously used to assess data quality
Table 3 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 3. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We included in the analysis Medicaid capitation records (identified by CLAIM_TYPE_CD=2). In the early years of T-MSIS, a few states reported expenditures for Medicaid beneficiaries covered under Title XIX on “other” (non-Medicaid, non-CHIP) capitation records (those with CLAIM_TYPE_CD=V). Because it is unclear what those records represent, we exclude them from this analysis.
Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
Type of service codes 138, 143, and 144 are valid starting in 2020.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the beneficiary had at least one month during the calendar year in which the CHIP code value was 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-75, we considered the beneficiary to have non-CHIP Medicaid enrollment during the year.
We used federally assigned service category code to identify and exclude DSH payments (13), and type of service code to identify and exclude supplemental payments (132, 133, 134) and EHR payments (135). Drug rebates were identified as records were any line had a type of service code of 131 or a service tracking type code (SRVC_TRKNG_TYPE_CD) of 01.
Medicare premiums are identified in TAF as records with a TOS code of 139, 140, 141, or 142 (valid values starting in 2020).
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
“Other” monthly beneficiary payments are identified in TAF as (1) records with claim type 2 where no claim line has a TOS code value of 119, 120, 121, 122, 123, 131, 132, or 135, or (2) any remaining records with claim type code 4 or 5 with a FASC code value of 12 or a TOS code value of 138, 143, or 144. TOS codes 138, 143, and 144 are valid starting in 2020. In contrast, all of the CMS-64 categories of service related to monthly beneficiary payments fall into one of the four specific types of monthly beneficiary payments.
Because the TAF type of service code does not provide a direct one-to-one mapping to the CMS-64 categories, we excluded payments for CMS-64 categories that did not map directly to the categories of CMC, PHP, PCCM, and premium assistance from these sub-analyses.
There are multiple variables that could be used to identify spending associated with different types of monthly beneficiary payments. The two that perform the best are the type of service code and the managed care plan type. The latter provides a finer level of detail than type of service code, but the user will have to link the TAF data from the other services (OT) file to the TAF Managed Care Plan file, which may not be available to all users. Another variable that can be used but that did not perform well in our analyses, is the Title XIX category of service code (XIX_SRVC_CTGRY_CD), which provides a direct one-to-one mapping of TAF line records to the CMS-64 categories of service. However, many states do not report this information consistently, and the high rates of missing values limit the use of this code for meaningful analyses of expenditures.
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: [(TAF – benchmark) / benchmark]*100.
For more information about the differences between TAF and CMS-64 expenditure data, see the methodology brief, “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We included in the analysis Medicaid capitation records (identified by CLAIM_TYPE_CD=2). In the early years of T-MSIS, a few states reported expenditures for Medicaid beneficiaries covered under Title XIX on \u201cother\u201d (non-Medicaid, non-CHIP) capitation records (those with CLAIM_TYPE_CD=V). Because it is unclear what those records represent, we exclude them from this analysis.
Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
Type of service codes 138, 143, and 144 are valid starting in 2020.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the beneficiary had at least one month during the calendar year in which the CHIP code value was 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-75, we considered the beneficiary to have non-CHIP Medicaid enrollment during the year.
We used federally assigned service category code to identify and exclude DSH payments (13), and type of service code to identify and exclude supplemental payments (132, 133, 134) and EHR payments (135). Drug rebates were identified as records were any line had a type of service code of 131 or a service tracking type code (SRVC_TRKNG_TYPE_CD) of 01.
Medicare premiums are identified in TAF as records with a TOS code of 139, 140, 141, or 142 (valid values starting in 2020).
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
\u201cOther\u201d monthly beneficiary payments are identified in TAF as (1) records with claim type 2 where no claim line has a TOS code value of 119, 120, 121, 122, 123, 131, 132, or 135, or (2) any remaining records with claim type code 4 or 5 with a FASC code value of 12 or a TOS code value of 138, 143, or 144. TOS codes 138, 143, and 144 are valid starting in 2020. In contrast, all of the CMS-64 categories of service related to monthly beneficiary payments fall into one of the four specific types of monthly beneficiary payments.
Because the TAF type of service code does not provide a direct one-to-one mapping to the CMS-64 categories, we excluded payments for CMS-64 categories that did not map directly to the categories of CMC, PHP, PCCM, and premium assistance from these sub-analyses.
There are multiple variables that could be used to identify spending associated with different types of monthly beneficiary payments. The two that perform the best are the type of service code and the managed care plan type. The latter provides a finer level of detail than type of service code, but the user will have to link the TAF data from the other services (OT) file to the TAF Managed Care Plan file, which may not be available to all users. Another variable that can be used but that did not perform well in our analyses, is the Title XIX category of service code (XIX_SRVC_CTGRY_CD), which provides a direct one-to-one mapping of TAF line records to the CMS-64 categories of service. However, many states do not report this information consistently, and the high rates of missing values limit the use of this code for meaningful analyses of expenditures.
The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: [(TAF \u2013 benchmark) / benchmark]*100.
For more information about the differences between TAF and CMS-64 expenditure data, see the methodology brief, \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
The TAF should include all state spending to provide or cover care for Medicaid beneficiaries, including premium assistance payments to enroll Medicaid beneficiaries in private plans. States can make these payments to enroll qualified beneficiaries in employer-based coverage or in individual private plans available through the state or the federal Health Insurance Exchange. This analysis examines how well the total expenditures for premium assistance payments on behalf of Medicaid beneficiaries align with an external benchmark, the CMS-64.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""6031"", ""relatedTopics"": [{""measureId"": 72, ""measureName"": ""Total Monthly Beneficiary Payments"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 0}, {""measureId"": 73, ""measureName"": ""CMC Payments"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 1}, {""measureId"": 74, ""measureName"": ""PHP Payments"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 2}, {""measureId"": 75, ""measureName"": ""PCCM Fees"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 3}]}" 77,"{""measureId"": 77, ""measureName"": ""Total Medicaid Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Total-Expenditures.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) provide a unique source of expenditure information on states’ Medicaid and CHIP programs. Although other data sources provide aggregate state-level expenditure information, the TAF allow users to analyze payments associated with managed care organizations, individual providers, subgroups of beneficiaries, and specific services. The Centers for Medicare & Medicaid Services (CMS) and other TAF users require accurate payment data to calculate total cost of care metrics; to compare costs for specific conditions, services, or populations; and to compare costs across states or different payment systems.
States may reimburse health care providers for services delivered to Medicaid beneficiaries by paying directly for each covered service on a fee-for-service (FFS) basis or by paying a flat monthly payment per beneficiary for a set of services that is contracted to another entity—such as a managed care plan—which then assumes responsibility for delivering care to the beneficiary. [1] FFS expenditures and monthly beneficiary payments—which can be linked to specific beneficiaries [2] —account for the majority of Medicaid expenditures in every state.
In addition to FFS claims and monthly beneficiary payments, TAF also includes other financial transactions, which are mostly captured on service tracking claims. Service tracking claims represent lump-sum payments made for services that cannot be attributed to a specific beneficiary, such as disproportionate share hospital (DSH) payments or payments to providers made under the Upper Payment Limit (UPL) demonstration. Table 1 summarizes the expenditure information available in TAF by expenditure type.
Table 1. Types of Medicaid expenditures in TAF
| Expenditure type | Description | Linked to specific beneficiaries? |
|---|---|---|
| FFS expenditures |
| Yes |
| Monthly beneficiary payments |
| Yes b |
| Other expenditures |
| No |
a Under federal law, states must pay federally qualified health centers (FQHCs) the difference between a managed care organization’s per-service payment and the amount determined under the state’s prospective payment system. States often report these “wraparound” payments on supplemental payment records (those with claim type code = 5). For more information, refer to the National Association of Community Health Centers: https://cdn1.digitellinc.com/uploads/nachc/articles/b9784c7326bea6f2202ef78a41872e03.pdf .
b Capitation payments reported on service tracking claims (which are a subset of total monthly beneficiary payments) cannot be linked to a specific beneficiary. All other monthly beneficiary payments can be linked to a specific beneficiary.
To claim federal matching funds for their medical assistance programs, states report aggregate expenditure data for Medicaid beneficiaries covered under Title XIX of the Social Security Act to CMS through the Medicaid Budget and Expenditure System (MBES). [3] The MBES system is designed to capture information states would report on the Quarterly Medicaid Statement of Expenditures for the Medical Assistance Program Form CMS-64; therefore, the data are commonly referred to as the CMS-64 data. Because of the oversight involved in ensuring the correct federal reimbursement to state Medicaid programs, the CMS-64 data are seen as a highly reliable source of expenditure information. If the expenditures in the TAF do not align with those in the CMS-64, that may suggest TAF data are incomplete or have other data quality errors that render them unrepresentative of actual Medicaid expenditures.
In this data quality assessment, we evaluate the completeness of total Medicaid expenditures in the TAF by comparing them to the total net expenditures states report in the CMS-64. However, benchmarking TAF expenditure data against CMS-64 data is not straightforward. Although the data in both systems are similar, there are some important differences: [4]
More information on FFS expenditure benchmarking can be found in the DQ Atlas single topic displays for Total FFS Expenditures , FFS Inpatient Expenditures , FFS Long-term Care Expenditures , FFS Other Medical Expenditures , and FFS Prescription Drug Expenditures . More information on expenditure benchmarking for monthly beneficiary payments can be found in the DQ Atlas single topic displays for Total Monthly Beneficiary Payments , CMC Payments , PHP Payments , PCCM Fees , and Premium Assistance Payments .
Capitation payments reported on service tracking claims (which are a subset of total monthly beneficiary payments) cannot be linked to a specific beneficiary but are still considered to be monthly beneficiary payments. All other monthly beneficiary payments, as well as FFS expenditures and wraparound payments, can be linked to a specific beneficiary.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
More information on FFS expenditure benchmarking can be found in the DQ Atlas single topic displays for Total FFS Expenditures , FFS Inpatient Expenditures , FFS Long-term Care Expenditures , FFS Other Medical Expenditures , and FFS Prescription Drug Expenditures . More information on expenditure benchmarking for monthly beneficiary payments can be found in the DQ Atlas single topic displays for Total Monthly Beneficiary Payments , CMC Payments , PHP Payments , PCCM Fees , and Premium Assistance Payments .
Capitation payments reported on service tracking claims (which are a subset of total monthly beneficiary payments) cannot be linked to a specific beneficiary but are still considered to be monthly beneficiary payments. All other monthly beneficiary payments, as well as FFS expenditures and wraparound payments, can be linked to a specific beneficiary.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
All Medicaid expenditures in TAF are used in this analysis. [5] , [6] We began by selecting all records from the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims files that represented FFS payments, capitation payments, service tracking claims, or supplemental payments by using the claim type code. [7] Managed care encounter records were excluded from the analysis because the payments recorded on those claims do not represent costs to the state Medicaid program and are not recorded on the CMS-64. [8]
Next, we further restricted FFS claims, capitation payments, and supplemental payments to those that matched to an eligibility record in the Demographic and Eligibility (DE) file that indicated that the beneficiary was enrolled in non-CHIP Medicaid at some point during the calendar year. [9] , [10] This is necessary because the CMS-64 benchmark includes only non-CHIP Medicaid expenditures, and the TAF claim type variable alone cannot distinguish payments made for Medicaid-expansion CHIP enrollees (whose costs are not reported on the CMS-64) from payments made for non-CHIP Medicaid beneficiaries (whose costs are reported on the CMS-64). [11]
Finally, we excluded TAF records that represent electronic health record (EHR) payments, Medicare premiums, and drug rebates. [12] EHR payments were excluded because they are considered administrative costs and thus not included in the CMS-64 data we use as a benchmark. Medicare premiums and drug rebates rarely appear in the TAF, and so were excluded from both the benchmark and TAF data for this analysis.
Once we had selected the set of TAF records included in the benchmarking analysis, we aggregated expenditures in two ways for each state: (1) total Medicaid expenditures —which includes FFS, monthly beneficiary payments, and other expenditures—and (2) Medicaid beneficiary expenditures , which only includes FFS and monthly beneficiary payments. For FFS claims in the IP file, we used the total Medicaid paid amount minus the DSH payment (both on the header record) to tabulate expenditures; for FFS claims in the OT, LT, and RX files, we used the total Medicaid paid amount only. For capitation and supplemental payment records, we used the total Medicaid paid amount on the header record to tabulate expenditures. For capitation payments on service tracking claims [13] , we used the service tracking payment amount on the header record; if that field was zero or missing, we used the total Medicaid paid amount. For all other expenditures on service tracking claims, we used the DSH payment field [14] ; if the DSH payment field was zero or missing, we used the service tracking payment amount. If the DSH payment field and service tracking payment amount were both zero or missing, we used the total Medicaid paid amount. [15]
We developed the total expenditure benchmark using four quarterly net expenditures reports from MBES that cover a calendar year. [16] We calculated total expenditures in the CMS-64 using the total net expenditures for medical assistance programs reported by each state (category of service code 50 on the CMS-64), minus expenditures that do not typically appear in TAF: Medicare premiums (category of service 17A, 17B, and 17C), drug rebates (category of service 7A1-7A7, 46A1-46A6 [17] ), and collections (adjustments related to third party liability, probate, fraud, waste, and abuse). We calculated Medicaid beneficiary expenditures in the CMS-64 by summing net expenditures for the categories of service that correspond to FFS and monthly beneficiary payments (Table 2).
Distribution of Medicaid spending by expenditure type
Each state’s expenditure data should reflect its unique Medicaid program characteristics, including the extent of Medicaid managed care and the use of DSH and other supplemental provider payments. To help users understand what types of expenditures are most common in each state, we tabulated the percentage of total CMS-64 spending that fell into each of three expenditure types: (1) FFS expenditures, (2) monthly beneficiary payments, and (3) other expenditures (see Table 1 in the background section).
Table 2 shows how we grouped the CMS-64 category of service codes into the three expenditure types used to calculate the distribution of spending in each state. The table also provides the set of TAF records that are expected to map to these three expenditure types. The Total Medicaid Expenditures topic includes all expenditures listed in Table 2, whereas the Medicaid Beneficiary Expenditures topic includes only the FFS and monthly beneficiary payments expenditure types.
Table 2. Assigning CMS-64 categories of service and TAF records to expenditure types
| Expenditure type | CMS-64 categories a | Corresponding TAF records b |
|---|---|---|
| FFS | Category of service 1A. This represents inpatient hospital services. Categories of service 2A, 3A, 4A, and 4B. These represent payments for mental health facility, nursing facility, and intermediate care facility services. Categories of service 2C, 5A, 5C, 5D, 6A, 7, 8, 9A, 10A, 11, 12, 13, 14, 15, 16, 19A, 19B, 19C, 19D, 23A, 23B, 24A, 24B, 26, 27, 28, 29A, 30, 31, 32, 33, 34, 34A, 35, 36, 37A, 38, 39, 40, 41, 42, 43, 44, 45, 46, 46B, 47, 48, 49 and 69. a These cover payments for Certified Community Behavior Health Clinics under the Section 1332 demonstration; physician and surgical services; outpatient hospital services; dental services; other practitioners services; clinic services; prescribed drugs; laboratory and radiological services; home health services; sterilizations; abortions; EPSDT screenings; rural health services; home and community-based services; personal care services; case management services; hospice benefits; emergency services for undocumented aliens; federally qualified health center services; non-emergency medical transportation (regular payments); physical therapy; occupational therapy; services for speech, hearing, and language; prosthetic devices, dentures, and eyeglasses; diagnostic screening and preventive services; preventive services vaccines and their administration; nurse mid-wife services; emergency hospital services; critical access hospital services (regular payments); nurse practitioner services; school-based services; rehabilitative services (not school based); private duty nursing services; freestanding birth center services; health home for enrollees with chronic conditions; tobacco cessation for pregnant women; health home for enrollees with substance abuse disorder; opioid use disorder medication-assisted treatment drugs and services; COVID-19 vaccines and vaccine administration, health homes for children with medically complex conditions, community-based mobile crisis intervention, and other care services. Categories of service 17D and 18D. These represent Medicaid payments on behalf of beneficiaries for deductibles and coinsurance on all services on which Medicare or a private insurance plan is the primary payer. | All LT claims with CLM_TYPE_CD=1, except those with a FED_SRVC_CTGRY_CD value of 13 or a TOS_CD value of 132, 133, or 134. This includes FFS claims in the LT file, except those that represent UPL supplemental payments and DSH payments, which correspond to the other expenditure type. All OT claims with CLM_TYPE_CD=1 and FED_SRVC_CTGRY_CD = 25-28 or 31-38, except those with a FED_SRVC_CTGRY_CD value of 13 or a TOS_CD value of 132, 133, or 134. This includes FFS claims in the OT file, except those that represent UPL supplemental payments and DSH payments, which correspond to the other expenditure type. All RX claims with CLM_TYPE_CD=1 and FED_SRVC_CTGRY_CD = 41, except those with a FED_SRVC_CTGRY_CD value of 13 or a TOS_CD value of 132, 133, or 134. This includes RX FFS claims for prescription drugs, except those that represent UPL supplemental payments and DSH payments, which correspond to the other expenditure type. IP claims with CLM_TYPE_CD=1 and a FED_SRVC_CTGRY_CD = 21-28, or 31-38 (or TOS_CD = 001 if missing), except those with a TOS_CD value of 132, 133, or 134, or those with a TOS_CD value for DSH payment c (123) as the only code on the claim. This includes IP FFS claims in the IP file, except those that represent UPL supplemental payments and DSH payments, which correspond to the other expenditure type. Records with CLM_TYPE_CD=5 that linked to a beneficiary, do not have a FED_SRVC_CTGRY_CD of 11 or 12 (or TOS_CD value of 119, 120, 121, 122, 138, 143, or 144 if missing), and do not have a TOS_CD value of 123, 132, 133, or 134. d This represents wraparound payments associated with a specific beneficiary above the negotiated per-service rate. |
| Categories of service 18A, 18A1, 18A2, 18A3, 18A4, 18A5, 18A6, and 22. a These represent payments to Medicaid managed care organizations and Programs of All-Inclusive Care for the Elderly (PACE) plans. Categories of service 18B1, 18B1a, 18B1b, 18B1c, 18B1d, 18B1e, 18B1f, 18B2, 18B2a, 18B2b, 18B2c, 18B2d, 18B2e, and 18B2f. a These represent payments to prepaid ambulatory health plans (PAHP) and prepaid inpatient health plans (PIHP). Categories of service 18C and 18E. These represent premium assistance for private plans. Category of service 25. This represents primary care case management (PCCM) payments. | All OT records with CLM_TYPE_CD=2, except those with a FED_SRVC_CTGRY_CD value of 13 or TOS_CD values of 132, 133, or 134. This includes all capitation payment records except those that represent DSH payments and UPL supplemental payments, which correspond to the other expenditure type. Any records with CLM_TYPE_CD=4 that also have FED_SRVC_CTGRY_CD values of 11 or 12 (or TOS_CD values of 119, 120, 121, 122, 138, 143, or 144 if missing). d These represent financial transaction records with a federally-assigned service category (or type of service if missing) indicating the lump-sum payment (not linkable to a specific beneficiary) was for a managed care capitation payment or other monthly beneficiary payment. Any records with CLM_TYPE_CD=5 that link to a Medicaid beneficiary and also have FED_SRVC_CTGRY_CD values of 11 or 12 (or TOS_CD values of 119, 120, 121, 122, 138, 143, or 144 if missing). d These represent payments that serve as adjustments to the negotiated monthly per-member rate. | |
| Other | Categories of service 1B and 2B. These represent inpatient hospital and mental health facility DSH payments. Categories of service 1C, 3B, 4C, 5B, 6B, 9B, 10B, 29B, 37B and 37C. a These represent supplemental payments for inpatient hospital, nursing facility services, intermediate care facility for individuals with intellectual disabilities, physician and surgical services, outpatient hospital services, other practitioner services, clinic services, non-emergency medical transportation, critical access hospital inpatient services, and critical access hospital outpatient services. Category of service 1D. This represents inpatient hospital graduate medical education (GME) payments. | All records with CLM_TYPE_CD=4 except those with a FED_SRVC_CTGRY_CD value of 11 or 12 (or TOS_CD value of 119, 120, 121, 122, 138, 143, or 144 if missing). d These represent all the lump-sum payments reported by the state that do not have a federally-assigned service category (or type of service if missing) indicating payment to a managed care plan. All records with CLM_TYPE_CD=5 and a missing or invalid MSIS ID, e except those with a FED_SRVC_CTGRY_CD value of 11 or 12 (or TOS_CD value of 119, 120, 121, 122, 138, 143, or 144 if missing). d These represent lump-sum payments (other than to managed care plans) that were classified as supplemental payments by the state in its TMSIS submission. Any records with CLM_TYPE_CD=1 that also had a FED_SRVC_CTGRY_CD value of 13, c and any records with CLM_TYPE_CD values of 1 or 2 that also had TOS_CD values of 132, 133, or 134. These represent DSH and UPL supplemental payments reported as FFS claims or capitation payment records. |
a CMS-64 category of service 45 (health home for enrollees with substance abuse disorder) is a valid value starting in 2019. CMS-64 categories of service 46 and 46B (medication-assisted treatment for opioid use disorder) are valid starting in 2020. CMS-64 categories of service 10A and 10B (clinic services), 18A6, 18B1f, and 18B2f (MCO services subject to electronic visit verification requirements), and 47 (COVID-19 vaccines and vaccine administration) are valid starting in 2021. CMS-64 categories of service 29A, 29B (regular and supplemental payments for non-emergency medical transportation), 37A, 37B and 37C (regular and supplemental payments for critical access hospital inpatient and outpatient services) are valid starting in 2022. CMS-64 category of service 48 (qualified community based mobile crisis intervention) is valid starting in 2022. CMS-64 category of service 49 (health homes for children with medically complex conditions) is valid starting in 2023.
b We first exclude TAF records that represent Medicare premiums, EHR payments to providers, and drug rebates.
c In some states, DSH payment amounts are included on FFS claim headers. For claims with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the total FFS expenditure and attributed the DSH payment amount to the other expenditure type.
d TOS codes 138, 143, and 144 are valid values starting in 2020.
e MSIS IDs that start with an “&” or are all 8-filled, 9-filled, or 0-filled are considered invalid.
f The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
Data quality assessment criteria
We categorized each state’s total Medicaid expenditures and Medicaid beneficiary expenditures in TAF as having high, moderate, low, or very low alignment with the CMS-64 based on the percent difference between the two data sources (Table 3). [18]
Table 3. Criteria for DQ assessment of TAF expenditures
| Percent difference between TAF and CMS-64 expenditures | Level of alignment | DQ assessment |
|---|---|---|
| x < 5 percent | High | Low concern |
| 5 percent ≤ x < 20 percent | Moderate | Medium concern |
| 20 percent ≤ x < 50 percent | Low | High concern |
| x ≥ 50 percent | Very low | Unusable |
States with a low absolute percentage difference between the expenditures captured in TAF and the CMS-64 are considered to present a low level of concern about the completeness and quality of their TAF expenditure data. However, TAF users should be aware that in states with large Medicaid programs, even a small percentage difference may represent a substantial difference in the dollar amount between the data sources. In addition, compared to its alignment for total expenditures, a state may have better alignment for certain types of spending (such as FFS spending or monthly beneficiary payments). When this occurs, users may be able to reliably examine some types of spending in the state even when overall expenditures are incomplete. [19]
There are several reasons other than the quality and completeness of TAF expenditure data that could result in a state’s total expenditures or Medicaid beneficiary expenditures in the TAF differing from the CMS-64: (1) structural differences in the data sets, such as organizing the data by service date versus claim payment date; (2) data quality problems in the TAF that affect expenditures, such as a low volume of claims in the TAF because of missing date values on claims or issues with assigning a final action status to claims; and (3) variation in how states report certain types of expenditures on the CMS-64 that are difficult to reproduce in the TAF data. [20] For instance, a state may make lump-sum reconciliation payments to hospitals as part of their cost-based FFS payments for inpatient services. On the CMS-64, the state may report these costs as part of the inpatient hospital base payment category of service, whereas in TAF these lump sum payments are reported as service tracking claims that cannot easily be differentiated from other lump sum payments such as DSH and UPL payments. As a result of these kinds of differences, some expenditures may fall into CMS-64 categories of service that are classified into the FFS expenditure type but appear in TAF as service records that are classified as the “other” expenditure type. When this occurs, we would expect to see a difference between the TAF-based Medicaid beneficiary expenditures and the CMS-64 benchmark even if TAF expenditure data are complete and accurate.
Methods previously used to assess data quality
Table 4 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 4. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1. In addition, the TAF RIF for all years and releases redacts supplemental payments in which the Medicaid identification number begins with a “&” because they cannot be attributed to a specific beneficiary. These are likely to represent lump sum payments that were misclassified as supplemental payments rather than service tracking claims. All other supplemental payment records are included in the TAF RIF. This analysis includes expenditures on supplemental payment records that are redacted from the RIF.
FFS claims representing Medicaid or M-CHIP benefits have a claim type code value of 1. Medicaid or M-CHIP monthly beneficiary payments have a claim type code value of 2, service tracking claims have a claim type code value of 4, and supplemental payment records have a claim type code value of 5.
Payment data on managed care encounter records are redacted from the TAF RIF.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the beneficiary had CHIP code value of 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-76, we considered the beneficiary to have non-CHIP Medicaid enrollment during the month. For FFS expenditures, we required claims to link to a beneficiary with non-CHIP Medicaid enrollment during the month of service. For monthly beneficiary payments, we required claims to link to a beneficiary with non-CHIP Medicaid enrollment in any month of the year. For other expenditures, records do not link to specific beneficiaries.
We did not require service tracking claims to match to a Medicaid eligibility record because these types of records are lump-sum payments that are not connected to any one beneficiary and should not have a valid beneficiary identifier.
The FFS expenditure amount is comprised of services provided in the month of eligibility; services for beneficiaries who switch from CHIP to non-CHIP Medicaid will only be counted in the months of non-CHIP Medicaid eligibility. The monthly beneficiary payment amount is comprised of payments in the year of service; all payments for beneficiaries who switch from CHIP to non-CHIP Medicaid within a year will be counted in the total.
EHR payments were identified as records where any line had a type of service code (TOS_CD) equal to 135. Drug rebates were identified as records were any line had a type of service code of 131 or a service tracking type code (SRVC_TRKNG_TYPE_CD) of 01. Medicare premiums were identified as records where any line had a type of service code of 139, 140, 141, or 142 (valid values starting in 2020).
We considered records with claim type 4 and a federally assigned service category code (11 or 12) or TOS code (119, 120,121, 122, 138, 143, or 144) that indicates a capitation payment to be capitation payments reported on service tracking claims.
The DSH payment field only appears on IP claims. For LT, OT, and RX claims, we only considered the service tracking payment amount or total Medicaid paid amount.
For service tracking claims, states sometimes entered the same payment amount in all three payment fields (or two of the three payment fields). To avoid double or triple counting, we used only one payment field per header claim.
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
CMS-64 categories of service 46A1-46A6 are valid starting in federal fiscal year 2021. CMS-64 category of service 7A7 is valid starting in federal fiscal year 2022.
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: (TAF - benchmark) / benchmark.
Users can find more information about each state’s alignment of FFS and monthly beneficiary payments in the DQ Atlas single topic displays for Total FFS Expenditures and Total Monthly Beneficiary Payments .
For more information about the differences between TAF and CMS-64 expenditure data, see the methodology brief, “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1. In addition, the TAF RIF for all years and releases redacts supplemental payments in which the Medicaid identification number begins with a \u201c&\u201d because they cannot be attributed to a specific beneficiary. These are likely to represent lump sum payments that were misclassified as supplemental payments rather than service tracking claims. All other supplemental payment records are included in the TAF RIF. This analysis includes expenditures on supplemental payment records that are redacted from the RIF.
FFS claims representing Medicaid or M-CHIP benefits have a claim type code value of 1. Medicaid or M-CHIP monthly beneficiary payments have a claim type code value of 2, service tracking claims have a claim type code value of 4, and supplemental payment records have a claim type code value of 5.
Payment data on managed care encounter records are redacted from the TAF RIF.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the beneficiary had CHIP code value of 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-76, we considered the beneficiary to have non-CHIP Medicaid enrollment during the month. For FFS expenditures, we required claims to link to a beneficiary with non-CHIP Medicaid enrollment during the month of service. For monthly beneficiary payments, we required claims to link to a beneficiary with non-CHIP Medicaid enrollment in any month of the year. For other expenditures, records do not link to specific beneficiaries.
We did not require service tracking claims to match to a Medicaid eligibility record because these types of records are lump-sum payments that are not connected to any one beneficiary and should not have a valid beneficiary identifier.
The FFS expenditure amount is comprised of services provided in the month of eligibility; services for beneficiaries who switch from CHIP to non-CHIP Medicaid will only be counted in the months of non-CHIP Medicaid eligibility. The monthly beneficiary payment amount is comprised of payments in the year of service; all payments for beneficiaries who switch from CHIP to non-CHIP Medicaid within a year will be counted in the total.
EHR payments were identified as records where any line had a type of service code (TOS_CD) equal to 135. Drug rebates were identified as records were any line had a type of service code of 131 or a service tracking type code (SRVC_TRKNG_TYPE_CD) of 01. Medicare premiums were identified as records where any line had a type of service code of 139, 140, 141, or 142 (valid values starting in 2020).
We considered records with claim type 4 and a federally assigned service category code (11 or 12) or TOS code (119, 120,121, 122, 138, 143, or 144) that indicates a capitation payment to be capitation payments reported on service tracking claims.
The DSH payment field only appears on IP claims. For LT, OT, and RX claims, we only considered the service tracking payment amount or total Medicaid paid amount.
For service tracking claims, states sometimes entered the same payment amount in all three payment fields (or two of the three payment fields). To avoid double or triple counting, we used only one payment field per header claim.
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
CMS-64 categories of service 46A1-46A6 are valid starting in federal fiscal year 2021. CMS-64 category of service 7A7 is valid starting in federal fiscal year 2022.
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: (TAF - benchmark) / benchmark.
Users can find more information about each state\u2019s alignment of FFS and monthly beneficiary payments in the DQ Atlas single topic displays for Total FFS Expenditures and Total Monthly Beneficiary Payments .
For more information about the differences between TAF and CMS-64 expenditure data, see the methodology brief, \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
The TAF should include all state spending to provide or cover care for Medicaid beneficiaries, including fee-for-service claims, monthly beneficiary payments, supplemental payments, and other lump sum payments to providers. This analysis examines how well the total Medicaid expenditures captured in TAF align with an external benchmark, the CMS-64.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""6041"", ""relatedTopics"": [{""measureId"": 87, ""measureName"": ""Medicaid Beneficiary Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 1}]}" 78,"{""measureId"": 78, ""measureName"": ""Payment Data Consistency - RX"", ""groupId"": 8, ""groupName"": ""Payments"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Pmt-Consistency-RX.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) for inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims are structured to capture one header record and one or more line records per claim. [1] Header-level records capture data that apply to the entire claim. Line-level records capture data about the specific goods or services provided to a Medicaid or Children’s Health Insurance Program (CHIP) beneficiary as part of the overall service and are summarized in the header record. In all four claims files, payments made by the Medicaid agency are captured at both the header and line level. On fee-for-service (FFS) claims, the total Medicaid payment amount on the header record represents the full payment that the Medicaid agency made for the entire claim; the Medicaid payment amount on the line-level records represents the portion of the payment associated with each particular service recorded within the claim. [2] Although payment amounts are reported on header and line records, states specify whether claims were processed and paid at either the header or the line level by using the payment-level indicator.
States vary in their approaches to Medicaid payment policy, [3] including whether claims for different types of services are paid at the header or line level, but the sum of the Medicaid payment amounts on the line-level records should always equal the claim’s total Medicaid payment amount on the header record. [4] Consistent payment amounts on claim headers and lines allow TAF users the maximum flexibility in tabulating and analyzing payment amounts within the claims files. [5] In this data quality assessment, we compare the consistency of Medicaid payments reported at the header and line levels within each claims file.
A header record summarizes the services provided that are captured on the claim lines, which provide details on each service covered by the claim. The TAF production algorithm includes final action claim headers and all their associated line records. The TAF algorithm only includes T-MSIS claim lines that can be linked to a T-MSIS claim header. In some cases, T-MSIS claim lines cannot be linked to a T-MSIS claim header. These claim lines are excluded from the TAF.
The TAF also include managed care encounter records which represent claims submitted by providers to non-state entities (for example, Medicaid managed care plans or MCOs) for care provided to Medicaid and CHIP beneficiaries who are enrolled in managed care arrangements. The Medicaid payment amount on encounter claims represents payments made by managed care entities to facilities and providers and does not represent a Medicaid or CHIP payment by the state (as it does on FFS claims). Managed care payment information is redacted from records in the TAF RIF. State payments for managed care services are reported in capitation claims, which represent the per member, per month premium payment from state Medicaid agencies to managed care entities.
Medicaid and CHIP Payment and Access Commission (MACPAC). “Chapter 5: Examining Medicaid Payment Policy.” Report to the Congress on Medicaid and CHIP. Washington, DC: MACPAC, March 2011, pp. 153–182. Available at https://www.macpac.gov/wp-content/uploads/2011/03/Examining-Medicaid-Payment-Policy.pdf . Accessed April 25, 2019.
CMS guidance issued to states regarding completeness and consistency of claim payment data elements is available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/52080 .
Certain data elements (for example, billing provider type) are available only on header records, whereas other data elements (such as servicing provider type) are available only on line records. If header- and line-level payment data are not consistent, TAF users who wish to examine spending by using data elements available only at the header level may need to aggregate line-level payments up to the header level before conducting their analysis. Alternatively, TAF users who wish to examine spending by using data elements available only at the line level may need to develop a method for disaggregating header-level payments across the lines within a claim. When payment amounts are consistent across header and line records, TAF users can rely on whichever payment variable is more convenient.
A header record summarizes the services provided that are captured on the claim lines, which provide details on each service covered by the claim. The TAF production algorithm includes final action claim headers and all their associated line records. The TAF algorithm only includes T-MSIS claim lines that can be linked to a T-MSIS claim header. In some cases, T-MSIS claim lines cannot be linked to a T-MSIS claim header. These claim lines are excluded from the TAF.
The TAF also include managed care encounter records which represent claims submitted by providers to non-state entities (for example, Medicaid managed care plans or MCOs) for care provided to Medicaid and CHIP beneficiaries who are enrolled in managed care arrangements. The Medicaid payment amount on encounter claims represents payments made by managed care entities to facilities and providers and does not represent a Medicaid or CHIP payment by the state (as it does on FFS claims). Managed care payment information is redacted from records in the TAF RIF. State payments for managed care services are reported in capitation claims, which represent the per member, per month premium payment from state Medicaid agencies to managed care entities.
Medicaid and CHIP Payment and Access Commission (MACPAC). \u201cChapter 5: Examining Medicaid Payment Policy.\u201d Report to the Congress on Medicaid and CHIP. Washington, DC: MACPAC, March 2011, pp. 153\u2013182. Available at https://www.macpac.gov/wp-content/uploads/2011/03/Examining-Medicaid-Payment-Policy.pdf . Accessed April 25, 2019.
CMS guidance issued to states regarding completeness and consistency of claim payment data elements is available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/52080 .
Certain data elements (for example, billing provider type) are available only on header records, whereas other data elements (such as servicing provider type) are available only on line records. If header- and line-level payment data are not consistent, TAF users who wish to examine spending by using data elements available only at the header level may need to aggregate line-level payments up to the header level before conducting their analysis. Alternatively, TAF users who wish to examine spending by using data elements available only at the line level may need to develop a method for disaggregating header-level payments across the lines within a claim. When payment amounts are consistent across header and line records, TAF users can rely on whichever payment variable is more convenient.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We selected FFS claims in the four TAF [6] claims files (IP, LT, OT, and RX) by using claim type code (CLM_TYPE_CD). [7] We further restricted the analysis to claim headers that had a positive total Medicaid payment amount [8] and at least one non-denied claim line. We tabulated the number of claim headers that met these criteria in each state and the proportion of claim headers (with at least one non-denied claim line) in which the header payment (TOT_MDCD_PD_AMT) equaled the sum of the line-level payments (MDCD_PD_AMT) among non-denied lines. [9] We considered claim headers with a total payment amount that equaled the sum of payments on the associated line records to be internally consistent, and states with higher proportions of internally consistent claims to present lower levels of concern about data quality (Table 1).
We also examined the indicator that states use to report whether payment is made at the header or the line level (PYMT_LVL_IND), to determine whether header- or line-level records are more likely to be accurate when a claim’s payment amounts are inconsistent across header and line records. We considered a state’s data to be unusable if more than 5 percent of claims had inconsistent header- and line-level payments and the state had a nontrivial proportion of total claims (20 percent or higher) with unknown payment level.
Table 1. Criteria for DQ assessment of payment consistency
| Percentage of claims with consistent payments across header and line records | Percentage of all claims with an unknown payment level | Level of consistency | DQ assessment |
|---|---|---|---|
| x ≥ 95 percent | Not assessed | Highly consistent | Low concern |
| 5 percent ≤ x < 95 percent | x < 20 percent | Mixed consistency | Medium concern a |
| x < 5 percent | x < 20 percent | Highly inconsistent | High concern a |
| x < 95 percent | x ≥ 20 percent | Highly inconsistent | Unusable a |
a Both criteria must be true for a state to receive the given DQ Assessment
Methods previously used to assess data quality
Table 2 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 2. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
FFS records (claim type 1 or A) were retained in the analysis. We excluded records with all other claim type values, including capitation payments, managed care encounters, service tracking claims, and supplemental payments.
Nearly all FFS claims should have a positive total Medicaid payment amount because the TAF only includes non-void, non-denied final action claims that incorporate adjustments made to payments. Although the TAF does include claims with unexpected payment amounts, the vast majority of FFS claims have positive payment amounts. More information can be found in the DQ Atlas single topic display for Missing Payment Data - FFS Claims .
In some states, Disproportionate Share Hospital (DSH) payment amounts are included on FFS claim headers. For claims in the IP file with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the header-level payment amount (TOT_MDCD_PD_AMT - MDCD_DSH_PD_AMT).
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
FFS records (claim type 1 or A) were retained in the analysis. We excluded records with all other claim type values, including capitation payments, managed care encounters, service tracking claims, and supplemental payments.
Nearly all FFS claims should have a positive total Medicaid payment amount because the TAF only includes non-void, non-denied final action claims that incorporate adjustments made to payments. Although the TAF does include claims with unexpected payment amounts, the vast majority of FFS claims have positive payment amounts. More information can be found in the DQ Atlas single topic display for Missing Payment Data - FFS Claims .
In some states, Disproportionate Share Hospital (DSH) payment amounts are included on FFS claim headers. For claims in the IP file with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the header-level payment amount (TOT_MDCD_PD_AMT - MDCD_DSH_PD_AMT).
States report Medicaid payments for a covered service both on claim headers and on claim lines. This analysis identifies how often the payment amounts captured in these two places are inconsistent on RX claims.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""6061"", ""relatedTopics"": [{""measureId"": 81, ""measureName"": ""Payment Data Consistency - IP"", ""groupId"": 8, ""groupName"": ""Payments"", ""order"": 0}, {""measureId"": 80, ""measureName"": ""Payment Data Consistency - LT"", ""groupId"": 8, ""groupName"": ""Payments"", ""order"": 1}, {""measureId"": 79, ""measureName"": ""Payment Data Consistency - OT"", ""groupId"": 8, ""groupName"": ""Payments"", ""order"": 2}]}" 79,"{""measureId"": 79, ""measureName"": ""Payment Data Consistency - OT"", ""groupId"": 8, ""groupName"": ""Payments"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Pmt-Consistency-OT.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) for inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims are structured to capture one header record and one or more line records per claim. [1] Header-level records capture data that apply to the entire claim. Line-level records capture data about the specific goods or services provided to a Medicaid or Children’s Health Insurance Program (CHIP) beneficiary as part of the overall service and are summarized in the header record. In all four claims files, payments made by the Medicaid agency are captured at both the header and line level. On fee-for-service (FFS) claims, the total Medicaid payment amount on the header record represents the full payment that the Medicaid agency made for the entire claim; the Medicaid payment amount on the line-level records represents the portion of the payment associated with each particular service recorded within the claim. [2] Although payment amounts are reported on header and line records, states specify whether claims were processed and paid at either the header or the line level by using the payment-level indicator.
States vary in their approaches to Medicaid payment policy, [3] including whether claims for different types of services are paid at the header or line level, but the sum of the Medicaid payment amounts on the line-level records should always equal the claim’s total Medicaid payment amount on the header record. [4] Consistent payment amounts on claim headers and lines allow TAF users the maximum flexibility in tabulating and analyzing payment amounts within the claims files. [5] In this data quality assessment, we compare the consistency of Medicaid payments reported at the header and line levels within each claims file.
A header record summarizes the services provided that are captured on the claim lines, which provide details on each service covered by the claim. The TAF production algorithm includes final action claim headers and all their associated line records. The TAF algorithm only includes T-MSIS claim lines that can be linked to a T-MSIS claim header. In some cases, T-MSIS claim lines cannot be linked to a T-MSIS claim header. These claim lines are excluded from the TAF.
The TAF also include managed care encounter records which represent claims submitted by providers to non-state entities (for example, Medicaid managed care plans or MCOs) for care provided to Medicaid and CHIP beneficiaries who are enrolled in managed care arrangements. The Medicaid payment amount on encounter claims represents payments made by managed care entities to facilities and providers and does not represent a Medicaid or CHIP payment by the state (as it does on FFS claims). Managed care payment information is redacted from records in the TAF RIF. State payments for managed care services are reported in capitation claims, which represent the per member, per month premium payment from state Medicaid agencies to managed care entities.
Medicaid and CHIP Payment and Access Commission (MACPAC). “Chapter 5: Examining Medicaid Payment Policy.” Report to the Congress on Medicaid and CHIP. Washington, DC: MACPAC, March 2011, pp. 153–182. Available at https://www.macpac.gov/wp-content/uploads/2011/03/Examining-Medicaid-Payment-Policy.pdf . Accessed April 25, 2019.
CMS guidance issued to states regarding completeness and consistency of claim payment data elements is available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/52080 .
Certain data elements (for example, billing provider type) are available only on header records, whereas other data elements (such as servicing provider type) are available only on line records. If header- and line-level payment data are not consistent, TAF users who wish to examine spending by using data elements available only at the header level may need to aggregate line-level payments up to the header level before conducting their analysis. Alternatively, TAF users who wish to examine spending by using data elements available only at the line level may need to develop a method for disaggregating header-level payments across the lines within a claim. When payment amounts are consistent across header and line records, TAF users can rely on whichever payment variable is more convenient.
A header record summarizes the services provided that are captured on the claim lines, which provide details on each service covered by the claim. The TAF production algorithm includes final action claim headers and all their associated line records. The TAF algorithm only includes T-MSIS claim lines that can be linked to a T-MSIS claim header. In some cases, T-MSIS claim lines cannot be linked to a T-MSIS claim header. These claim lines are excluded from the TAF.
The TAF also include managed care encounter records which represent claims submitted by providers to non-state entities (for example, Medicaid managed care plans or MCOs) for care provided to Medicaid and CHIP beneficiaries who are enrolled in managed care arrangements. The Medicaid payment amount on encounter claims represents payments made by managed care entities to facilities and providers and does not represent a Medicaid or CHIP payment by the state (as it does on FFS claims). Managed care payment information is redacted from records in the TAF RIF. State payments for managed care services are reported in capitation claims, which represent the per member, per month premium payment from state Medicaid agencies to managed care entities.
Medicaid and CHIP Payment and Access Commission (MACPAC). \u201cChapter 5: Examining Medicaid Payment Policy.\u201d Report to the Congress on Medicaid and CHIP. Washington, DC: MACPAC, March 2011, pp. 153\u2013182. Available at https://www.macpac.gov/wp-content/uploads/2011/03/Examining-Medicaid-Payment-Policy.pdf . Accessed April 25, 2019.
CMS guidance issued to states regarding completeness and consistency of claim payment data elements is available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/52080 .
Certain data elements (for example, billing provider type) are available only on header records, whereas other data elements (such as servicing provider type) are available only on line records. If header- and line-level payment data are not consistent, TAF users who wish to examine spending by using data elements available only at the header level may need to aggregate line-level payments up to the header level before conducting their analysis. Alternatively, TAF users who wish to examine spending by using data elements available only at the line level may need to develop a method for disaggregating header-level payments across the lines within a claim. When payment amounts are consistent across header and line records, TAF users can rely on whichever payment variable is more convenient.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We selected FFS claims in the four TAF [6] claims files (IP, LT, OT, and RX) by using claim type code (CLM_TYPE_CD). [7] We further restricted the analysis to claim headers that had a positive total Medicaid payment amount [8] and at least one non-denied claim line. We tabulated the number of claim headers that met these criteria in each state and the proportion of claim headers (with at least one non-denied claim line) in which the header payment (TOT_MDCD_PD_AMT) equaled the sum of the line-level payments (MDCD_PD_AMT) among non-denied lines. [9] We considered claim headers with a total payment amount that equaled the sum of payments on the associated line records to be internally consistent, and states with higher proportions of internally consistent claims to present lower levels of concern about data quality (Table 1).
We also examined the indicator that states use to report whether payment is made at the header or the line level (PYMT_LVL_IND), to determine whether header- or line-level records are more likely to be accurate when a claim’s payment amounts are inconsistent across header and line records. We considered a state’s data to be unusable if more than 5 percent of claims had inconsistent header- and line-level payments and the state had a nontrivial proportion of total claims (20 percent or higher) with unknown payment level.
Table 1. Criteria for DQ assessment of payment consistency
| Percentage of claims with consistent payments across header and line records | Percentage of all claims with an unknown payment level | Level of consistency | DQ assessment |
|---|---|---|---|
| x ≥ 95 percent | Not assessed | Highly consistent | Low concern |
| 5 percent ≤ x < 95 percent | x < 20 percent | Mixed consistency | Medium concern a |
| x < 5 percent | x < 20 percent | Highly inconsistent | High concern a |
| x < 95 percent | x ≥ 20 percent | Highly inconsistent | Unusable a |
a Both criteria must be true for a state to receive the given DQ Assessment
Methods previously used to assess data quality
Table 2 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 2. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
FFS records (claim type 1 or A) were retained in the analysis. We excluded records with all other claim type values, including capitation payments, managed care encounters, service tracking claims, and supplemental payments.
Nearly all FFS claims should have a positive total Medicaid payment amount because the TAF only includes non-void, non-denied final action claims that incorporate adjustments made to payments. Although the TAF does include claims with unexpected payment amounts, the vast majority of FFS claims have positive payment amounts. More information can be found in the DQ Atlas single topic display for Missing Payment Data - FFS Claims .
In some states, Disproportionate Share Hospital (DSH) payment amounts are included on FFS claim headers. For claims in the IP file with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the header-level payment amount (TOT_MDCD_PD_AMT - MDCD_DSH_PD_AMT).
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
FFS records (claim type 1 or A) were retained in the analysis. We excluded records with all other claim type values, including capitation payments, managed care encounters, service tracking claims, and supplemental payments.
Nearly all FFS claims should have a positive total Medicaid payment amount because the TAF only includes non-void, non-denied final action claims that incorporate adjustments made to payments. Although the TAF does include claims with unexpected payment amounts, the vast majority of FFS claims have positive payment amounts. More information can be found in the DQ Atlas single topic display for Missing Payment Data - FFS Claims .
In some states, Disproportionate Share Hospital (DSH) payment amounts are included on FFS claim headers. For claims in the IP file with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the header-level payment amount (TOT_MDCD_PD_AMT - MDCD_DSH_PD_AMT).
States report Medicaid payments for a covered service both on claim headers and on claim lines. This analysis identifies how often the payment amounts captured in these two places are inconsistent on OT claims.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""6061"", ""relatedTopics"": [{""measureId"": 81, ""measureName"": ""Payment Data Consistency - IP"", ""groupId"": 8, ""groupName"": ""Payments"", ""order"": 0}, {""measureId"": 80, ""measureName"": ""Payment Data Consistency - LT"", ""groupId"": 8, ""groupName"": ""Payments"", ""order"": 1}, {""measureId"": 78, ""measureName"": ""Payment Data Consistency - RX"", ""groupId"": 8, ""groupName"": ""Payments"", ""order"": 3}]}" 80,"{""measureId"": 80, ""measureName"": ""Payment Data Consistency - LT"", ""groupId"": 8, ""groupName"": ""Payments"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Pmt-Consistency-LT.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) for inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims are structured to capture one header record and one or more line records per claim. [1] Header-level records capture data that apply to the entire claim. Line-level records capture data about the specific goods or services provided to a Medicaid or Children’s Health Insurance Program (CHIP) beneficiary as part of the overall service and are summarized in the header record. In all four claims files, payments made by the Medicaid agency are captured at both the header and line level. On fee-for-service (FFS) claims, the total Medicaid payment amount on the header record represents the full payment that the Medicaid agency made for the entire claim; the Medicaid payment amount on the line-level records represents the portion of the payment associated with each particular service recorded within the claim. [2] Although payment amounts are reported on header and line records, states specify whether claims were processed and paid at either the header or the line level by using the payment-level indicator.
States vary in their approaches to Medicaid payment policy, [3] including whether claims for different types of services are paid at the header or line level, but the sum of the Medicaid payment amounts on the line-level records should always equal the claim’s total Medicaid payment amount on the header record. [4] Consistent payment amounts on claim headers and lines allow TAF users the maximum flexibility in tabulating and analyzing payment amounts within the claims files. [5] In this data quality assessment, we compare the consistency of Medicaid payments reported at the header and line levels within each claims file.
A header record summarizes the services provided that are captured on the claim lines, which provide details on each service covered by the claim. The TAF production algorithm includes final action claim headers and all their associated line records. The TAF algorithm only includes T-MSIS claim lines that can be linked to a T-MSIS claim header. In some cases, T-MSIS claim lines cannot be linked to a T-MSIS claim header. These claim lines are excluded from the TAF.
The TAF also include managed care encounter records which represent claims submitted by providers to non-state entities (for example, Medicaid managed care plans or MCOs) for care provided to Medicaid and CHIP beneficiaries who are enrolled in managed care arrangements. The Medicaid payment amount on encounter claims represents payments made by managed care entities to facilities and providers and does not represent a Medicaid or CHIP payment by the state (as it does on FFS claims). Managed care payment information is redacted from records in the TAF RIF. State payments for managed care services are reported in capitation claims, which represent the per member, per month premium payment from state Medicaid agencies to managed care entities.
Medicaid and CHIP Payment and Access Commission (MACPAC). “Chapter 5: Examining Medicaid Payment Policy.” Report to the Congress on Medicaid and CHIP. Washington, DC: MACPAC, March 2011, pp. 153–182. Available at https://www.macpac.gov/wp-content/uploads/2011/03/Examining-Medicaid-Payment-Policy.pdf . Accessed April 25, 2019.
CMS guidance issued to states regarding completeness and consistency of claim payment data elements is available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/52080 .
Certain data elements (for example, billing provider type) are available only on header records, whereas other data elements (such as servicing provider type) are available only on line records. If header- and line-level payment data are not consistent, TAF users who wish to examine spending by using data elements available only at the header level may need to aggregate line-level payments up to the header level before conducting their analysis. Alternatively, TAF users who wish to examine spending by using data elements available only at the line level may need to develop a method for disaggregating header-level payments across the lines within a claim. When payment amounts are consistent across header and line records, TAF users can rely on whichever payment variable is more convenient.
A header record summarizes the services provided that are captured on the claim lines, which provide details on each service covered by the claim. The TAF production algorithm includes final action claim headers and all their associated line records. The TAF algorithm only includes T-MSIS claim lines that can be linked to a T-MSIS claim header. In some cases, T-MSIS claim lines cannot be linked to a T-MSIS claim header. These claim lines are excluded from the TAF.
The TAF also include managed care encounter records which represent claims submitted by providers to non-state entities (for example, Medicaid managed care plans or MCOs) for care provided to Medicaid and CHIP beneficiaries who are enrolled in managed care arrangements. The Medicaid payment amount on encounter claims represents payments made by managed care entities to facilities and providers and does not represent a Medicaid or CHIP payment by the state (as it does on FFS claims). Managed care payment information is redacted from records in the TAF RIF. State payments for managed care services are reported in capitation claims, which represent the per member, per month premium payment from state Medicaid agencies to managed care entities.
Medicaid and CHIP Payment and Access Commission (MACPAC). \u201cChapter 5: Examining Medicaid Payment Policy.\u201d Report to the Congress on Medicaid and CHIP. Washington, DC: MACPAC, March 2011, pp. 153\u2013182. Available at https://www.macpac.gov/wp-content/uploads/2011/03/Examining-Medicaid-Payment-Policy.pdf . Accessed April 25, 2019.
CMS guidance issued to states regarding completeness and consistency of claim payment data elements is available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/52080 .
Certain data elements (for example, billing provider type) are available only on header records, whereas other data elements (such as servicing provider type) are available only on line records. If header- and line-level payment data are not consistent, TAF users who wish to examine spending by using data elements available only at the header level may need to aggregate line-level payments up to the header level before conducting their analysis. Alternatively, TAF users who wish to examine spending by using data elements available only at the line level may need to develop a method for disaggregating header-level payments across the lines within a claim. When payment amounts are consistent across header and line records, TAF users can rely on whichever payment variable is more convenient.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We selected FFS claims in the four TAF [6] claims files (IP, LT, OT, and RX) by using claim type code (CLM_TYPE_CD). [7] We further restricted the analysis to claim headers that had a positive total Medicaid payment amount [8] and at least one non-denied claim line. We tabulated the number of claim headers that met these criteria in each state and the proportion of claim headers (with at least one non-denied claim line) in which the header payment (TOT_MDCD_PD_AMT) equaled the sum of the line-level payments (MDCD_PD_AMT) among non-denied lines. [9] We considered claim headers with a total payment amount that equaled the sum of payments on the associated line records to be internally consistent, and states with higher proportions of internally consistent claims to present lower levels of concern about data quality (Table 1).
We also examined the indicator that states use to report whether payment is made at the header or the line level (PYMT_LVL_IND), to determine whether header- or line-level records are more likely to be accurate when a claim’s payment amounts are inconsistent across header and line records. We considered a state’s data to be unusable if more than 5 percent of claims had inconsistent header- and line-level payments and the state had a nontrivial proportion of total claims (20 percent or higher) with unknown payment level.
Table 1. Criteria for DQ assessment of payment consistency
| Percentage of claims with consistent payments across header and line records | Percentage of all claims with an unknown payment level | Level of consistency | DQ assessment |
|---|---|---|---|
| x ≥ 95 percent | Not assessed | Highly consistent | Low concern |
| 5 percent ≤ x < 95 percent | x < 20 percent | Mixed consistency | Medium concern a |
| x < 5 percent | x < 20 percent | Highly inconsistent | High concern a |
| x < 95 percent | x ≥ 20 percent | Highly inconsistent | Unusable a |
a Both criteria must be true for a state to receive the given DQ Assessment
Methods previously used to assess data quality
Table 2 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 2. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
FFS records (claim type 1 or A) were retained in the analysis. We excluded records with all other claim type values, including capitation payments, managed care encounters, service tracking claims, and supplemental payments.
Nearly all FFS claims should have a positive total Medicaid payment amount because the TAF only includes non-void, non-denied final action claims that incorporate adjustments made to payments. Although the TAF does include claims with unexpected payment amounts, the vast majority of FFS claims have positive payment amounts. More information can be found in the DQ Atlas single topic display for Missing Payment Data - FFS Claims .
In some states, Disproportionate Share Hospital (DSH) payment amounts are included on FFS claim headers. For claims in the IP file with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the header-level payment amount (TOT_MDCD_PD_AMT - MDCD_DSH_PD_AMT).
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
FFS records (claim type 1 or A) were retained in the analysis. We excluded records with all other claim type values, including capitation payments, managed care encounters, service tracking claims, and supplemental payments.
Nearly all FFS claims should have a positive total Medicaid payment amount because the TAF only includes non-void, non-denied final action claims that incorporate adjustments made to payments. Although the TAF does include claims with unexpected payment amounts, the vast majority of FFS claims have positive payment amounts. More information can be found in the DQ Atlas single topic display for Missing Payment Data - FFS Claims .
In some states, Disproportionate Share Hospital (DSH) payment amounts are included on FFS claim headers. For claims in the IP file with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the header-level payment amount (TOT_MDCD_PD_AMT - MDCD_DSH_PD_AMT).
States report Medicaid payments for a covered service both on claim headers and on claim lines. This analysis identifies how often the payment amounts captured in these two places are inconsistent on LT claims.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""6061"", ""relatedTopics"": [{""measureId"": 81, ""measureName"": ""Payment Data Consistency - IP"", ""groupId"": 8, ""groupName"": ""Payments"", ""order"": 0}, {""measureId"": 79, ""measureName"": ""Payment Data Consistency - OT"", ""groupId"": 8, ""groupName"": ""Payments"", ""order"": 2}, {""measureId"": 78, ""measureName"": ""Payment Data Consistency - RX"", ""groupId"": 8, ""groupName"": ""Payments"", ""order"": 3}]}" 81,"{""measureId"": 81, ""measureName"": ""Payment Data Consistency - IP"", ""groupId"": 8, ""groupName"": ""Payments"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Pmt-Consistency-IP.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) for inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims are structured to capture one header record and one or more line records per claim. [1] Header-level records capture data that apply to the entire claim. Line-level records capture data about the specific goods or services provided to a Medicaid or Children’s Health Insurance Program (CHIP) beneficiary as part of the overall service and are summarized in the header record. In all four claims files, payments made by the Medicaid agency are captured at both the header and line level. On fee-for-service (FFS) claims, the total Medicaid payment amount on the header record represents the full payment that the Medicaid agency made for the entire claim; the Medicaid payment amount on the line-level records represents the portion of the payment associated with each particular service recorded within the claim. [2] Although payment amounts are reported on header and line records, states specify whether claims were processed and paid at either the header or the line level by using the payment-level indicator.
States vary in their approaches to Medicaid payment policy, [3] including whether claims for different types of services are paid at the header or line level, but the sum of the Medicaid payment amounts on the line-level records should always equal the claim’s total Medicaid payment amount on the header record. [4] Consistent payment amounts on claim headers and lines allow TAF users the maximum flexibility in tabulating and analyzing payment amounts within the claims files. [5] In this data quality assessment, we compare the consistency of Medicaid payments reported at the header and line levels within each claims file.
A header record summarizes the services provided that are captured on the claim lines, which provide details on each service covered by the claim. The TAF production algorithm includes final action claim headers and all their associated line records. The TAF algorithm only includes T-MSIS claim lines that can be linked to a T-MSIS claim header. In some cases, T-MSIS claim lines cannot be linked to a T-MSIS claim header. These claim lines are excluded from the TAF.
The TAF also include managed care encounter records which represent claims submitted by providers to non-state entities (for example, Medicaid managed care plans or MCOs) for care provided to Medicaid and CHIP beneficiaries who are enrolled in managed care arrangements. The Medicaid payment amount on encounter claims represents payments made by managed care entities to facilities and providers and does not represent a Medicaid or CHIP payment by the state (as it does on FFS claims). Managed care payment information is redacted from records in the TAF RIF. State payments for managed care services are reported in capitation claims, which represent the per member, per month premium payment from state Medicaid agencies to managed care entities.
Medicaid and CHIP Payment and Access Commission (MACPAC). “Chapter 5: Examining Medicaid Payment Policy.” Report to the Congress on Medicaid and CHIP. Washington, DC: MACPAC, March 2011, pp. 153–182. Available at https://www.macpac.gov/wp-content/uploads/2011/03/Examining-Medicaid-Payment-Policy.pdf . Accessed April 25, 2019.
CMS guidance issued to states regarding completeness and consistency of claim payment data elements is available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/52080 .
Certain data elements (for example, billing provider type) are available only on header records, whereas other data elements (such as servicing provider type) are available only on line records. If header- and line-level payment data are not consistent, TAF users who wish to examine spending by using data elements available only at the header level may need to aggregate line-level payments up to the header level before conducting their analysis. Alternatively, TAF users who wish to examine spending by using data elements available only at the line level may need to develop a method for disaggregating header-level payments across the lines within a claim. When payment amounts are consistent across header and line records, TAF users can rely on whichever payment variable is more convenient.
A header record summarizes the services provided that are captured on the claim lines, which provide details on each service covered by the claim. The TAF production algorithm includes final action claim headers and all their associated line records. The TAF algorithm only includes T-MSIS claim lines that can be linked to a T-MSIS claim header. In some cases, T-MSIS claim lines cannot be linked to a T-MSIS claim header. These claim lines are excluded from the TAF.
The TAF also include managed care encounter records which represent claims submitted by providers to non-state entities (for example, Medicaid managed care plans or MCOs) for care provided to Medicaid and CHIP beneficiaries who are enrolled in managed care arrangements. The Medicaid payment amount on encounter claims represents payments made by managed care entities to facilities and providers and does not represent a Medicaid or CHIP payment by the state (as it does on FFS claims). Managed care payment information is redacted from records in the TAF RIF. State payments for managed care services are reported in capitation claims, which represent the per member, per month premium payment from state Medicaid agencies to managed care entities.
Medicaid and CHIP Payment and Access Commission (MACPAC). \u201cChapter 5: Examining Medicaid Payment Policy.\u201d Report to the Congress on Medicaid and CHIP. Washington, DC: MACPAC, March 2011, pp. 153\u2013182. Available at https://www.macpac.gov/wp-content/uploads/2011/03/Examining-Medicaid-Payment-Policy.pdf . Accessed April 25, 2019.
CMS guidance issued to states regarding completeness and consistency of claim payment data elements is available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/52080 .
Certain data elements (for example, billing provider type) are available only on header records, whereas other data elements (such as servicing provider type) are available only on line records. If header- and line-level payment data are not consistent, TAF users who wish to examine spending by using data elements available only at the header level may need to aggregate line-level payments up to the header level before conducting their analysis. Alternatively, TAF users who wish to examine spending by using data elements available only at the line level may need to develop a method for disaggregating header-level payments across the lines within a claim. When payment amounts are consistent across header and line records, TAF users can rely on whichever payment variable is more convenient.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We selected FFS claims in the four TAF [6] claims files (IP, LT, OT, and RX) by using claim type code (CLM_TYPE_CD). [7] We further restricted the analysis to claim headers that had a positive total Medicaid payment amount [8] and at least one non-denied claim line. We tabulated the number of claim headers that met these criteria in each state and the proportion of claim headers (with at least one non-denied claim line) in which the header payment (TOT_MDCD_PD_AMT) equaled the sum of the line-level payments (MDCD_PD_AMT) among non-denied lines. [9] We considered claim headers with a total payment amount that equaled the sum of payments on the associated line records to be internally consistent, and states with higher proportions of internally consistent claims to present lower levels of concern about data quality (Table 1).
We also examined the indicator that states use to report whether payment is made at the header or the line level (PYMT_LVL_IND), to determine whether header- or line-level records are more likely to be accurate when a claim’s payment amounts are inconsistent across header and line records. We considered a state’s data to be unusable if more than 5 percent of claims had inconsistent header- and line-level payments and the state had a nontrivial proportion of total claims (20 percent or higher) with unknown payment level.
Table 1. Criteria for DQ assessment of payment consistency
| Percentage of claims with consistent payments across header and line records | Percentage of all claims with an unknown payment level | Level of consistency | DQ assessment |
|---|---|---|---|
| x ≥ 95 percent | Not assessed | Highly consistent | Low concern |
| 5 percent ≤ x < 95 percent | x < 20 percent | Mixed consistency | Medium concern a |
| x < 5 percent | x < 20 percent | Highly inconsistent | High concern a |
| x < 95 percent | x ≥ 20 percent | Highly inconsistent | Unusable a |
a Both criteria must be true for a state to receive the given DQ Assessment
Methods previously used to assess data quality
Table 2 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 2. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
FFS records (claim type 1 or A) were retained in the analysis. We excluded records with all other claim type values, including capitation payments, managed care encounters, service tracking claims, and supplemental payments.
Nearly all FFS claims should have a positive total Medicaid payment amount because the TAF only includes non-void, non-denied final action claims that incorporate adjustments made to payments. Although the TAF does include claims with unexpected payment amounts, the vast majority of FFS claims have positive payment amounts. More information can be found in the DQ Atlas single topic display for Missing Payment Data - FFS Claims .
In some states, Disproportionate Share Hospital (DSH) payment amounts are included on FFS claim headers. For claims in the IP file with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the header-level payment amount (TOT_MDCD_PD_AMT - MDCD_DSH_PD_AMT).
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
FFS records (claim type 1 or A) were retained in the analysis. We excluded records with all other claim type values, including capitation payments, managed care encounters, service tracking claims, and supplemental payments.
Nearly all FFS claims should have a positive total Medicaid payment amount because the TAF only includes non-void, non-denied final action claims that incorporate adjustments made to payments. Although the TAF does include claims with unexpected payment amounts, the vast majority of FFS claims have positive payment amounts. More information can be found in the DQ Atlas single topic display for Missing Payment Data - FFS Claims .
In some states, Disproportionate Share Hospital (DSH) payment amounts are included on FFS claim headers. For claims in the IP file with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the header-level payment amount (TOT_MDCD_PD_AMT - MDCD_DSH_PD_AMT).
States report Medicaid payments for a covered service both on claim headers and on claim lines. This analysis identifies how often the payment amounts captured in these two places are inconsistent on IP claims.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""6061"", ""relatedTopics"": [{""measureId"": 80, ""measureName"": ""Payment Data Consistency - LT"", ""groupId"": 8, ""groupName"": ""Payments"", ""order"": 1}, {""measureId"": 79, ""measureName"": ""Payment Data Consistency - OT"", ""groupId"": 8, ""groupName"": ""Payments"", ""order"": 2}, {""measureId"": 78, ""measureName"": ""Payment Data Consistency - RX"", ""groupId"": 8, ""groupName"": ""Payments"", ""order"": 3}]}" 82,"{""measureId"": 82, ""measureName"": ""1915(c) Participation"", ""groupId"": 1, ""groupName"": ""Enrollment Benchmarking"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-1915-c-Participation.pdf"", ""background"": {""content"": ""Beneficiaries who use long-term services and supports (LTSS) account for a disproportionate amount of spending in the Medicaid program. In 2013, LTSS users made up approximately 6 percent of Medicaid beneficiaries, but accounted for nearly 42 percent of all Medicaid expenditures. [1] Analyses of LTSS, whether focused on the users of these services or on the services themselves, typically define two major categories of LTSS: (1) institutional-based care and (2) home- and community-based services (HCBS). The latter has been at the center of several different policies enacted since the launch of the 2001 New Freedom Initiative for individuals with disabilities.
This analysis focuses on 1915(c) waiver enrollment, the key pathway that Medicaid beneficiaries use to access HCBS in most states. Nearly half of the Medicaid beneficiaries who use HCBS access those services through 1915(c) waiver programs. [2] Most states operate more than one 1915(c) waiver program, with the programs serving different target groups and offering different services. An estimated 1.7 million beneficiaries participate in these programs across 47 states and the District of Columbia. [3] , [4] Section 1915(c) waivers also account for over half of Medicaid spending on HCBS, and for 29 percent of total Medicaid LTSS expenditures. [5]
The T-MSIS Analytic Files (TAF) were designed to make it easier to identify participants and track service use in 1915(c) waivers and other HCBS programs. These files include new data elements that enable in-depth study of HCBS, [6] but their utility is contingent on complete and accurate state reporting. In this analysis, we assess state reporting of 1915(c) participants in TAF and whether it aligns with their 1915(c) waiver status. We also explore how many of the TAF data elements that can be used to identify 1915(c) waiver participants were populated in each state. In years prior to 2020, we compare the count of 1915(c) waiver participants found in TAF to participant counts from states’ CMS 372 submissions. [7]
Medicaid and CHIP Payment and Access Commission (MACPAC). “MACStats: Medicaid and CHIP Data Book. Exhibit 20. Distribution of Medicaid Enrollment and Benefit Spending by Users and Non-Users of Long-Term Services and Supports, FY 2018.” Washington, DC: MACPAC. Available at. https://www.macpac.gov/wp-content/uploads/2015/01/EXHIBIT-20.-Distribution-of-Medicaid-Enrollment-and-Benefit-Spending-by-Users-and-Non-Users-of-Long-Term-Services-and-Supports-FY-2018.pdf Accessed April 26, 2019.
Eiken, S. “Medicaid Long-Term Services and Supports Beneficiaries, 2013.” Cambridge, MA: Truven Health Analytics, prepared for the Centers for Medicare & Medicaid Services, September 22, 2017. Available at https://www.medicaid.gov/sites/default/files/2019-12/ltss-beneficiaries-2013.pdf . Accessed March 28, 2019.
Arizona, Rhode Island, and Vermont did not have 1915(c) waiver programs during this time period, but they provided similar services through 1115 waiver authority.
Medicaid Section 1915(c) Waiver Programs Annual Expenditures and Beneficiaries Report: Analysis of CMS 372 Annual Reports, 2015-2017.” Chicago, IL: Mathematica, September 18, 2020.
Eiken, S., K. Sredl, B. Burwell, and R. Woodward. “Medicaid Expenditures for Long-Term Services and Supports in FY 2016.” Cambridge, MA: IBM Watson Health, prepared for the Centers for Medicare & Medicaid Services, May 2018. Available at https://www.medicaid.gov/sites/default/files/2019-12/ltssexpenditures2016.pdf . Accessed March 26, 2019.
For example, states can report the type of HCBS program at the claim line level (HCBS_SRVC_CD in the OT file).
The reports for 2015-2017, 2017-2018, and 2018-2019 can be found with other CMS 372 reports at https://www.medicaid.gov/medicaid/long-term-services-supports/reports-evaluations/index.html . The DQ Assessment for years prior to 2020 used these reports to obtain the CMS-372 counts of 1915(c) participants. Given that the reports will not be updated using CMS-372 data after 2019, different DQ assessment criteria are used for 2020 and later years.
Medicaid and CHIP Payment and Access Commission (MACPAC). \u201cMACStats: Medicaid and CHIP Data Book. Exhibit 20. Distribution of Medicaid Enrollment and Benefit Spending by Users and Non-Users of Long-Term Services and Supports, FY 2018.\u201d Washington, DC: MACPAC. Available at. https://www.macpac.gov/wp-content/uploads/2015/01/EXHIBIT-20.-Distribution-of-Medicaid-Enrollment-and-Benefit-Spending-by-Users-and-Non-Users-of-Long-Term-Services-and-Supports-FY-2018.pdf Accessed April 26, 2019.
Eiken, S. \u201cMedicaid Long-Term Services and Supports Beneficiaries, 2013.\u201d Cambridge, MA: Truven Health Analytics, prepared for the Centers for Medicare & Medicaid Services, September 22, 2017. Available at https://www.medicaid.gov/sites/default/files/2019-12/ltss-beneficiaries-2013.pdf . Accessed March 28, 2019.
Arizona, Rhode Island, and Vermont did not have 1915(c) waiver programs during this time period, but they provided similar services through 1115 waiver authority.
Medicaid Section 1915(c) Waiver Programs Annual Expenditures and Beneficiaries Report: Analysis of CMS 372 Annual Reports, 2015-2017.\u201d Chicago, IL: Mathematica, September 18, 2020.
Eiken, S., K. Sredl, B. Burwell, and R. Woodward. \u201cMedicaid Expenditures for Long-Term Services and Supports in FY 2016.\u201d Cambridge, MA: IBM Watson Health, prepared for the Centers for Medicare & Medicaid Services, May 2018. Available at https://www.medicaid.gov/sites/default/files/2019-12/ltssexpenditures2016.pdf . Accessed March 26, 2019.
For example, states can report the type of HCBS program at the claim line level (HCBS_SRVC_CD in the OT file).
The reports for 2015-2017, 2017-2018, and 2018-2019 can be found with other CMS 372 reports at https://www.medicaid.gov/medicaid/long-term-services-supports/reports-evaluations/index.html . The DQ Assessment for years prior to 2020 used these reports to obtain the CMS-372 counts of 1915(c) participants. Given that the reports will not be updated using CMS-372 data after 2019, different DQ assessment criteria are used for 2020 and later years.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
Using the TAF, [8] we examined records from the TAF annual Demographic and Eligibility (DE) file and the other services (OT) file. We included fee-for-service (FFS) claims and managed care encounter records for beneficiaries who were enrolled in non-CHIP Medicaid at some point during the calendar year. [9] , [10] The TAF data include five different data elements that can be used to identify 1915(c) waiver participants (Table 1).
Table 1. TAF data elements that could be used to identify participants in 1915(c) waiver programs
| TAF data element | File | How data element is used to identify 1915(c) waiver participants |
|---|---|---|
| _1915C_WVR_TYPE | DE | This data element represents the most recent value for type of 1915(c) waiver enrollment during the calendar year, and is not associated with a duration of enrollment (that is, the coverage could have been for any number of days during the year). The type refers to the target group served. We flagged the beneficiary as a 1915(c) waiver participant if there was a non-null value in this field. |
| PGM_TYPE_CD | OT | This code indicates the special program under which the service was provided. We considered the claim to represent a service provided through a 1915(c) waiver program if the value for this data element was “07: Home and Community Based Care Waiver Services.” |
| WVR_TYPE_CD | OT | This code indicates the waiver type under which the service was provided. We identified the service as provided through a 1915(c) waiver if the value was “06–20” or “33” (which represent 1915(c) waivers). |
| HCBS_SRVC_CD | OT | This code indicates that the service is an HCBS “for an individual with chronic medical and/or mental conditions” and serves “to help clearly delineate between acute care and long-term care provided in the home and community setting.” We considered the service to have been provided through a 1915(c) waiver if the value was “4: The HCBS service was provided under a 1915(c) HCBS Waiver.” |
| XIX_SRVC_CTGRY_CD | OT | This code indicates the category of service for the paid claim based on the CMS-64 form that states use to report their expenditures and request federal financial participation. We identified the service as provided through a 1915(c) waiver if the value for this data element was “19A: Home & Community-Based Services - Reg. Pay. (Waiv).” |
We calculated the number of Medicaid participants in 1915(c) waivers by using the TAF DE 1915(c) waiver flag (_1915C_WVR_TYPE) because states should report program participation in enrollment data.
For years prior to 2020, we compared the TAF DE participant counts with states’ CMS 372 reports for waiver programs ending in the corresponding time period. [11] States are required to submit the CMS 372 report each year to document that their 1915(c) waiver programs meet cost neutrality requirements. The reports contain the number of participants in each waiver program, the number of days participants received waiver services, waiver program Medicaid expenditures, and total Medicaid expenditures associated with waiver participants. [12] Although the CMS 372 report is the authoritative source for 1915(c) waiver participant data, it has several limitations for benchmarking purposes: (1) if a state’s reporting for a given year is incomplete or missing, the most recent, valid report submitted within the last four years is used; (2) because states report annual time periods based on their waiver effective dates, those periods do not necessarily align with the calendar year (CY); and (3) the reports are not audited, although CMS reviews them in comparison with each state’s approved waiver program and to determine whether 372 reporting requirements were met. [13] , [14] Only reports “accepted” by CMS are reflected in these benchmarks. For years prior to 2020, we compared the TAF to the benchmark by calculating a percent difference. Because the benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, the percent difference is calculated as a percent error or change: (TAF – CMS372) / CMS372. We then assessed the quality of the 1915(c) participation information based on the percent difference between the TAF and the benchmark (Table 2).
Table 2. Criteria for DQ assessment of 1915(c) waiver participants for years prior to 2020
| Percent difference in TAF and CMS 372 participant counts | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| > 50 percent or state does not report any participant counts where expected | Unusable |
Because the annual 1915(c) beneficiaries report will no longer use CMS-372 data after 2019 and cannot be used as a benchmark for those years, 11 the data quality of 1915(c) participation for years after 2019 is assessed based on two criteria: (1) the percentage of total Medicaid beneficiaries participating in 1915(c) waivers, and (2) the alignment between the expected reporting of 1915(c) participants in TAF and the state’s active 1915(c) waiver status. If a state has an unusually high percentage of Medicaid beneficiaries who are also 1915(c) enrollees, this likely indicates a data quality issue with the reporting of 1915(c) participants in TAF. [15] Additionally, the number 1915(c) participants that the state reports in TAF should appropriately align with the state’s 1915(c) waiver status, and misalignment likely signals a data quality issue.
We first determined the percentage of Medicaid enrollees in the state who are also 1915(c) participants in the DE file. We then compared the state’s active 1915(c) waiver status with the state’s reported 1915(c) participant count. A state that did not have any active waivers in the year but reported a non-zero number of 1915(c) participants in TAF was designated as “Not Aligned”. Similarly, a state that had at least one active waiver in the year but reported no 1915(c) participants in TAF was also designated as “Not Aligned”. We then assessed the quality of the 1915(c) participation information based on these two measures (Table 3).
Table 3. Criteria for DQ assessment of 1915(c) waiver participants for years after 2019
| Percentage of total Medicaid beneficiaries in TAF who are participating in 1915(c) waivers | Alignment with state’s 1915(c) waiver status | DQ assessment |
|---|---|---|
| x ≤ 10 percent | Aligned | Low concern |
| 10 percent < x ≤ 20 percent | Aligned | Medium concern |
| 20 percent < x ≤ 50 percent | Aligned | High concern |
| > 50 percent | Aligned | Unusable |
| Any value | Not Aligned | Unusable |
We also present information about the extent to which the four OT file data elements—program type (PGM_TYPE_CD), waiver type (WVR_TYPE_CD), HCBS service code (HCBS_SRVC_CD), and Title XIX service category (XIX_SRVC_CTGRY_CD)—can be used to identify 1915(c) participants, although this information was not used as part of the DQ assessment. All four data elements should be consistent within each OT record. In other words, if a claim or encounter is associated with a 1915(c) waiver service, the program type, waiver type, HCBS service code, and Title XIX service category should all be coded to show the service as covered by a 1915(c) waiver. However, states may vary with regard to which data elements are correctly populated. In the data table, we present the number of unduplicated beneficiaries identified using the DE and each of the four OT file data elements.
Methods previously used to assess data quality
Table 4 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 4. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Claim type code (CLM_TYPE_CD) was used to determine which OT records to include. Medicaid fee-for-service records (claim type 1) and managed care encounters (claim type 3) were retained in the analysis. We excluded fee-for-service records and managed care encounters associated with beneficiaries in the Children’s Health Insurance Program (claim types A and C) as well as other types of payments such as capitation, service tracking, and supplemental payment claims.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the beneficiary had CHIP code value of 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-75, we considered the beneficiary to have non-CHIP Medicaid enrollment during the month. For analyses of 2014 through 2017 TAF, we also use CHIP code 4 (Medicaid and S-CHIP) to identify Medicaid beneficiaries, because CHIP code 4 is a valid value for those TAF data years.
For example, for the 2016 analysis, we used TAF DE data from 2016 and CMS 372 reports for 2015-2016. CMS authorized the use of unpublished 2015–2016 CMS 372 data to use as a benchmark for the 2016 analysis.
“Medicaid Section 1915(c) Waiver Programs Annual Expenditures and Beneficiaries Report: Analysis of CMS 372 Annual Reports.” Chicago, IL: Mathematica, September 18, 2020. The reports for 2015-2017, 2017-2018, and 2018-2019 can be found with other CMS 372 reports at https://www.medicaid.gov/medicaid/long-term-services-supports/reports-evaluations/index.html . The DQ Assessment for years prior to 2020 used these reports to obtain the CMS-372 counts of 1915(c) participants. Given that the reports will not be updated using CMS-372 data after 2019, different DQ assessment criteria are used for 2020 and later years.
“Medicaid Section 1915(c) Waiver Programs Annual Expenditures and Beneficiaries Report: Analysis of CMS 372 Annual Reports, 2015-2017.” Chicago, IL: Mathematica, September 18, 2020. The 2015-2017 report can be found with previous CMS 372 reports at https://www.medicaid.gov/medicaid/long-term-services-supports/reports-evaluations/index.html .
Centers for Medicare & Medicaid Services. “Application for a §1915(c) Home and Community-Based Waiver: Instructions, Technical Guide and Review Criteria.” Version 3.6. January 2019.” Available at https://wms-mmdl.cms.gov/WMS/help/35/Instructions_TechnicalGuide_V3.6.pdf . Accessed October 13, 2020.
Preliminary investigations have shown that most states report <8% of their Medicaid beneficiaries as participating in 1915(c) waivers. Benchmark thresholds have been set to assess state reporting quality based on this finding.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Claim type code (CLM_TYPE_CD) was used to determine which OT records to include. Medicaid fee-for-service records (claim type 1) and managed care encounters (claim type 3) were retained in the analysis. We excluded fee-for-service records and managed care encounters associated with beneficiaries in the Children\u2019s Health Insurance Program (claim types A and C) as well as other types of payments such as capitation, service tracking, and supplemental payment claims.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the beneficiary had CHIP code value of 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-75, we considered the beneficiary to have non-CHIP Medicaid enrollment during the month. For analyses of 2014 through 2017 TAF, we also use CHIP code 4 (Medicaid and S-CHIP) to identify Medicaid beneficiaries, because CHIP code 4 is a valid value for those TAF data years.
For example, for the 2016 analysis, we used TAF DE data from 2016 and CMS 372 reports for 2015-2016. CMS authorized the use of unpublished 2015\u20132016 CMS 372 data to use as a benchmark for the 2016 analysis.
\u201cMedicaid Section 1915(c) Waiver Programs Annual Expenditures and Beneficiaries Report: Analysis of CMS 372 Annual Reports.\u201d Chicago, IL: Mathematica, September 18, 2020. The reports for 2015-2017, 2017-2018, and 2018-2019 can be found with other CMS 372 reports at https://www.medicaid.gov/medicaid/long-term-services-supports/reports-evaluations/index.html . The DQ Assessment for years prior to 2020 used these reports to obtain the CMS-372 counts of 1915(c) participants. Given that the reports will not be updated using CMS-372 data after 2019, different DQ assessment criteria are used for 2020 and later years.
\u201cMedicaid Section 1915(c) Waiver Programs Annual Expenditures and Beneficiaries Report: Analysis of CMS 372 Annual Reports, 2015-2017.\u201d Chicago, IL: Mathematica, September 18, 2020. The 2015-2017 report can be found with previous CMS 372 reports at https://www.medicaid.gov/medicaid/long-term-services-supports/reports-evaluations/index.html .
Centers for Medicare & Medicaid Services. \u201cApplication for a \u00a71915(c) Home and Community-Based Waiver: Instructions, Technical Guide and Review Criteria.\u201d Version 3.6. January 2019.\u201d Available at https://wms-mmdl.cms.gov/WMS/help/35/Instructions_TechnicalGuide_V3.6.pdf . Accessed October 13, 2020.
Preliminary investigations have shown that most states report <8% of their Medicaid beneficiaries as participating in 1915(c) waivers. Benchmark thresholds have been set to assess state reporting quality based on this finding.
In Medicaid, Section 1915(c) waivers are the key pathway for beneficiaries to access home- and community-based services in most states. This analysis examines how well the counts of participants in 1915(c) waivers align with an external benchmark, the CMS 372 reports.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""7041"", ""relatedTopics"": []}" 85,"{""measureId"": 85, ""measureName"": ""Service Tracking Claims"", ""groupId"": 7, ""groupName"": ""Non-Claim Records"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Service-Tracking-Claims.pdf"", ""background"": {""content"": ""States submit several types of service use and financial transaction records into T-MSIS: (1) fee-for-service (FFS) claims, which represent payments made directly to providers for services rendered to Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries; (2) capitated payments, which represent per-person-per-month payments made to managed care plans; (3) managed care encounters, which represent services rendered to beneficiaries covered under a capitation arrangement; (4) service tracking claims, which represent lump sum payments that cannot be attributed to a single beneficiary (for instance, drug rebates or disproportionate share hospital payments); and (5) supplemental payments, which represent additional payments for services provided to a specific beneficiary.
All of these records are available in the T-MSIS Analytic Files (TAF) Research Identifiable Files (RIFs). [1] Any analysis that uses these data should include only the types of records that will help answer the research question at hand. For example, tabulations of total state Medicaid expenditures should include FFS claims, capitated payments, service tracking claims, and supplemental payments, but they should exclude managed care encounters, which do not represent payments made by the state. In contrast, analyses of beneficiary service use should include FFS claims and managed care encounters (when applicable), but they should exclude capitated payments, service tracking claims, and supplemental payments.
TAF users can distinguish FFS claims, capitated payments, managed care encounters, service tracking claims, supplemental payments, and other payments by using the claim type code found on header records that summarize the claim. The claim type code can also be used to differentiate between service use and payments for Medicaid beneficiaries (including M-CHIP) and for separate CHIP beneficiaries. In addition, claim type code includes values for “other” program type records, which may or may not represent services that qualify for federal matching funds under Title XIX or Title XXI. These records are referred to as “other” or “non-program” records and have claim type code equal to U, V, W, X, or Y.
The distribution of claim type in a given state is driven by how that state implemented its Medicaid and CHIP programs as well as the capacity of the state’s Medicaid Management Information System to report each claim type.
This data quality assessment examines the reporting and usability of service tracking claims, supplemental payments and other, non-program claims. [2]
Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
For more information about what types of claims are in TAF and potential data quality issues, see “TAF Technical Guidance: Claims Files” on ResDAC.org.
Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
For more information about what types of claims are in TAF and potential data quality issues, see \u201cTAF Technical Guidance: Claims Files\u201d on ResDAC.org.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We examined service tracking claims, supplemental payments and other, non-program claims using claim type code (CLM_TYPE_CD) on records in the TAF inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims files. [3] Service tracking claims have claim type code values of 4 or D. Supplemental claims have claim type code values of 5 or E. Other, non-program claims have claim type code values of U, V, W, X, or Y. For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [4]
Service tracking claims
The MSIS identification number (MSIS_IDENT_NUM), service tracking type code (SRVC_TRKNG_TYPE_CD), and service tracking payment amount (SRVC_TRKNG_PYMT_AMT) are key data elements for identifying service tracking claims. Because these claims represent lump sum payments that cannot be attributed to a single Medicaid or CHIP beneficiary, the MSIS identification number should begin with an “&” or be missing. In addition, the service tracking type and payment amount needs to be populated for the claim to be useful for analytic purposes. We classified a claim as having a data quality problem if it met at least one of following conditions:
We grouped states into categories of low, medium, and high data quality concern based on the percentage of their records that were problematic according to any of the three conditions.
Table 1. Criteria for DQ assessment of service tracking claims
| Percentage of claims with a problematic value in the MSIS ID, service tracking type, or service tracking payment field | DQ assessment |
|---|---|
| 0 percent ≤ x < 10 percent | Low concern |
| 10 percent ≤ x < 20 percent | Medium concern |
| 20 percent ≤ x < 50 percent | High concern |
| x ≥ 50 percent | Unusable |
Supplemental payment records
The total Medicaid paid amount (TOT_MDCD_PD_AMT) is a key data element for supplemental payments. If there is no information in the payment field, the record cannot be used for research. These records should also have (1) a valid MSIS ID that does not begin with an “&”; (2) a missing service tracking type; and (3) a missing or zero service tracking payment amount. [5] We classified a claim as having a data quality problem if it met any one of the following conditions:
We grouped the states into low, medium, high, and unusable categories of concern based on the percentage of their supplemental payment records that were problematic according to any of the four conditions.
Table 2. Criteria for DQ assessment of supplemental payment records
| Percentage of claims with a problematic value in the total Medicaid paid amount, MSIS ID, service tracking type, or service tracking payment field | DQ assessment |
|---|---|
| 0 percent ≤ x < 10 percent | Low concern |
| 10 percent ≤ x < 20 percent | Medium concern |
| 20 percent ≤ x < 50 percent | High concern |
| x ≥ 50 percent | Unusable |
Other, non-program claims
TAF users should be aware if the data files for a state include records for other, non-program claims so that they can determine whether to include these claims in analyses. These records may represent services and payments that do not qualify for a federal match or may represent services that receive a federal match that the state wants to distinguish for some reason. Because states were not given specific guidance on what claims should be reported as other and because states are not required to justify why they have submitted a claim with the other, non-program code values, it is unclear what these claims represent in every state and whether a state’s use of this claim type changes over time. The lack of detail for these claims may make them unusable for many analytic purposes. We classified the state as high or low concern based on the percentage of all claims for the state that have an other, non-program claim type code (Table 3).
Table 3. Criteria for DQ assessment of other, non-program claims
| Percentage of claims that have an other, non-program claim type code | DQ assessment |
|---|---|
| x < 2 percent | Low concern |
| x ≥ 2 percent | High concern |
Methods previously used to assess data quality
Table 4 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 4. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND=0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
If the fields for service tracking type and service tracking amount are populated or if the MSIS ID is missing or starts with an “&”, then a TAF user cannot determine whether the record is a supplemental payment or a service tracking record.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND=0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
If the fields for service tracking type and service tracking amount are populated or if the MSIS ID is missing or starts with an \u201c&\u201d, then a TAF user cannot determine whether the record is a supplemental payment or a service tracking record.
States submit several types of service use and financial transaction records into T-MSIS. One type of financial transaction is service tracking claims, which represent lump sum payments that cannot be attributed to a single beneficiary. This analysis identifies unexpected coding patterns for service tracking claims that may indicate data quality problems.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5181"", ""relatedTopics"": [{""measureId"": 60, ""measureName"": ""Supplemental Payments"", ""groupId"": 7, ""groupName"": ""Non-Claim Records"", ""order"": 0}, {""measureId"": 61, ""measureName"": ""Non-Program (Other) Claims"", ""groupId"": 7, ""groupName"": ""Non-Claim Records"", ""order"": 1}]}" 86,"{""measureId"": 86, ""measureName"": ""Missing Payment Data - Encounters"", ""groupId"": 8, ""groupName"": ""Payments"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Missing-Pmt-Encounters.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) provide a unique source of expenditure information for all Medicaid and CHIP programs nationally and by state. Although other data sources, such as the Medicaid Budget & Expenditure System files, provide aggregate state-level expenditure information, T-MSIS is the only data source that allows users to analyze payments associated with managed care organizations, individual providers, subgroups of beneficiaries, and/or specific services. The Centers for Medicare & Medicaid Services (CMS) and other TAF users require accurate payment data to calculate total cost of care metrics; to compare costs for specific conditions, services, or populations; and to compare costs across states or different payment systems.
Medicaid and CHIP payments made on behalf of specific beneficiaries are captured in four types of records in the TAF claims files: (1) fee-for-service (FFS) claims, which represent payments to medical providers made directly by the state Medicaid or CHIP agency; (2) capitation payments, which reflect a set per member per month (PMPM) rate paid by the state Medicaid or CHIP agency to a managed care organization (MCO), prepaid health plan (PHP), or primary care provider; (3) managed care encounter records, which reflect payments made by MCOs or PHPs to providers for services rendered to covered beneficiaries; and (4) supplemental payments, which represent payments made in addition to a capitation payment or negotiated rate. [1] Because TAF claims records only include non-void, non-denied final action claims, nearly all of these records should have a positive total Medicaid paid amount. [2] , [3] A high percentage of claims reported with zero, missing, or negative payments may suggest a data quality or completeness issue, which may affect cost estimates based on TAF data.
This data quality assessment examines the extent to which states are reporting FFS claims and managed care encounters with missing or invalid payment data, as well as variation in the completeness and usability of the payment data on FFS claims and managed care encounters by file type. [4]
The TAF also include payment records that are not tied to specific beneficiaries, such as aggregate payments to transportation providers, called service tracking claims. Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
There are a few situations in which FFS claims are not expected to have a positive payment amount, including (1) claims for which Medicare or another liable third party already paid the full Medicaid allowable amount and (2) claims that are not paid on a FFS basis despite being processed at the individual claim level.
Previously, managed care encounters could be reported without positive payment amounts if the MCOs or PHPs had not agreed to provide payment data to the state. CMS expected all states to report provider payment amounts on managed care encounters no later than June 30, 2019, necessitating updates to some states’ managed care contracts. Because managed care payment data are proprietary, only TAF users with approval from CMS will be able to access information on what managed care plans pay providers for services. For more information, see: Reporting Paid and Billed Amounts on Managed Care Encounters in T-MSIS. July 2019. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/52005 .
Payments on managed care encounter records are redacted from the TAF RIF and require special permission to access.
The TAF also include payment records that are not tied to specific beneficiaries, such as aggregate payments to transportation providers, called service tracking claims. Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1.
There are a few situations in which FFS claims are not expected to have a positive payment amount, including (1) claims for which Medicare or another liable third party already paid the full Medicaid allowable amount and (2) claims that are not paid on a FFS basis despite being processed at the individual claim level.
Previously, managed care encounters could be reported without positive payment amounts if the MCOs or PHPs had not agreed to provide payment data to the state. CMS expected all states to report provider payment amounts on managed care encounters no later than June 30, 2019, necessitating updates to some states\u2019 managed care contracts. Because managed care payment data are proprietary, only TAF users with approval from CMS will be able to access information on what managed care plans pay providers for services. For more information, see: Reporting Paid and Billed Amounts on Managed Care Encounters in T-MSIS. July 2019. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/52005 .
Payments on managed care encounter records are redacted from the TAF RIF and require special permission to access.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We examined the values in the total Medicaid paid amount field (TOT_MDCD_PD_AMT) in the TAF [5] inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) header files. We included both Medicaid and CHIP FFS claims and managed care encounters, but we excluded (1) crossover claims [6] (those for which Medicare is the primary payer, and Medicaid is responsible only for covering the remaining cost-sharing on behalf of dually eligible beneficiaries); (2) capitation payments; (3) supplemental payments; and (4) service tracking payments. [7] , [8] We excluded states if the low volume of claims in the TAF rendered the data unusable for analysis. For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [9]
For each state, we calculated the percentage of records in each file where the total Medicaid paid amount was (1) missing, (2) zero dollars, or (3) any negative dollar amount. We then calculated the percentage of records that had a positive payment value. We grouped states into categories of low concern, medium concern, and high concern about the usability of their data, depending on the percentage of records that had a missing, zero, or negative payment value (Table 1). [10] , [11] The data quality assessment is based on a combination of the four claims file types, but the level of missingness can vary across file types.
Table 1. Criteria for DQ assessment of payment data
| Percentage of claims or encounters with zero, missing or negative payment in the total Medicaid paid amount field | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
Methods previously used to assess data quality
Table 2 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 2. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We classified crossover claims as records on which the crossover claim indicator (XOVR_IND) was equal to 1. If a claim was reported with a missing value in XOVR_IND, we retained it in our analysis.
We identified capitation, supplemental, and service tracking payments by using the claim type code (CLM_TYPE_CD). We also did not include any records with a claim type of “other,” which are records that the state did not classify as Medicaid or CHIP.
We limited this analysis to only FFS claims and managed care encounters because they are the two types of claims that will be most commonly analyzed by TAF users. Although capitation payments and supplemental payments also capture important payment data that can be attributed to individual beneficiaries, reviewing the usability of payment information on those claims was outside the scope of this assessment.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “TAF Technical Guidance: How to Use Illinois Claims Data,” on ResDAC.org.
It is not uncommon for a small percentage of TAF records to have payments equal to zero dollars, but missing or negative payment amounts are much less common. We recommend that TAF users interpret zero, missing, and negative payment amounts in the same manner—all three cases represent payment information that is likely unusable for research purposes.
The relatively generous threshold of 90 percent for a low level of concern should accommodate cases in which a zero payment is valid—for example, claims for which a third party already paid the Medicaid allowable amount.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We classified crossover claims as records on which the crossover claim indicator (XOVR_IND) was equal to 1. If a claim was reported with a missing value in XOVR_IND, we retained it in our analysis.
We identified capitation, supplemental, and service tracking payments by using the claim type code (CLM_TYPE_CD). We also did not include any records with a claim type of \u201cother,\u201d which are records that the state did not classify as Medicaid or CHIP.
We limited this analysis to only FFS claims and managed care encounters because they are the two types of claims that will be most commonly analyzed by TAF users. Although capitation payments and supplemental payments also capture important payment data that can be attributed to individual beneficiaries, reviewing the usability of payment information on those claims was outside the scope of this assessment.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cTAF Technical Guidance: How to Use Illinois Claims Data,\u201d on ResDAC.org.
It is not uncommon for a small percentage of TAF records to have payments equal to zero dollars, but missing or negative payment amounts are much less common. We recommend that TAF users interpret zero, missing, and negative payment amounts in the same manner\u2014all three cases represent payment information that is likely unusable for research purposes.
The relatively generous threshold of 90 percent for a low level of concern should accommodate cases in which a zero payment is valid\u2014for example, claims for which a third party already paid the Medicaid allowable amount.
The payment amount on TAF managed care encounters captures payments from managed care organizations or prepaid health plans to medical providers for services rendered to covered beneficiaries. Since the TAF excludes fully denied and voided claims, all managed care encounters should have a positive payment amount. This analysis examines the extent to which managed care encounters have missing, zero, or negative payment amounts, which are likely to represent data quality issues.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""6011"", ""relatedTopics"": [{""measureId"": 66, ""measureName"": ""Missing Payment Data - FFS Claims"", ""groupId"": 8, ""groupName"": ""Payments"", ""order"": 0}]}" 87,"{""measureId"": 87, ""measureName"": ""Medicaid Beneficiary Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Bene-Expenditures.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) provide a unique source of expenditure information on states’ Medicaid and CHIP programs. Although other data sources provide aggregate state-level expenditure information, the TAF allow users to analyze payments associated with managed care organizations, individual providers, subgroups of beneficiaries, and specific services. The Centers for Medicare & Medicaid Services (CMS) and other TAF users require accurate payment data to calculate total cost of care metrics; to compare costs for specific conditions, services, or populations; and to compare costs across states or different payment systems.
States may reimburse health care providers for services delivered to Medicaid beneficiaries by paying directly for each covered service on a fee-for-service (FFS) basis or by paying a flat monthly payment per beneficiary for a set of services that is contracted to another entity—such as a managed care plan—which then assumes responsibility for delivering care to the beneficiary. [1] FFS expenditures and monthly beneficiary payments—which can be linked to specific beneficiaries [2] —account for the majority of Medicaid expenditures in every state.
In addition to FFS claims and monthly beneficiary payments, TAF also includes other financial transactions, which are mostly captured on service tracking claims. Service tracking claims represent lump-sum payments made for services that cannot be attributed to a specific beneficiary, such as disproportionate share hospital (DSH) payments or payments to providers made under the Upper Payment Limit (UPL) demonstration. Table 1 summarizes the expenditure information available in TAF by expenditure type.
Table 1. Types of Medicaid expenditures in TAF
| Expenditure type | Description | Linked to specific beneficiaries? |
|---|---|---|
| FFS expenditures |
| Yes |
| Monthly beneficiary payments |
| Yes b |
| Other expenditures |
| No |
a Under federal law, states must pay federally qualified health centers (FQHCs) the difference between a managed care organization’s per-service payment and the amount determined under the state’s prospective payment system. States often report these “wraparound” payments on supplemental payment records (those with claim type code = 5). For more information, refer to the National Association of Community Health Centers: https://cdn1.digitellinc.com/uploads/nachc/articles/b9784c7326bea6f2202ef78a41872e03.pdf .
b Capitation payments reported on service tracking claims (which are a subset of total monthly beneficiary payments) cannot be linked to a specific beneficiary. All other monthly beneficiary payments can be linked to a specific beneficiary.
To claim federal matching funds for their medical assistance programs, states report aggregate expenditure data for Medicaid beneficiaries covered under Title XIX of the Social Security Act to CMS through the Medicaid Budget and Expenditure System (MBES). [3] The MBES system is designed to capture information states would report on the Quarterly Medicaid Statement of Expenditures for the Medical Assistance Program Form CMS-64; therefore, the data are commonly referred to as the CMS-64 data. Because of the oversight involved in ensuring the correct federal reimbursement to state Medicaid programs, the CMS-64 data are seen as a highly reliable source of expenditure information. If the expenditures in the TAF do not align with those in the CMS-64, that may suggest TAF data are incomplete or have other data quality errors that render them unrepresentative of actual Medicaid expenditures.
In this data quality assessment, we evaluate the completeness of total Medicaid expenditures in the TAF by comparing them to the total net expenditures states report in the CMS-64. However, benchmarking TAF expenditure data against CMS-64 data is not straightforward. Although the data in both systems are similar, there are some important differences: [4]
More information on FFS expenditure benchmarking can be found in the DQ Atlas single topic displays for Total FFS Expenditures , FFS Inpatient Expenditures , FFS Long-term Care Expenditures , FFS Other Medical Expenditures , and FFS Prescription Drug Expenditures . More information on expenditure benchmarking for monthly beneficiary payments can be found in the DQ Atlas single topic displays for Total Monthly Beneficiary Payments , CMC Payments , PHP Payments , PCCM Fees , and Premium Assistance Payments .
Capitation payments reported on service tracking claims (which are a subset of total monthly beneficiary payments) cannot be linked to a specific beneficiary but are still considered to be monthly beneficiary payments. All other monthly beneficiary payments, as well as FFS expenditures and wraparound payments, can be linked to a specific beneficiary.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
More information on FFS expenditure benchmarking can be found in the DQ Atlas single topic displays for Total FFS Expenditures , FFS Inpatient Expenditures , FFS Long-term Care Expenditures , FFS Other Medical Expenditures , and FFS Prescription Drug Expenditures . More information on expenditure benchmarking for monthly beneficiary payments can be found in the DQ Atlas single topic displays for Total Monthly Beneficiary Payments , CMC Payments , PHP Payments , PCCM Fees , and Premium Assistance Payments .
Capitation payments reported on service tracking claims (which are a subset of total monthly beneficiary payments) cannot be linked to a specific beneficiary but are still considered to be monthly beneficiary payments. All other monthly beneficiary payments, as well as FFS expenditures and wraparound payments, can be linked to a specific beneficiary.
Because CMS applies different federal matching rates to Medicaid and CHIP beneficiaries, they collect Medicaid and CHIP expenditure data separately. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act in the CHIP Budget and Expenditure System (CBES). States may cover children using CHIP funds by expanding their Medicaid programs (referred to as Medicaid expansion CHIP, or M-CHIP); creating a program separate from their existing Medicaid programs (referred to as separate CHIP, or S-CHIP); or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries are reported in CBES and not MBES.
For more information on the differences between the TAF and CMS-64 data on Medicaid expenditures, see the methodology brief \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
All Medicaid expenditures in TAF are used in this analysis. [5] , [6] We began by selecting all records from the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims files that represented FFS payments, capitation payments, service tracking claims, or supplemental payments by using the claim type code. [7] Managed care encounter records were excluded from the analysis because the payments recorded on those claims do not represent costs to the state Medicaid program and are not recorded on the CMS-64. [8]
Next, we further restricted FFS claims, capitation payments, and supplemental payments to those that matched to an eligibility record in the Demographic and Eligibility (DE) file that indicated that the beneficiary was enrolled in non-CHIP Medicaid at some point during the calendar year. [9] , [10] This is necessary because the CMS-64 benchmark includes only non-CHIP Medicaid expenditures, and the TAF claim type variable alone cannot distinguish payments made for Medicaid-expansion CHIP enrollees (whose costs are not reported on the CMS-64) from payments made for non-CHIP Medicaid beneficiaries (whose costs are reported on the CMS-64). [11]
Finally, we excluded TAF records that represent electronic health record (EHR) payments, Medicare premiums, and drug rebates. [12] EHR payments were excluded because they are considered administrative costs and thus not included in the CMS-64 data we use as a benchmark. Medicare premiums and drug rebates rarely appear in the TAF, and so were excluded from both the benchmark and TAF data for this analysis.
Once we had selected the set of TAF records included in the benchmarking analysis, we aggregated expenditures in two ways for each state: (1) total Medicaid expenditures —which includes FFS, monthly beneficiary payments, and other expenditures—and (2) Medicaid beneficiary expenditures , which only includes FFS and monthly beneficiary payments. For FFS claims in the IP file, we used the total Medicaid paid amount minus the DSH payment (both on the header record) to tabulate expenditures; for FFS claims in the OT, LT, and RX files, we used the total Medicaid paid amount only. For capitation and supplemental payment records, we used the total Medicaid paid amount on the header record to tabulate expenditures. For capitation payments on service tracking claims [13] , we used the service tracking payment amount on the header record; if that field was zero or missing, we used the total Medicaid paid amount. For all other expenditures on service tracking claims, we used the DSH payment field [14] ; if the DSH payment field was zero or missing, we used the service tracking payment amount. If the DSH payment field and service tracking payment amount were both zero or missing, we used the total Medicaid paid amount. [15]
We developed the total expenditure benchmark using four quarterly net expenditures reports from MBES that cover a calendar year. [16] We calculated total expenditures in the CMS-64 using the total net expenditures for medical assistance programs reported by each state (category of service code 50 on the CMS-64), minus expenditures that do not typically appear in TAF: Medicare premiums (category of service 17A, 17B, and 17C), drug rebates (category of service 7A1-7A7, 46A1-46A6 [17] ), and collections (adjustments related to third party liability, probate, fraud, waste, and abuse). We calculated Medicaid beneficiary expenditures in the CMS-64 by summing net expenditures for the categories of service that correspond to FFS and monthly beneficiary payments (Table 2).
Distribution of Medicaid spending by expenditure type
Each state’s expenditure data should reflect its unique Medicaid program characteristics, including the extent of Medicaid managed care and the use of DSH and other supplemental provider payments. To help users understand what types of expenditures are most common in each state, we tabulated the percentage of total CMS-64 spending that fell into each of three expenditure types: (1) FFS expenditures, (2) monthly beneficiary payments, and (3) other expenditures (see Table 1 in the background section).
Table 2 shows how we grouped the CMS-64 category of service codes into the three expenditure types used to calculate the distribution of spending in each state. The table also provides the set of TAF records that are expected to map to these three expenditure types. The Total Medicaid Expenditures topic includes all expenditures listed in Table 2, whereas the Medicaid Beneficiary Expenditures topic includes only the FFS and monthly beneficiary payments expenditure types.
Table 2. Assigning CMS-64 categories of service and TAF records to expenditure types
| Expenditure type | CMS-64 categories a | Corresponding TAF records b |
|---|---|---|
| FFS | Category of service 1A. This represents inpatient hospital services. Categories of service 2A, 3A, 4A, and 4B. These represent payments for mental health facility, nursing facility, and intermediate care facility services. Categories of service 2C, 5A, 5C, 5D, 6A, 7, 8, 9A, 10A, 11, 12, 13, 14, 15, 16, 19A, 19B, 19C, 19D, 23A, 23B, 24A, 24B, 26, 27, 28, 29A, 30, 31, 32, 33, 34, 34A, 35, 36, 37A, 38, 39, 40, 41, 42, 43, 44, 45, 46, 46B, 47, 48, 49 and 69. a These cover payments for Certified Community Behavior Health Clinics under the Section 1332 demonstration; physician and surgical services; outpatient hospital services; dental services; other practitioners services; clinic services; prescribed drugs; laboratory and radiological services; home health services; sterilizations; abortions; EPSDT screenings; rural health services; home and community-based services; personal care services; case management services; hospice benefits; emergency services for undocumented aliens; federally qualified health center services; non-emergency medical transportation (regular payments); physical therapy; occupational therapy; services for speech, hearing, and language; prosthetic devices, dentures, and eyeglasses; diagnostic screening and preventive services; preventive services vaccines and their administration; nurse mid-wife services; emergency hospital services; critical access hospital services (regular payments); nurse practitioner services; school-based services; rehabilitative services (not school based); private duty nursing services; freestanding birth center services; health home for enrollees with chronic conditions; tobacco cessation for pregnant women; health home for enrollees with substance abuse disorder; opioid use disorder medication-assisted treatment drugs and services; COVID-19 vaccines and vaccine administration, health homes for children with medically complex conditions, community-based mobile crisis intervention, and other care services. Categories of service 17D and 18D. These represent Medicaid payments on behalf of beneficiaries for deductibles and coinsurance on all services on which Medicare or a private insurance plan is the primary payer. | All LT claims with CLM_TYPE_CD=1, except those with a FED_SRVC_CTGRY_CD value of 13 or a TOS_CD value of 132, 133, or 134. This includes FFS claims in the LT file, except those that represent UPL supplemental payments and DSH payments, which correspond to the other expenditure type. All OT claims with CLM_TYPE_CD=1 and FED_SRVC_CTGRY_CD = 25-28 or 31-38, except those with a FED_SRVC_CTGRY_CD value of 13 or a TOS_CD value of 132, 133, or 134. This includes FFS claims in the OT file, except those that represent UPL supplemental payments and DSH payments, which correspond to the other expenditure type. All RX claims with CLM_TYPE_CD=1 and FED_SRVC_CTGRY_CD = 41, except those with a FED_SRVC_CTGRY_CD value of 13 or a TOS_CD value of 132, 133, or 134. This includes RX FFS claims for prescription drugs, except those that represent UPL supplemental payments and DSH payments, which correspond to the other expenditure type. IP claims with CLM_TYPE_CD=1 and a FED_SRVC_CTGRY_CD = 21-28, or 31-38 (or TOS_CD = 001 if missing), except those with a TOS_CD value of 132, 133, or 134, or those with a TOS_CD value for DSH payment c (123) as the only code on the claim. This includes IP FFS claims in the IP file, except those that represent UPL supplemental payments and DSH payments, which correspond to the other expenditure type. Records with CLM_TYPE_CD=5 that linked to a beneficiary, do not have a FED_SRVC_CTGRY_CD of 11 or 12 (or TOS_CD value of 119, 120, 121, 122, 138, 143, or 144 if missing), and do not have a TOS_CD value of 123, 132, 133, or 134. d This represents wraparound payments associated with a specific beneficiary above the negotiated per-service rate. |
| Categories of service 18A, 18A1, 18A2, 18A3, 18A4, 18A5, 18A6, and 22. a These represent payments to Medicaid managed care organizations and Programs of All-Inclusive Care for the Elderly (PACE) plans. Categories of service 18B1, 18B1a, 18B1b, 18B1c, 18B1d, 18B1e, 18B1f, 18B2, 18B2a, 18B2b, 18B2c, 18B2d, 18B2e, and 18B2f. a These represent payments to prepaid ambulatory health plans (PAHP) and prepaid inpatient health plans (PIHP). Categories of service 18C and 18E. These represent premium assistance for private plans. Category of service 25. This represents primary care case management (PCCM) payments. | All OT records with CLM_TYPE_CD=2, except those with a FED_SRVC_CTGRY_CD value of 13 or TOS_CD values of 132, 133, or 134. This includes all capitation payment records except those that represent DSH payments and UPL supplemental payments, which correspond to the other expenditure type. Any records with CLM_TYPE_CD=4 that also have FED_SRVC_CTGRY_CD values of 11 or 12 (or TOS_CD values of 119, 120, 121, 122, 138, 143, or 144 if missing). d These represent financial transaction records with a federally-assigned service category (or type of service if missing) indicating the lump-sum payment (not linkable to a specific beneficiary) was for a managed care capitation payment or other monthly beneficiary payment. Any records with CLM_TYPE_CD=5 that link to a Medicaid beneficiary and also have FED_SRVC_CTGRY_CD values of 11 or 12 (or TOS_CD values of 119, 120, 121, 122, 138, 143, or 144 if missing). d These represent payments that serve as adjustments to the negotiated monthly per-member rate. | |
| Other | Categories of service 1B and 2B. These represent inpatient hospital and mental health facility DSH payments. Categories of service 1C, 3B, 4C, 5B, 6B, 9B, 10B, 29B, 37B and 37C. a These represent supplemental payments for inpatient hospital, nursing facility services, intermediate care facility for individuals with intellectual disabilities, physician and surgical services, outpatient hospital services, other practitioner services, clinic services, non-emergency medical transportation, critical access hospital inpatient services, and critical access hospital outpatient services. Category of service 1D. This represents inpatient hospital graduate medical education (GME) payments. | All records with CLM_TYPE_CD=4 except those with a FED_SRVC_CTGRY_CD value of 11 or 12 (or TOS_CD value of 119, 120, 121, 122, 138, 143, or 144 if missing). d These represent all the lump-sum payments reported by the state that do not have a federally-assigned service category (or type of service if missing) indicating payment to a managed care plan. All records with CLM_TYPE_CD=5 and a missing or invalid MSIS ID, e except those with a FED_SRVC_CTGRY_CD value of 11 or 12 (or TOS_CD value of 119, 120, 121, 122, 138, 143, or 144 if missing). d These represent lump-sum payments (other than to managed care plans) that were classified as supplemental payments by the state in its TMSIS submission. Any records with CLM_TYPE_CD=1 that also had a FED_SRVC_CTGRY_CD value of 13, c and any records with CLM_TYPE_CD values of 1 or 2 that also had TOS_CD values of 132, 133, or 134. These represent DSH and UPL supplemental payments reported as FFS claims or capitation payment records. |
a CMS-64 category of service 45 (health home for enrollees with substance abuse disorder) is a valid value starting in 2019. CMS-64 categories of service 46 and 46B (medication-assisted treatment for opioid use disorder) are valid starting in 2020. CMS-64 categories of service 10A and 10B (clinic services), 18A6, 18B1f, and 18B2f (MCO services subject to electronic visit verification requirements), and 47 (COVID-19 vaccines and vaccine administration) are valid starting in 2021. CMS-64 categories of service 29A, 29B (regular and supplemental payments for non-emergency medical transportation), 37A, 37B and 37C (regular and supplemental payments for critical access hospital inpatient and outpatient services) are valid starting in 2022. CMS-64 category of service 48 (qualified community based mobile crisis intervention) is valid starting in 2022. CMS-64 category of service 49 (health homes for children with medically complex conditions) is valid starting in 2023.
b We first exclude TAF records that represent Medicare premiums, EHR payments to providers, and drug rebates.
c In some states, DSH payment amounts are included on FFS claim headers. For claims with a nonzero total Medicaid payment amount, we subtracted the DSH payment amount from the total Medicaid payment amount to obtain the total FFS expenditure and attributed the DSH payment amount to the other expenditure type.
d TOS codes 138, 143, and 144 are valid values starting in 2020.
e MSIS IDs that start with an “&” or are all 8-filled, 9-filled, or 0-filled are considered invalid.
f The federally assigned service category (FASC) code is available in TAF claims files produced in 2022 or later. More information can be found in the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
Data quality assessment criteria
We categorized each state’s total Medicaid expenditures and Medicaid beneficiary expenditures in TAF as having high, moderate, low, or very low alignment with the CMS-64 based on the percent difference between the two data sources (Table 3). [18]
Table 3. Criteria for DQ assessment of TAF expenditures
| Percent difference between TAF and CMS-64 expenditures | Level of alignment | DQ assessment |
|---|---|---|
| x < 5 percent | High | Low concern |
| 5 percent ≤ x < 20 percent | Moderate | Medium concern |
| 20 percent ≤ x < 50 percent | Low | High concern |
| x ≥ 50 percent | Very low | Unusable |
States with a low absolute percentage difference between the expenditures captured in TAF and the CMS-64 are considered to present a low level of concern about the completeness and quality of their TAF expenditure data. However, TAF users should be aware that in states with large Medicaid programs, even a small percentage difference may represent a substantial difference in the dollar amount between the data sources. In addition, compared to its alignment for total expenditures, a state may have better alignment for certain types of spending (such as FFS spending or monthly beneficiary payments). When this occurs, users may be able to reliably examine some types of spending in the state even when overall expenditures are incomplete. [19]
There are several reasons other than the quality and completeness of TAF expenditure data that could result in a state’s total expenditures or Medicaid beneficiary expenditures in the TAF differing from the CMS-64: (1) structural differences in the data sets, such as organizing the data by service date versus claim payment date; (2) data quality problems in the TAF that affect expenditures, such as a low volume of claims in the TAF because of missing date values on claims or issues with assigning a final action status to claims; and (3) variation in how states report certain types of expenditures on the CMS-64 that are difficult to reproduce in the TAF data. [20] For instance, a state may make lump-sum reconciliation payments to hospitals as part of their cost-based FFS payments for inpatient services. On the CMS-64, the state may report these costs as part of the inpatient hospital base payment category of service, whereas in TAF these lump sum payments are reported as service tracking claims that cannot easily be differentiated from other lump sum payments such as DSH and UPL payments. As a result of these kinds of differences, some expenditures may fall into CMS-64 categories of service that are classified into the FFS expenditure type but appear in TAF as service records that are classified as the “other” expenditure type. When this occurs, we would expect to see a difference between the TAF-based Medicaid beneficiary expenditures and the CMS-64 benchmark even if TAF expenditure data are complete and accurate.
Methods previously used to assess data quality
Table 4 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 4. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1. In addition, the TAF RIF for all years and releases redacts supplemental payments in which the Medicaid identification number begins with a “&” because they cannot be attributed to a specific beneficiary. These are likely to represent lump sum payments that were misclassified as supplemental payments rather than service tracking claims. All other supplemental payment records are included in the TAF RIF. This analysis includes expenditures on supplemental payment records that are redacted from the RIF.
FFS claims representing Medicaid or M-CHIP benefits have a claim type code value of 1. Medicaid or M-CHIP monthly beneficiary payments have a claim type code value of 2, service tracking claims have a claim type code value of 4, and supplemental payment records have a claim type code value of 5.
Payment data on managed care encounter records are redacted from the TAF RIF.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the beneficiary had CHIP code value of 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-76, we considered the beneficiary to have non-CHIP Medicaid enrollment during the month. For FFS expenditures, we required claims to link to a beneficiary with non-CHIP Medicaid enrollment during the month of service. For monthly beneficiary payments, we required claims to link to a beneficiary with non-CHIP Medicaid enrollment in any month of the year. For other expenditures, records do not link to specific beneficiaries.
We did not require service tracking claims to match to a Medicaid eligibility record because these types of records are lump-sum payments that are not connected to any one beneficiary and should not have a valid beneficiary identifier.
The FFS expenditure amount is comprised of services provided in the month of eligibility; services for beneficiaries who switch from CHIP to non-CHIP Medicaid will only be counted in the months of non-CHIP Medicaid eligibility. The monthly beneficiary payment amount is comprised of payments in the year of service; all payments for beneficiaries who switch from CHIP to non-CHIP Medicaid within a year will be counted in the total.
EHR payments were identified as records where any line had a type of service code (TOS_CD) equal to 135. Drug rebates were identified as records were any line had a type of service code of 131 or a service tracking type code (SRVC_TRKNG_TYPE_CD) of 01. Medicare premiums were identified as records where any line had a type of service code of 139, 140, 141, or 142 (valid values starting in 2020).
We considered records with claim type 4 and a federally assigned service category code (11 or 12) or TOS code (119, 120,121, 122, 138, 143, or 144) that indicates a capitation payment to be capitation payments reported on service tracking claims.
The DSH payment field only appears on IP claims. For LT, OT, and RX claims, we only considered the service tracking payment amount or total Medicaid paid amount.
For service tracking claims, states sometimes entered the same payment amount in all three payment fields (or two of the three payment fields). To avoid double or triple counting, we used only one payment field per header claim.
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
CMS-64 categories of service 46A1-46A6 are valid starting in federal fiscal year 2021. CMS-64 category of service 7A7 is valid starting in federal fiscal year 2022.
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: (TAF - benchmark) / benchmark.
Users can find more information about each state’s alignment of FFS and monthly beneficiary payments in the DQ Atlas single topic displays for Total FFS Expenditures and Total Monthly Beneficiary Payments .
For more information about the differences between TAF and CMS-64 expenditure data, see the methodology brief, “Medicaid Expenditure Data: TAF and the CMS-64”, available on the DQ Atlas Resources page.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
Service tracking claims are excluded from 2014-16 TAF RIF Releases 1 and 2 and 2017-18 TAF RIF Release 1. In addition, the TAF RIF for all years and releases redacts supplemental payments in which the Medicaid identification number begins with a \u201c&\u201d because they cannot be attributed to a specific beneficiary. These are likely to represent lump sum payments that were misclassified as supplemental payments rather than service tracking claims. All other supplemental payment records are included in the TAF RIF. This analysis includes expenditures on supplemental payment records that are redacted from the RIF.
FFS claims representing Medicaid or M-CHIP benefits have a claim type code value of 1. Medicaid or M-CHIP monthly beneficiary payments have a claim type code value of 2, service tracking claims have a claim type code value of 4, and supplemental payment records have a claim type code value of 5.
Payment data on managed care encounter records are redacted from the TAF RIF.
Non-CHIP Medicaid enrollment was determined by examining the CHIP code and eligibility group code on records in the DE file. If the beneficiary had CHIP code value of 1 (Medicaid), or if the CHIP code was missing but the eligibility group code was in the ranges 01-60 or 69-76, we considered the beneficiary to have non-CHIP Medicaid enrollment during the month. For FFS expenditures, we required claims to link to a beneficiary with non-CHIP Medicaid enrollment during the month of service. For monthly beneficiary payments, we required claims to link to a beneficiary with non-CHIP Medicaid enrollment in any month of the year. For other expenditures, records do not link to specific beneficiaries.
We did not require service tracking claims to match to a Medicaid eligibility record because these types of records are lump-sum payments that are not connected to any one beneficiary and should not have a valid beneficiary identifier.
The FFS expenditure amount is comprised of services provided in the month of eligibility; services for beneficiaries who switch from CHIP to non-CHIP Medicaid will only be counted in the months of non-CHIP Medicaid eligibility. The monthly beneficiary payment amount is comprised of payments in the year of service; all payments for beneficiaries who switch from CHIP to non-CHIP Medicaid within a year will be counted in the total.
EHR payments were identified as records where any line had a type of service code (TOS_CD) equal to 135. Drug rebates were identified as records were any line had a type of service code of 131 or a service tracking type code (SRVC_TRKNG_TYPE_CD) of 01. Medicare premiums were identified as records where any line had a type of service code of 139, 140, 141, or 142 (valid values starting in 2020).
We considered records with claim type 4 and a federally assigned service category code (11 or 12) or TOS code (119, 120,121, 122, 138, 143, or 144) that indicates a capitation payment to be capitation payments reported on service tracking claims.
The DSH payment field only appears on IP claims. For LT, OT, and RX claims, we only considered the service tracking payment amount or total Medicaid paid amount.
For service tracking claims, states sometimes entered the same payment amount in all three payment fields (or two of the three payment fields). To avoid double or triple counting, we used only one payment field per header claim.
Our tabulations of the CMS-64 expenditures (which are based on the calendar year) do not exactly match the CMS-64 expenditure reports (which are based on the federal fiscal year). CMS-64 expenditure reports based on the federal fiscal year are publicly available on Medicaid.gov: https://www.medicaid.gov/medicaid/finance/state-expenditure-reporting/expenditure-reports/index.html . To calculate CMS-64 expenditures for a given calendar year, we extracted data from the quarterly CMS-64 Financial Management Reports within the MBES for the last three quarters of the associated federal fiscal year and the first quarter of the next federal fiscal year. For example, calendar year 2016 corresponds to the last three quarters of fiscal year 2016 and the first quarter of fiscal year 2017.
CMS-64 categories of service 46A1-46A6 are valid starting in federal fiscal year 2021. CMS-64 category of service 7A7 is valid starting in federal fiscal year 2022.
All comparisons use the absolute value of the percent difference between TAF and CMS-64 expenditures for each state. For example, a state with a 2 percent difference from the benchmark would be categorized the same way as a state with a -2 percent difference. Because the CMS-64 benchmark data can be viewed as a baseline and the TAF-based calculations as the comparison, we calculated the percent difference as a percent error or change: (TAF - benchmark) / benchmark.
Users can find more information about each state\u2019s alignment of FFS and monthly beneficiary payments in the DQ Atlas single topic displays for Total FFS Expenditures and Total Monthly Beneficiary Payments .
For more information about the differences between TAF and CMS-64 expenditure data, see the methodology brief, \u201cMedicaid Expenditure Data: TAF and the CMS-64\u201d, available on the DQ Atlas Resources page.
The TAF should include all expenditures states make to provide or cover care for Medicaid beneficiaries, including fee-for-service claims, monthly beneficiary payments, supplemental payments, and other lump sum payments to providers. This analysis examines how well expenditures that can be linked to individual Medicaid beneficiaries captured in TAF align with an external benchmark, the CMS-64.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""6041"", ""relatedTopics"": [{""measureId"": 77, ""measureName"": ""Total Medicaid Expenditures"", ""groupId"": 9, ""groupName"": ""Expenditure Benchmarking"", ""order"": 0}]}" 88,"{""measureId"": 88, ""measureName"": ""Linking Beneficiaries to Managed Care Plans"", ""groupId"": 11, ""groupName"": ""Linking Across Files"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Link-Bene-MC-Plans.pdf"", ""background"": {""content"": ""Managed care plays a significant role in the delivery of care to Medicaid beneficiaries nationwide. Nearly all states and territories contract with managed care organizations (MCOs) to provide care to at least some of their beneficiaries. All states and territories contracting with one or more Medicaid managed care plans must report into T-MSIS (1) information about the managed care plans in which beneficiaries are enrolled each month, (2) service utilization (encounter) records and capitation payments, and (3) information on the characteristics of each managed care plan operating in the state. T-MSIS Analytic Files (TAF) users can find beneficiary-level plan enrollment information in the Demographics and Eligibility (DE) file, encounter records and capitation payments in the claims (IP, LT, OT, and RX) files, and managed care plan-level information in the Annual Managed Care Plan (APL file). Although the DE file includes limited information about each managed care plan in which a beneficiary is enrolled (including plan ID and plan type), the TAF APL file includes additional information about the characteristics, locations, enrolled populations, and service areas for all Medicaid and Children’s Health Insurance Program (CHIP) health plans and managed care entities. [1]
TAF users may wish to link beneficiary enrollment records in the DE file with additional information from the APL file about the plans that manage their care. To do so, the plan IDs in the beneficiary-level DE file must link to the plan IDs in the APL file. Alternatively, some users may wish to start with a specific plan that appears in the APL file and determine which beneficiaries are enrolled in that plan by linking to the DE file. This data quality assessment examines the extent to which managed care plan IDs in the DE file were present in the APL; and conversely, which managed care plan IDs in the APL file were present in the DE file. Additionally, the analysis examines the extent to which managed care plan IDs in the DE file were also present in the APL file for select plan type categories: comprehensive managed care plans (CMCs), behavioral health organizations (BHOs), managed long-term services and supports (MLTSS), Programs of All-Inclusive Care for the Elderly (PACE), and integrated care for dually eligible individuals.
The TAF APL consists of five files. The TAF APL base file includes basic plan-level information, whereas four additional supplemental files provide more detailed information on: plan geographic locations, service areas, populations enrolled, and operating authority. Each record in the APL TAF base file represents a managed care plan or entity; in some cases, states may report multiple records per plan or entity—for example, to distinguish between the subpopulations enrolled. Each record in the APL base file may link to more than one record in each supplemental file.
The TAF APL consists of five files. The TAF APL base file includes basic plan-level information, whereas four additional supplemental files provide more detailed information on: plan geographic locations, service areas, populations enrolled, and operating authority. Each record in the APL TAF base file represents a managed care plan or entity; in some cases, states may report multiple records per plan or entity\u2014for example, to distinguish between the subpopulations enrolled. Each record in the APL base file may link to more than one record in each supplemental file.
This analysis uses beneficiary enrollment records from the DE file and managed care plan records from the APL file. [2] From the DE file, we selected all records with one or more non-missing managed care plan ID variables. Beneficiaries enrolled in multiple plans over the year would have multiple plan IDs associated with their DE record. From the APL base file, we included all records, each of which represents a unique plan ID reported by the state during the calendar year. We excluded states without a managed care organization identified in the Medicaid Managed Care Enrollment and Program Characteristics (MMCEPC) report for the calendar year in this analysis.
We calculated the number of plan IDs for each state present in the DE file that were also present in the APL file for that state. [3] We reported this number as a percentage of plan IDs, using the total number of plan IDs present in the DE as the denominator. We also calculated the number of plan IDs in the APL file that were also present in the DE file. We reported this number as a percentage of plan IDs, using the total number of plan IDs in the APL as the denominator. We included these measures in the data table for this topic, but they were not used as part of the overall data quality assessment.
To assign states to a data quality concern level, we calculated the percentage of beneficiary-plan combinations in the DE file with a plan ID that links to an APL record. [4] This method gives a higher weight to larger plans with many beneficiaries enrolled and a lower weight to plans that have only a small number of beneficiaries enrolled. We grouped states into levels of concern about the usability of their APL data, based on the percentage of beneficiary-plan combinations in the DE file that link to an APL record (Table 1).
Table 1. Criteria for DQ assessment of linking the DE to the APL
| Percentage of beneficiary-plan combinations in the DE file that link to a plan ID in the APL file | DQ assessment |
|---|---|
| x > 90 percent | Low concern |
| 90 percent ≥ x > 80 percent | Medium concern |
| 80 percent ≥ x > 50 percent | High concern |
| x ≤ 50 percent | Unusable |
We also grouped similar plan type codes together and calculated the percentage of beneficiary-plan combinations in the DE file with a plan ID that links to an APL record for a subset of plan type categories: CMCs, BHOs, MLTSS, PACE, and integrated care for dually eligible individuals (Table 2). This information is presented in the table but was not used in the data quality assessment. We present these contextual data only for those states that had one of the plan types within each respective category during the year in the MMCEPC [5] or the Integrated Care Resource center (for plans providing integrated care for dually eligible individuals during the year).
Table 2. Plan type categories
| Plan type category | TAF plan type codes (MC_PLAN_TYPE_CD) |
|---|---|
| Comprehensive managed care (CMC) | 01: Comprehensive MCO 04: Health Insuring Organization (HIO) |
| Behavioral health organizations (BHO) | 08: MH PIHP 09: MH PAHP 10: SUD PIHP 11: SUD PAHP 12: MH and SUD PIHP 13: MH and SUD PAHP |
| Managed long-term services and supports (MLTSS) | 07: LTSS PIHP 19: LTSS and MH PHIP |
| Programs of All-Inclusive Care for the Elderly (PACE) | 17: PACE |
| Integrated care for dually eligible individuals | 80: Integrated care for dually eligible beneficiaries |
MH = mental health; SUD = substance use disorder; LTSS = long-term services and supports; PAHP = prepaid ambulatory health plan; PIHP = prepaid inpatient health plan.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIFs). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
States should include only managed care plans that were active in the relevant calendar year in their APL reporting. An active managed care plan record in T-MSIS has (a) an effective start date and end date that overlap with at least part of the calendar year, and (b) an indicator identifying the record as active if multiple records represent the same time period. However, states sometimes submit active managed care plan records that in fact were inactive because the plan’s contract was not in effect during the calendar year. If a state reports these inactive plans as being active in the calendar year, they are included in the APL file and in this analysis.
Beneficiaries may be enrolled in multiple plans within a given year. We counted beneficiaries once for every plan ID in which they were enrolled. As a result, a given beneficiary may be counted multiple times.
We considered a state to have each type of managed care plan within the year if at least one person was enrolled in the plan type according to the Managed Care Enrollment by Program and Population data on Medicaid.gov: https://www.medicaid.gov/medicaid/managed-care/enrollment-report/index.html .
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIFs). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
States should include only managed care plans that were active in the relevant calendar year in their APL reporting. An active managed care plan record in T-MSIS has (a) an effective start date and end date that overlap with at least part of the calendar year, and (b) an indicator identifying the record as active if multiple records represent the same time period. However, states sometimes submit active managed care plan records that in fact were inactive because the plan\u2019s contract was not in effect during the calendar year. If a state reports these inactive plans as being active in the calendar year, they are included in the APL file and in this analysis.
Beneficiaries may be enrolled in multiple plans within a given year. We counted beneficiaries once for every plan ID in which they were enrolled. As a result, a given beneficiary may be counted multiple times.
We considered a state to have each type of managed care plan within the year if at least one person was enrolled in the plan type according to the Managed Care Enrollment by Program and Population data on Medicaid.gov: https://www.medicaid.gov/medicaid/managed-care/enrollment-report/index.html .
The TAF Annual Managed Care Plan (APL) file contains plan-level information, including characteristics, locations, enrolled populations, and service areas for all Medicaid and Children's Health Insurance Program (CHIP) health plans and managed care entities. TAF users may wish to link beneficiary-level records in the TAF Demographics and Eligibility (DE) file with additional information found in the APL file about the plans that manage their care. This analysis examines the extent to which managed care plan IDs in the DE file were also present in the APL file and conversely, the extent to which managed care plan IDs in the APL file were also present in the DE file. Additionally, the analysis examines the extent to which managed care plan IDs in the DE file were also present in the APL file for select plan type categories: comprehensive managed care plans (CMCs), behavioral health organizations (BHOs), managed long-term services and supports (MLTSS), Programs of All-Inclusive Care for the Elderly (PACE), and integrated care for dually eligible individuals.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""8041"", ""relatedTopics"": []}" 89,"{""measureId"": 89, ""measureName"": ""National Drug Code - RX"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-National-Drug-Code-RX.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) Pharmacy (RX) file includes several data elements that provide information about the drug, device, or medical supply covered by a claim, as well as the duration and quantity of the treatment. The National Drug Code (NDC) is a universal identifier indicating the specific drug, device, or medical supply covered by a claim. An NDC is required on every claim in the RX file and on a small number of claims for prescription drugs, medical devices, or medical supplies in the Inpatient (IP), Other Services (OT), and Long-Term Care (LT) files. [1] The Food and Drug Administration (FDA) maintains NDCs reported by manufacturers as part of drug listing requirements under Section 510 of the Federal Food, Drug, and Cosmetics (FD&C) Act, 21 USC 360. [2] The complete universe of NDCs includes the following: (1) registered NDCs —codes for medications that have been officially registered in the FDA’s NDC file; (2) semi-official NDCs—codes for medications that have been properly assigned according to FDA rules but not yet registered in the FDA’s database; and (3) device/supply codes, which are excluded from the FDA database. [3] Data elements that appear in both the TAF RX claims and the FDA’s NDC Directory include the NDC code and the NDC unit of measure code.
The days’ supply data element captures the total number of days dispensed for the drug, device, or medical supply covered by a claim. Days’ supply can be used, along with the fill date, to approximate the time frame during which a beneficiary received drug treatment and to indirectly measure medication adherence. Days’ supply is a required data element for state reporting.
NDC quantity denotes the numerical quantity of the drug, device, or medical supply covered by a claim, whereas the NDC unit of measure code indicates the standard by which the value reported on the NDC quantity data element is measured. [4] For example, for a claim of 1.5 milliliters dispensed, a value of “1.5” would be reported in the NDC quantity data element, and a value of “ML” would be reported in the NDC unit of measure data element. NDC quantity and NDC unit of measure are often combined with days’ supply to calculate the daily supply of a prescription. Calculating the beneficiary-level daily supply of a prescription (or combinations of prescriptions) across RX claims can be used to assess medication utilization and identify prescription drug misuse and abuse. NDC quantity and NDC unit of measure are required data elements for state reporting on every claim in the TAF RX file.
The NDC unit of measure — along with additional information about the drug’s strength and dosage form — can also be identified using the NDC code itself. [5] Therefore, in cases where the NDC unit of measure is missing from TAF, this information may be obtained by cross-referencing the NDC code with the FDA’s NDC Directory. [6]
This data quality assessment examines the proportion of RX claim lines with a missing or invalid NDC. It also examines the proportion of RX claim lines with a missing, invalid, or unexpected value for days’ supply, or a missing or invalid value for NDC quantity. For context, the analysis also presents the percentage of RX claim lines with a missing or invalid value for the NDC unit of measure; because this information may be identified using the NDC in cases of missing or invalid data, this data element is excluded from the data quality assessments.
Medicaid and CHIP Business Information Solutions (MACBIS). “TAF Technical Guidance: Claims Files.” August 2020. Available at https://requests.resdac.org/sites/resdac.umn.edu/files/TAF_TechGuide_Claims_Files.pdf . Accessed January 18, 2021.
U.S. Food and Drug Administration. “National Drug Code Directory.” Available at https://www.fda.gov/drugs/drug-approvals-and-databases/national-drug-code-directory . Accessed March 5, 2021.
Simonaitis, Linas, and Clement J. McDonald. \""Using National Drug Codes and Drug Knowledge Bases to Organize Prescription Records from Multiple Sources.\"" American Journal of Health-System Pharmacy, vol. 66, no. 19, October 2009, pp. 1743−1753.
The quantity and unit of measure are required for every pharmacy claim. In a TAF RX record, such information is available in the NDC quantity and NDC unit of measure data elements.
For example, the NDC code 0378-6140-44 represents the prescription drug “Avita.” According to the NDC directory, it has a strength and unit of 0.25mg/g and its dosage form is a gel. The corresponding unit of measure for this drug is grams.
The NDC Directory is hosted by the Food and Drug Administration (FDA) and updated daily. The FDA does not publish a historical NDC directory.
Medicaid and CHIP Business Information Solutions (MACBIS). \u201cTAF Technical Guidance: Claims Files.\u201d August 2020. Available at https://requests.resdac.org/sites/resdac.umn.edu/files/TAF_TechGuide_Claims_Files.pdf . Accessed January 18, 2021.
U.S. Food and Drug Administration. \u201cNational Drug Code Directory.\u201d Available at https://www.fda.gov/drugs/drug-approvals-and-databases/national-drug-code-directory . Accessed March 5, 2021.
Simonaitis, Linas, and Clement J. McDonald. \""Using National Drug Codes and Drug Knowledge Bases to Organize Prescription Records from Multiple Sources.\"" American Journal of Health-System Pharmacy, vol. 66, no. 19, October 2009, pp. 1743\u22121753.
The quantity and unit of measure are required for every pharmacy claim. In a TAF RX record, such information is available in the NDC quantity and NDC unit of measure data elements.
For example, the NDC code 0378-6140-44 represents the prescription drug \u201cAvita.\u201d According to the NDC directory, it has a strength and unit of 0.25mg/g and its dosage form is a gel. The corresponding unit of measure for this drug is grams.
The NDC Directory is hosted by the Food and Drug Administration (FDA) and updated daily. The FDA does not publish a historical NDC directory.
We examined the NDC code, days’ supply, NDC quantity, and NDC unit of measure data elements in the TAF RX claim lines, which includes claims for both new prescriptions and refills. [7] We included fee-for-service (FFS) claims and managed care encounter records for both Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries in the analysis. [8] We excluded records with other claim type codes that represent capitation payments, supplemental claims, service tracking claims, or “other” non-program claims. [9] For Illinois, we restricted our analysis to the original version of the claim and excluded all subsequent adjustment records in the state’s TAF data. [10]
We calculated the percentage of RX claim lines with a missing or invalid NDC code (NDC_CD). [11] We considered an NDC to be missing if a TAF null condition was met (Table 1). [12] We classified an NDC as invalid if it was not present in the FDA Directory released at the end of the analysis year. [13]
We also calculated the percentage of RX claim lines with a missing, invalid, or unexpected value for days’ supply. We considered days’ supply to be missing if a TAF null condition was met (Table 1), and invalid if it contained non-numeric values or was less than or equal to 0. Although states may report valid values between -365 and 365 in the days’ supply data element, in practice, a negative or zero value for days’ supply represents data that are not usable for analytic purposes. We considered days’ supply to be an unexpected value if it was greater than 180. Records with a days’ supply above 180 present a possible data quality concern because they are outliers for any class of medication. [14] In addition, we calculated the percentage of claim lines that had a missing or invalid value for NDC quantity. We considered an NDC quantity to be missing if a TAF null condition was met, and invalid if it contained non-numeric values. Finally, we calculated the percentage of claim lines that had a missing or invalid value for NDC unit of measure, but this information is excluded from the data quality assessments. We considered an NDC unit of measure code to be missing if a TAF null condition was met and invalid if it contained any value not on the valid value list (Table 1).
Table 1. Missing, invalid, and unexpected values for NDC code, days’ supply, NDC quantity, and NDC unit of measure
| Data element | Missing values (TAF null conditions) | Invalid values | Unexpected values |
|---|---|---|---|
| NDC code (NDC_CD) | Blank, 8-filled, or 9-filled | The NDC was not present in the FDA’s NDC Directory at the end of the year of analysis | N/A |
| Days’ supply (SUPLY_DAYS_CNT) | Blank, 8-filled, or 9-filled | ≤ 0 or non-numeric | > 180 |
| NDC quantity (NDC_QTY) | Blank, 8-filled, or 9-filled | Non-numeric | N/A |
| NDC unit of measure (NDC_UOM_CD) | Blank | Any value not on the valid value list: EA: Each F2: International Unit GM, GR: Grams ME: Milligrams ML: Milliliters UN: Unit | N/A |
For the data quality assessment of the NDC code, we grouped states into categories of low, medium, and high concern about the usability of their data, based on the criteria shown in Table 2.
Table 2. Criteria for DQ assessment of NDC
| Percentage of RX claim lines with a missing or invalid NDC | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
For the data quality assessment of days’ supply and NDC quantity, we grouped states into categories of low, medium, and high concern about the usability of their data, based on the criteria shown in Table 3.
Table 3. Criteria for DQ assessment of days’ supply and NDC quantity
| Percentage of RX claim lines with a missing, invalid, or unexpected value for days’ supply, or a missing or invalid value for NDC quantity | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
This analysis used the TAF data released as TAF Research Identifiable Files (RIFs). During the transformation into RIF some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page ; a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We identified FFS claims and managed care encounters by selecting records with claim type code values of 1, A, 3, and C.
More information on other non-program claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits, rather than voiding the original claim and submitting a replacement record containing the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This practice means that in some cases, the TAF will include multiple versions of a single claim for Illinois. The original record is likely to contain the most complete nonpayment information (such as diagnosis code, procedure code, revenue center code, and so forth), and thus is likely to be the most usable record in the claim family for analytic purposes. Therefore, for Illinois, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide “How to Use Illinois Claims Data,” available on the DQ Atlas Resources page under “Technical guidance for using TAF RIF data.”
The NDC code data element is present only in the RX claim lines and not in the RX claim headers. The TAF RX claim file is structured with just one claim line per claim header.
For the NDC code, days’ supply, NDC quantity, and NDC unit of measure data elements, when a value is reported in the source data that meets one of the TAF null conditions, the value is re-coded to missing (null) in the TAF RX data.
The NDC Directory is hosted by the Food and Drug Administration (FDA) and updated daily. The FDA does not publish a historical NDC directory. As a result, the 2019 analysis references the NDC Directory published on December 8, 2019, as archived by the Way Back Machine tool, available at: https://web.archive.org/web/20191208061658/https://www.accessdata.fda.gov/cder/ndcxls.zip. Accessed January 18, 2021.
Analysis of the 2019 TAF RX claim file indicates that in almost all (52 of 53) states and territories, less than 0.05 percent of claims had a days’ supply above 180, indicating that prescriptions exceeding 180 days’ supply are either outliers or incorrectly reported.
This analysis used the TAF data released as TAF Research Identifiable Files (RIFs). During the transformation into RIF some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page ; a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We identified FFS claims and managed care encounters by selecting records with claim type code values of 1, A, 3, and C.
More information on other non-program claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits, rather than voiding the original claim and submitting a replacement record containing the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This practice means that in some cases, the TAF will include multiple versions of a single claim for Illinois. The original record is likely to contain the most complete nonpayment information (such as diagnosis code, procedure code, revenue center code, and so forth), and thus is likely to be the most usable record in the claim family for analytic purposes. Therefore, for Illinois, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide \u201cHow to Use Illinois Claims Data,\u201d available on the DQ Atlas Resources page under \u201cTechnical guidance for using TAF RIF data.\u201d
The NDC code data element is present only in the RX claim lines and not in the RX claim headers. The TAF RX claim file is structured with just one claim line per claim header.
For the NDC code, days\u2019 supply, NDC quantity, and NDC unit of measure data elements, when a value is reported in the source data that meets one of the TAF null conditions, the value is re-coded to missing (null) in the TAF RX data.
The NDC Directory is hosted by the Food and Drug Administration (FDA) and updated daily. The FDA does not publish a historical NDC directory. As a result, the 2019 analysis references the NDC Directory published on December 8, 2019, as archived by the Way Back Machine tool, available at: https://web.archive.org/web/20191208061658/https://www.accessdata.fda.gov/cder/ndcxls.zip. Accessed January 18, 2021.
Analysis of the 2019 TAF RX claim file indicates that in almost all (52 of 53) states and territories, less than 0.05 percent of claims had a days\u2019 supply above 180, indicating that prescriptions exceeding 180 days\u2019 supply are either outliers or incorrectly reported.
A National Drug Code (NDC) is a universal identifier indicating the drug, device, or medical supply covered by a pharmacy (RX) claim. TAF users can use the NDC to identify the specific treatment prescribed and obtained by a beneficiary. This analysis examines how often the NDC is missing or invalid in the TAF RX claims records.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5231"", ""relatedTopics"": [{""measureId"": 90, ""measureName"": ""Days' Supply, Quantity, and Units - RX"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 1}]}" 90,"{""measureId"": 90, ""measureName"": ""Days' Supply, Quantity, and Units - RX"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Day-Supply-Qty-Unit-RX.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) Pharmacy (RX) file includes several data elements that provide information about the drug, device, or medical supply covered by a claim, as well as the duration and quantity of the treatment. The National Drug Code (NDC) is a universal identifier indicating the specific drug, device, or medical supply covered by a claim. An NDC is required on every claim in the RX file and on a small number of claims for prescription drugs, medical devices, or medical supplies in the Inpatient (IP), Other Services (OT), and Long-Term Care (LT) files. [1] The Food and Drug Administration (FDA) maintains NDCs reported by manufacturers as part of drug listing requirements under Section 510 of the Federal Food, Drug, and Cosmetics (FD&C) Act, 21 USC 360. [2] The complete universe of NDCs includes the following: (1) registered NDCs —codes for medications that have been officially registered in the FDA’s NDC file; (2) semi-official NDCs—codes for medications that have been properly assigned according to FDA rules but not yet registered in the FDA’s database; and (3) device/supply codes, which are excluded from the FDA database. [3] Data elements that appear in both the TAF RX claims and the FDA’s NDC Directory include the NDC code and the NDC unit of measure code.
The days’ supply data element captures the total number of days dispensed for the drug, device, or medical supply covered by a claim. Days’ supply can be used, along with the fill date, to approximate the time frame during which a beneficiary received drug treatment and to indirectly measure medication adherence. Days’ supply is a required data element for state reporting.
NDC quantity denotes the numerical quantity of the drug, device, or medical supply covered by a claim, whereas the NDC unit of measure code indicates the standard by which the value reported on the NDC quantity data element is measured. [4] For example, for a claim of 1.5 milliliters dispensed, a value of “1.5” would be reported in the NDC quantity data element, and a value of “ML” would be reported in the NDC unit of measure data element. NDC quantity and NDC unit of measure are often combined with days’ supply to calculate the daily supply of a prescription. Calculating the beneficiary-level daily supply of a prescription (or combinations of prescriptions) across RX claims can be used to assess medication utilization and identify prescription drug misuse and abuse. NDC quantity and NDC unit of measure are required data elements for state reporting on every claim in the TAF RX file.
The NDC unit of measure — along with additional information about the drug’s strength and dosage form — can also be identified using the NDC code itself. [5] Therefore, in cases where the NDC unit of measure is missing from TAF, this information may be obtained by cross-referencing the NDC code with the FDA’s NDC Directory. [6]
This data quality assessment examines the proportion of RX claim lines with a missing or invalid NDC. It also examines the proportion of RX claim lines with a missing, invalid, or unexpected value for days’ supply, or a missing or invalid value for NDC quantity. For context, the analysis also presents the percentage of RX claim lines with a missing or invalid value for the NDC unit of measure; because this information may be identified using the NDC in cases of missing or invalid data, this data element is excluded from the data quality assessments.
Medicaid and CHIP Business Information Solutions (MACBIS). “TAF Technical Guidance: Claims Files.” August 2020. Available at https://requests.resdac.org/sites/resdac.umn.edu/files/TAF_TechGuide_Claims_Files.pdf . Accessed January 18, 2021.
U.S. Food and Drug Administration. “National Drug Code Directory.” Available at https://www.fda.gov/drugs/drug-approvals-and-databases/national-drug-code-directory . Accessed March 5, 2021.
Simonaitis, Linas, and Clement J. McDonald. \""Using National Drug Codes and Drug Knowledge Bases to Organize Prescription Records from Multiple Sources.\"" American Journal of Health-System Pharmacy, vol. 66, no. 19, October 2009, pp. 1743−1753.
The quantity and unit of measure are required for every pharmacy claim. In a TAF RX record, such information is available in the NDC quantity and NDC unit of measure data elements.
For example, the NDC code 0378-6140-44 represents the prescription drug “Avita.” According to the NDC directory, it has a strength and unit of 0.25mg/g and its dosage form is a gel. The corresponding unit of measure for this drug is grams.
The NDC Directory is hosted by the Food and Drug Administration (FDA) and updated daily. The FDA does not publish a historical NDC directory.
Medicaid and CHIP Business Information Solutions (MACBIS). \u201cTAF Technical Guidance: Claims Files.\u201d August 2020. Available at https://requests.resdac.org/sites/resdac.umn.edu/files/TAF_TechGuide_Claims_Files.pdf . Accessed January 18, 2021.
U.S. Food and Drug Administration. \u201cNational Drug Code Directory.\u201d Available at https://www.fda.gov/drugs/drug-approvals-and-databases/national-drug-code-directory . Accessed March 5, 2021.
Simonaitis, Linas, and Clement J. McDonald. \""Using National Drug Codes and Drug Knowledge Bases to Organize Prescription Records from Multiple Sources.\"" American Journal of Health-System Pharmacy, vol. 66, no. 19, October 2009, pp. 1743\u22121753.
The quantity and unit of measure are required for every pharmacy claim. In a TAF RX record, such information is available in the NDC quantity and NDC unit of measure data elements.
For example, the NDC code 0378-6140-44 represents the prescription drug \u201cAvita.\u201d According to the NDC directory, it has a strength and unit of 0.25mg/g and its dosage form is a gel. The corresponding unit of measure for this drug is grams.
The NDC Directory is hosted by the Food and Drug Administration (FDA) and updated daily. The FDA does not publish a historical NDC directory.
We examined the NDC code, days’ supply, NDC quantity, and NDC unit of measure data elements in the TAF RX claim lines, which includes claims for both new prescriptions and refills. [7] We included fee-for-service (FFS) claims and managed care encounter records for both Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries in the analysis. [8] We excluded records with other claim type codes that represent capitation payments, supplemental claims, service tracking claims, or “other” non-program claims. [9] For Illinois, we restricted our analysis to the original version of the claim and excluded all subsequent adjustment records in the state’s TAF data. [10]
We calculated the percentage of RX claim lines with a missing or invalid NDC code (NDC_CD). [11] We considered an NDC to be missing if a TAF null condition was met (Table 1). [12] We classified an NDC as invalid if it was not present in the FDA Directory released at the end of the analysis year. [13]
We also calculated the percentage of RX claim lines with a missing, invalid, or unexpected value for days’ supply. We considered days’ supply to be missing if a TAF null condition was met (Table 1), and invalid if it contained non-numeric values or was less than or equal to 0. Although states may report valid values between -365 and 365 in the days’ supply data element, in practice, a negative or zero value for days’ supply represents data that are not usable for analytic purposes. We considered days’ supply to be an unexpected value if it was greater than 180. Records with a days’ supply above 180 present a possible data quality concern because they are outliers for any class of medication. [14] In addition, we calculated the percentage of claim lines that had a missing or invalid value for NDC quantity. We considered an NDC quantity to be missing if a TAF null condition was met, and invalid if it contained non-numeric values. Finally, we calculated the percentage of claim lines that had a missing or invalid value for NDC unit of measure, but this information is excluded from the data quality assessments. We considered an NDC unit of measure code to be missing if a TAF null condition was met and invalid if it contained any value not on the valid value list (Table 1).
Table 1. Missing, invalid, and unexpected values for NDC code, days’ supply, NDC quantity, and NDC unit of measure
| Data element | Missing values (TAF null conditions) | Invalid values | Unexpected values |
|---|---|---|---|
| NDC code (NDC_CD) | Blank, 8-filled, or 9-filled | The NDC was not present in the FDA’s NDC Directory at the end of the year of analysis | N/A |
| Days’ supply (SUPLY_DAYS_CNT) | Blank, 8-filled, or 9-filled | ≤ 0 or non-numeric | > 180 |
| NDC quantity (NDC_QTY) | Blank, 8-filled, or 9-filled | Non-numeric | N/A |
| NDC unit of measure (NDC_UOM_CD) | Blank | Any value not on the valid value list: EA: Each F2: International Unit GM, GR: Grams ME: Milligrams ML: Milliliters UN: Unit | N/A |
For the data quality assessment of the NDC code, we grouped states into categories of low, medium, and high concern about the usability of their data, based on the criteria shown in Table 2.
Table 2. Criteria for DQ assessment of NDC
| Percentage of RX claim lines with a missing or invalid NDC | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
For the data quality assessment of days’ supply and NDC quantity, we grouped states into categories of low, medium, and high concern about the usability of their data, based on the criteria shown in Table 3.
Table 3. Criteria for DQ assessment of days’ supply and NDC quantity
| Percentage of RX claim lines with a missing, invalid, or unexpected value for days’ supply, or a missing or invalid value for NDC quantity | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
This analysis used the TAF data released as TAF Research Identifiable Files (RIFs). During the transformation into RIF some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page ; a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We identified FFS claims and managed care encounters by selecting records with claim type code values of 1, A, 3, and C.
More information on other non-program claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits, rather than voiding the original claim and submitting a replacement record containing the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This practice means that in some cases, the TAF will include multiple versions of a single claim for Illinois. The original record is likely to contain the most complete nonpayment information (such as diagnosis code, procedure code, revenue center code, and so forth), and thus is likely to be the most usable record in the claim family for analytic purposes. Therefore, for Illinois, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide “How to Use Illinois Claims Data,” available on the DQ Atlas Resources page under “Technical guidance for using TAF RIF data.”
The NDC code data element is present only in the RX claim lines and not in the RX claim headers. The TAF RX claim file is structured with just one claim line per claim header.
For the NDC code, days’ supply, NDC quantity, and NDC unit of measure data elements, when a value is reported in the source data that meets one of the TAF null conditions, the value is re-coded to missing (null) in the TAF RX data.
The NDC Directory is hosted by the Food and Drug Administration (FDA) and updated daily. The FDA does not publish a historical NDC directory. As a result, the 2019 analysis references the NDC Directory published on December 8, 2019, as archived by the Way Back Machine tool, available at: https://web.archive.org/web/20191208061658/https://www.accessdata.fda.gov/cder/ndcxls.zip. Accessed January 18, 2021.
Analysis of the 2019 TAF RX claim file indicates that in almost all (52 of 53) states and territories, less than 0.05 percent of claims had a days’ supply above 180, indicating that prescriptions exceeding 180 days’ supply are either outliers or incorrectly reported.
This analysis used the TAF data released as TAF Research Identifiable Files (RIFs). During the transformation into RIF some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page ; a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We identified FFS claims and managed care encounters by selecting records with claim type code values of 1, A, 3, and C.
More information on other non-program claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits, rather than voiding the original claim and submitting a replacement record containing the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This practice means that in some cases, the TAF will include multiple versions of a single claim for Illinois. The original record is likely to contain the most complete nonpayment information (such as diagnosis code, procedure code, revenue center code, and so forth), and thus is likely to be the most usable record in the claim family for analytic purposes. Therefore, for Illinois, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide \u201cHow to Use Illinois Claims Data,\u201d available on the DQ Atlas Resources page under \u201cTechnical guidance for using TAF RIF data.\u201d
The NDC code data element is present only in the RX claim lines and not in the RX claim headers. The TAF RX claim file is structured with just one claim line per claim header.
For the NDC code, days\u2019 supply, NDC quantity, and NDC unit of measure data elements, when a value is reported in the source data that meets one of the TAF null conditions, the value is re-coded to missing (null) in the TAF RX data.
The NDC Directory is hosted by the Food and Drug Administration (FDA) and updated daily. The FDA does not publish a historical NDC directory. As a result, the 2019 analysis references the NDC Directory published on December 8, 2019, as archived by the Way Back Machine tool, available at: https://web.archive.org/web/20191208061658/https://www.accessdata.fda.gov/cder/ndcxls.zip. Accessed January 18, 2021.
Analysis of the 2019 TAF RX claim file indicates that in almost all (52 of 53) states and territories, less than 0.05 percent of claims had a days\u2019 supply above 180, indicating that prescriptions exceeding 180 days\u2019 supply are either outliers or incorrectly reported.
Days' supply in the pharmacy (RX) file captures the total number of days dispensed for the drug, device, or medical supply covered by a claim. The National Drug Code (NDC) quantity indicates the numerical quantity of the drug, device, or medical supply covered by a claim. The NDC unit of measure code - which can also be identified using the NDC code itself - indicates the standard by which the value reported on the NDC quantity data element is measured. This analysis examines how often the days' supply is missing, invalid, or an unexpected value, or the NDC quantity is missing or invalid. The percentage of claims with a missing or invalid NDC unit of measure is also presented for context.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5231"", ""relatedTopics"": [{""measureId"": 89, ""measureName"": ""National Drug Code - RX"", ""groupId"": 5, ""groupName"": ""Service Use Information"", ""order"": 0}]}" 91,"{""measureId"": 91, ""measureName"": ""Active Enrollment Status Indicator"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Enroll-Status-Ind.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) Annual Provider File (APR) captures detailed information about each provider authorized by a state to provide services to its Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries, as well as providers whose approval is pending, denied, or has been terminated. Providers are included in the TAF APR regardless of whether or how often the provider has billed the state for services. The APR file encompasses all types of providers, including facilities, groups of practitioners, and individual practitioners. States also indicate the status of each provider’s enrollment in Medicaid and/or CHIP on a monthly basis, which may be classified as “active,” “pending,” “denied,” or “terminated.” [1] Most users of the TAF will want to restrict their analysis to providers who were ever active in the year or active for the month(s) of the analysis.
The APR is structured as one base file with several supplemental files that cover information regarding the characteristics, classifications, enrollment, affiliated groups, affiliated programs, locations, licensing and accreditations, identifiers, and bed types for every state-assigned provider identifier. [2] Each record in the APR TAF base file is intended to represent a single Medicaid or CHIP provider in the state and can link to more than one record in each APR supplemental file. [3] For example, one provider in the APR base file may have several records in the APR identifiers supplemental file—a National Provider Identifier (NPI), Medicare identifier, and federal tax identifier. Most TAF users should start with the APR base file to identify providers of interest and merge information from the APR supplemental files as needed for their analysis. [4]
TAF users can also use the APR file to obtain more complete information on providers who rendered or billed for a given service present in the TAF claims files by linking the claim or encounter record to the APR base file. Although TAF fee-for-service (FFS) and encounter records in the inpatient (IP), long-term care (LT), and other services (OT) files include information on the taxonomy, specialty, and state-identified provider type associated with the claim, the TAF APR offers additional details about the characteristics, locations, classifications, affiliated groups, affiliated programs, licensing/accreditations, and (for facility providers) bed types for Medicaid- or CHIP-eligible providers, as well as other identifiers associated with the provider. [5] Alternatively, TAF users may want to link provider identifiers from the APR base file to claims and encounters to obtain more information on provider activity and availability. For example, TAF users may be interested in assessing the types and quantity of services Medicaid-enrolled providers are delivering to beneficiaries as measures of network adequacy and access to care.
Users can link the APR TAF and TAF claims files using the state-assigned unique provider identifier, the provider NPI, or both, depending on the analysis. [6] For example, whereas the state-assigned provider identifier is the most commonly used means of linking claims and the APR file, TAF users may need to use the NPI instead if the analysis requires additional linkage to external files that contain NPIs, such as the National Plan and Provider Enumeration System (NPPES). [7]
This data quality assessment evaluates how well the APR TAF and TAF claims files can be linked using the state-assigned unique provider identifier to determine the following: (1) how completely the APR base file captures all providers who billed for Medicaid- and CHIP-funded medical services, and (2) the validity of the active enrollment status indicator in the APR TAF.
A provider’s Medicaid and/or CHIP enrollment can be denied due to multiple provider numbers, an invalid license, not meeting eligibility requirements, or other reasons. A provider’s enrollment status can be pending license verification, rate determination, a valid National Provider Identifier (NPI), status approval, documentation, or fulfillment of other requirements. A provider’s Medicaid/CHIP enrollment status can also be terminated for various reasons, including noncompliance, change of ownership, an expired or revoked license, voluntary termination, the provider is deceased, or other reasons. If the provider is eligible and their enrollment in Medicaid and/or CHIP is approved and not pending, denied, or terminated, the state considers the provider as actively enrolled in the state’s Medicaid and/or CHIP program.
The APR TAF includes nine files: Base, Affiliated Groups, Affiliated Programs, Taxonomy, Enrollment, Location, Licensing, Identifiers, and Bed Type. The TAF APR base file includes basic provider characteristics, and the eight additional supplemental files provide more detailed information .
TAF APR records present unique combinations of submitting state and provider identifiers. In most cases, each APR base record will represent a single provider per state. However, because providers can participate with more than one state’s Medicaid program, a given provider may have more than one TAF APR base file record. For example, it may be possible for a provider with one National Provider Identifier (NPI) to have more than one state-assigned provider identifier, which would be represented as multiple records in the APR base file. Conversely, a provider may have more than one NPI in the identifiers supplemental file but only a single state-assigned provider identifier, which would be represented as a single record in the APR base file. In addition, multiple APR base file records for the same provider are possible when a state submits separate records for each T-MSIS submission type (that is, Medicaid or separate CHIP programs).
Not all supplemental files are relevant to all providers. For example, only facility providers would be expected to have information in the APR Bed Type supplemental file.
More information on the completeness of the provider NPI on claims can be found in the DQ Atlas single-topic displays for the following: Billing Provider NPI - IP , Billing Provider NPI - LT , Billing Provider NPI - OT , Billing Provider NPI - RX , Servicing Provider NPI - OT , Prescribing Provider NPI - RX , and Dispensing Provider NPI - RX .
TAF users are advised to exercise caution when linking with NPI. Given the relative reliability and completeness of the state-assigned identifier compared to the NPI, the former is more commonly used for linking the APR TAF to other TAF files and is used in this analysis. In addition, although it is less common, some providers have more than one NPI assigned to them, and users who are interested in the provider NPI would need to link files using all available NPI numbers from the APR TAF assigned to a given state-assigned identifier. However, TAF users will need to determine which identifier is best suited for their analysis.
The NPPES NPI downloadable file can be found at https://download.cms.gov/nppes/NPI_Files.html .
A provider\u2019s Medicaid and/or CHIP enrollment can be denied due to multiple provider numbers, an invalid license, not meeting eligibility requirements, or other reasons. A provider\u2019s enrollment status can be pending license verification, rate determination, a valid National Provider Identifier (NPI), status approval, documentation, or fulfillment of other requirements. A provider\u2019s Medicaid/CHIP enrollment status can also be terminated for various reasons, including noncompliance, change of ownership, an expired or revoked license, voluntary termination, the provider is deceased, or other reasons. If the provider is eligible and their enrollment in Medicaid and/or CHIP is approved and not pending, denied, or terminated, the state considers the provider as actively enrolled in the state\u2019s Medicaid and/or CHIP program.
The APR TAF includes nine files: Base, Affiliated Groups, Affiliated Programs, Taxonomy, Enrollment, Location, Licensing, Identifiers, and Bed Type. The TAF APR base file includes basic provider characteristics, and the eight additional supplemental files provide more detailed information .
TAF APR records present unique combinations of submitting state and provider identifiers. In most cases, each APR base record will represent a single provider per state. However, because providers can participate with more than one state\u2019s Medicaid program, a given provider may have more than one TAF APR base file record. For example, it may be possible for a provider with one National Provider Identifier (NPI) to have more than one state-assigned provider identifier, which would be represented as multiple records in the APR base file. Conversely, a provider may have more than one NPI in the identifiers supplemental file but only a single state-assigned provider identifier, which would be represented as a single record in the APR base file. In addition, multiple APR base file records for the same provider are possible when a state submits separate records for each T-MSIS submission type (that is, Medicaid or separate CHIP programs).
Not all supplemental files are relevant to all providers. For example, only facility providers would be expected to have information in the APR Bed Type supplemental file.
More information on the completeness of the provider NPI on claims can be found in the DQ Atlas single-topic displays for the following: Billing Provider NPI - IP , Billing Provider NPI - LT , Billing Provider NPI - OT , Billing Provider NPI - RX , Servicing Provider NPI - OT , Prescribing Provider NPI - RX , and Dispensing Provider NPI - RX .
TAF users are advised to exercise caution when linking with NPI. Given the relative reliability and completeness of the state-assigned identifier compared to the NPI, the former is more commonly used for linking the APR TAF to other TAF files and is used in this analysis. In addition, although it is less common, some providers have more than one NPI assigned to them, and users who are interested in the provider NPI would need to link files using all available NPI numbers from the APR TAF assigned to a given state-assigned identifier. However, TAF users will need to determine which identifier is best suited for their analysis.
The NPPES NPI downloadable file can be found at https://download.cms.gov/nppes/NPI_Files.html .
This analysis uses the APR TAF base file and the monthly TAF claims files, including the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims files. [8] We limited this analysis to FFS claims and managed care encounters to represent services delivered to beneficiaries and funded through Medicaid or CHIP. [9] For Illinois, we restricted our analysis to the original version of the claim and excluded all subsequent adjustment records in the state’s TAF data. [10]
Using SUBMTG_STATE_PRVDR_ID in the APR base file, we identified all unique state-assigned provider identifiers present in the APR TAF for each state. Because of the way the APR is constructed, the number of unique state-assigned provider identifiers in the APR is equal to the number of base file records for most states. [11] We also identified all unique, state-assigned provider identifiers that appeared on FFS claims and managed care encounters in the IP, LT, OT, and RX files for each state. A single claim or encounter can include information for up to six providers, depending on the file type. [12] When developing our list of unique state-assigned provider identifiers, we included the values present in all fields specified in Table 1. [13]
Table 1. State-assigned provider identifiers used to evaluate the linkage of the TAF APR and claims files
| Provider identifier | TAF file | TAF field name |
|---|---|---|
| Submitting State Provider ID | APR base file | SUBMTG_STATE_PRVDR_ID |
| Admitting Provider | IP and LT header files | ADMTG_PRVDR_NUM |
| Billing Provider | IP, LT, OT, and RX header files | BLG_PRVDR_NUM |
| Referring Provider | IP, LT, and OT header files | RFRG_PRVDR_NUM |
| Servicing Provider | OT line file | SRVCNG_PRVDR_NUM |
| Prescribing Provider | RX header files | PRSCRBNG_PRVDR_NUM |
| Dispensing Provider | RX header files | DSPNSNG_PD_PRVDR_NUM |
Linkage of claims to APR
To determine how completely the APR TAF captures the universe of providers rendering medical services to Medicaid and CHIP beneficiaries, we evaluated the extent to which provider identifiers in the TAF IP, LT, and OT claims files could be found in the APR TAF file.
For this analysis, we identified the full set of state-assigned provider identifiers present on medical claims (IP, LT and OT claims) and calculated the percentage of these identifiers that could link to a provider identifier in the APR base file. [14] We assigned an overall level of data quality concern for the linkage of claims to APR based on this measure (Table 2). [15] We then identified the subset of unique state-assigned provider identifiers present in the claims for each claims file type (IP, LT, OT, and RX) and calculated the percentage of these identifiers that matched a provider identifier in the APR base file. This information is presented in the data table but was not used to assign a concern level. TAF users interested in linking pharmacy claims or another subset of claims to the TAF APR can refer to the file-specific results relevant to their analyses. [16]
Table 2. Criteria for DQ assessment of the linkage of claims to APR
| Percentage of unique provider IDs on medical claims that link to the APR file | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
Active enrollment status indicator
To further assess the validity of the active enrollment status indicator, we examined whether the active enrollment status for a provider in the APR file aligned with information in the claims files. For example, we would expect most providers identified as actively enrolled in the APR file to appear on at least one FFS claim or managed care encounter in the year to indicate that the provider delivered or billed for a Medicaid or CHIP-funded service. Conversely, we would not expect any Medicaid or CHIP claim records in the year for providers whose Medicaid and CHIP enrollment status was classified as pending, denied, or terminated for the entire year.
To identify any unexpected patterns that might indicate a data quality problem in the active enrollment status indicator, we evaluated two scenarios—(1) the provider had “active” enrollment in the APR file but did not appear on any claim in the year, or (2) the provider had “non-active” enrollment in the APR file and did appear on a claim in the year. Both combinations are unexpected and should occur infrequently.
We first identified the subset of provider records in the APR base file ever active during the year, using the variable PRVDR_ENRLMT_STUS_ACTV_IND. [17] Next, we calculated the percentage of all unique provider identifiers in the APR base file that were ever-active and did not link to an FFS claim or managed care encounter record in the IP, LT, OT, or RX files in the year.
We then identified the subset of provider records in the APR base file that had a non-active enrollment status (pending, denied, or terminated) for the entire year using the PRVDR_ENRLMT_STUS_ACTV_IND variable. Next, we calculated the percentage of all unique provider identifiers in the APR base file that had a non-active enrollment status and linked to an FFS claim or managed care encounter record in the IP, LT, OT, or RX files in the year.
Finally, we calculated the overall percentage of all unique provider identifiers in the APR file for which neither of the two unexpected scenarios applied. This measure represents provider IDs where (1) the provider had “active” enrollment in the APR file and appeared on any claim in the year, or (2) the provider had “non-active” enrollment in the APR file and did not appear on a claim in the year. Both combinations indicate alignment between the active enrollment status indicator in the APR file and what appears on the claims. [18] We then assigned the overall level of concern for the active enrollment status indicator based on the percentage of all provider IDs where the active enrollment status indicator aligns with the claims (Table 3). [19]
Table 3. Criteria for DQ assessment of active provider indicator
| Percentage of unique APR provider IDs where the active enrollment status indicator aligns with the claims | DQ assessment |
|---|---|
| x ≥ 85 percent | Low concern |
| 75 percent ≤ x < 85 percent | Medium concern |
| 50 percent ≤ x < 75 percent | High concern |
| x < 50 percent | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIFs). During the transformation into RIFs, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We implemented all claims exclusions by using the claim type code (CLM_TYPE_CD). We excluded capitation payments, supplemental payments, and service tracking payments. We also excluded records with a claim type of “other,” which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Titles XIX or XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits, rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts were correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide “How to Use Illinois Claims Data,” available on ResDAC.org.
Some states may submit separate provider records for each data submission type in T-MSIS (that is, Medicaid or separate CHIP programs).
The TAF claims files contain numerous provider identifier variables that represent the roles providers play in beneficiaries’ health care. The IP file includes information about billing, referring, servicing, admitting, and operating providers. The LT file includes information about billing, referring, servicing, and admitting providers. The OT file includes information about billing, referring, servicing, health home, directing, and supervising providers. The RX file includes information about billing, dispensing, and prescribing providers.
State-assigned provider identifier fields are available on a claim for admitting, billing, servicing, referring, dispensing, and prescribing providers. Although claims also include information for health home, supervising, operating, and directing providers, the provider identifier fields available for APR linkage for these providers are limited to NPI only. However, more than 95 percent of provider NPIs belonging to health home, supervising, operating, and directing providers can also be found elsewhere in the claims files in another NPI field where a state-assigned provider identifier is available (admitting, billing, servicing, referring, and dispensing NPIs). Therefore, we excluded operating, health home, directing, and supervising provider identifiers from this analysis. However, TAF users interested in these specific types of providers have the option to link the APR TAF and claims using the NPI fields only.
This measure represents all unique provider identifiers on medical claims (in the IP, LT, and OT files) that appear in any of the following provider identifier fields: admitting, billing, servicing, and referring.
The data quality assessment was limited to the linkage of medical claims to the APR because we anticipate most TAF users will want to use the APR to study the types of providers found in medical claims, such as physicians, hospitals, and other medical providers. While billing, prescribing and dispensing providers on pharmacy claims (RX file) were excluded from the overall data quality assessment of the linkage of claims to the APR, this information is presented separately in the data table for TAF users who are interested in providers appearing on pharmacy claims.
TAF users may be interested in using only a subset of claims to identify providers, such as providers delivering or billing for long-term care services.
States report enrollment status for provider records monthly as a part of the provider file. In the APR TAF, the value of the active provider indicator (PRVDR_ENRLMT_STUS_ACTV_IND) is 1 when the provider’s enrollment in the state’s Medicaid or CHIP program is reported as active for at least one month in the year and 0 when the provider’s enrollment is pending, denied, or terminated for all months in the year.
If a state has a noticeably low proportion of provider records in the APR file in which the active enrollment status matches information in the claims, the state’s APR file may not be suitable for analyses of provider participation and network adequacy. For instance, if only a small proportion of active providers have at least one claim in the year, it may indicate that some providers might not be enrolled in the state’s Medicaid and/or CHIP program despite being reported as active in the state’s APR file. Conversely, if only a small percentage of non-active providers are without a claim in the year, it may indicate that some providers may be actively enrolled in the state’s Medicaid and/or CHIP program despite their enrollment being identified as pending, denied, or terminated in the state’s APR file.
Though uncommon, there may be providers who are in fact actively enrolled but do not have a claim in a year because they did not render a Medicaid- or CHIP-funded service in the year. When determining our thresholds for data quality assessment, we accounted for a small percentage of actively enrolled providers in the APR file who may not have a claim in the year for valid reasons unrelated to data quality.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIFs). During the transformation into RIFs, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We implemented all claims exclusions by using the claim type code (CLM_TYPE_CD). We excluded capitation payments, supplemental payments, and service tracking payments. We also excluded records with a claim type of \u201cother,\u201d which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Titles XIX or XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits, rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts were correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide \u201cHow to Use Illinois Claims Data,\u201d available on ResDAC.org.
Some states may submit separate provider records for each data submission type in T-MSIS (that is, Medicaid or separate CHIP programs).
The TAF claims files contain numerous provider identifier variables that represent the roles providers play in beneficiaries\u2019 health care. The IP file includes information about billing, referring, servicing, admitting, and operating providers. The LT file includes information about billing, referring, servicing, and admitting providers. The OT file includes information about billing, referring, servicing, health home, directing, and supervising providers. The RX file includes information about billing, dispensing, and prescribing providers.
State-assigned provider identifier fields are available on a claim for admitting, billing, servicing, referring, dispensing, and prescribing providers. Although claims also include information for health home, supervising, operating, and directing providers, the provider identifier fields available for APR linkage for these providers are limited to NPI only. However, more than 95 percent of provider NPIs belonging to health home, supervising, operating, and directing providers can also be found elsewhere in the claims files in another NPI field where a state-assigned provider identifier is available (admitting, billing, servicing, referring, and dispensing NPIs). Therefore, we excluded operating, health home, directing, and supervising provider identifiers from this analysis. However, TAF users interested in these specific types of providers have the option to link the APR TAF and claims using the NPI fields only.
This measure represents all unique provider identifiers on medical claims (in the IP, LT, and OT files) that appear in any of the following provider identifier fields: admitting, billing, servicing, and referring.
The data quality assessment was limited to the linkage of medical claims to the APR because we anticipate most TAF users will want to use the APR to study the types of providers found in medical claims, such as physicians, hospitals, and other medical providers. While billing, prescribing and dispensing providers on pharmacy claims (RX file) were excluded from the overall data quality assessment of the linkage of claims to the APR, this information is presented separately in the data table for TAF users who are interested in providers appearing on pharmacy claims.
TAF users may be interested in using only a subset of claims to identify providers, such as providers delivering or billing for long-term care services.
States report enrollment status for provider records monthly as a part of the provider file. In the APR TAF, the value of the active provider indicator (PRVDR_ENRLMT_STUS_ACTV_IND) is 1 when the provider\u2019s enrollment in the state\u2019s Medicaid or CHIP program is reported as active for at least one month in the year and 0 when the provider\u2019s enrollment is pending, denied, or terminated for all months in the year.
If a state has a noticeably low proportion of provider records in the APR file in which the active enrollment status matches information in the claims, the state\u2019s APR file may not be suitable for analyses of provider participation and network adequacy. For instance, if only a small proportion of active providers have at least one claim in the year, it may indicate that some providers might not be enrolled in the state\u2019s Medicaid and/or CHIP program despite being reported as active in the state\u2019s APR file. Conversely, if only a small percentage of non-active providers are without a claim in the year, it may indicate that some providers may be actively enrolled in the state\u2019s Medicaid and/or CHIP program despite their enrollment being identified as pending, denied, or terminated in the state\u2019s APR file.
Though uncommon, there may be providers who are in fact actively enrolled but do not have a claim in a year because they did not render a Medicaid- or CHIP-funded service in the year. When determining our thresholds for data quality assessment, we accounted for a small percentage of actively enrolled providers in the APR file who may not have a claim in the year for valid reasons unrelated to data quality.
The TAF Annual Provider (APR) file contains detailed information about each provider authorized to deliver services to Medicaid and CHIP beneficiaries. This topic examines important elements of provider information in the TAF APR. This analysis evaluates the validity of the active enrollment status indicator in the TAF APR.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""9041"", ""relatedTopics"": [{""measureId"": 92, ""measureName"": ""Linking Claims to Providers"", ""groupId"": 11, ""groupName"": ""Linking Across Files"", ""order"": 0}]}" 92,"{""measureId"": 92, ""measureName"": ""Linking Claims to Providers"", ""groupId"": 11, ""groupName"": ""Linking Across Files"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Link-Claims-Providers.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) Annual Provider File (APR) captures detailed information about each provider authorized by a state to provide services to its Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries, as well as providers whose approval is pending, denied, or has been terminated. Providers are included in the TAF APR regardless of whether or how often the provider has billed the state for services. The APR file encompasses all types of providers, including facilities, groups of practitioners, and individual practitioners. States also indicate the status of each provider’s enrollment in Medicaid and/or CHIP on a monthly basis, which may be classified as “active,” “pending,” “denied,” or “terminated.” [1] Most users of the TAF will want to restrict their analysis to providers who were ever active in the year or active for the month(s) of the analysis.
The APR is structured as one base file with several supplemental files that cover information regarding the characteristics, classifications, enrollment, affiliated groups, affiliated programs, locations, licensing and accreditations, identifiers, and bed types for every state-assigned provider identifier. [2] Each record in the APR TAF base file is intended to represent a single Medicaid or CHIP provider in the state and can link to more than one record in each APR supplemental file. [3] For example, one provider in the APR base file may have several records in the APR identifiers supplemental file—a National Provider Identifier (NPI), Medicare identifier, and federal tax identifier. Most TAF users should start with the APR base file to identify providers of interest and merge information from the APR supplemental files as needed for their analysis. [4]
TAF users can also use the APR file to obtain more complete information on providers who rendered or billed for a given service present in the TAF claims files by linking the claim or encounter record to the APR base file. Although TAF fee-for-service (FFS) and encounter records in the inpatient (IP), long-term care (LT), and other services (OT) files include information on the taxonomy, specialty, and state-identified provider type associated with the claim, the TAF APR offers additional details about the characteristics, locations, classifications, affiliated groups, affiliated programs, licensing/accreditations, and (for facility providers) bed types for Medicaid- or CHIP-eligible providers, as well as other identifiers associated with the provider. [5] Alternatively, TAF users may want to link provider identifiers from the APR base file to claims and encounters to obtain more information on provider activity and availability. For example, TAF users may be interested in assessing the types and quantity of services Medicaid-enrolled providers are delivering to beneficiaries as measures of network adequacy and access to care.
Users can link the APR TAF and TAF claims files using the state-assigned unique provider identifier, the provider NPI, or both, depending on the analysis. [6] For example, whereas the state-assigned provider identifier is the most commonly used means of linking claims and the APR file, TAF users may need to use the NPI instead if the analysis requires additional linkage to external files that contain NPIs, such as the National Plan and Provider Enumeration System (NPPES). [7]
This data quality assessment evaluates how well the APR TAF and TAF claims files can be linked using the state-assigned unique provider identifier to determine the following: (1) how completely the APR base file captures all providers who billed for Medicaid- and CHIP-funded medical services, and (2) the validity of the active enrollment status indicator in the APR TAF.
A provider’s Medicaid and/or CHIP enrollment can be denied due to multiple provider numbers, an invalid license, not meeting eligibility requirements, or other reasons. A provider’s enrollment status can be pending license verification, rate determination, a valid National Provider Identifier (NPI), status approval, documentation, or fulfillment of other requirements. A provider’s Medicaid/CHIP enrollment status can also be terminated for various reasons, including noncompliance, change of ownership, an expired or revoked license, voluntary termination, the provider is deceased, or other reasons. If the provider is eligible and their enrollment in Medicaid and/or CHIP is approved and not pending, denied, or terminated, the state considers the provider as actively enrolled in the state’s Medicaid and/or CHIP program.
The APR TAF includes nine files: Base, Affiliated Groups, Affiliated Programs, Taxonomy, Enrollment, Location, Licensing, Identifiers, and Bed Type. The TAF APR base file includes basic provider characteristics, and the eight additional supplemental files provide more detailed information .
TAF APR records present unique combinations of submitting state and provider identifiers. In most cases, each APR base record will represent a single provider per state. However, because providers can participate with more than one state’s Medicaid program, a given provider may have more than one TAF APR base file record. For example, it may be possible for a provider with one National Provider Identifier (NPI) to have more than one state-assigned provider identifier, which would be represented as multiple records in the APR base file. Conversely, a provider may have more than one NPI in the identifiers supplemental file but only a single state-assigned provider identifier, which would be represented as a single record in the APR base file. In addition, multiple APR base file records for the same provider are possible when a state submits separate records for each T-MSIS submission type (that is, Medicaid or separate CHIP programs).
Not all supplemental files are relevant to all providers. For example, only facility providers would be expected to have information in the APR Bed Type supplemental file.
More information on the completeness of the provider NPI on claims can be found in the DQ Atlas single-topic displays for the following: Billing Provider NPI - IP , Billing Provider NPI - LT , Billing Provider NPI - OT , Billing Provider NPI - RX , Servicing Provider NPI - OT , Prescribing Provider NPI - RX , and Dispensing Provider NPI - RX .
TAF users are advised to exercise caution when linking with NPI. Given the relative reliability and completeness of the state-assigned identifier compared to the NPI, the former is more commonly used for linking the APR TAF to other TAF files and is used in this analysis. In addition, although it is less common, some providers have more than one NPI assigned to them, and users who are interested in the provider NPI would need to link files using all available NPI numbers from the APR TAF assigned to a given state-assigned identifier. However, TAF users will need to determine which identifier is best suited for their analysis.
The NPPES NPI downloadable file can be found at https://download.cms.gov/nppes/NPI_Files.html .
A provider\u2019s Medicaid and/or CHIP enrollment can be denied due to multiple provider numbers, an invalid license, not meeting eligibility requirements, or other reasons. A provider\u2019s enrollment status can be pending license verification, rate determination, a valid National Provider Identifier (NPI), status approval, documentation, or fulfillment of other requirements. A provider\u2019s Medicaid/CHIP enrollment status can also be terminated for various reasons, including noncompliance, change of ownership, an expired or revoked license, voluntary termination, the provider is deceased, or other reasons. If the provider is eligible and their enrollment in Medicaid and/or CHIP is approved and not pending, denied, or terminated, the state considers the provider as actively enrolled in the state\u2019s Medicaid and/or CHIP program.
The APR TAF includes nine files: Base, Affiliated Groups, Affiliated Programs, Taxonomy, Enrollment, Location, Licensing, Identifiers, and Bed Type. The TAF APR base file includes basic provider characteristics, and the eight additional supplemental files provide more detailed information .
TAF APR records present unique combinations of submitting state and provider identifiers. In most cases, each APR base record will represent a single provider per state. However, because providers can participate with more than one state\u2019s Medicaid program, a given provider may have more than one TAF APR base file record. For example, it may be possible for a provider with one National Provider Identifier (NPI) to have more than one state-assigned provider identifier, which would be represented as multiple records in the APR base file. Conversely, a provider may have more than one NPI in the identifiers supplemental file but only a single state-assigned provider identifier, which would be represented as a single record in the APR base file. In addition, multiple APR base file records for the same provider are possible when a state submits separate records for each T-MSIS submission type (that is, Medicaid or separate CHIP programs).
Not all supplemental files are relevant to all providers. For example, only facility providers would be expected to have information in the APR Bed Type supplemental file.
More information on the completeness of the provider NPI on claims can be found in the DQ Atlas single-topic displays for the following: Billing Provider NPI - IP , Billing Provider NPI - LT , Billing Provider NPI - OT , Billing Provider NPI - RX , Servicing Provider NPI - OT , Prescribing Provider NPI - RX , and Dispensing Provider NPI - RX .
TAF users are advised to exercise caution when linking with NPI. Given the relative reliability and completeness of the state-assigned identifier compared to the NPI, the former is more commonly used for linking the APR TAF to other TAF files and is used in this analysis. In addition, although it is less common, some providers have more than one NPI assigned to them, and users who are interested in the provider NPI would need to link files using all available NPI numbers from the APR TAF assigned to a given state-assigned identifier. However, TAF users will need to determine which identifier is best suited for their analysis.
The NPPES NPI downloadable file can be found at https://download.cms.gov/nppes/NPI_Files.html .
This analysis uses the APR TAF base file and the monthly TAF claims files, including the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims files. [8] We limited this analysis to FFS claims and managed care encounters to represent services delivered to beneficiaries and funded through Medicaid or CHIP. [9] For Illinois, we restricted our analysis to the original version of the claim and excluded all subsequent adjustment records in the state’s TAF data. [10]
Using SUBMTG_STATE_PRVDR_ID in the APR base file, we identified all unique state-assigned provider identifiers present in the APR TAF for each state. Because of the way the APR is constructed, the number of unique state-assigned provider identifiers in the APR is equal to the number of base file records for most states. [11] We also identified all unique, state-assigned provider identifiers that appeared on FFS claims and managed care encounters in the IP, LT, OT, and RX files for each state. A single claim or encounter can include information for up to six providers, depending on the file type. [12] When developing our list of unique state-assigned provider identifiers, we included the values present in all fields specified in Table 1. [13]
Table 1. State-assigned provider identifiers used to evaluate the linkage of the TAF APR and claims files
| Provider identifier | TAF file | TAF field name |
|---|---|---|
| Submitting State Provider ID | APR base file | SUBMTG_STATE_PRVDR_ID |
| Admitting Provider | IP and LT header files | ADMTG_PRVDR_NUM |
| Billing Provider | IP, LT, OT, and RX header files | BLG_PRVDR_NUM |
| Referring Provider | IP, LT, and OT header files | RFRG_PRVDR_NUM |
| Servicing Provider | OT line file | SRVCNG_PRVDR_NUM |
| Prescribing Provider | RX header files | PRSCRBNG_PRVDR_NUM |
| Dispensing Provider | RX header files | DSPNSNG_PD_PRVDR_NUM |
Linkage of claims to APR
To determine how completely the APR TAF captures the universe of providers rendering medical services to Medicaid and CHIP beneficiaries, we evaluated the extent to which provider identifiers in the TAF IP, LT, and OT claims files could be found in the APR TAF file.
For this analysis, we identified the full set of state-assigned provider identifiers present on medical claims (IP, LT and OT claims) and calculated the percentage of these identifiers that could link to a provider identifier in the APR base file. [14] We assigned an overall level of data quality concern for the linkage of claims to APR based on this measure (Table 2). [15] We then identified the subset of unique state-assigned provider identifiers present in the claims for each claims file type (IP, LT, OT, and RX) and calculated the percentage of these identifiers that matched a provider identifier in the APR base file. This information is presented in the data table but was not used to assign a concern level. TAF users interested in linking pharmacy claims or another subset of claims to the TAF APR can refer to the file-specific results relevant to their analyses. [16]
Table 2. Criteria for DQ assessment of the linkage of claims to APR
| Percentage of unique provider IDs on medical claims that link to the APR file | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
Active enrollment status indicator
To further assess the validity of the active enrollment status indicator, we examined whether the active enrollment status for a provider in the APR file aligned with information in the claims files. For example, we would expect most providers identified as actively enrolled in the APR file to appear on at least one FFS claim or managed care encounter in the year to indicate that the provider delivered or billed for a Medicaid or CHIP-funded service. Conversely, we would not expect any Medicaid or CHIP claim records in the year for providers whose Medicaid and CHIP enrollment status was classified as pending, denied, or terminated for the entire year.
To identify any unexpected patterns that might indicate a data quality problem in the active enrollment status indicator, we evaluated two scenarios—(1) the provider had “active” enrollment in the APR file but did not appear on any claim in the year, or (2) the provider had “non-active” enrollment in the APR file and did appear on a claim in the year. Both combinations are unexpected and should occur infrequently.
We first identified the subset of provider records in the APR base file ever active during the year, using the variable PRVDR_ENRLMT_STUS_ACTV_IND. [17] Next, we calculated the percentage of all unique provider identifiers in the APR base file that were ever-active and did not link to an FFS claim or managed care encounter record in the IP, LT, OT, or RX files in the year.
We then identified the subset of provider records in the APR base file that had a non-active enrollment status (pending, denied, or terminated) for the entire year using the PRVDR_ENRLMT_STUS_ACTV_IND variable. Next, we calculated the percentage of all unique provider identifiers in the APR base file that had a non-active enrollment status and linked to an FFS claim or managed care encounter record in the IP, LT, OT, or RX files in the year.
Finally, we calculated the overall percentage of all unique provider identifiers in the APR file for which neither of the two unexpected scenarios applied. This measure represents provider IDs where (1) the provider had “active” enrollment in the APR file and appeared on any claim in the year, or (2) the provider had “non-active” enrollment in the APR file and did not appear on a claim in the year. Both combinations indicate alignment between the active enrollment status indicator in the APR file and what appears on the claims. [18] We then assigned the overall level of concern for the active enrollment status indicator based on the percentage of all provider IDs where the active enrollment status indicator aligns with the claims (Table 3). [19]
Table 3. Criteria for DQ assessment of active provider indicator
| Percentage of unique APR provider IDs where the active enrollment status indicator aligns with the claims | DQ assessment |
|---|---|
| x ≥ 85 percent | Low concern |
| 75 percent ≤ x < 85 percent | Medium concern |
| 50 percent ≤ x < 75 percent | High concern |
| x < 50 percent | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIFs). During the transformation into RIFs, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We implemented all claims exclusions by using the claim type code (CLM_TYPE_CD). We excluded capitation payments, supplemental payments, and service tracking payments. We also excluded records with a claim type of “other,” which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Titles XIX or XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits, rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts were correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide “How to Use Illinois Claims Data,” available on ResDAC.org.
Some states may submit separate provider records for each data submission type in T-MSIS (that is, Medicaid or separate CHIP programs).
The TAF claims files contain numerous provider identifier variables that represent the roles providers play in beneficiaries’ health care. The IP file includes information about billing, referring, servicing, admitting, and operating providers. The LT file includes information about billing, referring, servicing, and admitting providers. The OT file includes information about billing, referring, servicing, health home, directing, and supervising providers. The RX file includes information about billing, dispensing, and prescribing providers.
State-assigned provider identifier fields are available on a claim for admitting, billing, servicing, referring, dispensing, and prescribing providers. Although claims also include information for health home, supervising, operating, and directing providers, the provider identifier fields available for APR linkage for these providers are limited to NPI only. However, more than 95 percent of provider NPIs belonging to health home, supervising, operating, and directing providers can also be found elsewhere in the claims files in another NPI field where a state-assigned provider identifier is available (admitting, billing, servicing, referring, and dispensing NPIs). Therefore, we excluded operating, health home, directing, and supervising provider identifiers from this analysis. However, TAF users interested in these specific types of providers have the option to link the APR TAF and claims using the NPI fields only.
This measure represents all unique provider identifiers on medical claims (in the IP, LT, and OT files) that appear in any of the following provider identifier fields: admitting, billing, servicing, and referring.
The data quality assessment was limited to the linkage of medical claims to the APR because we anticipate most TAF users will want to use the APR to study the types of providers found in medical claims, such as physicians, hospitals, and other medical providers. While billing, prescribing and dispensing providers on pharmacy claims (RX file) were excluded from the overall data quality assessment of the linkage of claims to the APR, this information is presented separately in the data table for TAF users who are interested in providers appearing on pharmacy claims.
TAF users may be interested in using only a subset of claims to identify providers, such as providers delivering or billing for long-term care services.
States report enrollment status for provider records monthly as a part of the provider file. In the APR TAF, the value of the active provider indicator (PRVDR_ENRLMT_STUS_ACTV_IND) is 1 when the provider’s enrollment in the state’s Medicaid or CHIP program is reported as active for at least one month in the year and 0 when the provider’s enrollment is pending, denied, or terminated for all months in the year.
If a state has a noticeably low proportion of provider records in the APR file in which the active enrollment status matches information in the claims, the state’s APR file may not be suitable for analyses of provider participation and network adequacy. For instance, if only a small proportion of active providers have at least one claim in the year, it may indicate that some providers might not be enrolled in the state’s Medicaid and/or CHIP program despite being reported as active in the state’s APR file. Conversely, if only a small percentage of non-active providers are without a claim in the year, it may indicate that some providers may be actively enrolled in the state’s Medicaid and/or CHIP program despite their enrollment being identified as pending, denied, or terminated in the state’s APR file.
Though uncommon, there may be providers who are in fact actively enrolled but do not have a claim in a year because they did not render a Medicaid- or CHIP-funded service in the year. When determining our thresholds for data quality assessment, we accounted for a small percentage of actively enrolled providers in the APR file who may not have a claim in the year for valid reasons unrelated to data quality.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIFs). During the transformation into RIFs, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We implemented all claims exclusions by using the claim type code (CLM_TYPE_CD). We excluded capitation payments, supplemental payments, and service tracking payments. We also excluded records with a claim type of \u201cother,\u201d which are records that the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Titles XIX or XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits, rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts were correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide \u201cHow to Use Illinois Claims Data,\u201d available on ResDAC.org.
Some states may submit separate provider records for each data submission type in T-MSIS (that is, Medicaid or separate CHIP programs).
The TAF claims files contain numerous provider identifier variables that represent the roles providers play in beneficiaries\u2019 health care. The IP file includes information about billing, referring, servicing, admitting, and operating providers. The LT file includes information about billing, referring, servicing, and admitting providers. The OT file includes information about billing, referring, servicing, health home, directing, and supervising providers. The RX file includes information about billing, dispensing, and prescribing providers.
State-assigned provider identifier fields are available on a claim for admitting, billing, servicing, referring, dispensing, and prescribing providers. Although claims also include information for health home, supervising, operating, and directing providers, the provider identifier fields available for APR linkage for these providers are limited to NPI only. However, more than 95 percent of provider NPIs belonging to health home, supervising, operating, and directing providers can also be found elsewhere in the claims files in another NPI field where a state-assigned provider identifier is available (admitting, billing, servicing, referring, and dispensing NPIs). Therefore, we excluded operating, health home, directing, and supervising provider identifiers from this analysis. However, TAF users interested in these specific types of providers have the option to link the APR TAF and claims using the NPI fields only.
This measure represents all unique provider identifiers on medical claims (in the IP, LT, and OT files) that appear in any of the following provider identifier fields: admitting, billing, servicing, and referring.
The data quality assessment was limited to the linkage of medical claims to the APR because we anticipate most TAF users will want to use the APR to study the types of providers found in medical claims, such as physicians, hospitals, and other medical providers. While billing, prescribing and dispensing providers on pharmacy claims (RX file) were excluded from the overall data quality assessment of the linkage of claims to the APR, this information is presented separately in the data table for TAF users who are interested in providers appearing on pharmacy claims.
TAF users may be interested in using only a subset of claims to identify providers, such as providers delivering or billing for long-term care services.
States report enrollment status for provider records monthly as a part of the provider file. In the APR TAF, the value of the active provider indicator (PRVDR_ENRLMT_STUS_ACTV_IND) is 1 when the provider\u2019s enrollment in the state\u2019s Medicaid or CHIP program is reported as active for at least one month in the year and 0 when the provider\u2019s enrollment is pending, denied, or terminated for all months in the year.
If a state has a noticeably low proportion of provider records in the APR file in which the active enrollment status matches information in the claims, the state\u2019s APR file may not be suitable for analyses of provider participation and network adequacy. For instance, if only a small proportion of active providers have at least one claim in the year, it may indicate that some providers might not be enrolled in the state\u2019s Medicaid and/or CHIP program despite being reported as active in the state\u2019s APR file. Conversely, if only a small percentage of non-active providers are without a claim in the year, it may indicate that some providers may be actively enrolled in the state\u2019s Medicaid and/or CHIP program despite their enrollment being identified as pending, denied, or terminated in the state\u2019s APR file.
Though uncommon, there may be providers who are in fact actively enrolled but do not have a claim in a year because they did not render a Medicaid- or CHIP-funded service in the year. When determining our thresholds for data quality assessment, we accounted for a small percentage of actively enrolled providers in the APR file who may not have a claim in the year for valid reasons unrelated to data quality.
The TAF Annual Provider (APR) file contains detailed information about each provider authorized to deliver services to Medicaid and CHIP beneficiaries. This topic examines important elements of provider information in the TAF APR. This analysis evaluates how well the state-assigned provider identifiers in the TAF claims files can be linked with provider identifiers in the TAF APR.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""9041"", ""relatedTopics"": [{""measureId"": 91, ""measureName"": ""Active Enrollment Status Indicator"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 1}]}" 94,"{""measureId"": 94, ""measureName"": ""Facility/Group/Individual Code"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Facility-Group-Indiv-Cd.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) Annual Provider File (APR) captures detailed information about each provider authorized by a state to provide services to its Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries, regardless of whether or how often the provider billed the state for services. [1] The APR file includes more detailed information about those providers rendering or billing for the service than the fee-for-service and encounter records in the TAF claims files, which contain limited information about them. For example, TAF fee-for-service and encounter records on the inpatient (IP), long-term care (LT), and other services (OT) files include information on each claim about the billing, servicing, and rendering provider’s taxonomy, specialty, and state-identified provider type. [2] The TAF APR, on the other hand, includes details about the characteristics, locations, taxonomies/classifications, affiliated groups, affiliated programs, licensing/accreditations, other identifiers associated with the provider, and—for facility providers—the bed types linked with the facility. [3]
Each record in the TAF APR represents a provider enrolled in the state’s Medicaid or CHIP program. A provider could be a facility, a group of providers, or an individual provider. In the TAF APR, a facility is defined as an organization, institution, place, building, or agency that furnishes, conducts, and operates health care services for the prevention, diagnosis, or treatment of human disease, pain, or injury. Examples include hospitals, nursing facilities, home health agencies, schools, or transportation organizations. A group is defined as two or more physicians, advanced practice nurses, and/or physician’s assistants who work together and share facilities. [4] An individual provider is defined as an individual who provides medical or non-medical services. [5] , [6]
States must report information about areas of specialization for all providers included on the TAF APR, using at least one of four possible classification types: (1) state-reported taxonomy code, (2) specialty code, (3) provider type code, and (4) authorized category of service code. In addition to the four classification types reported by the state, TAF APR files produced in 2022 and later years also include a constructed taxonomy variable, the NPPES primary taxonomy code, that represents the provider’s primary taxonomy information as reported in National Plan and Provider Enumeration System (NPPES) based on the provider’s National Provider Identifier (NPI).
Maintained by the National Uniform Claim Committee, taxonomy codes are standardized, unique 10-character codes that designate a provider’s classification or area of specialization. Providers select one or more taxonomy codes that most closely describe their classification or area of specialization and must submit the selected code when applying for an NPI. Provider specialty, provider type, and authorized category of service also designate a provider’s classification or area of specialization, but with less granularity than taxonomy code does. Analysis of various topics, such as the geographic distribution of specialists or primary care physicians, requires accurate information about provider specialization in the TAF.
States are encouraged to report all classification types for which information is available. At a minimum, states must report taxonomy codes for providers with an NPI and authorized categories of service for providers without an NPI. In addition, most states should have relatively complete data for provider specialty or taxonomy because state Medicaid programs typically set or modify provider payments based on this information. Depending on the state, providers submit their taxonomy or specialty when enrolling with the state Medicaid or CHIP program, include it on Medicaid claim submissions, or do a combination of the two.
Other variables in the APR provide information about provider characteristics that are particularly relevant for facility providers. For instance, ownership code denotes a provider’s ownership interest and/or managing control information, while provider profit status code indicates whether the provider is a 501(c)(3) non-profit, a for-profit either closely held or publicly traded, or has any other profit status.
This analysis examines the extent to which APR records have a valid facility/group/individual code and the extent to which the proportion of providers that are facilities, groups, and individuals falls within the expected range. The analysis also explores the extent to which TAF APR records representing group and individual providers—and, separately, those representing facility providers—have a valid, usable, and applicable value for at least one of the four provider classification types. Finally, the analysis examines the extent to which TAF APR records representing facility providers have a usable ownership code and provider profit status code.
TAF APR records present unique combinations of submitting state and provider. Because providers can participate with more than one state’s Medicaid program, more than one TAF APR record may represent a given provider.
In addition, fee-for-service and encounter records on the IP and LT files include information on each claim about the admitting provider’s taxonomy, specialty, and state-identified provider type, and the fee-for-service and encounter records on the prescription drug (RX) file include information on each claim about the billing provider’s taxonomy and specialty.
The TAF APR consists of nine files. The TAF APR base file includes basic provider characteristics. Eight additional supplemental files provide more detailed information on the following topics: the provider’s taxonomy, Medicaid and/or CHIP program enrollment, group affiliation, health plan and program affiliation, geographic location(s), licensing and accreditation, other provider identifiers, and bed type. Each record in the APR TAF base file represents a provider enrolled with the state’s Medicaid or CHIP program. Each provider record on the base file may link to more than one record on each supplemental file.
Physicians within a group may have different specialties.
Individual providers include two distinct groups. The first, incorporated sole practitioners, are sole practitioners with legal separation between themselves and their practice. The second, non-incorporated sole practitioners (also known as sole proprietors), are sole practitioners who operate with no legal distinction between themselves and their practice. The latter have characteristics associated with both individuals and organizations.
Centers for Medicare & Medicaid Services (CMS). “CMS Guidance: Reporting Provider Facility-Group-Individual-Code in T-MSIS.” Baltimore, MD: CMS, n.d. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/51240 .
TAF APR records present unique combinations of submitting state and provider. Because providers can participate with more than one state\u2019s Medicaid program, more than one TAF APR record may represent a given provider.
In addition, fee-for-service and encounter records on the IP and LT files include information on each claim about the admitting provider\u2019s taxonomy, specialty, and state-identified provider type, and the fee-for-service and encounter records on the prescription drug (RX) file include information on each claim about the billing provider\u2019s taxonomy and specialty.
The TAF APR consists of nine files. The TAF APR base file includes basic provider characteristics. Eight additional supplemental files provide more detailed information on the following topics: the provider\u2019s taxonomy, Medicaid and/or CHIP program enrollment, group affiliation, health plan and program affiliation, geographic location(s), licensing and accreditation, other provider identifiers, and bed type. Each record in the APR TAF base file represents a provider enrolled with the state\u2019s Medicaid or CHIP program. Each provider record on the base file may link to more than one record on each supplemental file.
Physicians within a group may have different specialties.
Individual providers include two distinct groups. The first, incorporated sole practitioners, are sole practitioners with legal separation between themselves and their practice. The second, non-incorporated sole practitioners (also known as sole proprietors), are sole practitioners who operate with no legal distinction between themselves and their practice. The latter have characteristics associated with both individuals and organizations.
Centers for Medicare & Medicaid Services (CMS). \u201cCMS Guidance: Reporting Provider Facility-Group-Individual-Code in T-MSIS.\u201d Baltimore, MD: CMS, n.d. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/51240 .
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
This analysis included all records in the APR base file, including those classified as actively enrolled, pending, denied, or terminated. [7] We linked information from the taxonomy supplemental file to the base file records for some analyses. We first examined the percentage of APR base file records that include a valid facility/group/individual code and the extent to which the share of providers in the state’s base file that are facilities or groups versus individuals falls within the expected range. [8]
To determine the expected range of providers that are facilities or groups versus individuals, we identified all providers in the NPPES database with an active NPI in the calendar year. We then identified the subset of these providers that were Medicaid or CHIP providers as those that linked, via NPI, to at least one TAF claim (fee-for-service or encounter record) in the calendar year. [9]
In each state, we calculated the percentage of these active NPPES providers linking to a TAF claim who are individuals (which aligns with the providers whom the TAF APR defines as individuals) and that are organizations (which aligns with the providers that the TAF APR defines as groups and facilities). We set the expected range for the percentage of each state’s APR records that represent individuals equal to the range of state-level percentages of active NPPES providers linking to a TAF claim that are individuals (70 percent to 93 percent). Likewise, we set the expected range for the percentage of each state’s APR records that represent facilities and groups equal to the range of state-level percentages of active NPPES providers linking to a TAF claim that are organizations (7 percent to 30 percent). We set tiers for percentages moderately above and below and far above and below the expected range for group and facility providers and individual providers, respectively (Table 1).
Table 1. Expected distribution of group/facility and individual providers in the TAF APR
| Classification | Percentage of APR records identified as group or facility providers | Percentage of APR records identified as individual providers |
|---|---|---|
| Far above expected range | x > 40 percent | x > 97 percent |
| Moderately above expected range | 30 percent < x ≤ 40 percent | 93 percent < x ≤ 97 percent |
| Within expected range | 7 percent < x ≤ 30 percent | 70 percent < x ≤ 93 percent |
| Moderately below expected range | 3 percent < x ≤ 7 percent | 60 percent < x ≤ 70 percent |
| Far below expected range | x ≤ 3 percent | x ≤ 60 percent |
We based the data quality assessment of the facility/group/individual code on three criteria: (1) the percentage of APR records with a valid facility/group/individual code, (2) the extent to which the reported percentage of individual providers falls within the expected range, and (3) the extent to which the reported percentage of facility and group providers falls within the expected range. [10] We assigned states to a low level of concern about data quality if all three specified criteria were true (Table 2). If any of the three criteria specified for the “low concern” category were not true, we assigned the state to the highest concern category for which at least one of the three criteria were true.
Table 2. Criteria for DQ assessment of facility/group/individual code
| Percent of APR records with a valid facility-group-individual code | Percentage of APR records identified as group or facility providers | Percentage of APR records identified as individual providers | DQ assessment |
|---|---|---|---|
| x ≥ 90 percent | Within expected range | Within expected range | Low concern a |
| 80 percent ≤ x < 90 percent | Moderately above or below expected range | Moderately above or below expected range | Medium concern b |
| 50 percent ≤ x < 80 percent | Far above or below expected range | Far above or below expected range | High concern b |
| x < 50 percent | x < 1 percent (facility) or x < 1 percent (group) | x < 1 percent | Unusable b |
a All three criteria must be true for a state to receive the given DQ Assessment.
b One of the three criteria must be true for a state to receive the given DQ Assessment.
For group and individual providers—and separately for facility providers—we examined the extent to which APR records have a valid, usable, and applicable value for at least one of the five provider classification fields (state-reported taxonomy, NPPES primary taxonomy, [11] provider specialty, provider type, or authorized category of service). For a record to count as having a valid, usable, and applicable value for a given classification type, it must meet all of the following criteria:
For each topic, we grouped states into categories of concern about the usability of their data, based on the total percentage of APR records that represent group and individual providers and facility providers, respectively, that have a valid, usable, and applicable value reported for at least one of the five classification types (Table 3). For context, we also calculated the percentage of APR records with valid, usable, and applicable codes for more than one classification type.
TAF APR users may want to conduct additional data quality assessments before using the provider specialty, type, and authorized category of service fields to classify providers. A state could possibly use an improper code set (for instance, the Medicare provider type code set instead of the TAF provider type code set) with valid values that overlap with the TAF provider specialty, type, and authorized category of service codes. [13] TAF APR users can screen for this issue by examining the most prevalent provider specialties, types, or authorized category of service codes for the provider category of interest (for instance, facilities) to ensure that the most commonly reported codes align with expectations.
Table 3. Criteria for DQ assessment of provider classification types
| Percentage of APR records with a valid, usable, and applicable value reported for at least one of the five classification types | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
For facility providers, we also examined the extent to which APR records include usable information for two variables that provide more information about provider characteristics: ownership code and provider profit status code. We grouped states into categories of concern about the usability of their data, based on the total percentage of APR records that represent facility providers with a valid value reported for both data elements (Table 4). [14]
Table 4. Criteria for DQ assessment of facility provider characteristics
| Percentage of APR records that represent facilities with a valid value reported for ownership code and provider profit status code | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
Methods previously used to assess data quality
Table 5 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. Current methods are used to assess all data years and versions not listed in the table, aside from data years and versions for which this data quality analysis is unavailable.
Table 5. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
|
|
This analysis drew on the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. More details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the “Production of the TAF Research Identifiable Files” guide.
A valid facility/group/individual code (FAC_GRP_INDVDL_CD) is one of the following values: “01” (facility), “02” (group), or “03” (individual).
We linked active NPIs in NPPES (those without a deactivation date prior to the start of the analysis year) to the NPIs on FFS claims and managed care encounters funded through Medicaid or CHIP (CLM_TYPE_CD = 1, 3, A, or C) on the monthly TAF claims files for the year of analysis, including the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims files. We excluded capitation payments, supplemental payments, service tracking payments, and records with a claim type of “other.” For Illinois, we only linked to the original version of the claim and did not link to subsequent adjustment records in the state’s TAF data. For more information, see the guide “How to Use Illinois Claims Data,” available on ResDAC.org. The TAF claims files contain numerous provider identifier variables that represent the roles providers play in beneficiaries’ health care. This analysis linked the NPIs in NPPES to the following NPIs on the TAF claims files: (1) the IP and LT billing (BLG_PRVDR_NPI_NUM), referring (RFRG_PRVDR_NPI_NUM), servicing (SRVCNG_PRVDR_NPI_NUM), and admitting (ADMTG_PRVDR_NPI_NUM) provider NPIs, (2) the OT billing (BLG_PRVDR_NPI_NUM), referring (RFRG_PRVDR_NPI_NUM), and servicing (SRVCNG_PRVDR_NPI_NUM) provider NPIs, and (3) the RX billing (BLG_PRVDR_NPI_NUM), prescribing (SRVCNG_PRVDR_NPI_NUM), and dispensing (DSPNSNG_PD_PRVDR_NPI_NUM) provider NPIs.
This criterion also includes an assessment of the extent to which the reported percentage of group providers and facility providers, evaluated separately, has face validity. States reporting less than 1 percent of providers that are groups, less than 1 percent that are facilities, or less than 1 percent that are individuals were considered to lack face validity and were classified as “unusable” for the facility/group/individual code.
In addition to the state-reported provider taxonomy codes, TAF APR files produced in 2022 and later years also include a constructed taxonomy variable, the NPPES primary taxonomy code, that represents the provider’s primary taxonomy information as reported in NPPES based on the provider’s NPI.
The current provider taxonomy code list is available at www.wpc-edi.com/reference . The provider specialty, type, and authorized category of service code sets are available in the appendixes of the TAF APR data dictionary.
CMS. “CMS Guidance: Best Practice for Reporting PROV‐CLASSIFICATION‐TYPE and PROV‐CLASSIFICATION‐CODE in the T‐MSIS Provider File.” Baltimore, MD: CMS, n.d. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/47562 .
A valid ownership code (OWNRSHP_CD) is in the range “01” – “19”. A valid provider profit status code (PRVDR_PRFT_STUS_CD) is in the range “01” – “04”.
This analysis drew on the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. More details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the \u201cProduction of the TAF Research Identifiable Files\u201d guide.
A valid facility/group/individual code (FAC_GRP_INDVDL_CD) is one of the following values: \u201c01\u201d (facility), \u201c02\u201d (group), or \u201c03\u201d (individual).
We linked active NPIs in NPPES (those without a deactivation date prior to the start of the analysis year) to the NPIs on FFS claims and managed care encounters funded through Medicaid or CHIP (CLM_TYPE_CD = 1, 3, A, or C) on the monthly TAF claims files for the year of analysis, including the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims files. We excluded capitation payments, supplemental payments, service tracking payments, and records with a claim type of \u201cother.\u201d For Illinois, we only linked to the original version of the claim and did not link to subsequent adjustment records in the state\u2019s TAF data. For more information, see the guide \u201cHow to Use Illinois Claims Data,\u201d available on ResDAC.org. The TAF claims files contain numerous provider identifier variables that represent the roles providers play in beneficiaries\u2019 health care. This analysis linked the NPIs in NPPES to the following NPIs on the TAF claims files: (1) the IP and LT billing (BLG_PRVDR_NPI_NUM), referring (RFRG_PRVDR_NPI_NUM), servicing (SRVCNG_PRVDR_NPI_NUM), and admitting (ADMTG_PRVDR_NPI_NUM) provider NPIs, (2) the OT billing (BLG_PRVDR_NPI_NUM), referring (RFRG_PRVDR_NPI_NUM), and servicing (SRVCNG_PRVDR_NPI_NUM) provider NPIs, and (3) the RX billing (BLG_PRVDR_NPI_NUM), prescribing (SRVCNG_PRVDR_NPI_NUM), and dispensing (DSPNSNG_PD_PRVDR_NPI_NUM) provider NPIs.
This criterion also includes an assessment of the extent to which the reported percentage of group providers and facility providers, evaluated separately, has face validity. States reporting less than 1 percent of providers that are groups, less than 1 percent that are facilities, or less than 1 percent that are individuals were considered to lack face validity and were classified as \u201cunusable\u201d for the facility/group/individual code.
In addition to the state-reported provider taxonomy codes, TAF APR files produced in 2022 and later years also include a constructed taxonomy variable, the NPPES primary taxonomy code, that represents the provider\u2019s primary taxonomy information as reported in NPPES based on the provider\u2019s NPI.
The current provider taxonomy code list is available at www.wpc-edi.com/reference . The provider specialty, type, and authorized category of service code sets are available in the appendixes of the TAF APR data dictionary.
CMS. \u201cCMS Guidance: Best Practice for Reporting PROV\u2010CLASSIFICATION\u2010TYPE and PROV\u2010CLASSIFICATION\u2010CODE in the T\u2010MSIS Provider File.\u201d Baltimore, MD: CMS, n.d. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/47562 .
A valid ownership code (OWNRSHP_CD) is in the range \u201c01\u201d \u2013 \u201c19\u201d. A valid provider profit status code (PRVDR_PRFT_STUS_CD) is in the range \u201c01\u201d \u2013 \u201c04\u201d.
The TAF Annual Provider (APR) file contains detailed information about each provider authorized to deliver services to Medicaid and CHIP beneficiaries at any point during the calendar year. The facility/group/individual code indicates whether a record in the APR represents a facility, a group of practitioners, or an individual practitioner. This data quality assessment examines the extent to which the TAF APR records have a valid facility/group/individual code and the extent to which the proportion of providers that are facilities, groups, and individuals falls within the expected range.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""9071"", ""relatedTopics"": [{""measureId"": 95, ""measureName"": ""Group and Individual Providers - Classification Types"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 1}, {""measureId"": 97, ""measureName"": ""Facilities - Classification Types"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 2}, {""measureId"": 96, ""measureName"": ""Facility Characteristics"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 3}]}" 95,"{""measureId"": 95, ""measureName"": ""Group and Individual Providers - Classification Types"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Group-Indiv-Class-Type.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) Annual Provider File (APR) captures detailed information about each provider authorized by a state to provide services to its Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries, regardless of whether or how often the provider billed the state for services. [1] The APR file includes more detailed information about those providers rendering or billing for the service than the fee-for-service and encounter records in the TAF claims files, which contain limited information about them. For example, TAF fee-for-service and encounter records on the inpatient (IP), long-term care (LT), and other services (OT) files include information on each claim about the billing, servicing, and rendering provider’s taxonomy, specialty, and state-identified provider type. [2] The TAF APR, on the other hand, includes details about the characteristics, locations, taxonomies/classifications, affiliated groups, affiliated programs, licensing/accreditations, other identifiers associated with the provider, and—for facility providers—the bed types linked with the facility. [3]
Each record in the TAF APR represents a provider enrolled in the state’s Medicaid or CHIP program. A provider could be a facility, a group of providers, or an individual provider. In the TAF APR, a facility is defined as an organization, institution, place, building, or agency that furnishes, conducts, and operates health care services for the prevention, diagnosis, or treatment of human disease, pain, or injury. Examples include hospitals, nursing facilities, home health agencies, schools, or transportation organizations. A group is defined as two or more physicians, advanced practice nurses, and/or physician’s assistants who work together and share facilities. [4] An individual provider is defined as an individual who provides medical or non-medical services. [5] , [6]
States must report information about areas of specialization for all providers included on the TAF APR, using at least one of four possible classification types: (1) state-reported taxonomy code, (2) specialty code, (3) provider type code, and (4) authorized category of service code. In addition to the four classification types reported by the state, TAF APR files produced in 2022 and later years also include a constructed taxonomy variable, the NPPES primary taxonomy code, that represents the provider’s primary taxonomy information as reported in National Plan and Provider Enumeration System (NPPES) based on the provider’s National Provider Identifier (NPI).
Maintained by the National Uniform Claim Committee, taxonomy codes are standardized, unique 10-character codes that designate a provider’s classification or area of specialization. Providers select one or more taxonomy codes that most closely describe their classification or area of specialization and must submit the selected code when applying for an NPI. Provider specialty, provider type, and authorized category of service also designate a provider’s classification or area of specialization, but with less granularity than taxonomy code does. Analysis of various topics, such as the geographic distribution of specialists or primary care physicians, requires accurate information about provider specialization in the TAF.
States are encouraged to report all classification types for which information is available. At a minimum, states must report taxonomy codes for providers with an NPI and authorized categories of service for providers without an NPI. In addition, most states should have relatively complete data for provider specialty or taxonomy because state Medicaid programs typically set or modify provider payments based on this information. Depending on the state, providers submit their taxonomy or specialty when enrolling with the state Medicaid or CHIP program, include it on Medicaid claim submissions, or do a combination of the two.
Other variables in the APR provide information about provider characteristics that are particularly relevant for facility providers. For instance, ownership code denotes a provider’s ownership interest and/or managing control information, while provider profit status code indicates whether the provider is a 501(c)(3) non-profit, a for-profit either closely held or publicly traded, or has any other profit status.
This analysis examines the extent to which APR records have a valid facility/group/individual code and the extent to which the proportion of providers that are facilities, groups, and individuals falls within the expected range. The analysis also explores the extent to which TAF APR records representing group and individual providers—and, separately, those representing facility providers—have a valid, usable, and applicable value for at least one of the four provider classification types. Finally, the analysis examines the extent to which TAF APR records representing facility providers have a usable ownership code and provider profit status code.
TAF APR records present unique combinations of submitting state and provider. Because providers can participate with more than one state’s Medicaid program, more than one TAF APR record may represent a given provider.
In addition, fee-for-service and encounter records on the IP and LT files include information on each claim about the admitting provider’s taxonomy, specialty, and state-identified provider type, and the fee-for-service and encounter records on the prescription drug (RX) file include information on each claim about the billing provider’s taxonomy and specialty.
The TAF APR consists of nine files. The TAF APR base file includes basic provider characteristics. Eight additional supplemental files provide more detailed information on the following topics: the provider’s taxonomy, Medicaid and/or CHIP program enrollment, group affiliation, health plan and program affiliation, geographic location(s), licensing and accreditation, other provider identifiers, and bed type. Each record in the APR TAF base file represents a provider enrolled with the state’s Medicaid or CHIP program. Each provider record on the base file may link to more than one record on each supplemental file.
Physicians within a group may have different specialties.
Individual providers include two distinct groups. The first, incorporated sole practitioners, are sole practitioners with legal separation between themselves and their practice. The second, non-incorporated sole practitioners (also known as sole proprietors), are sole practitioners who operate with no legal distinction between themselves and their practice. The latter have characteristics associated with both individuals and organizations.
Centers for Medicare & Medicaid Services (CMS). “CMS Guidance: Reporting Provider Facility-Group-Individual-Code in T-MSIS.” Baltimore, MD: CMS, n.d. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/51240 .
TAF APR records present unique combinations of submitting state and provider. Because providers can participate with more than one state\u2019s Medicaid program, more than one TAF APR record may represent a given provider.
In addition, fee-for-service and encounter records on the IP and LT files include information on each claim about the admitting provider\u2019s taxonomy, specialty, and state-identified provider type, and the fee-for-service and encounter records on the prescription drug (RX) file include information on each claim about the billing provider\u2019s taxonomy and specialty.
The TAF APR consists of nine files. The TAF APR base file includes basic provider characteristics. Eight additional supplemental files provide more detailed information on the following topics: the provider\u2019s taxonomy, Medicaid and/or CHIP program enrollment, group affiliation, health plan and program affiliation, geographic location(s), licensing and accreditation, other provider identifiers, and bed type. Each record in the APR TAF base file represents a provider enrolled with the state\u2019s Medicaid or CHIP program. Each provider record on the base file may link to more than one record on each supplemental file.
Physicians within a group may have different specialties.
Individual providers include two distinct groups. The first, incorporated sole practitioners, are sole practitioners with legal separation between themselves and their practice. The second, non-incorporated sole practitioners (also known as sole proprietors), are sole practitioners who operate with no legal distinction between themselves and their practice. The latter have characteristics associated with both individuals and organizations.
Centers for Medicare & Medicaid Services (CMS). \u201cCMS Guidance: Reporting Provider Facility-Group-Individual-Code in T-MSIS.\u201d Baltimore, MD: CMS, n.d. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/51240 .
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
This analysis included all records in the APR base file, including those classified as actively enrolled, pending, denied, or terminated. [7] We linked information from the taxonomy supplemental file to the base file records for some analyses. We first examined the percentage of APR base file records that include a valid facility/group/individual code and the extent to which the share of providers in the state’s base file that are facilities or groups versus individuals falls within the expected range. [8]
To determine the expected range of providers that are facilities or groups versus individuals, we identified all providers in the NPPES database with an active NPI in the calendar year. We then identified the subset of these providers that were Medicaid or CHIP providers as those that linked, via NPI, to at least one TAF claim (fee-for-service or encounter record) in the calendar year. [9]
In each state, we calculated the percentage of these active NPPES providers linking to a TAF claim who are individuals (which aligns with the providers whom the TAF APR defines as individuals) and that are organizations (which aligns with the providers that the TAF APR defines as groups and facilities). We set the expected range for the percentage of each state’s APR records that represent individuals equal to the range of state-level percentages of active NPPES providers linking to a TAF claim that are individuals (70 percent to 93 percent). Likewise, we set the expected range for the percentage of each state’s APR records that represent facilities and groups equal to the range of state-level percentages of active NPPES providers linking to a TAF claim that are organizations (7 percent to 30 percent). We set tiers for percentages moderately above and below and far above and below the expected range for group and facility providers and individual providers, respectively (Table 1).
Table 1. Expected distribution of group/facility and individual providers in the TAF APR
| Classification | Percentage of APR records identified as group or facility providers | Percentage of APR records identified as individual providers |
|---|---|---|
| Far above expected range | x > 40 percent | x > 97 percent |
| Moderately above expected range | 30 percent < x ≤ 40 percent | 93 percent < x ≤ 97 percent |
| Within expected range | 7 percent < x ≤ 30 percent | 70 percent < x ≤ 93 percent |
| Moderately below expected range | 3 percent < x ≤ 7 percent | 60 percent < x ≤ 70 percent |
| Far below expected range | x ≤ 3 percent | x ≤ 60 percent |
We based the data quality assessment of the facility/group/individual code on three criteria: (1) the percentage of APR records with a valid facility/group/individual code, (2) the extent to which the reported percentage of individual providers falls within the expected range, and (3) the extent to which the reported percentage of facility and group providers falls within the expected range. [10] We assigned states to a low level of concern about data quality if all three specified criteria were true (Table 2). If any of the three criteria specified for the “low concern” category were not true, we assigned the state to the highest concern category for which at least one of the three criteria were true.
Table 2. Criteria for DQ assessment of facility/group/individual code
| Percent of APR records with a valid facility-group-individual code | Percentage of APR records identified as group or facility providers | Percentage of APR records identified as individual providers | DQ assessment |
|---|---|---|---|
| x ≥ 90 percent | Within expected range | Within expected range | Low concern a |
| 80 percent ≤ x < 90 percent | Moderately above or below expected range | Moderately above or below expected range | Medium concern b |
| 50 percent ≤ x < 80 percent | Far above or below expected range | Far above or below expected range | High concern b |
| x < 50 percent | x < 1 percent (facility) or x < 1 percent (group) | x < 1 percent | Unusable b |
a All three criteria must be true for a state to receive the given DQ Assessment.
b One of the three criteria must be true for a state to receive the given DQ Assessment.
For group and individual providers—and separately for facility providers—we examined the extent to which APR records have a valid, usable, and applicable value for at least one of the five provider classification fields (state-reported taxonomy, NPPES primary taxonomy, [11] provider specialty, provider type, or authorized category of service). For a record to count as having a valid, usable, and applicable value for a given classification type, it must meet all of the following criteria:
For each topic, we grouped states into categories of concern about the usability of their data, based on the total percentage of APR records that represent group and individual providers and facility providers, respectively, that have a valid, usable, and applicable value reported for at least one of the five classification types (Table 3). For context, we also calculated the percentage of APR records with valid, usable, and applicable codes for more than one classification type.
TAF APR users may want to conduct additional data quality assessments before using the provider specialty, type, and authorized category of service fields to classify providers. A state could possibly use an improper code set (for instance, the Medicare provider type code set instead of the TAF provider type code set) with valid values that overlap with the TAF provider specialty, type, and authorized category of service codes. [13] TAF APR users can screen for this issue by examining the most prevalent provider specialties, types, or authorized category of service codes for the provider category of interest (for instance, facilities) to ensure that the most commonly reported codes align with expectations.
Table 3. Criteria for DQ assessment of provider classification types
| Percentage of APR records with a valid, usable, and applicable value reported for at least one of the five classification types | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
For facility providers, we also examined the extent to which APR records include usable information for two variables that provide more information about provider characteristics: ownership code and provider profit status code. We grouped states into categories of concern about the usability of their data, based on the total percentage of APR records that represent facility providers with a valid value reported for both data elements (Table 4). [14]
Table 4. Criteria for DQ assessment of facility provider characteristics
| Percentage of APR records that represent facilities with a valid value reported for ownership code and provider profit status code | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
Methods previously used to assess data quality
Table 5 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. Current methods are used to assess all data years and versions not listed in the table, aside from data years and versions for which this data quality analysis is unavailable.
Table 5. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
|
|
This analysis drew on the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. More details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the “Production of the TAF Research Identifiable Files” guide.
A valid facility/group/individual code (FAC_GRP_INDVDL_CD) is one of the following values: “01” (facility), “02” (group), or “03” (individual).
We linked active NPIs in NPPES (those without a deactivation date prior to the start of the analysis year) to the NPIs on FFS claims and managed care encounters funded through Medicaid or CHIP (CLM_TYPE_CD = 1, 3, A, or C) on the monthly TAF claims files for the year of analysis, including the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims files. We excluded capitation payments, supplemental payments, service tracking payments, and records with a claim type of “other.” For Illinois, we only linked to the original version of the claim and did not link to subsequent adjustment records in the state’s TAF data. For more information, see the guide “How to Use Illinois Claims Data,” available on ResDAC.org. The TAF claims files contain numerous provider identifier variables that represent the roles providers play in beneficiaries’ health care. This analysis linked the NPIs in NPPES to the following NPIs on the TAF claims files: (1) the IP and LT billing (BLG_PRVDR_NPI_NUM), referring (RFRG_PRVDR_NPI_NUM), servicing (SRVCNG_PRVDR_NPI_NUM), and admitting (ADMTG_PRVDR_NPI_NUM) provider NPIs, (2) the OT billing (BLG_PRVDR_NPI_NUM), referring (RFRG_PRVDR_NPI_NUM), and servicing (SRVCNG_PRVDR_NPI_NUM) provider NPIs, and (3) the RX billing (BLG_PRVDR_NPI_NUM), prescribing (SRVCNG_PRVDR_NPI_NUM), and dispensing (DSPNSNG_PD_PRVDR_NPI_NUM) provider NPIs.
This criterion also includes an assessment of the extent to which the reported percentage of group providers and facility providers, evaluated separately, has face validity. States reporting less than 1 percent of providers that are groups, less than 1 percent that are facilities, or less than 1 percent that are individuals were considered to lack face validity and were classified as “unusable” for the facility/group/individual code.
In addition to the state-reported provider taxonomy codes, TAF APR files produced in 2022 and later years also include a constructed taxonomy variable, the NPPES primary taxonomy code, that represents the provider’s primary taxonomy information as reported in NPPES based on the provider’s NPI.
The current provider taxonomy code list is available at www.wpc-edi.com/reference . The provider specialty, type, and authorized category of service code sets are available in the appendixes of the TAF APR data dictionary.
CMS. “CMS Guidance: Best Practice for Reporting PROV‐CLASSIFICATION‐TYPE and PROV‐CLASSIFICATION‐CODE in the T‐MSIS Provider File.” Baltimore, MD: CMS, n.d. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/47562 .
A valid ownership code (OWNRSHP_CD) is in the range “01” – “19”. A valid provider profit status code (PRVDR_PRFT_STUS_CD) is in the range “01” – “04”.
This analysis drew on the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. More details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the \u201cProduction of the TAF Research Identifiable Files\u201d guide.
A valid facility/group/individual code (FAC_GRP_INDVDL_CD) is one of the following values: \u201c01\u201d (facility), \u201c02\u201d (group), or \u201c03\u201d (individual).
We linked active NPIs in NPPES (those without a deactivation date prior to the start of the analysis year) to the NPIs on FFS claims and managed care encounters funded through Medicaid or CHIP (CLM_TYPE_CD = 1, 3, A, or C) on the monthly TAF claims files for the year of analysis, including the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims files. We excluded capitation payments, supplemental payments, service tracking payments, and records with a claim type of \u201cother.\u201d For Illinois, we only linked to the original version of the claim and did not link to subsequent adjustment records in the state\u2019s TAF data. For more information, see the guide \u201cHow to Use Illinois Claims Data,\u201d available on ResDAC.org. The TAF claims files contain numerous provider identifier variables that represent the roles providers play in beneficiaries\u2019 health care. This analysis linked the NPIs in NPPES to the following NPIs on the TAF claims files: (1) the IP and LT billing (BLG_PRVDR_NPI_NUM), referring (RFRG_PRVDR_NPI_NUM), servicing (SRVCNG_PRVDR_NPI_NUM), and admitting (ADMTG_PRVDR_NPI_NUM) provider NPIs, (2) the OT billing (BLG_PRVDR_NPI_NUM), referring (RFRG_PRVDR_NPI_NUM), and servicing (SRVCNG_PRVDR_NPI_NUM) provider NPIs, and (3) the RX billing (BLG_PRVDR_NPI_NUM), prescribing (SRVCNG_PRVDR_NPI_NUM), and dispensing (DSPNSNG_PD_PRVDR_NPI_NUM) provider NPIs.
This criterion also includes an assessment of the extent to which the reported percentage of group providers and facility providers, evaluated separately, has face validity. States reporting less than 1 percent of providers that are groups, less than 1 percent that are facilities, or less than 1 percent that are individuals were considered to lack face validity and were classified as \u201cunusable\u201d for the facility/group/individual code.
In addition to the state-reported provider taxonomy codes, TAF APR files produced in 2022 and later years also include a constructed taxonomy variable, the NPPES primary taxonomy code, that represents the provider\u2019s primary taxonomy information as reported in NPPES based on the provider\u2019s NPI.
The current provider taxonomy code list is available at www.wpc-edi.com/reference . The provider specialty, type, and authorized category of service code sets are available in the appendixes of the TAF APR data dictionary.
CMS. \u201cCMS Guidance: Best Practice for Reporting PROV\u2010CLASSIFICATION\u2010TYPE and PROV\u2010CLASSIFICATION\u2010CODE in the T\u2010MSIS Provider File.\u201d Baltimore, MD: CMS, n.d. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/47562 .
A valid ownership code (OWNRSHP_CD) is in the range \u201c01\u201d \u2013 \u201c19\u201d. A valid provider profit status code (PRVDR_PRFT_STUS_CD) is in the range \u201c01\u201d \u2013 \u201c04\u201d.
The TAF Annual Provider (APR) file contains detailed information about each provider authorized to deliver services to Medicaid and CHIP beneficiaries at any point during the calendar year. States may categorize providers using four possible classification types in the TAF APR: taxonomy, provider specialty, provider type, and authorized category of service. This data quality assessment examines the extent to which the TAF APR records that represent group and individual providers have a valid, usable, and applicable value for at least one of the four classification types.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""9071"", ""relatedTopics"": [{""measureId"": 94, ""measureName"": ""Facility/Group/Individual Code"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 0}, {""measureId"": 97, ""measureName"": ""Facilities - Classification Types"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 2}, {""measureId"": 96, ""measureName"": ""Facility Characteristics"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 3}]}" 96,"{""measureId"": 96, ""measureName"": ""Facility Characteristics"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Facility-Characteristic.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) Annual Provider File (APR) captures detailed information about each provider authorized by a state to provide services to its Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries, regardless of whether or how often the provider billed the state for services. [1] The APR file includes more detailed information about those providers rendering or billing for the service than the fee-for-service and encounter records in the TAF claims files, which contain limited information about them. For example, TAF fee-for-service and encounter records on the inpatient (IP), long-term care (LT), and other services (OT) files include information on each claim about the billing, servicing, and rendering provider’s taxonomy, specialty, and state-identified provider type. [2] The TAF APR, on the other hand, includes details about the characteristics, locations, taxonomies/classifications, affiliated groups, affiliated programs, licensing/accreditations, other identifiers associated with the provider, and—for facility providers—the bed types linked with the facility. [3]
Each record in the TAF APR represents a provider enrolled in the state’s Medicaid or CHIP program. A provider could be a facility, a group of providers, or an individual provider. In the TAF APR, a facility is defined as an organization, institution, place, building, or agency that furnishes, conducts, and operates health care services for the prevention, diagnosis, or treatment of human disease, pain, or injury. Examples include hospitals, nursing facilities, home health agencies, schools, or transportation organizations. A group is defined as two or more physicians, advanced practice nurses, and/or physician’s assistants who work together and share facilities. [4] An individual provider is defined as an individual who provides medical or non-medical services. [5] , [6]
States must report information about areas of specialization for all providers included on the TAF APR, using at least one of four possible classification types: (1) state-reported taxonomy code, (2) specialty code, (3) provider type code, and (4) authorized category of service code. In addition to the four classification types reported by the state, TAF APR files produced in 2022 and later years also include a constructed taxonomy variable, the NPPES primary taxonomy code, that represents the provider’s primary taxonomy information as reported in National Plan and Provider Enumeration System (NPPES) based on the provider’s National Provider Identifier (NPI).
Maintained by the National Uniform Claim Committee, taxonomy codes are standardized, unique 10-character codes that designate a provider’s classification or area of specialization. Providers select one or more taxonomy codes that most closely describe their classification or area of specialization and must submit the selected code when applying for an NPI. Provider specialty, provider type, and authorized category of service also designate a provider’s classification or area of specialization, but with less granularity than taxonomy code does. Analysis of various topics, such as the geographic distribution of specialists or primary care physicians, requires accurate information about provider specialization in the TAF.
States are encouraged to report all classification types for which information is available. At a minimum, states must report taxonomy codes for providers with an NPI and authorized categories of service for providers without an NPI. In addition, most states should have relatively complete data for provider specialty or taxonomy because state Medicaid programs typically set or modify provider payments based on this information. Depending on the state, providers submit their taxonomy or specialty when enrolling with the state Medicaid or CHIP program, include it on Medicaid claim submissions, or do a combination of the two.
Other variables in the APR provide information about provider characteristics that are particularly relevant for facility providers. For instance, ownership code denotes a provider’s ownership interest and/or managing control information, while provider profit status code indicates whether the provider is a 501(c)(3) non-profit, a for-profit either closely held or publicly traded, or has any other profit status.
This analysis examines the extent to which APR records have a valid facility/group/individual code and the extent to which the proportion of providers that are facilities, groups, and individuals falls within the expected range. The analysis also explores the extent to which TAF APR records representing group and individual providers—and, separately, those representing facility providers—have a valid, usable, and applicable value for at least one of the four provider classification types. Finally, the analysis examines the extent to which TAF APR records representing facility providers have a usable ownership code and provider profit status code.
TAF APR records present unique combinations of submitting state and provider. Because providers can participate with more than one state’s Medicaid program, more than one TAF APR record may represent a given provider.
In addition, fee-for-service and encounter records on the IP and LT files include information on each claim about the admitting provider’s taxonomy, specialty, and state-identified provider type, and the fee-for-service and encounter records on the prescription drug (RX) file include information on each claim about the billing provider’s taxonomy and specialty.
The TAF APR consists of nine files. The TAF APR base file includes basic provider characteristics. Eight additional supplemental files provide more detailed information on the following topics: the provider’s taxonomy, Medicaid and/or CHIP program enrollment, group affiliation, health plan and program affiliation, geographic location(s), licensing and accreditation, other provider identifiers, and bed type. Each record in the APR TAF base file represents a provider enrolled with the state’s Medicaid or CHIP program. Each provider record on the base file may link to more than one record on each supplemental file.
Physicians within a group may have different specialties.
Individual providers include two distinct groups. The first, incorporated sole practitioners, are sole practitioners with legal separation between themselves and their practice. The second, non-incorporated sole practitioners (also known as sole proprietors), are sole practitioners who operate with no legal distinction between themselves and their practice. The latter have characteristics associated with both individuals and organizations.
Centers for Medicare & Medicaid Services (CMS). “CMS Guidance: Reporting Provider Facility-Group-Individual-Code in T-MSIS.” Baltimore, MD: CMS, n.d. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/51240 .
TAF APR records present unique combinations of submitting state and provider. Because providers can participate with more than one state\u2019s Medicaid program, more than one TAF APR record may represent a given provider.
In addition, fee-for-service and encounter records on the IP and LT files include information on each claim about the admitting provider\u2019s taxonomy, specialty, and state-identified provider type, and the fee-for-service and encounter records on the prescription drug (RX) file include information on each claim about the billing provider\u2019s taxonomy and specialty.
The TAF APR consists of nine files. The TAF APR base file includes basic provider characteristics. Eight additional supplemental files provide more detailed information on the following topics: the provider\u2019s taxonomy, Medicaid and/or CHIP program enrollment, group affiliation, health plan and program affiliation, geographic location(s), licensing and accreditation, other provider identifiers, and bed type. Each record in the APR TAF base file represents a provider enrolled with the state\u2019s Medicaid or CHIP program. Each provider record on the base file may link to more than one record on each supplemental file.
Physicians within a group may have different specialties.
Individual providers include two distinct groups. The first, incorporated sole practitioners, are sole practitioners with legal separation between themselves and their practice. The second, non-incorporated sole practitioners (also known as sole proprietors), are sole practitioners who operate with no legal distinction between themselves and their practice. The latter have characteristics associated with both individuals and organizations.
Centers for Medicare & Medicaid Services (CMS). \u201cCMS Guidance: Reporting Provider Facility-Group-Individual-Code in T-MSIS.\u201d Baltimore, MD: CMS, n.d. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/51240 .
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
This analysis included all records in the APR base file, including those classified as actively enrolled, pending, denied, or terminated. [7] We linked information from the taxonomy supplemental file to the base file records for some analyses. We first examined the percentage of APR base file records that include a valid facility/group/individual code and the extent to which the share of providers in the state’s base file that are facilities or groups versus individuals falls within the expected range. [8]
To determine the expected range of providers that are facilities or groups versus individuals, we identified all providers in the NPPES database with an active NPI in the calendar year. We then identified the subset of these providers that were Medicaid or CHIP providers as those that linked, via NPI, to at least one TAF claim (fee-for-service or encounter record) in the calendar year. [9]
In each state, we calculated the percentage of these active NPPES providers linking to a TAF claim who are individuals (which aligns with the providers whom the TAF APR defines as individuals) and that are organizations (which aligns with the providers that the TAF APR defines as groups and facilities). We set the expected range for the percentage of each state’s APR records that represent individuals equal to the range of state-level percentages of active NPPES providers linking to a TAF claim that are individuals (70 percent to 93 percent). Likewise, we set the expected range for the percentage of each state’s APR records that represent facilities and groups equal to the range of state-level percentages of active NPPES providers linking to a TAF claim that are organizations (7 percent to 30 percent). We set tiers for percentages moderately above and below and far above and below the expected range for group and facility providers and individual providers, respectively (Table 1).
Table 1. Expected distribution of group/facility and individual providers in the TAF APR
| Classification | Percentage of APR records identified as group or facility providers | Percentage of APR records identified as individual providers |
|---|---|---|
| Far above expected range | x > 40 percent | x > 97 percent |
| Moderately above expected range | 30 percent < x ≤ 40 percent | 93 percent < x ≤ 97 percent |
| Within expected range | 7 percent < x ≤ 30 percent | 70 percent < x ≤ 93 percent |
| Moderately below expected range | 3 percent < x ≤ 7 percent | 60 percent < x ≤ 70 percent |
| Far below expected range | x ≤ 3 percent | x ≤ 60 percent |
We based the data quality assessment of the facility/group/individual code on three criteria: (1) the percentage of APR records with a valid facility/group/individual code, (2) the extent to which the reported percentage of individual providers falls within the expected range, and (3) the extent to which the reported percentage of facility and group providers falls within the expected range. [10] We assigned states to a low level of concern about data quality if all three specified criteria were true (Table 2). If any of the three criteria specified for the “low concern” category were not true, we assigned the state to the highest concern category for which at least one of the three criteria were true.
Table 2. Criteria for DQ assessment of facility/group/individual code
| Percent of APR records with a valid facility-group-individual code | Percentage of APR records identified as group or facility providers | Percentage of APR records identified as individual providers | DQ assessment |
|---|---|---|---|
| x ≥ 90 percent | Within expected range | Within expected range | Low concern a |
| 80 percent ≤ x < 90 percent | Moderately above or below expected range | Moderately above or below expected range | Medium concern b |
| 50 percent ≤ x < 80 percent | Far above or below expected range | Far above or below expected range | High concern b |
| x < 50 percent | x < 1 percent (facility) or x < 1 percent (group) | x < 1 percent | Unusable b |
a All three criteria must be true for a state to receive the given DQ Assessment.
b One of the three criteria must be true for a state to receive the given DQ Assessment.
For group and individual providers—and separately for facility providers—we examined the extent to which APR records have a valid, usable, and applicable value for at least one of the five provider classification fields (state-reported taxonomy, NPPES primary taxonomy, [11] provider specialty, provider type, or authorized category of service). For a record to count as having a valid, usable, and applicable value for a given classification type, it must meet all of the following criteria:
For each topic, we grouped states into categories of concern about the usability of their data, based on the total percentage of APR records that represent group and individual providers and facility providers, respectively, that have a valid, usable, and applicable value reported for at least one of the five classification types (Table 3). For context, we also calculated the percentage of APR records with valid, usable, and applicable codes for more than one classification type.
TAF APR users may want to conduct additional data quality assessments before using the provider specialty, type, and authorized category of service fields to classify providers. A state could possibly use an improper code set (for instance, the Medicare provider type code set instead of the TAF provider type code set) with valid values that overlap with the TAF provider specialty, type, and authorized category of service codes. [13] TAF APR users can screen for this issue by examining the most prevalent provider specialties, types, or authorized category of service codes for the provider category of interest (for instance, facilities) to ensure that the most commonly reported codes align with expectations.
Table 3. Criteria for DQ assessment of provider classification types
| Percentage of APR records with a valid, usable, and applicable value reported for at least one of the five classification types | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
For facility providers, we also examined the extent to which APR records include usable information for two variables that provide more information about provider characteristics: ownership code and provider profit status code. We grouped states into categories of concern about the usability of their data, based on the total percentage of APR records that represent facility providers with a valid value reported for both data elements (Table 4). [14]
Table 4. Criteria for DQ assessment of facility provider characteristics
| Percentage of APR records that represent facilities with a valid value reported for ownership code and provider profit status code | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
Methods previously used to assess data quality
Table 5 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. Current methods are used to assess all data years and versions not listed in the table, aside from data years and versions for which this data quality analysis is unavailable.
Table 5. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
|
|
This analysis drew on the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. More details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the “Production of the TAF Research Identifiable Files” guide.
A valid facility/group/individual code (FAC_GRP_INDVDL_CD) is one of the following values: “01” (facility), “02” (group), or “03” (individual).
We linked active NPIs in NPPES (those without a deactivation date prior to the start of the analysis year) to the NPIs on FFS claims and managed care encounters funded through Medicaid or CHIP (CLM_TYPE_CD = 1, 3, A, or C) on the monthly TAF claims files for the year of analysis, including the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims files. We excluded capitation payments, supplemental payments, service tracking payments, and records with a claim type of “other.” For Illinois, we only linked to the original version of the claim and did not link to subsequent adjustment records in the state’s TAF data. For more information, see the guide “How to Use Illinois Claims Data,” available on ResDAC.org. The TAF claims files contain numerous provider identifier variables that represent the roles providers play in beneficiaries’ health care. This analysis linked the NPIs in NPPES to the following NPIs on the TAF claims files: (1) the IP and LT billing (BLG_PRVDR_NPI_NUM), referring (RFRG_PRVDR_NPI_NUM), servicing (SRVCNG_PRVDR_NPI_NUM), and admitting (ADMTG_PRVDR_NPI_NUM) provider NPIs, (2) the OT billing (BLG_PRVDR_NPI_NUM), referring (RFRG_PRVDR_NPI_NUM), and servicing (SRVCNG_PRVDR_NPI_NUM) provider NPIs, and (3) the RX billing (BLG_PRVDR_NPI_NUM), prescribing (SRVCNG_PRVDR_NPI_NUM), and dispensing (DSPNSNG_PD_PRVDR_NPI_NUM) provider NPIs.
This criterion also includes an assessment of the extent to which the reported percentage of group providers and facility providers, evaluated separately, has face validity. States reporting less than 1 percent of providers that are groups, less than 1 percent that are facilities, or less than 1 percent that are individuals were considered to lack face validity and were classified as “unusable” for the facility/group/individual code.
In addition to the state-reported provider taxonomy codes, TAF APR files produced in 2022 and later years also include a constructed taxonomy variable, the NPPES primary taxonomy code, that represents the provider’s primary taxonomy information as reported in NPPES based on the provider’s NPI.
The current provider taxonomy code list is available at www.wpc-edi.com/reference . The provider specialty, type, and authorized category of service code sets are available in the appendixes of the TAF APR data dictionary.
CMS. “CMS Guidance: Best Practice for Reporting PROV‐CLASSIFICATION‐TYPE and PROV‐CLASSIFICATION‐CODE in the T‐MSIS Provider File.” Baltimore, MD: CMS, n.d. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/47562 .
A valid ownership code (OWNRSHP_CD) is in the range “01” – “19”. A valid provider profit status code (PRVDR_PRFT_STUS_CD) is in the range “01” – “04”.
This analysis drew on the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. More details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the \u201cProduction of the TAF Research Identifiable Files\u201d guide.
A valid facility/group/individual code (FAC_GRP_INDVDL_CD) is one of the following values: \u201c01\u201d (facility), \u201c02\u201d (group), or \u201c03\u201d (individual).
We linked active NPIs in NPPES (those without a deactivation date prior to the start of the analysis year) to the NPIs on FFS claims and managed care encounters funded through Medicaid or CHIP (CLM_TYPE_CD = 1, 3, A, or C) on the monthly TAF claims files for the year of analysis, including the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims files. We excluded capitation payments, supplemental payments, service tracking payments, and records with a claim type of \u201cother.\u201d For Illinois, we only linked to the original version of the claim and did not link to subsequent adjustment records in the state\u2019s TAF data. For more information, see the guide \u201cHow to Use Illinois Claims Data,\u201d available on ResDAC.org. The TAF claims files contain numerous provider identifier variables that represent the roles providers play in beneficiaries\u2019 health care. This analysis linked the NPIs in NPPES to the following NPIs on the TAF claims files: (1) the IP and LT billing (BLG_PRVDR_NPI_NUM), referring (RFRG_PRVDR_NPI_NUM), servicing (SRVCNG_PRVDR_NPI_NUM), and admitting (ADMTG_PRVDR_NPI_NUM) provider NPIs, (2) the OT billing (BLG_PRVDR_NPI_NUM), referring (RFRG_PRVDR_NPI_NUM), and servicing (SRVCNG_PRVDR_NPI_NUM) provider NPIs, and (3) the RX billing (BLG_PRVDR_NPI_NUM), prescribing (SRVCNG_PRVDR_NPI_NUM), and dispensing (DSPNSNG_PD_PRVDR_NPI_NUM) provider NPIs.
This criterion also includes an assessment of the extent to which the reported percentage of group providers and facility providers, evaluated separately, has face validity. States reporting less than 1 percent of providers that are groups, less than 1 percent that are facilities, or less than 1 percent that are individuals were considered to lack face validity and were classified as \u201cunusable\u201d for the facility/group/individual code.
In addition to the state-reported provider taxonomy codes, TAF APR files produced in 2022 and later years also include a constructed taxonomy variable, the NPPES primary taxonomy code, that represents the provider\u2019s primary taxonomy information as reported in NPPES based on the provider\u2019s NPI.
The current provider taxonomy code list is available at www.wpc-edi.com/reference . The provider specialty, type, and authorized category of service code sets are available in the appendixes of the TAF APR data dictionary.
CMS. \u201cCMS Guidance: Best Practice for Reporting PROV\u2010CLASSIFICATION\u2010TYPE and PROV\u2010CLASSIFICATION\u2010CODE in the T\u2010MSIS Provider File.\u201d Baltimore, MD: CMS, n.d. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/47562 .
A valid ownership code (OWNRSHP_CD) is in the range \u201c01\u201d \u2013 \u201c19\u201d. A valid provider profit status code (PRVDR_PRFT_STUS_CD) is in the range \u201c01\u201d \u2013 \u201c04\u201d.
The TAF Annual Provider (APR) file contains detailed information about each provider authorized to deliver services to Medicaid and CHIP beneficiaries at any point during the calendar year. The TAF APR includes several data elements that offer more information about a provider's characteristics and are particularly relevant to facility providers, including ownership code and provider profit status code. This data quality assessment examines the extent to which the TAF APR records that represent facilities include usable information for ownership code and provider profit status code.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""9071"", ""relatedTopics"": [{""measureId"": 94, ""measureName"": ""Facility/Group/Individual Code"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 0}, {""measureId"": 95, ""measureName"": ""Group and Individual Providers - Classification Types"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 1}, {""measureId"": 97, ""measureName"": ""Facilities - Classification Types"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 2}]}" 97,"{""measureId"": 97, ""measureName"": ""Facilities - Classification Types"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Facility-Class-Types.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) Annual Provider File (APR) captures detailed information about each provider authorized by a state to provide services to its Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries, regardless of whether or how often the provider billed the state for services. [1] The APR file includes more detailed information about those providers rendering or billing for the service than the fee-for-service and encounter records in the TAF claims files, which contain limited information about them. For example, TAF fee-for-service and encounter records on the inpatient (IP), long-term care (LT), and other services (OT) files include information on each claim about the billing, servicing, and rendering provider’s taxonomy, specialty, and state-identified provider type. [2] The TAF APR, on the other hand, includes details about the characteristics, locations, taxonomies/classifications, affiliated groups, affiliated programs, licensing/accreditations, other identifiers associated with the provider, and—for facility providers—the bed types linked with the facility. [3]
Each record in the TAF APR represents a provider enrolled in the state’s Medicaid or CHIP program. A provider could be a facility, a group of providers, or an individual provider. In the TAF APR, a facility is defined as an organization, institution, place, building, or agency that furnishes, conducts, and operates health care services for the prevention, diagnosis, or treatment of human disease, pain, or injury. Examples include hospitals, nursing facilities, home health agencies, schools, or transportation organizations. A group is defined as two or more physicians, advanced practice nurses, and/or physician’s assistants who work together and share facilities. [4] An individual provider is defined as an individual who provides medical or non-medical services. [5] , [6]
States must report information about areas of specialization for all providers included on the TAF APR, using at least one of four possible classification types: (1) state-reported taxonomy code, (2) specialty code, (3) provider type code, and (4) authorized category of service code. In addition to the four classification types reported by the state, TAF APR files produced in 2022 and later years also include a constructed taxonomy variable, the NPPES primary taxonomy code, that represents the provider’s primary taxonomy information as reported in National Plan and Provider Enumeration System (NPPES) based on the provider’s National Provider Identifier (NPI).
Maintained by the National Uniform Claim Committee, taxonomy codes are standardized, unique 10-character codes that designate a provider’s classification or area of specialization. Providers select one or more taxonomy codes that most closely describe their classification or area of specialization and must submit the selected code when applying for an NPI. Provider specialty, provider type, and authorized category of service also designate a provider’s classification or area of specialization, but with less granularity than taxonomy code does. Analysis of various topics, such as the geographic distribution of specialists or primary care physicians, requires accurate information about provider specialization in the TAF.
States are encouraged to report all classification types for which information is available. At a minimum, states must report taxonomy codes for providers with an NPI and authorized categories of service for providers without an NPI. In addition, most states should have relatively complete data for provider specialty or taxonomy because state Medicaid programs typically set or modify provider payments based on this information. Depending on the state, providers submit their taxonomy or specialty when enrolling with the state Medicaid or CHIP program, include it on Medicaid claim submissions, or do a combination of the two.
Other variables in the APR provide information about provider characteristics that are particularly relevant for facility providers. For instance, ownership code denotes a provider’s ownership interest and/or managing control information, while provider profit status code indicates whether the provider is a 501(c)(3) non-profit, a for-profit either closely held or publicly traded, or has any other profit status.
This analysis examines the extent to which APR records have a valid facility/group/individual code and the extent to which the proportion of providers that are facilities, groups, and individuals falls within the expected range. The analysis also explores the extent to which TAF APR records representing group and individual providers—and, separately, those representing facility providers—have a valid, usable, and applicable value for at least one of the four provider classification types. Finally, the analysis examines the extent to which TAF APR records representing facility providers have a usable ownership code and provider profit status code.
TAF APR records present unique combinations of submitting state and provider. Because providers can participate with more than one state’s Medicaid program, more than one TAF APR record may represent a given provider.
In addition, fee-for-service and encounter records on the IP and LT files include information on each claim about the admitting provider’s taxonomy, specialty, and state-identified provider type, and the fee-for-service and encounter records on the prescription drug (RX) file include information on each claim about the billing provider’s taxonomy and specialty.
The TAF APR consists of nine files. The TAF APR base file includes basic provider characteristics. Eight additional supplemental files provide more detailed information on the following topics: the provider’s taxonomy, Medicaid and/or CHIP program enrollment, group affiliation, health plan and program affiliation, geographic location(s), licensing and accreditation, other provider identifiers, and bed type. Each record in the APR TAF base file represents a provider enrolled with the state’s Medicaid or CHIP program. Each provider record on the base file may link to more than one record on each supplemental file.
Physicians within a group may have different specialties.
Individual providers include two distinct groups. The first, incorporated sole practitioners, are sole practitioners with legal separation between themselves and their practice. The second, non-incorporated sole practitioners (also known as sole proprietors), are sole practitioners who operate with no legal distinction between themselves and their practice. The latter have characteristics associated with both individuals and organizations.
Centers for Medicare & Medicaid Services (CMS). “CMS Guidance: Reporting Provider Facility-Group-Individual-Code in T-MSIS.” Baltimore, MD: CMS, n.d. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/51240 .
TAF APR records present unique combinations of submitting state and provider. Because providers can participate with more than one state\u2019s Medicaid program, more than one TAF APR record may represent a given provider.
In addition, fee-for-service and encounter records on the IP and LT files include information on each claim about the admitting provider\u2019s taxonomy, specialty, and state-identified provider type, and the fee-for-service and encounter records on the prescription drug (RX) file include information on each claim about the billing provider\u2019s taxonomy and specialty.
The TAF APR consists of nine files. The TAF APR base file includes basic provider characteristics. Eight additional supplemental files provide more detailed information on the following topics: the provider\u2019s taxonomy, Medicaid and/or CHIP program enrollment, group affiliation, health plan and program affiliation, geographic location(s), licensing and accreditation, other provider identifiers, and bed type. Each record in the APR TAF base file represents a provider enrolled with the state\u2019s Medicaid or CHIP program. Each provider record on the base file may link to more than one record on each supplemental file.
Physicians within a group may have different specialties.
Individual providers include two distinct groups. The first, incorporated sole practitioners, are sole practitioners with legal separation between themselves and their practice. The second, non-incorporated sole practitioners (also known as sole proprietors), are sole practitioners who operate with no legal distinction between themselves and their practice. The latter have characteristics associated with both individuals and organizations.
Centers for Medicare & Medicaid Services (CMS). \u201cCMS Guidance: Reporting Provider Facility-Group-Individual-Code in T-MSIS.\u201d Baltimore, MD: CMS, n.d. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/51240 .
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
This analysis included all records in the APR base file, including those classified as actively enrolled, pending, denied, or terminated. [7] We linked information from the taxonomy supplemental file to the base file records for some analyses. We first examined the percentage of APR base file records that include a valid facility/group/individual code and the extent to which the share of providers in the state’s base file that are facilities or groups versus individuals falls within the expected range. [8]
To determine the expected range of providers that are facilities or groups versus individuals, we identified all providers in the NPPES database with an active NPI in the calendar year. We then identified the subset of these providers that were Medicaid or CHIP providers as those that linked, via NPI, to at least one TAF claim (fee-for-service or encounter record) in the calendar year. [9]
In each state, we calculated the percentage of these active NPPES providers linking to a TAF claim who are individuals (which aligns with the providers whom the TAF APR defines as individuals) and that are organizations (which aligns with the providers that the TAF APR defines as groups and facilities). We set the expected range for the percentage of each state’s APR records that represent individuals equal to the range of state-level percentages of active NPPES providers linking to a TAF claim that are individuals (70 percent to 93 percent). Likewise, we set the expected range for the percentage of each state’s APR records that represent facilities and groups equal to the range of state-level percentages of active NPPES providers linking to a TAF claim that are organizations (7 percent to 30 percent). We set tiers for percentages moderately above and below and far above and below the expected range for group and facility providers and individual providers, respectively (Table 1).
Table 1. Expected distribution of group/facility and individual providers in the TAF APR
| Classification | Percentage of APR records identified as group or facility providers | Percentage of APR records identified as individual providers |
|---|---|---|
| Far above expected range | x > 40 percent | x > 97 percent |
| Moderately above expected range | 30 percent < x ≤ 40 percent | 93 percent < x ≤ 97 percent |
| Within expected range | 7 percent < x ≤ 30 percent | 70 percent < x ≤ 93 percent |
| Moderately below expected range | 3 percent < x ≤ 7 percent | 60 percent < x ≤ 70 percent |
| Far below expected range | x ≤ 3 percent | x ≤ 60 percent |
We based the data quality assessment of the facility/group/individual code on three criteria: (1) the percentage of APR records with a valid facility/group/individual code, (2) the extent to which the reported percentage of individual providers falls within the expected range, and (3) the extent to which the reported percentage of facility and group providers falls within the expected range. [10] We assigned states to a low level of concern about data quality if all three specified criteria were true (Table 2). If any of the three criteria specified for the “low concern” category were not true, we assigned the state to the highest concern category for which at least one of the three criteria were true.
Table 2. Criteria for DQ assessment of facility/group/individual code
| Percent of APR records with a valid facility-group-individual code | Percentage of APR records identified as group or facility providers | Percentage of APR records identified as individual providers | DQ assessment |
|---|---|---|---|
| x ≥ 90 percent | Within expected range | Within expected range | Low concern a |
| 80 percent ≤ x < 90 percent | Moderately above or below expected range | Moderately above or below expected range | Medium concern b |
| 50 percent ≤ x < 80 percent | Far above or below expected range | Far above or below expected range | High concern b |
| x < 50 percent | x < 1 percent (facility) or x < 1 percent (group) | x < 1 percent | Unusable b |
a All three criteria must be true for a state to receive the given DQ Assessment.
b One of the three criteria must be true for a state to receive the given DQ Assessment.
For group and individual providers—and separately for facility providers—we examined the extent to which APR records have a valid, usable, and applicable value for at least one of the five provider classification fields (state-reported taxonomy, NPPES primary taxonomy, [11] provider specialty, provider type, or authorized category of service). For a record to count as having a valid, usable, and applicable value for a given classification type, it must meet all of the following criteria:
For each topic, we grouped states into categories of concern about the usability of their data, based on the total percentage of APR records that represent group and individual providers and facility providers, respectively, that have a valid, usable, and applicable value reported for at least one of the five classification types (Table 3). For context, we also calculated the percentage of APR records with valid, usable, and applicable codes for more than one classification type.
TAF APR users may want to conduct additional data quality assessments before using the provider specialty, type, and authorized category of service fields to classify providers. A state could possibly use an improper code set (for instance, the Medicare provider type code set instead of the TAF provider type code set) with valid values that overlap with the TAF provider specialty, type, and authorized category of service codes. [13] TAF APR users can screen for this issue by examining the most prevalent provider specialties, types, or authorized category of service codes for the provider category of interest (for instance, facilities) to ensure that the most commonly reported codes align with expectations.
Table 3. Criteria for DQ assessment of provider classification types
| Percentage of APR records with a valid, usable, and applicable value reported for at least one of the five classification types | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
For facility providers, we also examined the extent to which APR records include usable information for two variables that provide more information about provider characteristics: ownership code and provider profit status code. We grouped states into categories of concern about the usability of their data, based on the total percentage of APR records that represent facility providers with a valid value reported for both data elements (Table 4). [14]
Table 4. Criteria for DQ assessment of facility provider characteristics
| Percentage of APR records that represent facilities with a valid value reported for ownership code and provider profit status code | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
Methods previously used to assess data quality
Table 5 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. Current methods are used to assess all data years and versions not listed in the table, aside from data years and versions for which this data quality analysis is unavailable.
Table 5. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
|
|
This analysis drew on the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. More details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the “Production of the TAF Research Identifiable Files” guide.
A valid facility/group/individual code (FAC_GRP_INDVDL_CD) is one of the following values: “01” (facility), “02” (group), or “03” (individual).
We linked active NPIs in NPPES (those without a deactivation date prior to the start of the analysis year) to the NPIs on FFS claims and managed care encounters funded through Medicaid or CHIP (CLM_TYPE_CD = 1, 3, A, or C) on the monthly TAF claims files for the year of analysis, including the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims files. We excluded capitation payments, supplemental payments, service tracking payments, and records with a claim type of “other.” For Illinois, we only linked to the original version of the claim and did not link to subsequent adjustment records in the state’s TAF data. For more information, see the guide “How to Use Illinois Claims Data,” available on ResDAC.org. The TAF claims files contain numerous provider identifier variables that represent the roles providers play in beneficiaries’ health care. This analysis linked the NPIs in NPPES to the following NPIs on the TAF claims files: (1) the IP and LT billing (BLG_PRVDR_NPI_NUM), referring (RFRG_PRVDR_NPI_NUM), servicing (SRVCNG_PRVDR_NPI_NUM), and admitting (ADMTG_PRVDR_NPI_NUM) provider NPIs, (2) the OT billing (BLG_PRVDR_NPI_NUM), referring (RFRG_PRVDR_NPI_NUM), and servicing (SRVCNG_PRVDR_NPI_NUM) provider NPIs, and (3) the RX billing (BLG_PRVDR_NPI_NUM), prescribing (SRVCNG_PRVDR_NPI_NUM), and dispensing (DSPNSNG_PD_PRVDR_NPI_NUM) provider NPIs.
This criterion also includes an assessment of the extent to which the reported percentage of group providers and facility providers, evaluated separately, has face validity. States reporting less than 1 percent of providers that are groups, less than 1 percent that are facilities, or less than 1 percent that are individuals were considered to lack face validity and were classified as “unusable” for the facility/group/individual code.
In addition to the state-reported provider taxonomy codes, TAF APR files produced in 2022 and later years also include a constructed taxonomy variable, the NPPES primary taxonomy code, that represents the provider’s primary taxonomy information as reported in NPPES based on the provider’s NPI.
The current provider taxonomy code list is available at www.wpc-edi.com/reference . The provider specialty, type, and authorized category of service code sets are available in the appendixes of the TAF APR data dictionary.
CMS. “CMS Guidance: Best Practice for Reporting PROV‐CLASSIFICATION‐TYPE and PROV‐CLASSIFICATION‐CODE in the T‐MSIS Provider File.” Baltimore, MD: CMS, n.d. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/47562 .
A valid ownership code (OWNRSHP_CD) is in the range “01” – “19”. A valid provider profit status code (PRVDR_PRFT_STUS_CD) is in the range “01” – “04”.
This analysis drew on the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. More details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the \u201cProduction of the TAF Research Identifiable Files\u201d guide.
A valid facility/group/individual code (FAC_GRP_INDVDL_CD) is one of the following values: \u201c01\u201d (facility), \u201c02\u201d (group), or \u201c03\u201d (individual).
We linked active NPIs in NPPES (those without a deactivation date prior to the start of the analysis year) to the NPIs on FFS claims and managed care encounters funded through Medicaid or CHIP (CLM_TYPE_CD = 1, 3, A, or C) on the monthly TAF claims files for the year of analysis, including the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims files. We excluded capitation payments, supplemental payments, service tracking payments, and records with a claim type of \u201cother.\u201d For Illinois, we only linked to the original version of the claim and did not link to subsequent adjustment records in the state\u2019s TAF data. For more information, see the guide \u201cHow to Use Illinois Claims Data,\u201d available on ResDAC.org. The TAF claims files contain numerous provider identifier variables that represent the roles providers play in beneficiaries\u2019 health care. This analysis linked the NPIs in NPPES to the following NPIs on the TAF claims files: (1) the IP and LT billing (BLG_PRVDR_NPI_NUM), referring (RFRG_PRVDR_NPI_NUM), servicing (SRVCNG_PRVDR_NPI_NUM), and admitting (ADMTG_PRVDR_NPI_NUM) provider NPIs, (2) the OT billing (BLG_PRVDR_NPI_NUM), referring (RFRG_PRVDR_NPI_NUM), and servicing (SRVCNG_PRVDR_NPI_NUM) provider NPIs, and (3) the RX billing (BLG_PRVDR_NPI_NUM), prescribing (SRVCNG_PRVDR_NPI_NUM), and dispensing (DSPNSNG_PD_PRVDR_NPI_NUM) provider NPIs.
This criterion also includes an assessment of the extent to which the reported percentage of group providers and facility providers, evaluated separately, has face validity. States reporting less than 1 percent of providers that are groups, less than 1 percent that are facilities, or less than 1 percent that are individuals were considered to lack face validity and were classified as \u201cunusable\u201d for the facility/group/individual code.
In addition to the state-reported provider taxonomy codes, TAF APR files produced in 2022 and later years also include a constructed taxonomy variable, the NPPES primary taxonomy code, that represents the provider\u2019s primary taxonomy information as reported in NPPES based on the provider\u2019s NPI.
The current provider taxonomy code list is available at www.wpc-edi.com/reference . The provider specialty, type, and authorized category of service code sets are available in the appendixes of the TAF APR data dictionary.
CMS. \u201cCMS Guidance: Best Practice for Reporting PROV\u2010CLASSIFICATION\u2010TYPE and PROV\u2010CLASSIFICATION\u2010CODE in the T\u2010MSIS Provider File.\u201d Baltimore, MD: CMS, n.d. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/entry/47562 .
A valid ownership code (OWNRSHP_CD) is in the range \u201c01\u201d \u2013 \u201c19\u201d. A valid provider profit status code (PRVDR_PRFT_STUS_CD) is in the range \u201c01\u201d \u2013 \u201c04\u201d.
The TAF Annual Provider (APR) file contains detailed information about each provider authorized to deliver services to Medicaid and CHIP beneficiaries at any point during the calendar year. States may categorize providers using four possible classification types in the TAF APR: taxonomy, provider specialty, provider type, and authorized category of service. This data quality assessment examines the extent to which the TAF APR records that represent facilities have a valid, usable, and applicable value for at least one of the four classification types.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""9071"", ""relatedTopics"": [{""measureId"": 94, ""measureName"": ""Facility/Group/Individual Code"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 0}, {""measureId"": 95, ""measureName"": ""Group and Individual Providers - Classification Types"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 1}, {""measureId"": 96, ""measureName"": ""Facility Characteristics"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 3}]}" 98,"{""measureId"": 98, ""measureName"": ""Managed Care Plan Program and Population Characteristics"", ""groupId"": 10, ""groupName"": ""Managed Care Plans"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-MC-Plan-Prgm-Pop.pdf"", ""background"": {""content"": ""Managed care plays a significant role in the delivery of care to Medicaid beneficiaries nationwide. Nearly all states and territories contract with managed care organizations (MCOs) to provide care to at least some of their beneficiaries. All states and territories contracting with one or more Medicaid managed care plans must report into T-MSIS (1) information about the managed care plans in which beneficiaries are enrolled each month, (2) service utilization (encounter) records and capitation payments, and (3) information on the characteristics of each managed care plan operating in the state. T-MSIS Analytic Files (TAF) users can find beneficiary-level plan enrollment information in the Demographics and Eligibility (DE) file; encounter records and capitation payments in the inpatient (IP), long-term care (LT), other services (OT), and prescription drug (RX) claims files; and managed care plan-level information in the Annual Managed Care Plan (APL) file.
Although the TAF DE file includes limited information about each managed care plan in which a beneficiary is enrolled (including plan ID and type), the APL file includes additional information about the characteristics, locations, enrolled populations, and service areas for all Medicaid and Children’s Health Insurance Program (CHIP) health plans and managed care entities. [1] , [2] The APL file also includes additional identifying information, such as plan name and plan type, that TAF users need if they want to identify a specific managed care plan in the TAF data. However, the completeness of data elements in the APL may vary across states. In addition, some states may report plan names that are not unique across plan IDs or APL records. [3]
This analysis examines the extent to which the APL records include missing information for select data elements that identify program and population-related characteristics of these managed care entities. Separately, it examines the extent to which the APL records include missing information for select data elements that identify operational characteristics of these managed care entities. Finally, it examines the extent to which APL records are missing a plan name or plan type. For context, the analysis also shows the percentage of records with a non-unique plan name or plan ID and the distribution of plan types represented in the APL records.
The APL consists of five files. The APL base file includes basic plan-level information, whereas four additional supplemental files provide more detailed information on: plan geographic locations, service areas, populations enrolled, and operating authority. Each record in the APL base file represents a managed care plan or entity; in some cases, states may report multiple records per plan or entity—for example, to distinguish between the sub-populations enrolled. Each record in the APL base file may link to more than one record in each supplemental file.
States associate a plan identifier (plan ID) with each managed care entity or program operating in the state. The APL file is organized by plan ID.
For example, some states assign a single ID to a managed care plan and then use the managed care location number to identify the plan’s various locations throughout the state. However, other states assign a unique ID to each location for the same plan and then do not assign a location number or use the plan ID as the managed care location number.
The APL consists of five files. The APL base file includes basic plan-level information, whereas four additional supplemental files provide more detailed information on: plan geographic locations, service areas, populations enrolled, and operating authority. Each record in the APL base file represents a managed care plan or entity; in some cases, states may report multiple records per plan or entity\u2014for example, to distinguish between the sub-populations enrolled. Each record in the APL base file may link to more than one record in each supplemental file.
States associate a plan identifier (plan ID) with each managed care entity or program operating in the state. The APL file is organized by plan ID.
For example, some states assign a single ID to a managed care plan and then use the managed care location number to identify the plan\u2019s various locations throughout the state. However, other states assign a unique ID to each location for the same plan and then do not assign a location number or use the plan ID as the managed care location number.
This analysis uses all records in the TAF APL, including those representing both active and non-active plans. [4] , [5] We included in the analysis only states classified as using managed care in the Medicaid Managed Care Enrollment and Program Characteristics (MMCEPC) report for the calendar year. We did not evaluate the small number of states with no managed care program in the MMCEPC, even if these states had records present in the APL file. Additionally, if a state’s Medicaid program did not contract with any managed care plans during the year or a state’s Medicaid program contracted with one or more managed care plans during the year but the state had no records in the APL, we described it as unclassified.
We calculated the percentage of APL records with missing values across several data elements that convey information about program, population, and operational characteristics of the managed care entities and identifying information (Table 1). [6]
Table 1. TAF elements assessed
| TAF data element | Element description | Valid values |
|---|---|---|
| Managed care program code (MC_PGM_CD) | The state program through which a managed care plan is approved to operate | 1: Medicaid State Plan 2: CHIP State Plan 3: Both Medicaid and CHIP |
| Reimbursement arrangement category (REIMBRSMT_ARNGMT_CAT) | Aggregated categories of managed care entity reimbursement arrangement values | 1: Risk-based capitation 2: Non-risk capitation 3: Fee-for-service 4: Primary care case management (PCCM) 5: Other |
| Eligible population variables (POP_*) | Flag indicating whether the managed care entity is authorized to enroll a given eligibility group; 11 instances for 11 specific eligibility group(s) | 0: No 1: Yes |
| Operating authority variables (OPRTG_AUTHRTY_*) | Flag indicating the operating authority for the plan, when applicable; 15 instances for up to 15 potential operating authorities for a given managed care entity | 0: No 1: Yes |
| Profit status (MC_PRFT_STUS_CD) | The profit status of the managed care entity | 01: 501(C)(3) non-profit 02: For-profit, closely held 03: For-profit, publicly traded 04: Other |
| Managed care plan name (MC_NAME) | The name of the managed care entity under contract with the state Medicaid agency | The managed care name reported by the managed care entity |
| Managed care plan ID (MC_PLAN_ID) | The ID number the state issued to the managed care entity | State-assigned unique managed care identification number |
| Managed care plan type (MC_PLAN_TYPE_CD) | The type of managed care plan that corresponds to MC_PLAN_ID | 01: Comprehensive MCO 02: Traditional PCCM provider arrangement 03: Enhanced PCCM provider arrangement 04: Health insuring organization (HIO) 05: Medical-only prepaid inpatient health plan (PIHP) (risk or non-risk/non-comprehensive/with inpatient hospital or institutional services) 06: Medical-only prepaid ambulatory health plan (PAHP) (risk or non-risk/non-comprehensive/no inpatient hospital or institutional services) 07: Long-term care services and supports (LTSS) PIHP 08: Mental health (MH) PIHP 09: MH PAHP 10: Substance use disorders (SUD) PIHP 11: SUD PAHP 12: MH and SUD PIHP 13: MH and SUD PAHP 14: Dental PAHP 15: Transportation PAHP 16: Disease management PAHP 17: Program of All-Inclusive Care for the Elderly (PACE) 18: Pharmacy PAHP 19: Individual is enrolled in LTSS and MH PIHP 20: Other 60: Accountable care organization (ACO) 70: Health/medical home 80: Integrated care for dual eligibles |
We conducted three data quality assessments. The first assessment captured the percentage of APL records with a missing value for at least one of the data elements that identify program and population-related characteristics of the managed care entity: managed care program code and eligible population identifiers. [7] The second assessment captured the percentage of APL records with a missing value for at least one of the data elements that identify operational characteristics of the managed care entity: reimbursement arrangement category, operating authority identifiers, and profit status. [8] For each of these assessments, we grouped states into levels of concern about the usability of their data based on this percentage (Tables 2 and 3).
Table 2. Criteria for DQ assessment of managed care plan program and population characteristics
| Percentage of APL records with a missing value for managed care program code or eligible population identifiers | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
Table 3. Criteria for DQ assessment of managed care plan operational characteristics
| Percentage of APL records with a missing value for reimbursement arrangement category, operating authority identifiers, or profit status | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
The third assessment captured the percentage of APL records with a missing value for either plan name or plan type. We grouped states into levels of concern about the usability of their data based on this percentage (Table 4).
Table 4. Criteria for DQ assessment of plan name and type
| Percentage of APL records with a missing value for plan name or plan type | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
To provide users with additional information about how plans are represented in each state’s APL file, we present the percentage of APL records that had non-unique plan names. [9] This information is available in the table, but we did not use it in the data quality assessment for plan name because the T-MSIS data reporting requirements allow multiple plan IDs for a single plan name.
We also categorized and reported the distribution of managed care plan types. Table 5 describes the plan type categories used for reporting the distribution of APL records across plan type in each state.
Table 5. Plan types categories
| Plan type category | TAF plan type codes (MC_PLAN_TYPE_CD) |
|---|---|
| Comprehensive managed care (CMC) | 01: Comprehensive MCO 04: HIO 80: Integrated care for dually eligible individuals |
| Primary care case management (PCCM) | 02: Traditional PCCM provider arrangement 03: Enhanced PCCM provider arrangement |
| Managed long-term care services and supports (MLTSS)-only | 07: LTSS PIHP 19: Individual is enrolled in LTSS and MH PIHP |
| Behavioral health organization (BHO) | 08: MH PIHP 09: MH PAHP 10: SUD PIHP 11: SUD PAHP 12: MH and SUD PIHP 13: MH and SUD PAHP |
| Dental only | 14: Dental PAHP |
| Non-emergency medical transportation | 15: Transportation PAHP |
| Programs of All-Inclusive Care for the Elderly (PACE) | 17: PACE |
| Accountable care organizations (ACO) | 60: ACO |
| Other health plans | 05: Medical-only PIHP (non-comprehensive, with inpatient hospital or institutional services) 06: Medical-only PAHP (non-comprehensive; no inpatient hospital or institutional services) 16: Disease management PAHP 18: Pharmacy PAHP 20: Other 70: Health/medical home |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
States should include only those managed care plans that were active in the relevant calendar year in their APL reporting. An active managed care plan record in T-MSIS has (a) an effective start date and end date that overlap with at least part of the calendar year, and (b) an indicator identifying the record as active if multiple records represent the same time period. However, states sometimes submit active managed care plan records that in fact were inactive because the plan’s contract was not in effect during the calendar year. If a state reports these inactive plans as being active in the calendar year, they are included in the APL file.
During the TAF production process any invalid values submitted by the state are recoded to null (missing). As a result, missing values in TAF may actually represent either missing or invalid T-MSIS data submitted by states.
The eligible population identifiers were considered missing if there was not at least one eligible population indicated.
The operating authority identifiers were considered missing if there was not at least one operating authority indicated.
To assess non-uniqueness for plan names, we started by processing the names—removing excess spacing, punctuation, and other items that could complicate the matching process. We then determined how many records shared a plan name with one or more other records.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
States should include only those managed care plans that were active in the relevant calendar year in their APL reporting. An active managed care plan record in T-MSIS has (a) an effective start date and end date that overlap with at least part of the calendar year, and (b) an indicator identifying the record as active if multiple records represent the same time period. However, states sometimes submit active managed care plan records that in fact were inactive because the plan\u2019s contract was not in effect during the calendar year. If a state reports these inactive plans as being active in the calendar year, they are included in the APL file.
During the TAF production process any invalid values submitted by the state are recoded to null (missing). As a result, missing values in TAF may actually represent either missing or invalid T-MSIS data submitted by states.
The eligible population identifiers were considered missing if there was not at least one eligible population indicated.
The operating authority identifiers were considered missing if there was not at least one operating authority indicated.
To assess non-uniqueness for plan names, we started by processing the names\u2014removing excess spacing, punctuation, and other items that could complicate the matching process. We then determined how many records shared a plan name with one or more other records.
The TAF Annual Managed Care Plan (APL) file contains plan-level information, including characteristics, locations, enrolled populations, and service areas for all Medicaid and Children's Health Insurance Program (CHIP) health plans and managed care entities. This analysis examines the extent to which the APL records include missing information for selected data elements that identify program and population-related characteristics of these managed care entities: whether the plan serves Medicaid, CHIP, or both programs and the populations eligible to enroll.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""8051"", ""relatedTopics"": [{""measureId"": 99, ""measureName"": ""Managed Care Plan Operational Characteristics"", ""groupId"": 10, ""groupName"": ""Managed Care Plans"", ""order"": 1}, {""measureId"": 100, ""measureName"": ""Plan Name and Type"", ""groupId"": 10, ""groupName"": ""Managed Care Plans"", ""order"": 2}]}" 99,"{""measureId"": 99, ""measureName"": ""Managed Care Plan Operational Characteristics"", ""groupId"": 10, ""groupName"": ""Managed Care Plans"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-MC-Plan-Operational.pdf"", ""background"": {""content"": ""Managed care plays a significant role in the delivery of care to Medicaid beneficiaries nationwide. Nearly all states and territories contract with managed care organizations (MCOs) to provide care to at least some of their beneficiaries. All states and territories contracting with one or more Medicaid managed care plans must report into T-MSIS (1) information about the managed care plans in which beneficiaries are enrolled each month, (2) service utilization (encounter) records and capitation payments, and (3) information on the characteristics of each managed care plan operating in the state. T-MSIS Analytic Files (TAF) users can find beneficiary-level plan enrollment information in the Demographics and Eligibility (DE) file; encounter records and capitation payments in the inpatient (IP), long-term care (LT), other services (OT), and prescription drug (RX) claims files; and managed care plan-level information in the Annual Managed Care Plan (APL) file.
Although the TAF DE file includes limited information about each managed care plan in which a beneficiary is enrolled (including plan ID and type), the APL file includes additional information about the characteristics, locations, enrolled populations, and service areas for all Medicaid and Children’s Health Insurance Program (CHIP) health plans and managed care entities. [1] , [2] The APL file also includes additional identifying information, such as plan name and plan type, that TAF users need if they want to identify a specific managed care plan in the TAF data. However, the completeness of data elements in the APL may vary across states. In addition, some states may report plan names that are not unique across plan IDs or APL records. [3]
This analysis examines the extent to which the APL records include missing information for select data elements that identify program and population-related characteristics of these managed care entities. Separately, it examines the extent to which the APL records include missing information for select data elements that identify operational characteristics of these managed care entities. Finally, it examines the extent to which APL records are missing a plan name or plan type. For context, the analysis also shows the percentage of records with a non-unique plan name or plan ID and the distribution of plan types represented in the APL records.
The APL consists of five files. The APL base file includes basic plan-level information, whereas four additional supplemental files provide more detailed information on: plan geographic locations, service areas, populations enrolled, and operating authority. Each record in the APL base file represents a managed care plan or entity; in some cases, states may report multiple records per plan or entity—for example, to distinguish between the sub-populations enrolled. Each record in the APL base file may link to more than one record in each supplemental file.
States associate a plan identifier (plan ID) with each managed care entity or program operating in the state. The APL file is organized by plan ID.
For example, some states assign a single ID to a managed care plan and then use the managed care location number to identify the plan’s various locations throughout the state. However, other states assign a unique ID to each location for the same plan and then do not assign a location number or use the plan ID as the managed care location number.
The APL consists of five files. The APL base file includes basic plan-level information, whereas four additional supplemental files provide more detailed information on: plan geographic locations, service areas, populations enrolled, and operating authority. Each record in the APL base file represents a managed care plan or entity; in some cases, states may report multiple records per plan or entity\u2014for example, to distinguish between the sub-populations enrolled. Each record in the APL base file may link to more than one record in each supplemental file.
States associate a plan identifier (plan ID) with each managed care entity or program operating in the state. The APL file is organized by plan ID.
For example, some states assign a single ID to a managed care plan and then use the managed care location number to identify the plan\u2019s various locations throughout the state. However, other states assign a unique ID to each location for the same plan and then do not assign a location number or use the plan ID as the managed care location number.
This analysis uses all records in the TAF APL, including those representing both active and non-active plans. [4] , [5] We included in the analysis only states classified as using managed care in the Medicaid Managed Care Enrollment and Program Characteristics (MMCEPC) report for the calendar year. We did not evaluate the small number of states with no managed care program in the MMCEPC, even if these states had records present in the APL file. Additionally, if a state’s Medicaid program did not contract with any managed care plans during the year or a state’s Medicaid program contracted with one or more managed care plans during the year but the state had no records in the APL, we described it as unclassified.
We calculated the percentage of APL records with missing values across several data elements that convey information about program, population, and operational characteristics of the managed care entities and identifying information (Table 1). [6]
Table 1. TAF elements assessed
| TAF data element | Element description | Valid values |
|---|---|---|
| Managed care program code (MC_PGM_CD) | The state program through which a managed care plan is approved to operate | 1: Medicaid State Plan 2: CHIP State Plan 3: Both Medicaid and CHIP |
| Reimbursement arrangement category (REIMBRSMT_ARNGMT_CAT) | Aggregated categories of managed care entity reimbursement arrangement values | 1: Risk-based capitation 2: Non-risk capitation 3: Fee-for-service 4: Primary care case management (PCCM) 5: Other |
| Eligible population variables (POP_*) | Flag indicating whether the managed care entity is authorized to enroll a given eligibility group; 11 instances for 11 specific eligibility group(s) | 0: No 1: Yes |
| Operating authority variables (OPRTG_AUTHRTY_*) | Flag indicating the operating authority for the plan, when applicable; 15 instances for up to 15 potential operating authorities for a given managed care entity | 0: No 1: Yes |
| Profit status (MC_PRFT_STUS_CD) | The profit status of the managed care entity | 01: 501(C)(3) non-profit 02: For-profit, closely held 03: For-profit, publicly traded 04: Other |
| Managed care plan name (MC_NAME) | The name of the managed care entity under contract with the state Medicaid agency | The managed care name reported by the managed care entity |
| Managed care plan ID (MC_PLAN_ID) | The ID number the state issued to the managed care entity | State-assigned unique managed care identification number |
| Managed care plan type (MC_PLAN_TYPE_CD) | The type of managed care plan that corresponds to MC_PLAN_ID | 01: Comprehensive MCO 02: Traditional PCCM provider arrangement 03: Enhanced PCCM provider arrangement 04: Health insuring organization (HIO) 05: Medical-only prepaid inpatient health plan (PIHP) (risk or non-risk/non-comprehensive/with inpatient hospital or institutional services) 06: Medical-only prepaid ambulatory health plan (PAHP) (risk or non-risk/non-comprehensive/no inpatient hospital or institutional services) 07: Long-term care services and supports (LTSS) PIHP 08: Mental health (MH) PIHP 09: MH PAHP 10: Substance use disorders (SUD) PIHP 11: SUD PAHP 12: MH and SUD PIHP 13: MH and SUD PAHP 14: Dental PAHP 15: Transportation PAHP 16: Disease management PAHP 17: Program of All-Inclusive Care for the Elderly (PACE) 18: Pharmacy PAHP 19: Individual is enrolled in LTSS and MH PIHP 20: Other 60: Accountable care organization (ACO) 70: Health/medical home 80: Integrated care for dual eligibles |
We conducted three data quality assessments. The first assessment captured the percentage of APL records with a missing value for at least one of the data elements that identify program and population-related characteristics of the managed care entity: managed care program code and eligible population identifiers. [7] The second assessment captured the percentage of APL records with a missing value for at least one of the data elements that identify operational characteristics of the managed care entity: reimbursement arrangement category, operating authority identifiers, and profit status. [8] For each of these assessments, we grouped states into levels of concern about the usability of their data based on this percentage (Tables 2 and 3).
Table 2. Criteria for DQ assessment of managed care plan program and population characteristics
| Percentage of APL records with a missing value for managed care program code or eligible population identifiers | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
Table 3. Criteria for DQ assessment of managed care plan operational characteristics
| Percentage of APL records with a missing value for reimbursement arrangement category, operating authority identifiers, or profit status | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
The third assessment captured the percentage of APL records with a missing value for either plan name or plan type. We grouped states into levels of concern about the usability of their data based on this percentage (Table 4).
Table 4. Criteria for DQ assessment of plan name and type
| Percentage of APL records with a missing value for plan name or plan type | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
To provide users with additional information about how plans are represented in each state’s APL file, we present the percentage of APL records that had non-unique plan names. [9] This information is available in the table, but we did not use it in the data quality assessment for plan name because the T-MSIS data reporting requirements allow multiple plan IDs for a single plan name.
We also categorized and reported the distribution of managed care plan types. Table 5 describes the plan type categories used for reporting the distribution of APL records across plan type in each state.
Table 5. Plan types categories
| Plan type category | TAF plan type codes (MC_PLAN_TYPE_CD) |
|---|---|
| Comprehensive managed care (CMC) | 01: Comprehensive MCO 04: HIO 80: Integrated care for dually eligible individuals |
| Primary care case management (PCCM) | 02: Traditional PCCM provider arrangement 03: Enhanced PCCM provider arrangement |
| Managed long-term care services and supports (MLTSS)-only | 07: LTSS PIHP 19: Individual is enrolled in LTSS and MH PIHP |
| Behavioral health organization (BHO) | 08: MH PIHP 09: MH PAHP 10: SUD PIHP 11: SUD PAHP 12: MH and SUD PIHP 13: MH and SUD PAHP |
| Dental only | 14: Dental PAHP |
| Non-emergency medical transportation | 15: Transportation PAHP |
| Programs of All-Inclusive Care for the Elderly (PACE) | 17: PACE |
| Accountable care organizations (ACO) | 60: ACO |
| Other health plans | 05: Medical-only PIHP (non-comprehensive, with inpatient hospital or institutional services) 06: Medical-only PAHP (non-comprehensive; no inpatient hospital or institutional services) 16: Disease management PAHP 18: Pharmacy PAHP 20: Other 70: Health/medical home |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
States should include only those managed care plans that were active in the relevant calendar year in their APL reporting. An active managed care plan record in T-MSIS has (a) an effective start date and end date that overlap with at least part of the calendar year, and (b) an indicator identifying the record as active if multiple records represent the same time period. However, states sometimes submit active managed care plan records that in fact were inactive because the plan’s contract was not in effect during the calendar year. If a state reports these inactive plans as being active in the calendar year, they are included in the APL file.
During the TAF production process any invalid values submitted by the state are recoded to null (missing). As a result, missing values in TAF may actually represent either missing or invalid T-MSIS data submitted by states.
The eligible population identifiers were considered missing if there was not at least one eligible population indicated.
The operating authority identifiers were considered missing if there was not at least one operating authority indicated.
To assess non-uniqueness for plan names, we started by processing the names—removing excess spacing, punctuation, and other items that could complicate the matching process. We then determined how many records shared a plan name with one or more other records.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
States should include only those managed care plans that were active in the relevant calendar year in their APL reporting. An active managed care plan record in T-MSIS has (a) an effective start date and end date that overlap with at least part of the calendar year, and (b) an indicator identifying the record as active if multiple records represent the same time period. However, states sometimes submit active managed care plan records that in fact were inactive because the plan\u2019s contract was not in effect during the calendar year. If a state reports these inactive plans as being active in the calendar year, they are included in the APL file.
During the TAF production process any invalid values submitted by the state are recoded to null (missing). As a result, missing values in TAF may actually represent either missing or invalid T-MSIS data submitted by states.
The eligible population identifiers were considered missing if there was not at least one eligible population indicated.
The operating authority identifiers were considered missing if there was not at least one operating authority indicated.
To assess non-uniqueness for plan names, we started by processing the names\u2014removing excess spacing, punctuation, and other items that could complicate the matching process. We then determined how many records shared a plan name with one or more other records.
The TAF Annual Managed Care Plan (APL) file contains plan-level information, including characteristics, locations, enrolled populations, and service areas for all Medicaid and Children's Health Insurance Program (CHIP) health plans and managed care entities. This analysis examines the extent to which the APL records include missing information for selected data elements that identify operational characteristics of these managed care entities. These selected data elements include the waiver or state plan authority under which the plan operates; the reimbursement arrangement between the plan and its providers; and whether the plan is for profit or not for profit.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""8051"", ""relatedTopics"": [{""measureId"": 98, ""measureName"": ""Managed Care Plan Program and Population Characteristics"", ""groupId"": 10, ""groupName"": ""Managed Care Plans"", ""order"": 0}, {""measureId"": 100, ""measureName"": ""Plan Name and Type"", ""groupId"": 10, ""groupName"": ""Managed Care Plans"", ""order"": 2}]}" 100,"{""measureId"": 100, ""measureName"": ""Plan Name and Type"", ""groupId"": 10, ""groupName"": ""Managed Care Plans"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Plan-Name-Type.pdf"", ""background"": {""content"": ""Managed care plays a significant role in the delivery of care to Medicaid beneficiaries nationwide. Nearly all states and territories contract with managed care organizations (MCOs) to provide care to at least some of their beneficiaries. All states and territories contracting with one or more Medicaid managed care plans must report into T-MSIS (1) information about the managed care plans in which beneficiaries are enrolled each month, (2) service utilization (encounter) records and capitation payments, and (3) information on the characteristics of each managed care plan operating in the state. T-MSIS Analytic Files (TAF) users can find beneficiary-level plan enrollment information in the Demographics and Eligibility (DE) file; encounter records and capitation payments in the inpatient (IP), long-term care (LT), other services (OT), and prescription drug (RX) claims files; and managed care plan-level information in the Annual Managed Care Plan (APL) file.
Although the TAF DE file includes limited information about each managed care plan in which a beneficiary is enrolled (including plan ID and type), the APL file includes additional information about the characteristics, locations, enrolled populations, and service areas for all Medicaid and Children’s Health Insurance Program (CHIP) health plans and managed care entities. [1] , [2] The APL file also includes additional identifying information, such as plan name and plan type, that TAF users need if they want to identify a specific managed care plan in the TAF data. However, the completeness of data elements in the APL may vary across states. In addition, some states may report plan names that are not unique across plan IDs or APL records. [3]
This analysis examines the extent to which the APL records include missing information for select data elements that identify program and population-related characteristics of these managed care entities. Separately, it examines the extent to which the APL records include missing information for select data elements that identify operational characteristics of these managed care entities. Finally, it examines the extent to which APL records are missing a plan name or plan type. For context, the analysis also shows the percentage of records with a non-unique plan name or plan ID and the distribution of plan types represented in the APL records.
The APL consists of five files. The APL base file includes basic plan-level information, whereas four additional supplemental files provide more detailed information on: plan geographic locations, service areas, populations enrolled, and operating authority. Each record in the APL base file represents a managed care plan or entity; in some cases, states may report multiple records per plan or entity—for example, to distinguish between the sub-populations enrolled. Each record in the APL base file may link to more than one record in each supplemental file.
States associate a plan identifier (plan ID) with each managed care entity or program operating in the state. The APL file is organized by plan ID.
For example, some states assign a single ID to a managed care plan and then use the managed care location number to identify the plan’s various locations throughout the state. However, other states assign a unique ID to each location for the same plan and then do not assign a location number or use the plan ID as the managed care location number.
The APL consists of five files. The APL base file includes basic plan-level information, whereas four additional supplemental files provide more detailed information on: plan geographic locations, service areas, populations enrolled, and operating authority. Each record in the APL base file represents a managed care plan or entity; in some cases, states may report multiple records per plan or entity\u2014for example, to distinguish between the sub-populations enrolled. Each record in the APL base file may link to more than one record in each supplemental file.
States associate a plan identifier (plan ID) with each managed care entity or program operating in the state. The APL file is organized by plan ID.
For example, some states assign a single ID to a managed care plan and then use the managed care location number to identify the plan\u2019s various locations throughout the state. However, other states assign a unique ID to each location for the same plan and then do not assign a location number or use the plan ID as the managed care location number.
This analysis uses all records in the TAF APL, including those representing both active and non-active plans. [4] , [5] We included in the analysis only states classified as using managed care in the Medicaid Managed Care Enrollment and Program Characteristics (MMCEPC) report for the calendar year. We did not evaluate the small number of states with no managed care program in the MMCEPC, even if these states had records present in the APL file. Additionally, if a state’s Medicaid program did not contract with any managed care plans during the year or a state’s Medicaid program contracted with one or more managed care plans during the year but the state had no records in the APL, we described it as unclassified.
We calculated the percentage of APL records with missing values across several data elements that convey information about program, population, and operational characteristics of the managed care entities and identifying information (Table 1). [6]
Table 1. TAF elements assessed
| TAF data element | Element description | Valid values |
|---|---|---|
| Managed care program code (MC_PGM_CD) | The state program through which a managed care plan is approved to operate | 1: Medicaid State Plan 2: CHIP State Plan 3: Both Medicaid and CHIP |
| Reimbursement arrangement category (REIMBRSMT_ARNGMT_CAT) | Aggregated categories of managed care entity reimbursement arrangement values | 1: Risk-based capitation 2: Non-risk capitation 3: Fee-for-service 4: Primary care case management (PCCM) 5: Other |
| Eligible population variables (POP_*) | Flag indicating whether the managed care entity is authorized to enroll a given eligibility group; 11 instances for 11 specific eligibility group(s) | 0: No 1: Yes |
| Operating authority variables (OPRTG_AUTHRTY_*) | Flag indicating the operating authority for the plan, when applicable; 15 instances for up to 15 potential operating authorities for a given managed care entity | 0: No 1: Yes |
| Profit status (MC_PRFT_STUS_CD) | The profit status of the managed care entity | 01: 501(C)(3) non-profit 02: For-profit, closely held 03: For-profit, publicly traded 04: Other |
| Managed care plan name (MC_NAME) | The name of the managed care entity under contract with the state Medicaid agency | The managed care name reported by the managed care entity |
| Managed care plan ID (MC_PLAN_ID) | The ID number the state issued to the managed care entity | State-assigned unique managed care identification number |
| Managed care plan type (MC_PLAN_TYPE_CD) | The type of managed care plan that corresponds to MC_PLAN_ID | 01: Comprehensive MCO 02: Traditional PCCM provider arrangement 03: Enhanced PCCM provider arrangement 04: Health insuring organization (HIO) 05: Medical-only prepaid inpatient health plan (PIHP) (risk or non-risk/non-comprehensive/with inpatient hospital or institutional services) 06: Medical-only prepaid ambulatory health plan (PAHP) (risk or non-risk/non-comprehensive/no inpatient hospital or institutional services) 07: Long-term care services and supports (LTSS) PIHP 08: Mental health (MH) PIHP 09: MH PAHP 10: Substance use disorders (SUD) PIHP 11: SUD PAHP 12: MH and SUD PIHP 13: MH and SUD PAHP 14: Dental PAHP 15: Transportation PAHP 16: Disease management PAHP 17: Program of All-Inclusive Care for the Elderly (PACE) 18: Pharmacy PAHP 19: Individual is enrolled in LTSS and MH PIHP 20: Other 60: Accountable care organization (ACO) 70: Health/medical home 80: Integrated care for dual eligibles |
We conducted three data quality assessments. The first assessment captured the percentage of APL records with a missing value for at least one of the data elements that identify program and population-related characteristics of the managed care entity: managed care program code and eligible population identifiers. [7] The second assessment captured the percentage of APL records with a missing value for at least one of the data elements that identify operational characteristics of the managed care entity: reimbursement arrangement category, operating authority identifiers, and profit status. [8] For each of these assessments, we grouped states into levels of concern about the usability of their data based on this percentage (Tables 2 and 3).
Table 2. Criteria for DQ assessment of managed care plan program and population characteristics
| Percentage of APL records with a missing value for managed care program code or eligible population identifiers | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
Table 3. Criteria for DQ assessment of managed care plan operational characteristics
| Percentage of APL records with a missing value for reimbursement arrangement category, operating authority identifiers, or profit status | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
The third assessment captured the percentage of APL records with a missing value for either plan name or plan type. We grouped states into levels of concern about the usability of their data based on this percentage (Table 4).
Table 4. Criteria for DQ assessment of plan name and type
| Percentage of APL records with a missing value for plan name or plan type | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
To provide users with additional information about how plans are represented in each state’s APL file, we present the percentage of APL records that had non-unique plan names. [9] This information is available in the table, but we did not use it in the data quality assessment for plan name because the T-MSIS data reporting requirements allow multiple plan IDs for a single plan name.
We also categorized and reported the distribution of managed care plan types. Table 5 describes the plan type categories used for reporting the distribution of APL records across plan type in each state.
Table 5. Plan types categories
| Plan type category | TAF plan type codes (MC_PLAN_TYPE_CD) |
|---|---|
| Comprehensive managed care (CMC) | 01: Comprehensive MCO 04: HIO 80: Integrated care for dually eligible individuals |
| Primary care case management (PCCM) | 02: Traditional PCCM provider arrangement 03: Enhanced PCCM provider arrangement |
| Managed long-term care services and supports (MLTSS)-only | 07: LTSS PIHP 19: Individual is enrolled in LTSS and MH PIHP |
| Behavioral health organization (BHO) | 08: MH PIHP 09: MH PAHP 10: SUD PIHP 11: SUD PAHP 12: MH and SUD PIHP 13: MH and SUD PAHP |
| Dental only | 14: Dental PAHP |
| Non-emergency medical transportation | 15: Transportation PAHP |
| Programs of All-Inclusive Care for the Elderly (PACE) | 17: PACE |
| Accountable care organizations (ACO) | 60: ACO |
| Other health plans | 05: Medical-only PIHP (non-comprehensive, with inpatient hospital or institutional services) 06: Medical-only PAHP (non-comprehensive; no inpatient hospital or institutional services) 16: Disease management PAHP 18: Pharmacy PAHP 20: Other 70: Health/medical home |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
States should include only those managed care plans that were active in the relevant calendar year in their APL reporting. An active managed care plan record in T-MSIS has (a) an effective start date and end date that overlap with at least part of the calendar year, and (b) an indicator identifying the record as active if multiple records represent the same time period. However, states sometimes submit active managed care plan records that in fact were inactive because the plan’s contract was not in effect during the calendar year. If a state reports these inactive plans as being active in the calendar year, they are included in the APL file.
During the TAF production process any invalid values submitted by the state are recoded to null (missing). As a result, missing values in TAF may actually represent either missing or invalid T-MSIS data submitted by states.
The eligible population identifiers were considered missing if there was not at least one eligible population indicated.
The operating authority identifiers were considered missing if there was not at least one operating authority indicated.
To assess non-uniqueness for plan names, we started by processing the names—removing excess spacing, punctuation, and other items that could complicate the matching process. We then determined how many records shared a plan name with one or more other records.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
States should include only those managed care plans that were active in the relevant calendar year in their APL reporting. An active managed care plan record in T-MSIS has (a) an effective start date and end date that overlap with at least part of the calendar year, and (b) an indicator identifying the record as active if multiple records represent the same time period. However, states sometimes submit active managed care plan records that in fact were inactive because the plan\u2019s contract was not in effect during the calendar year. If a state reports these inactive plans as being active in the calendar year, they are included in the APL file.
During the TAF production process any invalid values submitted by the state are recoded to null (missing). As a result, missing values in TAF may actually represent either missing or invalid T-MSIS data submitted by states.
The eligible population identifiers were considered missing if there was not at least one eligible population indicated.
The operating authority identifiers were considered missing if there was not at least one operating authority indicated.
To assess non-uniqueness for plan names, we started by processing the names\u2014removing excess spacing, punctuation, and other items that could complicate the matching process. We then determined how many records shared a plan name with one or more other records.
The TAF Annual Managed Care Plan (APL) file contains plan-level information, including characteristics, locations, enrolled populations, and service areas for all Medicaid and Children's Health Insurance Program (CHIP) health plans and managed care entities. The APL file is organized by plan identifier but also includes information on the plan type and name, which TAF users might use to identify a specific managed care plan in the data. This analysis examines the prevalence of missing plan name and missing plan type codes in the APL file.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""8051"", ""relatedTopics"": [{""measureId"": 98, ""measureName"": ""Managed Care Plan Program and Population Characteristics"", ""groupId"": 10, ""groupName"": ""Managed Care Plans"", ""order"": 0}, {""measureId"": 99, ""measureName"": ""Managed Care Plan Operational Characteristics"", ""groupId"": 10, ""groupName"": ""Managed Care Plans"", ""order"": 1}]}" 101,"{""measureId"": 101, ""measureName"": ""Bundled Payments for Prenatal Care"", ""groupId"": 8, ""groupName"": ""Payments"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Bundled-Pmts-Prenatal.pdf"", ""background"": {""content"": ""Maternity care services, including prenatal care, delivery, and postpartum care, can be billed either separately for each service or as a “bundle” of services with a bundled payment procedure code. Two types of Current Procedural Terminology (CPT) bundled payment codes include delivery: (1) those that include standard prenatal care and the mother’s charges for an uncomplicated delivery and postpartum care, and (2) those that include only the charges for an uncomplicated delivery and postpartum care (Table 1). [1] , [2] These bundled payment codes are used by many payers, including state Medicaid agencies and Medicaid managed care organizations, and they can be used in fee-for-service (FFS) or managed care arrangements.
If bundled payment codes are used to bill for deliveries, TAF users may need to incorporate supplemental data when they rely on administrative claims data to assess the use of maternity care services. Specifically, a study or a quality measure for assessing the number or timing of prenatal care visits might underestimate service use or may not be able to identify the dates of prenatal visits. As a result, TAF users may inadvertently form inaccurate conclusions about the timing and number of prenatal care visits a person receives. Furthermore, if a person receives prenatal care covered only by a bundled payment code, they may not have a claim related to the pregnancy until delivery, complicating whether and the extent to which claims data can be used to identify or count beneficiaries who are pregnant but have not yet delivered. One option for addressing limitations in measuring prenatal care services is to link the administrative claims data to electronic health or vital records that include information on timing and number of prenatal visits or information on the date of delivery and delivery outcome.
If sufficient supplemental data is not available, TAF users who are measuring prenatal service use should exclude from their analysis any delivery that uses a CPT code that bundles prenatal care into the payment for delivery (Table 1), even in states that fall into the low concern category. Use of these codes indicate the provider is not expected to submit separate claims for each prenatal visit associated with that pregnancy, and the lack of prenatal claims cannot be interpreted as the beneficiary failing to receive prenatal services. This assessment provides information on the extent to which TAF users would need to account for bundled payments in each state when measuring the timeliness and utilization of prenatal care services.
Table 1. CPT bundled payment codes
| Type | Code and description [3] |
|---|---|
| Codes that include prenatal care, delivery services, and postnatal care |
|
| Codes that include delivery services and postnatal care |
|
The bundled payment claim is submitted to the payer at the time of delivery, and the code is in the claim’s procedure code field.
Per CPT coding guidance, bundled payments for maternity care cover only uncomplicated prenatal and postpartum care, along with labor and delivery; any complications, laboratory and imaging services, or additional testing are billed separately. See, for example, Ballard Jr., Dawson. “From Antepartum to Postpartum, Get the CPT ® OB Basics: Simplifying Coding by Knowing What Is Packaged into Obstetrics Care.” August 1, 2013. Available at https://www.aapc.com/blog/25857-from-antepartum-to-postpartum-get-the-cpt-ob-basics/. Accessed January 16, 2019.
This list of CPT bundled payment codes is based on Mathematica’s review of the CPT® 2016 Professional Edition (American Medical Association 2016). We consulted two additional sources to ensure that we included the correct list of CPT bundled payment codes: (1) the Medicaid Innovation Accelerator Program Maternal and Infant Health Pregnant and Postpartum Beneficiary Tool and (2) the American Academy of Professional Coders Knowledge Center. See Verhovshek, John. “Coding Maternity Care with Insurance Change.” June 7, 2018. Available at https://www.aapc.com/blog/42591-coding-maternity-care-insurance-change/ . Accessed January 16, 2019.
The bundled payment claim is submitted to the payer at the time of delivery, and the code is in the claim\u2019s procedure code field.
Per CPT coding guidance, bundled payments for maternity care cover only uncomplicated prenatal and postpartum care, along with labor and delivery; any complications, laboratory and imaging services, or additional testing are billed separately. See, for example, Ballard Jr., Dawson. \u201cFrom Antepartum to Postpartum, Get the CPT \u00ae OB Basics: Simplifying Coding by Knowing What Is Packaged into Obstetrics Care.\u201d August 1, 2013. Available at https://www.aapc.com/blog/25857-from-antepartum-to-postpartum-get-the-cpt-ob-basics/. Accessed January 16, 2019.
This list of CPT bundled payment codes is based on Mathematica\u2019s review of the CPT\u00ae 2016 Professional Edition (American Medical Association 2016). We consulted two additional sources to ensure that we included the correct list of CPT bundled payment codes: (1) the Medicaid Innovation Accelerator Program Maternal and Infant Health Pregnant and Postpartum Beneficiary Tool and (2) the American Academy of Professional Coders Knowledge Center. See Verhovshek, John. \u201cCoding Maternity Care with Insurance Change.\u201d June 7, 2018. Available at https://www.aapc.com/blog/42591-coding-maternity-care-insurance-change/ . Accessed January 16, 2019.
We used the T-MSIS Analytic Files (TAF) annual Demographic and Eligibility (DE) file and included FFS claims and managed care encounter records from the other services (OT) and inpatient (IP) files for both Medicaid and CHIP beneficiaries. [4] , [5] States were excluded from this analysis if the volume of header records in either the IP or OT file was unusably low or if the IP or OT file had high levels of missing or invalid diagnosis codes, procedure codes, or admission dates.
We first limited the analysis to female beneficiaries ages 8 through 64 and identified all deliveries in the OT and IP files using procedure, diagnosis, and revenue codes for live births, stillbirths, and labor and delivery with an unknown outcome. [6] , [7] , [8] We then assessed the percentage of deliveries that had a bundled payment code that include prenatal care on the claim by state.
We also measured the extent to which each state used bundled payment codes that excluded prenatal care to bill for deliveries. This information is presented in the data table to help TAF users understand the extent to which they should account for bundled payments of delivery services and postnatal care but was not used to assign the concern level.
We categorized states into levels of concern about the usability of their TAF data for measuring prenatal care visits. The state was assigned a level of concern based on the percentage of deliveries that include a bundled payment code that include prenatal care (Table 2).
For states in which bundled payment codes that include prenatal care account for a high proportion of payments for deliveries, TAF users who are interested in assessing the timing and use of prenatal care services should either incorporate a supplemental data source, such as electronic health or vital records, or exclude deliveries that are billed using a bundled payment code from their analyses.
Table 2. Criteria for DQ assessment of usability of TAF data for measuring the number and timing of prenatal care services
| DQ assessment | |
|---|---|
| x < 5 percent | Low concern |
| 5 percent ≤ x < 20 percent | Medium concern |
| 20 percent ≤ x | High concern |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into the RIF, some TAF data elements were suppressed, changed, or renamed. For more details on the difference between the pre-RIF and RIF versions of the TAF data, including a crosswalk of variable names, see TAF DQ Brief #9010, “Production of the TAF Research Identifiable Files (RIFs).”
We included both the IP and OT files in the analysis after finding that claims for deliveries were present in both files. The claims in the OT file included, for example, charges for professional services related to a delivery, with the place of service being an inpatient hospital. Researchers interested in deliveries billed by using a bundled payment code will therefore have to include the OT file in their analysis.
We used an age range of 8 to 64 to capture as many pregnancies as possible, including those outside the typical reproductive age range, which is 15 to 49. Excluding beneficiaries younger than 8 or older than 64 also eliminates obvious coding errors for beneficiaries of pediatric or geriatric age.
An individual with more than one delivery during the year, which is relatively rare, is counted only once in the analysis.
The list of codes used to identify all deliveries was developed for the Medicaid Innovation Accelerator Program Maternal and Infant Health Pregnant and Postpartum Beneficiary Tool. The approach relies on accuracy in the diagnosis, procedure, and revenue code data elements in the IP and OT claims files.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into the RIF, some TAF data elements were suppressed, changed, or renamed. For more details on the difference between the pre-RIF and RIF versions of the TAF data, including a crosswalk of variable names, see TAF DQ Brief #9010, \u201cProduction of the TAF Research Identifiable Files (RIFs).\u201d
We included both the IP and OT files in the analysis after finding that claims for deliveries were present in both files. The claims in the OT file included, for example, charges for professional services related to a delivery, with the place of service being an inpatient hospital. Researchers interested in deliveries billed by using a bundled payment code will therefore have to include the OT file in their analysis.
We used an age range of 8 to 64 to capture as many pregnancies as possible, including those outside the typical reproductive age range, which is 15 to 49. Excluding beneficiaries younger than 8 or older than 64 also eliminates obvious coding errors for beneficiaries of pediatric or geriatric age.
An individual with more than one delivery during the year, which is relatively rare, is counted only once in the analysis.
The list of codes used to identify all deliveries was developed for the Medicaid Innovation Accelerator Program Maternal and Infant Health Pregnant and Postpartum Beneficiary Tool. The approach relies on accuracy in the diagnosis, procedure, and revenue code data elements in the IP and OT claims files.
Maternity care services, including prenatal care, delivery, and postpartum care, can be billed either separately for each service or as a \""bundle\"" of services with a bundled payment procedure code. This analysis provides information on the extent to which TAF users would need to account for bundled payments when measuring the timeliness and utilization of prenatal care services.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""7031"", ""relatedTopics"": []}" 102,"{""measureId"": 102, ""measureName"": ""Provider Location"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Provider-Location.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) Annual Provider File (APR) captures detailed information about each provider authorized by a state to provide services to its Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries, as well as providers whose approval is pending, denied, or has been terminated. Providers are included in the APR file regardless of whether or how often they have billed the state for services. [1] The APR file includes more detailed information about those providers rendering or billing for the service than do the fee-for-service (FFS) and encounter records in the TAF claims files. For example, FFS and encounter records in the inpatient (IP), long-term care (LT), and other services (OT) files include information on each claim regarding the billing, servicing, and rendering provider’s taxonomy, specialty, and state-identified provider type. [2] The APR file, on the other hand, includes details about the characteristics, locations, taxonomies/classifications, affiliated groups, affiliated programs, licensing/accreditations, and—for facility providers—the bed types linked with the facility, as well as other identifiers associated with a provider. [3]
Each record in the APR base file should link to one or more records in the APR location supplemental file containing information about the geographic location (including address, ZIP code, and county code) from which a provider bills, practices, or provides services. TAF users may want to use this information to evaluate the number of health care providers available to serve Medicaid and CHIP populations by geographic area. The Centers for Medicare & Medicaid Services requires states that contract with managed care organizations, prepaid inpatient health plans (PIHPs), or prepaid ambulatory health plans (PAHPs) to develop and enforce network adequacy standards, thus ensuring that beneficiaries have access to primary care, OB/GYN services, behavioral health care, specialists, hospitals, pharmacies, pediatric dental providers, and long-term services and supports. States are also required to monitor access to care to ensure that FFS rates are sufficient to enlist enough providers to ensure care is available to beneficiaries in all areas of the state. Using TAF data to support these types of access-to-care analyses requires detailed information on practice and servicing location for Medicaid- or CHIP-enrolled providers.
This data quality assessment examines the extent to which two key geographic identifiers of the provider’s service or practice location—county code and ZIP code—are present in the location supplemental file for the subset of providers in the APR base file that are actively participating in Medicaid or CHIP. [4] It also examines the completeness of location information separately for facilities, groups, and individual providers.
TAF APR records present unique combinations of submitting state and provider. Because providers can participate with more than one state’s Medicaid program, more than one APR record may represent a given provider.
In addition, FFS and encounter records in the IP and LT files include information about the admitting provider’s taxonomy, specialty, and state-identified provider type for each claim, and the FFS and encounter records in the pharmacy (RX) file include information about the billing provider’s taxonomy and specialty for each claim.
The TAF APR consists of nine files, including the base file and eight additional supplemental files: the affiliated groups file, affiliated programs file, taxonomy file, enrollment file, location file, licensing file, identifiers file, and bed type file. The TAF APR base file includes basic provider characteristics, and the eight additional supplemental files provide more detailed information on the following topics: the provider’s taxonomy, Medicaid and/or CHIP program enrollment, group affiliation, health plan and program affiliation, geographic location(s), licensing and accreditation, other provider identifiers, and bed type. Each record in the TAF APR base file represents a provider enrolled with the state’s Medicaid or CHIP program. Each provider record in the base file may link to more than one record in each supplemental file.
Street address information is also available in the TAF APR, but we did not examine it for this DQ assessment.
TAF APR records present unique combinations of submitting state and provider. Because providers can participate with more than one state\u2019s Medicaid program, more than one APR record may represent a given provider.
In addition, FFS and encounter records in the IP and LT files include information about the admitting provider\u2019s taxonomy, specialty, and state-identified provider type for each claim, and the FFS and encounter records in the pharmacy (RX) file include information about the billing provider\u2019s taxonomy and specialty for each claim.
The TAF APR consists of nine files, including the base file and eight additional supplemental files: the affiliated groups file, affiliated programs file, taxonomy file, enrollment file, location file, licensing file, identifiers file, and bed type file. The TAF APR base file includes basic provider characteristics, and the eight additional supplemental files provide more detailed information on the following topics: the provider\u2019s taxonomy, Medicaid and/or CHIP program enrollment, group affiliation, health plan and program affiliation, geographic location(s), licensing and accreditation, other provider identifiers, and bed type. Each record in the TAF APR base file represents a provider enrolled with the state\u2019s Medicaid or CHIP program. Each provider record in the base file may link to more than one record in each supplemental file.
Street address information is also available in the TAF APR, but we did not examine it for this DQ assessment.
This analysis was restricted to providers in the APR base file who saw Medicaid or CHIP patients during the calendar year. [5] We chose to include only providers actively participating in Medicaid or CHIP because we anticipate these records are the ones most likely to be used for analyses of provider location and health care access. To identify actively participating providers, we restricted the analysis to APR base file records that linked to at least one FFS claim or managed care encounter during the year. [6] We used the state-assigned provider IDs (SUBMTG_STATE_PRVDR_ID) in the APR base file to link to the state-assigned provider IDs on FFS claims and managed care encounters in the monthly TAF claims files, including the inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) claims files. This linking was limited to FFS claims and managed care encounters (CLM_TYPE_CD = 1, 3, A, or C) to represent services delivered to beneficiaries and funded through Medicaid or CHIP. [7] For Illinois, we linked only to the original version of the claim, not subsequent adjustment records. [8] A single claim or encounter can include up to six state-assigned provider IDs, depending on the file type; this analysis linked the state-assigned provider IDs specified in Table 1. [9] , [10]
Table 1. State-assigned provider identifiers used to link the TAF APR base file and claims files
| Provider identifier | TAF file | TAF field name |
|---|---|---|
| Submitting State Provider ID | APR base file | SUBMTG_STATE_PRVDR_ID |
| Admitting Provider | IP and LT header files | ADMTG_PRVDR_NUM |
| Billing Provider | IP, LT, OT, and RX header files | BLG_PRVDR_NUM |
| Referring Provider | IP, LT, and OT header files | RFRG_PRVDR_NUM |
| Servicing Provider | OT line file | SRVCNG_PRVDR_NUM |
| Prescribing Provider | RX header files | PRSCRBNG_PRVDR_NUM |
| Dispensing Provider | RX header files | DSPNSNG_PD_PRVDR_NUM |
We linked information from the location supplemental file to this subset of base file records representing providers actively participating in Medicaid or CHIP. Providers may have multiple geographic locations indicated in the location supplemental file because of different service, practice, and billing locations; we examined only the location supplemental file records representing provider service and practice locations. [11]
We calculated the percentage of providers actively participating in Medicaid or CHIP that did not link to either a service or practice location with usable data for either ZIP or county code. Table 2 presents the criteria used to define a usable county and ZIP code, respectively.
Table 2. Criteria used to define a usable county and ZIP code
| Variable | TAF data element name | Requirements to be considered usable |
|---|---|---|
| Provider ZIP code | ADR_ZIP_CD |
|
| Provider county code | ADR_CNTY_CD |
|
a The U.S. Census Bureau publishes FIPS files for each year, available at: https://www.census.gov/geographies/reference-files.html
We grouped states into categories of data quality concern based on the percentage of providers participating in Medicaid or CHIP that did not link to either a service or practice location with a usable ZIP code or a usable county code in the location supplemental file (Table 3).
Table 3. Criteria for DQ assessment of provider location
| Percentage of Medicaid or CHIP-participating providers in the APR that did not link to a usable service or practice location (ZIP or county code) | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent | Unusable |
| State did not have any Medicaid or CHIP-participating providers in the APR | Unclassified |
We also repeated this analysis on facilities, groups, and individual providers for those TAF users interested in analysis specific to one of these broader categories of providers. [12] We present this information but did not use it for the DQ assessment.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIFs). During the transformation into RIFs, some TAF data elements were suppressed, changed, or renamed. More details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the “Production of the TAF Research Identifiable Files” guide.
The APR file includes an active enrollment indicator, but it was found to be an unreliable indicator of provider participation in Medicaid or CHIP in many states in 2019. As a result, this analysis defined providers actively participating in Medicaid or CHIP during the year as those who linked to an FFS claim or managed care encounter. More information on the validity of the APR’s active enrollment indicator can be found in the DQ Atlas single-topic displays for Active Enrollment Status Indicator .
We excluded capitation, supplemental, and service tracking payments. We also excluded records with a claim type of “other,” which are records the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Titles XIX or XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits, rather than voiding the original claim and submitting a replacement record containing the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This practice means that in some cases, the TAF will include multiple versions of a single claim for Illinois. The original record is likely to contain the most complete nonpayment information (such as diagnosis code, procedure code, revenue center code, and so forth), and thus is likely to be the most usable record in the claim family for analytic purposes. Therefore, for Illinois, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide “How to Use Illinois Claims Data,” available on ResDAC.org.
The TAF claims files contain numerous provider identifier variables that represent the roles providers play in beneficiaries’ health care. The IP file includes information about billing, referring, servicing, admitting, and operating providers. The LT file includes information about billing, referring, servicing, and admitting providers. The OT file includes information about billing, referring, servicing, health home, directing, and supervising providers. The RX file includes information about billing, dispensing, and prescribing providers.
State-assigned provider identifier fields are available on a claim for admitting, billing, servicing, referring, dispensing, and prescribing providers. Although claims also include information for health home, supervising, operating, and directing providers, the provider identifier fields available for APR linkage for these providers are limited to National Provider Identifiers (NPIs) only. However, more than 95 percent of provider NPIs belonging to health home, supervising, operating, and directing providers can also be found elsewhere in the claims files in another NPI field where a state-assigned provider identifier is available (admitting, billing, servicing, referring, and dispensing NPIs). Therefore, we excluded operating, health home, directing, and supervising provider identifiers from this analysis. TAF users interested in these specific types of providers have the option to link the TAF APR and claims using the NPI fields only.
Location records representing a provider’s service address were identified as those in which PRVDR_ADR_SRVC_IND = 1; location records representing a provider’s practice address were identified as those in which PRVDR_ADR_PRCTC_IND = 1.
Facility providers are those with a facility/group/individual code (FAC_GRP_INDVDL_IND) of “01.” Group providers are those with a facility/group/individual code of “02,” and individual providers are those with a facility/group/individual code of “03.”
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIFs). During the transformation into RIFs, some TAF data elements were suppressed, changed, or renamed. More details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the \u201cProduction of the TAF Research Identifiable Files\u201d guide.
The APR file includes an active enrollment indicator, but it was found to be an unreliable indicator of provider participation in Medicaid or CHIP in many states in 2019. As a result, this analysis defined providers actively participating in Medicaid or CHIP during the year as those who linked to an FFS claim or managed care encounter. More information on the validity of the APR\u2019s active enrollment indicator can be found in the DQ Atlas single-topic displays for Active Enrollment Status Indicator .
We excluded capitation, supplemental, and service tracking payments. We also excluded records with a claim type of \u201cother,\u201d which are records the state did not classify as Medicaid or CHIP; these records may capture services that do not qualify for federal matching funds under Titles XIX or XXI.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits, rather than voiding the original claim and submitting a replacement record containing the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This practice means that in some cases, the TAF will include multiple versions of a single claim for Illinois. The original record is likely to contain the most complete nonpayment information (such as diagnosis code, procedure code, revenue center code, and so forth), and thus is likely to be the most usable record in the claim family for analytic purposes. Therefore, for Illinois, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide \u201cHow to Use Illinois Claims Data,\u201d available on ResDAC.org.
The TAF claims files contain numerous provider identifier variables that represent the roles providers play in beneficiaries\u2019 health care. The IP file includes information about billing, referring, servicing, admitting, and operating providers. The LT file includes information about billing, referring, servicing, and admitting providers. The OT file includes information about billing, referring, servicing, health home, directing, and supervising providers. The RX file includes information about billing, dispensing, and prescribing providers.
State-assigned provider identifier fields are available on a claim for admitting, billing, servicing, referring, dispensing, and prescribing providers. Although claims also include information for health home, supervising, operating, and directing providers, the provider identifier fields available for APR linkage for these providers are limited to National Provider Identifiers (NPIs) only. However, more than 95 percent of provider NPIs belonging to health home, supervising, operating, and directing providers can also be found elsewhere in the claims files in another NPI field where a state-assigned provider identifier is available (admitting, billing, servicing, referring, and dispensing NPIs). Therefore, we excluded operating, health home, directing, and supervising provider identifiers from this analysis. TAF users interested in these specific types of providers have the option to link the TAF APR and claims using the NPI fields only.
Location records representing a provider\u2019s service address were identified as those in which PRVDR_ADR_SRVC_IND = 1; location records representing a provider\u2019s practice address were identified as those in which PRVDR_ADR_PRCTC_IND = 1.
Facility providers are those with a facility/group/individual code (FAC_GRP_INDVDL_IND) of \u201c01.\u201d Group providers are those with a facility/group/individual code of \u201c02,\u201d and individual providers are those with a facility/group/individual code of \u201c03.\u201d
The TAF Annual Provider File (APR) contains detailed information about each provider authorized to deliver services to Medicaid and CHIP beneficiaries at any point during the calendar year. The APR location supplemental file captures information about each provider's billing, practice, and service locations. This data quality assessment examines the extent to which the APR location supplemental file contains valid information about the ZIP or county code of a service or practice location for providers actively participating in Medicaid or CHIP.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""9061"", ""relatedTopics"": []}" 103,"{""measureId"": 103, ""measureName"": ""Billing Provider Specialty and Taxonomy - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Billing-Prov-Type-RX.pdf"", ""background"": {""content"": ""Users of the T-MSIS Analytic Files (TAF) may want to identify the providers or categories of providers that deliver services to Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries. A claim in the TAF could include information for up to six providers, depending on the file type. [1] A provider could be a facility, a group of individual practitioners, or an individual practitioner. The provider types most often used for claims-based analyses include the billing, servicing, prescribing, and dispensing providers.
The billing provider represents the entity that submits the claim and is reimbursed by the Medicaid or CHIP agency. Information about the billing provider is available in all TAF claims files: inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX).
The servicing provider, in contrast, represents the individual practitioner who was responsible for or provided direct care to the beneficiary (that is, the rendering provider). Information about servicing provider is available in the IP, LT, and OT files. In some cases, the servicing provider is the same as the billing provider. In other cases, the servicing and billing provider differ, such as when a hospital or large group practice bills for services but an individual physician employed by the hospital or group provides the direct care and is the serving provider. Because institutional claims in the IP and LT files represent facility costs and are often submitted by the hospitals and skilled nursing facilities where the care occurred, analyses using these claims would typically only use the billing provider information.
For prescription drug claims, the billing provider is the pharmacy where the prescription was filled, the prescribing provider is the individual practitioner who prescribed a prescription drug to a beneficiary, and the dispensing provider is the pharmacist who filled the prescription or was responsible for overseeing the filling of the prescription. Because the billing provider represents the entity that submits a claim and is reimbursed by the Medicaid or CHIP agency, analyses using prescription drug claims typically focus on the billing provider information.
Data elements related to the billing provider can offer insight into the characteristics of providers who receive payment for Medicaid- and CHIP-funded services. Likewise, data elements related to the servicing provider can offer insight into the characteristics of the individual practitioner who was responsible for or provided direct care to a Medicaid or CHIP beneficiary. There are multiple systems available in TAF claim records for classifying providers, including the following:
Each of these classification systems may be best suited to different types of analyses. Although states are encouraged to populate all these fields on all claims, only one classification type is required. In practice, some states only submit information related to some of the classification systems, which may require TAF users to adjust their methodology across states. If the preferred data element has high rates of missingness, TAF users may be able to link the claims record to the provider’s record in the Annual Provider File (APR) to obtain the needed information. [2]
This data quality assessment examines whether any information on billing or servicing provider characteristics is available in claims records, and which type (taxonomy, specialty, or provider type). This information can help users design their analysis by selecting the fields populated in each state or identify states for which it may be necessary to link to the APR to obtain complete information about providers, rather than rely on claims alone.
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
States are required to submit information about all providers eligible to provide Medicaid- and CHIP-funded services; these records are captured in the APR. It is possible that missing information in the claim about provider taxonomy, specialty, or type could be imputed by linking to the provider’s record in the APR and obtaining the information from that file. In addition, the APR includes information on how states classify providers by “authorized category of service,” a classification scheme that can be used to identify certain non-medical provider categories not captured in the other classification systems, such as transportation or personal care service providers. For more information on the completeness of these fields in each state’s APR file, see the DQ Atlas single-topic displays for Group and Individual Providers - Classification Types and Facilities - Classification Types .
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
States are required to submit information about all providers eligible to provide Medicaid- and CHIP-funded services; these records are captured in the APR. It is possible that missing information in the claim about provider taxonomy, specialty, or type could be imputed by linking to the provider\u2019s record in the APR and obtaining the information from that file. In addition, the APR includes information on how states classify providers by \u201cauthorized category of service,\u201d a classification scheme that can be used to identify certain non-medical provider categories not captured in the other classification systems, such as transportation or personal care service providers. For more information on the completeness of these fields in each state\u2019s APR file, see the DQ Atlas single-topic displays for Group and Individual Providers - Classification Types and Facilities - Classification Types .
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We examined the values for provider type, specialty, and taxonomy codes for the providers most examined in analyses for the TAF IP, LT, OT, and RX files (Table 1). [3] We included both fee-for-service (FFS) claims and managed care encounter records, which represent claims paid by managed care organizations and should generally follow the same reporting standards for billing and servicing providers. We excluded financial transaction records, supplemental payments, and “other” records that the state did not classify as being covered by either the Medicaid or CHIP programs. [4] We also excluded states from the analysis of each file if the number of header records in the TAF was low enough to be unusable. [5] For Illinois, we restricted our analysis to the original version of the claim and excluded all subsequent adjustment records in the state’s TAF data. [6]
Table 1. Crosswalk between file type, data element description, and TAF variable name
| File type | Data element description | Variable name |
|---|---|---|
| IP, LT, OT | Billing provider | BLG_PRVDR_TYPE_CD |
| IP, LT, OT, RX | Billing provider specialty code | BLG_PRVDR_SPCLTY_CD |
| IP, LT, OT, RX | Billing provider state-reported taxonomy code | BLG_PRVDR_TXNMY_CD |
| IP, LT, OT | Billing provider NPPES primary taxonomy code | BLG_PRVDR_NPPES_TXNMY_CD |
| OT | Servicing provider type code | SRVCNG_PRVDR_TYPE_CD |
| OT | Servicing provider specialty code | SRVCNG_PRVDR_SPCLTY_CD |
| OT | Servicing provider state-reported taxonomy code | SRVCNG_PRVDR_TXNMY_CD |
| OT | Servicing provider NPPES primary taxonomy code | SRVCNG_PRVDR_NPPES_TXNMY_CD |
The provider type code includes 57 different valid values, covering both facilities and professionals. As shown in Table 2, we grouped these values into those that are expected, unexpected, and unusable for billing providers for each of the three file types in which this data element is present (IP, LT, OT) and for servicing providers in the OT file.
In the IP file, we would expect to see only general hospitals and Indian Health Service facilities as billing providers. In the LT file, we would expect to see nursing facilities, intermediate care facilities for individuals with intellectual or developmental disabilities, and psychiatric facilities as billing providers, as well as general hospitals in selected cases. [7] In the OT file, we would expect to see a broad range of both facilities and professionals billing for and directly providing services.
For all three files, we examined the “all other” code separately from the expected and unexpected categories. In some instances, use of the non-specific “all other” code may reflect the possibility that the provider is not represented in the list of valid provider type codes, which is the case for certain types of home- and community-based services (HCBS) and non-emergency medical transport providers. Although this code is valid and not considered an unexpected value in any file, it does not provide any information about the billing or servicing provider and thus renders the data element unusable for analysis.
Table 2. Classification of provider type codes, by file and provider field
| File and provider field | Expected codes | Unusable codes (“all other” value) | Unexpected codes |
|---|---|---|---|
| IP—billing provider | 42, 51 | 57 | All other values |
| LT—billing provider | 42, 43, 44, 45 | 57 | All other values |
| OT—billing and servicing provider | 01−56 | 57 | None |
Source: A full list of values is available in the TAF Claims Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries . The provider type code values are identical across all three files.
The provider specialty code includes 115 different valid values, covering both facilities and professionals. As shown in Table 3, we grouped these values into those that are expected and unexpected for each of the file types and provider fields. In the IP file, we would expect to see only general hospitals as billing providers. In the LT file, we would expect to see nursing facilities, intermediate care facilities for individuals with intellectual or developmental disabilities, and psychiatric facilities as billing providers, as well as general hospitals in selected cases. In the OT file, we would expect to see a broad range of both facilities and professionals billing for and directly providing services. In the RX file, we would expect to see pharmacies, medical supply companies, department stores, and grocery stores as billing providers. For all files and provider fields, we examined the “all other” codes (billing provider specialty code 87—All Other Suppliers) separately from the expected and unexpected categories. Although this code is valid and not considered an unexpected value in any file, it does not provide any information about the provider and thus renders the data element unusable for analysis.
Table 3. Classification of provider specialty codes, by file and provider field
| File and provider field | Expected codes | Unusable codes (“all other” value) | Unexpected codes |
|---|---|---|---|
| IP—billing provider | A0 | 87 | All other values except 88 and 99, which are treated as missing |
| LT—billing provider | A0, A1, A2, A3, B4 | 87 | All other values except 88 and 99, which are treated as missing |
| OT—billing and servicing provider | 1−86, 89−98, A0−B5 | 87 | All other values except 88 and 99, which are treated as missing |
| RX—billing provider | 51–54, 58, 73, A5, A6, A9, B1, B3, B4 | 87 | All other values except 88 and 99, which are treated as missing |
Source: A full list of values is available in the TAF Claims Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries . The provider specialty code values are identical across all claims files.
The state-reported billing provider taxonomy code is a 10-character alphanumeric code that identifies the provider’s area of specialization. As shown in Table 4, we grouped these values into those that are expected and unexpected for each of the file types. In the IP file, we would expect to see inpatient hospitals as billing providers. In the LT files we would expect to see nursing facilities, intermediate care facility services for individuals with intellectual disabilities, mental health facility services, and independent (free-standing) psychiatric wings of acute care hospitals as billing providers. In the OT file, we would expect to see a broad range of both facilities and professionals billing for and directly providing services. In the RX file, we would expect to see pharmacies and other suppliers as billing providers.
In addition to the state-reported provider taxonomy codes, IP, OT and LT files produced in 2022 and later years also include a constructed taxonomy variable for billing and servicing providers, the NPPES primary taxonomy code (*_NPPES_TXNMY_CD), that represents the provider’s primary taxonomy information as reported in the National Plan and Provider Enumeration System (NPPES) based on the provider’s National Provider Identifier. We used the same criteria as the state-reported taxonomy codes to determine expected and unexpected values for each file type.
Table 4. Classification of provider taxonomy codes, by file and provider field
| File and provider field | Expected codes | Unexpected codes |
|---|---|---|
| IP—billing provider | First 2 digits of taxonomy code are 27 or 28 | All other values |
| LT—billing provider | Taxonomy code begins with 283Q, 283X, 282E, 31, 32, 385H, or 281P | All other values |
| OT—billing and servicing provider | All valid taxonomy codes | None |
| RX—billing provider | First 2 digits of taxonomy code are 33, first 3 digits are 183, or 302R00000X [8] | All other values |
Source: The health care provider taxonomy code set can be found at http://www.wpc-edi.com/reference/codelists/healthcare/health-care-provider-taxonomy-code-set/ . The provider taxonomy code values are identical across all claims files.
We grouped states into categories of low, medium, and high concern about the usability of their data, based on the percentage of header records (or for servicing providers in the OT file, claim lines) that had an expected provider type, specialty, state-reported taxonomy code, or NPPES primary taxonomy code for the file and provider fields (Table 5).
Because users of the OT file may be interested in professional or institutional (facility) claims, we conducted the DQ assessment separately for each of these claim types. Table 6 shows the methodology used to classify each claim header as a professional or institutional claim. Any claim that did not meet at least one of the criteria for a professional or institutional claim was considered unclassified and is not included in this analysis. Note that the servicing provider analyses were conducted at the claim line level, but the claims were classified as professional or institutional claims at the claim header level.
Table 5. Criteria for DQ assessment of provider type, specialty, and taxonomy code
| Percentage of records with an expected provider type, specialty, state-reported taxonomy code, or NPPES primary taxonomy code | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
Table 6. Classification of OT claim headers as professional or facility claims
| Claim category | Identification rules |
|---|---|
| Professional | The record must meet at least one of the following criteria:
|
| Institutional | The record must meet at least one of the following criteria:
|
Methods previously used to assess data quality
Table 7 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 7. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
|
|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We used claim type code (CLM_TYPE_CD) to determine which records to include and exclude in our analysis. Records included are those with a claim type code indicating FFS claims (values of 1 and A) and managed care encounters (3 and C). We excluded capitated payments (2 and B), supplemental payments (5 and E), and service-tracking claims (4 and D). We also used claim type code to exclude “other” records (values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information on supplemental payments and other claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
More information about the completeness of each state’s TAF data can be found in the DQ Atlas single topic displays for Claims Volume—IP , Claims Volume—LT , Claims Volume—OT , and Claims Volume—RX .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This approach means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only those records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
General hospitals may provide sub-acute care through “swing beds” or other non-acute units within the hospital, and we would expect these claims to be found in the LT file rather than the IP file.
At least one state (Florida) allows Health Maintenance Organizations (HMOs) to dispense pharmaceuticals to enrollees. Therefore, HMOs are included as an “expected” taxonomy for RX billing providers.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We used claim type code (CLM_TYPE_CD) to determine which records to include and exclude in our analysis. Records included are those with a claim type code indicating FFS claims (values of 1 and A) and managed care encounters (3 and C). We excluded capitated payments (2 and B), supplemental payments (5 and E), and service-tracking claims (4 and D). We also used claim type code to exclude \u201cother\u201d records (values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information on supplemental payments and other claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
More information about the completeness of each state\u2019s TAF data can be found in the DQ Atlas single topic displays for Claims Volume\u2014IP , Claims Volume\u2014LT , Claims Volume\u2014OT , and Claims Volume\u2014RX .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This approach means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only those records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
General hospitals may provide sub-acute care through \u201cswing beds\u201d or other non-acute units within the hospital, and we would expect these claims to be found in the LT file rather than the IP file.
At least one state (Florida) allows Health Maintenance Organizations (HMOs) to dispense pharmaceuticals to enrollees. Therefore, HMOs are included as an \u201cexpected\u201d taxonomy for RX billing providers.
The billing provider specialty and taxonomy codes allow TAF users to examine the characteristics of providers who bill and receive payments for Medicaid- and CHIP-funded services. This analysis examines the extent to which records in the RX file have a billing provider specialty or taxonomy missing or coded with unexpected or unusable values.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5081"", ""relatedTopics"": [{""measureId"": 37, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - IP"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 0}, {""measureId"": 38, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - LT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 1}, {""measureId"": 104, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - OT Professional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 2}, {""measureId"": 105, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - OT Institutional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 3}, {""measureId"": 106, ""measureName"": ""Servicing Provider Type, Specialty, and Taxonomy - OT Professional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 5}, {""measureId"": 107, ""measureName"": ""Servicing Provider Type, Specialty, and Taxonomy - OT Institutional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 6}]}" 104,"{""measureId"": 104, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - OT Professional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Bill-Prov-Type-OT-Prof.pdf"", ""background"": {""content"": ""Users of the T-MSIS Analytic Files (TAF) may want to identify the providers or categories of providers that deliver services to Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries. A claim in the TAF could include information for up to six providers, depending on the file type. [1] A provider could be a facility, a group of individual practitioners, or an individual practitioner. The provider types most often used for claims-based analyses include the billing, servicing, prescribing, and dispensing providers.
The billing provider represents the entity that submits the claim and is reimbursed by the Medicaid or CHIP agency. Information about the billing provider is available in all TAF claims files: inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX).
The servicing provider, in contrast, represents the individual practitioner who was responsible for or provided direct care to the beneficiary (that is, the rendering provider). Information about servicing provider is available in the IP, LT, and OT files. In some cases, the servicing provider is the same as the billing provider. In other cases, the servicing and billing provider differ, such as when a hospital or large group practice bills for services but an individual physician employed by the hospital or group provides the direct care and is the serving provider. Because institutional claims in the IP and LT files represent facility costs and are often submitted by the hospitals and skilled nursing facilities where the care occurred, analyses using these claims would typically only use the billing provider information.
For prescription drug claims, the billing provider is the pharmacy where the prescription was filled, the prescribing provider is the individual practitioner who prescribed a prescription drug to a beneficiary, and the dispensing provider is the pharmacist who filled the prescription or was responsible for overseeing the filling of the prescription. Because the billing provider represents the entity that submits a claim and is reimbursed by the Medicaid or CHIP agency, analyses using prescription drug claims typically focus on the billing provider information.
Data elements related to the billing provider can offer insight into the characteristics of providers who receive payment for Medicaid- and CHIP-funded services. Likewise, data elements related to the servicing provider can offer insight into the characteristics of the individual practitioner who was responsible for or provided direct care to a Medicaid or CHIP beneficiary. There are multiple systems available in TAF claim records for classifying providers, including the following:
Each of these classification systems may be best suited to different types of analyses. Although states are encouraged to populate all these fields on all claims, only one classification type is required. In practice, some states only submit information related to some of the classification systems, which may require TAF users to adjust their methodology across states. If the preferred data element has high rates of missingness, TAF users may be able to link the claims record to the provider’s record in the Annual Provider File (APR) to obtain the needed information. [2]
This data quality assessment examines whether any information on billing or servicing provider characteristics is available in claims records, and which type (taxonomy, specialty, or provider type). This information can help users design their analysis by selecting the fields populated in each state or identify states for which it may be necessary to link to the APR to obtain complete information about providers, rather than rely on claims alone.
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
States are required to submit information about all providers eligible to provide Medicaid- and CHIP-funded services; these records are captured in the APR. It is possible that missing information in the claim about provider taxonomy, specialty, or type could be imputed by linking to the provider’s record in the APR and obtaining the information from that file. In addition, the APR includes information on how states classify providers by “authorized category of service,” a classification scheme that can be used to identify certain non-medical provider categories not captured in the other classification systems, such as transportation or personal care service providers. For more information on the completeness of these fields in each state’s APR file, see the DQ Atlas single-topic displays for Group and Individual Providers - Classification Types and Facilities - Classification Types .
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
States are required to submit information about all providers eligible to provide Medicaid- and CHIP-funded services; these records are captured in the APR. It is possible that missing information in the claim about provider taxonomy, specialty, or type could be imputed by linking to the provider\u2019s record in the APR and obtaining the information from that file. In addition, the APR includes information on how states classify providers by \u201cauthorized category of service,\u201d a classification scheme that can be used to identify certain non-medical provider categories not captured in the other classification systems, such as transportation or personal care service providers. For more information on the completeness of these fields in each state\u2019s APR file, see the DQ Atlas single-topic displays for Group and Individual Providers - Classification Types and Facilities - Classification Types .
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We examined the values for provider type, specialty, and taxonomy codes for the providers most examined in analyses for the TAF IP, LT, OT, and RX files (Table 1). [3] We included both fee-for-service (FFS) claims and managed care encounter records, which represent claims paid by managed care organizations and should generally follow the same reporting standards for billing and servicing providers. We excluded financial transaction records, supplemental payments, and “other” records that the state did not classify as being covered by either the Medicaid or CHIP programs. [4] We also excluded states from the analysis of each file if the number of header records in the TAF was low enough to be unusable. [5] For Illinois, we restricted our analysis to the original version of the claim and excluded all subsequent adjustment records in the state’s TAF data. [6]
Table 1. Crosswalk between file type, data element description, and TAF variable name
| File type | Data element description | Variable name |
|---|---|---|
| IP, LT, OT | Billing provider | BLG_PRVDR_TYPE_CD |
| IP, LT, OT, RX | Billing provider specialty code | BLG_PRVDR_SPCLTY_CD |
| IP, LT, OT, RX | Billing provider state-reported taxonomy code | BLG_PRVDR_TXNMY_CD |
| IP, LT, OT | Billing provider NPPES primary taxonomy code | BLG_PRVDR_NPPES_TXNMY_CD |
| OT | Servicing provider type code | SRVCNG_PRVDR_TYPE_CD |
| OT | Servicing provider specialty code | SRVCNG_PRVDR_SPCLTY_CD |
| OT | Servicing provider state-reported taxonomy code | SRVCNG_PRVDR_TXNMY_CD |
| OT | Servicing provider NPPES primary taxonomy code | SRVCNG_PRVDR_NPPES_TXNMY_CD |
The provider type code includes 57 different valid values, covering both facilities and professionals. As shown in Table 2, we grouped these values into those that are expected, unexpected, and unusable for billing providers for each of the three file types in which this data element is present (IP, LT, OT) and for servicing providers in the OT file.
In the IP file, we would expect to see only general hospitals and Indian Health Service facilities as billing providers. In the LT file, we would expect to see nursing facilities, intermediate care facilities for individuals with intellectual or developmental disabilities, and psychiatric facilities as billing providers, as well as general hospitals in selected cases. [7] In the OT file, we would expect to see a broad range of both facilities and professionals billing for and directly providing services.
For all three files, we examined the “all other” code separately from the expected and unexpected categories. In some instances, use of the non-specific “all other” code may reflect the possibility that the provider is not represented in the list of valid provider type codes, which is the case for certain types of home- and community-based services (HCBS) and non-emergency medical transport providers. Although this code is valid and not considered an unexpected value in any file, it does not provide any information about the billing or servicing provider and thus renders the data element unusable for analysis.
Table 2. Classification of provider type codes, by file and provider field
| File and provider field | Expected codes | Unusable codes (“all other” value) | Unexpected codes |
|---|---|---|---|
| IP—billing provider | 42, 51 | 57 | All other values |
| LT—billing provider | 42, 43, 44, 45 | 57 | All other values |
| OT—billing and servicing provider | 01−56 | 57 | None |
Source: A full list of values is available in the TAF Claims Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries . The provider type code values are identical across all three files.
The provider specialty code includes 115 different valid values, covering both facilities and professionals. As shown in Table 3, we grouped these values into those that are expected and unexpected for each of the file types and provider fields. In the IP file, we would expect to see only general hospitals as billing providers. In the LT file, we would expect to see nursing facilities, intermediate care facilities for individuals with intellectual or developmental disabilities, and psychiatric facilities as billing providers, as well as general hospitals in selected cases. In the OT file, we would expect to see a broad range of both facilities and professionals billing for and directly providing services. In the RX file, we would expect to see pharmacies, medical supply companies, department stores, and grocery stores as billing providers. For all files and provider fields, we examined the “all other” codes (billing provider specialty code 87—All Other Suppliers) separately from the expected and unexpected categories. Although this code is valid and not considered an unexpected value in any file, it does not provide any information about the provider and thus renders the data element unusable for analysis.
Table 3. Classification of provider specialty codes, by file and provider field
| File and provider field | Expected codes | Unusable codes (“all other” value) | Unexpected codes |
|---|---|---|---|
| IP—billing provider | A0 | 87 | All other values except 88 and 99, which are treated as missing |
| LT—billing provider | A0, A1, A2, A3, B4 | 87 | All other values except 88 and 99, which are treated as missing |
| OT—billing and servicing provider | 1−86, 89−98, A0−B5 | 87 | All other values except 88 and 99, which are treated as missing |
| RX—billing provider | 51–54, 58, 73, A5, A6, A9, B1, B3, B4 | 87 | All other values except 88 and 99, which are treated as missing |
Source: A full list of values is available in the TAF Claims Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries . The provider specialty code values are identical across all claims files.
The state-reported billing provider taxonomy code is a 10-character alphanumeric code that identifies the provider’s area of specialization. As shown in Table 4, we grouped these values into those that are expected and unexpected for each of the file types. In the IP file, we would expect to see inpatient hospitals as billing providers. In the LT files we would expect to see nursing facilities, intermediate care facility services for individuals with intellectual disabilities, mental health facility services, and independent (free-standing) psychiatric wings of acute care hospitals as billing providers. In the OT file, we would expect to see a broad range of both facilities and professionals billing for and directly providing services. In the RX file, we would expect to see pharmacies and other suppliers as billing providers.
In addition to the state-reported provider taxonomy codes, IP, OT and LT files produced in 2022 and later years also include a constructed taxonomy variable for billing and servicing providers, the NPPES primary taxonomy code (*_NPPES_TXNMY_CD), that represents the provider’s primary taxonomy information as reported in the National Plan and Provider Enumeration System (NPPES) based on the provider’s National Provider Identifier. We used the same criteria as the state-reported taxonomy codes to determine expected and unexpected values for each file type.
Table 4. Classification of provider taxonomy codes, by file and provider field
| File and provider field | Expected codes | Unexpected codes |
|---|---|---|
| IP—billing provider | First 2 digits of taxonomy code are 27 or 28 | All other values |
| LT—billing provider | Taxonomy code begins with 283Q, 283X, 282E, 31, 32, 385H, or 281P | All other values |
| OT—billing and servicing provider | All valid taxonomy codes | None |
| RX—billing provider | First 2 digits of taxonomy code are 33, first 3 digits are 183, or 302R00000X [8] | All other values |
Source: The health care provider taxonomy code set can be found at http://www.wpc-edi.com/reference/codelists/healthcare/health-care-provider-taxonomy-code-set/ . The provider taxonomy code values are identical across all claims files.
We grouped states into categories of low, medium, and high concern about the usability of their data, based on the percentage of header records (or for servicing providers in the OT file, claim lines) that had an expected provider type, specialty, state-reported taxonomy code, or NPPES primary taxonomy code for the file and provider fields (Table 5).
Because users of the OT file may be interested in professional or institutional (facility) claims, we conducted the DQ assessment separately for each of these claim types. Table 6 shows the methodology used to classify each claim header as a professional or institutional claim. Any claim that did not meet at least one of the criteria for a professional or institutional claim was considered unclassified and is not included in this analysis. Note that the servicing provider analyses were conducted at the claim line level, but the claims were classified as professional or institutional claims at the claim header level.
Table 5. Criteria for DQ assessment of provider type, specialty, and taxonomy code
| Percentage of records with an expected provider type, specialty, state-reported taxonomy code, or NPPES primary taxonomy code | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
Table 6. Classification of OT claim headers as professional or facility claims
| Claim category | Identification rules |
|---|---|
| Professional | The record must meet at least one of the following criteria:
|
| Institutional | The record must meet at least one of the following criteria:
|
Methods previously used to assess data quality
Table 7 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 7. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
|
|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We used claim type code (CLM_TYPE_CD) to determine which records to include and exclude in our analysis. Records included are those with a claim type code indicating FFS claims (values of 1 and A) and managed care encounters (3 and C). We excluded capitated payments (2 and B), supplemental payments (5 and E), and service-tracking claims (4 and D). We also used claim type code to exclude “other” records (values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information on supplemental payments and other claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
More information about the completeness of each state’s TAF data can be found in the DQ Atlas single topic displays for Claims Volume—IP , Claims Volume—LT , Claims Volume—OT , and Claims Volume—RX .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This approach means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only those records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
General hospitals may provide sub-acute care through “swing beds” or other non-acute units within the hospital, and we would expect these claims to be found in the LT file rather than the IP file.
At least one state (Florida) allows Health Maintenance Organizations (HMOs) to dispense pharmaceuticals to enrollees. Therefore, HMOs are included as an “expected” taxonomy for RX billing providers.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We used claim type code (CLM_TYPE_CD) to determine which records to include and exclude in our analysis. Records included are those with a claim type code indicating FFS claims (values of 1 and A) and managed care encounters (3 and C). We excluded capitated payments (2 and B), supplemental payments (5 and E), and service-tracking claims (4 and D). We also used claim type code to exclude \u201cother\u201d records (values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information on supplemental payments and other claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
More information about the completeness of each state\u2019s TAF data can be found in the DQ Atlas single topic displays for Claims Volume\u2014IP , Claims Volume\u2014LT , Claims Volume\u2014OT , and Claims Volume\u2014RX .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This approach means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only those records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
General hospitals may provide sub-acute care through \u201cswing beds\u201d or other non-acute units within the hospital, and we would expect these claims to be found in the LT file rather than the IP file.
At least one state (Florida) allows Health Maintenance Organizations (HMOs) to dispense pharmaceuticals to enrollees. Therefore, HMOs are included as an \u201cexpected\u201d taxonomy for RX billing providers.
The billing provider type, specialty, and taxonomy codes allow TAF users to examine the characteristics of providers who bill and receive payments for Medicaid- and CHIP-funded services. This analysis examines the extent to which professional claims in the OT file have a billing provider type, specialty, or taxonomy missing or coded with unexpected or unusable values.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5081"", ""relatedTopics"": [{""measureId"": 37, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - IP"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 0}, {""measureId"": 38, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - LT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 1}, {""measureId"": 105, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - OT Institutional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 3}, {""measureId"": 103, ""measureName"": ""Billing Provider Specialty and Taxonomy - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 4}, {""measureId"": 106, ""measureName"": ""Servicing Provider Type, Specialty, and Taxonomy - OT Professional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 5}, {""measureId"": 107, ""measureName"": ""Servicing Provider Type, Specialty, and Taxonomy - OT Institutional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 6}]}" 105,"{""measureId"": 105, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - OT Institutional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Bill-Prov-Type-OT-Inst.pdf"", ""background"": {""content"": ""Users of the T-MSIS Analytic Files (TAF) may want to identify the providers or categories of providers that deliver services to Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries. A claim in the TAF could include information for up to six providers, depending on the file type. [1] A provider could be a facility, a group of individual practitioners, or an individual practitioner. The provider types most often used for claims-based analyses include the billing, servicing, prescribing, and dispensing providers.
The billing provider represents the entity that submits the claim and is reimbursed by the Medicaid or CHIP agency. Information about the billing provider is available in all TAF claims files: inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX).
The servicing provider, in contrast, represents the individual practitioner who was responsible for or provided direct care to the beneficiary (that is, the rendering provider). Information about servicing provider is available in the IP, LT, and OT files. In some cases, the servicing provider is the same as the billing provider. In other cases, the servicing and billing provider differ, such as when a hospital or large group practice bills for services but an individual physician employed by the hospital or group provides the direct care and is the serving provider. Because institutional claims in the IP and LT files represent facility costs and are often submitted by the hospitals and skilled nursing facilities where the care occurred, analyses using these claims would typically only use the billing provider information.
For prescription drug claims, the billing provider is the pharmacy where the prescription was filled, the prescribing provider is the individual practitioner who prescribed a prescription drug to a beneficiary, and the dispensing provider is the pharmacist who filled the prescription or was responsible for overseeing the filling of the prescription. Because the billing provider represents the entity that submits a claim and is reimbursed by the Medicaid or CHIP agency, analyses using prescription drug claims typically focus on the billing provider information.
Data elements related to the billing provider can offer insight into the characteristics of providers who receive payment for Medicaid- and CHIP-funded services. Likewise, data elements related to the servicing provider can offer insight into the characteristics of the individual practitioner who was responsible for or provided direct care to a Medicaid or CHIP beneficiary. There are multiple systems available in TAF claim records for classifying providers, including the following:
Each of these classification systems may be best suited to different types of analyses. Although states are encouraged to populate all these fields on all claims, only one classification type is required. In practice, some states only submit information related to some of the classification systems, which may require TAF users to adjust their methodology across states. If the preferred data element has high rates of missingness, TAF users may be able to link the claims record to the provider’s record in the Annual Provider File (APR) to obtain the needed information. [2]
This data quality assessment examines whether any information on billing or servicing provider characteristics is available in claims records, and which type (taxonomy, specialty, or provider type). This information can help users design their analysis by selecting the fields populated in each state or identify states for which it may be necessary to link to the APR to obtain complete information about providers, rather than rely on claims alone.
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
States are required to submit information about all providers eligible to provide Medicaid- and CHIP-funded services; these records are captured in the APR. It is possible that missing information in the claim about provider taxonomy, specialty, or type could be imputed by linking to the provider’s record in the APR and obtaining the information from that file. In addition, the APR includes information on how states classify providers by “authorized category of service,” a classification scheme that can be used to identify certain non-medical provider categories not captured in the other classification systems, such as transportation or personal care service providers. For more information on the completeness of these fields in each state’s APR file, see the DQ Atlas single-topic displays for Group and Individual Providers - Classification Types and Facilities - Classification Types .
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
States are required to submit information about all providers eligible to provide Medicaid- and CHIP-funded services; these records are captured in the APR. It is possible that missing information in the claim about provider taxonomy, specialty, or type could be imputed by linking to the provider\u2019s record in the APR and obtaining the information from that file. In addition, the APR includes information on how states classify providers by \u201cauthorized category of service,\u201d a classification scheme that can be used to identify certain non-medical provider categories not captured in the other classification systems, such as transportation or personal care service providers. For more information on the completeness of these fields in each state\u2019s APR file, see the DQ Atlas single-topic displays for Group and Individual Providers - Classification Types and Facilities - Classification Types .
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We examined the values for provider type, specialty, and taxonomy codes for the providers most examined in analyses for the TAF IP, LT, OT, and RX files (Table 1). [3] We included both fee-for-service (FFS) claims and managed care encounter records, which represent claims paid by managed care organizations and should generally follow the same reporting standards for billing and servicing providers. We excluded financial transaction records, supplemental payments, and “other” records that the state did not classify as being covered by either the Medicaid or CHIP programs. [4] We also excluded states from the analysis of each file if the number of header records in the TAF was low enough to be unusable. [5] For Illinois, we restricted our analysis to the original version of the claim and excluded all subsequent adjustment records in the state’s TAF data. [6]
Table 1. Crosswalk between file type, data element description, and TAF variable name
| File type | Data element description | Variable name |
|---|---|---|
| IP, LT, OT | Billing provider | BLG_PRVDR_TYPE_CD |
| IP, LT, OT, RX | Billing provider specialty code | BLG_PRVDR_SPCLTY_CD |
| IP, LT, OT, RX | Billing provider state-reported taxonomy code | BLG_PRVDR_TXNMY_CD |
| IP, LT, OT | Billing provider NPPES primary taxonomy code | BLG_PRVDR_NPPES_TXNMY_CD |
| OT | Servicing provider type code | SRVCNG_PRVDR_TYPE_CD |
| OT | Servicing provider specialty code | SRVCNG_PRVDR_SPCLTY_CD |
| OT | Servicing provider state-reported taxonomy code | SRVCNG_PRVDR_TXNMY_CD |
| OT | Servicing provider NPPES primary taxonomy code | SRVCNG_PRVDR_NPPES_TXNMY_CD |
The provider type code includes 57 different valid values, covering both facilities and professionals. As shown in Table 2, we grouped these values into those that are expected, unexpected, and unusable for billing providers for each of the three file types in which this data element is present (IP, LT, OT) and for servicing providers in the OT file.
In the IP file, we would expect to see only general hospitals and Indian Health Service facilities as billing providers. In the LT file, we would expect to see nursing facilities, intermediate care facilities for individuals with intellectual or developmental disabilities, and psychiatric facilities as billing providers, as well as general hospitals in selected cases. [7] In the OT file, we would expect to see a broad range of both facilities and professionals billing for and directly providing services.
For all three files, we examined the “all other” code separately from the expected and unexpected categories. In some instances, use of the non-specific “all other” code may reflect the possibility that the provider is not represented in the list of valid provider type codes, which is the case for certain types of home- and community-based services (HCBS) and non-emergency medical transport providers. Although this code is valid and not considered an unexpected value in any file, it does not provide any information about the billing or servicing provider and thus renders the data element unusable for analysis.
Table 2. Classification of provider type codes, by file and provider field
| File and provider field | Expected codes | Unusable codes (“all other” value) | Unexpected codes |
|---|---|---|---|
| IP—billing provider | 42, 51 | 57 | All other values |
| LT—billing provider | 42, 43, 44, 45 | 57 | All other values |
| OT—billing and servicing provider | 01−56 | 57 | None |
Source: A full list of values is available in the TAF Claims Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries . The provider type code values are identical across all three files.
The provider specialty code includes 115 different valid values, covering both facilities and professionals. As shown in Table 3, we grouped these values into those that are expected and unexpected for each of the file types and provider fields. In the IP file, we would expect to see only general hospitals as billing providers. In the LT file, we would expect to see nursing facilities, intermediate care facilities for individuals with intellectual or developmental disabilities, and psychiatric facilities as billing providers, as well as general hospitals in selected cases. In the OT file, we would expect to see a broad range of both facilities and professionals billing for and directly providing services. In the RX file, we would expect to see pharmacies, medical supply companies, department stores, and grocery stores as billing providers. For all files and provider fields, we examined the “all other” codes (billing provider specialty code 87—All Other Suppliers) separately from the expected and unexpected categories. Although this code is valid and not considered an unexpected value in any file, it does not provide any information about the provider and thus renders the data element unusable for analysis.
Table 3. Classification of provider specialty codes, by file and provider field
| File and provider field | Expected codes | Unusable codes (“all other” value) | Unexpected codes |
|---|---|---|---|
| IP—billing provider | A0 | 87 | All other values except 88 and 99, which are treated as missing |
| LT—billing provider | A0, A1, A2, A3, B4 | 87 | All other values except 88 and 99, which are treated as missing |
| OT—billing and servicing provider | 1−86, 89−98, A0−B5 | 87 | All other values except 88 and 99, which are treated as missing |
| RX—billing provider | 51–54, 58, 73, A5, A6, A9, B1, B3, B4 | 87 | All other values except 88 and 99, which are treated as missing |
Source: A full list of values is available in the TAF Claims Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries . The provider specialty code values are identical across all claims files.
The state-reported billing provider taxonomy code is a 10-character alphanumeric code that identifies the provider’s area of specialization. As shown in Table 4, we grouped these values into those that are expected and unexpected for each of the file types. In the IP file, we would expect to see inpatient hospitals as billing providers. In the LT files we would expect to see nursing facilities, intermediate care facility services for individuals with intellectual disabilities, mental health facility services, and independent (free-standing) psychiatric wings of acute care hospitals as billing providers. In the OT file, we would expect to see a broad range of both facilities and professionals billing for and directly providing services. In the RX file, we would expect to see pharmacies and other suppliers as billing providers.
In addition to the state-reported provider taxonomy codes, IP, OT and LT files produced in 2022 and later years also include a constructed taxonomy variable for billing and servicing providers, the NPPES primary taxonomy code (*_NPPES_TXNMY_CD), that represents the provider’s primary taxonomy information as reported in the National Plan and Provider Enumeration System (NPPES) based on the provider’s National Provider Identifier. We used the same criteria as the state-reported taxonomy codes to determine expected and unexpected values for each file type.
Table 4. Classification of provider taxonomy codes, by file and provider field
| File and provider field | Expected codes | Unexpected codes |
|---|---|---|
| IP—billing provider | First 2 digits of taxonomy code are 27 or 28 | All other values |
| LT—billing provider | Taxonomy code begins with 283Q, 283X, 282E, 31, 32, 385H, or 281P | All other values |
| OT—billing and servicing provider | All valid taxonomy codes | None |
| RX—billing provider | First 2 digits of taxonomy code are 33, first 3 digits are 183, or 302R00000X [8] | All other values |
Source: The health care provider taxonomy code set can be found at http://www.wpc-edi.com/reference/codelists/healthcare/health-care-provider-taxonomy-code-set/ . The provider taxonomy code values are identical across all claims files.
We grouped states into categories of low, medium, and high concern about the usability of their data, based on the percentage of header records (or for servicing providers in the OT file, claim lines) that had an expected provider type, specialty, state-reported taxonomy code, or NPPES primary taxonomy code for the file and provider fields (Table 5).
Because users of the OT file may be interested in professional or institutional (facility) claims, we conducted the DQ assessment separately for each of these claim types. Table 6 shows the methodology used to classify each claim header as a professional or institutional claim. Any claim that did not meet at least one of the criteria for a professional or institutional claim was considered unclassified and is not included in this analysis. Note that the servicing provider analyses were conducted at the claim line level, but the claims were classified as professional or institutional claims at the claim header level.
Table 5. Criteria for DQ assessment of provider type, specialty, and taxonomy code
| Percentage of records with an expected provider type, specialty, state-reported taxonomy code, or NPPES primary taxonomy code | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
Table 6. Classification of OT claim headers as professional or facility claims
| Claim category | Identification rules |
|---|---|
| Professional | The record must meet at least one of the following criteria:
|
| Institutional | The record must meet at least one of the following criteria:
|
Methods previously used to assess data quality
Table 7 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 7. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
|
|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We used claim type code (CLM_TYPE_CD) to determine which records to include and exclude in our analysis. Records included are those with a claim type code indicating FFS claims (values of 1 and A) and managed care encounters (3 and C). We excluded capitated payments (2 and B), supplemental payments (5 and E), and service-tracking claims (4 and D). We also used claim type code to exclude “other” records (values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information on supplemental payments and other claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
More information about the completeness of each state’s TAF data can be found in the DQ Atlas single topic displays for Claims Volume—IP , Claims Volume—LT , Claims Volume—OT , and Claims Volume—RX .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This approach means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only those records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
General hospitals may provide sub-acute care through “swing beds” or other non-acute units within the hospital, and we would expect these claims to be found in the LT file rather than the IP file.
At least one state (Florida) allows Health Maintenance Organizations (HMOs) to dispense pharmaceuticals to enrollees. Therefore, HMOs are included as an “expected” taxonomy for RX billing providers.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We used claim type code (CLM_TYPE_CD) to determine which records to include and exclude in our analysis. Records included are those with a claim type code indicating FFS claims (values of 1 and A) and managed care encounters (3 and C). We excluded capitated payments (2 and B), supplemental payments (5 and E), and service-tracking claims (4 and D). We also used claim type code to exclude \u201cother\u201d records (values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information on supplemental payments and other claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
More information about the completeness of each state\u2019s TAF data can be found in the DQ Atlas single topic displays for Claims Volume\u2014IP , Claims Volume\u2014LT , Claims Volume\u2014OT , and Claims Volume\u2014RX .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This approach means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only those records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
General hospitals may provide sub-acute care through \u201cswing beds\u201d or other non-acute units within the hospital, and we would expect these claims to be found in the LT file rather than the IP file.
At least one state (Florida) allows Health Maintenance Organizations (HMOs) to dispense pharmaceuticals to enrollees. Therefore, HMOs are included as an \u201cexpected\u201d taxonomy for RX billing providers.
The billing provider type, specialty, and taxonomy codes allow TAF users to examine the characteristics of providers who bill and receive payments for Medicaid- and CHIP-funded services. This analysis examines the extent to which facility claims in the OT file have a billing provider type, specialty, or taxonomy missing or coded with unexpected or unusable values.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5081"", ""relatedTopics"": [{""measureId"": 37, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - IP"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 0}, {""measureId"": 38, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - LT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 1}, {""measureId"": 104, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - OT Professional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 2}, {""measureId"": 103, ""measureName"": ""Billing Provider Specialty and Taxonomy - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 4}, {""measureId"": 106, ""measureName"": ""Servicing Provider Type, Specialty, and Taxonomy - OT Professional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 5}, {""measureId"": 107, ""measureName"": ""Servicing Provider Type, Specialty, and Taxonomy - OT Institutional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 6}]}" 106,"{""measureId"": 106, ""measureName"": ""Servicing Provider Type, Specialty, and Taxonomy - OT Professional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Serv-Prov-Type-OT-Prof.pdf"", ""background"": {""content"": ""Users of the T-MSIS Analytic Files (TAF) may want to identify the providers or categories of providers that deliver services to Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries. A claim in the TAF could include information for up to six providers, depending on the file type. [1] A provider could be a facility, a group of individual practitioners, or an individual practitioner. The provider types most often used for claims-based analyses include the billing, servicing, prescribing, and dispensing providers.
The billing provider represents the entity that submits the claim and is reimbursed by the Medicaid or CHIP agency. Information about the billing provider is available in all TAF claims files: inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX).
The servicing provider, in contrast, represents the individual practitioner who was responsible for or provided direct care to the beneficiary (that is, the rendering provider). Information about servicing provider is available in the IP, LT, and OT files. In some cases, the servicing provider is the same as the billing provider. In other cases, the servicing and billing provider differ, such as when a hospital or large group practice bills for services but an individual physician employed by the hospital or group provides the direct care and is the serving provider. Because institutional claims in the IP and LT files represent facility costs and are often submitted by the hospitals and skilled nursing facilities where the care occurred, analyses using these claims would typically only use the billing provider information.
For prescription drug claims, the billing provider is the pharmacy where the prescription was filled, the prescribing provider is the individual practitioner who prescribed a prescription drug to a beneficiary, and the dispensing provider is the pharmacist who filled the prescription or was responsible for overseeing the filling of the prescription. Because the billing provider represents the entity that submits a claim and is reimbursed by the Medicaid or CHIP agency, analyses using prescription drug claims typically focus on the billing provider information.
Data elements related to the billing provider can offer insight into the characteristics of providers who receive payment for Medicaid- and CHIP-funded services. Likewise, data elements related to the servicing provider can offer insight into the characteristics of the individual practitioner who was responsible for or provided direct care to a Medicaid or CHIP beneficiary. There are multiple systems available in TAF claim records for classifying providers, including the following:
Each of these classification systems may be best suited to different types of analyses. Although states are encouraged to populate all these fields on all claims, only one classification type is required. In practice, some states only submit information related to some of the classification systems, which may require TAF users to adjust their methodology across states. If the preferred data element has high rates of missingness, TAF users may be able to link the claims record to the provider’s record in the Annual Provider File (APR) to obtain the needed information. [2]
This data quality assessment examines whether any information on billing or servicing provider characteristics is available in claims records, and which type (taxonomy, specialty, or provider type). This information can help users design their analysis by selecting the fields populated in each state or identify states for which it may be necessary to link to the APR to obtain complete information about providers, rather than rely on claims alone.
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
States are required to submit information about all providers eligible to provide Medicaid- and CHIP-funded services; these records are captured in the APR. It is possible that missing information in the claim about provider taxonomy, specialty, or type could be imputed by linking to the provider’s record in the APR and obtaining the information from that file. In addition, the APR includes information on how states classify providers by “authorized category of service,” a classification scheme that can be used to identify certain non-medical provider categories not captured in the other classification systems, such as transportation or personal care service providers. For more information on the completeness of these fields in each state’s APR file, see the DQ Atlas single-topic displays for Group and Individual Providers - Classification Types and Facilities - Classification Types .
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
States are required to submit information about all providers eligible to provide Medicaid- and CHIP-funded services; these records are captured in the APR. It is possible that missing information in the claim about provider taxonomy, specialty, or type could be imputed by linking to the provider\u2019s record in the APR and obtaining the information from that file. In addition, the APR includes information on how states classify providers by \u201cauthorized category of service,\u201d a classification scheme that can be used to identify certain non-medical provider categories not captured in the other classification systems, such as transportation or personal care service providers. For more information on the completeness of these fields in each state\u2019s APR file, see the DQ Atlas single-topic displays for Group and Individual Providers - Classification Types and Facilities - Classification Types .
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We examined the values for provider type, specialty, and taxonomy codes for the providers most examined in analyses for the TAF IP, LT, OT, and RX files (Table 1). [3] We included both fee-for-service (FFS) claims and managed care encounter records, which represent claims paid by managed care organizations and should generally follow the same reporting standards for billing and servicing providers. We excluded financial transaction records, supplemental payments, and “other” records that the state did not classify as being covered by either the Medicaid or CHIP programs. [4] We also excluded states from the analysis of each file if the number of header records in the TAF was low enough to be unusable. [5] For Illinois, we restricted our analysis to the original version of the claim and excluded all subsequent adjustment records in the state’s TAF data. [6]
Table 1. Crosswalk between file type, data element description, and TAF variable name
| File type | Data element description | Variable name |
|---|---|---|
| IP, LT, OT | Billing provider | BLG_PRVDR_TYPE_CD |
| IP, LT, OT, RX | Billing provider specialty code | BLG_PRVDR_SPCLTY_CD |
| IP, LT, OT, RX | Billing provider state-reported taxonomy code | BLG_PRVDR_TXNMY_CD |
| IP, LT, OT | Billing provider NPPES primary taxonomy code | BLG_PRVDR_NPPES_TXNMY_CD |
| OT | Servicing provider type code | SRVCNG_PRVDR_TYPE_CD |
| OT | Servicing provider specialty code | SRVCNG_PRVDR_SPCLTY_CD |
| OT | Servicing provider state-reported taxonomy code | SRVCNG_PRVDR_TXNMY_CD |
| OT | Servicing provider NPPES primary taxonomy code | SRVCNG_PRVDR_NPPES_TXNMY_CD |
The provider type code includes 57 different valid values, covering both facilities and professionals. As shown in Table 2, we grouped these values into those that are expected, unexpected, and unusable for billing providers for each of the three file types in which this data element is present (IP, LT, OT) and for servicing providers in the OT file.
In the IP file, we would expect to see only general hospitals and Indian Health Service facilities as billing providers. In the LT file, we would expect to see nursing facilities, intermediate care facilities for individuals with intellectual or developmental disabilities, and psychiatric facilities as billing providers, as well as general hospitals in selected cases. [7] In the OT file, we would expect to see a broad range of both facilities and professionals billing for and directly providing services.
For all three files, we examined the “all other” code separately from the expected and unexpected categories. In some instances, use of the non-specific “all other” code may reflect the possibility that the provider is not represented in the list of valid provider type codes, which is the case for certain types of home- and community-based services (HCBS) and non-emergency medical transport providers. Although this code is valid and not considered an unexpected value in any file, it does not provide any information about the billing or servicing provider and thus renders the data element unusable for analysis.
Table 2. Classification of provider type codes, by file and provider field
| File and provider field | Expected codes | Unusable codes (“all other” value) | Unexpected codes |
|---|---|---|---|
| IP—billing provider | 42, 51 | 57 | All other values |
| LT—billing provider | 42, 43, 44, 45 | 57 | All other values |
| OT—billing and servicing provider | 01−56 | 57 | None |
Source: A full list of values is available in the TAF Claims Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries . The provider type code values are identical across all three files.
The provider specialty code includes 115 different valid values, covering both facilities and professionals. As shown in Table 3, we grouped these values into those that are expected and unexpected for each of the file types and provider fields. In the IP file, we would expect to see only general hospitals as billing providers. In the LT file, we would expect to see nursing facilities, intermediate care facilities for individuals with intellectual or developmental disabilities, and psychiatric facilities as billing providers, as well as general hospitals in selected cases. In the OT file, we would expect to see a broad range of both facilities and professionals billing for and directly providing services. In the RX file, we would expect to see pharmacies, medical supply companies, department stores, and grocery stores as billing providers. For all files and provider fields, we examined the “all other” codes (billing provider specialty code 87—All Other Suppliers) separately from the expected and unexpected categories. Although this code is valid and not considered an unexpected value in any file, it does not provide any information about the provider and thus renders the data element unusable for analysis.
Table 3. Classification of provider specialty codes, by file and provider field
| File and provider field | Expected codes | Unusable codes (“all other” value) | Unexpected codes |
|---|---|---|---|
| IP—billing provider | A0 | 87 | All other values except 88 and 99, which are treated as missing |
| LT—billing provider | A0, A1, A2, A3, B4 | 87 | All other values except 88 and 99, which are treated as missing |
| OT—billing and servicing provider | 1−86, 89−98, A0−B5 | 87 | All other values except 88 and 99, which are treated as missing |
| RX—billing provider | 51–54, 58, 73, A5, A6, A9, B1, B3, B4 | 87 | All other values except 88 and 99, which are treated as missing |
Source: A full list of values is available in the TAF Claims Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries . The provider specialty code values are identical across all claims files.
The state-reported billing provider taxonomy code is a 10-character alphanumeric code that identifies the provider’s area of specialization. As shown in Table 4, we grouped these values into those that are expected and unexpected for each of the file types. In the IP file, we would expect to see inpatient hospitals as billing providers. In the LT files we would expect to see nursing facilities, intermediate care facility services for individuals with intellectual disabilities, mental health facility services, and independent (free-standing) psychiatric wings of acute care hospitals as billing providers. In the OT file, we would expect to see a broad range of both facilities and professionals billing for and directly providing services. In the RX file, we would expect to see pharmacies and other suppliers as billing providers.
In addition to the state-reported provider taxonomy codes, IP, OT and LT files produced in 2022 and later years also include a constructed taxonomy variable for billing and servicing providers, the NPPES primary taxonomy code (*_NPPES_TXNMY_CD), that represents the provider’s primary taxonomy information as reported in the National Plan and Provider Enumeration System (NPPES) based on the provider’s National Provider Identifier. We used the same criteria as the state-reported taxonomy codes to determine expected and unexpected values for each file type.
Table 4. Classification of provider taxonomy codes, by file and provider field
| File and provider field | Expected codes | Unexpected codes |
|---|---|---|
| IP—billing provider | First 2 digits of taxonomy code are 27 or 28 | All other values |
| LT—billing provider | Taxonomy code begins with 283Q, 283X, 282E, 31, 32, 385H, or 281P | All other values |
| OT—billing and servicing provider | All valid taxonomy codes | None |
| RX—billing provider | First 2 digits of taxonomy code are 33, first 3 digits are 183, or 302R00000X [8] | All other values |
Source: The health care provider taxonomy code set can be found at http://www.wpc-edi.com/reference/codelists/healthcare/health-care-provider-taxonomy-code-set/ . The provider taxonomy code values are identical across all claims files.
We grouped states into categories of low, medium, and high concern about the usability of their data, based on the percentage of header records (or for servicing providers in the OT file, claim lines) that had an expected provider type, specialty, state-reported taxonomy code, or NPPES primary taxonomy code for the file and provider fields (Table 5).
Because users of the OT file may be interested in professional or institutional (facility) claims, we conducted the DQ assessment separately for each of these claim types. Table 6 shows the methodology used to classify each claim header as a professional or institutional claim. Any claim that did not meet at least one of the criteria for a professional or institutional claim was considered unclassified and is not included in this analysis. Note that the servicing provider analyses were conducted at the claim line level, but the claims were classified as professional or institutional claims at the claim header level.
Table 5. Criteria for DQ assessment of provider type, specialty, and taxonomy code
| Percentage of records with an expected provider type, specialty, state-reported taxonomy code, or NPPES primary taxonomy code | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
Table 6. Classification of OT claim headers as professional or facility claims
| Claim category | Identification rules |
|---|---|
| Professional | The record must meet at least one of the following criteria:
|
| Institutional | The record must meet at least one of the following criteria:
|
Methods previously used to assess data quality
Table 7 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 7. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
|
|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We used claim type code (CLM_TYPE_CD) to determine which records to include and exclude in our analysis. Records included are those with a claim type code indicating FFS claims (values of 1 and A) and managed care encounters (3 and C). We excluded capitated payments (2 and B), supplemental payments (5 and E), and service-tracking claims (4 and D). We also used claim type code to exclude “other” records (values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information on supplemental payments and other claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
More information about the completeness of each state’s TAF data can be found in the DQ Atlas single topic displays for Claims Volume—IP , Claims Volume—LT , Claims Volume—OT , and Claims Volume—RX .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This approach means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only those records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
General hospitals may provide sub-acute care through “swing beds” or other non-acute units within the hospital, and we would expect these claims to be found in the LT file rather than the IP file.
At least one state (Florida) allows Health Maintenance Organizations (HMOs) to dispense pharmaceuticals to enrollees. Therefore, HMOs are included as an “expected” taxonomy for RX billing providers.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We used claim type code (CLM_TYPE_CD) to determine which records to include and exclude in our analysis. Records included are those with a claim type code indicating FFS claims (values of 1 and A) and managed care encounters (3 and C). We excluded capitated payments (2 and B), supplemental payments (5 and E), and service-tracking claims (4 and D). We also used claim type code to exclude \u201cother\u201d records (values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information on supplemental payments and other claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
More information about the completeness of each state\u2019s TAF data can be found in the DQ Atlas single topic displays for Claims Volume\u2014IP , Claims Volume\u2014LT , Claims Volume\u2014OT , and Claims Volume\u2014RX .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This approach means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only those records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
General hospitals may provide sub-acute care through \u201cswing beds\u201d or other non-acute units within the hospital, and we would expect these claims to be found in the LT file rather than the IP file.
At least one state (Florida) allows Health Maintenance Organizations (HMOs) to dispense pharmaceuticals to enrollees. Therefore, HMOs are included as an \u201cexpected\u201d taxonomy for RX billing providers.
The servicing provider type, specialty, and taxonomy codes allow TAF users to examine the characteristics of the individual practitioner who was responsible for or provided direct care to a Medicaid or CHIP beneficiary. This analysis examines the extent to which professional claims in the OT file have a servicing provider type, specialty, and taxonomy missing or coded with unexpected or unusable values.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5081"", ""relatedTopics"": [{""measureId"": 37, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - IP"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 0}, {""measureId"": 38, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - LT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 1}, {""measureId"": 104, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - OT Professional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 2}, {""measureId"": 105, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - OT Institutional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 3}, {""measureId"": 103, ""measureName"": ""Billing Provider Specialty and Taxonomy - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 4}, {""measureId"": 107, ""measureName"": ""Servicing Provider Type, Specialty, and Taxonomy - OT Institutional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 6}]}" 107,"{""measureId"": 107, ""measureName"": ""Servicing Provider Type, Specialty, and Taxonomy - OT Institutional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Serv-Prov-Type-OT-Inst.pdf"", ""background"": {""content"": ""Users of the T-MSIS Analytic Files (TAF) may want to identify the providers or categories of providers that deliver services to Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries. A claim in the TAF could include information for up to six providers, depending on the file type. [1] A provider could be a facility, a group of individual practitioners, or an individual practitioner. The provider types most often used for claims-based analyses include the billing, servicing, prescribing, and dispensing providers.
The billing provider represents the entity that submits the claim and is reimbursed by the Medicaid or CHIP agency. Information about the billing provider is available in all TAF claims files: inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX).
The servicing provider, in contrast, represents the individual practitioner who was responsible for or provided direct care to the beneficiary (that is, the rendering provider). Information about servicing provider is available in the IP, LT, and OT files. In some cases, the servicing provider is the same as the billing provider. In other cases, the servicing and billing provider differ, such as when a hospital or large group practice bills for services but an individual physician employed by the hospital or group provides the direct care and is the serving provider. Because institutional claims in the IP and LT files represent facility costs and are often submitted by the hospitals and skilled nursing facilities where the care occurred, analyses using these claims would typically only use the billing provider information.
For prescription drug claims, the billing provider is the pharmacy where the prescription was filled, the prescribing provider is the individual practitioner who prescribed a prescription drug to a beneficiary, and the dispensing provider is the pharmacist who filled the prescription or was responsible for overseeing the filling of the prescription. Because the billing provider represents the entity that submits a claim and is reimbursed by the Medicaid or CHIP agency, analyses using prescription drug claims typically focus on the billing provider information.
Data elements related to the billing provider can offer insight into the characteristics of providers who receive payment for Medicaid- and CHIP-funded services. Likewise, data elements related to the servicing provider can offer insight into the characteristics of the individual practitioner who was responsible for or provided direct care to a Medicaid or CHIP beneficiary. There are multiple systems available in TAF claim records for classifying providers, including the following:
Each of these classification systems may be best suited to different types of analyses. Although states are encouraged to populate all these fields on all claims, only one classification type is required. In practice, some states only submit information related to some of the classification systems, which may require TAF users to adjust their methodology across states. If the preferred data element has high rates of missingness, TAF users may be able to link the claims record to the provider’s record in the Annual Provider File (APR) to obtain the needed information. [2]
This data quality assessment examines whether any information on billing or servicing provider characteristics is available in claims records, and which type (taxonomy, specialty, or provider type). This information can help users design their analysis by selecting the fields populated in each state or identify states for which it may be necessary to link to the APR to obtain complete information about providers, rather than rely on claims alone.
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
States are required to submit information about all providers eligible to provide Medicaid- and CHIP-funded services; these records are captured in the APR. It is possible that missing information in the claim about provider taxonomy, specialty, or type could be imputed by linking to the provider’s record in the APR and obtaining the information from that file. In addition, the APR includes information on how states classify providers by “authorized category of service,” a classification scheme that can be used to identify certain non-medical provider categories not captured in the other classification systems, such as transportation or personal care service providers. For more information on the completeness of these fields in each state’s APR file, see the DQ Atlas single-topic displays for Group and Individual Providers - Classification Types and Facilities - Classification Types .
The inpatient file includes information about billing, referring, servicing, admitting, and operating providers. The long-term care file includes information about billing, referring, servicing, and admitting providers. The other services file includes information about billing, referring, servicing, health home, directing, and supervising providers. The pharmacy claims file includes information about billing, dispensing, and prescribing providers.
States are required to submit information about all providers eligible to provide Medicaid- and CHIP-funded services; these records are captured in the APR. It is possible that missing information in the claim about provider taxonomy, specialty, or type could be imputed by linking to the provider\u2019s record in the APR and obtaining the information from that file. In addition, the APR includes information on how states classify providers by \u201cauthorized category of service,\u201d a classification scheme that can be used to identify certain non-medical provider categories not captured in the other classification systems, such as transportation or personal care service providers. For more information on the completeness of these fields in each state\u2019s APR file, see the DQ Atlas single-topic displays for Group and Individual Providers - Classification Types and Facilities - Classification Types .
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We examined the values for provider type, specialty, and taxonomy codes for the providers most examined in analyses for the TAF IP, LT, OT, and RX files (Table 1). [3] We included both fee-for-service (FFS) claims and managed care encounter records, which represent claims paid by managed care organizations and should generally follow the same reporting standards for billing and servicing providers. We excluded financial transaction records, supplemental payments, and “other” records that the state did not classify as being covered by either the Medicaid or CHIP programs. [4] We also excluded states from the analysis of each file if the number of header records in the TAF was low enough to be unusable. [5] For Illinois, we restricted our analysis to the original version of the claim and excluded all subsequent adjustment records in the state’s TAF data. [6]
Table 1. Crosswalk between file type, data element description, and TAF variable name
| File type | Data element description | Variable name |
|---|---|---|
| IP, LT, OT | Billing provider | BLG_PRVDR_TYPE_CD |
| IP, LT, OT, RX | Billing provider specialty code | BLG_PRVDR_SPCLTY_CD |
| IP, LT, OT, RX | Billing provider state-reported taxonomy code | BLG_PRVDR_TXNMY_CD |
| IP, LT, OT | Billing provider NPPES primary taxonomy code | BLG_PRVDR_NPPES_TXNMY_CD |
| OT | Servicing provider type code | SRVCNG_PRVDR_TYPE_CD |
| OT | Servicing provider specialty code | SRVCNG_PRVDR_SPCLTY_CD |
| OT | Servicing provider state-reported taxonomy code | SRVCNG_PRVDR_TXNMY_CD |
| OT | Servicing provider NPPES primary taxonomy code | SRVCNG_PRVDR_NPPES_TXNMY_CD |
The provider type code includes 57 different valid values, covering both facilities and professionals. As shown in Table 2, we grouped these values into those that are expected, unexpected, and unusable for billing providers for each of the three file types in which this data element is present (IP, LT, OT) and for servicing providers in the OT file.
In the IP file, we would expect to see only general hospitals and Indian Health Service facilities as billing providers. In the LT file, we would expect to see nursing facilities, intermediate care facilities for individuals with intellectual or developmental disabilities, and psychiatric facilities as billing providers, as well as general hospitals in selected cases. [7] In the OT file, we would expect to see a broad range of both facilities and professionals billing for and directly providing services.
For all three files, we examined the “all other” code separately from the expected and unexpected categories. In some instances, use of the non-specific “all other” code may reflect the possibility that the provider is not represented in the list of valid provider type codes, which is the case for certain types of home- and community-based services (HCBS) and non-emergency medical transport providers. Although this code is valid and not considered an unexpected value in any file, it does not provide any information about the billing or servicing provider and thus renders the data element unusable for analysis.
Table 2. Classification of provider type codes, by file and provider field
| File and provider field | Expected codes | Unusable codes (“all other” value) | Unexpected codes |
|---|---|---|---|
| IP—billing provider | 42, 51 | 57 | All other values |
| LT—billing provider | 42, 43, 44, 45 | 57 | All other values |
| OT—billing and servicing provider | 01−56 | 57 | None |
Source: A full list of values is available in the TAF Claims Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries . The provider type code values are identical across all three files.
The provider specialty code includes 115 different valid values, covering both facilities and professionals. As shown in Table 3, we grouped these values into those that are expected and unexpected for each of the file types and provider fields. In the IP file, we would expect to see only general hospitals as billing providers. In the LT file, we would expect to see nursing facilities, intermediate care facilities for individuals with intellectual or developmental disabilities, and psychiatric facilities as billing providers, as well as general hospitals in selected cases. In the OT file, we would expect to see a broad range of both facilities and professionals billing for and directly providing services. In the RX file, we would expect to see pharmacies, medical supply companies, department stores, and grocery stores as billing providers. For all files and provider fields, we examined the “all other” codes (billing provider specialty code 87—All Other Suppliers) separately from the expected and unexpected categories. Although this code is valid and not considered an unexpected value in any file, it does not provide any information about the provider and thus renders the data element unusable for analysis.
Table 3. Classification of provider specialty codes, by file and provider field
| File and provider field | Expected codes | Unusable codes (“all other” value) | Unexpected codes |
|---|---|---|---|
| IP—billing provider | A0 | 87 | All other values except 88 and 99, which are treated as missing |
| LT—billing provider | A0, A1, A2, A3, B4 | 87 | All other values except 88 and 99, which are treated as missing |
| OT—billing and servicing provider | 1−86, 89−98, A0−B5 | 87 | All other values except 88 and 99, which are treated as missing |
| RX—billing provider | 51–54, 58, 73, A5, A6, A9, B1, B3, B4 | 87 | All other values except 88 and 99, which are treated as missing |
Source: A full list of values is available in the TAF Claims Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries . The provider specialty code values are identical across all claims files.
The state-reported billing provider taxonomy code is a 10-character alphanumeric code that identifies the provider’s area of specialization. As shown in Table 4, we grouped these values into those that are expected and unexpected for each of the file types. In the IP file, we would expect to see inpatient hospitals as billing providers. In the LT files we would expect to see nursing facilities, intermediate care facility services for individuals with intellectual disabilities, mental health facility services, and independent (free-standing) psychiatric wings of acute care hospitals as billing providers. In the OT file, we would expect to see a broad range of both facilities and professionals billing for and directly providing services. In the RX file, we would expect to see pharmacies and other suppliers as billing providers.
In addition to the state-reported provider taxonomy codes, IP, OT and LT files produced in 2022 and later years also include a constructed taxonomy variable for billing and servicing providers, the NPPES primary taxonomy code (*_NPPES_TXNMY_CD), that represents the provider’s primary taxonomy information as reported in the National Plan and Provider Enumeration System (NPPES) based on the provider’s National Provider Identifier. We used the same criteria as the state-reported taxonomy codes to determine expected and unexpected values for each file type.
Table 4. Classification of provider taxonomy codes, by file and provider field
| File and provider field | Expected codes | Unexpected codes |
|---|---|---|
| IP—billing provider | First 2 digits of taxonomy code are 27 or 28 | All other values |
| LT—billing provider | Taxonomy code begins with 283Q, 283X, 282E, 31, 32, 385H, or 281P | All other values |
| OT—billing and servicing provider | All valid taxonomy codes | None |
| RX—billing provider | First 2 digits of taxonomy code are 33, first 3 digits are 183, or 302R00000X [8] | All other values |
Source: The health care provider taxonomy code set can be found at http://www.wpc-edi.com/reference/codelists/healthcare/health-care-provider-taxonomy-code-set/ . The provider taxonomy code values are identical across all claims files.
We grouped states into categories of low, medium, and high concern about the usability of their data, based on the percentage of header records (or for servicing providers in the OT file, claim lines) that had an expected provider type, specialty, state-reported taxonomy code, or NPPES primary taxonomy code for the file and provider fields (Table 5).
Because users of the OT file may be interested in professional or institutional (facility) claims, we conducted the DQ assessment separately for each of these claim types. Table 6 shows the methodology used to classify each claim header as a professional or institutional claim. Any claim that did not meet at least one of the criteria for a professional or institutional claim was considered unclassified and is not included in this analysis. Note that the servicing provider analyses were conducted at the claim line level, but the claims were classified as professional or institutional claims at the claim header level.
Table 5. Criteria for DQ assessment of provider type, specialty, and taxonomy code
| Percentage of records with an expected provider type, specialty, state-reported taxonomy code, or NPPES primary taxonomy code | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
Table 6. Classification of OT claim headers as professional or facility claims
| Claim category | Identification rules |
|---|---|
| Professional | The record must meet at least one of the following criteria:
|
| Institutional | The record must meet at least one of the following criteria:
|
Methods previously used to assess data quality
Table 7 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 7. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
|
|
|
|
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We used claim type code (CLM_TYPE_CD) to determine which records to include and exclude in our analysis. Records included are those with a claim type code indicating FFS claims (values of 1 and A) and managed care encounters (3 and C). We excluded capitated payments (2 and B), supplemental payments (5 and E), and service-tracking claims (4 and D). We also used claim type code to exclude “other” records (values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information on supplemental payments and other claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
More information about the completeness of each state’s TAF data can be found in the DQ Atlas single topic displays for Claims Volume—IP , Claims Volume—LT , Claims Volume—OT , and Claims Volume—RX .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This approach means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only those records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
General hospitals may provide sub-acute care through “swing beds” or other non-acute units within the hospital, and we would expect these claims to be found in the LT file rather than the IP file.
At least one state (Florida) allows Health Maintenance Organizations (HMOs) to dispense pharmaceuticals to enrollees. Therefore, HMOs are included as an “expected” taxonomy for RX billing providers.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We used claim type code (CLM_TYPE_CD) to determine which records to include and exclude in our analysis. Records included are those with a claim type code indicating FFS claims (values of 1 and A) and managed care encounters (3 and C). We excluded capitated payments (2 and B), supplemental payments (5 and E), and service-tracking claims (4 and D). We also used claim type code to exclude \u201cother\u201d records (values of U, V, W, X, and Y) that the state did not classify as either Medicaid or CHIP payment records. More information on supplemental payments and other claims can be found in the DQ Atlas single topic display for Non-Program (Other) Claims .
More information about the completeness of each state\u2019s TAF data can be found in the DQ Atlas single topic displays for Claims Volume\u2014IP , Claims Volume\u2014LT , Claims Volume\u2014OT , and Claims Volume\u2014RX .
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim, as it does in all other states. To ensure that the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This approach means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only those records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
General hospitals may provide sub-acute care through \u201cswing beds\u201d or other non-acute units within the hospital, and we would expect these claims to be found in the LT file rather than the IP file.
At least one state (Florida) allows Health Maintenance Organizations (HMOs) to dispense pharmaceuticals to enrollees. Therefore, HMOs are included as an \u201cexpected\u201d taxonomy for RX billing providers.
The servicing provider type, specialty, and taxonomy codes allow TAF users to examine the characteristics of the individual practitioner who was responsible for or provided direct care to a Medicaid or CHIP beneficiary. This analysis examines the extent to which facility claims in the OT file have a servicing provider type, specialty, and taxonomy missing or coded with unexpected or unusable values.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5081"", ""relatedTopics"": [{""measureId"": 37, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - IP"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 0}, {""measureId"": 38, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - LT"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 1}, {""measureId"": 104, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - OT Professional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 2}, {""measureId"": 105, ""measureName"": ""Billing Provider Type, Specialty, and Taxonomy - OT Institutional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 3}, {""measureId"": 103, ""measureName"": ""Billing Provider Specialty and Taxonomy - RX"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 4}, {""measureId"": 106, ""measureName"": ""Servicing Provider Type, Specialty, and Taxonomy - OT Professional"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""order"": 5}]}" 108,"{""measureId"": 108, ""measureName"": ""Linking Claims to Beneficiaries"", ""groupId"": 11, ""groupName"": ""Linking Across Files"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Link-Claims-Bene.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) consist of data on enrollment and service use for all individuals enrolled in Medicaid or in the Children’s Health Insurance Program (CHIP). All service use records, including both fee-for-service (FFS) claims paid by the state Medicaid agency and managed care encounters paid by a Medicaid managed care plan, should link to a beneficiary eligibility record. The TAF Research Identifiable Files (RIF) include two beneficiary identifiers that can be used to make this linkage: (1) the state-assigned MSIS identifier (MSIS ID), which is in the eligibility and service use data that states submit into T-MSIS, and (2) the federally assigned beneficiary identifier (BENE ID). [1] , [2]
State Medicaid agencies and managed care plans should only pay for services delivered while an individual is enrolled in Medicaid or CHIP. Because the TAF excludes denied claims, [3] all FFS claims and managed care encounter records in the inpatient (IP), other services (OT), long-term care (LT), and pharmacy (RX) files should link to an eligibility record in the TAF annual Demographic and Eligibility (DE) file that indicates the beneficiary was enrolled and eligible for services. [4]
Certain Medicaid policies could lead to predictable data quality errors in the enrollment start and end dates on eligibility records. When this occurs, service dates on paid claims may not match to enrollment dates on beneficiary eligibility records. For instance, Medicaid has a retroactive coverage policy that allows states to cover the costs of unpaid services provided up to three months before the beneficiary applied for coverage as long as the state can determine that the beneficiary would have been eligible during that period had he or she applied at that time. For a beneficiary who qualifies for retroactive eligibility, it is possible that a state will incorrectly code the start date of enrollment based on when an application was approved as opposed to when the coverage period started. If that occurs, the TAF may include paid claims with service dates in the three months before the beneficiary’s enrollment date.
Additionally, many states have adopted presumptive eligibility for certain low-income individuals, which allows providers such as hospitals to enroll patients into temporary Medicaid coverage for up to a month before the Medicaid agency makes a final eligibility determination. [5] , [6] Although these individuals are enrolled in Medicaid during the period of presumptive eligibility, it is possible that a portion of them may later be determined ineligible for some or all of the months in which they received services; as a result, the state may submit their service use records but not their enrollment data.
Sometimes, states may submit claims in T-MSIS with MSIS IDs for which they do not submit any corresponding eligibility information. There are a variety of reasons this could happen (such as administrative errors or an unintended consequence of presumptive eligibility policies), but only a small proportion of records in the DE file represent MSIS IDs for which the state submitted no eligibility information (less than 0.5 percent in the majority of states).
This analysis examines the proportion of service use records in the TAF that can be linked to an eligibility record using the state-assigned MSIS ID. The ability to correctly link eligibility and service use records in the TAF is critical for analyses in which diagnoses, services, or expenditures must be attributed to specific individuals, such as counting the number of beneficiaries with certain health conditions, calculating quality measures, or calculating per-beneficiary costs for a specific subgroup of beneficiaries. Service use records that do not link to eligibility records indicate data quality issues with either the enrollment or the service use information reported by states.
The BENE ID is created from the state-assigned MSIS ID and other person-specific variables (such as birth date and sex) by looking across records that have different state-assigned unique identifiers to determine whether they represent the same person. The BENE ID can be used to identify the same individual enrolled in Medicaid or CHIP in more than one state. It can also be used to link to the Medicare data for dually eligible beneficiaries.
More information on the availability and quality of MSIS ID and BENE ID is available in the brief, \""Unique Beneficiary Identifiers in TAF,\"" available on the DQ Atlas Resources page
TAF excludes denied header records and all the line records associated with it. However, it includes denied claim lines associated with non-denied header records.
The DE file includes \""dummy\"" records with beneficiary identification numbers that are observed on claims but were not reported by states in their eligibility data submissions. As a result, every service use record in the IP, LT, OT, and RX files will link to a DE record when using MSIS ID and state to make the linkage. However, these dummy records do not include any information regarding enrollment or beneficiary characteristics.
Historically, presumptive eligibility policies allowed certain qualified providers to screen and temporarily enroll children and pregnant women into Medicaid or CHIP. Under the Affordable Care Act, states have the option to extend presumptive eligibility to cover low-income parents and other adults. For more information, see 42 CFR 435.11 (Options for Coverage of Special Groups under Presumptive Eligibility).
A list of states that provide presumptive eligibility in Medicaid and/or CHIP can be found here: https://www.kff.org/health-reform/state-indicator/presumptive-eligibility-in-medicaid-chip/?currentTimeframe=0&sortModel=%7B%22colId%22:%22Location%22,%22sort%22:%22asc%22%7D
The BENE ID is created from the state-assigned MSIS ID and other person-specific variables (such as birth date and sex) by looking across records that have different state-assigned unique identifiers to determine whether they represent the same person. The BENE ID can be used to identify the same individual enrolled in Medicaid or CHIP in more than one state. It can also be used to link to the Medicare data for dually eligible beneficiaries.
More information on the availability and quality of MSIS ID and BENE ID is available in the brief, \""Unique Beneficiary Identifiers in TAF,\"" available on the DQ Atlas Resources page
TAF excludes denied header records and all the line records associated with it. However, it includes denied claim lines associated with non-denied header records.
The DE file includes \""dummy\"" records with beneficiary identification numbers that are observed on claims but were not reported by states in their eligibility data submissions. As a result, every service use record in the IP, LT, OT, and RX files will link to a DE record when using MSIS ID and state to make the linkage. However, these dummy records do not include any information regarding enrollment or beneficiary characteristics.
Historically, presumptive eligibility policies allowed certain qualified providers to screen and temporarily enroll children and pregnant women into Medicaid or CHIP. Under the Affordable Care Act, states have the option to extend presumptive eligibility to cover low-income parents and other adults. For more information, see 42 CFR 435.11 (Options for Coverage of Special Groups under Presumptive Eligibility).
A list of states that provide presumptive eligibility in Medicaid and/or CHIP can be found here: https://www.kff.org/health-reform/state-indicator/presumptive-eligibility-in-medicaid-chip/?currentTimeframe=0&sortModel=%7B%22colId%22:%22Location%22,%22sort%22:%22asc%22%7D
We used the claim type code (CLM_TYPE_CD) to select all header records [7] classified as fee-for-service (FFS) claims or managed care encounters from the four TAF claim files. [8] , [9] From the Demographic and Eligibility (DE) TAF, we selected all non-dummy records, which represent Medicaid and CHIP beneficiaries who the state indicated were enrolled at any time during the calendar year. [10]
We matched header records from the claims files to eligibility records by using the state-assigned beneficiary identifier (MSIS_IDENT_NUM) and the submitting state code (SUBMTG_STATE_CD). If a header record matched to a DE record that indicated that the person was enrolled during the month in which the service occurred, we counted it as matching to a beneficiary enrolled at the time of service (same-month match). [11] , [12]
We grouped states into categories of concern about the usability of their data, depending on the percentage of service records that do not link to an eligibility record in the same month of service (Table 1).
Table 1. Criteria for DQ assessment of the linkage of claims to beneficiaries
| Percentage of service use records that do not link to an eligibility record in the same month of service | DQ assessment |
|---|---|
| x ≤ 1 percent | Low concern |
| 1 percent < x ≤ 10 percent | Medium concern |
| 10 percent < x ≤ 20 percent | High concern |
| x > 20 percent | Unusable |
Header records summarize the service or the set of linked services provided to a beneficiary. Line records have more detailed information about the services provided. We used the header record for this analysis because the majority of studies in which claims data are used will be based on the summary information in the header record.
We selected records with claim type code = 1, 3, A, C, U, or W.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \""Production of the TAF Research Identifiable Files.\""
We excluded the \""dummy\"" records from the DE file that represent MSIS IDs present on claims but for which the state submitted no enrollment information. Dummy records can be identified as those with MISG_ELGBLTY_DATA_IND = 1.
If it was available, we used the monthly CHIP code (CHIP_CD) to identify beneficiaries enrolled in Medicaid or CHIP that month (CHIP_CD values of 1–4). CHIP code 4 (Medicaid and S-CHIP) is a valid value for 2014 through 2017 TAF. If the CHIP code was missing, we considered the beneficiary as enrolled in the month only if the eligibility group variable (ELGBLTY_GRP_CD) was not missing.
TAF claims data are organized by month and year of service. For IP and LT claims, the discharge date (or if that is missing, the service end date) is used to determine the month and year of the service. For OT claims, the service end date is used to determine the month and year of the service. If the service end date is missing, then the service start date is used, and if the start date is missing, then the most recent service end date from all the claim lines is used. For RX claims, the prescription fill date is used to determine the month and year of the service.
Header records summarize the service or the set of linked services provided to a beneficiary. Line records have more detailed information about the services provided. We used the header record for this analysis because the majority of studies in which claims data are used will be based on the summary information in the header record.
We selected records with claim type code = 1, 3, A, C, U, or W.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \""Production of the TAF Research Identifiable Files.\""
We excluded the \""dummy\"" records from the DE file that represent MSIS IDs present on claims but for which the state submitted no enrollment information. Dummy records can be identified as those with MISG_ELGBLTY_DATA_IND = 1.
If it was available, we used the monthly CHIP code (CHIP_CD) to identify beneficiaries enrolled in Medicaid or CHIP that month (CHIP_CD values of 1\u20134). CHIP code 4 (Medicaid and S-CHIP) is a valid value for 2014 through 2017 TAF. If the CHIP code was missing, we considered the beneficiary as enrolled in the month only if the eligibility group variable (ELGBLTY_GRP_CD) was not missing.
TAF claims data are organized by month and year of service. For IP and LT claims, the discharge date (or if that is missing, the service end date) is used to determine the month and year of the service. For OT claims, the service end date is used to determine the month and year of the service. If the service end date is missing, then the service start date is used, and if the start date is missing, then the most recent service end date from all the claim lines is used. For RX claims, the prescription fill date is used to determine the month and year of the service.
All service use records in the TAF claims files should link to a beneficiary using the unique identifier that states assign to each beneficiary, the MSIS ID. This analysis examines the proportion of service use records in the TAF that can be linked to an eligibility record using the MSIS ID.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""3141"", ""relatedTopics"": []}" 109,"{""measureId"": 109, ""measureName"": ""National Provider Identifier"", ""groupId"": 6, ""groupName"": ""Provider Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-APR-NPI.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) Annual Provider File (APR) captures detailed information about each provider authorized by a state to provide services to its Medicaid and Children’s Health Insurance Program (CHIP) beneficiaries, as well as providers whose approval is pending, denied, or has been terminated. Providers are included in the TAF APR regardless of whether or how often the provider billed the state for services. [1] A provider could be a facility, a group of practitioners, or an individual practitioner. The TAF APR includes more detailed information about those providers rendering or billing for services compared to the limited provider information found on the fee-for-service (FFS) and encounter records in the TAF claims files. For example, FFS and encounter records on the TAF inpatient (IP), long-term care (LT), and other services (OT) files include information on each claim about the billing, servicing, and rendering provider’s taxonomy, specialty, and state-identified provider type. [2] The TAF APR, on the other hand, includes details about the characteristics, locations, taxonomies and classifications, affiliated groups, affiliated programs, licensing and accreditations, and—for facility providers—bed types for Medicaid- or CHIP-eligible providers, as well as other identifiers associated with the provider. [3]
Each TAF APR base file record includes two provider identifiers: (1) the National Provider Identifier (NPI), which is the unique, 10-digit identification number that the National Plan and Provider Enumeration System (NPPES) assigns to each Health Insurance Portability and Accountability Act (HIPAA)-covered health care provider; and (2) the state-assigned provider identifier used in the state’s Medicaid Management Information System. [4] The state-assigned provider identifier, if accurately reported in the TAF APR records, can be used to link to most provider data elements on TAF claims records to examine beneficiary service use and related payments. [5] However, in the event these records do not link well on state-assigned provider identifiers, the NPI can be used as a secondary linking mechanism. In addition, because a provider can participate with more than one state Medicaid program, and therefore more than one TAF APR record may represent a given provider, the NPI is needed to de-duplicate providers for cross-state or national-level analyses. Also, certain provider data elements on the TAF claims files—those indicating the supervising provider, operating provider, directing provider, and home health provider on a claim (where applicable)—specify only the NPI as a provider identifier. TAF users interested in linking the APR to these provider fields on claims must do so using the NPI. The NPI can also be used to link to other non-TAF data sources. TAF users may be interested in the quality of the NPI in order to link to the publicly available NPPES NPI registry to obtain additional information about a provider’s characteristics. [6]
States are required to report NPIs for all providers that have been assigned an NPI by the NPPES. Certain \""atypical\"" providers - who are authorized to deliver services to Medicaid and CHIP beneficiaries, but do not provide \""health care\"", as defined in HIPAA 45 C.F.R. 160.103 - are not eligible to receive an NPI. Therefore, we do not expect the NPI field to be populated for TAF APR records representing these providers. Among atypical providers that are reimbursed by Medicaid programs are those who offer taxi services, home and vehicle modifications, and respite services. [7] There is significant variation across state Medicaid programs in the prevalence of atypical providers and the ability of states to report these providers into T-MSIS. In some states, atypical providers may represent a sizeable proportion of all APR records, whereas in other states they represent a very small percentage. This data quality assessment examines the extent to which a provider’s NPI is available on the APR records and can be matched to an active NPI on the NPPES NPI registry, regardless of whether they are an atypical provider. [8]
TAF APR records present unique combinations of submitting state and provider identifier. Because a provider can participate with more than one state Medicaid program, more than one TAF APR record may represent a given provider.
FFS and encounter records on the RX file also include information on each claim about the billing provider’s taxonomy and specialty. In addition to billing and servicing provider information, FFS and encounter records on the IP and LT files include information on each claim about the admitting provider’s taxonomy, specialty, and state-identified provider type.
The TAF APR consists of nine files. The TAF APR base file includes basic provider characteristics, and eight additional supplemental files provide more detailed information on the following topics: the provider’s classifications (for example, taxonomy code), Medicaid and/or CHIP program enrollment, group affiliation, health plan and program affiliation, geographic location(s), licensing and accreditation, other provider identifiers, and bed type. Each record in the APR TAF base file represents a provider enrolled with the state’s Medicaid or CHIP program. Each provider record on the base file may link to more than one record on each supplemental file.
Centers for Medicare & Medicaid Services (CMS). \""CMS Guidance: Reporting Provider Identifiers in T-MSIS. \"" Baltimore, MD: CMS, January 2019. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=50507 .
More information on how well the state-assigned provider identifiers in the TAF claims files link to the TAF APR can be found in the DQ Atlas single-topic displays for Linking Claims to Providers .
The NPPES NPI registry is available at: https://npiregistry.cms.hhs.gov/ .
See Smith, Dennis G. SMDL #06-020. Letter to State Medicaid Directors, September 19, 2006. Available at: https://www.medicaid.gov/Federal-Policy-Guidance/downloads/SMD091906b.pdf . Some providers of non-healthcare services may be eligible for an NPI, for example if the provider renders other services that qualify them as a healthcare provider (or has done so in the past).
An active NPI is a provider identifier that is present on the NPPES NPI registry and is either (1) not flagged as being deactivated before the start of the calendar year, or (2) is flagged as being deactivated before the start of the calendar year but subsequently reactivated during that calendar year. An NPI may be deactivated because of a provider’s retirement or death, disbandment of a provider entity, or in cases of identity theft or fraudulent use.
TAF APR records present unique combinations of submitting state and provider identifier. Because a provider can participate with more than one state Medicaid program, more than one TAF APR record may represent a given provider.
FFS and encounter records on the RX file also include information on each claim about the billing provider\u2019s taxonomy and specialty. In addition to billing and servicing provider information, FFS and encounter records on the IP and LT files include information on each claim about the admitting provider\u2019s taxonomy, specialty, and state-identified provider type.
The TAF APR consists of nine files. The TAF APR base file includes basic provider characteristics, and eight additional supplemental files provide more detailed information on the following topics: the provider\u2019s classifications (for example, taxonomy code), Medicaid and/or CHIP program enrollment, group affiliation, health plan and program affiliation, geographic location(s), licensing and accreditation, other provider identifiers, and bed type. Each record in the APR TAF base file represents a provider enrolled with the state\u2019s Medicaid or CHIP program. Each provider record on the base file may link to more than one record on each supplemental file.
Centers for Medicare & Medicaid Services (CMS). \""CMS Guidance: Reporting Provider Identifiers in T-MSIS. \"" Baltimore, MD: CMS, January 2019. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=50507 .
More information on how well the state-assigned provider identifiers in the TAF claims files link to the TAF APR can be found in the DQ Atlas single-topic displays for Linking Claims to Providers .
The NPPES NPI registry is available at: https://npiregistry.cms.hhs.gov/ .
See Smith, Dennis G. SMDL #06-020. Letter to State Medicaid Directors, September 19, 2006. Available at: https://www.medicaid.gov/Federal-Policy-Guidance/downloads/SMD091906b.pdf . Some providers of non-healthcare services may be eligible for an NPI, for example if the provider renders other services that qualify them as a healthcare provider (or has done so in the past).
An active NPI is a provider identifier that is present on the NPPES NPI registry and is either (1) not flagged as being deactivated before the start of the calendar year, or (2) is flagged as being deactivated before the start of the calendar year but subsequently reactivated during that calendar year. An NPI may be deactivated because of a provider\u2019s retirement or death, disbandment of a provider entity, or in cases of identity theft or fraudulent use.
We examined the extent to which NPIs are both available on the APR records and can be matched to an active NPI on the NPPES NPI registry. For each provider record on the TAF APR, up to two NPIs can be reported on the APR base file; these NPIs, plus any additional ones beyond the first two, are reported on the APR identifiers supplemental file. [9] However, only one NPI is expected to be active at any given time, and only a small percentage of providers will ever have more than one NPI. [10] For each submitting state, we calculated (1) the percentage of records with only missing or invalidly formatted NPI [11] (that is, both NPI fields on the APR base file as well as any additional NPIs present on the identifiers supplemental file contain either a missing NPI value or an invalidly formatted NPI); (2) the percentage of records with a validly formatted identifier that does not match to an active NPI on the NPPES registry; [12] and (3) the total percentage of APR records with a valid NPI, defined as identifiers that successfully match to an active NPI on the NPPES NPI registry. [13]
For additional context, we examined the percentage of APR records with more than one valid NPI reported. For the subset of APR records that do not have a valid NPI, we also examined the percentage of these records that represent atypical providers (and are therefore not expected to have an NPI).
Table 1. Criteria for DQ assessment of NPI in the TAF APR
| Percentage of APR records with a valid NPI | DQ assessment |
|---|---|
| x > 90 percent | Low concern |
| 80 percent < x ≤ 90 percent | Medium concern |
| 40 percent < x ≤ 80 percent | High concern |
| x ≤ 40 percent | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \""Production of the TAF Research Identifiable Files.\""
Centers for Medicare & Medicaid Services (CMS). \""CMS Guidance: Reporting Provider Identifiers in T-MSIS.\"" Baltimore, MD: CMS, January 2019. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=50507 .
An invalidly formatted NPI is one that does not meet all of the following criteria: (1) the NPI is 10-digit numeric, (2) the first digit of the NPI is a \""1\"" or a \""2\"" and (3) the NPI’s 10th \""check\"" digit is valid according to the Luhn formula. More information is available at https://www.cms.gov/Regulations-and-Guidance/Administrative-Simplification/NationalProvIdentStand/Downloads/NPIcheckdigit.pdf .
For this data quality assessment, identifiers that did not match to an active NPI on the NPPES registry include both (1) identifiers not present on the NPPES NPI registry and (2) identifiers flagged on the NPPES as being deactivated before the start of the calendar year (and not subsequently reactivated during that calendar year).
APR records representing atypical providers were identified by linking APR base file records to the APR taxonomy supplemental file records. Providers that met both of the following criteria were counted as atypical provider: (1) the provider had one or more taxonomy, specialty, provider type, or authorized category of service codes applicable to atypical providers, and (2) the provider did not have one or more additional taxonomy, specialty, provider type, or authorized category of service codes applicable to HIPPA-defined health care providers (which would qualify for an NPI).
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \""Production of the TAF Research Identifiable Files.\""
Centers for Medicare & Medicaid Services (CMS). \""CMS Guidance: Reporting Provider Identifiers in T-MSIS.\"" Baltimore, MD: CMS, January 2019. Available at https://www.medicaid.gov/medicaid/data-and-systems/macbis/tmsis/tmsis-blog/?entry=50507 .
An invalidly formatted NPI is one that does not meet all of the following criteria: (1) the NPI is 10-digit numeric, (2) the first digit of the NPI is a \""1\"" or a \""2\"" and (3) the NPI\u2019s 10th \""check\"" digit is valid according to the Luhn formula. More information is available at https://www.cms.gov/Regulations-and-Guidance/Administrative-Simplification/NationalProvIdentStand/Downloads/NPIcheckdigit.pdf .
For this data quality assessment, identifiers that did not match to an active NPI on the NPPES registry include both (1) identifiers not present on the NPPES NPI registry and (2) identifiers flagged on the NPPES as being deactivated before the start of the calendar year (and not subsequently reactivated during that calendar year).
APR records representing atypical providers were identified by linking APR base file records to the APR taxonomy supplemental file records. Providers that met both of the following criteria were counted as atypical provider: (1) the provider had one or more taxonomy, specialty, provider type, or authorized category of service codes applicable to atypical providers, and (2) the provider did not have one or more additional taxonomy, specialty, provider type, or authorized category of service codes applicable to HIPPA-defined health care providers (which would qualify for an NPI).
The TAF Annual Provider File (APR) contains detailed information about each provider authorized to deliver services to Medicaid and CHIP beneficiaries at any point during the calendar year. Two provider identifiers are available on the TAF APR: the National Provider Identifier (NPI) and the state-assigned provider identifier used in the state's claims processing system. This data quality assessment examines the extent to which the TAF APR records include a valid NPI.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""9051"", ""relatedTopics"": []}" 110,"{""measureId"": 110, ""measureName"": ""Linking Expenditures to Beneficiaries"", ""groupId"": 11, ""groupName"": ""Linking Across Files"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Link-Bene-Expenditures.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) consist of enrollment data in the eligibility file and spending data in the claims files for all individuals enrolled in Medicaid or in the Children’s Health Insurance Program (CHIP). The TAF Research Identifiable Files (RIF) include two beneficiary identifiers that can be used to link expenditures to eligibility records: (1) the state-assigned MSIS identifier (MSIS ID), which is in the eligibility and service use data that states submit into T-MSIS, and (2) the federally assigned beneficiary identifier (BENE ID). [1] , [2]
States can make Medicaid and CHIP payments on behalf of specific beneficiaries or in bulk for covered services using one of the following types of records: [3]
Fee-for-service (FFS) claims , which represent payments to medical providers made directly by the state Medicaid or CHIP agency. These claims should correspond to specific individuals and services.
Capitation payment records , which reflect a set per member per month (PMPM) rate paid by the state Medicaid or CHIP agency to a managed care organization (MCO), prepaid health plan (PHP), or primary care provider. These records should correspond to specific individuals. [4]
Supplemental payment records , which represent payments made in addition to a capitation payment or negotiated rate; they should correspond to specific individuals but not always to specific services.
Service tracking claims , which represent lump-sum payments made for services that cannot be attributed to a specific beneficiary, such as disproportionate share hospital (DSH) payments, payments to providers made under the Upper Payment Limit demonstration, or aggregate payments to transportation providers. [5] Some states report capitation payments covering a group of individuals as service tracking claims.
All FFS expenditures, PMPM expenditures on capitation claims, and supplemental payments in the inpatient (IP), other services (OT), long-term care (LT), and pharmacy (RX) files should link to an eligibility record in the TAF annual Demographic and Eligibility (DE) file that indicates the beneficiary was enrolled and eligible for services. [6] Non-claim based financial transactions reported in T-MSIS, which are most often found on service tracking claims, are not expected to link to beneficiaries but are helpful to include in this analysis so that TAF users can see the proportion of overall expenditures that can be linked to individual beneficiaries.
Some circumstances might generate administrative data errors that cause a mismatch between service dates on paid claims and enrollment dates on beneficiary eligibility records. For instance, many states have adopted retroactive coverage or presumptive eligibility policies that allow them to cover costs of unpaid services for a certain amount of time before a beneficiary’s final eligibility determination. [7] , [8] For beneficiaries affected by these policies, a state might incorrectly code the start date of enrollment based on the date an application was approved rather than the date the coverage period started, or a state might submit an individual’s service use records but not their enrollment data.
Because TAF claims records only include non-void, non-denied final action claims, nearly all FFS claims should have a positive total Medicaid paid amount. [9] , [10] In contrast, states may report negative payment values on non-claim-based financial transaction records, which include capitated payment records, service tracking claims, and supplemental payment records. A negative payment amount on a capitated payment record typically reflects a rate adjustment or corrections for beneficiaries who moved out of the state or died. A negative payment amount on a service tracking claim or supplemental payment record typically reflects an adjustment to the original payment amount. TAF users should be able to sum positive and negative payments across all relevant records to obtain beneficiary-level net expenditures for a particular time period or service type. If a state properly assigns valid beneficiary identifiers to records with positive payment values but reports adjustments or claw-backs as lump sums that cannot be linked to individual beneficiaries, TAF users will not be able to accurately calculate the net expenditures associated with each individual.
To account for both positive and negative payment amounts, this analysis examines the absolute value of expenditures in TAF that can be linked to an eligibility record using the MSIS ID. States with a high proportion of expenditures that do not link to eligibility records might have data quality issues with their enrollment or payment information or might rely heavily on lump sum payments for Medicaid program operations. Both options limit the usability of TAF for beneficiary-level expenditure analyses.
The BENE ID is created from the state-assigned MSIS ID and other person-specific variables (such as birth date and sex) by looking across records that have different state-assigned unique identifiers to determine whether they represent the same person. The BENE ID can be used to identify the same individual enrolled in Medicaid or CHIP in more than one state. It can also be used to link to the Medicare data for dually eligible beneficiaries.
More information on the availability and quality of MSIS ID and BENE ID is available in the data quality brief, “Unique Beneficiary Identifiers in TAF,” which can be found on the DQ Atlas Resources page
Payments on managed care encounter records reflect payments made by MCOs or PHPs to providers for services rendered to covered beneficiaries and are therefore not included in this analysis.
Capitation payments reported on service tracking claims cannot be linked to a specific beneficiary. All other monthly beneficiary payments can be linked to a specific beneficiary.
Service tracking claims are excluded from 2014–2016 TAF RIF Releases 1 and 2 and 2017–2018 TAF RIF Release 1.
The DE file includes “dummy” records with beneficiary identification numbers that are observed on claims but were not reported by states in their eligibility data submissions. As a result, every service use record in the IP, LT, OT, and RX files will link to a DE record when using MSIS ID and state to make the linkage. However, these dummy records do not include information on enrollment or beneficiary characteristics.
Historically, presumptive eligibility policies allowed certain qualified providers to screen and temporarily enroll children and pregnant women into Medicaid or CHIP. Under the Affordable Care Act, states have the option to extend presumptive eligibility to cover low-income parents and other adults. For more information, see 42 CFR 435.11 (Options for Coverage of Special Groups under Presumptive Eligibility).
A list of states that provide presumptive eligibility in Medicaid and/or CHIP is available at https://www.kff.org/health-reform/state-indicator/presumptive-eligibility-in-medicaid-chip/?currentTimeframe=0&sortModel=%7B%22colId%22:%22Location%22,%22sort%22:%22asc%22%7D .
There are a few situations in which FFS claims are not expected to have a positive payment amount, including (1) claims for which Medicare or another liable third party already paid the full Medicaid allowable amount and (2) claims that are not paid on an FFS basis despite being processed at the individual claim level.
It is not unusual for FFS claims in Illinois to have negative payment amounts due to the state’s approach to processing claims. Instead of adjusting original claims through void and resubmission records, Illinois submits marginal adjustments to the original claim.
The BENE ID is created from the state-assigned MSIS ID and other person-specific variables (such as birth date and sex) by looking across records that have different state-assigned unique identifiers to determine whether they represent the same person. The BENE ID can be used to identify the same individual enrolled in Medicaid or CHIP in more than one state. It can also be used to link to the Medicare data for dually eligible beneficiaries.
More information on the availability and quality of MSIS ID and BENE ID is available in the data quality brief, \u201cUnique Beneficiary Identifiers in TAF,\u201d which can be found on the DQ Atlas Resources page
Payments on managed care encounter records reflect payments made by MCOs or PHPs to providers for services rendered to covered beneficiaries and are therefore not included in this analysis.
Capitation payments reported on service tracking claims cannot be linked to a specific beneficiary. All other monthly beneficiary payments can be linked to a specific beneficiary.
Service tracking claims are excluded from 2014\u20132016 TAF RIF Releases 1 and 2 and 2017\u20132018 TAF RIF Release 1.
The DE file includes \u201cdummy\u201d records with beneficiary identification numbers that are observed on claims but were not reported by states in their eligibility data submissions. As a result, every service use record in the IP, LT, OT, and RX files will link to a DE record when using MSIS ID and state to make the linkage. However, these dummy records do not include information on enrollment or beneficiary characteristics.
Historically, presumptive eligibility policies allowed certain qualified providers to screen and temporarily enroll children and pregnant women into Medicaid or CHIP. Under the Affordable Care Act, states have the option to extend presumptive eligibility to cover low-income parents and other adults. For more information, see 42 CFR 435.11 (Options for Coverage of Special Groups under Presumptive Eligibility).
A list of states that provide presumptive eligibility in Medicaid and/or CHIP is available at https://www.kff.org/health-reform/state-indicator/presumptive-eligibility-in-medicaid-chip/?currentTimeframe=0&sortModel=%7B%22colId%22:%22Location%22,%22sort%22:%22asc%22%7D .
There are a few situations in which FFS claims are not expected to have a positive payment amount, including (1) claims for which Medicare or another liable third party already paid the full Medicaid allowable amount and (2) claims that are not paid on an FFS basis despite being processed at the individual claim level.
It is not unusual for FFS claims in Illinois to have negative payment amounts due to the state\u2019s approach to processing claims. Instead of adjusting original claims through void and resubmission records, Illinois submits marginal adjustments to the original claim.
The following describes current methods used to assess data quality. Information about methods previously used to assess data quality can be found at the bottom of this section.
We used the claim type code (CLM_TYPE_CD) to select all header records classified as FFS claims, capitated payment records, supplemental payment records, or service tracking claims from the four TAF claim files. [11] , [12] , [13] From the DE TAF, we selected all non-dummy records, which represent Medicaid and CHIP beneficiaries who the state indicated were enrolled at any time during the calendar year. [14]
We matched header records from the claims files to eligibility records by using the state-assigned beneficiary identifier (MSIS_IDENT_NUM) and the submitting state code (SUBMTG_STATE_CD). If a header record matched to a DE record that indicated that the person was enrolled during the month in which the service occurred, we counted it as matching to a beneficiary enrolled at the time of service (same-month match). [15] , [16] We also calculated whether a header record matched a beneficiary enrolled at any time in the year of service to understand whether administrative errors found at the monthly level (such as those driven by retroactive and presumptive eligibility policies) affected linkage rates.
We tabulated expenditures on FFS, capitated payment, and supplemental payment header records (those with claim type code = 1, 2, 5, A, B, or E) using the absolute value of the total Medicaid paid amount (TOT_MDCD_PD_AMT). We tabulated expenditures on service tracking claims (those with claim type code = 4 or D) using the absolute value of either the DSH payment amount (MDCD_DSH_PD_AMT), service tracking payment amount (SRVC_TRKNG_PYMT_AMT), or total Medicaid paid amount to tabulate expenditures. [17] , [18] , [19]
We grouped states into categories of concern about the usability of their data, depending on the percentage of the absolute value of expenditures that do not link to an eligibility record in the same month of service (Table 1).
Table 1. Criteria for DQ assessment of the linkage of expenditures to beneficiaries
To provide TAF users with additional detail about whether certain payment types are more or less likely to link to individual beneficiaries, we used the federally assigned service category (FED_SRVC_CTGRY_CD) to differentiate between FFS expenditures (payments that are for a specific service) and PMPM expenditures (non-claims-based financial transactions representing premiums or monthly fees). [20] We tabulated FFS expenditures among records where the federally assigned service category is 21–28, 31–38, or 41. We tabulated PMPM expenditures among records where the federally assigned service category is 11 or 12. We separately list the proportion of FFS expenditures and PMPM expenditures that can be linked to eligibility records as contextual information in the table view for this topic.
Methods previously used to assess data quality
Table 2 includes information about methods previously used to assess data quality and the data years and versions assessed using those methods. Each table record describes how the assessment methods for the listed data years and versions differ from current methods. Aside from those differences, the assessments for these data years and versions align with current methods. All data years and versions not listed in the table are assessed using current methods.
Table 2. Previously used methods and applicable data years and versions
| Data year(s) and version(s) | Description of difference(s) from current methods |
|---|---|
|
|
Header records summarize the service or the set of linked services provided to a beneficiary. Line records have more detailed information about the services provided. We used the header record for this analysis because most studies in which claims data are used will be based on the summary information in the header record.
We selected records with claim type code = 1, 2, 4, 5, A, B, D, or E.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We excluded the “dummy” records from the DE file that represent MSIS IDs present on claims but for which the state submitted no enrollment information. Dummy records can be identified as those with MISG_ELGBLTY_DATA_IND = 1.
If it was available, we used the monthly CHIP code (CHIP_CD) to identify beneficiaries enrolled in Medicaid or CHIP that month (CHIP_CD values of 1–4). CHIP code 4 (Medicaid and S-CHIP) is a valid value for 2014 through 2017 TAF. If the CHIP code was missing, we considered the beneficiary as enrolled in the month only if the eligibility group variable (ELGBLTY_GRP_CD) was not missing.
TAF claims data are organized by month and year of service. For IP and LT claims, the discharge date (or if that is missing, the service end date) is used to determine the month and year of the service. For OT claims, the service end date is used to determine the month and year of the service. If the service end date is missing, the service start date is used, and if the start date is missing, the most recent service end date from all the claim lines is used. For RX claims, the prescription fill date is used to determine the month and year of the service.
The DSH payment field appears only on IP claims. For LT, OT, and RX claims, we considered only the service tracking payment amount or total Medicaid paid amount.
States sometimes entered the same payment amount in all three payment fields (or two of the three payment fields). To avoid double or triple counting, we used only one payment field per header claim.
If the DSH payment field was zero or missing, we used the service tracking payment amount. If the DSH payment field and service tracking payment amount were both zero or missing, we used the total Medicaid paid amount.
More information about the FASC code can be found in the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page .
Header records summarize the service or the set of linked services provided to a beneficiary. Line records have more detailed information about the services provided. We used the header record for this analysis because most studies in which claims data are used will be based on the summary information in the header record.
We selected records with claim type code = 1, 2, 4, 5, A, B, D, or E.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We excluded the \u201cdummy\u201d records from the DE file that represent MSIS IDs present on claims but for which the state submitted no enrollment information. Dummy records can be identified as those with MISG_ELGBLTY_DATA_IND = 1.
If it was available, we used the monthly CHIP code (CHIP_CD) to identify beneficiaries enrolled in Medicaid or CHIP that month (CHIP_CD values of 1\u20134). CHIP code 4 (Medicaid and S-CHIP) is a valid value for 2014 through 2017 TAF. If the CHIP code was missing, we considered the beneficiary as enrolled in the month only if the eligibility group variable (ELGBLTY_GRP_CD) was not missing.
TAF claims data are organized by month and year of service. For IP and LT claims, the discharge date (or if that is missing, the service end date) is used to determine the month and year of the service. For OT claims, the service end date is used to determine the month and year of the service. If the service end date is missing, the service start date is used, and if the start date is missing, the most recent service end date from all the claim lines is used. For RX claims, the prescription fill date is used to determine the month and year of the service.
The DSH payment field appears only on IP claims. For LT, OT, and RX claims, we considered only the service tracking payment amount or total Medicaid paid amount.
States sometimes entered the same payment amount in all three payment fields (or two of the three payment fields). To avoid double or triple counting, we used only one payment field per header claim.
If the DSH payment field was zero or missing, we used the service tracking payment amount. If the DSH payment field and service tracking payment amount were both zero or missing, we used the total Medicaid paid amount.
More information about the FASC code can be found in the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page .
Most expenditures in the TAF claims files should link to a beneficiary using the unique identifier that states assign to each beneficiary, the MSIS ID. This analysis examines the proportion of expenditures in the TAF that can be linked to an eligibility record using the MSIS ID.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""6071"", ""relatedTopics"": []}" 111,"{""measureId"": 111, ""measureName"": ""Category of Service Code - IP"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Srvc-Cat-Cd-IP.pdf"", ""background"": {""content"": ""Medicaid and the Children’s Health Insurance Program (CHIP) are joint state-federal medical assistance programs in which the state administers the program and pays providers or health plans directly for covered services, and the federal government reimburses a set percentage of the state’s expenditures (referred to as the \""federal match\""). Because the federal government applies different federal matching rates to Medicaid and CHIP beneficiaries, states report Medicaid and CHIP expenditure data separately.
To claim federal matching funds for their medical assistance programs, states report aggregate expenditure data for Medicaid beneficiaries covered under Title XIX of the Social Security Act to the Centers for Medicare & Medicaid Services (CMS) on Form CMS-64, Quarterly Medicaid Statement of Expenditures. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act on Form CMS-21, Quarterly CHIP Statement of Expenditures . States may cover children using CHIP funds by expanding their Medicaid programs (called \""Medicaid expansion CHIP,\"" or M-CHIP), creating a program separate from their existing Medicaid programs (called \""separate CHIP,\"" or S-CHIP), or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries, which qualify for the higher CHIP federal match rate, are reported on Form CMS-21 and not Form CMS-64.
There are three data elements in the T-MSIS Analytic Files (TAF) inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) files that contain information about how states report Medicaid and CHIP expenditures to claim federal matching funds: the federal reimbursement category code, the CMS-64 category of service code, and the CMS-21 category of service code (Table 1). [1] The federal reimbursement category code indicates whether the expenditure was reimbursed under Title XIX, Title XXI, the Affordable Care Act (ACA), or other legislation. As noted earlier, if the expenditure was covered by Title XIX of the Social Security Act, it should be reported on Form CMS-64; if the expenditure was covered by Title XXI of the Social Security Act it should be reported on Form CMS-21. The CMS-64 category of service code indicates the specific category of service on Form CMS-64 into which the state will report the Medicaid expenditures associated with a paid claim. The CMS-21 category of service code indicates the specific category of service on Form CMS-21 into which the state will report the CHIP expenditures associated with a paid claim.
Table 1. TAF variables that classify expenditures into federal reimbursement categories
| Variable | TAF data element name (TAF RIF name in parentheses if different) | IP, LT, OT, and RX records expected to have non-missing values | Description |
|---|---|---|---|
| Federal reimbursement category code | CMS_64_FED_REIMBRSMT_CTGRY_CD (CMS_64_FED_CTGRY_CD) | All records | This code indicates whether the claim was matched with federal funding under Title XIX (value of ‘01’), Title XXI (‘02’), the Affordable Care Act (‘03’), a or federal funding under other legislation (‘04’). |
| CMS-64 category of service code | XIX_SRVC_CTGRY_CD | Only records representing expenditures that the state will report on the CMS-64, which should link to beneficiaries enrolled in non-CHIP Medicaid | A four-digit code (from 001A to 0050) indicating the line on the Form CMS-64 on which the state reported the expenditures associated with the claim. The Form CMS-64 is used to report expenditures that qualify for federal matching funds under Title XXI. A full list of the category of service and definition for each line can be found at https://www.medicaid.gov/medicaid/downloads/cms-649-base-category-of-services-definition-2-14.pdf . |
| CMS-21 category of service code | XIX_SRVC_CTGRY_CD | Only records representing expenditures that the state will report on the CMS-21, which should link to beneficiaries enrolled in Medicaid-expansion CHIP or Separate CHIP | A three-digit code (from 01A to 35B) indicating the category of service (or line) on the Form CMS-21 on which the state reported the expenditures associated with the claim. The Form CMS-21 is used to report expenditures that qualify for federal matching funds under Title XXI. A full list of the category of service and definition for each line can be found at https://www.medicaid.gov/medicaid/downloads/cms-21-base-category-of-services-definition-2-14.pdf . |
a The federal reimbursement category code \""03\"" was previously a valid value used by some states on claims for their ACA adult expansion population. However, CMS retired the \""03\"" value in September 2020 and states are expected to use the \""01\"" value for this population moving forward since the enhanced match rate for these enrollees is authorized under Title XIX.
If states are populating these data elements correctly, claim records that link to beneficiaries enrolled in Title XIX Medicaid will be coded to indicate the expenditure is reportable on Form CMS-64. [2] Similarly, claim records that link to beneficiaries enrolled in Title XXI CHIP—including both M-CHIP and S-CHIP programs—will be coded to indicate the expenditure is reportable on Form CMS-21.
Other TAF data elements can be used to summarize Medicaid and CHIP spending associated with certain services or benefit categories, including the state-assigned type of service code, the federally-assigned service category code, and the benefit type code. [3] However, those data elements will not necessarily line up with the CMS-64 or CMS-21 categories that states use to report expenditures for the purposes of federal reimbursement.
In this data quality assessment, we evaluate the percentage of records in each claims file reported with a federal reimbursement category code that is consistent with the beneficiary’s enrollment in the Medicaid or CHIP programs. We also evaluate the percentage of records in each claims file reported with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid or CHIP program.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
More information about differences between expenditures in the TAF and the CMS-64 data can be found in the DQ Atlas single topic displays under the Expenditure Benchmarking topic area.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX . For more information about the FASC code, see the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
More information about differences between expenditures in the TAF and the CMS-64 data can be found in the DQ Atlas single topic displays under the Expenditure Benchmarking topic area.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX . For more information about the FASC code, see the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
We began by selecting all records from the IP, LT, OT, and RX claims files that represented Medicaid and S-CHIP fee-for-service (FFS), capitation, and supplemental payments based on the claim type code. [4] , [5] We excluded managed care encounter records from the analysis because the payments recorded on those claims do not represent state expenditures and are not recorded on the CMS-64 or CMS-21. [6] We also excluded service tracking claims from the analysis because these records are lump-sum payments that that cannot be linked to an eligibility record. [7]
To determine whether the beneficiary was enrolled in Medicaid, M-CHIP, or S-CHIP on the date of service, we linked the claims records to the matching eligibility record using the unique beneficiary identifier coded on the claim (MSIS ID). We determined Medicaid, M-CHIP, and S-CHIP enrollment by examining the CHIP code on records in the TAF Demographic and Eligibility (DE) file from the month that corresponded to the date of service on the claim. We excluded from the analysis claims records with a missing or invalid beneficiary identifier. Additionally, we excluded claims that did not link to an eligibility record, as well as those that linked to an eligibility record but did not have a CHIP code populated (indicating the individual was enrolled in Medicaid or CHIP) during the date of service on the claim. [8]
Finally, we calculated the percentage of records in each file with a federal reimbursement category code consistent with the beneficiary’s reported enrollment in Medicaid, M-CHIP, or S-CHIP. The expected coding for federal reimbursement category code based on CHIP code is shown in Table 2. Because Medicaid beneficiaries enrolled in an adult expansion eligibility group may have a different federal reimbursement category code than Medicaid beneficiaries not enrolled in such a group, we examined these two groups of Medicaid beneficiaries separately. [9] We also calculated the percentage of records with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid or CHIP program. The expected coding for CMS-64 and CMS-21 category of service code based on CHIP code is shown in Table 3.
Table 2. Expected coding for federal reimbursement category code
| Enrollment type | CHIP code on eligibility record (CHIP_CD) | Latest eligibility group on eligibility record (ELGBLTY_GRP_CD_LTST) | Expected federal reimbursement category code on claim record |
|---|---|---|---|
| Medicaid – Adult expansion group | 1 (Medicaid) | 72 (Adult Group - Individuals at or below 133% FPL Age 19 through 64 - newly eligible for all states) 73 (Adult Group - Individuals at or below 133% FPL Age 19 through 64- not newly eligible for non 1905z(3) states), 74 (Adult Group - Individuals at or below 133% FPL Age 19 through 64 – not newly eligible parent/ caretaker relative(s) in 1905z(3) states), or 75 (Adult Group - Individuals at or below 133% FPL Age 19 through 64- not newly eligible non-parent/ caretaker-relative(s) in 1905z(3) states) | 01 (Federal funding under Title XIX) or 03 (Federal funding under ACA) a |
| Medicaid – All other | 1 (Medicaid) | Any valid code or null/missing, except: 72, 73, 74, 75, or 61-68 (CHIP eligibility groups) | 01 (Federal funding under Title XIX) |
| M-CHIP | 2 (M-CHIP) | N/A | 02 (Federal funding under Title XXI) |
| S-CHIP | 3 (S-CHIP) | N/A | 02 (Federal funding under Title XXI) |
a In this assessment, we consider both \""01\"" and \""03\"" to be consistent values for Medicaid-enrolled beneficiaries in the adult expansion eligibility groups. The value \""03\"" (Federal funding under ACA) was retired as a valid value on September 30, 2020 and while the value is still present in the data before that date, we would expect to see relatively few claims with this value after that date.
ACA = Affordable Care Act; FPL = Federal Poverty Level; N/A = not available.
Table 3. Expected coding for CMS-64 and CMS-21 category of service codes
| CHIP code on eligibility record (CHIP_CD) | Expected CMS-64 category of service code on claims record | Expected CMS-21 category of service code on claims record |
|---|---|---|
| 1 (Medicaid) |
|
|
| 2 (M-CHIP) | Field should be blank |
|
| 3 (S-CHIP) | Field should be blank |
|
Data quality assessment criteria
We categorized each state as having low, medium, or high data quality concern for the federal reimbursement category code, depending on the percentage of records with a federal reimbursement category code consistent with the beneficiary’s enrollment in the Medicaid, M-CHIP, or S-CHIP programs, as shown in Table 4.
Table 4. Criteria for DQ assessment of federal reimbursement category code
| Percentage of claim lines with a federal reimbursement category code consistent with program enrollment | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
We categorized each state as having low, medium, or high data quality concern for the category of service code, depending on the percentage of records with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid, M-CHIP, or S-CHIP programs, as shown in Table 5. As shown in Table 3, only the CMS-64 or CMS-21 category of service code is expected to be populated on a single claim record. When both fields were populated on the same claim (even if one or both values were invalid), that record was inconsistent with enrollment in the Medicaid, M-CHIP, or S-CHIP programs for the purposes of this analysis.
Table 5. Criteria for DQ assessment of category of service codes
| Percentage of claim lines with a CMS-64 category of service code or a CMS-21 category of service code consistent with program enrollment | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
FFS claims representing Medicaid or M-CHIP services have a claim type code (CLM_TYPE_CD) value of 1, capitation payments have a claim type code value of 2, and supplemental payments have a claim type code value of 5. FFS claims representing S-CHIP benefits have a claim type code value of A, capitation payments have a claim type code value of B, and supplemental payments have a claim type code value of E.
Payment data on managed care encounter records are redacted from the TAF RIFs.
Managed care encounter records representing Medicaid or M-CHIP services have a claim type code value of 3 and service tracking claims have a claim type code value of 4. Managed care encounter records representing S-CHIP benefits have a claim type code value of C, and service tracking claims have a claim type code value of D.
Once an individual is determined eligible for Medicaid, coverage is effective either on the date of application or the first day of the month of application. Benefits also may be covered retroactively for up to three months prior to the month of application if the individual would have been eligible during that period had he or she applied. States are required to update beneficiaries’ eligibility records in T-MSIS to reflect the retroactive coverage period. However, if a state did not update the eligibility record of a beneficiary to reflect the retroactive coverage period, enrollment information like CHIP code may be missing during the date of service on a paid claim.
We determined the eligibility group by using ELGBLTY_GRP_CD_LTST, the most recent eligibility group reported for the beneficiary in the calendar year.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
FFS claims representing Medicaid or M-CHIP services have a claim type code (CLM_TYPE_CD) value of 1, capitation payments have a claim type code value of 2, and supplemental payments have a claim type code value of 5. FFS claims representing S-CHIP benefits have a claim type code value of A, capitation payments have a claim type code value of B, and supplemental payments have a claim type code value of E.
Payment data on managed care encounter records are redacted from the TAF RIFs.
Managed care encounter records representing Medicaid or M-CHIP services have a claim type code value of 3 and service tracking claims have a claim type code value of 4. Managed care encounter records representing S-CHIP benefits have a claim type code value of C, and service tracking claims have a claim type code value of D.
Once an individual is determined eligible for Medicaid, coverage is effective either on the date of application or the first day of the month of application. Benefits also may be covered retroactively for up to three months prior to the month of application if the individual would have been eligible during that period had he or she applied. States are required to update beneficiaries\u2019 eligibility records in T-MSIS to reflect the retroactive coverage period. However, if a state did not update the eligibility record of a beneficiary to reflect the retroactive coverage period, enrollment information like CHIP code may be missing during the date of service on a paid claim.
We determined the eligibility group by using ELGBLTY_GRP_CD_LTST, the most recent eligibility group reported for the beneficiary in the calendar year.
States report quarterly Medicaid expenditures on Form CMS-64 and quarterly Children's Health Insurance Program (CHIP) expenditures on Form CMS-21 to claim federal matching funds. In the T-MSIS Analytic Files (TAF), the CMS-64 category of service code data element on paid claim records should reflect the category where expenditures for the service are reported on Form CMS-64. Similarly, the CMS-21 category of service code on a paid claim record should reflect the category where expenditures for the service are reported on Form CMS-21. This analysis examines the extent to which records in the IP file are reported with a CMS-64 or CMS-21 category of service code value consistent with the beneficiary's enrollment in the Medicaid or CHIP program.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5221"", ""relatedTopics"": [{""measureId"": 112, ""measureName"": ""Category of Service Code - LT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 1}, {""measureId"": 113, ""measureName"": ""Category of Service Code - OT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 2}, {""measureId"": 114, ""measureName"": ""Category of Service Code - RX"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 3}, {""measureId"": 115, ""measureName"": ""Federal Reimbursement Category Code - IP"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 4}, {""measureId"": 116, ""measureName"": ""Federal Reimbursement Category Code - LT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 5}, {""measureId"": 117, ""measureName"": ""Federal Reimbursement Category Code - OT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 6}, {""measureId"": 118, ""measureName"": ""Federal Reimbursement Category Code - RX"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 7}]}" 112,"{""measureId"": 112, ""measureName"": ""Category of Service Code - LT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Srvc-Cat-Cd-LT.pdf"", ""background"": {""content"": ""Medicaid and the Children’s Health Insurance Program (CHIP) are joint state-federal medical assistance programs in which the state administers the program and pays providers or health plans directly for covered services, and the federal government reimburses a set percentage of the state’s expenditures (referred to as the \""federal match\""). Because the federal government applies different federal matching rates to Medicaid and CHIP beneficiaries, states report Medicaid and CHIP expenditure data separately.
To claim federal matching funds for their medical assistance programs, states report aggregate expenditure data for Medicaid beneficiaries covered under Title XIX of the Social Security Act to the Centers for Medicare & Medicaid Services (CMS) on Form CMS-64, Quarterly Medicaid Statement of Expenditures. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act on Form CMS-21, Quarterly CHIP Statement of Expenditures . States may cover children using CHIP funds by expanding their Medicaid programs (called \""Medicaid expansion CHIP,\"" or M-CHIP), creating a program separate from their existing Medicaid programs (called \""separate CHIP,\"" or S-CHIP), or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries, which qualify for the higher CHIP federal match rate, are reported on Form CMS-21 and not Form CMS-64.
There are three data elements in the T-MSIS Analytic Files (TAF) inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) files that contain information about how states report Medicaid and CHIP expenditures to claim federal matching funds: the federal reimbursement category code, the CMS-64 category of service code, and the CMS-21 category of service code (Table 1). [1] The federal reimbursement category code indicates whether the expenditure was reimbursed under Title XIX, Title XXI, the Affordable Care Act (ACA), or other legislation. As noted earlier, if the expenditure was covered by Title XIX of the Social Security Act, it should be reported on Form CMS-64; if the expenditure was covered by Title XXI of the Social Security Act it should be reported on Form CMS-21. The CMS-64 category of service code indicates the specific category of service on Form CMS-64 into which the state will report the Medicaid expenditures associated with a paid claim. The CMS-21 category of service code indicates the specific category of service on Form CMS-21 into which the state will report the CHIP expenditures associated with a paid claim.
Table 1. TAF variables that classify expenditures into federal reimbursement categories
| Variable | TAF data element name (TAF RIF name in parentheses if different) | IP, LT, OT, and RX records expected to have non-missing values | Description |
|---|---|---|---|
| Federal reimbursement category code | CMS_64_FED_REIMBRSMT_CTGRY_CD (CMS_64_FED_CTGRY_CD) | All records | This code indicates whether the claim was matched with federal funding under Title XIX (value of ‘01’), Title XXI (‘02’), the Affordable Care Act (‘03’), a or federal funding under other legislation (‘04’). |
| CMS-64 category of service code | XIX_SRVC_CTGRY_CD | Only records representing expenditures that the state will report on the CMS-64, which should link to beneficiaries enrolled in non-CHIP Medicaid | A four-digit code (from 001A to 0050) indicating the line on the Form CMS-64 on which the state reported the expenditures associated with the claim. The Form CMS-64 is used to report expenditures that qualify for federal matching funds under Title XXI. A full list of the category of service and definition for each line can be found at https://www.medicaid.gov/medicaid/downloads/cms-649-base-category-of-services-definition-2-14.pdf . |
| CMS-21 category of service code | XIX_SRVC_CTGRY_CD | Only records representing expenditures that the state will report on the CMS-21, which should link to beneficiaries enrolled in Medicaid-expansion CHIP or Separate CHIP | A three-digit code (from 01A to 35B) indicating the category of service (or line) on the Form CMS-21 on which the state reported the expenditures associated with the claim. The Form CMS-21 is used to report expenditures that qualify for federal matching funds under Title XXI. A full list of the category of service and definition for each line can be found at https://www.medicaid.gov/medicaid/downloads/cms-21-base-category-of-services-definition-2-14.pdf . |
a The federal reimbursement category code \""03\"" was previously a valid value used by some states on claims for their ACA adult expansion population. However, CMS retired the \""03\"" value in September 2020 and states are expected to use the \""01\"" value for this population moving forward since the enhanced match rate for these enrollees is authorized under Title XIX.
If states are populating these data elements correctly, claim records that link to beneficiaries enrolled in Title XIX Medicaid will be coded to indicate the expenditure is reportable on Form CMS-64. [2] Similarly, claim records that link to beneficiaries enrolled in Title XXI CHIP—including both M-CHIP and S-CHIP programs—will be coded to indicate the expenditure is reportable on Form CMS-21.
Other TAF data elements can be used to summarize Medicaid and CHIP spending associated with certain services or benefit categories, including the state-assigned type of service code, the federally-assigned service category code, and the benefit type code. [3] However, those data elements will not necessarily line up with the CMS-64 or CMS-21 categories that states use to report expenditures for the purposes of federal reimbursement.
In this data quality assessment, we evaluate the percentage of records in each claims file reported with a federal reimbursement category code that is consistent with the beneficiary’s enrollment in the Medicaid or CHIP programs. We also evaluate the percentage of records in each claims file reported with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid or CHIP program.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
More information about differences between expenditures in the TAF and the CMS-64 data can be found in the DQ Atlas single topic displays under the Expenditure Benchmarking topic area.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX . For more information about the FASC code, see the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
More information about differences between expenditures in the TAF and the CMS-64 data can be found in the DQ Atlas single topic displays under the Expenditure Benchmarking topic area.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX . For more information about the FASC code, see the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
We began by selecting all records from the IP, LT, OT, and RX claims files that represented Medicaid and S-CHIP fee-for-service (FFS), capitation, and supplemental payments based on the claim type code. [4] , [5] We excluded managed care encounter records from the analysis because the payments recorded on those claims do not represent state expenditures and are not recorded on the CMS-64 or CMS-21. [6] We also excluded service tracking claims from the analysis because these records are lump-sum payments that that cannot be linked to an eligibility record. [7]
To determine whether the beneficiary was enrolled in Medicaid, M-CHIP, or S-CHIP on the date of service, we linked the claims records to the matching eligibility record using the unique beneficiary identifier coded on the claim (MSIS ID). We determined Medicaid, M-CHIP, and S-CHIP enrollment by examining the CHIP code on records in the TAF Demographic and Eligibility (DE) file from the month that corresponded to the date of service on the claim. We excluded from the analysis claims records with a missing or invalid beneficiary identifier. Additionally, we excluded claims that did not link to an eligibility record, as well as those that linked to an eligibility record but did not have a CHIP code populated (indicating the individual was enrolled in Medicaid or CHIP) during the date of service on the claim. [8]
Finally, we calculated the percentage of records in each file with a federal reimbursement category code consistent with the beneficiary’s reported enrollment in Medicaid, M-CHIP, or S-CHIP. The expected coding for federal reimbursement category code based on CHIP code is shown in Table 2. Because Medicaid beneficiaries enrolled in an adult expansion eligibility group may have a different federal reimbursement category code than Medicaid beneficiaries not enrolled in such a group, we examined these two groups of Medicaid beneficiaries separately. [9] We also calculated the percentage of records with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid or CHIP program. The expected coding for CMS-64 and CMS-21 category of service code based on CHIP code is shown in Table 3.
Table 2. Expected coding for federal reimbursement category code
| Enrollment type | CHIP code on eligibility record (CHIP_CD) | Latest eligibility group on eligibility record (ELGBLTY_GRP_CD_LTST) | Expected federal reimbursement category code on claim record |
|---|---|---|---|
| Medicaid – Adult expansion group | 1 (Medicaid) | 72 (Adult Group - Individuals at or below 133% FPL Age 19 through 64 - newly eligible for all states) 73 (Adult Group - Individuals at or below 133% FPL Age 19 through 64- not newly eligible for non 1905z(3) states), 74 (Adult Group - Individuals at or below 133% FPL Age 19 through 64 – not newly eligible parent/ caretaker relative(s) in 1905z(3) states), or 75 (Adult Group - Individuals at or below 133% FPL Age 19 through 64- not newly eligible non-parent/ caretaker-relative(s) in 1905z(3) states) | 01 (Federal funding under Title XIX) or 03 (Federal funding under ACA) a |
| Medicaid – All other | 1 (Medicaid) | Any valid code or null/missing, except: 72, 73, 74, 75, or 61-68 (CHIP eligibility groups) | 01 (Federal funding under Title XIX) |
| M-CHIP | 2 (M-CHIP) | N/A | 02 (Federal funding under Title XXI) |
| S-CHIP | 3 (S-CHIP) | N/A | 02 (Federal funding under Title XXI) |
a In this assessment, we consider both \""01\"" and \""03\"" to be consistent values for Medicaid-enrolled beneficiaries in the adult expansion eligibility groups. The value \""03\"" (Federal funding under ACA) was retired as a valid value on September 30, 2020 and while the value is still present in the data before that date, we would expect to see relatively few claims with this value after that date.
ACA = Affordable Care Act; FPL = Federal Poverty Level; N/A = not available.
Table 3. Expected coding for CMS-64 and CMS-21 category of service codes
| CHIP code on eligibility record (CHIP_CD) | Expected CMS-64 category of service code on claims record | Expected CMS-21 category of service code on claims record |
|---|---|---|
| 1 (Medicaid) |
|
|
| 2 (M-CHIP) | Field should be blank |
|
| 3 (S-CHIP) | Field should be blank |
|
Data quality assessment criteria
We categorized each state as having low, medium, or high data quality concern for the federal reimbursement category code, depending on the percentage of records with a federal reimbursement category code consistent with the beneficiary’s enrollment in the Medicaid, M-CHIP, or S-CHIP programs, as shown in Table 4.
Table 4. Criteria for DQ assessment of federal reimbursement category code
| Percentage of claim lines with a federal reimbursement category code consistent with program enrollment | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
We categorized each state as having low, medium, or high data quality concern for the category of service code, depending on the percentage of records with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid, M-CHIP, or S-CHIP programs, as shown in Table 5. As shown in Table 3, only the CMS-64 or CMS-21 category of service code is expected to be populated on a single claim record. When both fields were populated on the same claim (even if one or both values were invalid), that record was inconsistent with enrollment in the Medicaid, M-CHIP, or S-CHIP programs for the purposes of this analysis.
Table 5. Criteria for DQ assessment of category of service codes
| Percentage of claim lines with a CMS-64 category of service code or a CMS-21 category of service code consistent with program enrollment | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
FFS claims representing Medicaid or M-CHIP services have a claim type code (CLM_TYPE_CD) value of 1, capitation payments have a claim type code value of 2, and supplemental payments have a claim type code value of 5. FFS claims representing S-CHIP benefits have a claim type code value of A, capitation payments have a claim type code value of B, and supplemental payments have a claim type code value of E.
Payment data on managed care encounter records are redacted from the TAF RIFs.
Managed care encounter records representing Medicaid or M-CHIP services have a claim type code value of 3 and service tracking claims have a claim type code value of 4. Managed care encounter records representing S-CHIP benefits have a claim type code value of C, and service tracking claims have a claim type code value of D.
Once an individual is determined eligible for Medicaid, coverage is effective either on the date of application or the first day of the month of application. Benefits also may be covered retroactively for up to three months prior to the month of application if the individual would have been eligible during that period had he or she applied. States are required to update beneficiaries’ eligibility records in T-MSIS to reflect the retroactive coverage period. However, if a state did not update the eligibility record of a beneficiary to reflect the retroactive coverage period, enrollment information like CHIP code may be missing during the date of service on a paid claim.
We determined the eligibility group by using ELGBLTY_GRP_CD_LTST, the most recent eligibility group reported for the beneficiary in the calendar year.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
FFS claims representing Medicaid or M-CHIP services have a claim type code (CLM_TYPE_CD) value of 1, capitation payments have a claim type code value of 2, and supplemental payments have a claim type code value of 5. FFS claims representing S-CHIP benefits have a claim type code value of A, capitation payments have a claim type code value of B, and supplemental payments have a claim type code value of E.
Payment data on managed care encounter records are redacted from the TAF RIFs.
Managed care encounter records representing Medicaid or M-CHIP services have a claim type code value of 3 and service tracking claims have a claim type code value of 4. Managed care encounter records representing S-CHIP benefits have a claim type code value of C, and service tracking claims have a claim type code value of D.
Once an individual is determined eligible for Medicaid, coverage is effective either on the date of application or the first day of the month of application. Benefits also may be covered retroactively for up to three months prior to the month of application if the individual would have been eligible during that period had he or she applied. States are required to update beneficiaries\u2019 eligibility records in T-MSIS to reflect the retroactive coverage period. However, if a state did not update the eligibility record of a beneficiary to reflect the retroactive coverage period, enrollment information like CHIP code may be missing during the date of service on a paid claim.
We determined the eligibility group by using ELGBLTY_GRP_CD_LTST, the most recent eligibility group reported for the beneficiary in the calendar year.
States report quarterly Medicaid expenditures on Form CMS-64 and quarterly Children's Health Insurance Program (CHIP) expenditures on Form CMS-21 to claim federal matching funds. In the T-MSIS Analytic Files (TAF), the CMS-64 category of service code data element on paid claim records should reflect the category where expenditures for the service are reported on Form CMS-64. Similarly, the CMS-21 category of service code on a paid claim record should reflect the category where expenditures for the service are reported on Form CMS-21. This analysis examines the extent to which records in the LT file are reported with a CMS-64 or CMS-21 category of service code value consistent with the beneficiary's enrollment in the Medicaid or CHIP program.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5221"", ""relatedTopics"": [{""measureId"": 111, ""measureName"": ""Category of Service Code - IP"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 0}, {""measureId"": 113, ""measureName"": ""Category of Service Code - OT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 2}, {""measureId"": 114, ""measureName"": ""Category of Service Code - RX"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 3}, {""measureId"": 115, ""measureName"": ""Federal Reimbursement Category Code - IP"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 4}, {""measureId"": 116, ""measureName"": ""Federal Reimbursement Category Code - LT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 5}, {""measureId"": 117, ""measureName"": ""Federal Reimbursement Category Code - OT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 6}, {""measureId"": 118, ""measureName"": ""Federal Reimbursement Category Code - RX"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 7}]}" 113,"{""measureId"": 113, ""measureName"": ""Category of Service Code - OT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Srvc-Cat-Cd-OT.pdf"", ""background"": {""content"": ""Medicaid and the Children’s Health Insurance Program (CHIP) are joint state-federal medical assistance programs in which the state administers the program and pays providers or health plans directly for covered services, and the federal government reimburses a set percentage of the state’s expenditures (referred to as the \""federal match\""). Because the federal government applies different federal matching rates to Medicaid and CHIP beneficiaries, states report Medicaid and CHIP expenditure data separately.
To claim federal matching funds for their medical assistance programs, states report aggregate expenditure data for Medicaid beneficiaries covered under Title XIX of the Social Security Act to the Centers for Medicare & Medicaid Services (CMS) on Form CMS-64, Quarterly Medicaid Statement of Expenditures. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act on Form CMS-21, Quarterly CHIP Statement of Expenditures . States may cover children using CHIP funds by expanding their Medicaid programs (called \""Medicaid expansion CHIP,\"" or M-CHIP), creating a program separate from their existing Medicaid programs (called \""separate CHIP,\"" or S-CHIP), or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries, which qualify for the higher CHIP federal match rate, are reported on Form CMS-21 and not Form CMS-64.
There are three data elements in the T-MSIS Analytic Files (TAF) inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) files that contain information about how states report Medicaid and CHIP expenditures to claim federal matching funds: the federal reimbursement category code, the CMS-64 category of service code, and the CMS-21 category of service code (Table 1). [1] The federal reimbursement category code indicates whether the expenditure was reimbursed under Title XIX, Title XXI, the Affordable Care Act (ACA), or other legislation. As noted earlier, if the expenditure was covered by Title XIX of the Social Security Act, it should be reported on Form CMS-64; if the expenditure was covered by Title XXI of the Social Security Act it should be reported on Form CMS-21. The CMS-64 category of service code indicates the specific category of service on Form CMS-64 into which the state will report the Medicaid expenditures associated with a paid claim. The CMS-21 category of service code indicates the specific category of service on Form CMS-21 into which the state will report the CHIP expenditures associated with a paid claim.
Table 1. TAF variables that classify expenditures into federal reimbursement categories
| Variable | TAF data element name (TAF RIF name in parentheses if different) | IP, LT, OT, and RX records expected to have non-missing values | Description |
|---|---|---|---|
| Federal reimbursement category code | CMS_64_FED_REIMBRSMT_CTGRY_CD (CMS_64_FED_CTGRY_CD) | All records | This code indicates whether the claim was matched with federal funding under Title XIX (value of ‘01’), Title XXI (‘02’), the Affordable Care Act (‘03’), a or federal funding under other legislation (‘04’). |
| CMS-64 category of service code | XIX_SRVC_CTGRY_CD | Only records representing expenditures that the state will report on the CMS-64, which should link to beneficiaries enrolled in non-CHIP Medicaid | A four-digit code (from 001A to 0050) indicating the line on the Form CMS-64 on which the state reported the expenditures associated with the claim. The Form CMS-64 is used to report expenditures that qualify for federal matching funds under Title XXI. A full list of the category of service and definition for each line can be found at https://www.medicaid.gov/medicaid/downloads/cms-649-base-category-of-services-definition-2-14.pdf . |
| CMS-21 category of service code | XIX_SRVC_CTGRY_CD | Only records representing expenditures that the state will report on the CMS-21, which should link to beneficiaries enrolled in Medicaid-expansion CHIP or Separate CHIP | A three-digit code (from 01A to 35B) indicating the category of service (or line) on the Form CMS-21 on which the state reported the expenditures associated with the claim. The Form CMS-21 is used to report expenditures that qualify for federal matching funds under Title XXI. A full list of the category of service and definition for each line can be found at https://www.medicaid.gov/medicaid/downloads/cms-21-base-category-of-services-definition-2-14.pdf . |
a The federal reimbursement category code \""03\"" was previously a valid value used by some states on claims for their ACA adult expansion population. However, CMS retired the \""03\"" value in September 2020 and states are expected to use the \""01\"" value for this population moving forward since the enhanced match rate for these enrollees is authorized under Title XIX.
If states are populating these data elements correctly, claim records that link to beneficiaries enrolled in Title XIX Medicaid will be coded to indicate the expenditure is reportable on Form CMS-64. [2] Similarly, claim records that link to beneficiaries enrolled in Title XXI CHIP—including both M-CHIP and S-CHIP programs—will be coded to indicate the expenditure is reportable on Form CMS-21.
Other TAF data elements can be used to summarize Medicaid and CHIP spending associated with certain services or benefit categories, including the state-assigned type of service code, the federally-assigned service category code, and the benefit type code. [3] However, those data elements will not necessarily line up with the CMS-64 or CMS-21 categories that states use to report expenditures for the purposes of federal reimbursement.
In this data quality assessment, we evaluate the percentage of records in each claims file reported with a federal reimbursement category code that is consistent with the beneficiary’s enrollment in the Medicaid or CHIP programs. We also evaluate the percentage of records in each claims file reported with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid or CHIP program.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
More information about differences between expenditures in the TAF and the CMS-64 data can be found in the DQ Atlas single topic displays under the Expenditure Benchmarking topic area.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX . For more information about the FASC code, see the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
More information about differences between expenditures in the TAF and the CMS-64 data can be found in the DQ Atlas single topic displays under the Expenditure Benchmarking topic area.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX . For more information about the FASC code, see the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
We began by selecting all records from the IP, LT, OT, and RX claims files that represented Medicaid and S-CHIP fee-for-service (FFS), capitation, and supplemental payments based on the claim type code. [4] , [5] We excluded managed care encounter records from the analysis because the payments recorded on those claims do not represent state expenditures and are not recorded on the CMS-64 or CMS-21. [6] We also excluded service tracking claims from the analysis because these records are lump-sum payments that that cannot be linked to an eligibility record. [7]
To determine whether the beneficiary was enrolled in Medicaid, M-CHIP, or S-CHIP on the date of service, we linked the claims records to the matching eligibility record using the unique beneficiary identifier coded on the claim (MSIS ID). We determined Medicaid, M-CHIP, and S-CHIP enrollment by examining the CHIP code on records in the TAF Demographic and Eligibility (DE) file from the month that corresponded to the date of service on the claim. We excluded from the analysis claims records with a missing or invalid beneficiary identifier. Additionally, we excluded claims that did not link to an eligibility record, as well as those that linked to an eligibility record but did not have a CHIP code populated (indicating the individual was enrolled in Medicaid or CHIP) during the date of service on the claim. [8]
Finally, we calculated the percentage of records in each file with a federal reimbursement category code consistent with the beneficiary’s reported enrollment in Medicaid, M-CHIP, or S-CHIP. The expected coding for federal reimbursement category code based on CHIP code is shown in Table 2. Because Medicaid beneficiaries enrolled in an adult expansion eligibility group may have a different federal reimbursement category code than Medicaid beneficiaries not enrolled in such a group, we examined these two groups of Medicaid beneficiaries separately. [9] We also calculated the percentage of records with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid or CHIP program. The expected coding for CMS-64 and CMS-21 category of service code based on CHIP code is shown in Table 3.
Table 2. Expected coding for federal reimbursement category code
| Enrollment type | CHIP code on eligibility record (CHIP_CD) | Latest eligibility group on eligibility record (ELGBLTY_GRP_CD_LTST) | Expected federal reimbursement category code on claim record |
|---|---|---|---|
| Medicaid – Adult expansion group | 1 (Medicaid) | 72 (Adult Group - Individuals at or below 133% FPL Age 19 through 64 - newly eligible for all states) 73 (Adult Group - Individuals at or below 133% FPL Age 19 through 64- not newly eligible for non 1905z(3) states), 74 (Adult Group - Individuals at or below 133% FPL Age 19 through 64 – not newly eligible parent/ caretaker relative(s) in 1905z(3) states), or 75 (Adult Group - Individuals at or below 133% FPL Age 19 through 64- not newly eligible non-parent/ caretaker-relative(s) in 1905z(3) states) | 01 (Federal funding under Title XIX) or 03 (Federal funding under ACA) a |
| Medicaid – All other | 1 (Medicaid) | Any valid code or null/missing, except: 72, 73, 74, 75, or 61-68 (CHIP eligibility groups) | 01 (Federal funding under Title XIX) |
| M-CHIP | 2 (M-CHIP) | N/A | 02 (Federal funding under Title XXI) |
| S-CHIP | 3 (S-CHIP) | N/A | 02 (Federal funding under Title XXI) |
a In this assessment, we consider both \""01\"" and \""03\"" to be consistent values for Medicaid-enrolled beneficiaries in the adult expansion eligibility groups. The value \""03\"" (Federal funding under ACA) was retired as a valid value on September 30, 2020 and while the value is still present in the data before that date, we would expect to see relatively few claims with this value after that date.
ACA = Affordable Care Act; FPL = Federal Poverty Level; N/A = not available.
Table 3. Expected coding for CMS-64 and CMS-21 category of service codes
| CHIP code on eligibility record (CHIP_CD) | Expected CMS-64 category of service code on claims record | Expected CMS-21 category of service code on claims record |
|---|---|---|
| 1 (Medicaid) |
|
|
| 2 (M-CHIP) | Field should be blank |
|
| 3 (S-CHIP) | Field should be blank |
|
Data quality assessment criteria
We categorized each state as having low, medium, or high data quality concern for the federal reimbursement category code, depending on the percentage of records with a federal reimbursement category code consistent with the beneficiary’s enrollment in the Medicaid, M-CHIP, or S-CHIP programs, as shown in Table 4.
Table 4. Criteria for DQ assessment of federal reimbursement category code
| Percentage of claim lines with a federal reimbursement category code consistent with program enrollment | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
We categorized each state as having low, medium, or high data quality concern for the category of service code, depending on the percentage of records with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid, M-CHIP, or S-CHIP programs, as shown in Table 5. As shown in Table 3, only the CMS-64 or CMS-21 category of service code is expected to be populated on a single claim record. When both fields were populated on the same claim (even if one or both values were invalid), that record was inconsistent with enrollment in the Medicaid, M-CHIP, or S-CHIP programs for the purposes of this analysis.
Table 5. Criteria for DQ assessment of category of service codes
| Percentage of claim lines with a CMS-64 category of service code or a CMS-21 category of service code consistent with program enrollment | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
FFS claims representing Medicaid or M-CHIP services have a claim type code (CLM_TYPE_CD) value of 1, capitation payments have a claim type code value of 2, and supplemental payments have a claim type code value of 5. FFS claims representing S-CHIP benefits have a claim type code value of A, capitation payments have a claim type code value of B, and supplemental payments have a claim type code value of E.
Payment data on managed care encounter records are redacted from the TAF RIFs.
Managed care encounter records representing Medicaid or M-CHIP services have a claim type code value of 3 and service tracking claims have a claim type code value of 4. Managed care encounter records representing S-CHIP benefits have a claim type code value of C, and service tracking claims have a claim type code value of D.
Once an individual is determined eligible for Medicaid, coverage is effective either on the date of application or the first day of the month of application. Benefits also may be covered retroactively for up to three months prior to the month of application if the individual would have been eligible during that period had he or she applied. States are required to update beneficiaries’ eligibility records in T-MSIS to reflect the retroactive coverage period. However, if a state did not update the eligibility record of a beneficiary to reflect the retroactive coverage period, enrollment information like CHIP code may be missing during the date of service on a paid claim.
We determined the eligibility group by using ELGBLTY_GRP_CD_LTST, the most recent eligibility group reported for the beneficiary in the calendar year.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
FFS claims representing Medicaid or M-CHIP services have a claim type code (CLM_TYPE_CD) value of 1, capitation payments have a claim type code value of 2, and supplemental payments have a claim type code value of 5. FFS claims representing S-CHIP benefits have a claim type code value of A, capitation payments have a claim type code value of B, and supplemental payments have a claim type code value of E.
Payment data on managed care encounter records are redacted from the TAF RIFs.
Managed care encounter records representing Medicaid or M-CHIP services have a claim type code value of 3 and service tracking claims have a claim type code value of 4. Managed care encounter records representing S-CHIP benefits have a claim type code value of C, and service tracking claims have a claim type code value of D.
Once an individual is determined eligible for Medicaid, coverage is effective either on the date of application or the first day of the month of application. Benefits also may be covered retroactively for up to three months prior to the month of application if the individual would have been eligible during that period had he or she applied. States are required to update beneficiaries\u2019 eligibility records in T-MSIS to reflect the retroactive coverage period. However, if a state did not update the eligibility record of a beneficiary to reflect the retroactive coverage period, enrollment information like CHIP code may be missing during the date of service on a paid claim.
We determined the eligibility group by using ELGBLTY_GRP_CD_LTST, the most recent eligibility group reported for the beneficiary in the calendar year.
States report quarterly Medicaid expenditures on Form CMS-64 and quarterly Children's Health Insurance Program (CHIP) expenditures on Form CMS-21 to claim federal matching funds. In the T-MSIS Analytic Files (TAF), the CMS-64 category of service code data element on paid claim records should reflect the category where expenditures for the service are reported on Form CMS-64. Similarly, the CMS-21 category of service code on a paid claim record should reflect the category where expenditures for the service are reported on Form CMS-21. This analysis examines the extent to which records in the OT file are reported with a CMS-64 or CMS-21 category of service code value consistent with the beneficiary's enrollment in the Medicaid or CHIP program.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5221"", ""relatedTopics"": [{""measureId"": 111, ""measureName"": ""Category of Service Code - IP"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 0}, {""measureId"": 112, ""measureName"": ""Category of Service Code - LT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 1}, {""measureId"": 114, ""measureName"": ""Category of Service Code - RX"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 3}, {""measureId"": 115, ""measureName"": ""Federal Reimbursement Category Code - IP"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 4}, {""measureId"": 116, ""measureName"": ""Federal Reimbursement Category Code - LT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 5}, {""measureId"": 117, ""measureName"": ""Federal Reimbursement Category Code - OT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 6}, {""measureId"": 118, ""measureName"": ""Federal Reimbursement Category Code - RX"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 7}]}" 114,"{""measureId"": 114, ""measureName"": ""Category of Service Code - RX"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Srvc-Cat-Cd-RX.pdf"", ""background"": {""content"": ""Medicaid and the Children’s Health Insurance Program (CHIP) are joint state-federal medical assistance programs in which the state administers the program and pays providers or health plans directly for covered services, and the federal government reimburses a set percentage of the state’s expenditures (referred to as the \""federal match\""). Because the federal government applies different federal matching rates to Medicaid and CHIP beneficiaries, states report Medicaid and CHIP expenditure data separately.
To claim federal matching funds for their medical assistance programs, states report aggregate expenditure data for Medicaid beneficiaries covered under Title XIX of the Social Security Act to the Centers for Medicare & Medicaid Services (CMS) on Form CMS-64, Quarterly Medicaid Statement of Expenditures. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act on Form CMS-21, Quarterly CHIP Statement of Expenditures . States may cover children using CHIP funds by expanding their Medicaid programs (called \""Medicaid expansion CHIP,\"" or M-CHIP), creating a program separate from their existing Medicaid programs (called \""separate CHIP,\"" or S-CHIP), or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries, which qualify for the higher CHIP federal match rate, are reported on Form CMS-21 and not Form CMS-64.
There are three data elements in the T-MSIS Analytic Files (TAF) inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) files that contain information about how states report Medicaid and CHIP expenditures to claim federal matching funds: the federal reimbursement category code, the CMS-64 category of service code, and the CMS-21 category of service code (Table 1). [1] The federal reimbursement category code indicates whether the expenditure was reimbursed under Title XIX, Title XXI, the Affordable Care Act (ACA), or other legislation. As noted earlier, if the expenditure was covered by Title XIX of the Social Security Act, it should be reported on Form CMS-64; if the expenditure was covered by Title XXI of the Social Security Act it should be reported on Form CMS-21. The CMS-64 category of service code indicates the specific category of service on Form CMS-64 into which the state will report the Medicaid expenditures associated with a paid claim. The CMS-21 category of service code indicates the specific category of service on Form CMS-21 into which the state will report the CHIP expenditures associated with a paid claim.
Table 1. TAF variables that classify expenditures into federal reimbursement categories
| Variable | TAF data element name (TAF RIF name in parentheses if different) | IP, LT, OT, and RX records expected to have non-missing values | Description |
|---|---|---|---|
| Federal reimbursement category code | CMS_64_FED_REIMBRSMT_CTGRY_CD (CMS_64_FED_CTGRY_CD) | All records | This code indicates whether the claim was matched with federal funding under Title XIX (value of ‘01’), Title XXI (‘02’), the Affordable Care Act (‘03’), a or federal funding under other legislation (‘04’). |
| CMS-64 category of service code | XIX_SRVC_CTGRY_CD | Only records representing expenditures that the state will report on the CMS-64, which should link to beneficiaries enrolled in non-CHIP Medicaid | A four-digit code (from 001A to 0050) indicating the line on the Form CMS-64 on which the state reported the expenditures associated with the claim. The Form CMS-64 is used to report expenditures that qualify for federal matching funds under Title XXI. A full list of the category of service and definition for each line can be found at https://www.medicaid.gov/medicaid/downloads/cms-649-base-category-of-services-definition-2-14.pdf . |
| CMS-21 category of service code | XIX_SRVC_CTGRY_CD | Only records representing expenditures that the state will report on the CMS-21, which should link to beneficiaries enrolled in Medicaid-expansion CHIP or Separate CHIP | A three-digit code (from 01A to 35B) indicating the category of service (or line) on the Form CMS-21 on which the state reported the expenditures associated with the claim. The Form CMS-21 is used to report expenditures that qualify for federal matching funds under Title XXI. A full list of the category of service and definition for each line can be found at https://www.medicaid.gov/medicaid/downloads/cms-21-base-category-of-services-definition-2-14.pdf . |
a The federal reimbursement category code \""03\"" was previously a valid value used by some states on claims for their ACA adult expansion population. However, CMS retired the \""03\"" value in September 2020 and states are expected to use the \""01\"" value for this population moving forward since the enhanced match rate for these enrollees is authorized under Title XIX.
If states are populating these data elements correctly, claim records that link to beneficiaries enrolled in Title XIX Medicaid will be coded to indicate the expenditure is reportable on Form CMS-64. [2] Similarly, claim records that link to beneficiaries enrolled in Title XXI CHIP—including both M-CHIP and S-CHIP programs—will be coded to indicate the expenditure is reportable on Form CMS-21.
Other TAF data elements can be used to summarize Medicaid and CHIP spending associated with certain services or benefit categories, including the state-assigned type of service code, the federally-assigned service category code, and the benefit type code. [3] However, those data elements will not necessarily line up with the CMS-64 or CMS-21 categories that states use to report expenditures for the purposes of federal reimbursement.
In this data quality assessment, we evaluate the percentage of records in each claims file reported with a federal reimbursement category code that is consistent with the beneficiary’s enrollment in the Medicaid or CHIP programs. We also evaluate the percentage of records in each claims file reported with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid or CHIP program.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
More information about differences between expenditures in the TAF and the CMS-64 data can be found in the DQ Atlas single topic displays under the Expenditure Benchmarking topic area.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX . For more information about the FASC code, see the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
More information about differences between expenditures in the TAF and the CMS-64 data can be found in the DQ Atlas single topic displays under the Expenditure Benchmarking topic area.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX . For more information about the FASC code, see the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
We began by selecting all records from the IP, LT, OT, and RX claims files that represented Medicaid and S-CHIP fee-for-service (FFS), capitation, and supplemental payments based on the claim type code. [4] , [5] We excluded managed care encounter records from the analysis because the payments recorded on those claims do not represent state expenditures and are not recorded on the CMS-64 or CMS-21. [6] We also excluded service tracking claims from the analysis because these records are lump-sum payments that that cannot be linked to an eligibility record. [7]
To determine whether the beneficiary was enrolled in Medicaid, M-CHIP, or S-CHIP on the date of service, we linked the claims records to the matching eligibility record using the unique beneficiary identifier coded on the claim (MSIS ID). We determined Medicaid, M-CHIP, and S-CHIP enrollment by examining the CHIP code on records in the TAF Demographic and Eligibility (DE) file from the month that corresponded to the date of service on the claim. We excluded from the analysis claims records with a missing or invalid beneficiary identifier. Additionally, we excluded claims that did not link to an eligibility record, as well as those that linked to an eligibility record but did not have a CHIP code populated (indicating the individual was enrolled in Medicaid or CHIP) during the date of service on the claim. [8]
Finally, we calculated the percentage of records in each file with a federal reimbursement category code consistent with the beneficiary’s reported enrollment in Medicaid, M-CHIP, or S-CHIP. The expected coding for federal reimbursement category code based on CHIP code is shown in Table 2. Because Medicaid beneficiaries enrolled in an adult expansion eligibility group may have a different federal reimbursement category code than Medicaid beneficiaries not enrolled in such a group, we examined these two groups of Medicaid beneficiaries separately. [9] We also calculated the percentage of records with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid or CHIP program. The expected coding for CMS-64 and CMS-21 category of service code based on CHIP code is shown in Table 3.
Table 2. Expected coding for federal reimbursement category code
| Enrollment type | CHIP code on eligibility record (CHIP_CD) | Latest eligibility group on eligibility record (ELGBLTY_GRP_CD_LTST) | Expected federal reimbursement category code on claim record |
|---|---|---|---|
| Medicaid – Adult expansion group | 1 (Medicaid) | 72 (Adult Group - Individuals at or below 133% FPL Age 19 through 64 - newly eligible for all states) 73 (Adult Group - Individuals at or below 133% FPL Age 19 through 64- not newly eligible for non 1905z(3) states), 74 (Adult Group - Individuals at or below 133% FPL Age 19 through 64 – not newly eligible parent/ caretaker relative(s) in 1905z(3) states), or 75 (Adult Group - Individuals at or below 133% FPL Age 19 through 64- not newly eligible non-parent/ caretaker-relative(s) in 1905z(3) states) | 01 (Federal funding under Title XIX) or 03 (Federal funding under ACA) a |
| Medicaid – All other | 1 (Medicaid) | Any valid code or null/missing, except: 72, 73, 74, 75, or 61-68 (CHIP eligibility groups) | 01 (Federal funding under Title XIX) |
| M-CHIP | 2 (M-CHIP) | N/A | 02 (Federal funding under Title XXI) |
| S-CHIP | 3 (S-CHIP) | N/A | 02 (Federal funding under Title XXI) |
a In this assessment, we consider both \""01\"" and \""03\"" to be consistent values for Medicaid-enrolled beneficiaries in the adult expansion eligibility groups. The value \""03\"" (Federal funding under ACA) was retired as a valid value on September 30, 2020 and while the value is still present in the data before that date, we would expect to see relatively few claims with this value after that date.
ACA = Affordable Care Act; FPL = Federal Poverty Level; N/A = not available.
Table 3. Expected coding for CMS-64 and CMS-21 category of service codes
| CHIP code on eligibility record (CHIP_CD) | Expected CMS-64 category of service code on claims record | Expected CMS-21 category of service code on claims record |
|---|---|---|
| 1 (Medicaid) |
|
|
| 2 (M-CHIP) | Field should be blank |
|
| 3 (S-CHIP) | Field should be blank |
|
Data quality assessment criteria
We categorized each state as having low, medium, or high data quality concern for the federal reimbursement category code, depending on the percentage of records with a federal reimbursement category code consistent with the beneficiary’s enrollment in the Medicaid, M-CHIP, or S-CHIP programs, as shown in Table 4.
Table 4. Criteria for DQ assessment of federal reimbursement category code
| Percentage of claim lines with a federal reimbursement category code consistent with program enrollment | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
We categorized each state as having low, medium, or high data quality concern for the category of service code, depending on the percentage of records with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid, M-CHIP, or S-CHIP programs, as shown in Table 5. As shown in Table 3, only the CMS-64 or CMS-21 category of service code is expected to be populated on a single claim record. When both fields were populated on the same claim (even if one or both values were invalid), that record was inconsistent with enrollment in the Medicaid, M-CHIP, or S-CHIP programs for the purposes of this analysis.
Table 5. Criteria for DQ assessment of category of service codes
| Percentage of claim lines with a CMS-64 category of service code or a CMS-21 category of service code consistent with program enrollment | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
FFS claims representing Medicaid or M-CHIP services have a claim type code (CLM_TYPE_CD) value of 1, capitation payments have a claim type code value of 2, and supplemental payments have a claim type code value of 5. FFS claims representing S-CHIP benefits have a claim type code value of A, capitation payments have a claim type code value of B, and supplemental payments have a claim type code value of E.
Payment data on managed care encounter records are redacted from the TAF RIFs.
Managed care encounter records representing Medicaid or M-CHIP services have a claim type code value of 3 and service tracking claims have a claim type code value of 4. Managed care encounter records representing S-CHIP benefits have a claim type code value of C, and service tracking claims have a claim type code value of D.
Once an individual is determined eligible for Medicaid, coverage is effective either on the date of application or the first day of the month of application. Benefits also may be covered retroactively for up to three months prior to the month of application if the individual would have been eligible during that period had he or she applied. States are required to update beneficiaries’ eligibility records in T-MSIS to reflect the retroactive coverage period. However, if a state did not update the eligibility record of a beneficiary to reflect the retroactive coverage period, enrollment information like CHIP code may be missing during the date of service on a paid claim.
We determined the eligibility group by using ELGBLTY_GRP_CD_LTST, the most recent eligibility group reported for the beneficiary in the calendar year.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
FFS claims representing Medicaid or M-CHIP services have a claim type code (CLM_TYPE_CD) value of 1, capitation payments have a claim type code value of 2, and supplemental payments have a claim type code value of 5. FFS claims representing S-CHIP benefits have a claim type code value of A, capitation payments have a claim type code value of B, and supplemental payments have a claim type code value of E.
Payment data on managed care encounter records are redacted from the TAF RIFs.
Managed care encounter records representing Medicaid or M-CHIP services have a claim type code value of 3 and service tracking claims have a claim type code value of 4. Managed care encounter records representing S-CHIP benefits have a claim type code value of C, and service tracking claims have a claim type code value of D.
Once an individual is determined eligible for Medicaid, coverage is effective either on the date of application or the first day of the month of application. Benefits also may be covered retroactively for up to three months prior to the month of application if the individual would have been eligible during that period had he or she applied. States are required to update beneficiaries\u2019 eligibility records in T-MSIS to reflect the retroactive coverage period. However, if a state did not update the eligibility record of a beneficiary to reflect the retroactive coverage period, enrollment information like CHIP code may be missing during the date of service on a paid claim.
We determined the eligibility group by using ELGBLTY_GRP_CD_LTST, the most recent eligibility group reported for the beneficiary in the calendar year.
States report quarterly Medicaid expenditures on Form CMS-64 and quarterly Children's Health Insurance Program (CHIP) expenditures on Form CMS-21 to claim federal matching funds. In the T-MSIS Analytic Files (TAF), the CMS-64 category of service code data element on paid claim records should reflect the category where expenditures for the service are reported on Form CMS-64. Similarly, the CMS-21 category of service code on a paid claim record should reflect the category where expenditures for the service are reported on Form CMS-21. This analysis examines the extent to which records in the RX file are reported with a CMS-64 or CMS-21 category of service code value consistent with the beneficiary's enrollment in the Medicaid or CHIP program.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5221"", ""relatedTopics"": [{""measureId"": 111, ""measureName"": ""Category of Service Code - IP"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 0}, {""measureId"": 112, ""measureName"": ""Category of Service Code - LT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 1}, {""measureId"": 113, ""measureName"": ""Category of Service Code - OT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 2}, {""measureId"": 115, ""measureName"": ""Federal Reimbursement Category Code - IP"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 4}, {""measureId"": 116, ""measureName"": ""Federal Reimbursement Category Code - LT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 5}, {""measureId"": 117, ""measureName"": ""Federal Reimbursement Category Code - OT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 6}, {""measureId"": 118, ""measureName"": ""Federal Reimbursement Category Code - RX"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 7}]}" 115,"{""measureId"": 115, ""measureName"": ""Federal Reimbursement Category Code - IP"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Fed-Reimb-Cat-Cd-IP.pdf"", ""background"": {""content"": ""Medicaid and the Children’s Health Insurance Program (CHIP) are joint state-federal medical assistance programs in which the state administers the program and pays providers or health plans directly for covered services, and the federal government reimburses a set percentage of the state’s expenditures (referred to as the \""federal match\""). Because the federal government applies different federal matching rates to Medicaid and CHIP beneficiaries, states report Medicaid and CHIP expenditure data separately.
To claim federal matching funds for their medical assistance programs, states report aggregate expenditure data for Medicaid beneficiaries covered under Title XIX of the Social Security Act to the Centers for Medicare & Medicaid Services (CMS) on Form CMS-64, Quarterly Medicaid Statement of Expenditures. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act on Form CMS-21, Quarterly CHIP Statement of Expenditures . States may cover children using CHIP funds by expanding their Medicaid programs (called \""Medicaid expansion CHIP,\"" or M-CHIP), creating a program separate from their existing Medicaid programs (called \""separate CHIP,\"" or S-CHIP), or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries, which qualify for the higher CHIP federal match rate, are reported on Form CMS-21 and not Form CMS-64.
There are three data elements in the T-MSIS Analytic Files (TAF) inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) files that contain information about how states report Medicaid and CHIP expenditures to claim federal matching funds: the federal reimbursement category code, the CMS-64 category of service code, and the CMS-21 category of service code (Table 1). [1] The federal reimbursement category code indicates whether the expenditure was reimbursed under Title XIX, Title XXI, the Affordable Care Act (ACA), or other legislation. As noted earlier, if the expenditure was covered by Title XIX of the Social Security Act, it should be reported on Form CMS-64; if the expenditure was covered by Title XXI of the Social Security Act it should be reported on Form CMS-21. The CMS-64 category of service code indicates the specific category of service on Form CMS-64 into which the state will report the Medicaid expenditures associated with a paid claim. The CMS-21 category of service code indicates the specific category of service on Form CMS-21 into which the state will report the CHIP expenditures associated with a paid claim.
Table 1. TAF variables that classify expenditures into federal reimbursement categories
| Variable | TAF data element name (TAF RIF name in parentheses if different) | IP, LT, OT, and RX records expected to have non-missing values | Description |
|---|---|---|---|
| Federal reimbursement category code | CMS_64_FED_REIMBRSMT_CTGRY_CD (CMS_64_FED_CTGRY_CD) | All records | This code indicates whether the claim was matched with federal funding under Title XIX (value of ‘01’), Title XXI (‘02’), the Affordable Care Act (‘03’), a or federal funding under other legislation (‘04’). |
| CMS-64 category of service code | XIX_SRVC_CTGRY_CD | Only records representing expenditures that the state will report on the CMS-64, which should link to beneficiaries enrolled in non-CHIP Medicaid | A four-digit code (from 001A to 0050) indicating the line on the Form CMS-64 on which the state reported the expenditures associated with the claim. The Form CMS-64 is used to report expenditures that qualify for federal matching funds under Title XXI. A full list of the category of service and definition for each line can be found at https://www.medicaid.gov/medicaid/downloads/cms-649-base-category-of-services-definition-2-14.pdf . |
| CMS-21 category of service code | XIX_SRVC_CTGRY_CD | Only records representing expenditures that the state will report on the CMS-21, which should link to beneficiaries enrolled in Medicaid-expansion CHIP or Separate CHIP | A three-digit code (from 01A to 35B) indicating the category of service (or line) on the Form CMS-21 on which the state reported the expenditures associated with the claim. The Form CMS-21 is used to report expenditures that qualify for federal matching funds under Title XXI. A full list of the category of service and definition for each line can be found at https://www.medicaid.gov/medicaid/downloads/cms-21-base-category-of-services-definition-2-14.pdf . |
a The federal reimbursement category code \""03\"" was previously a valid value used by some states on claims for their ACA adult expansion population. However, CMS retired the \""03\"" value in September 2020 and states are expected to use the \""01\"" value for this population moving forward since the enhanced match rate for these enrollees is authorized under Title XIX.
If states are populating these data elements correctly, claim records that link to beneficiaries enrolled in Title XIX Medicaid will be coded to indicate the expenditure is reportable on Form CMS-64. [2] Similarly, claim records that link to beneficiaries enrolled in Title XXI CHIP—including both M-CHIP and S-CHIP programs—will be coded to indicate the expenditure is reportable on Form CMS-21.
Other TAF data elements can be used to summarize Medicaid and CHIP spending associated with certain services or benefit categories, including the state-assigned type of service code, the federally-assigned service category code, and the benefit type code. [3] However, those data elements will not necessarily line up with the CMS-64 or CMS-21 categories that states use to report expenditures for the purposes of federal reimbursement.
In this data quality assessment, we evaluate the percentage of records in each claims file reported with a federal reimbursement category code that is consistent with the beneficiary’s enrollment in the Medicaid or CHIP programs. We also evaluate the percentage of records in each claims file reported with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid or CHIP program.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
More information about differences between expenditures in the TAF and the CMS-64 data can be found in the DQ Atlas single topic displays under the Expenditure Benchmarking topic area.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX . For more information about the FASC code, see the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
More information about differences between expenditures in the TAF and the CMS-64 data can be found in the DQ Atlas single topic displays under the Expenditure Benchmarking topic area.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX . For more information about the FASC code, see the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
We began by selecting all records from the IP, LT, OT, and RX claims files that represented Medicaid and S-CHIP fee-for-service (FFS), capitation, and supplemental payments based on the claim type code. [4] , [5] We excluded managed care encounter records from the analysis because the payments recorded on those claims do not represent state expenditures and are not recorded on the CMS-64 or CMS-21. [6] We also excluded service tracking claims from the analysis because these records are lump-sum payments that that cannot be linked to an eligibility record. [7]
To determine whether the beneficiary was enrolled in Medicaid, M-CHIP, or S-CHIP on the date of service, we linked the claims records to the matching eligibility record using the unique beneficiary identifier coded on the claim (MSIS ID). We determined Medicaid, M-CHIP, and S-CHIP enrollment by examining the CHIP code on records in the TAF Demographic and Eligibility (DE) file from the month that corresponded to the date of service on the claim. We excluded from the analysis claims records with a missing or invalid beneficiary identifier. Additionally, we excluded claims that did not link to an eligibility record, as well as those that linked to an eligibility record but did not have a CHIP code populated (indicating the individual was enrolled in Medicaid or CHIP) during the date of service on the claim. [8]
Finally, we calculated the percentage of records in each file with a federal reimbursement category code consistent with the beneficiary’s reported enrollment in Medicaid, M-CHIP, or S-CHIP. The expected coding for federal reimbursement category code based on CHIP code is shown in Table 2. Because Medicaid beneficiaries enrolled in an adult expansion eligibility group may have a different federal reimbursement category code than Medicaid beneficiaries not enrolled in such a group, we examined these two groups of Medicaid beneficiaries separately. [9] We also calculated the percentage of records with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid or CHIP program. The expected coding for CMS-64 and CMS-21 category of service code based on CHIP code is shown in Table 3.
Table 2. Expected coding for federal reimbursement category code
| Enrollment type | CHIP code on eligibility record (CHIP_CD) | Latest eligibility group on eligibility record (ELGBLTY_GRP_CD_LTST) | Expected federal reimbursement category code on claim record |
|---|---|---|---|
| Medicaid – Adult expansion group | 1 (Medicaid) | 72 (Adult Group - Individuals at or below 133% FPL Age 19 through 64 - newly eligible for all states) 73 (Adult Group - Individuals at or below 133% FPL Age 19 through 64- not newly eligible for non 1905z(3) states), 74 (Adult Group - Individuals at or below 133% FPL Age 19 through 64 – not newly eligible parent/ caretaker relative(s) in 1905z(3) states), or 75 (Adult Group - Individuals at or below 133% FPL Age 19 through 64- not newly eligible non-parent/ caretaker-relative(s) in 1905z(3) states) | 01 (Federal funding under Title XIX) or 03 (Federal funding under ACA) a |
| Medicaid – All other | 1 (Medicaid) | Any valid code or null/missing, except: 72, 73, 74, 75, or 61-68 (CHIP eligibility groups) | 01 (Federal funding under Title XIX) |
| M-CHIP | 2 (M-CHIP) | N/A | 02 (Federal funding under Title XXI) |
| S-CHIP | 3 (S-CHIP) | N/A | 02 (Federal funding under Title XXI) |
a In this assessment, we consider both \""01\"" and \""03\"" to be consistent values for Medicaid-enrolled beneficiaries in the adult expansion eligibility groups. The value \""03\"" (Federal funding under ACA) was retired as a valid value on September 30, 2020 and while the value is still present in the data before that date, we would expect to see relatively few claims with this value after that date.
ACA = Affordable Care Act; FPL = Federal Poverty Level; N/A = not available.
Table 3. Expected coding for CMS-64 and CMS-21 category of service codes
| CHIP code on eligibility record (CHIP_CD) | Expected CMS-64 category of service code on claims record | Expected CMS-21 category of service code on claims record |
|---|---|---|
| 1 (Medicaid) |
|
|
| 2 (M-CHIP) | Field should be blank |
|
| 3 (S-CHIP) | Field should be blank |
|
Data quality assessment criteria
We categorized each state as having low, medium, or high data quality concern for the federal reimbursement category code, depending on the percentage of records with a federal reimbursement category code consistent with the beneficiary’s enrollment in the Medicaid, M-CHIP, or S-CHIP programs, as shown in Table 4.
Table 4. Criteria for DQ assessment of federal reimbursement category code
| Percentage of claim lines with a federal reimbursement category code consistent with program enrollment | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
We categorized each state as having low, medium, or high data quality concern for the category of service code, depending on the percentage of records with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid, M-CHIP, or S-CHIP programs, as shown in Table 5. As shown in Table 3, only the CMS-64 or CMS-21 category of service code is expected to be populated on a single claim record. When both fields were populated on the same claim (even if one or both values were invalid), that record was inconsistent with enrollment in the Medicaid, M-CHIP, or S-CHIP programs for the purposes of this analysis.
Table 5. Criteria for DQ assessment of category of service codes
| Percentage of claim lines with a CMS-64 category of service code or a CMS-21 category of service code consistent with program enrollment | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
FFS claims representing Medicaid or M-CHIP services have a claim type code (CLM_TYPE_CD) value of 1, capitation payments have a claim type code value of 2, and supplemental payments have a claim type code value of 5. FFS claims representing S-CHIP benefits have a claim type code value of A, capitation payments have a claim type code value of B, and supplemental payments have a claim type code value of E.
Payment data on managed care encounter records are redacted from the TAF RIFs.
Managed care encounter records representing Medicaid or M-CHIP services have a claim type code value of 3 and service tracking claims have a claim type code value of 4. Managed care encounter records representing S-CHIP benefits have a claim type code value of C, and service tracking claims have a claim type code value of D.
Once an individual is determined eligible for Medicaid, coverage is effective either on the date of application or the first day of the month of application. Benefits also may be covered retroactively for up to three months prior to the month of application if the individual would have been eligible during that period had he or she applied. States are required to update beneficiaries’ eligibility records in T-MSIS to reflect the retroactive coverage period. However, if a state did not update the eligibility record of a beneficiary to reflect the retroactive coverage period, enrollment information like CHIP code may be missing during the date of service on a paid claim.
We determined the eligibility group by using ELGBLTY_GRP_CD_LTST, the most recent eligibility group reported for the beneficiary in the calendar year.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
FFS claims representing Medicaid or M-CHIP services have a claim type code (CLM_TYPE_CD) value of 1, capitation payments have a claim type code value of 2, and supplemental payments have a claim type code value of 5. FFS claims representing S-CHIP benefits have a claim type code value of A, capitation payments have a claim type code value of B, and supplemental payments have a claim type code value of E.
Payment data on managed care encounter records are redacted from the TAF RIFs.
Managed care encounter records representing Medicaid or M-CHIP services have a claim type code value of 3 and service tracking claims have a claim type code value of 4. Managed care encounter records representing S-CHIP benefits have a claim type code value of C, and service tracking claims have a claim type code value of D.
Once an individual is determined eligible for Medicaid, coverage is effective either on the date of application or the first day of the month of application. Benefits also may be covered retroactively for up to three months prior to the month of application if the individual would have been eligible during that period had he or she applied. States are required to update beneficiaries\u2019 eligibility records in T-MSIS to reflect the retroactive coverage period. However, if a state did not update the eligibility record of a beneficiary to reflect the retroactive coverage period, enrollment information like CHIP code may be missing during the date of service on a paid claim.
We determined the eligibility group by using ELGBLTY_GRP_CD_LTST, the most recent eligibility group reported for the beneficiary in the calendar year.
The federal reimbursement category code indicates whether a claim represents state expenditures that qualify for federal matching funds under Title XIX (Medicaid), Title XXI (CHIP), or other legislation. This analysis examines the extent to which records in the IP file are reported with a federal reimbursement category code consistent with the beneficiary's enrollment in the Medicaid or CHIP program.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5221"", ""relatedTopics"": [{""measureId"": 111, ""measureName"": ""Category of Service Code - IP"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 0}, {""measureId"": 112, ""measureName"": ""Category of Service Code - LT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 1}, {""measureId"": 113, ""measureName"": ""Category of Service Code - OT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 2}, {""measureId"": 114, ""measureName"": ""Category of Service Code - RX"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 3}, {""measureId"": 116, ""measureName"": ""Federal Reimbursement Category Code - LT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 5}, {""measureId"": 117, ""measureName"": ""Federal Reimbursement Category Code - OT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 6}, {""measureId"": 118, ""measureName"": ""Federal Reimbursement Category Code - RX"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 7}]}" 116,"{""measureId"": 116, ""measureName"": ""Federal Reimbursement Category Code - LT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Fed-Reimb-Cat-Cd-LT.pdf"", ""background"": {""content"": ""Medicaid and the Children’s Health Insurance Program (CHIP) are joint state-federal medical assistance programs in which the state administers the program and pays providers or health plans directly for covered services, and the federal government reimburses a set percentage of the state’s expenditures (referred to as the \""federal match\""). Because the federal government applies different federal matching rates to Medicaid and CHIP beneficiaries, states report Medicaid and CHIP expenditure data separately.
To claim federal matching funds for their medical assistance programs, states report aggregate expenditure data for Medicaid beneficiaries covered under Title XIX of the Social Security Act to the Centers for Medicare & Medicaid Services (CMS) on Form CMS-64, Quarterly Medicaid Statement of Expenditures. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act on Form CMS-21, Quarterly CHIP Statement of Expenditures . States may cover children using CHIP funds by expanding their Medicaid programs (called \""Medicaid expansion CHIP,\"" or M-CHIP), creating a program separate from their existing Medicaid programs (called \""separate CHIP,\"" or S-CHIP), or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries, which qualify for the higher CHIP federal match rate, are reported on Form CMS-21 and not Form CMS-64.
There are three data elements in the T-MSIS Analytic Files (TAF) inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) files that contain information about how states report Medicaid and CHIP expenditures to claim federal matching funds: the federal reimbursement category code, the CMS-64 category of service code, and the CMS-21 category of service code (Table 1). [1] The federal reimbursement category code indicates whether the expenditure was reimbursed under Title XIX, Title XXI, the Affordable Care Act (ACA), or other legislation. As noted earlier, if the expenditure was covered by Title XIX of the Social Security Act, it should be reported on Form CMS-64; if the expenditure was covered by Title XXI of the Social Security Act it should be reported on Form CMS-21. The CMS-64 category of service code indicates the specific category of service on Form CMS-64 into which the state will report the Medicaid expenditures associated with a paid claim. The CMS-21 category of service code indicates the specific category of service on Form CMS-21 into which the state will report the CHIP expenditures associated with a paid claim.
Table 1. TAF variables that classify expenditures into federal reimbursement categories
| Variable | TAF data element name (TAF RIF name in parentheses if different) | IP, LT, OT, and RX records expected to have non-missing values | Description |
|---|---|---|---|
| Federal reimbursement category code | CMS_64_FED_REIMBRSMT_CTGRY_CD (CMS_64_FED_CTGRY_CD) | All records | This code indicates whether the claim was matched with federal funding under Title XIX (value of ‘01’), Title XXI (‘02’), the Affordable Care Act (‘03’), a or federal funding under other legislation (‘04’). |
| CMS-64 category of service code | XIX_SRVC_CTGRY_CD | Only records representing expenditures that the state will report on the CMS-64, which should link to beneficiaries enrolled in non-CHIP Medicaid | A four-digit code (from 001A to 0050) indicating the line on the Form CMS-64 on which the state reported the expenditures associated with the claim. The Form CMS-64 is used to report expenditures that qualify for federal matching funds under Title XXI. A full list of the category of service and definition for each line can be found at https://www.medicaid.gov/medicaid/downloads/cms-649-base-category-of-services-definition-2-14.pdf . |
| CMS-21 category of service code | XIX_SRVC_CTGRY_CD | Only records representing expenditures that the state will report on the CMS-21, which should link to beneficiaries enrolled in Medicaid-expansion CHIP or Separate CHIP | A three-digit code (from 01A to 35B) indicating the category of service (or line) on the Form CMS-21 on which the state reported the expenditures associated with the claim. The Form CMS-21 is used to report expenditures that qualify for federal matching funds under Title XXI. A full list of the category of service and definition for each line can be found at https://www.medicaid.gov/medicaid/downloads/cms-21-base-category-of-services-definition-2-14.pdf . |
a The federal reimbursement category code \""03\"" was previously a valid value used by some states on claims for their ACA adult expansion population. However, CMS retired the \""03\"" value in September 2020 and states are expected to use the \""01\"" value for this population moving forward since the enhanced match rate for these enrollees is authorized under Title XIX.
If states are populating these data elements correctly, claim records that link to beneficiaries enrolled in Title XIX Medicaid will be coded to indicate the expenditure is reportable on Form CMS-64. [2] Similarly, claim records that link to beneficiaries enrolled in Title XXI CHIP—including both M-CHIP and S-CHIP programs—will be coded to indicate the expenditure is reportable on Form CMS-21.
Other TAF data elements can be used to summarize Medicaid and CHIP spending associated with certain services or benefit categories, including the state-assigned type of service code, the federally-assigned service category code, and the benefit type code. [3] However, those data elements will not necessarily line up with the CMS-64 or CMS-21 categories that states use to report expenditures for the purposes of federal reimbursement.
In this data quality assessment, we evaluate the percentage of records in each claims file reported with a federal reimbursement category code that is consistent with the beneficiary’s enrollment in the Medicaid or CHIP programs. We also evaluate the percentage of records in each claims file reported with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid or CHIP program.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
More information about differences between expenditures in the TAF and the CMS-64 data can be found in the DQ Atlas single topic displays under the Expenditure Benchmarking topic area.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX . For more information about the FASC code, see the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
More information about differences between expenditures in the TAF and the CMS-64 data can be found in the DQ Atlas single topic displays under the Expenditure Benchmarking topic area.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX . For more information about the FASC code, see the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
We began by selecting all records from the IP, LT, OT, and RX claims files that represented Medicaid and S-CHIP fee-for-service (FFS), capitation, and supplemental payments based on the claim type code. [4] , [5] We excluded managed care encounter records from the analysis because the payments recorded on those claims do not represent state expenditures and are not recorded on the CMS-64 or CMS-21. [6] We also excluded service tracking claims from the analysis because these records are lump-sum payments that that cannot be linked to an eligibility record. [7]
To determine whether the beneficiary was enrolled in Medicaid, M-CHIP, or S-CHIP on the date of service, we linked the claims records to the matching eligibility record using the unique beneficiary identifier coded on the claim (MSIS ID). We determined Medicaid, M-CHIP, and S-CHIP enrollment by examining the CHIP code on records in the TAF Demographic and Eligibility (DE) file from the month that corresponded to the date of service on the claim. We excluded from the analysis claims records with a missing or invalid beneficiary identifier. Additionally, we excluded claims that did not link to an eligibility record, as well as those that linked to an eligibility record but did not have a CHIP code populated (indicating the individual was enrolled in Medicaid or CHIP) during the date of service on the claim. [8]
Finally, we calculated the percentage of records in each file with a federal reimbursement category code consistent with the beneficiary’s reported enrollment in Medicaid, M-CHIP, or S-CHIP. The expected coding for federal reimbursement category code based on CHIP code is shown in Table 2. Because Medicaid beneficiaries enrolled in an adult expansion eligibility group may have a different federal reimbursement category code than Medicaid beneficiaries not enrolled in such a group, we examined these two groups of Medicaid beneficiaries separately. [9] We also calculated the percentage of records with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid or CHIP program. The expected coding for CMS-64 and CMS-21 category of service code based on CHIP code is shown in Table 3.
Table 2. Expected coding for federal reimbursement category code
| Enrollment type | CHIP code on eligibility record (CHIP_CD) | Latest eligibility group on eligibility record (ELGBLTY_GRP_CD_LTST) | Expected federal reimbursement category code on claim record |
|---|---|---|---|
| Medicaid – Adult expansion group | 1 (Medicaid) | 72 (Adult Group - Individuals at or below 133% FPL Age 19 through 64 - newly eligible for all states) 73 (Adult Group - Individuals at or below 133% FPL Age 19 through 64- not newly eligible for non 1905z(3) states), 74 (Adult Group - Individuals at or below 133% FPL Age 19 through 64 – not newly eligible parent/ caretaker relative(s) in 1905z(3) states), or 75 (Adult Group - Individuals at or below 133% FPL Age 19 through 64- not newly eligible non-parent/ caretaker-relative(s) in 1905z(3) states) | 01 (Federal funding under Title XIX) or 03 (Federal funding under ACA) a |
| Medicaid – All other | 1 (Medicaid) | Any valid code or null/missing, except: 72, 73, 74, 75, or 61-68 (CHIP eligibility groups) | 01 (Federal funding under Title XIX) |
| M-CHIP | 2 (M-CHIP) | N/A | 02 (Federal funding under Title XXI) |
| S-CHIP | 3 (S-CHIP) | N/A | 02 (Federal funding under Title XXI) |
a In this assessment, we consider both \""01\"" and \""03\"" to be consistent values for Medicaid-enrolled beneficiaries in the adult expansion eligibility groups. The value \""03\"" (Federal funding under ACA) was retired as a valid value on September 30, 2020 and while the value is still present in the data before that date, we would expect to see relatively few claims with this value after that date.
ACA = Affordable Care Act; FPL = Federal Poverty Level; N/A = not available.
Table 3. Expected coding for CMS-64 and CMS-21 category of service codes
| CHIP code on eligibility record (CHIP_CD) | Expected CMS-64 category of service code on claims record | Expected CMS-21 category of service code on claims record |
|---|---|---|
| 1 (Medicaid) |
|
|
| 2 (M-CHIP) | Field should be blank |
|
| 3 (S-CHIP) | Field should be blank |
|
Data quality assessment criteria
We categorized each state as having low, medium, or high data quality concern for the federal reimbursement category code, depending on the percentage of records with a federal reimbursement category code consistent with the beneficiary’s enrollment in the Medicaid, M-CHIP, or S-CHIP programs, as shown in Table 4.
Table 4. Criteria for DQ assessment of federal reimbursement category code
| Percentage of claim lines with a federal reimbursement category code consistent with program enrollment | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
We categorized each state as having low, medium, or high data quality concern for the category of service code, depending on the percentage of records with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid, M-CHIP, or S-CHIP programs, as shown in Table 5. As shown in Table 3, only the CMS-64 or CMS-21 category of service code is expected to be populated on a single claim record. When both fields were populated on the same claim (even if one or both values were invalid), that record was inconsistent with enrollment in the Medicaid, M-CHIP, or S-CHIP programs for the purposes of this analysis.
Table 5. Criteria for DQ assessment of category of service codes
| Percentage of claim lines with a CMS-64 category of service code or a CMS-21 category of service code consistent with program enrollment | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
FFS claims representing Medicaid or M-CHIP services have a claim type code (CLM_TYPE_CD) value of 1, capitation payments have a claim type code value of 2, and supplemental payments have a claim type code value of 5. FFS claims representing S-CHIP benefits have a claim type code value of A, capitation payments have a claim type code value of B, and supplemental payments have a claim type code value of E.
Payment data on managed care encounter records are redacted from the TAF RIFs.
Managed care encounter records representing Medicaid or M-CHIP services have a claim type code value of 3 and service tracking claims have a claim type code value of 4. Managed care encounter records representing S-CHIP benefits have a claim type code value of C, and service tracking claims have a claim type code value of D.
Once an individual is determined eligible for Medicaid, coverage is effective either on the date of application or the first day of the month of application. Benefits also may be covered retroactively for up to three months prior to the month of application if the individual would have been eligible during that period had he or she applied. States are required to update beneficiaries’ eligibility records in T-MSIS to reflect the retroactive coverage period. However, if a state did not update the eligibility record of a beneficiary to reflect the retroactive coverage period, enrollment information like CHIP code may be missing during the date of service on a paid claim.
We determined the eligibility group by using ELGBLTY_GRP_CD_LTST, the most recent eligibility group reported for the beneficiary in the calendar year.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
FFS claims representing Medicaid or M-CHIP services have a claim type code (CLM_TYPE_CD) value of 1, capitation payments have a claim type code value of 2, and supplemental payments have a claim type code value of 5. FFS claims representing S-CHIP benefits have a claim type code value of A, capitation payments have a claim type code value of B, and supplemental payments have a claim type code value of E.
Payment data on managed care encounter records are redacted from the TAF RIFs.
Managed care encounter records representing Medicaid or M-CHIP services have a claim type code value of 3 and service tracking claims have a claim type code value of 4. Managed care encounter records representing S-CHIP benefits have a claim type code value of C, and service tracking claims have a claim type code value of D.
Once an individual is determined eligible for Medicaid, coverage is effective either on the date of application or the first day of the month of application. Benefits also may be covered retroactively for up to three months prior to the month of application if the individual would have been eligible during that period had he or she applied. States are required to update beneficiaries\u2019 eligibility records in T-MSIS to reflect the retroactive coverage period. However, if a state did not update the eligibility record of a beneficiary to reflect the retroactive coverage period, enrollment information like CHIP code may be missing during the date of service on a paid claim.
We determined the eligibility group by using ELGBLTY_GRP_CD_LTST, the most recent eligibility group reported for the beneficiary in the calendar year.
The federal reimbursement category code indicates whether a claim represents state expenditures that qualify for federal matching funds under Title XIX (Medicaid), Title XXI (CHIP), or other legislation. This analysis examines the extent to which records in the LT file are reported with a federal reimbursement category code consistent with the beneficiary's enrollment in the Medicaid or CHIP program.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5221"", ""relatedTopics"": [{""measureId"": 111, ""measureName"": ""Category of Service Code - IP"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 0}, {""measureId"": 112, ""measureName"": ""Category of Service Code - LT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 1}, {""measureId"": 113, ""measureName"": ""Category of Service Code - OT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 2}, {""measureId"": 114, ""measureName"": ""Category of Service Code - RX"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 3}, {""measureId"": 115, ""measureName"": ""Federal Reimbursement Category Code - IP"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 4}, {""measureId"": 117, ""measureName"": ""Federal Reimbursement Category Code - OT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 6}, {""measureId"": 118, ""measureName"": ""Federal Reimbursement Category Code - RX"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 7}]}" 117,"{""measureId"": 117, ""measureName"": ""Federal Reimbursement Category Code - OT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Fed-Srvc-Reimb-Cd-OT.pdf"", ""background"": {""content"": ""Medicaid and the Children’s Health Insurance Program (CHIP) are joint state-federal medical assistance programs in which the state administers the program and pays providers or health plans directly for covered services, and the federal government reimburses a set percentage of the state’s expenditures (referred to as the \""federal match\""). Because the federal government applies different federal matching rates to Medicaid and CHIP beneficiaries, states report Medicaid and CHIP expenditure data separately.
To claim federal matching funds for their medical assistance programs, states report aggregate expenditure data for Medicaid beneficiaries covered under Title XIX of the Social Security Act to the Centers for Medicare & Medicaid Services (CMS) on Form CMS-64, Quarterly Medicaid Statement of Expenditures. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act on Form CMS-21, Quarterly CHIP Statement of Expenditures . States may cover children using CHIP funds by expanding their Medicaid programs (called \""Medicaid expansion CHIP,\"" or M-CHIP), creating a program separate from their existing Medicaid programs (called \""separate CHIP,\"" or S-CHIP), or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries, which qualify for the higher CHIP federal match rate, are reported on Form CMS-21 and not Form CMS-64.
There are three data elements in the T-MSIS Analytic Files (TAF) inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) files that contain information about how states report Medicaid and CHIP expenditures to claim federal matching funds: the federal reimbursement category code, the CMS-64 category of service code, and the CMS-21 category of service code (Table 1). [1] The federal reimbursement category code indicates whether the expenditure was reimbursed under Title XIX, Title XXI, the Affordable Care Act (ACA), or other legislation. As noted earlier, if the expenditure was covered by Title XIX of the Social Security Act, it should be reported on Form CMS-64; if the expenditure was covered by Title XXI of the Social Security Act it should be reported on Form CMS-21. The CMS-64 category of service code indicates the specific category of service on Form CMS-64 into which the state will report the Medicaid expenditures associated with a paid claim. The CMS-21 category of service code indicates the specific category of service on Form CMS-21 into which the state will report the CHIP expenditures associated with a paid claim.
Table 1. TAF variables that classify expenditures into federal reimbursement categories
| Variable | TAF data element name (TAF RIF name in parentheses if different) | IP, LT, OT, and RX records expected to have non-missing values | Description |
|---|---|---|---|
| Federal reimbursement category code | CMS_64_FED_REIMBRSMT_CTGRY_CD (CMS_64_FED_CTGRY_CD) | All records | This code indicates whether the claim was matched with federal funding under Title XIX (value of ‘01’), Title XXI (‘02’), the Affordable Care Act (‘03’), a or federal funding under other legislation (‘04’). |
| CMS-64 category of service code | XIX_SRVC_CTGRY_CD | Only records representing expenditures that the state will report on the CMS-64, which should link to beneficiaries enrolled in non-CHIP Medicaid | A four-digit code (from 001A to 0050) indicating the line on the Form CMS-64 on which the state reported the expenditures associated with the claim. The Form CMS-64 is used to report expenditures that qualify for federal matching funds under Title XXI. A full list of the category of service and definition for each line can be found at https://www.medicaid.gov/medicaid/downloads/cms-649-base-category-of-services-definition-2-14.pdf . |
| CMS-21 category of service code | XIX_SRVC_CTGRY_CD | Only records representing expenditures that the state will report on the CMS-21, which should link to beneficiaries enrolled in Medicaid-expansion CHIP or Separate CHIP | A three-digit code (from 01A to 35B) indicating the category of service (or line) on the Form CMS-21 on which the state reported the expenditures associated with the claim. The Form CMS-21 is used to report expenditures that qualify for federal matching funds under Title XXI. A full list of the category of service and definition for each line can be found at https://www.medicaid.gov/medicaid/downloads/cms-21-base-category-of-services-definition-2-14.pdf . |
a The federal reimbursement category code \""03\"" was previously a valid value used by some states on claims for their ACA adult expansion population. However, CMS retired the \""03\"" value in September 2020 and states are expected to use the \""01\"" value for this population moving forward since the enhanced match rate for these enrollees is authorized under Title XIX.
If states are populating these data elements correctly, claim records that link to beneficiaries enrolled in Title XIX Medicaid will be coded to indicate the expenditure is reportable on Form CMS-64. [2] Similarly, claim records that link to beneficiaries enrolled in Title XXI CHIP—including both M-CHIP and S-CHIP programs—will be coded to indicate the expenditure is reportable on Form CMS-21.
Other TAF data elements can be used to summarize Medicaid and CHIP spending associated with certain services or benefit categories, including the state-assigned type of service code, the federally-assigned service category code, and the benefit type code. [3] However, those data elements will not necessarily line up with the CMS-64 or CMS-21 categories that states use to report expenditures for the purposes of federal reimbursement.
In this data quality assessment, we evaluate the percentage of records in each claims file reported with a federal reimbursement category code that is consistent with the beneficiary’s enrollment in the Medicaid or CHIP programs. We also evaluate the percentage of records in each claims file reported with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid or CHIP program.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
More information about differences between expenditures in the TAF and the CMS-64 data can be found in the DQ Atlas single topic displays under the Expenditure Benchmarking topic area.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX . For more information about the FASC code, see the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
More information about differences between expenditures in the TAF and the CMS-64 data can be found in the DQ Atlas single topic displays under the Expenditure Benchmarking topic area.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX . For more information about the FASC code, see the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
We began by selecting all records from the IP, LT, OT, and RX claims files that represented Medicaid and S-CHIP fee-for-service (FFS), capitation, and supplemental payments based on the claim type code. [4] , [5] We excluded managed care encounter records from the analysis because the payments recorded on those claims do not represent state expenditures and are not recorded on the CMS-64 or CMS-21. [6] We also excluded service tracking claims from the analysis because these records are lump-sum payments that that cannot be linked to an eligibility record. [7]
To determine whether the beneficiary was enrolled in Medicaid, M-CHIP, or S-CHIP on the date of service, we linked the claims records to the matching eligibility record using the unique beneficiary identifier coded on the claim (MSIS ID). We determined Medicaid, M-CHIP, and S-CHIP enrollment by examining the CHIP code on records in the TAF Demographic and Eligibility (DE) file from the month that corresponded to the date of service on the claim. We excluded from the analysis claims records with a missing or invalid beneficiary identifier. Additionally, we excluded claims that did not link to an eligibility record, as well as those that linked to an eligibility record but did not have a CHIP code populated (indicating the individual was enrolled in Medicaid or CHIP) during the date of service on the claim. [8]
Finally, we calculated the percentage of records in each file with a federal reimbursement category code consistent with the beneficiary’s reported enrollment in Medicaid, M-CHIP, or S-CHIP. The expected coding for federal reimbursement category code based on CHIP code is shown in Table 2. Because Medicaid beneficiaries enrolled in an adult expansion eligibility group may have a different federal reimbursement category code than Medicaid beneficiaries not enrolled in such a group, we examined these two groups of Medicaid beneficiaries separately. [9] We also calculated the percentage of records with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid or CHIP program. The expected coding for CMS-64 and CMS-21 category of service code based on CHIP code is shown in Table 3.
Table 2. Expected coding for federal reimbursement category code
| Enrollment type | CHIP code on eligibility record (CHIP_CD) | Latest eligibility group on eligibility record (ELGBLTY_GRP_CD_LTST) | Expected federal reimbursement category code on claim record |
|---|---|---|---|
| Medicaid – Adult expansion group | 1 (Medicaid) | 72 (Adult Group - Individuals at or below 133% FPL Age 19 through 64 - newly eligible for all states) 73 (Adult Group - Individuals at or below 133% FPL Age 19 through 64- not newly eligible for non 1905z(3) states), 74 (Adult Group - Individuals at or below 133% FPL Age 19 through 64 – not newly eligible parent/ caretaker relative(s) in 1905z(3) states), or 75 (Adult Group - Individuals at or below 133% FPL Age 19 through 64- not newly eligible non-parent/ caretaker-relative(s) in 1905z(3) states) | 01 (Federal funding under Title XIX) or 03 (Federal funding under ACA) a |
| Medicaid – All other | 1 (Medicaid) | Any valid code or null/missing, except: 72, 73, 74, 75, or 61-68 (CHIP eligibility groups) | 01 (Federal funding under Title XIX) |
| M-CHIP | 2 (M-CHIP) | N/A | 02 (Federal funding under Title XXI) |
| S-CHIP | 3 (S-CHIP) | N/A | 02 (Federal funding under Title XXI) |
a In this assessment, we consider both \""01\"" and \""03\"" to be consistent values for Medicaid-enrolled beneficiaries in the adult expansion eligibility groups. The value \""03\"" (Federal funding under ACA) was retired as a valid value on September 30, 2020 and while the value is still present in the data before that date, we would expect to see relatively few claims with this value after that date.
ACA = Affordable Care Act; FPL = Federal Poverty Level; N/A = not available.
Table 3. Expected coding for CMS-64 and CMS-21 category of service codes
| CHIP code on eligibility record (CHIP_CD) | Expected CMS-64 category of service code on claims record | Expected CMS-21 category of service code on claims record |
|---|---|---|
| 1 (Medicaid) |
|
|
| 2 (M-CHIP) | Field should be blank |
|
| 3 (S-CHIP) | Field should be blank |
|
Data quality assessment criteria
We categorized each state as having low, medium, or high data quality concern for the federal reimbursement category code, depending on the percentage of records with a federal reimbursement category code consistent with the beneficiary’s enrollment in the Medicaid, M-CHIP, or S-CHIP programs, as shown in Table 4.
Table 4. Criteria for DQ assessment of federal reimbursement category code
| Percentage of claim lines with a federal reimbursement category code consistent with program enrollment | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
We categorized each state as having low, medium, or high data quality concern for the category of service code, depending on the percentage of records with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid, M-CHIP, or S-CHIP programs, as shown in Table 5. As shown in Table 3, only the CMS-64 or CMS-21 category of service code is expected to be populated on a single claim record. When both fields were populated on the same claim (even if one or both values were invalid), that record was inconsistent with enrollment in the Medicaid, M-CHIP, or S-CHIP programs for the purposes of this analysis.
Table 5. Criteria for DQ assessment of category of service codes
| Percentage of claim lines with a CMS-64 category of service code or a CMS-21 category of service code consistent with program enrollment | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
FFS claims representing Medicaid or M-CHIP services have a claim type code (CLM_TYPE_CD) value of 1, capitation payments have a claim type code value of 2, and supplemental payments have a claim type code value of 5. FFS claims representing S-CHIP benefits have a claim type code value of A, capitation payments have a claim type code value of B, and supplemental payments have a claim type code value of E.
Payment data on managed care encounter records are redacted from the TAF RIFs.
Managed care encounter records representing Medicaid or M-CHIP services have a claim type code value of 3 and service tracking claims have a claim type code value of 4. Managed care encounter records representing S-CHIP benefits have a claim type code value of C, and service tracking claims have a claim type code value of D.
Once an individual is determined eligible for Medicaid, coverage is effective either on the date of application or the first day of the month of application. Benefits also may be covered retroactively for up to three months prior to the month of application if the individual would have been eligible during that period had he or she applied. States are required to update beneficiaries’ eligibility records in T-MSIS to reflect the retroactive coverage period. However, if a state did not update the eligibility record of a beneficiary to reflect the retroactive coverage period, enrollment information like CHIP code may be missing during the date of service on a paid claim.
We determined the eligibility group by using ELGBLTY_GRP_CD_LTST, the most recent eligibility group reported for the beneficiary in the calendar year.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
FFS claims representing Medicaid or M-CHIP services have a claim type code (CLM_TYPE_CD) value of 1, capitation payments have a claim type code value of 2, and supplemental payments have a claim type code value of 5. FFS claims representing S-CHIP benefits have a claim type code value of A, capitation payments have a claim type code value of B, and supplemental payments have a claim type code value of E.
Payment data on managed care encounter records are redacted from the TAF RIFs.
Managed care encounter records representing Medicaid or M-CHIP services have a claim type code value of 3 and service tracking claims have a claim type code value of 4. Managed care encounter records representing S-CHIP benefits have a claim type code value of C, and service tracking claims have a claim type code value of D.
Once an individual is determined eligible for Medicaid, coverage is effective either on the date of application or the first day of the month of application. Benefits also may be covered retroactively for up to three months prior to the month of application if the individual would have been eligible during that period had he or she applied. States are required to update beneficiaries\u2019 eligibility records in T-MSIS to reflect the retroactive coverage period. However, if a state did not update the eligibility record of a beneficiary to reflect the retroactive coverage period, enrollment information like CHIP code may be missing during the date of service on a paid claim.
We determined the eligibility group by using ELGBLTY_GRP_CD_LTST, the most recent eligibility group reported for the beneficiary in the calendar year.
The federal reimbursement category code indicates whether a claim represents state expenditures that qualify for federal matching funds under Title XIX (Medicaid), Title XXI (CHIP), or other legislation. This analysis examines the extent to which records in the OT file are reported with a federal reimbursement category code consistent with the beneficiary's enrollment in the Medicaid or CHIP program.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5221"", ""relatedTopics"": [{""measureId"": 111, ""measureName"": ""Category of Service Code - IP"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 0}, {""measureId"": 112, ""measureName"": ""Category of Service Code - LT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 1}, {""measureId"": 113, ""measureName"": ""Category of Service Code - OT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 2}, {""measureId"": 114, ""measureName"": ""Category of Service Code - RX"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 3}, {""measureId"": 115, ""measureName"": ""Federal Reimbursement Category Code - IP"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 4}, {""measureId"": 116, ""measureName"": ""Federal Reimbursement Category Code - LT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 5}, {""measureId"": 118, ""measureName"": ""Federal Reimbursement Category Code - RX"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 7}]}" 118,"{""measureId"": 118, ""measureName"": ""Federal Reimbursement Category Code - RX"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Fed-Srvc-Reimb-Cd-RX.pdf"", ""background"": {""content"": ""Medicaid and the Children’s Health Insurance Program (CHIP) are joint state-federal medical assistance programs in which the state administers the program and pays providers or health plans directly for covered services, and the federal government reimburses a set percentage of the state’s expenditures (referred to as the \""federal match\""). Because the federal government applies different federal matching rates to Medicaid and CHIP beneficiaries, states report Medicaid and CHIP expenditure data separately.
To claim federal matching funds for their medical assistance programs, states report aggregate expenditure data for Medicaid beneficiaries covered under Title XIX of the Social Security Act to the Centers for Medicare & Medicaid Services (CMS) on Form CMS-64, Quarterly Medicaid Statement of Expenditures. States report aggregate expenditure data for the CHIP beneficiaries covered under Title XXI of the Social Security Act on Form CMS-21, Quarterly CHIP Statement of Expenditures . States may cover children using CHIP funds by expanding their Medicaid programs (called \""Medicaid expansion CHIP,\"" or M-CHIP), creating a program separate from their existing Medicaid programs (called \""separate CHIP,\"" or S-CHIP), or adopting a combination of both approaches. Expenditures for both M-CHIP and S-CHIP beneficiaries, which qualify for the higher CHIP federal match rate, are reported on Form CMS-21 and not Form CMS-64.
There are three data elements in the T-MSIS Analytic Files (TAF) inpatient (IP), long-term care (LT), other services (OT), and pharmacy (RX) files that contain information about how states report Medicaid and CHIP expenditures to claim federal matching funds: the federal reimbursement category code, the CMS-64 category of service code, and the CMS-21 category of service code (Table 1). [1] The federal reimbursement category code indicates whether the expenditure was reimbursed under Title XIX, Title XXI, the Affordable Care Act (ACA), or other legislation. As noted earlier, if the expenditure was covered by Title XIX of the Social Security Act, it should be reported on Form CMS-64; if the expenditure was covered by Title XXI of the Social Security Act it should be reported on Form CMS-21. The CMS-64 category of service code indicates the specific category of service on Form CMS-64 into which the state will report the Medicaid expenditures associated with a paid claim. The CMS-21 category of service code indicates the specific category of service on Form CMS-21 into which the state will report the CHIP expenditures associated with a paid claim.
Table 1. TAF variables that classify expenditures into federal reimbursement categories
| Variable | TAF data element name (TAF RIF name in parentheses if different) | IP, LT, OT, and RX records expected to have non-missing values | Description |
|---|---|---|---|
| Federal reimbursement category code | CMS_64_FED_REIMBRSMT_CTGRY_CD (CMS_64_FED_CTGRY_CD) | All records | This code indicates whether the claim was matched with federal funding under Title XIX (value of ‘01’), Title XXI (‘02’), the Affordable Care Act (‘03’), a or federal funding under other legislation (‘04’). |
| CMS-64 category of service code | XIX_SRVC_CTGRY_CD | Only records representing expenditures that the state will report on the CMS-64, which should link to beneficiaries enrolled in non-CHIP Medicaid | A four-digit code (from 001A to 0050) indicating the line on the Form CMS-64 on which the state reported the expenditures associated with the claim. The Form CMS-64 is used to report expenditures that qualify for federal matching funds under Title XXI. A full list of the category of service and definition for each line can be found at https://www.medicaid.gov/medicaid/downloads/cms-649-base-category-of-services-definition-2-14.pdf . |
| CMS-21 category of service code | XIX_SRVC_CTGRY_CD | Only records representing expenditures that the state will report on the CMS-21, which should link to beneficiaries enrolled in Medicaid-expansion CHIP or Separate CHIP | A three-digit code (from 01A to 35B) indicating the category of service (or line) on the Form CMS-21 on which the state reported the expenditures associated with the claim. The Form CMS-21 is used to report expenditures that qualify for federal matching funds under Title XXI. A full list of the category of service and definition for each line can be found at https://www.medicaid.gov/medicaid/downloads/cms-21-base-category-of-services-definition-2-14.pdf . |
a The federal reimbursement category code \""03\"" was previously a valid value used by some states on claims for their ACA adult expansion population. However, CMS retired the \""03\"" value in September 2020 and states are expected to use the \""01\"" value for this population moving forward since the enhanced match rate for these enrollees is authorized under Title XIX.
If states are populating these data elements correctly, claim records that link to beneficiaries enrolled in Title XIX Medicaid will be coded to indicate the expenditure is reportable on Form CMS-64. [2] Similarly, claim records that link to beneficiaries enrolled in Title XXI CHIP—including both M-CHIP and S-CHIP programs—will be coded to indicate the expenditure is reportable on Form CMS-21.
Other TAF data elements can be used to summarize Medicaid and CHIP spending associated with certain services or benefit categories, including the state-assigned type of service code, the federally-assigned service category code, and the benefit type code. [3] However, those data elements will not necessarily line up with the CMS-64 or CMS-21 categories that states use to report expenditures for the purposes of federal reimbursement.
In this data quality assessment, we evaluate the percentage of records in each claims file reported with a federal reimbursement category code that is consistent with the beneficiary’s enrollment in the Medicaid or CHIP programs. We also evaluate the percentage of records in each claims file reported with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid or CHIP program.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
More information about differences between expenditures in the TAF and the CMS-64 data can be found in the DQ Atlas single topic displays under the Expenditure Benchmarking topic area.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX . For more information about the FASC code, see the methodology brief, “Assigning TAF Records to a Federally Assigned Service Category,” on the DQ Atlas Resources page.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
More information about differences between expenditures in the TAF and the CMS-64 data can be found in the DQ Atlas single topic displays under the Expenditure Benchmarking topic area.
More information on states with missing or invalid type of service code can be found in the DQ Atlas single topic displays for Type of Service - IP , Type of Service - OT , Type of Service - LT , and Type of Service - RX . For more information about the FASC code, see the methodology brief, \u201cAssigning TAF Records to a Federally Assigned Service Category,\u201d on the DQ Atlas Resources page.
We began by selecting all records from the IP, LT, OT, and RX claims files that represented Medicaid and S-CHIP fee-for-service (FFS), capitation, and supplemental payments based on the claim type code. [4] , [5] We excluded managed care encounter records from the analysis because the payments recorded on those claims do not represent state expenditures and are not recorded on the CMS-64 or CMS-21. [6] We also excluded service tracking claims from the analysis because these records are lump-sum payments that that cannot be linked to an eligibility record. [7]
To determine whether the beneficiary was enrolled in Medicaid, M-CHIP, or S-CHIP on the date of service, we linked the claims records to the matching eligibility record using the unique beneficiary identifier coded on the claim (MSIS ID). We determined Medicaid, M-CHIP, and S-CHIP enrollment by examining the CHIP code on records in the TAF Demographic and Eligibility (DE) file from the month that corresponded to the date of service on the claim. We excluded from the analysis claims records with a missing or invalid beneficiary identifier. Additionally, we excluded claims that did not link to an eligibility record, as well as those that linked to an eligibility record but did not have a CHIP code populated (indicating the individual was enrolled in Medicaid or CHIP) during the date of service on the claim. [8]
Finally, we calculated the percentage of records in each file with a federal reimbursement category code consistent with the beneficiary’s reported enrollment in Medicaid, M-CHIP, or S-CHIP. The expected coding for federal reimbursement category code based on CHIP code is shown in Table 2. Because Medicaid beneficiaries enrolled in an adult expansion eligibility group may have a different federal reimbursement category code than Medicaid beneficiaries not enrolled in such a group, we examined these two groups of Medicaid beneficiaries separately. [9] We also calculated the percentage of records with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid or CHIP program. The expected coding for CMS-64 and CMS-21 category of service code based on CHIP code is shown in Table 3.
Table 2. Expected coding for federal reimbursement category code
| Enrollment type | CHIP code on eligibility record (CHIP_CD) | Latest eligibility group on eligibility record (ELGBLTY_GRP_CD_LTST) | Expected federal reimbursement category code on claim record |
|---|---|---|---|
| Medicaid – Adult expansion group | 1 (Medicaid) | 72 (Adult Group - Individuals at or below 133% FPL Age 19 through 64 - newly eligible for all states) 73 (Adult Group - Individuals at or below 133% FPL Age 19 through 64- not newly eligible for non 1905z(3) states), 74 (Adult Group - Individuals at or below 133% FPL Age 19 through 64 – not newly eligible parent/ caretaker relative(s) in 1905z(3) states), or 75 (Adult Group - Individuals at or below 133% FPL Age 19 through 64- not newly eligible non-parent/ caretaker-relative(s) in 1905z(3) states) | 01 (Federal funding under Title XIX) or 03 (Federal funding under ACA) a |
| Medicaid – All other | 1 (Medicaid) | Any valid code or null/missing, except: 72, 73, 74, 75, or 61-68 (CHIP eligibility groups) | 01 (Federal funding under Title XIX) |
| M-CHIP | 2 (M-CHIP) | N/A | 02 (Federal funding under Title XXI) |
| S-CHIP | 3 (S-CHIP) | N/A | 02 (Federal funding under Title XXI) |
a In this assessment, we consider both \""01\"" and \""03\"" to be consistent values for Medicaid-enrolled beneficiaries in the adult expansion eligibility groups. The value \""03\"" (Federal funding under ACA) was retired as a valid value on September 30, 2020 and while the value is still present in the data before that date, we would expect to see relatively few claims with this value after that date.
ACA = Affordable Care Act; FPL = Federal Poverty Level; N/A = not available.
Table 3. Expected coding for CMS-64 and CMS-21 category of service codes
| CHIP code on eligibility record (CHIP_CD) | Expected CMS-64 category of service code on claims record | Expected CMS-21 category of service code on claims record |
|---|---|---|
| 1 (Medicaid) |
|
|
| 2 (M-CHIP) | Field should be blank |
|
| 3 (S-CHIP) | Field should be blank |
|
Data quality assessment criteria
We categorized each state as having low, medium, or high data quality concern for the federal reimbursement category code, depending on the percentage of records with a federal reimbursement category code consistent with the beneficiary’s enrollment in the Medicaid, M-CHIP, or S-CHIP programs, as shown in Table 4.
Table 4. Criteria for DQ assessment of federal reimbursement category code
| Percentage of claim lines with a federal reimbursement category code consistent with program enrollment | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
We categorized each state as having low, medium, or high data quality concern for the category of service code, depending on the percentage of records with a CMS-64 or CMS-21 category of service code consistent with the beneficiary’s enrollment in the Medicaid, M-CHIP, or S-CHIP programs, as shown in Table 5. As shown in Table 3, only the CMS-64 or CMS-21 category of service code is expected to be populated on a single claim record. When both fields were populated on the same claim (even if one or both values were invalid), that record was inconsistent with enrollment in the Medicaid, M-CHIP, or S-CHIP programs for the purposes of this analysis.
Table 5. Criteria for DQ assessment of category of service codes
| Percentage of claim lines with a CMS-64 category of service code or a CMS-21 category of service code consistent with program enrollment | DQ assessment |
|---|---|
| x ≥ 90 percent | Low concern |
| 80 percent ≤ x < 90 percent | Medium concern |
| 50 percent ≤ x < 80 percent | High concern |
| x < 50 percent | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
FFS claims representing Medicaid or M-CHIP services have a claim type code (CLM_TYPE_CD) value of 1, capitation payments have a claim type code value of 2, and supplemental payments have a claim type code value of 5. FFS claims representing S-CHIP benefits have a claim type code value of A, capitation payments have a claim type code value of B, and supplemental payments have a claim type code value of E.
Payment data on managed care encounter records are redacted from the TAF RIFs.
Managed care encounter records representing Medicaid or M-CHIP services have a claim type code value of 3 and service tracking claims have a claim type code value of 4. Managed care encounter records representing S-CHIP benefits have a claim type code value of C, and service tracking claims have a claim type code value of D.
Once an individual is determined eligible for Medicaid, coverage is effective either on the date of application or the first day of the month of application. Benefits also may be covered retroactively for up to three months prior to the month of application if the individual would have been eligible during that period had he or she applied. States are required to update beneficiaries’ eligibility records in T-MSIS to reflect the retroactive coverage period. However, if a state did not update the eligibility record of a beneficiary to reflect the retroactive coverage period, enrollment information like CHIP code may be missing during the date of service on a paid claim.
We determined the eligibility group by using ELGBLTY_GRP_CD_LTST, the most recent eligibility group reported for the beneficiary in the calendar year.
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page, and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
FFS claims representing Medicaid or M-CHIP services have a claim type code (CLM_TYPE_CD) value of 1, capitation payments have a claim type code value of 2, and supplemental payments have a claim type code value of 5. FFS claims representing S-CHIP benefits have a claim type code value of A, capitation payments have a claim type code value of B, and supplemental payments have a claim type code value of E.
Payment data on managed care encounter records are redacted from the TAF RIFs.
Managed care encounter records representing Medicaid or M-CHIP services have a claim type code value of 3 and service tracking claims have a claim type code value of 4. Managed care encounter records representing S-CHIP benefits have a claim type code value of C, and service tracking claims have a claim type code value of D.
Once an individual is determined eligible for Medicaid, coverage is effective either on the date of application or the first day of the month of application. Benefits also may be covered retroactively for up to three months prior to the month of application if the individual would have been eligible during that period had he or she applied. States are required to update beneficiaries\u2019 eligibility records in T-MSIS to reflect the retroactive coverage period. However, if a state did not update the eligibility record of a beneficiary to reflect the retroactive coverage period, enrollment information like CHIP code may be missing during the date of service on a paid claim.
We determined the eligibility group by using ELGBLTY_GRP_CD_LTST, the most recent eligibility group reported for the beneficiary in the calendar year.
The federal reimbursement category code indicates whether a claim represents state expenditures that qualify for federal matching funds under Title XIX (Medicaid), Title XXI (CHIP), or other legislation. This analysis examines the extent to which records in the RX file are reported with a federal reimbursement category code consistent with the beneficiary's enrollment in the Medicaid or CHIP program.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5221"", ""relatedTopics"": [{""measureId"": 111, ""measureName"": ""Category of Service Code - IP"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 0}, {""measureId"": 112, ""measureName"": ""Category of Service Code - LT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 1}, {""measureId"": 113, ""measureName"": ""Category of Service Code - OT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 2}, {""measureId"": 114, ""measureName"": ""Category of Service Code - RX"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 3}, {""measureId"": 115, ""measureName"": ""Federal Reimbursement Category Code - IP"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 4}, {""measureId"": 116, ""measureName"": ""Federal Reimbursement Category Code - LT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 5}, {""measureId"": 117, ""measureName"": ""Federal Reimbursement Category Code - OT"", ""groupId"": 12, ""groupName"": ""Financial Reporting"", ""order"": 6}]}" 119,"{""measureId"": 119, ""measureName"": ""Primary Language"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Primary-Lang.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) are research-optimized data on beneficiaries enrolled in Medicaid and the Children’s Health Insurance Program (CHIP). The Annual Demographic and Eligibility (DE) file contains information on beneficiary demographic characteristics, including primary or preferred language and English language proficiency. States often gather this information via applications for Medicaid and CHIP benefits. However, states may vary in whether and how they ask these questions because language information is not required for eligibility determination. [1] Although the U.S. Department of Health and Human Services (HHS) recommends that states give at minimum two language options other than English ( Spanish or Other language ), some states give applicants a range of languages from which to choose.
States use one data element to submit information on a beneficiary’s language preference in T-MSIS: the primary language code, [2] which uses a valid value set of approximately 500 three-letter categories from the International Organization for Standardization (ISO) codes for the representation of names of languages, Part 2 (ISO 639-2). [3] When eligibility records are created in TAF, this source data element is used to create two TAF data elements: (1) the primary language code, [4] which takes on all of the ISO 639-2 language codes as valid values; and (2) the constructed primary language group code, with 14 valid values. [5] States vary in whether they report the English ISO language code or leave the data element blank when a beneficiary’s primary language is English.
In addition, states use one data element to submit information on how well a beneficiary speaks English: the English language proficiency code, [6] which uses a four-point scale (no spoken proficiency, does not speak well, speaks well, speaks very well).
This analysis assesses the usability of the constructed primary language group code [7] by measuring the extent to which certain language categories (English, Spanish, Other) align with an external benchmark, the American Community Survey (ACS). We also assess the usability of the English language proficiency code among beneficiaries with Spanish or Other primary language in TAF by measuring the extent to which certain English proficiency categories (\""not well\"" or \""no spoken proficiency\"") differ substantively from the ACS.
More information about the collection of race, ethnicity, and language data in Medicaid Applications is available at https://www.shvs.org/wp-content/uploads/2021/05/SHVS-50-State-Review-EDITED.pdf .
The definition of the T-MSIS primary language code (and in turn, the TAF language code and constructed primary language group code) changed slightly in 2022 to represent \""the individual’s preferred spoken or written language\"" (which may or may not be English) rather than the individual’s language \""other than English.\"" The updated definition reflects what already had been observed in T-MSIS data: most states report a majority of eligibility records with English as the primary language code, suggesting they were not following the original variable definition but rather reporting what they gather on Medicaid applications, which offer English as an option for primary language.
For a list of valid values, see Codes for Representation of Names of Languages at https://www.loc.gov/standards/iso639-2/php/code_list.php . The ISO reviews its code sets every five years; the ISO 639-2 lifecycle can be viewed at https://www.iso.org/standard/4767.html .
In the internal version of the TAF data within the CMCS data environment, the primary language code is stored as \""Language (Other Than English) Code\"" (OTHR_LANG_HOME_CD).\""
In the internal version of the TAF data within the CMCS data environment, the constructed primary language group code is stored as \""Constructed Primary Language (Other Than English) Group Code\"" (PRMRY_LANG_FLAG).\"" The categories in the constructed primary language group code come from the Social Security Administration Master Beneficiary Record variable for preferred written language. Although English is a valid value, it is widely accepted that users should assume English is the preferred language if the field is left blank.
The name of the T-MSIS English language proficiency code (and in turn, the TAF English language proficiency code) changed slightly in 2022 to clarify that states should capture the level of spoken English proficiency regardless of whether it is the individual’s preferred spoken or written language. Previously, it was unclear whether the field should be populated for beneficiaries whose primary language was not English. The updated name reflects what states likely already gather on Medicaid applications.
We were not able to assess the validity of each language value available under the language code. Users interested in the level of specificity offered by the language code should adapt the methods used for this data quality assessment using the three-letter values for the languages of interest.
More information about the collection of race, ethnicity, and language data in Medicaid Applications is available at https://www.shvs.org/wp-content/uploads/2021/05/SHVS-50-State-Review-EDITED.pdf .
The definition of the T-MSIS primary language code (and in turn, the TAF language code and constructed primary language group code) changed slightly in 2022 to represent \""the individual\u2019s preferred spoken or written language\"" (which may or may not be English) rather than the individual\u2019s language \""other than English.\"" The updated definition reflects what already had been observed in T-MSIS data: most states report a majority of eligibility records with English as the primary language code, suggesting they were not following the original variable definition but rather reporting what they gather on Medicaid applications, which offer English as an option for primary language.
For a list of valid values, see Codes for Representation of Names of Languages at https://www.loc.gov/standards/iso639-2/php/code_list.php . The ISO reviews its code sets every five years; the ISO 639-2 lifecycle can be viewed at https://www.iso.org/standard/4767.html .
In the internal version of the TAF data within the CMCS data environment, the primary language code is stored as \""Language (Other Than English) Code\"" (OTHR_LANG_HOME_CD).\""
In the internal version of the TAF data within the CMCS data environment, the constructed primary language group code is stored as \""Constructed Primary Language (Other Than English) Group Code\"" (PRMRY_LANG_FLAG).\"" The categories in the constructed primary language group code come from the Social Security Administration Master Beneficiary Record variable for preferred written language. Although English is a valid value, it is widely accepted that users should assume English is the preferred language if the field is left blank.
The name of the T-MSIS English language proficiency code (and in turn, the TAF English language proficiency code) changed slightly in 2022 to clarify that states should capture the level of spoken English proficiency regardless of whether it is the individual\u2019s preferred spoken or written language. Previously, it was unclear whether the field should be populated for beneficiaries whose primary language was not English. The updated name reflects what states likely already gather on Medicaid applications.
We were not able to assess the validity of each language value available under the language code. Users interested in the level of specificity offered by the language code should adapt the methods used for this data quality assessment using the three-letter values for the languages of interest.
For the primary language analysis, we examined the constructed primary language group code (PRMRY_LANG_FLAG) on non-dummy enrollment records [8] in the TAF DE file. [9] We tabulated the proportion of records that fell into one of three categories: English, Spanish, and Other languages. [10] Following the Social Security Administration (SSA) and ACS convention, we categorized missing values as English; however, we calculated the percentage with null values separately for informational purposes.
To construct the benchmark, we used the ACS five-year estimates [11] Public Use Microdata Sample (PUMS) [12] for a given year. The ACS data, which are collected annually from a nationally representative random sample of households, contain information on language spoken at home, English proficiency only for those respondents who do not speak English at home, and health insurance coverage. After pulling the ACS microdata from PUMS, we selected all individuals who reported having \""Medicaid, Medical Assistance, or any kind of government-assistance plan for those with low incomes or a disability\"" at the time of the survey. For individuals in that group, we calculated the percentage who were in the three aggregate language categories (English, Spanish, Other languages).
The ACS asks respondents whether they speak a language other than English at home and, if so, what language they speak and how well they speak English. Users can assume that if the subsequent language and proficiency questions are not answered, the person speaks only English.
ACS data are used by many stakeholders, including federal and state government agencies for policy and program-funding activities, and are considered a highly reliable source of demographic data. However, self-reporting of health insurance coverage for the ACS often results in an undercount compared to the number of Medicaid beneficiaries who appear in administrative data. Therefore, in this data quality assessment, we compare the percentage of Medicaid beneficiaries in each language category in TAF to the comparable distribution in the ACS, rather than comparing the count of individuals in each category.
Table 1 shows the level of concern for the TAF primary language variable based on how well the percentage of beneficiaries in each of the three aggregate language categories aligned with the ACS. For language categories that accounted for more than 10 percent of a state’s Medicaid population in the ACS, we deemed a \""substantive difference\"" between TAF and ACS to be more than 10 percentage points. For language categories that accounted for more than 2 percent and less than or equal to 10 percent of a state’s Medicaid population in the ACS, we considered a substantive difference between the data sources to exist if there were no TAF records in the given language category. We assessed each state on the number of language categories in which TAF differs substantively from the ACS (either a 10 percentage point difference or zero TAF records, depending on the language category proportion in the ACS). We considered states with a missing primary language value for all beneficiaries in TAF to be unusable for language analyses.
Table 1. Criteria for DQ assessment of the constructed primary language group code
| Number of language categories (out of three) in which TAF differs substantively from ACS | Percentage of beneficiaries with missing primary language in TAF | DQ assessment |
|---|---|---|
| 0 | x < 100 percent | Low concern |
| 1 | x < 100 percent | Medium concern |
| 2 | x < 100 percent | High concern |
| 3 | x < 100 percent | Unusable |
| Any value | x = 100 percent | Unusable |
For the spoken English proficiency analysis, we examined the English language proficiency code (PRMRY_LANG_ENGLSH_PRFCNCY_CD) on non-dummy enrollment records in the TAF DE file. We limited the analysis to beneficiaries who indicated a non-English primary language. [13] We tabulated the proportion of these records that fell into each one of two aggregate proficiency categories (\""very well\"" or \""well\"" versus \""not well\"" or \""no spoken proficiency\""), as well as the proportion with missing values. The assessment focuses on beneficiaries with limited English proficiency—those with proficiency categories \""not well\"" or \""no spoken proficiency.\""
We used two criteria to assess each state’s spoken English proficiency data, both measured only among beneficiaries that reported a non-English primary language. First, we calculated the percentage of these beneficiaries with missing English proficiency data. Second, we assessed the extent to which the percentage of beneficiaries with limited English proficiency—that is, those in the categories \""not well\"" or \""no spoken proficiency\""—aligns with the ACS benchmark data for the state (Table 2).
Table 2. Criteria for DQ assessment of English language proficiency code (assessed only among beneficiaries with a non-English primary language)
We excluded DE records representing beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state (those with MISG_ELGBLTY_DATA_IND = 1).
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \""Production of the TAF Research Identifiable Files.\""
We chose only three groups (English, Spanish, and Other) for the assessment because they align with the HHS minimum recommendations.
ACS five-year estimates are more reliable and complete than ACS one-year estimates and the Current Population Survey because they include smaller geographic areas and have a larger sample size. For more details, see https://www.census.gov/programs-surveys/acs/guidance/estimates.html and https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2677056/ .
Although CMS guidance is to report English proficiency for all beneficiaries (even those whose primary language is English), most analytic users will be interested in the ability to speak English of beneficiaries whose primary language is not English.
We excluded DE records representing beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state (those with MISG_ELGBLTY_DATA_IND = 1).
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \""Production of the TAF Research Identifiable Files.\""
We chose only three groups (English, Spanish, and Other) for the assessment because they align with the HHS minimum recommendations.
ACS five-year estimates are more reliable and complete than ACS one-year estimates and the Current Population Survey because they include smaller geographic areas and have a larger sample size. For more details, see https://www.census.gov/programs-surveys/acs/guidance/estimates.html and https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2677056/ .
Although CMS guidance is to report English proficiency for all beneficiaries (even those whose primary language is English), most analytic users will be interested in the ability to speak English of beneficiaries whose primary language is not English.
The TAF eligibility files include information on select demographic characteristics of beneficiaries enrolled in Medicaid or CHIP. This analysis examines the completeness and reliability of language information in the TAF. It also examines how well the TAF data on primary or preferred language and English language proficiency align with an external benchmark, the U.S. Census Bureau's American Community Survey.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4181"", ""relatedTopics"": [{""measureId"": 120, ""measureName"": ""English Language Proficiency"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""order"": 1}]}" 120,"{""measureId"": 120, ""measureName"": ""English Language Proficiency"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-Eng-Lang-Prof.pdf"", ""background"": {""content"": ""The T-MSIS Analytic Files (TAF) are research-optimized data on beneficiaries enrolled in Medicaid and the Children’s Health Insurance Program (CHIP). The Annual Demographic and Eligibility (DE) file contains information on beneficiary demographic characteristics, including primary or preferred language and English language proficiency. States often gather this information via applications for Medicaid and CHIP benefits. However, states may vary in whether and how they ask these questions because language information is not required for eligibility determination. [1] Although the U.S. Department of Health and Human Services (HHS) recommends that states give at minimum two language options other than English ( Spanish or Other language ), some states give applicants a range of languages from which to choose.
States use one data element to submit information on a beneficiary’s language preference in T-MSIS: the primary language code, [2] which uses a valid value set of approximately 500 three-letter categories from the International Organization for Standardization (ISO) codes for the representation of names of languages, Part 2 (ISO 639-2). [3] When eligibility records are created in TAF, this source data element is used to create two TAF data elements: (1) the primary language code, [4] which takes on all of the ISO 639-2 language codes as valid values; and (2) the constructed primary language group code, with 14 valid values. [5] States vary in whether they report the English ISO language code or leave the data element blank when a beneficiary’s primary language is English.
In addition, states use one data element to submit information on how well a beneficiary speaks English: the English language proficiency code, [6] which uses a four-point scale (no spoken proficiency, does not speak well, speaks well, speaks very well).
This analysis assesses the usability of the constructed primary language group code [7] by measuring the extent to which certain language categories (English, Spanish, Other) align with an external benchmark, the American Community Survey (ACS). We also assess the usability of the English language proficiency code among beneficiaries with Spanish or Other primary language in TAF by measuring the extent to which certain English proficiency categories (\""not well\"" or \""no spoken proficiency\"") differ substantively from the ACS.
More information about the collection of race, ethnicity, and language data in Medicaid Applications is available at https://www.shvs.org/wp-content/uploads/2021/05/SHVS-50-State-Review-EDITED.pdf .
The definition of the T-MSIS primary language code (and in turn, the TAF language code and constructed primary language group code) changed slightly in 2022 to represent \""the individual’s preferred spoken or written language\"" (which may or may not be English) rather than the individual’s language \""other than English.\"" The updated definition reflects what already had been observed in T-MSIS data: most states report a majority of eligibility records with English as the primary language code, suggesting they were not following the original variable definition but rather reporting what they gather on Medicaid applications, which offer English as an option for primary language.
For a list of valid values, see Codes for Representation of Names of Languages at https://www.loc.gov/standards/iso639-2/php/code_list.php . The ISO reviews its code sets every five years; the ISO 639-2 lifecycle can be viewed at https://www.iso.org/standard/4767.html .
In the internal version of the TAF data within the CMCS data environment, the primary language code is stored as \""Language (Other Than English) Code\"" (OTHR_LANG_HOME_CD).\""
In the internal version of the TAF data within the CMCS data environment, the constructed primary language group code is stored as \""Constructed Primary Language (Other Than English) Group Code\"" (PRMRY_LANG_FLAG).\"" The categories in the constructed primary language group code come from the Social Security Administration Master Beneficiary Record variable for preferred written language. Although English is a valid value, it is widely accepted that users should assume English is the preferred language if the field is left blank.
The name of the T-MSIS English language proficiency code (and in turn, the TAF English language proficiency code) changed slightly in 2022 to clarify that states should capture the level of spoken English proficiency regardless of whether it is the individual’s preferred spoken or written language. Previously, it was unclear whether the field should be populated for beneficiaries whose primary language was not English. The updated name reflects what states likely already gather on Medicaid applications.
We were not able to assess the validity of each language value available under the language code. Users interested in the level of specificity offered by the language code should adapt the methods used for this data quality assessment using the three-letter values for the languages of interest.
More information about the collection of race, ethnicity, and language data in Medicaid Applications is available at https://www.shvs.org/wp-content/uploads/2021/05/SHVS-50-State-Review-EDITED.pdf .
The definition of the T-MSIS primary language code (and in turn, the TAF language code and constructed primary language group code) changed slightly in 2022 to represent \""the individual\u2019s preferred spoken or written language\"" (which may or may not be English) rather than the individual\u2019s language \""other than English.\"" The updated definition reflects what already had been observed in T-MSIS data: most states report a majority of eligibility records with English as the primary language code, suggesting they were not following the original variable definition but rather reporting what they gather on Medicaid applications, which offer English as an option for primary language.
For a list of valid values, see Codes for Representation of Names of Languages at https://www.loc.gov/standards/iso639-2/php/code_list.php . The ISO reviews its code sets every five years; the ISO 639-2 lifecycle can be viewed at https://www.iso.org/standard/4767.html .
In the internal version of the TAF data within the CMCS data environment, the primary language code is stored as \""Language (Other Than English) Code\"" (OTHR_LANG_HOME_CD).\""
In the internal version of the TAF data within the CMCS data environment, the constructed primary language group code is stored as \""Constructed Primary Language (Other Than English) Group Code\"" (PRMRY_LANG_FLAG).\"" The categories in the constructed primary language group code come from the Social Security Administration Master Beneficiary Record variable for preferred written language. Although English is a valid value, it is widely accepted that users should assume English is the preferred language if the field is left blank.
The name of the T-MSIS English language proficiency code (and in turn, the TAF English language proficiency code) changed slightly in 2022 to clarify that states should capture the level of spoken English proficiency regardless of whether it is the individual\u2019s preferred spoken or written language. Previously, it was unclear whether the field should be populated for beneficiaries whose primary language was not English. The updated name reflects what states likely already gather on Medicaid applications.
We were not able to assess the validity of each language value available under the language code. Users interested in the level of specificity offered by the language code should adapt the methods used for this data quality assessment using the three-letter values for the languages of interest.
For the primary language analysis, we examined the constructed primary language group code (PRMRY_LANG_FLAG) on non-dummy enrollment records [8] in the TAF DE file. [9] We tabulated the proportion of records that fell into one of three categories: English, Spanish, and Other languages. [10] Following the Social Security Administration (SSA) and ACS convention, we categorized missing values as English; however, we calculated the percentage with null values separately for informational purposes.
To construct the benchmark, we used the ACS five-year estimates [11] Public Use Microdata Sample (PUMS) [12] for a given year. The ACS data, which are collected annually from a nationally representative random sample of households, contain information on language spoken at home, English proficiency only for those respondents who do not speak English at home, and health insurance coverage. After pulling the ACS microdata from PUMS, we selected all individuals who reported having \""Medicaid, Medical Assistance, or any kind of government-assistance plan for those with low incomes or a disability\"" at the time of the survey. For individuals in that group, we calculated the percentage who were in the three aggregate language categories (English, Spanish, Other languages).
The ACS asks respondents whether they speak a language other than English at home and, if so, what language they speak and how well they speak English. Users can assume that if the subsequent language and proficiency questions are not answered, the person speaks only English.
ACS data are used by many stakeholders, including federal and state government agencies for policy and program-funding activities, and are considered a highly reliable source of demographic data. However, self-reporting of health insurance coverage for the ACS often results in an undercount compared to the number of Medicaid beneficiaries who appear in administrative data. Therefore, in this data quality assessment, we compare the percentage of Medicaid beneficiaries in each language category in TAF to the comparable distribution in the ACS, rather than comparing the count of individuals in each category.
Table 1 shows the level of concern for the TAF primary language variable based on how well the percentage of beneficiaries in each of the three aggregate language categories aligned with the ACS. For language categories that accounted for more than 10 percent of a state’s Medicaid population in the ACS, we deemed a \""substantive difference\"" between TAF and ACS to be more than 10 percentage points. For language categories that accounted for more than 2 percent and less than or equal to 10 percent of a state’s Medicaid population in the ACS, we considered a substantive difference between the data sources to exist if there were no TAF records in the given language category. We assessed each state on the number of language categories in which TAF differs substantively from the ACS (either a 10 percentage point difference or zero TAF records, depending on the language category proportion in the ACS). We considered states with a missing primary language value for all beneficiaries in TAF to be unusable for language analyses.
Table 1. Criteria for DQ assessment of the constructed primary language group code
| Number of language categories (out of three) in which TAF differs substantively from ACS | Percentage of beneficiaries with missing primary language in TAF | DQ assessment |
|---|---|---|
| 0 | x < 100 percent | Low concern |
| 1 | x < 100 percent | Medium concern |
| 2 | x < 100 percent | High concern |
| 3 | x < 100 percent | Unusable |
| Any value | x = 100 percent | Unusable |
For the spoken English proficiency analysis, we examined the English language proficiency code (PRMRY_LANG_ENGLSH_PRFCNCY_CD) on non-dummy enrollment records in the TAF DE file. We limited the analysis to beneficiaries who indicated a non-English primary language. [13] We tabulated the proportion of these records that fell into each one of two aggregate proficiency categories (\""very well\"" or \""well\"" versus \""not well\"" or \""no spoken proficiency\""), as well as the proportion with missing values. The assessment focuses on beneficiaries with limited English proficiency—those with proficiency categories \""not well\"" or \""no spoken proficiency.\""
We used two criteria to assess each state’s spoken English proficiency data, both measured only among beneficiaries that reported a non-English primary language. First, we calculated the percentage of these beneficiaries with missing English proficiency data. Second, we assessed the extent to which the percentage of beneficiaries with limited English proficiency—that is, those in the categories \""not well\"" or \""no spoken proficiency\""—aligns with the ACS benchmark data for the state (Table 2).
Table 2. Criteria for DQ assessment of English language proficiency code (assessed only among beneficiaries with a non-English primary language)
We excluded DE records representing beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state (those with MISG_ELGBLTY_DATA_IND = 1).
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \""Production of the TAF Research Identifiable Files.\""
We chose only three groups (English, Spanish, and Other) for the assessment because they align with the HHS minimum recommendations.
ACS five-year estimates are more reliable and complete than ACS one-year estimates and the Current Population Survey because they include smaller geographic areas and have a larger sample size. For more details, see https://www.census.gov/programs-surveys/acs/guidance/estimates.html and https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2677056/ .
Although CMS guidance is to report English proficiency for all beneficiaries (even those whose primary language is English), most analytic users will be interested in the ability to speak English of beneficiaries whose primary language is not English.
We excluded DE records representing beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state (those with MISG_ELGBLTY_DATA_IND = 1).
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \""Production of the TAF Research Identifiable Files.\""
We chose only three groups (English, Spanish, and Other) for the assessment because they align with the HHS minimum recommendations.
ACS five-year estimates are more reliable and complete than ACS one-year estimates and the Current Population Survey because they include smaller geographic areas and have a larger sample size. For more details, see https://www.census.gov/programs-surveys/acs/guidance/estimates.html and https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2677056/ .
Although CMS guidance is to report English proficiency for all beneficiaries (even those whose primary language is English), most analytic users will be interested in the ability to speak English of beneficiaries whose primary language is not English.
The TAF eligibility files include information on select demographic characteristics of beneficiaries enrolled in Medicaid or CHIP. This analysis examines the completeness and reliability of language information in the TAF. It also examines how well the TAF data on primary or preferred language and English language proficiency align with an external benchmark, the U.S. Census Bureau's American Community Survey.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4181"", ""relatedTopics"": [{""measureId"": 119, ""measureName"": ""Primary Language"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""order"": 0}]}" 121,"{""measureId"": 121, ""measureName"": ""Availability of CMC Plan Encounter Data"", ""groupId"": 4, ""groupName"": ""Claim Files Completeness"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-CMC-Plan-Encounters-All.pdf"", ""background"": {""content"": ""In 2020, 72 percent of Medicaid beneficiaries received the majority of their care through comprehensive managed care (CMC) plans. [1] States are required to report the services provided to beneficiaries through managed care plans in their monthly T-MSIS claims records. [2] CMS strengthened the requirement to report complete encounter data in 2018. [3] Complete reporting of CMC encounter data ensures that TAF users can accurately identify all services received by Medicaid and CHIP beneficiaries and their diagnosed health conditions, regardless of whether those services were delivered through managed care or the fee-for-service system.
Historically, some states have under-reported encounter records due to challenges with collecting information from CMC plans, as data reporting requirements must be written into each contract between the State and CMC plan. As states have continued to renegotiate CMC contracts in the time after the strengthened 2018 guidance on encounter data reporting, the volume and quality of encounter records should improve. Where CMC data are un- or under-reported, TAF users will underestimate health care utilization and the prevalence of medical conditions in beneficiaries who are enrolled in managed care. Additionally, qualify of care and access to care measures are likely to be unreliable in states with missing encounter data for some or all plans.
This data quality assessment examines the number of comprehensive managed care plans for which the state is reporting enrollment but not reporting any encounter data.
Centers for Medicare & Medicaid Services. “Medicaid Managed Care Enrollment and Program Characteristics, 2020.” Spring 2022. Available at https://www.medicaid.gov/medicaid/managed-care/enrollment-report/index.html Accessed August 11, 2022.
Subpart J: Conditions for Federal Financial Participation (FFP). Enrollee encounter data. 42 CFR §438.818 (May 6, 2016). Available at https://www.govinfo.gov/content/pkg/CFR-2018-title42-vol4/xml/CFR-2018-title42-vol4-part438.xml#seqnum438.818 . Accessed May 29, 2020.
Centers for Medicare & Medicaid Services. \""State Health Official Letter #18-008: Transformed-Medicaid Statistical Information System (T-MSIS).\"" August 10, 2018.
Centers for Medicare & Medicaid Services. \u201cMedicaid Managed Care Enrollment and Program Characteristics, 2020.\u201d Spring 2022. Available at https://www.medicaid.gov/medicaid/managed-care/enrollment-report/index.html Accessed August 11, 2022.
Subpart J: Conditions for Federal Financial Participation (FFP). Enrollee encounter data. 42 CFR \u00a7438.818 (May 6, 2016). Available at https://www.govinfo.gov/content/pkg/CFR-2018-title42-vol4/xml/CFR-2018-title42-vol4-part438.xml#seqnum438.818 . Accessed May 29, 2020.
Centers for Medicare & Medicaid Services. \""State Health Official Letter #18-008: Transformed-Medicaid Statistical Information System (T-MSIS).\"" August 10, 2018.
Using the TAF, [4] we tabulated the number of CMC plans that had Medicaid or CHIP beneficiaries enrolled in the calendar year but did not have any encounters [5] in the inpatient (IP) or other services (OT) claims files that could be linked to the plan. [6] This pattern suggests that specific CMC plans are entirely missing encounter records.
The analysis included enrollment and managed care encounter records for all CMC plans serving Medicaid and CHIP beneficiaries. We did not analyze states with no CMC program in operation during the year, as captured in the Medicaid Managed Care Enrollment and Program Characteristics (MMCEPC) report. [7] For Illinois, we restricted our analysis to the original version of the claim, and we excluded all subsequent adjustment records in the state’s TAF data. [8]
To assess potential issues with the completeness or quality of the CMC encounter data in each state, we calculated: (1) total number of CMC plans, (2) total number of CMC-enrolled months, and (3) number of CMC plans with no IP or OT encounters. The first two measures allow us to identify whether there are any CMCs in the state and whether there is any enrollment in CMC plans and are provided as contextual information for the assessment.
To calculate the number of CMC plans in each state with enrollment during the year, we identified the number of unique managed care plan ID (MC_PLAN_ID) values with a managed care plan type code (MC_PLAN_TYPE_CD) of ‘01’ or ‘04’ in the Annual Demographic & Eligibility (ADE) file. We then tabulated the number of CMC plans that had beneficiaries enrolled in the calendar year but did not have any encounters in either the IP or OT claims file that could be linked to the plan. Encounter records for a particular plan were identified as those with a claim type code of ‘3’ or ‘C’ and a matching plan ID present on the header record. We did not require a header record to link to an enrollment record that was coded as participating in the CMC plan to be counted in the measure. However, we did not a count a header record as representing encounter data from a particular plan if it had a missing beneficiary identifier or beneficiary identifier with an invalid format (values that start with “&”), because these records may represent capitation payments or other lump-sum payments being made to the plan rather than encounter records.
To identify states with incomplete managed care encounter reporting, we used the criteria presented in Table 1. States were only assessed as low concern if they had encounter data present for all CMC plans. More information on whether the encounter data were correctly formatted and whether the volume of encounter data was as expected given the size of the enrolled population can be found in other topics in DQ Atlas . [9]
Table 1. Criteria for DQ assessment of the availability of encounter data for CMC plans
| Number of CMC plans with no encounter records | DQ assessment |
|---|---|
| None | Low concern |
| At least one but not more than half of CMC plans in the state | Medium concern |
| More than half but not all CMC plans in the state | High concern |
| All CMC plans in the state | Unusable |
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
We identified managed care encounters by using claim type code (CLM_TYPE_CD) values of 3 and C. We limited the analysis to encounters with a managed care plan ID (MC_PLAN_ID) that linked to a managed care plan type code (MC_PLAN_TYPE_CD) of 01 (comprehensive managed care) or 04 (Health Insuring Organization).
States have the option to carve institutional long-term care services and prescription drugs out of their comprehensive managed care plan benefit packages, so the absence of encounter data in the LT or RX files does not necessarily indicate missing service use information. However, all comprehensive managed care plans are required to cover inpatient services and select outpatient services, and so IP and OT encounters are expected to be present in TAF for every comprehensive managed care plan.
States with no CMC program include those with zero enrollment in CMC plans, as well as those where less than one percent of the Medicaid program is enrolled in a CMC.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, “How to Use Illinois Claims Data,” on ResDAC.org.
See the DQ Atlas single-topic displays for the following topics: CMC Plan Encounters – IP , CMC Plan Encounters – LT , CMC Plan Encounters – OT , and CMC Plan Encounters – RX .
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
We identified managed care encounters by using claim type code (CLM_TYPE_CD) values of 3 and C. We limited the analysis to encounters with a managed care plan ID (MC_PLAN_ID) that linked to a managed care plan type code (MC_PLAN_TYPE_CD) of 01 (comprehensive managed care) or 04 (Health Insuring Organization).
States have the option to carve institutional long-term care services and prescription drugs out of their comprehensive managed care plan benefit packages, so the absence of encounter data in the LT or RX files does not necessarily indicate missing service use information. However, all comprehensive managed care plans are required to cover inpatient services and select outpatient services, and so IP and OT encounters are expected to be present in TAF for every comprehensive managed care plan.
States with no CMC program include those with zero enrollment in CMC plans, as well as those where less than one percent of the Medicaid program is enrolled in a CMC.
Because of limitations in its claims processing system, Illinois captures adjustments to original claims as incremental credits or debits rather than voiding the original claim and submitting a replacement record with the new payment amount. As a result, the version of a record with the latest adjudication date may not represent the final action claim as it does in all other states. To ensure the TAF correctly captures all expenditures reported by Illinois into T-MSIS, all service use records are included in the IP, OT, LT, and RX files. This means that in some cases, the TAF will include multiple versions of a single claim for Illinois, so including all records in an analysis will overcount service utilization. To ensure that utilization counts are correct, we restricted our analysis to the original claim from each claim family by selecting only records where ADJSTMT_IND = 0 and ADJSTMT_CLM_NUM = null. For more information, see the guide, \u201cHow to Use Illinois Claims Data,\u201d on ResDAC.org.
See the DQ Atlas single-topic displays for the following topics: CMC Plan Encounters \u2013 IP , CMC Plan Encounters \u2013 LT , CMC Plan Encounters \u2013 OT , and CMC Plan Encounters \u2013 RX .
States are required to report into T-MSIS the encounter records that reflect services provided to Medicaid and CHIP beneficiaries by managed care organizations. In states with managed care programs, the inclusion of encounter data in the T-MSIS submissions is critical for understanding the full picture of beneficiary service use and health status. This assessment shows which states are reporting encounter data for all comprehensive managed care plans that enrolled any Medicaid or CHIP beneficiaries during the year.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""5251"", ""relatedTopics"": []}" 122,"{""measureId"": 122, ""measureName"": ""SSI Participation"", ""groupId"": 1, ""groupName"": ""Enrollment Benchmarking"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-SSI-Participation.pdf"", ""background"": {""content"": ""Most individuals who qualify for Medicaid coverage do so through one of the major eligibility groups, each of which have different requirements: children, pregnant women, low-income adults, older adults (those age 65 and older), and adults who have a disabling condition. For instance, individuals covered under the Supplemental Security Income (SSI) program often qualify for Medicaid because they have limited financial means and a long-lasting disabling condition; SSI receipt is one of the primary pathways for individuals with disabilities to qualify for Medicaid. People with disabilities who are not receiving SSI may still enroll in Medicaid via other eligibility groups, such as the working disabled or disabled adult children groups (which are considered disability pathways) or the adult expansion group (which is not considered a disability pathway).
The SSI program provides financial support to individuals with long-lasting disabilities (including blindness) who have limited work history, income, and assets. [1] To qualify as having a disability or blindness, an individual must have a chronic physical or mental impairment that prevents them from doing any substantial gainful activity (if age 18 or older) or results in marked and severe functional limitations (if less than age 18). Social Security Administration (SSA) field offices manage SSI applications and verify non-medical eligibility information (such as age and employment) and state Disability Determination Services offices review the medical evidence to determine whether the person has a disability or is blind under the law. [2] On average, it takes three to five months to receive an initial decision on an SSI application; it can take substantially longer if an individual goes through the appeals process, which is common. [3] , [4] Once an individual is enrolled in SSI, they must report earnings or other changes (like increased assets or incarceration) that might disqualify them for SSI benefits as often as on a monthly basis. Additionally, SSA periodically conducts \""continuing disability reviews\"" to confirm that a beneficiary’s disability has not sufficiently improved to make them ineligible for benefits. [5]
Federal law requires states to provide Medicaid coverage to most individuals with disabilities who are receiving SSI payments. [6] Most states have a 1634 agreement with SSA, under which SSA makes Medicaid eligibility decisions in conjunction with SSI determinations (Table 1). In these states, SSI recipients are automatically conferred Medicaid coverage. Other states, called \""SSI criteria states,\"" offer Medicaid to all SSI recipients, but require a separate application for each program. Finally, a few states elect an option provided in section 209(b) of the Social Security Act Amendments of 1972 [7] which enables them to impose more stringent criteria for Medicaid eligibility than for SSI eligibility; these states are known as \""209(b) states.\"" Although federal law prohibits 209(b) states from using more restrictive standards for Medicaid eligibility than those in effect on January 1, 1972, some SSI recipients may not be eligible for Medicaid coverage in those states.
Table 1. State policy options to determine Medicaid eligibility for SSI recipients
| Policy | Description | Number of states |
|---|---|---|
| 1634 agreement |
| 35 |
| SSI criteria |
| 7 |
| 209(b) |
| 9 |
Sources: https://secure.ssa.gov/poms.nsf/lnx/0501715010 and https://www.gao.gov/assets/gao-21-473r.pdf
The eligibility group code in the T-MSIS Analytic Files (TAF), the research-ready version of T-MSIS, can be used to identify the basis on which an individual was deemed eligible for Medicaid or CHIP. [8] , [9] Thirty-four of the eligibility group codes can be used to identify beneficiaries who qualified on the basis of disability, although several of these codes are also used to indicate beneficiaries qualified on the basis of age (Table 2). Restricting to beneficiaries under age 65 for the eligibility groups that include both disabled and aged beneficiaries enables TAF users to restrict their analysis to those who qualify for Medicaid on the basis of disability.
Among the 34 eligibility group codes related to disability pathways for Medicaid eligibility, three codes appear to only include SSI recipients: 11, 12, and 13. These three codes account for more than half of all beneficiaries who qualify on the basis of disability. [10] Other eligibility group codes related to disability may include some SSI recipients but also include many non-recipients, such as Qualified Medicare Beneficiaries [11] (code 23) and Optional State Supplement Recipients [12] (codes 40 and 41).
Table 2. Eligibility group codes associated with disability pathways for Medicaid and CHIP eligibility
| Eligibility Group Code | Age requirement |
|---|---|
| 11: Individuals Receiving SSI a 12: Aged, Blind and Disabled Individuals in 209(b) States a 13: Individuals Receiving Mandatory State Supplements a 15: Institutionalized Individuals Continuously Eligible Since 1973 a 16: Blind or Disabled Individuals Eligible in 1973 a 17: Individuals Who Lost Eligibility for SSI/SSP Due to an Increase in OASDI Benefits in 1972 a 18: Individuals Who Would be Eligible for SSI/SSP but for OASDI COLA increases since April, 1977 a 19: Disabled Widows and Widowers Ineligible for SSI due to Increase in OASDI a 20: Disabled Widows and Widowers Ineligible for SSI due to Early Receipt of Social Security a 22: Disabled Adult Children a 23: Qualified Medicare Beneficiaries a 25: Specified Low Income Medicare Beneficiaries a 26: Qualifying Individuals a 37: Aged, Blind or Disabled Individuals Eligible for but Not Receiving Cash Assistance 38: Individuals Eligible for Cash Assistance except for Institutionalization 39: Individuals Receiving Home and Community Based Services under Institutional Rules 40: Optional State Supplement Recipients - 1634 States, and SSI Criteria States with 1616 Agreements 41: Optional State Supplement Recipients - 209(b) States, and SSI Criteria States without 1616 Agreements 42: Institutionalized Individuals Eligible under a Special Income Level 43: Individuals participating in a PACE Program under Institutional Rules 44: Individuals Receiving Hospice Care 46: Poverty Level Aged or Disabled 51: Individuals Eligible for Home and Community-Based Services 52: Individuals Eligible for Home and Community-Based Services – Special Income Level 59: Medically Needy Aged, Blind or Disabled b 60: Medically Needy Blind or Disabled Individuals Eligible in 1973 b | If age < 65 years old |
| 21: Working Disabled under 1619(b) a 24: Qualified Disabled and Working Individuals a 45: Qualified Disabled Children under Age 19 47: Work Incentives Eligibility Group 48: Ticket to Work Basic Group 49: Ticket to Work Medical Improvements Group 50: Family Opportunity Act Children with Disabilities 69: Individuals with Mental Health Conditions (expansion group) | Any age |
a Eligibility group code represents Medicaid mandatory coverage
b Eligibility group code represents optional Medicaid medically needy coverage
States must assign every Medicaid and CHIP beneficiary to one of the 72 eligibility groups, even if they meet the qualifications of more than one group. In cases where an applicant meets the qualifications of more than one eligibility group, Medicaid enrollment staff may decide the pathway through which the person gains eligibility. For example, a person who meets SSI eligibility criteria but has not already applied for SSI benefits or must go through the lengthy SSI appeals process may find it easier and/or quicker to obtain Medicaid coverage through an age- or income-based pathway. It may also be difficult to distinguish SSI recipients in states that rely on Medicaid eligibility groups that contain both recipients and non-recipients of SSI payments. For example, states may include SSI recipients and/or non-recipients in the Optional State Supplement Recipients eligibility groups (codes 40 and 41).
There are two other variables in TAF that could be used to identify SSI receipt among Medicaid beneficiaries: SSI indicator and SSI status code. States use information collected in their eligibility and enrollment systems to populate these fields, but they are not necessarily part of the Medicaid eligibility determination process. It may be possible for individuals to be correctly coded in these fields as receiving SSI even if they qualify for Medicaid through a non-SSI eligibility pathway. For example, there could be cases where beneficiaries are identified as SSI recipients using SSI indicator or SSI status code and also identified as part of an Optional State Supplement Recipients eligibility group, which includes both SSI recipients and non-recipients.
To understand the usability of eligibility group code and the two SSI variables in identifying Medicaid beneficiaries who receive SSI payments, we measure the extent to which the SSI population counts in TAF align with an external benchmark, the SSI Annual Statistical Report published by the SSA.
The SSI Annual Statistical Report contains the number of all SSI beneficiaries, not all of which are enrolled in Medicaid. This is especially true in 209(b) states where Medicaid eligibility may only to apply to a subset of SSI recipients. In SSI criteria states, we would expect the number of SSI recipients in the TAF to be closer to but lower than in the SSA report (in theory, all SSI recipients in those states could receive Medicaid if they applied). However, in states that use 1634 agreements, we would expect very close alignment between the TAF and SSA numbers of SSI recipients.
The SSI program also supports \""aged\"" individuals from age 65. In this analysis, we focus on SSI recipients under age 65 because we can better disentangle eligibility based on disability versus eligibility based on age.
SSA. Accessible at: https://www.ssa.gov/disability/determination.htm .
SSA. Accessible at: https://www.ssa.gov/disability/Documents/Factsheet-AD.pdf .
SSA. Accessible at: https://www.ssa.gov/ssi/text-appeals-ussi.htm .
Continuing disability reviews occur more frequently if the person’s medical condition is expected to improve or other circumstances change (for example, the person returns to work or experiences an increase in wages).
MACPAC. Accessible at: https://www.macpac.gov/subtopic/people-with-disabilities/ .
The 1972 amendments to the Social Security Act are accessible at: https://www.govinfo.gov/content/pkg/STATUTE-86/pdf/STATUTE-86-Pg1329.pdf .
Historically, states reported the basis of eligibility in the legacy MSIS in two fields populated with the Maintenance Assistance Status (MAS) and Basis of Eligibility (BOE) codes. These codes were combined in T-MSIS but are no longer required fields. Although MAS/BOE may continue to be reported, fewer states are reporting this data element over time. In place of MAS and BOE, CMS developed a new coding system for classifying eligibility, known as the Eligibility Group, which is the focus of this assessment.
For a full description of the eligibility groups, see the T-MSIS Data Dictionary Appendices, Version 2.4, Appendix F, p. 67, at https://www.medicaid.gov/medicaid/data-systems/downloads/tmsis-data-appendices.docx .
Using 2019 TAF data, an average of 56 percent of beneficiaries under age 65 with disability-related eligibility group codes have one of the SSI eligibility codes (11, 12, or 13); however, the proportion attributed to codes 11, 12, and 13 varies from 2 percent (in Illinois) to 86 percent (in Missouri).
Medicaid.gov. Accessible at: https://www.medicaid.gov/resources-for-states/downloads/macpro-ig-qualified-medicare-beneficiaries.pdf .
Medicaid.gov. Accessible at: https://www.medicaid.gov/resources-for-states/downloads/macpro-ig-optional-state-supplement-beneficiaries.pdf .
The SSI program also supports \""aged\"" individuals from age 65. In this analysis, we focus on SSI recipients under age 65 because we can better disentangle eligibility based on disability versus eligibility based on age.
SSA. Accessible at: https://www.ssa.gov/disability/determination.htm .
SSA. Accessible at: https://www.ssa.gov/disability/Documents/Factsheet-AD.pdf .
SSA. Accessible at: https://www.ssa.gov/ssi/text-appeals-ussi.htm .
Continuing disability reviews occur more frequently if the person\u2019s medical condition is expected to improve or other circumstances change (for example, the person returns to work or experiences an increase in wages).
MACPAC. Accessible at: https://www.macpac.gov/subtopic/people-with-disabilities/ .
The 1972 amendments to the Social Security Act are accessible at: https://www.govinfo.gov/content/pkg/STATUTE-86/pdf/STATUTE-86-Pg1329.pdf .
Historically, states reported the basis of eligibility in the legacy MSIS in two fields populated with the Maintenance Assistance Status (MAS) and Basis of Eligibility (BOE) codes. These codes were combined in T-MSIS but are no longer required fields. Although MAS/BOE may continue to be reported, fewer states are reporting this data element over time. In place of MAS and BOE, CMS developed a new coding system for classifying eligibility, known as the Eligibility Group, which is the focus of this assessment.
For a full description of the eligibility groups, see the T-MSIS Data Dictionary Appendices, Version 2.4, Appendix F, p. 67, at https://www.medicaid.gov/medicaid/data-systems/downloads/tmsis-data-appendices.docx .
Using 2019 TAF data, an average of 56 percent of beneficiaries under age 65 with disability-related eligibility group codes have one of the SSI eligibility codes (11, 12, or 13); however, the proportion attributed to codes 11, 12, and 13 varies from 2 percent (in Illinois) to 86 percent (in Missouri).
Medicaid.gov. Accessible at: https://www.medicaid.gov/resources-for-states/downloads/macpro-ig-qualified-medicare-beneficiaries.pdf .
Medicaid.gov. Accessible at: https://www.medicaid.gov/resources-for-states/downloads/macpro-ig-optional-state-supplement-beneficiaries.pdf .
We examined non-dummy enrollment records [13] in the TAF annual Demographic and Eligibility (DE) file. [14] To align with the benchmark data source, we identified SSI recipients in TAF during December of the given year. We limited the analysis to beneficiaries less than age 65, then tabulated the number of SSI recipients using each of three TAF variables as described in Table 3.
Table 3. Identification of SSI recipients in TAF
| Variable label | Variable name | Values to identify SSI recipients |
|---|---|---|
| Eligibility group code (month 12) | ELGBLTY_GRP_CD_12 | 11, 12, 13 a |
| SSI indicator (month 12) | SSI_IND_12 | 1 (Yes) |
| SSI status code (month 12) | SSI_STUS_CD_12 | 001 (SSI) or 002 (SSI eligible spouse) |
a Eligibility group codes 11, 12, and 13 should only be used for SSI recipients. We did not include beneficiaries in other eligibility groups that could contain SSI recipients because we could not distinguish them from non-recipients.
To construct the benchmark, we used the Social Security Administration (SSA) SSI Annual Statistical Report, which contains information on the number of SSI recipients by age and state in December for a given year. [15] We summed the “under 18” and “18-64” age groups to determine the number of SSI recipients less than age 65.
We first assessed data quality for each of the three TAF measures based on how well the number of SSI recipients aligned with the SSI Annual Statistical Report (Table 4). Then, the overall data quality assessment for each state was assigned based on the measure with the lowest data quality concern. For example, states in which any of the three measures were less than 10 percent different from the benchmark were classified as low concern, because there is at least one data element in TAF that can accurately identify SSI recipients with disabilities.
Table 4. Criteria for DQ assessment of ability to identify SSI recipients
| Percent difference between TAF and SSI Annual Statistical Report | Level of alignment | DQ assessment |
|---|---|---|
| x < 10 percent | High | Low concern |
| 10 percent ≤ x < 20 percent | Moderate | Medium concern |
| 20 percent ≤ x < 50 percent | Low | High concern |
| x ≥ 50 percent | Very low | Unusable |
We excluded DE records representing beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state (those with MISG_ELGBLTY_DATA_IND = 1).
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
The SSI Annual Statistical Report is available at https://www.ssa.gov/policy/docs/statcomps/ssi_asr/index.html . In the 2020 report, this information was available in Table 10.
We excluded DE records representing beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state (those with MISG_ELGBLTY_DATA_IND = 1).
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
The SSI Annual Statistical Report is available at https://www.ssa.gov/policy/docs/statcomps/ssi_asr/index.html . In the 2020 report, this information was available in Table 10.
One of the major pathways to qualify for Medicaid is through having a disabling condition. Low-income individuals with significant disabling conditions often receive Supplemental Security Income (SSI), which qualifies them for Medicaid in most states. This analysis examines how well the TAF can be used to identify Medicaid beneficiaries who receive SSI payments. Specifically, we compare the TAF to an external benchmark from the Social Security Administration (SSA), the SSI Annual Statistical Report.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4191"", ""relatedTopics"": []}" 123,"{""measureId"": 123, ""measureName"": ""1115 Demonstration Identification"", ""groupId"": 1, ""groupName"": ""Enrollment Benchmarking"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-1115-Demo-ID.pdf"", ""background"": {""content"": ""Section 1115 of the Social Security Act provides the Secretary of Health and Human Services with broad authority to waive federal Medicaid requirements and allow states to make changes to their Medicaid programs as long as they are likely to promote the objectives of the Medicaid program. As of October 2022, there were 79 approved section 1115 demonstrations in 47 states and the District of Columbia. [1] These demonstrations vary greatly in size and scope. States can change policies for existing Medicaid populations, expand coverage for certain groups or benefits, authorize new types of Medicaid payments, and test other approaches. Some section 1115 demonstrations encompass most or all Medicaid beneficiaries in the states, and others focus on only a small subset of Medicaid beneficiaries or a discrete feature of the program. [2]
TAF were designed to make it easier to identify participants in all types of waiver programs, including section 1115 demonstrations and 1915(b) and 1915(c) waivers. Each beneficiary enrolled in at least one Medicaid waiver in any month during the calendar year has a record in the annual Demographic and Eligibility (DE) Waiver Supplemental file. The Waiver Supplemental file includes information about each specific waiver type (WVR_TYPE_CD) and waiver ID (WVR_ID) for up to 10 waivers under which the eligible beneficiary received services for each month during the calendar year. It also includes high-level summary data elements to represent the most recent Medicaid section 1115 demonstration type (_1115_WVR_TYPE) during the year and how many months a beneficiary was enrolled in various types of section 1115 demonstrations (Table 1).
Table 1. TAF data elements that could serve to identify participants in section 1115 demonstration programs
| TAF data element | File | How data element helps identify section 1115 demonstration IDs |
|---|---|---|
| WVR_TYPE_CD | DE (Waiver) BSF IP (Header) LT (Header) OT (Header) RX (Header) | In the DE and BSF files, this code indicates the waiver type under which the eligible beneficiary is covered and receiving services for up to 10 waivers for each month during the calendar year. In the claims files, this code indicates the waiver type under which the claim was paid. Waiver codes that correspond to section 1115 demonstrations include \""01\"" and \""21-30\"". |
| _1115_WVR_TYPE a | DE (Waiver) | This data element represents the type of section 1115 demonstration under which the beneficiary most recently received coverage, and it is not associated with a duration of enrollment (that is, the coverage could have been for any number of days during the year). A beneficiary can be identified as a section 1115 demonstration participant during the calendar year if there is a non-null value in this field. The DE Waiver Supplemental file also includes how many months a beneficiary was enrolled in various types of section 1115 demonstrations: Pharmacy Plus demonstrations (_1115_PHRMCY_PLUS_WVR_MOS), Disaster-related demonstrations (_1115_DSTR_REL_WVR_MOS), Family Planning only demonstrations (_1115_FP_ONLY_WVR_MOS), Health Insurance Flexibility and Accountability (HIFA) demonstrations (_1115_HIFA_WVR_MOS), and Other research and demonstration (_1115_OTHR_WVR_MOS). |
| WVR_ID | DE (Waiver) APL (Operating Authority) BSF IP (Header) LT (Header) MCP (Base) OT (Header) RX (Header) | In the DE and BSF files, this data element provides the specific federal waiver ID in which the beneficiary was enrolled. In the APL file, this data element is the operating authority(or authorities) through which the managed care entity receives its contract authority and specifies the federal waiver ID that authorized payment for a claim. In the claims files, this data element provides the specific federal waiver ID that authorized payment for a claim. These IDs must be the approved full federal waiver ID number assigned during the state submission and the approval process of the Centers for Medicare & Medicaid Services. |
a In the Annual DE Waiver Supplemental File, Section 1115A demonstration participation indicator (SECT_1115A_DEMO_IND_01-12) refers to a CMS Innovation Center demonstration and should not be confused with the Medicaid Section 1115(a) waiver type (_1115_WVR_TYPE).
Note: The waiver type and waiver ID data elements are available monthly in the TAF DE, with the number of waivers (up to 10) and each month (1-12) appended to the end of the data element name (for instance, WVR_ID1_01 for the first waiver ID in January, WVR_ID10_12 for the 10th waiver ID in December, and so on). For simplicity, we did not list the monthly indicators in this analysis because we used all months of data. A list of valid values and descriptions of these data elements is available in the TAF Demographic and Eligibility Codebook at https://www2.ccwdata.org/web/guest/data-dictionaries .
Although the waiver type code is currently designed to identify 11 different types of section 1115 demonstrations, TAF users are discouraged from relying upon it to identify participants in specific types of section 1115 demonstrations. Many states code all their section 1115 demonstrations as code 01 (representing the non-specific \""Other 1115(a) Medicaid research and evaluation demonstrations\""), and several waiver type codes are not reported by any state. [3] Instead, TAF users are encouraged to use the waiver ID data element to identify Medicaid beneficiaries participating in specific section 1115 demonstrations.
This data quality assessment compares the waiver IDs representing section 1115 demonstrations found in TAF with administrative data that states report to the Centers for Medicare & Medicaid Services (CMS) on active section 1115 demonstrations in each state. We confirm whether the full federal demonstration number assigned to each unique section 1115 demonstration is accurately reported in TAF. We also identify when a state reports a waiver ID in T-MSIS that couldn’t be matched with the set of active section 1115 demonstration numbers in the CMS administrative data.
Centers for Medicare & Medicaid Services. \""State Waiver List.\"" 2021. https://www.medicaid.gov/medicaid/section-1115-demo/demonstration-and-waiver-list/index.html . Accessed October 12, 2022.
Medicaid and CHIP Payment and Access Commission. \""Section 1115 Demonstration Budget Neutrality.\"" December 2021. https://www.macpac.gov/wp-content/uploads/2021/12/Section-1115-Demonstration-Budget-Neutrality.pdf . Accessed October 12, 2022.
The other 10 section 1115 waiver type codes are 21 - 1115 Health Insurance Flexibility and Accountability (HIFA) demonstration; 22 - 1115 Pharmacy demonstration; 23 - 1115 Disaster-related demonstration; 24 - 1115 Family planning demonstration; 25 - 1115 Substance use demonstration; 26 - 1115 Premium Assistance demonstration; 27 - 1115 Beneficiary engagement demonstration; 28 - 1115 Former foster care youth from another state; 29 - 1115 Managed long term services and support; and 30 - 1115 Delivery system reform.
Centers for Medicare & Medicaid Services. \""State Waiver List.\"" 2021. https://www.medicaid.gov/medicaid/section-1115-demo/demonstration-and-waiver-list/index.html . Accessed October 12, 2022.
Medicaid and CHIP Payment and Access Commission. \""Section 1115 Demonstration Budget Neutrality.\"" December 2021. https://www.macpac.gov/wp-content/uploads/2021/12/Section-1115-Demonstration-Budget-Neutrality.pdf . Accessed October 12, 2022.
The other 10 section 1115 waiver type codes are 21 - 1115 Health Insurance Flexibility and Accountability (HIFA) demonstration; 22 - 1115 Pharmacy demonstration; 23 - 1115 Disaster-related demonstration; 24 - 1115 Family planning demonstration; 25 - 1115 Substance use demonstration; 26 - 1115 Premium Assistance demonstration; 27 - 1115 Beneficiary engagement demonstration; 28 - 1115 Former foster care youth from another state; 29 - 1115 Managed long term services and support; and 30 - 1115 Delivery system reform.
Using the TAF, [4] we examine records from the TAF annual DE Waiver Supplemental file. To identify beneficiaries covered under a section 1115 demonstration, we looked at WVR_TYPE_CD and counted the beneficiary as a section 1115 demonstration participant if there is at least one waiver type code indicating enrollment in a section 1115 demonstration (WVR_TYPE_CD_01_01 - WVR_TYPE_CD_10_12 =\""01\"" or \""21-30\"") in any month during the year. [5] For each waiver type code identified as a section 1115 demonstration, we looked up the corresponding waiver ID for the same beneficiary in the same month. We then compiled and counted all unique waiver IDs associated with a section 1115 demonstration by state.
We used an extract from the section 1115 demonstration PMDA as the benchmark data to examine the accuracy of the waiver IDs representing section 1115 demonstrations found in TAF. The PMDA system is a web-based application that allows states to submit data on section 1115 demonstrations to the Centers for Medicaid and CHIP Services (CMCS). CMCS tracks state applications for new demonstrations, amendments, and extensions and monitors post-approval demonstrations for whether states achieved desired outcomes and projected cost savings through PMDA. Although the PMDA system is the authoritative source for section 1115 demonstration administration, it has several limitations for benchmarking purposes: (1) because states report quarter or annual time periods based on their demonstration implementation dates, those periods do not necessarily align with the calendar year; (2) reports are submitted quarterly, but states often run into delays submitting data for a variety of reasons; and (3) the reports are not audited, although CMS staff review them in comparison with each state’s approved demonstration program to determine whether reporting requirements were met. Only reports CMS accepts are reflected in these benchmarks.
A benchmark data set is created by extracting the demonstration name, demonstration number, and start and end dates of demonstration performance period from the PMDA system for every section 1115 demonstration active during the calendar year. We then perform a two-part assessment for each state that compares the demonstration number for the active section 1115 demonstrations in a state in the benchmark data with the waiver ID representing section 1115 demonstrations as reported in the DE Waiver Supplemental file. First, we assessed whether every section 1115 demonstration number in the benchmark data were present for at least one beneficiary in that state’s TAF data. Second, we assessed whether there were any unexpected waiver IDs in TAF that were not present in the benchmark data. States in which expected demonstration numbers were missing from TAF, or for which waiver ID values were present in TAF that were not found in the benchmark data, were assessed as having higher levels of data quality concern (Table 2). Occasionally, a demonstration starts late in a year and might have little time to enroll beneficiaries. Regardless of the effective dates, all active section 1115 demonstrations during a calendar year are counted in this analysis. We make an exception, however, to the data quality assessment criteria for demonstrations in the first year of performance if they have a late effective date (defined as October 1 or later)— if all a state has in the PMDA system are section 1115 demonstrations with late effective date, we consider it unclassified (that is, its data quality will be assessed in the next calendar year). If a state has multiple section 1115 demonstrations and only some have a late effective date, whether we find a match in TAF for these demonstrations with late effective date does not carry any weight in the data quality assessment. [6]
Table 2. Criteria for DQ assessment of 1115 Demonstration Identification
| Proportion of unique waiver IDs present in TAF that are expected a | Number of unique waiver IDs present in TAF that are unexpected | DQ assessment |
|---|---|---|
| All | None | Low concern |
| All | Some (1 or more) | Medium concern |
| More than half but not all | Any value | Medium concern |
| Half or fewer | Any value | High concern |
| None | Any value | Unusable |
a If a state has no section 1115 demonstrations in the TAF or the PMDA system, or only “late start” demonstrations in the PMDA system, the DQ assessment is unclassified.
This analysis used the TAF data released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. More details are available on the DQ Atlas Resources page , and a crosswalk of variable names are available in the guide called Production of the TAF Research Identifiable Files. https://www.medicaid.gov/dq-atlas/downloads/supplemental/9010-Production-of-TAF-RIF.pdf . Accessed October 12, 2022.
States are not reporting the specific type of section 1115 waiver code reliably. Although waiver type code cannot be used to identify the specific type of section 1115 waiver for a state, it can serve to differentiate a section 1115 waiver from other waiver programs.
There could be multiple reasons why a state has not reported an expected section 1115 demonstration number in its T-MSIS data, some of which relate to data quality (for example, a typo in reporting the demonstration number or not being able to capture enrollment in specific demonstration at all), but others might be driven by a specific policy or program context, and having a late effective date during a calendar year is only one of them.
This analysis used the TAF data released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. More details are available on the DQ Atlas Resources page , and a crosswalk of variable names are available in the guide called Production of the TAF Research Identifiable Files. https://www.medicaid.gov/dq-atlas/downloads/supplemental/9010-Production-of-TAF-RIF.pdf . Accessed October 12, 2022.
States are not reporting the specific type of section 1115 waiver code reliably. Although waiver type code cannot be used to identify the specific type of section 1115 waiver for a state, it can serve to differentiate a section 1115 waiver from other waiver programs.
There could be multiple reasons why a state has not reported an expected section 1115 demonstration number in its T-MSIS data, some of which relate to data quality (for example, a typo in reporting the demonstration number or not being able to capture enrollment in specific demonstration at all), but others might be driven by a specific policy or program context, and having a late effective date during a calendar year is only one of them.
Medicaid section 1115 demonstration waivers encourage innovation by letting states test policies that depart from existing federal rules as long as they promote the objectives of the Medicaid program and are budget neutral to the federal government. When beneficiaries participate in a section 1115 demonstration, states must report the waiver type and the waiver ID number as part of the beneficiary's eligibility information submitted in the Transformed Medicaid Statistical Information System (T-MSIS). This analysis examines how well the section 1115 demonstration IDs in the T-MSIS Analytic Files (TAF) align with an external benchmark, the state-submitted demonstration monitoring data on the Performance Metrics Database and Analytics (PMDA) system.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4211"", ""relatedTopics"": []}" 124,"{""measureId"": 124, ""measureName"": ""American Indian and Alaska Native Indicator"", ""groupId"": 3, ""groupName"": ""Beneficiary Information"", ""pdfVersionLink"": ""downloads/background-and-methods/TAF-DQ-AI-AN-Ind.pdf"", ""background"": {""content"": ""The T MSIS Analytic Files (TAF) are research-optimized data on beneficiaries enrolled in Medicaid and the Children’s Health Insurance Program (CHIP). The Annual Demographic and Eligibility (DE) file contains information on beneficiary demographic characteristics, including whether the beneficiary meets the definition of an American Indian/Alaska Native.
States use the American Indian or Alaska Native (AI/AN) indicator to report a beneficiary’s status as an AI/AN individual. The valid values for these data elements are listed in Table 1 below.
Table 1. Valid values for the AI/AN indicator in TAF
| Variable | Description | Valid Values b |
|---|---|---|
| CRTFD_AMRCN_INDN_ALSKN_NTV_IND a | “American Indian or Alaska Native” means any beneficiary defined at 25 USC 1603(13), 1603(28), or 1679(a), or who has been determined eligible as an Indian, pursuant to 42 CFR § 136.12. This means the beneficiary:
| 0 = Individual does not meet the definition of an American Indian/Alaska Native 1 = Individual meets the definition of an American Indian/Alaska Native |
a In the TAF RIF, this data element is called CRTFD_AMRCN_INDN_ALSKN_NTV_CD. Starting in 2025, this data element name will be updated in TAF and TAF RIF.
b As of February 14, 2020 the value of 2 for “Yes individual does have CDIB” (Certificate of Degree of Indian Blood) has been retired from T-MSIS and is no longer considered valid.
Members of federally recognized Indian tribes and individuals who are otherwise eligible for services from an Indian health care provider are eligible for certain Medicaid and CHIP protections related to premiums, enrollment fees, and cost sharing. [1] While AI/AN identity is self-reported on Medicaid applications, states typically verify the beneficiary’s AI/AN status before reporting the AI/AN indicator in TAF. However, each state undergoes a different verification process, which may result in variation or inconsistency in reporting the data element. For example, some states require proof of tribal membership or affiliation, evidence of receipt of services or referrals to an Indian health provider, or other types of documentation to be submitted, whereas other states may accept an individual’s self-attestation. Although states are expected to report the information that they receive on the beneficiary’s AI/AN status, some states may not submit complete information because the data were not collected or technical difficulties arose in reporting.
This data quality assessment tabulates the proportion of records in the DE file that have missing information for the certified AI/AN indicator. For additional context, this analysis also measures the extent to which state reporting of the AI/AN indicator aligns with values reported using the race/ethnicity code in TAF and with an external benchmark, the American Community Survey (ACS) 5-year estimates.
For more information about Medicaid and CHIP coverage for American Indian and Alaska Native individuals, refer to: https://www.cms.gov/marketplace/technical-assistance-resources/AIAN-health-coverage-options.pdf
For more information about Medicaid and CHIP coverage for American Indian and Alaska Native individuals, refer to: https://www.cms.gov/marketplace/technical-assistance-resources/AIAN-health-coverage-options.pdf
We examined the American Indian or Alaska Native (AI/AN) Indicator (CRTFD_AMRCN_INDN_ALSKN_NTV_IND) on non-dummy enrollment records [2] in the TAF annual Demographic and Eligibility (DE) file. [3] We tabulated proportion with missing or invalid values for AI/AN indicator and assessed data quality for the AI/AN indictor based on this measure (Table 2). We considered any value that is not 0 (Individual does not meet the definition of an American Indian/Alaska Native) or 1 (Individual meets the definition of an American Indian/Alaska Native) to be missing or invalid. In addition, if a state reported no AI/AN beneficiaries in TAF based on the AI/AN indicator, we deemed the data quality assessment to be unusable. [4]
Table 2. Criteria for DQ assessment of AI/AN Indicator
| Percentage of records with missing or invalid AI/AN indicator values | DQ assessment |
|---|---|
| x ≤ 10 percent | Low concern |
| 10 percent < x ≤ 20 percent | Medium concern |
| 20 percent < x ≤ 50 percent | High concern |
| x > 50 percent or no AI/AN beneficiaries reported in TAF | Unusable |
We also calculated two sets of contextual measures to investigate the extent to which state reporting of American Indian and Alaska Native individuals using the AI/AN indicator aligns with (1) reporting of the condensed race/ethnicity code (RACE_ETHNCTY_FLAG) in TAF and (2) American Community Survey (ACS) data.
It is expected that for most cases, the value reported in the AI/AN indicator variable should align with an individual’s race/ethnicity reported in TAF (a race/ethnicity code value of 4 in TAF indicates American Indian/Alaska Native). However, the AI/AN indicator variable does not separate beneficiaries based on ethnicity or multiracial identity, whereas the AI/AN value for the combined race/ethnicity variable in TAF specifies non-Hispanic, mono-racial AI/AN individuals. In addition, while both the AI/AN indicator and race/ethnicity rely primarily on self-identification, states often require supplemental verification when reporting the AI/AN indicator. Therefore, it is expected that there will be some level of expected misalignment between AI/AN indicator and race/ethnicity in TAF. For context, we calculated and displayed: (1) the percentage of beneficiaries reported as AI/AN based on the AI/AN indicator value without a corresponding race/ethnicity value, (2) the percentage of beneficiaries reported as AI/AN based on the race/ethnicity code value without a corresponding AI/AN indicator value, and (3) the overall percentage of all TAF beneficiary records for which either of the two previously described scenarios applied. Given the known differences between the AI/AN indicator and race/ethnicity variables, these measures are displayed for context only to help TAF users explore additional potential data quality issues and are not included in calculation of the data quality assessment.
We also compared the AI/AN beneficiary counts using the AI/AN indicator in TAF to an external benchmark, the ACS. We used the ACS 5-year estimates [5] Public Use Microdata Sample (PUMS) [6] for a given year. [7] The ACS data, which are collected annually from a nationally representative random sample of households, contains information on self-reported race, ethnicity, and health insurance coverage. After pulling the ACS microdata from PUMS, we selected all individuals who reported having “Medicaid, Medical Assistance, or any kind of government-assistance plan for those with low incomes or a disability” at the time of the survey. For individuals in that group, we calculated the percentage who were AI/AN (Table 3). We did not subset this population by ethnicity.
ACS data is used by many stakeholders, including by federal and state government agencies for policy and program funding activities, and are considered a highly reliable source of demographic data. There are, however, several reasons for there to be a discrepancy in AI/AN identification between ACS and TAF data. The ACS data do not include individuals who identify with two or more races in the AI/AN count. Self-reporting of health insurance coverage and AI/AN status for the ACS can result in an undercount or overcount of the number of AI/AN Medicaid beneficiaries who appear in administrative data, whereas reporting of AI/AN indicator in TAF often relies on verification of an individual’s AI/AN status. In this analysis, we compare the percentage of Medicaid beneficiaries with AI/AN indicator value of 1 in TAF to the comparable distribution in the ACS. Given the known differences between the TAF and ACS definitions of AI/AN, we presented these measures for context only and did not include them in the calculation of the data quality assessment.
Table 3. AI/AN categories in TAF and ACS
| AI/AN indicator value in TAF | Race/ethnicity code value in TAF a | Race variables in the ACS |
|---|---|---|
| 1 = Individual meets the definition of an American Indian/Alaska Native | 4 = American Indian and Alaska Native (AIAN), non-Hispanic |
|
a For the race/ethnicity flag TAF, a “non-Hispanic” value may indicate that (1) the beneficiary is not Hispanic or (2) the beneficiary’s ethnicity or Hispanic status is not reported.
We excluded DE records representing beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state (those with MISG_ELGBLTY_DATA_IND = 1).
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide “Production of the TAF Research Identifiable Files.”
All states included in the ACS have a non-zero number of AI/AN individuals who also identified as a Medicaid enrollee. Therefore, if a state reports zero AI/AN beneficiaries in TAF, this likely indicates a data quality issue.
ACS 5-year estimates are more reliable and complete than ACS 1-year estimates and the Current Population Survey, as it includes smaller geographic areas and has a larger sample size. For more details, see: https://www.census.gov/programs-surveys/acs/guidance/estimates.html and https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2677056/
For example, to compare against the 2022 TAF data, we used the ACS 5-year estimates for 2022 that are based on data collected between 2018 and 2022.
We excluded DE records representing beneficiary IDs that are present on claims but were not included in the eligibility records submitted by the state (those with MISG_ELGBLTY_DATA_IND = 1).
This analysis used the TAF data that were released as TAF Research Identifiable Files (RIF). During the transformation into RIF, some TAF data elements were suppressed, changed, or renamed. Additional details are available on the DQ Atlas Resources page , and a crosswalk of variable names can be found in the guide \u201cProduction of the TAF Research Identifiable Files.\u201d
All states included in the ACS have a non-zero number of AI/AN individuals who also identified as a Medicaid enrollee. Therefore, if a state reports zero AI/AN beneficiaries in TAF, this likely indicates a data quality issue.
ACS 5-year estimates are more reliable and complete than ACS 1-year estimates and the Current Population Survey, as it includes smaller geographic areas and has a larger sample size. For more details, see: https://www.census.gov/programs-surveys/acs/guidance/estimates.html and https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2677056/
For example, to compare against the 2022 TAF data, we used the ACS 5-year estimates for 2022 that are based on data collected between 2018 and 2022.
The TAF eligibility files include information on select demographic characteristics of beneficiaries enrolled in Medicaid or CHIP. This analysis examines the completeness of the American Indian and Alaska Native indicator information in the TAF. For additional context, this analysis also presents information on how well the TAF data on American Indian and Alaska Native indicator align with an external benchmark, the U.S. Census Bureau's American Community Survey, and race/ethnicity values in TAF.
"", ""footnotes"": []}, ""originalIssueBriefId"": ""4221"", ""relatedTopics"": []}"