Performance of the Phoenix bacterial identification system compared with disc diffusion methods for identifying extended-spectrum β-lactamase, AmpC and KPC producers

Phenotypic identification of AmpC, KPC and extended-spectrum beta-lactamases (ESBLs) among members of the Enterobacteriaceae remains challenging. This study compared the Phoenix Automated Microbiology System (BD Diagnostics) with the Clinical and Laboratory Standards Institute confirmatory method to identify ESBL production among 200 Escherichia coli and Klebsiella pneumoniae clinical isolates. The Phoenix system misclassified nearly half of the isolates as ESBL-positive, requiring manual testing for confirmation. Inclusion of aztreonam +/- clavulanic acid (CA) and cefpodoxime +/- CA in the testing algorithm increased the ESBL detection rate by 6%. Boronic acid-based screening identified 24 isolates as AmpC(+), but in a subset of genotypically characterized isolates, appeared to have a high false-positivity rate. PCR screening revealed eight KPC(+) isolates, all of which tested as ESBL(+) or ESBL(+) AmpC(+) by phenotypic methods, but half were reported as carbapenem-susceptible by the Phoenix system. Overall, these results indicate that laboratories should use the Phoenix ESBL results only as an initial screen followed by confirmation with an alternative method.