Integrating healthcare system context to improve risk prediction and assess racial disparities among dual-eligible Medicare–Medicaid beneficiaries: a retrospective cohort study using national fee-for-service claims

Objectives Medicare–Medicaid dual-eligible beneficiaries comprise <15% of enrollees but account for disproportionate spending. Current models used to predict future spending rely on patient demographics and diagnoses, potentially missing healthcare system context (provider availability, care fragmentation, facility characteristics). We estimated the degree to which healthcare context data improve risk prediction and what portion of risk and risk disparities among racial/ethnic groups is predicted by such data. Design Retrospective cohort study with 6-month feature window (January–June 2023) and 6-month outcome window (July–December 2023). Setting National fee-for-service Medicare programme across 50 US states, District of Columbia and Puerto Rico. Participants 3 877 563 dual-eligible beneficiaries with ≥8 months enrolment in Medicare Parts A, B and D during 2023. Mean age 64.9 years; 56% female; 65% White, 15% Black, 7% Hispanic and 6% Asian. Outcome measures Primary: prospective Medicare spending (per-member-per-month) and acute care utilisation (≥1 hospitalisation/emergency department visit) July–December 2023. Secondary: concurrent spending January–June 2023. Results Adding healthcare system context to patient data improved model performance from R²=0.454 (0.450–0.458) to 0.615 (95% CI 0.611 to 0.619) for prospective spending prediction (a 35% improvement, p<0.001). The sensitivity of predicting acute care visits improved from 25.0% (24.2–25.8%) to 33.8% (32.9–34.7%, p<0.001), while maintaining specificity (>97%). System context reduced predicted risk for 86% of beneficiaries (mean −8.8 percentage points), while increasing risk for 14%. Black beneficiaries experienced higher rates of contextual risk amplification (15.8% vs 13.6% White) and showed distinct vulnerability to large practice coordination and workforce composition factors, while Hispanic populations’ risk was primarily associated with geographic access barriers. Conclusions Healthcare delivery context substantially improves prediction of future spending among dual-eligible beneficiaries. These findings suggest risk adjustment incorporating context should account for differential racial impacts.