Abstract
<jats:p>Background Proton pump inhibitors (PPIs) have been proposed as a risk factor for initial and recurrent Clostridioides difficile infection (CDI) based on observational data. Whether desprescribing PPIs reduces the odds of incident CDI is unknown. Methods This was a secondary analysis of the cluster randomized MedSafer clinical trial dataset, in which hospitalized older adults taking five or more medications were randomized to usual care (a medication review) or electronic deprescribing decision support prior to hospital discharge. For this substudy, all patients were included regardless of trial assignment and were categorized according to their PPI use at hospitalization and discharge. The 30-day occurrence of CDI was compared as a function of PPI use using a Bayesian mixed-effect logistic regression adjusted for prespecified confounders. Results 4990 patients were included in the analysis, including 405 (8.1%) patients who were deprescribed their home PPI and 325 (6.5%) new PPI users. 30-day CDI occurred in 32 (0.6%) patients. PPI deprescription was associated with a 99.1% probability of increased odds of CDI (adjusted odds ratio[aOR]=3.66, 95%Credible interval[95%CrI]=1.25-10.11). New PPI use, however, was not associated with a probable increase in CDI (aOR=1.14, 95%CrI=0.20-4.67). Conclusion Unexpectedly, PPI deprescription was associated with a higher odds of CDI. This challenges the practice of deprescribing unnecessary PPIs in hospitalized patients for CDI prevention. However, it is possible this finding is artefactual due to confounding by indication. Further mechanistic and confirmatory studies are needed.</jats:p>