ABSTRACT
We evaluated the abilities of 10 commercially available antimicrobial susceptibility testing methods and four reference methods (agar dilution, broth microdilution, disk diffusion, and the agar screen plate) to classify enterococci correctly as vancomycin susceptible or resistant using 50 well-characterized strains of enterococci. There was a high level of agreement of category classification data obtained with broth-based systems (Sceptor, MicroMedia, Pasco, and Sensititre), agar dilution, and an antibiotic gradient method (E test) with data obtained by reference broth microdilution; no very major or major errors were seen, and minor errors were < or = 6%. Increased minor error rates were observed with disk diffusion (12%), Alamar (16%), Uniscept (16%), and conventional (overnight) MicroScan panels (16%). The errors were primarily with Enterococcus casseliflavus strains and organisms containing the vanB vancomycin resistance gene. Very major error rates of 10.3 and 20.7% were observed with Vitek and MicroScan Rapid (MS/Rapid) systems, respectively; however, only the MS/Rapid system produced major errors (13.3%). On repeat testing of discrepant isolates, the very major error rate with the Vitek system dropped to 3.4%, while the very major error rate with the MS/Rapid system increased to 27.6%; major errors with the MS/Rapid system were not resolved. Many of the commercial systems had only 4 dilutions of vancomycin, which resulted in up to 84% of values being off scale (e.g., Uniscept). Of the methods tested, most conventional broth- and agar-based methods proved to be highly accurate when incubation was done for a full 24 h, although several of the tests had high minor error rates. Automated systems continued to demonstrate problems in detecting low-level resistance.
