LAUSR

Background. Bloodstream infection (BSI) to due vancomycin-resistant Enterococcus (VRE) is an important complication of hematologic malignancy. Determining when to use empiric anti-VRE antibiotic therapy in this population remains a clinical challenge. Methods. A single-center cohort representing 664 admissions for induction or hematopoietic stem-cell transplant (HSCT) from 2006 to 2014 was selected. We derived a prediction score using risk factors for VRE BSI and evaluated the model's predictive performance by calculating it for each of 16232 BSI at-risk inpatient days. Results. VRE BSI incidence was 6.5% of admissions (2.7 VRE BSI per 1000 BSI at-risk days). Adjusted 1-year mortality and length of stay were significantly higher in patients with VRE BSI. VRE colonization (adjusted odds ratio [aOR] = 8.4; 95% confidence interval [CI] = 3.4-20.6; P < .0001), renal insufficiency (aOR = 2.4; 95% CI = 1.0-5.8; P = .046), aminoglycoside use (aOR = 4.7; 95% CI = 2.2-9.8; P < .0001), and antianaerobic antibiotic use (aOR = 2.8; 95% CI = 1.3-5.8; P = .007) correlated most closely with VRE BSI. A prediction model with optimal performance included these factors plus gastrointestinal disturbance, severe neutropenia, and prior beta-lactam antibiotic use. The score effectively risk-stratified patients (area under the receiver operating curve = 0.84; 95% CI = 0.79-0.89). At a threshold of ≥5 points, per day probability of VRE BSI was increased nearly 4-fold. Conclusions. This novel predictive score is based on risk factors reflecting a plausible pathophysiological model for VRE BSI in patients with hematological malignancy. Integrating VRE colonization status with risk factors for developing BSI is a promising method of guiding rational use of empiric anti-VRE antimicrobial therapy in patients with hematological malignancy. Validation of this novel predictive score is needed to confirm clinical utility.