Validity of the Current susceptibility breakpoint criteria for 9 beta-lactam antimicrobials and performance of proposed alternative breakpoints to improve prediction of clinical outcomes were analyzed by testing a contemporary collection of 350 Enterobacteriaceae, enriched for an overrepresentative collection of 70 (20.0%) strains producing extended-spectrum beta-lactamases (ESBLs). The majority of the strains were isolated from bloodstream infections (83.7% of the entire collection and 85.7% of the ESBL subset). The 9 beta-lactam antimicrobials analyzed were aztreonam, cefepime, cefotaxime, cefotetan, cefoxitin, ceftazidime, ceftriaxone, ceftizoxime, and cefuroxime. Reference broth microdilution MIC results were compared with those zone diameters obtained by the standardized disk diffusion test. The correlation coefficient (r) was acceptable for all antimicrobials, ranging from 0.87 (cefotetan) to 0.97 (aztreonam, cefotaxime, and ceflazidime). Using the current susceptible breakpoint criteria for Enterobacteriaceae, the intermethod categorical agreement ranged from 90.6% (cefotaxime) to 98.0% (ceftazidime). Very major (false-susceptible by disk test) and major errors (false-resistant) were nil (0.0%) for 6 of the 9 beta-lactams. Minor error rates ranged from only 0.9% (cefotetan) to 9.4% (cefotaxime). The proposed MIC breakpoint criteria (generally lower) adjusted to levels to accurately detect ESBL-producing strains and better predict clinical outcomes, also had acceptable intermethod concordance ranging from 90.6% (cefotetan) to 100.0% (ceftazidime). Remarkably, an improvement in the intermethod categorical agreement ranging from + 1.7% to + 8.3% was observed for 7 of the 9 antimicrobials, including the ESBL index or screening compounds (aztreonam, ceftazidime, cefotaxime, and ceftriaxone). No change in the breakpoint criteria, with removal of the intermediate category was tentatively proposed for cefoxitin and cefuroxime, resulting in an increase of the serious intermethod errors (very major and major), but the absolute intermethod agreement remained highly acceptable at 90.6% to 93.4%. Although the current breakpoint criteria remain acceptable in minimizing intermethod discords, the alternative susceptible breakpoint criteria proposed by combining pharmacokinetic/pharmacodynamic (PK/PD), microbiology MIC population analyses, and clinical success parameters possess improved intermethod agreement for the ESBL screening drugs and 5 other broad-spectrum beta-lactani compounds. The Clinical Laboratory Standards Institute (formerly, the National Committee for Clinical Laboratory Standards) should consider these changes to facilitate the detection of all Enterobacteriaceae with low PK/PD target attainment rates, therefore having the potential for suboptimal responses with usual therapeutic dosing. (c) 2005 Elsevier Inc. All rights reserved.
