Correlation of various in vitro testing methods with clinical outcomes in patients with Bacteroides fragilis group infections treated with cefoxitin: a retrospective analysis.

There is limited information regarding the correlation of anaerobic susceptibility testing and outcome in the treatment of Bacteroides fragilis infections. We retrospectively analyzed the clinical outcomes of B. fragilis infections in patients treated with cefoxitin; the analysis was blinded for susceptibility results. Isolates of B. fragilis were tested by multiple agar dilution methods, disk elution, and commercial broth microdilution methods. Of 19 patients analyzed, 11 were cured and 8 were treatment failures. No significant differences existed between the groups with respect to age, sex distribution, weight, APACHE II score, dose of cefoxitin, or bacteremia. Failure was associated with a longer cefoxitin dosing interval (P = 0.019), a longer duration of hospitalization (P = 0.038), and decreased duration of cefoxitin treatment (P = 0.05). Four agar dilution systems (brucella plus blood, Wilkins-Chalgren, Wilkins-Chalgren plus blood, brain heart infusion plus blood) and two broth systems (Wilkins-Chalgren microdilution and a commercial system [Micromedia; Beckman, Carlsbad, Calif.]) all demonstrated lower geometric mean MICs for isolates from the group of patients that could be cured. Only the commercial broth microdilution medium (Micromedia) demonstrated a significantly reduced geometric mean MIC (P = 0.056). By using a logistic regression analysis, the shorter cefoxitin dosing interval (P = 0.0004) and the lower geometric mean MIC (P = 0.0088) in the commercial broth microdilution system were shown to be independent predictors of treatment success. These data suggest that the time that the concentration of cefoxitin is over the MIC for B. fragilis may be an important predictor of treatment success.