Objective: The aim of this study was to establish the degree of comparability between the NCCLS broth microdilution and BSAC agar dilution MIC methods of antimicrobial susceptibility testing.
Methods: Six hundred and sixty-one clinical isolates of Streptococcus pneumoniae, 936 Haemophilus influenzae and 421 Moraxella catarrhalis, collected in the winter of 1999-2000 by 20 laboratories in the UK and Eire from patients with presumed community-acquired lower respiratory infections, were tested by the two methods. MIC agreement was defined as excellent (good) if results were within +/- 1 doubling dilution for > or =90% (> or =80%) of isolates and within +/- 2 doubling dilutions for > or =95%. Isolates were categorized as susceptible, intermediate or resistant using the breakpoints appropriate to the testing method.
Results: MIC agreement was good or excellent in 27 of 36 organism-agent combinations. Agreement was less for M. catarrhalis than for other species, and lower in all three species for cefaclor and trimethoprim than for other antimicrobials. Discrepancies in categorization occurred only occasionally, and were generally explained by differences in breakpoints rather than in measured MICs. One exception was S. pneumoniae with penicillin. Despite excellent MIC agreement and identical breakpoints, 9% of these had minor discrepancies, mainly because 7% of isolates were found intermediate by the BSAC method but resistant by the NCCLS method.
Conclusion: There is generally very good agreement between the MICs obtained by the BSAC agar dilution and NCCLS broth microdilution methods in this population of isolates, comparable to the level of agreement achieved between different laboratories using a single method. Breakpoint differences contribute to most of the discrepancies in susceptibility categorization.