OBJECTIVES: An important feature of the US healthcare system is the risk prediction and adjustment of payments for managed care organizations or accountable care organizations. Risk adjustment models use patients’ demographic characteristics and health status, and provider-related factors to adjust the risks not due to the quality of care provided. These models allow care management services to identify high-risk patients and compensate providers for the additional costs associated with high-risk patients. Recent literature provided strong support for the inclusion of Social Determinants of Health (SDoHs) in risk prediction and adjustment. This study aims to assess the impact of SDoHs on providers’ Hierarchical Conditions Categories (HCC) scores and added predictive power of HCC scores on medical cost after incorporating SDoHs.

METHODS: We link area and beneficiary-level HCC scores with area-level SDoH indicators to estimate the improvements in model accuracy after incorporating SDoH variables. Beneficiary-level HCC scores will be acquired from the CMS through RESDAC. SDoH data will be assembled through multiple sources including the Agency for Healthcare Research and Quality SDoH database, Area Health Resource Files, and various waves of the American Community Surveys.

RESULTS: We conducted preliminary analyses using publicly available data from CMS. We find providers in counties with a higher proportion of older adults tend to have higher HCC scores. However, providers in rural counties have lower HCC scores, which contradicts to the common wisdom that rural residents are less healthy. Additional analyses are in process.

CONCLUSIONS: Our proposed research has critical policy implications for Medicare, the largest health security program in the world for older adults. Incorporating SDoH in risk prediction and adjustment may have an important role in reducing health disparities.