ABSTRACT
OBJECTIVES: In severe infections, time to appropriate therapy is decisive for survival. Patients with bloodstream infection caused by extended-spectrum ß-lactamase-producing Enterobacterales (EPE-BSI) often receive inadequate empirical treatment. This study aimed to identify risk factors, to evaluate a previously suggested risk score and to suggest a new score for facilitating empirical treatment choice. METHODS: Predictors for EPE-BSI were assessed through a retrospective case-control design. The diagnostic performance of the two scores was evaluated. Included patients had blood cultures sampled at four EDs in Stockholm (2012-2015), were admitted, and received antibiotics with activity against Gram-negative bacilli. RESULTS: A total of 277 EPE-BSI cases and 400 controls were included. The strongest predictor of EPE-BSI was prior EPE-positive culture (cases 33% vs. controls 3%; multivariate (MV) ORâ¯=â¯19.1). Recent EPE-positivity within ≤3 months had a univariate OR of 32.8. Other major predictors were recent prostate biopsy (14% vs. 1%; MV ORâ¯=â¯22.2) and healthcare abroad (6% vs. 2%; MV ORâ¯=â¯3.9). Several previously suggested risk factors were not associated with EPE-BSI. The previously developed Utrecht score had a sensitivity of 54% and a specificity of 77%. The Stockholm score suggested herein (prior EPE-positive culture/prostate biopsy/healthcare abroad) showed comparable sensitivity (50%) but better specificity (96%). Prediction in patients lacking major predictors was difficult and caused high false-positive rates, which would cause unnecessary overtreatment. CONCLUSIONS: Prior EPE-positive culture, especially recently sampled, prostate biopsy and healthcare abroad were the strongest risk factors for community-onset EPE-BSI in Stockholm. Local data are needed when evaluating risk-scoring models before implementation.