Medication risk factors associated with healthcare-associated Clostridium difficile infection: a multilevel model case-control study among 64 US academic medical centres.

OBJECTIVES: The main objective of this study was to determine patient- and hospital-level medication risk factors associated with Clostridium difficile infection (CDI) occurrence among patients clustered within hospitals using a multilevel model. METHODS: Patients with healthcare-associated (HA)-CDI were identified from among 64 academic medical centres in 2009. A frequency match was conducted; for each case, up to two controls were selected, matched on similar pre-infection length of stay and clinical service line. Patient- and hospital-level medication use, including antibacterial and gastric acid-suppressant agents, was assessed using a two-level logistic regression model. RESULTS: A total of 5967 CDI cases and 8167 controls were included in the analysis. The odds of acquiring HA-CDI increased with the following medications [OR (95% CI)]: anti-methicillin-resistant Staphylococcus aureus agents [1.38 (1.22-1.56)]; third- or fourth-generation cephalosporins [1.75 (1.62-1.89)]; carbapenems [1.60 (1.44-1.79)]; ß-lactam/ß-lactamase inhibitor combinations [1.49 (1.36-1.64)]; vancomycin [1.73 (1.57-1.89)]; and proton pump inhibitors [1.43 (1.30-1.57)]. The odds of acquiring HA-CDI decreased with the following medications: clindamycin [0.74 (0.63-0.87)]; and macrolides [0.88 (0.77-0.99)]. Controlling for patient-level covariates, no hospital-level medication covariates that we analysed had statistically significant effects on HA-CDI. The odds of acquiring HA-CDI increased with the hospital proportion of patients aged ≥ 65 years [1.01 (1.00-1.02)]. CONCLUSIONS: We found several medications that were associated with the risk of patients developing HA-CDI, including ß-lactam/ß-lactamase inhibitor combinations, third- or fourth-generation cephalosporins, carbapenems, vancomycin, proton pump inhibitors and anti-methicillin-resistant S. aureus agents. There were no medication effects significant at the hospital level.