External validation of clinical scores to predict complications of Clostridium difficile infection

BACKGROUND: Clostridium difficile infection (CDI) is the most common cause of nosocomial diarrhea. About one in 5 patients with CDI (median 18%) develop a complication (cCDI), including mortality. Many predictive scores have been published to identify patients at risk of cCDI but none is currently recommended for clinical use and few were validated. We conducted an external validation study of predictive tools for cCDI. METHODS: Predictive tools were identified through a systematic review. We included those reporting at least an internal validation process. We performed the external validation on a multicenter prospective cohort of 1380 Canadian adults with confirmed CDI. Most cases were elderly (median age 71), had a healthcare facility-associated CDI (90%), and cCDI occurred in 8%. NAP1 strain was found in 52%. The performance of each scoring system was analyzed using individual outcomes. Modifications in predictors were made to match available data in the validation cohort. RESULTS: We assessed 3 predictive scores and one predictive model. The performance (95% CI) of higher thresholds are shown in the Table. All scores had a low sensitivity and PPV, and moderate specificity and NPV. The model of Shivashankar 2013 (age, WBC> 15, narcotic use, antacids use, creatinine ratio > 1.5) predicted 25% probability of cCDI. All showed similar AUC (0.63–0.67). CONCLUSION: The predictive tools included in our study showed moderate performance in a validation cohort with a low rate of cCDI and high proportion of NAP1 strains. Further research is needed to develop an accurate predictive tool to guide clinicians in the management of CDI. DISCLOSURES: J. Powis, Merck: Grant Investigator, Research grant; GSK: Grant Investigator, Research grant; Roche: Grant Investigator, Research grant; Synthetic Biologicals: Investigator, Research grant