A Comprehensive Assessment Across the Healthcare Continuum: Risk of Hospital-Associated Clostridium difficile Infection Due to Outpatient and Inpatient Antibiotic Exposure

Abstract

BACKGROUNDLimitations in sample size, overly inclusive antibiotic classes, lack of adjustment of key risk variables, and inadequate assessment of cases contribute to widely ranging estimates of risk factors for Clostridium difficile infection (CDI).OBJECTIVETo incorporate all key CDI risk factors in addition to 27 antibiotic classes into a single comprehensive model.DESIGNRetrospective cohort study.SETTINGKaiser Permanente Southern California.PATIENTSMembers of Kaiser Permanente Southern California at least 18 years old admitted to any of its 14 hospitals from January 1, 2011, through December 31, 2012.METHODSHospital-acquired CDI cases were identified by polymerase chain reaction assay. Exposure to major outpatient antibiotics (10 classes) and those administered during inpatient stays (27 classes) was assessed. Age, sex, self-identified race/ethnicity, Charlson Comorbidity Score, previous hospitalization, transfer from a skilled nursing facility, number of different antibiotic classes, statin use, and proton pump inhibitor use were also assessed. Poisson regression estimated adjusted risk of CDI.RESULTSA total of 401,234 patients with 2,638 cases of incident CDI (0.7%) were detected. The final model demonstrated highest CDI risk associated with increasing age, exposure to multiple antibiotic classes, and skilled nursing facility transfer. Factors conferring the most reduced CDI risk were inpatient exposure to tetracyclines and first-generation cephalosporins, and outpatient macrolides.CONCLUSIONSAlthough type and aggregate antibiotic exposure are important, the factors that increase the likelihood of environmental spore acquisition should not be underestimated. Operationally, our findings have implications for antibiotic stewardship efforts and can inform empirical and culture-driven treatment approaches.Infect. Control Hosp. Epidemiol. 2015;36(12):1409–1416