Clinical prediction rule for identifying patients with vancomycin-resistant enterococci (VRE) at the time of admission to the intensive care unit in a low VRE prevalence setting

Abstract

The purpose of this study was to develop and validate a clinical prediction rule to screen patients at risk of vancomycin-resistant enterococci (VRE) carriage at intensive care unit (ICU) admission in a hospital setting with low VRE prevalence. This study was retrospectively conducted in the ICUs of a university-affiliated hospital in Korea, where active surveillance cultures for VRE had been run at ICU admission and weekly thereafter. In the derivation cohort from April 2008 to September 2010, risk factors for VRE carriage at ICU admission were determined and assigned weighted point values using a multivariate logistic regression model. In the validation cohort from October 2010 to March 2011, predictability of the prediction rule was evaluated. Of a total of 4445 cultures taken from patients at ICU admission, 153 (3.4) patients carried VRE. In the derivation cohort, independent risk factors (assigned points) for VRE carriage at ICU admission were ICU readmission during hospitalization (1 point), chronic obstructive lung disease (2 points), recent antibiotic treatment (3 points) and recent vancomycin use (2 points). In the validation cohort, the sensitivity, specificity, and positive and negative predictive values of the prediction rule, on the basis of risk scores epsilon 3 points, were 84.2, 82.5, 15.2 and 99.3, respectively. This clinical prediction rule for identifying VRE carriage at the time of ICU admission is expected to markedly reduce the screening volume (by 80.1) in our healthcare facility. For use in clinical practice, the rule needs to be prospectively validated in other settings.