2D-01 - Effectiveness of electronic health record tools for deprescribing in older adults: A 3-arm pragmatic database-randomized trial

Background: High-risk medications, such as benzodiazepines, sedative hypnotic Z-drugs, and strongly anticholinergic medications continue to be overprescribed in older adults. Electronic health record (EHR)-based tools have been used to promote the use of evidence-based medications but been far less commonly studied to reduce chronic medication overuse.

Objectives: We recently completed a large 16-arm adaptive trial of EHR deprescribing tools to identify those that were potentially effective. Using this as preliminary evidence, we conducted a definitive parallel group trial testing them in a different health system, a replicability step not commonly done.

Methods: In this 3-arm parallel group pragmatic trial, we cluster randomized 201 primary care providers (PCPs) in a large health system (1:1:1) to receive: 1) “pre-commitment” EHR tools that triggered when PCPs opened patient encounters and used a two-visit approach of reviewing risks and subsequently encouraging deprescribing, 2) “boostering” EHR tools that triggered when PCPs opened patient encounters and provided a subsequent reminder four weeks later, and 3) usual care (no tool). We identified their eligible patients ≥65 years of age chronically prescribed these high-risk medications who had a PCP visit from EHR data for 16 months (ending March 2024). The primary outcome was patient-level deprescribing, defined as a composite of provider-directed discontinuation or tapering over follow-up beginning on the first visit, measured using EHR data. We used generalized estimating equations with a log link and binary-distributed errors adjusting for clustering and multiple testing using Holm-Bonferroni corrections.

Results: In total, 1146 eligible patients were included; their mean age was 73.6 years, and 69.7% were female, with characteristics well-balanced. Across the three arms, 32.5% of patients had their eligible medications deprescribed: 36.8% in the pre-commitment arm, 34.3% in the boostering arm, and 26.8% in usual care. Compared with usual care, pre-commitment EHR tools were 40% more likely to lead to deprescribing (adjusted relative risk [RR]: 1.40, 95%CI: 1.11-1.73). Boostering EHR tools were 26% more likely to lead to deprescribing (adjusted RR: 1.26, 95%CI: 1.01-1.57). Differences were largely driven by rates of PCP-directed discontinuation (Pre-commitment: 26.9%; Booster: 26.4%; Usual care: 18.5%).

Conclusions: Deploying EHR deprescribing tools when opening patient encounters and that enable PCPs to address deprescribing over multiple visits are effective in older adults, even in a different health system. This rigorous multi-trial process emphasizes the need for and success of EHR deprescribing tools to incorporate a longitudinal approach to addressing chronic high-risk medications.