Hospitalization and emergency department visits associated with potentially inappropriate medication in older adults: self-controlled case series analysis.

Abstract

Potentially inappropriate medications (PIM) and resulting adverse health outcomes in older adults are a common occurrence. However, PIM prescriptions are still frequent for vulnerable older adults. Here, we sought to estimate the risk of hospitalization and emergency department (ED) visits associated with PIM prescriptions over different exposure periods and PIM drug categories. We used the National Health Insurance Service-Elderly Cohort Database (NHIS-ECDB) to construct the cohort and implemented a Self-Controlled Case Series (SCCS) method. Hospitalization or ED visits during the exposure and post-exposure periods were compared to those during the non-exposure period, and six PIM drug categories were evaluated. A conditional Poisson regression model was applied, and the risk of outcomes was presented as the incidence rate ratio (IRR). All potential time-varying covariates were adjusted by year. A total of 43,942 older adults aged ≥65 y who had at least one PIM prescription and the events of either hospitalization or ED visits between Jan 2016 and Dec 2019 were selected.. Mean days of each exposure period was 46 d (±123); risk was highest in exposure1 (1-7 d, 37.8%), whereas it was similar during exposure2 (15-28 d), and exposure3 (29-56 d) (16.6%). The mean number of total PIM drugs administered during the study period was 7.34 (±4.60). Both hospitalization and ED visits were significantly higher in both exposure (adjusted IRR 2.14, 95% Confidence Interval (CI):2.11-2.17) and post-exposure periods (adjusted IRR 1.41, 95% CI:1.38-1.44) in comparison to non-exposure period. The risk of adverse health outcomes was highest during the first exposure period (1-14 d), but decreased gradually over time. Among the PIM categories, pain medication was used the most, followed by anticholinergics. All PIM categories significantly increased the risk of hospitalization and ED visits, ranging from 1.18 (other PIM) to 2.85 (pain medication). Sensitivity analyses using the first incidence of PIM exposure demonstrated similar results. All PIM categories significantly increased the risk of hospitalization and ED visits, with the initial period of PIM prescriptions showing the highest risk. In subgroup analysis stratified by the number of medications, PIM effects on the risk of hospitalization and ED visits remained significant but gradually attenuated by the increased number of medications. Therefore, the development of deprescribing strategies to control PIM and polypharmacy collectively is urgent and essential.