Abstract
BACKGROUND: Of the patients with a Clostridium difficile infection, 2% to 8% will progress to fulminant C. difficile colitis (fCDC), which carries high morbidity and mortality. No system exists to rapidly identify patients at risk for developing fCDC and possibly in need of surgical intervention. Our aim was to design a simple and accurate risk scoring system (RSS) for daily clinical practice.
METHODS: We prospectively enrolled all patients diagnosed with a C. difficile infection and compared patients with and without fCDC. An expert panel, combined with data derived from previous studies, identified four risk factors, and a multivariable logistic regression model was performed to determine their effect in predicting fCDC. The RSS was created based on the predictive power of each factor, and calibration, discrimination, and test characteristics were subsequently determined. In addition, the RSS was compared with a previously proposed severity scoring system.
RESULTS: A total of 746 patients diagnosed with C. difficile infection were enrolled between November 2010 and October 2012. Based on the log (odds ratio) of each risk factor, age greater than 70 years was assigned 2 points, white blood cell count equal to or greater than 20,000/mu L or equal to or less than 2,000/mu L was assigned 1 point, cardiorespiratory failure was assigned 7 points, and diffuse abdominal tenderness on physical examination was assigned 6 points. With the use of this system, the discriminatory value of the RSS (c statistic) was 0.98 (95% confidence interval, 0.96-1).The Ho smer-Lemeshow goodness-of-fit test showed a p value of 0.78, and the Brier score was 0.019. A value of 6 points was determined to be the threshold for reliably dividing low-risk (<6) from high-risk (>= 6) patients.
CONCLUSION: The RSS is a valid and reliable tool to identify at the bedside patients who are at risk for developing fCDC. External validation is needed before widespread implementation. Copyright (C) 2014 by Lippincott Williams & Wilkins
