Abstract
The aim of the study was to evaluate the EUCAST RAST method by extending analysis to 16−20 h reading time and performance with new β-lactam/β-lactamase inhibitor combinations. A total of 676 positive blood cultures (BCs) were enrolled. Results at 4 h, 6 h, 8 h and 16−20 h were interpreted according to bacterial species using EUCAST RAST breakpoints (version 5.1). For species for which no breakpoints were available, tentative breakpoints were used. Categorical agreement with the Microscan microdilution system was analysed. Among the 676 BCs enrolled, 641 were monomicrobial and were included in the analysis. Categorical agreement ranged from 98.9% at 4 h to 99.4% at 16−20 h. The rates of very major errors were 3.3%, 3.7% and 3.4% at 4 h, 6 h and 8 h, respectively, and decreased to 1% at 16−20 h (p < 0.001). The number of major errors was low for each reading time (0.2% and 0.4% at 4 h and 6 h, respectively, and 0.3% at both 8 h and 16−20 h). The proportions of results in the area of technical uncertainty were 9.9%, 5.9%, 5% and 5.2% for readings at 4 h, 6 h, 8 h and 16−20 h, respectively. Tentative breakpoints proposed for Enterobacterales other than E.coli/K.pneumoniae and coagulase-negative staphylococci showed overall performances comparable to those observed for E. coli/K. pneumoniae and S. aureus. In conclusion, EUCAST RAST has been shown to be reliable to determine microbial susceptibility to main antimicrobials, including ceftazidime/avibactam and ceftolozane/tazobactam. A poorer performance was observed for certain species/antimicrobial agent combinations. The better performance observed at 16−20 h compared to the early readings may confer to the method greater potential for antimicrobial de-escalation interventions.