OBJECTIVES: The Covid-19 pandemic highlighted the importance of considering Social Determinants of Health (SDoH) in healthcare research. Administrative claims databases are widely used for research, but often lack SDoH data or sufficient transparency in how these data were obtained. This study describes innovative methods for integrating SDoH data with administrative claims to facilitate health equity research.

METHODS: The HealthCore Integrated Research Database (HIRD) contains medical and pharmacy claims from a large, national US payer starting in 2006 and includes commercial (Comm), Medicare Advantage (MCare), and Medicaid (MCaid) populations. The HIRD includes individually identifiable information, which was used for linking with SDoH data from the following sources: national neighborhood-level data from the American Community Survey, the Food Access Research Atlas, and the National Center for Health Statistics’ urbanicity classification; and member-level data on race/ethnicity from enrollment files, medical records, self-attestation, and imputation algorithms. We examined SDoH metrics for members enrolled as of 05-July-2022 and compared them to the respective US national data using descriptive statistics. We also examined telehealth utilization in 2022.

RESULTS: SDoH data were available for ~95% of currently active members in the HIRD (Comm/MCare/MCaid 12.5m/1m/7.6m). Socioeconomic characteristics at the neighborhood-level differed by membership type and vs. national data: % of members with at least a high-school education (90/88/84 vs. 87); median family income ($98k/$76k/$70k vs. $82k); % of members living in low-income low-food-access tracts (9/14/18 vs. 13); urban (57/52/47 vs. 61). At the member-level, the % of White Non-Hispanics, Black Non-Hispanics, Asian Non-Hispanics, and Hispanics were 61/6/5/6 (Comm), 76/12/2/2 (MCare), and 45/26/5/19 (MCaid). Imputation contributed 15-60% of race/ethnicity values across membership types. Telehealth utilization increased with socioeconomic status.

CONCLUSIONS: We successfully integrated SDoH data from a variety of sources with administrative claims. SDoH characteristics differed by type of insurance coverage and were associated with differences in telehealth utilization.