Predictors of Hospitalization in Community-acquired Clostridium difficile Infection: A Population-based Study

Abstract

Purpose: Community-acquired Clostridium difficile infection (CA-CDI) is an increasingly recognized cause of diarrhea, including those who were previously thought to be at low risk or lacked traditional risk factors. Outcomes of CA-CDI have not been routinely examined. Therefore, we assessed predictors of hospitalization in a population-based cohort of CA-CDI. Methods: A computerized diagnostic index, which captured all inpatient and outpatient medical encounters for Olmsted County residents, was used to identify CDI cases from 1991-2005. Medical records were reviewed to confirm the diagnosis, document demographic data, and assess treatment outcomes. CDI was defined as community-acquired if symptom onset occurred in the community or within 48 hours of hospital admission, provided symptom onset was more than 4 weeks after the last hospital discharge. Severe CDI was defined by a peripheral white blood cell count ≥ 15,000/uL or a serum creatinine rise of ≥ 50% from baseline. Recurrent CDI was defined if diagnostic criteria were met within 8 weeks of initial diagnosis after documented symptom resolution. Treatment failure was defined as change in treatment within 14 days due to non-response or intolerable drug side effects. Results: Of 157 CA-CDI cases, median age was 50 years and 75.3% were female. Forty percent of CA-CDI cases were subsequently hospitalized for CDI management and the remaining were treated as outpatients. On univariate analysis, patients hospitalized for CA-CDI were significantly older (median age 64 vs 44 years, p<0.001), had higher rates of severe disease (33.3% vs 11.7%, p=0.001), higher Charlson Comorbidity index scores (p=0.001) and less likely to have antibiotic exposure (68.2% vs 85.2%, p=0.01). There were no differences in sex, recent gastrointestinal endoscopy procedures, initial CDI treatment, treatment failure or risk of recurrence in those hospitalized compared to managed as an outpatient. On multivariate logistic regression analysis, increasing age (p=0.02), higher Charlson Comorbidity index (p=0.04), disease severity (p=0.02) and lack of antibiotic exposure (p=0.02) predicted the need for hospitalization in patients with CA-CDI. Conclusion: In this population-based cohort, forty percent of CA-CDI cases were hospitalized for management of CDI. Patients hospitalized were older and had more severe CDI. Interestingly, those hospitalized were less likely to have received prior antibiotics, suggesting that other undefined risk factors associated with CA-CDI were operative, and could be predictors of hospitalization. Future studies are needed to better characterize determinants of infection risk and need for hospitalization in CA-CDI. This research was supported by an industry grant from ViroPharma.