The diabetes health plan and medication adherence among individuals with low incomes

Objective To test the impact of the Diabetes Health Plan (DHP), a diabetes-specific insurance plan that lowers out-of-pocket costs for diabetes-related medications and clinical visits, on adherence to oral hypoglycemic medications among low-income adults with Type 2 Diabetes (T2DM). Data Sources and Study Setting Cohort of adults (18-64) with T2DM, an annual household income <USD 30,000, and who were continuously enrolled in an employer-sponsored UnitedHealthcare plan for at least two years between 2009 and 2014. Study Design We employed a linear regression Difference-In-Differences (DID) approach with a matched comparison group. To assess for differential DHP effects across adherent versus non-adherent patients, we ran a Difference-in-Difference-in-Differences (DDD) analysis by including an interaction term that included indicators for DHP exposure status and time, and low versus high baseline medication adherence. Data Collection The analytic data set is limited to employer groups that purchased the DHP and standard benefit plans from UnitedHealthcare, had internal pharmacy contracts; complete pharmacy claims data, and sufficient medical claims and lab data to identify employees and their dependents with T2DM. Principal Findings Our DID analysis did not show improved medication adherence associated with employer DHP adoption. However, the DDD model suggested a difference between DHP-exposed and comparison beneficiaries when comparing the relative effect on individuals who were adherent versus non-adherent at baseline, as suggested by the significant three-way interaction term (10.2,p = 0.028). This effect was driven by the 8.2 percentage point increase in medication adherence for the DHP subsample that was non-adherent at baseline. Conclusions The DHP may benefit low-income patients with low baseline medication adherence. Value-based insurance design may be an important strategy for mitigating income disparities in T2DM outcomes.