Deprescribing Intervention and Reduction of Proton Pump Inhibitor Use in Primary Care: A Cluster Randomized Clinical Trial
Mailing education to patients plus guidance to clinicians roughly doubled meaningful dose reduction of long-term proton pump inhibitor use without worsening reflux symptoms.
*Pragmatic cluster randomized clinical trial; Level 1b (OCEBM).
*Pragmatic cluster randomized clinical trial; Level 1b (OCEBM).
Citation
Fournier J-P, Gaultier A, Riche V-P, Tessier P, Rat C, Nguyen-Soenen J. Deprescribing intervention and reduction of proton pump inhibitor use in primary care: a cluster randomized clinical trial. JAMA Internal Medicine. 2026;186(6):668-676. doi:10.1001/jamainternmed.2026.0584
Background
Long-term proton pump inhibitor use is often unnecessary and may cause harm and higher costs. This study tested a health system–scale approach to encourage safe reduction in primary care.
Patients
Adults (≥18 years) in western France with at least 1 year of high proton pump inhibitor use (>300 defined daily doses in the prior year). Excluded: patients at higher ulcer-bleeding risk (older than 65 years using anti-inflammatory pain medicines) and those using steroids, blood thinners, or antiplatelet medicines.
Intervention
Mailing to patients: educational brochure plus motivational letter; mailing to clinicians: deprescribing letter plus step-by-step dose-reduction algorithm.
Control
Clinician-only mailing (letter + algorithm) or usual care.
Outcome
Primary: ≥50% reduction in annual proton pump inhibitor use (from reimbursement claims). Secondary: reflux symptom scores in a 10% sample using a 1-to-4 scale (1=never, 4=daily); a change ≥0.5 is clinically meaningful.
Follow-up Period
1 year
Results
| Comparison | Patients with ≥50% annual dose reduction | Adjusted absolute difference | NNT |
|---|---|---|---|
| Patient+clinician mailings vs usual care | 14.9% vs 7.0% | +6.9% | 15 |
| Patient+clinician mailings vs clinician-only mailing | 14.9% vs 7.7% | +6.7% | 15 |
Reflux symptom scores did not differ between groups and changes were not clinically meaningful. Analysis used a modified intention-to-treat approach.
Limitations
Medication use was inferred from reimbursements, not confirmed swallowing of pills. Reflux symptoms were measured only in a small sample with incomplete responses and only at 1 year, which may miss short-term rebound symptoms. Claims data could not fully rule out appropriate long-term use (such as severe esophagus disease).
Funding
French Ministry of Health grant (PREPS-19-0040); no funder role.
Clinical Application
For long-term users without high bleeding risk, consider mailing patient education plus clinician guidance; expect modest but meaningful reductions without worsening reflux symptoms.
Discussion
Sign in to join the discussion.
In this pragmatic, open-label cluster-randomized clinical trial, a mailed patient brochure plus a GP deprescribing algorithm increased ≥50% annual PPI dose reduction at 1 year (14.9% vs 7.0%; adjusted absolute difference 6.9%). What threats to validity or implementation barriers would stop you from adopting this system-wide in your practice? This study randomized entire GP practices (clusters) to receive a patient-and-GP-facing deprescribing intervention vs GP-only vs usual care. What is the primary methodological reason to use cluster randomization in this context?