Using prescription drug data for timely assessments of state insurance coverage rates: a validation study.

Introduction: Timely assessments of insurance coverage are limited by lags and unpredictable data availability. We assessed the performance of real-time prescription data as a proxy for state-level coverage changes. Methods: We analyzed correlations between quarterly state insurance coverage (Medicaid, private insurance, and uninsured) and counts of filled prescriptions per capita by payer (Medicaid, private insurance, and cash-pay/assistance programs) from 2013 to 2024. Regression models measured how predictive prescription counts were of within-state coverage changes. Results: Medicaid prescriptions per capita were strongly correlated with Medicaid coverage (ρ = 0.62). Cash-pay prescriptions per capita were moderately correlated with uninsured rates (ρ = 0.49). Private coverage correlations were weak. Correlations were stronger for adults (ρ = 0.74 for Medicaid, ρ = 0. 52 for uninsured) than children. Among adults, each 10% within-state increase in Medicaid prescriptions was associated with a 7.5% relative increase (95% confidence interval [CI], 5.7%-9.3%) in Medicaid coverage; each 10% increase in cash-pay prescriptions was associated with a 2.5% increase (95% CI, 1.0%-4.1%) in uninsured rates. Conclusions: Real-time prescription drug data are moderately to highly correlated with state-level coverage changes in Medicaid and uninsured rates, particularly for adults. These data may be useful for timely assessments of upcoming policies affecting Medicaid coverage.