The goal of this internal EHealth project was to evaluate whether using Medicaid data with missing race/ethnicity information results in biased estimates of racial/ethnic disparities in health. Medicaid data are frequently used to investigate racial and ethnic disparities in health and health care. However, there is considerable variation in the completeness of race/ethnicity in Medicaid data across US states. Missing race/ethnicity data impacts our ability to accurately understand population health and respond to health disparities.

Medicaid enrollment data (T-MSIS) was linked with a variety of Census Bureau data sources (e.g., Decennial Census, American Community Survey). This type of data linkage allowed us to populate race/ethnicity information for beneficiaries missing this information and to validate race/ethnicity information in both data sources.

Initial findings from this project suggest that linking Medicaid enrollment data and Census Bureau records is both a feasible and effective tool for filling in missing race/ethnicity information for Medicaid beneficiaries and can help to advance our understanding of population health. Furthermore, this project demonstrates the significant potential of using Census Bureau data to complement existing health records that may lack demographic characteristics, including race/ethnicity. We are currently working to disseminate our findings. 

Below is an Excel document containing disclosed state-level data from SEHSD Working Paper 2023-05 for ease of use. The first sheet is a Readme providing information regarding data layout. The second sheet contains the actual data and can be used in Excel or imported into another statistical software.