Antibiotic exposure and risk of community-associated Clostridium difficile infection: A self-controlled case series analysis

Abstract

Community-associated Clostridium difficile infection is inconsistently associated with antibiotic exposure. This study uses a self-controlled case series (SCCS) design to estimate antibiotic exposure effect sizes and compare them with those estimated from previous case-control studies. We estimated the association between antibiotic exposure and community-associated Clostridium difficile infection among 139,000 patients registered to the Barrie Family Health Team from January 1, 2011, to May 1, 2017, using an SCCS design. Poisson regression analysis was used to estimate the incidence rate ratio (IRR) between antibiotic exposure versus nonexposure periods within individuals. Antibiotic exposure was categorized as either high risk (fluoroquinolone, clindamycin, or cephalosporin) or low risk (all other antibiotic classes). The final analysis included 189 cases. The pooled IRR for high-risk antibiotics was 2.26 (95% confidence interval [CI] 1.29, 3.98) and 2.03 (95% CI 1.19, 3.47) for lower-risk antibiotics. There was no difference between high-risk and lower-risk antibiotics (IRR 1.11, 95% CI 0.53, 2.36). The IRRs were smaller than the odds ratios reported in previous case-control studies, suggesting a less biased estimate because SCCS designs control for time-invariant confounders. Compared with case-control studies, SCCS designs are underused in infection prevention and control studies.