Diagnostic uncertainty is common in the emergency room and multidrug-resistant bacteria emerge in the community setting, implying to establish the most efficient empirical antibiotic therapy (eEAT). Our aim was to identify such eEAT, considering that in case of DU with severe clinical presentation, most prescribers would propose an empiric combination (EC). The medical dashboard of our ward records prospectively 28 characteristics of each hospitalization including hospitalization motive, final diagnosis, and all antibiotics prescribed. All patients with community-acquired bacteremia (CAB) were included. DU was defined by a discrepancy between suspected diagnosis in the emergency room and final diagnosis. eEAT was defined by in vitro activity of at least one prescribed compound. Finally, independently from the dashboard, we retrospectively compared 2 CTs: amoxicillin/clavulanic acid (AMC)+gentamicin (G) and cefotaxime (3GC)+G. One thousand thirty-four patients with a final diagnosis of CAB were identified from July 2005 to June 2018, including 357 DU (35%) at baseline. eEAT (n = 553) was associated with a trend towards a lower death rate compared to inefficient therapies: 5.4 vs 10.0% (p = 0.053), and effective antibiotic reassessment was the most protective factor against an unfavorable outcome: 0.34 (0.16-0.71). Bacteria involved in case of UD were resistant to AMC+G and to 3GC+G in 8.1% and 12.8% of patients, respectively. Diagnostic uncertainty was a frequent event requiring antibiotic reassessment. As the latter was not systematically realized, the best eEAT is required and AMC+aminoglycoside should be considered.
Keywords: Bacteremia; Community-acquired infection; ESBL Enterobacteriaceae; Empirical antibiotic therapy; Outcome.