Brigham and Women’s Hospital and Harvard Medical School, United States
Background: Reductions in high-risk medication use in older adults has been shown to help decrease medication-related risks, but little is known about the relative rates of discontinuing long-term high-risk medication across sociodemographic groups of older adults.
Objectives: Our goal was to examine associations between sociodemographic factors and the discontinuation of high-risk medications in older adults.
Methods: We conducted a retrospective cohort study using administrative claims data from 2017 to 2023, including adults aged ≥65 years enrolled in a large national U.S. health insurer who were recently classified as long-term users of high-risk medications. The primary exposure variables included database-recorded age, gender, ethnicity, and race (measured separately). The primary outcome was discontinuation of high-risk medications for ≥90 days and no subsequent fills. We used multivariable Cox regression to estimate the association between sociodemographic characteristics and time until discontinuation, adjusting for patient-level clustering and baseline clinical and healthcare utilization covariates. We repeated analyses by individual medication class and added interaction terms to assess whether associations with sociodemographic variables varied across subgroups.
Results: Across 729,705 eligible older adults (mean age: 74 years [SD:7], 59% female), 18% discontinued their high-risk medication (mean follow-up: 626 days). Black older adults (HR=1.07, 95%CI 1.03-1.11) were more likely to discontinue high-risk medications than their White counterparts, highest in 2020-2021. Men (HR=0.89, 95%CI 0.87-0.91) and adults aged ≥75 years (HR=0.86, 95%CI 0.84-0.91) were less likely to discontinue high-risk medications compared with women and those 65-74 years, respectively. Although these two characteristics were each associated with lower high-risk medication discontinuation rates, their combination was linked to a higher likelihood of discontinuation (HR=1.04, 95% CI: 1.02-1.06). No combinations of other sociodemographic characteristics were associated with differences in discontinuation.
Conclusions: Despite variations in sociodemographic characteristics and high-risk medication classes, discontinuation rates were relatively low. Our findings emphasize the need for effective interventions to reduce high-risk medication use in older adults, where discontinuation is recommended.
