A variety of clinical anaerobic isolates were tested against cefoperazone (216 strains), cefoxitin (120 strains), and cefotaxime (120 strains) by the thioglycolate anaerobic broth disk method, and the results were compared with the National Committee for Clinical Laboratory Standards reference agar dilution method. The broth disk and reference breakpoint concentrations were as follows: cefoperazone, 60 and 64 or 30 and 32 micrograms/ml; cefotaxime, 30 and 32 micrograms/ml; cefoxitin, 18 and 16 micrograms/ml, respectively. Discrepant results were retested to obtain a mode. There was 99% agreement between the broth disk and reference methods for cefotaxime, 98% for cefoperazone with 60- and 64-micrograms/ml breakpoints and 91% with 30- and 32-micrograms/ml breakpoints, and 75% for cefoxitin. All but one of the strains that produced false susceptibility results by broth disk were members of the Bacteroides fragilis group, 1 with cefoperazone using the 60-micrograms/ml concentration, 14 with cefoperazone at the 30-micrograms/ml concentration, and 27 with cefoxitin. One strain of Clostridium difficile produced false susceptibility results to cefoperazone at the 30-micrograms/ml concentration. The lack of agreement between the broth disk and reference methods with cefoxitin may be a reflection of the number of isolates at the 16-micrograms/ml level and that the broth disk breakpoint was slightly higher than this concentration. Increased incubation time did not improve the results significantly.
