Carbapenem-resistant Enterobacteriaceae colonization and infection in critically ill patients: a retrospective matched cohort comparison with non-carriers

Abstract

To examine whether carbapenem-resistant Enterobacteriaceae (CRE) carriage is associated with incidence of clinical infection as a means of assessing whether the morbidity and mortality associated with these bacteria are mediated by underlying conditions or intrinsic properties of CRE. This retrospective matched cohort study compared the incidence of invasive infections in CRE-colonized patients and matched non-carriers in the intensive care unit (ICU). The primary outcome was infection caused by CRE of the same species as the colonizing strain among CRE carriers, and infections caused by carbapenem-sensitive strains of the same organism in non-carriers. Hospital discharge and death were considered as competing events. Competing-risks hazard analysis was performed for the entire cohort and for a nested cohort matched by Acute Physiology and Chronic Health Evaluation (APACHE) II scores, stratified by matching. In total, 146 CRE carriers were compared with 292 non-carriers. Patients were well matched for most risk factors for Enterobacteriaceae infection, including age, renal failure, previous invasive infection, previous hospitalization, APACHE II score, length of mechanical ventilation, length of hospitalization and CRE carriage. On regression analysis, colonization with CRE was independently associated with Enterobacteriaceae infection {cause-specific hazard ratio (CSHR) 2.06 [95% confidence interval (CI) 1.03-4.09]}. On regression analysis of the APACHE-II-matched cohort (N=284), colonization with CRE remained significantly associated with Enterobacteriaceae infection [CSHR 3.32 (95% CI 1.31-8.43)]. Colonization with CRE was associated with at least a two-fold increased risk of infection by the colonizing strain amongst ICU patients.