Social Determinants of Health and Incident Type 2 Diabetes Risk: Real-World Evidence From the Houston Methodist Cardiovascular Learning Health System Registry

OBJECTIVE Modern electronic medical record (EMR) registries offer unique opportunities to assess distinct social determinants of health (SDoH) phenotypes of patients with diverse cardiovascular disease risk profiles. Although SDoH has been correlated with type 2 diabetes mellitus (T2DM) risk, little is known about the longitudinal impact of aggregate social burden on new-onset T2DM using real-world EMR data in a contemporary, integrated, health system setting. RESEARCH DESIGN AND METHODS We conducted a retrospective cohort study of patients aged 18 years or older from the 2016–2023 Houston Methodist Cardiovascular Learning Health System (HM-CV-LHS) registry. SDoH domains included difficulty affording medication, food insecurity, transportation barriers, and mental health, aggregated to create a composite index and categorized into high, low, and optimal risk. Cox proportional hazards models were used to predict the risk of new-onset T2DM in those without diabetes at baseline, based on individual and aggregate SDoH burden. RESULTS Among the 1.2 million patients in the HM-CV-LHS registry, 156,677 with available data on all four SDoH domains were included in the final analytic sample. During mean follow-up of 3.5 years (7 years total), 8% developed new-onset T2DM. T2DM incidence exhibited a consistent, dose-response relationship with increasing SDoH burden. In models fully adjusted for traditional risk factors, patients experiencing the highest SDoH burden were nearly twice as likely to experience incident diabetes compared with those with an optimal social profile (adjusted hazard ratio 1.95; 95% CI 1.81, 2.11). CONCLUSIONS SDoH are robust, independent predictors of T2DM. Proactive incorporation of SDoH data into patient care could optimize social interventions to prevent T2DM risk and reduce health disparities.