A predictive model of mortality in patients with bloodstream infections due to carbapenemase-producing Enterobacteriaceae

Objective To develop a score to predict mortality in patients with bloodstream infections (BSIs) due to carbapenemase-producing Enterobacteriaceae (CPE). Patients and Methods A multinational retrospective cohort study (INCREMENT project) was performed from January 1, 2004, through December 31, 2013. Patients with clinically relevant monomicrobial BSIs due to CPE were included and randomly assigned to either a derivation cohort (DC) or a validation cohort (VC). The variables were assessed on the day the susceptibility results were available, and the predictive score was developed using hierarchical logistic regression. The main outcome variable was 14-day all-cause mortality. The predictive ability of the model and scores were measured by calculating the area under the receiver operating characteristic curve. The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were calculated for different cutoffs of the score. Results The DC and VC included 314 and 154 patients, respectively. The final logistic regression model of the DC included the following variables: severe sepsis or shock at presentation (5 points); Pitt score of 6 or more (4 points); Charlson comorbidity index of 2 or more (3 points); source of BSI other than urinary or biliary tract (3 points); inappropriate empirical therapy and inappropriate early targeted therapy (2 points). The score exhibited an area under the receiver operating characteristic curve of 0.80 (95% CI, 0.74-0.85) in the DC and 0.80 (95% CI, 0.73-0.88) in the VC. The results for 30-day all-cause mortality were similar. Conclusion A validated score predictive of early mortality in patients with BSIs due to CPE was developed.