Patient-Directed Education to Promote Deprescribing: A Nonrandomized Clinical Trial

Abstract

Importance: Patient-directed educational materials are a promising implementation strategy to expand deprescribing reach and adoption, but little is known about the impact across medication groups with potentially different perceived risks.

Objective: To examine the impact of a patient-directed education intervention on clinician deprescribing of potentially low-benefit (proton pump inhibitors) or high-risk medications (high-dose gabapentin, diabetes agents with hypoglycemia risks).

Design, setting, and participants: This pragmatic multisite nonrandomized clinical trial took place at 3 geographically distinct US Veterans Affairs (VA) medical centers from April 2021 to October 2022. The total study sample was composed of the intervention cohort and the historical control cohort cared for by 103 primary care practitioners (PCPs).

Intervention: The primary intervention component was a medication-specific brochure, mailed during the intervention time frame to all eligible patients 2 to 3 weeks prior to upcoming primary care appointments. Patients seen by the same PCPs at the same sites 1 year prior to the study intervention served as controls.

Main outcome and measures: The primary binary outcome variable was deprescribing 6 months after the intervention, defined as complete cessation or any dose reduction of the target medication using VA pharmacy dispensing data.

Results: The total study sample included 5071 patients. The overall rate of deprescribing among the intervention cohort (n = 2539) was 29.5% compared with 25.8% among the controls (n = 2532). In an unadjusted model, the intervention cohort was statistically significantly more likely to have deprescribing (odds ratio [OR], 1.17 [95% CI, 1.03-1.33]; P = .02). In a multivariable logistic regression model nesting patients within PCPs within sites and controlling for patient and PCP characteristics, the odds of deprescribing in the intervention cohort were 1.21 times that of the control cohort (95% CI, 1.05-1.38; P = .008). The difference in deprescribing prevalence between the intervention and control cohorts (proton pump inhibitors: 29.4% vs 25.4%; gabapentin: 40.2% vs 36.2%; hypoglycemia risk: 27.3% vs 25.1%) did not statistically significantly differ by medication group (P = .90).

Conclusion and relevance: This nonrandomized clinical trial found that patient-directed educational materials provided prior to scheduled primary care appointments can effectively promote deprescribing for potentially low-benefit and high-risk medication groups.

Trial registration: ClinicalTrials.gov Identifier: NCT0429490.

Figure 1.. CONSORT Diagram

^aFor the intervention cohort, the index date was the date of the scheduled primary care practitioner appointment from April 2021 to October 2022. ^bFor the historical control cohort, the index date was the date of the scheduled primary care practitioner appointment from October 2019 to April 2020. PPI indicates proton pump inhibitor.

Figure 2.. Prevalence of Deprescribing: Medication Cessation or Reduction at 6 Months, Overall and by Medication Group