State Insulin Copay Cap Laws and Working-Age Diabetes Mortality: A Difference-in-Differences Analysis

Abstract

This paper estimates the causal effect of state-level insulin copay cap laws on diabetes mortality among working-age adults (25–64) in the United States. Prior analyses using all-ages mortality data found null effects, but these results were mechanically driven by outcome dilution: copay caps affect only commercially insured insulin users, who represent roughly 3% of all-ages diabetes decedents but approximately 15–20% of working-age decedents. I construct a state-year panel of age-restricted diabetes mortality (ICD-10 E10–E14) from NCHS vital statistics and CDC provisional mortality data for 1999–2017 and 2020–2023 (with a two-year gap in 2018–2019). Exploiting the staggered adoption of copay cap laws across twenty-six states using the difference-in-differences estimator, I find that the aggregate effect of insulin copay caps on working-age diabetes mortality is not statistically significant over the short post-treatment horizon available in current data (1–4 years post-adoption for the earliest cohorts), though one cohort (2023 adopters) shows a significant negative estimate that warrants monitoring as post-treatment data accumulate. However, the working-age specification substantially reduces outcome dilution—the treated population share rises from $s \approx 3%$ (all-ages) to $s \approx 15$–$20%$ (working-age)—yielding minimum detectable effects that are three to five times smaller and bringing plausible treatment effects within the range of statistical detectability. Event-study estimates show no evidence of differential pre-trends, and HonestDiD sensitivity analysis (both relative-magnitudes and smoothness/FLCI approaches) confirms robustness to plausible violations of parallel trends. Vermont sensitivity analysis (excluded, as-treated, and as-control specifications) shows results are invariant to Vermont's classification. The working-age null is more informative than the all-ages null: it narrows the range of plausible treatment effects and provides a stronger test of the hypothesis that copay caps reduce diabetes mortality among the directly affected population.