In this data quality report, we investigate Medicaid spending data on people dually enrolled in Medicare and Medicaid (hereafter “dual enrollees”) in the Transformed Medicaid Statistical Information System (T-MSIS) Analytic Files (TAF). For select states, we investigate capitation payments among Medicaid managed-care plan enrollees by type of comprehensive plan, including the Financial Alignment Initiative’s MedicareMedicaid Plans, and by type of more limited benefit plans, including behavioral health plans, long-term care plans, and transportation service plans. For those who receive benefits on a fee-for-service basis, we investigate Medicaid spending for specific services for which Medicaid is typically the primary payer among dual enrollees: nursing home care; behavioral health services; and long-term services and supports, such as personal care, nonemergency transportation, and other home- and community-based services. Our analysis produced several key findings.
- Analyzing Medicaid managed-care plan capitation payments, we find that
- most managed-care plan enrollees had corresponding capitation claims with a positive payment amount, which is necessary for the data to be usable (i.e., missing or negative amounts are unusable);
- the most common data quality issue is a lack of corresponding capitation claims for certain combinations of enrollees and plan types; and
- the proportion of enrollees whose plans had no capitation claim was almost 100 percent for some plan types and states, making the data unusable, and was very moderate (e.g., less than 1 percent) in other plan types and states, leading to few data quality concerns.
- Analyzing Medicaid fee-for-service noncrossover (meaning no Medicare responsibility) claim payments, we find that
- most states’ claims had positive Medicaid payments, which are necessary for the data to be usable;
- the most common data quality issue is claim lines with a reported $0 Medicaid payment amount; and » data quality across the services studied varies significantly within states, though Montana and Iowa had consistently high shares of $0 payment amounts across four of the five services studied.
Taken together, this analysis shows Medicaid spending contained in the TAF can be used for various analyses of care delivered to dual enrollees. However, given the significant state-by-state variation in data quality by type of service, researchers will need to use these results on a case-bycase basis to determine the quality of the data for a given state-specific application. Further, the data will not support national studies without exclusions.