Frequently Asked Questions

  ADA/508 friendly site



FAQ

Are there any Medicaid Analytic eXtract (MAX) standard sample files (e.g. 5% or 10%) that are available to researchers?

The MSIS instructions to State Medicaid agencies are that they are to report data for all Medicaid enrollees and all services in MSIS, whether provided under fee-for-service or prepaid managed care. However, full reporting of all enrollees and all services may not be a goal that is met for all States and all years, especially for “encounter” records of services provide by prepaid managed care plans. Nevertheless, the MAX data are a very rich data set that allows users to conduct research for a wide variety of Medicaid enrollee groups and Medicaid covered services. The possibilities are quite diverse, including: nursing home residents, dual enrollees with full Medicaid benefits, persons with HIV/AIDS, foster care children, Medicare ESRD enrollees in Medicaid, low birth-weight newborns, persons with injuries, Medicaid CHIP (M-CHIP) expansions, and more. A simple random sample would include a very small numbers of records for many important Medicaid population-based and service-based research questions, for many states. This is an extremely important issue in Medicaid because of the diversity across state programs. The small number of cases would not provide users with the estimating power they desire. For these reasons, CMS has not created standard MAX sample files in the past. Under the MAX-PDQ contract, a feasibility study for creating a standard MAX sample, known as Mini-MAX, has been completed. We expect that CMS will design and create Mini-MAX in the near future. Until then, starting with the MAX “universe” files, users may want to devise a sampling strategy of their own to select a representative sample tailored to their specific research requirements. However, users should carefully determine sample selection criteria in view of the quality of data reporting for data elements that are key to their research. The MAX Validation Tables are an excellent tool for users in this process.
(FAQ2459)

Was this answer helpful?