Background: Significant policy efforts have been implemented to lower the out-of-pocket payments (OOP) for insulin to lower access barriers and enhance adherence. We employed a machine learning approach to explore the relationship between OOP and insulin adherence among Medicare beneficiaries.
Methods: 48,613 Medicare beneficiaries with continuous use of long-acting insulin analog were identified from Medicare Fee-for-Service data (2017-2019). People covered by low-income subsidy programs and Medicaid were excluded. Adherence was measured by the proportion of days covered (PDC) by insulin in each year, and monthly OOP for a 30-day supply of insulin was estimated at a plan level. We used a localized fixed-effect model with recursive partitioning to depict the local response of OOP changes on PDC changes at different levels of baseline OOP.
Results: For individuals with monthly OOP exceeding $150, a 1.51% increase in PDC (95%CI: 1.03% to 1.99%) with every $10 reduction was observed, and for those with monthly OOP between $75 and $150, every $10 reduction correlated with a 0.69% increase in PDC (95%CI: 0.43% to 0.94%). For those with monthly OOP below $75, the further reduction did not impact the PDC.
Conclusion: The price elasticity of demand for insulin is non-linear. Reducing monthly OOP for insulin to levels lower than the cut points is vital to remove the financial barrier to adherence and ensure access to proper care.
D. Guan: None. P. Li: None. T. Jiao: None. X. Hu: Research Support; PhRMA Foundation. Y. Zhang: None. P.K. Chehal: None. J. Lee: None. K. Narayan: None. M.K. Ali: Advisory Panel; Eli Lilly and Company. H. Shao: Consultant; Eli Lilly and Company.
