Presentation (CTI)

OBJECTIVES: To quantify the economic burden of prescription opioid use across U.S. counties and evaluate how pharmacy benefit management (PBM) levers and social determinants of health (SDOH) jointly influence opioid-related costs.
METHODS: Ten years (2013-2023) of Medicare Part D prescribing and drug-cost data were merged with County Health Rankings SDOH indicators. A Random Forest regression model was trained to predict opioid cost per capita using unemployment, income inequality, provider availability, obesity, and other social indicators. SHAP analysis was used for interpretability. Counterfactual simulations modeled PBM interventions (e.g., formulary tightening, 3% cost-per-claim reduction, utilization guidance) and SDOH improvements (e.g., +20% mental-health providers, +10% primary care access).
RESULTS: The model achieved R² = 0.97 and RMSE ≈ $375, indicating strong predictive accuracy. Unemployment rate, mental-health-provider density, and income inequality were the leading predictors, jointly explaining ~80% of variance. Simulations revealed that PBM levers alone reduced national opioid costs by 4-6%, while integrated PBM-SDOH strategies achieved reductions up to 17%, equivalent to ≈ $23 billion annually.
CONCLUSIONS: Opioid cost burden in the U.S. is driven by both economic access barriers and PBM pricing dynamics. Combining fiscal policy levers with community-level SDOH investments yields compounding cost reductions and more equitable outcomes. This work demonstrates that integrating fiscal levers (PBM) with population-level health determinants (SDOH) provides a scalable framework for optimizing national drug spending.