Mortality in Clostridium difficile infection: a prospective analysis of risk predictors

Objectives To date, the vast majority of studies investigating risk factors for mortality in Clostridium difficile infection (CDI) have been based on retrospective, routinely collected data, and have not specifically tested the capacity of risk factors to predict outcome. We aimed to prospectively evaluate predictors of mortality in patients with CDI, utilizing established metrics of risk prediction to assess their ability to prognosticate. Patients and methods We collected a cohort of all patients diagnosed with CDI at Addenbrooke's Hospital in 2010. Univariate associations between several parameters and all-cause 30-day in-hospital mortality were assessed, with statistically significant parameters entered into a Cox regression model. A backwards selection procedure was used to derive a final multivariate model. Results The cohort consisted of 131 patients. From the univariate analyses white blood cell count (WBC) > 15 x 10(9)/l, serum albumin <25 g/l, serum creatinine >200 mu mol/l and C-reactive protein >100 nmol/l met criteria for entry into the multivariate model. WBC > 15 x 10(9)/l (hazard ratio 5.3, 95% confidence interval 1.7-16.8) and serum albumin level <25 g/l (hazard ratio 9.5, 95% confidence interval 1.2-74.5), were significantly associated with mortality in the final multivariate model. The model containing these variables had a C-index of 0.79, D-statistic of 2.1 and (RD)-D-2 measure of 0.52. Conclusion We have demonstrated in a prospective cohort of patients diagnosed with CDI that WBC and serum albumin, when used together, offer good risk predictive ability for mortality. Our results support the inclusion of these parameters in a clinically useful risk prediction model. (C) 2013 Wolters Kluwer Health vertical bar Lippincott Williams & Wilkins.