# A Vanished Association Between Proton Pump Inhibitors and Clostridioides Difficile Infection After Minimizing Bias

## TL;DR

This study finds that the link between proton pump inhibitors and Clostridioides difficile infection is likely due to co-use of antibacterial drugs, not the inhibitors themselves.

## Contribution

The study introduces a novel approach to minimize competition bias in pharmacovigilance data to reassess PPI-CDI associations.

## Key findings

- Initial analysis showed a significant PPI-CDI association with ROR = 2.36 and IC = 1.21.
- After adjusting for antibacterial drug competition, the CDI signal strength dropped significantly.
- Age-stratified analyses showed complete signal loss after antibacterial drug adjustment across all age groups.

## Abstract

Background: The gut microbiome might be affected by proton-pump inhibitors (PPIs), increasing the risk of Clostridioides difficile infection (CDI); however, the association between PPIs and Clostridioides difficile infection (CDI) remains controversial. Aim: The aim of this study is to reevaluate the association between PPIs and CDI based on pharmacovigilance data, taking competition bias into account. Methods: PPI-related CDI adverse event reports, based on the Food and Drug Administration adverse event reporting system database from 2004 to 2023, were analyzed. Included PPI cases were stratified into CDI and non-CDI groups. Disproportionality analysis was performed using the reporting odds ratio (ROR) and information component (IC). The effect of competition bias on signal detection was quantitatively investigated. Age-stratified analyses were conducted to assess residual confounding. Results: A total of 238,470 PPI reports were included, with 1268 cases in the CDI group and 237,202 cases in the non-CDI group. Initial analysis revealed a significant PPI-CDI association (ROR = 2.36, 95% confidence interval (95%CI) 2.19 to 2.53; IC = 1.21, 95%CI 0.97 to 1.45), with CDI signals detected for five PPI agents, including pantoprazole, omeprazole, lansoprazole, rabeprazole, and dexlansoprazole. After excluding competition from antibacterial drugs, CDI signal strength decreased substantially (ROR = 1.47, 95%CI 1.34 to 1.62; IC = 0.55, 95%CI 0.23 to 0.87), retaining a significant CDI signal only for rabeprazole and pantoprazole. Upon further exclusion of antibacterial or immunosuppressive drug users and renal injury event cases, CDI signal strength decreased (ROR = 1.48, 95%CI 1.32 to 1.66; IC = 0.56, 95%CI 0.18 to 0.94), with pantoprazole as the sole CDI signal drug. Age-stratified analyses demonstrated complete signal loss after antibacterial drug adjustment across all age groups. Conclusions: The current large-scale pharmacovigilance study indicated that the observed PPI-CDI association may be mediated predominantly by antibacterial drug co-exposure rather than PPI direct causation.