Trends in annual medication possession in older adults with type 2 diabetes

Older adults (≥65 years) make up a substantial portion of individuals with type 2 diabetes (T2D) and frequently experience multimorbidity requiring complex, costly medication regimens. Cost-related nonadherence is prevalent in this population because of fixed incomes, polypharmacy, and a historically complex Medicare Part D coverage structure with variable out-of-pocket expenses over the year, which saw significant changes in 2025. We hypothesized that, among Medicare beneficiaries, brand medication possession would decline more rapidly than generic possession over the calendar year and rebound in January of the subsequent year when coverage resets.

To examine comparative trends in medication possession for brand and generic classes of glucose-lowering medications among older adults over the calendar year stratified by insurance type.

We conducted a retrospective cohort study using electronic health record and claims data from 6 US health systems. For each year from 2015 through 2020, adults aged at least 65 years with T2D who possessed at least 1 glucose-lowering medication on January 1 (index), filled the medication at least once during the following year, and maintained continuous prescription insurance coverage for 15 months from index were included; individuals could contribute multiple observations across years and medications. Medication possession was determined on January 1 and, thereafter, based on dispense date and days supply for each glucose-lowering medication. We calculated the daily proportion of individuals in possession of each medication class over 15 months and compared trends by brand/generic classification, medication class, and insurance. Medications were classified into brand and generic status based on class availability and prescribing trends during the study period. Segmented linear regression estimated slope changes before and after day 365. Difference-in-differences-in-slope approach compared brand vs generic medications before and in the first month after the new year. Additional analyses compared individual branded classes with biguanides.

Among 88,060 older adults with T2D (240,542 medication observations), mean age for the population was 73.7 years, 46% were female, 60% were White, 83% of medications were covered by Medicare insurance, and 92% were generic classes of glucose-lowering medications. Medication possession declined uniformly during the first 3 months of each year across all insurance types, followed by steeper declines for brand vs generic classes of medications (slopes -19.6% and -7.9% per year) for the Medicare-insured cohort. Brand classes exhibited a more significant positive slope change compared with generics (18.2% and 1.3% per year; P < 0.001) in the first month of the new year (days 366-395) for Medicare-insured cohort, a pattern differing from the commercial and Medicaid cohorts. Individual brand medication possession differed significantly from biguanides. Sensitivity analyses using a 3-month post-new year segment yielded similar findings.

Older adults appear vulnerable to cost-related nonadherence, with lower year-end medication possession of brand classes of T2D medications than generics. The modest rebound in possession after coverage reset and similar end-of-year possession across insurance types suggest that the coverage gap is not the sole contributor to reduced brand medication possession.