LAUSR

Objectives To compare results of amoxicillin/clavulanate susceptibility testing using CLSI and EUCAST methodologies and to evaluate their impact on outcome in patients with bacteraemia caused by Enterobacteriaceae. Patients and methods A prospective observational cohort study was conducted in 13 Spanish hospitals. Patients with bacteraemia due to Enterobacteriaceae who received empirical intravenous amoxicillin/clavulanate treatment for at least 48 h were included. MICs were determined following CLSI and EUCAST recommendations. Outcome variables were: failure at the end of treatment with amoxicillin/clavulanate (FEAMC); failure at day 21; and 30 day mortality. Classification and regression tree (CART) analysis and logistic regression were performed. Results Overall, 264 episodes were included; the urinary tract was the most common source (64.7%) and Escherichia coli the most frequent pathogen (76.5%). Fifty-two isolates (19.7%) showed resistance according to CLSI and 141 (53.4%) according to EUCAST. The kappa index for the concordance between the results of both committees was only 0.24. EUCAST-derived, but not CLSI-derived, MICs were associated with failure when considered as continuous variables. CART analysis suggested a 'resistance' breakpoint of > 8/4 mg/L for CLSI-derived MICs; it predicted FEAMC in adjusted analysis (OR = 1.96; 95% CI: 0.98-3.90). Isolates with EUCAST-derived MICs >16/2 mg/L independently predicted FEAMC (OR = 2.10; 95% CI: 1.05-4.21) and failure at day 21 (OR= 3.01; 95% CI: 0.93-9.67). MICs >32/2 mg/L were only predictive of failure among patients with bacteraemia from urinary or biliary tract sources. Conclusions CLSI and EUCAST methodologies showed low agreement for determining the MIC of amoxicillin/clavulanate. EUCAST-derived MICs seemed more predictive of failure than CLSI-derived ones. EUCAST-derived MICs >16/2 mg/L were independently associated with therapeutic failure.