Agreement And Error Rates In Antimicrobial Susceptibility Testing For Three Commercial Automated Systems: A Systematic Literature Review And Meta-Analysis

BACKGROUNDAutomated antimicrobial susceptibility testing (AST) systems are crucial for accurate, timely detection of drug-resistant microbial isolates. This meta-analysis assessed the performance of the BD Phoenix ("Phoenix", BD Diagnostic Solutions), Vitek(R) 2 ("Vitek 2", bioMerieux), and DxM MicroScan WalkAway ("MicroScan", Beckman Coulter, Inc.) AST systems relative to common reference methodology. METHODSA systematic literature search in Ovid (MEDLINE and Embase) yielded 275 unique (not duplicated) records, with 44 additional records retrieved from handsearching; 39 studies met inclusion criteria. Categorical agreement (CA), essential agreement (EA), very major errors (VMEs), and major errors (MEs) for the three instruments were compared to a common reference method. Ratios of proportions were analyzed using random-effect meta-regression. RESULTSThe instruments did not differ significantly in CA, EA, or ME. Vitek 2 showed a higher overall VME rate than Phoenix ([~]44% higher; Vitek 2-to-Phoenix ratio = 1.44; p=0.062 [approaching significance]) and MicroScan (74% higher; ratio = 1.74; p=0.045). No appreciable difference was observed for VME between Phoenix and MicroScan. Subgroup analyses should be interpreted cautiously due to limited overall significance indicating varying performance across systems. Vitek 2 generally had higher relative VMEs for gram-negative organisms and lower relative VMEs for gram-positive organisms, whereas Phoenix showed the opposite pattern. MicroScan had relatively low VMEs when stratified by Clinical and Laboratory Standards Institute (CLSI) criteria; no differences in VMEs were observed using European Committee on Antimicrobial Susceptibility Testing (EUCAST) criteria. CONCLUSIONAlthough some VME differences were noted, overall performance of the three systems was comparable. Organism- and drug-specific VME patterns--and updates to CLSI criteria over time--highlight the importance of continued monitoring of current breakpoints for all three instruments.