Risk Factors and Clinical Outcomes of Community-Onset Bacteremia Caused by Multidrug-Resistant Escherichia coli and Klebsiella pneumoniae

Background The rapid emergence of antibiotics resistance is a major global health threat. Multidrug-resistant (MDR) organisms are associated with receiving ineffective empirical antibiotics, longer hospital stays, higher medical costs and increased mortality in hospital settings. However, the prevalence, impact on outcomes and risk factors of common MDR gram-negative organisms causing bacteremia among community patients has seldom been elucidated. Objective To clarify the temporal trend of prevalence and impact on outcomes of community-onset bacteremia caused by MDR Escherichia coli and Klebsiella pneumoniae and develop an easy-to-use predictive rule by identifying the risk factors to assist physicians in empirical antibiotics selection. Methods This retrospective cohort study enrolled all emergency department adult patients with E. coli and K. pneumoniae bacteremia in three study years: 2001, 2006, and 2011. MDR isolate was defined as resistance to at least 3 of different antimicrobial classes. Baseline demographic, clinical characteristics and treatment outcomes were compared. Temporal trend of MDR isolates were analyzed and prognostic factors associated with 30-day mortality were determined by Cox proportional hazard regression model. For prediction model, four-fifths of patients were randomly allocated to a derivation set and the others to a validation set for model training and testing. Independent risk factors determined by logistic regression model were included for model construction. A simplified coefficient-base scoring model was also established for the ease of clinical application. Results A total of 1770 bacteremia episodes (E. coli: 1243, K. pneumoniae: 527) were enrolled. Increasing in prevalence of MDR isolates was observed (year 2001: 4.0%, year 2006: 7.6%, year 2011: 10.1%, p < 0.001 by test for trend). Patients with community-onset bacteremia caused by MDR E. coli and K. pneumoniae were significantly higher risk of septic shock (27.6% vs. 11.9%, p < 0.001), more likely to receive inappropriate empirical antibiotics (70.2% vs. 10.5%, p < 0.001), increased 30-day mortality (19.8% vs. 13.5%, p = 0.047) and longer length of hospital stay (19.9 ± 23.7 days vs. 15.6 ± 17.1 days, p = 0.042). MDR isolates (adjusted HR, 1.92; 95% CI, 1.06-3.45), malignancy (adjusted HR, 3.95; 95% CI, 2.74-5.71), liver cirrhosis (adjusted HR, 1.84; 95% CI, 1.17-2.90), bacteremia due to K. pneumoniae (adjusted HR, 2.10; 95% CI, 1.45-3.06) and urinary tract infection (adjusted HR, 0.47; 95% CI, 0.30-0.74) are the independent prognostic factors associated with 30-day mortality in patients without septic shock. Malignancy (adjusted HR, 2.30; 95% CI, 1.56-3.40) and urinary tract infection (adjusted HR, 0.44; 95% CI, 0.26-0.75) are the independent prognostic factors associated with 30-day mortality in patients with septic shock. For prediction model, the independent risk factors of bacteremia due to MDR E. coli and K. pneumoniae identified from derivation set were nursing home residence (adjusted OR, 11.14; 95% CI, 5.57-22.31), OPD invasive procedure in the past 30 days (adjusted OR, 1.86; 1.05-3.29), prior hospitalization in the past 90 days (adjusted OR, 3.53; 95% CI, 2.30-5.42), congestive heart failure (adjusted OR, 2.87; 95% CI, 1.39-5.94) and cerebrovascular accident (adjusted OR, 1.90; 95% CI, 1.10-3.28) (CI, confidence interval). A clinical risk score was derived and the area under receiver operating characteristic(ROC) curve was 0.75 (95% CI, 0.70-0.80). The prediction model successfully identified 30.0% to 38.8% of patients with MDR infection in high clinical risk score ≥ 4 points. Conclusions MDR E. coli and K. pneumoniae causing community-onset bacteremia is alarming, and prevalence has been increasing over time. Impact of community-onset bacteremia caused by MDR E. coli and K. pneumoniae include higher risk of septic shock, more likely received inappropriate empirical antibiotics, increased 30-day mortality, longer hospital length of stay, and independently associated with increased 30-day mortality in patients without septic shock. Risk stratification by simple clinical decision rule is important for optimizing patient care and antibiotics stewardship.