The carbapenem breakpoints set by different organizations for Acinetobacter are discordant, but supporting clinical data are lacking. This study aimed to provide the first clinical outcome data to support the carbapenem breakpoints for Acinetobacter baumannii (Ab) group in patients with bacteremia. This study included 117 adults who received carbapenems for treatment of Ab group bacteremia in Taipei Veterans General Hospital over an 8-year period. We analyzed 30-day mortality rates among patient groups acquiring isolates with different carbapenem minimal inhibitory concentrations (MICs). The carbapenem MIC breakpoint derived from classification and regression tree (CART) analysis to delineate the risk of 30-day mortality was between MICs of </= 4 mg/L and >/= 8 mg/L. Mortality rate was higher in patients acquiring isolates with carbapenem MIC >/= 8 mg/L than </= 4 mg/L, by bivariate (54.9% [28/51] vs 25.8% [17/66]; P = 0.003) and survival analysis (P = 0.001 by log-rank test). Multivariate analysis using logistic regression and Cox regression models including severity of illness indices demonstrated that treating patients with Ab group bacteremia caused by isolates with a carbapenem MIC >/= 8 mg/L with carbapenem was an independent predictor of 30-day mortality (odds ratio, 5.125; 95% confidence interval [CI], 1.946-13.498; P = 0.001, and hazard ratio, 2.630; 95% CI, 1.431-4.834; P = 0.002, respectively). The clinical outcome data confirmed that isolates with MIC </= 4 mg/L were susceptible to carbapenem, and those with MIC >/= 8 mg/L were resistant in patients with Ab group bacteremia.
