Background: This study was designed to evaluate the ability of the Vitek and MicroScan ESBL test by comparing with NCCLS ESBL phenotypic confirmatory test by disk diffusion and to know the frequency of ESBL producers in the Seoul Veterans Hospital.

Methods: A total of 1,261 isolates(Escherichia coli 705, Klebsiella pneumoniae 502, K. oxytoca 54) from 883 patients were included in ESBL screening test by Vitek (494 strains) and MicroScan (767 strains). After excluding repetitive isolates from same patients, NCCLS ESBL confirmatory test was performed for 197 ESBL screening positives and 184 ESBL screening negatives.

Results:The overall frequency of ESBL screening positives was 22.3% (by MicroScan 26.2%, by Vitek 15.6%), and that of NCCLS ESBL positives was 18.9%(18.3% in E. coli, 21% in K. pneumoniae). MicroScan and Vitek ESBL test showed 100% and 92.3% sensitivity, 77.1% and 95.5% specificity, respectively. Among the 158 NCCLS ESBL positives, 17.7% showed clavulanic acid effect in cefotaxime only, 10.1% in ceftazidime only, and 72.2% in both. MicroScan Neg ComboPanel Type 21 test revealed that 91.4% of suspicious ESBL producers flagged by one or two antimicrobials were erroneous. In contrast, 96.2% of strains flagged by all five antimicrobials were correct.

Conclusion: Suspicious ESBL producers by MicroScan showing three or four antimicrobial flags should be retested by NCCLS ESBLconfirmatory test. But strains with two or less flags and strains with all 5 flags can be reported as Non-ESBL producers and ESBL producers, respectively. (Korean J Clin Microbiol 2003;6(1):41-46)