Electronic Decision Support for Deprescribing in Older Adults Living in Long-Term Care: A Stepped-Wedge Cluster Randomized Trial.

Importance: Potentially inappropriate prescribing (PIP) is common, costly, and harmful. Deprescribing potentially inappropriate medications (PIMs) is a priority for improving the health outcomes of older adults. PIP is especially common in long-term care homes, with up to 88% of residents affected. Objective: To assess the efficacy of electronic decision support for deprescribing in long-term care. Design, Setting, and Participants: This stepped-wedge cluster randomized trial took place from August 1, 2021, to October 31, 2022, during the COVID-19 pandemic. The study assessed older adults residing in 1 of 5 long-term care homes in New Brunswick, Canada, at the start of the study who were prescribed 1 or more PIMs. The 5 long-term care homes were divided into 3 clusters. All clusters spent at least 3 months in a control phase; every 3 months a cluster was randomized to enter the intervention phase. Data analysis was performed from October 15, 2023, to March 24, 2025. Interventions: Electronically generated, individualized reports that contained prioritized opportunities for deprescribing in older adults were paired with preexisting quarterly medication reviews. Deprescribing reports were accessed through a secure viewer. Main Outcomes and Measures: The primary outcome was the proportion of residents with 1 or more PIMs deprescribed in the control phase vs intervention measured every 3 months after a medication review. For the primary outcome, an adjusted odds ratio (AOR) was calculated using a generalized linear model with a logit link, controlling for the effect of the intervention and adjusted for the number of PIMs, age, sex, language, and period as fixed effects and participants nested within sites as random effects. Results: A total of 725 residents participated in the study (median [IQR] age, 84 [76-90] years; 478 [65.9%] female). The median (IQR) number of medications was 10 (7-13), and the median (IQR) number of PIMs was 3 (2-4). In the control phase, the proportion of residents with 1 or more PIMs deprescribed was 92 of 725 (12.7%) compared with 226 of 621 (36.4%) during the intervention (AOR, 1.58; 95% CI, 1.07-2.34), in favor of the intervention. Conclusions and Relevance: This study found that electronic decision support paired with the usual workflow could render the deprescribing process scalable and effective. These results suggest that medication reviews should incorporate deprescribing as part of usual care. Trial Registration: ClinicalTrials.gov Identifier: NCT04762303.