Proton Pump Inhibitors and Infection-Related Hospitalizations Among Residents of Long-Term Care Facilities: A Case-Control Study

Objective Our objective was to investigate associations between proton pump inhibitor (PPIs) use and infection- related hospitalizations among residents of long-term care facilities ( LTCFs). Methods This was a case-control study of residents aged = 65 years admitted to hospital between July 2013 and June 2015. Residents admitted for infections (cases) and falls or fall-related injuries (controls) were matched for age (+/- 2 years), sex, and index date of admission (+/- 6 months). Conditional logistic regression was used to estimate crude and adjusted odds ratios (aORs) and 95% confidence intervals (CIs) for associations between PPI use and infection-related hospitalizations. Analyses were adjusted for age, sex, polypharmacy, diabetes, heart failure, chronic obstructive pulmonary disease, myocardial infarction, cerebrovascular accident, and concomitant use of cancer and immunosuppressant medications. Subgroup analyses were performed for high- and low/moderate-intensity PPIs and for respiratory and non-respiratory infections. Logistic regression was used to compare the odds of infection- related hospitalizations among users of high- and low/moderate-intensity PPIs. Results Overall, 181 cases were matched to 354 controls. Preadmission PPI use was associated with infection-related hospitalizations (aOR 1.66; 95% CI 1.11-2.48). In subgroup analyses, the association was apparent only for respiratory infections (aOR 2.26; 95% CI 1.37-3.73) and high-intensity PPIs (aOR 1.93; 95% CI 1.23-3.04). However, the risk of infection-related hospitalization was not significantly higher among users of high- versus low/moderate-intensity PPIs (aOR 1.25; 95% CI 0.74-2.13). Conclusion Residents who use PPIs may be at increased risk of infection-related hospitalizations, particularly respiratory infections. Study findings provide further support for initiatives to minimize unnecessary PPI use in the LTCF setting.