The Burden of E. coli Blood Stream Infections: Large Impact of Third Generation Cephalosporin Resistance on Mortality and Length of Stay in Europe

Background: The prevalence of infections caused by resistant E. coli is increasing worldwide. For prioritization in health care determination of the burden of this infection is important. Blood stream infections (BSI) caused by third generation cephalosporin (G3CEP) resistant E. coli (RECO) affect patients that are older, more ill and had a longer hospital stay before infection than patients with G3CEP susceptible E. coli (SECO) BSI, making direct comparisons difficult. In this study modern sampling techniques were used to avoid this problem.

Objective: Determine the excess mortality and the excess length of stay attributable to RECO BSI in Europe.

Methods: In this study a parallel matched cohort design was used, where patients with an E. coli BSI (exposed) were compared to patients without such a BSI (controls). Two parallel cohorts were constructed, comparing either RECO or SECO exposure. Matching was based on duration of admission prior to enrolment. Thirteen European hospitals, participated in this study from July 2007 to July 2008.  Hospital patients were routinely sampled and all patients above 18 years with an E. coli BSI were included as exposed patients. Modern statistical methods were used for comparing mortality and length of stay within the two parallel cohorts: multivariate Cox's regression for competing events and a multivariate generalized linear model with gamma distribution.

Results: In total 3509 patients could be included into the study: 111 and 1110 patients had a BSI caused by RECO and SECO, respectively, 2288 patients were included as controls. In-hospital 36% of the RECO BSI patients died compared to 5% of the controls. In the SECO cohort 17% of the exposed patients and 7% of the controls died in the hospital. Cox's regression showed that the hazard for hospital mortality was 5.7 times larger for patients with a RECO BSI than for the controls (confidence interval (CI) 2.5-13.0). These exposed patients had a reduced discharge hazard rate, meaning that they stayed much longer in hospital than the controls. The overall excess length of stay after enrolment for the patients with a RECO BSI was 7.9 days (interquartile range (IQR) 3.5-13.0). In the SECO cohort, the same effects were seen, although the differences were smaller. More exposed patients died (hazard ratio (HR); 2.0, CI 1.5-2.5) and the overall excess length of stay after enrolment for the patients with a SECO BSI was 2.9 days (IQR 1.7-4.0). Overall, G3CEP resistance was associated with excess hospital mortality (HR 2.9, CI 1.2-6.9), as well as an excess length of stay (5.0 days, IQR 0.4-10.2).

Conclusions: BSI caused by E. coli increases hospital mortality, as well as the length of stay after enrolment, especially if the pathogen is resistant to G3CEP. Prevention and improved management of infections caused by, especially resistant, gram-negative pathogens within hospitals could reduce a large burden.