Multicenter Study of Clinical Features of Breakthrough Acinetobacter Bacteremia during Carbapenem Therapy

Abstract

Breakthrough <i>Acinetobacter</i> bacteremia during carbapenem therapy is not uncommon, and it creates therapeutic dilemmas for clinicians. This study was conducted to evaluate the clinical and microbiological characteristics of breakthrough <i>Acinetobacter</i> bacteremia during carbapenem therapy and to assess the efficacy of various antimicrobial therapies. We analyzed 100 adults who developed breakthrough <i>Acinetobacter</i> bacteremia during carbapenem therapy at 4 medical centers over a 6-year period. Their 30-day mortality rate was 57.0%, and the carbapenem resistance rate of their isolates was 87.0%. Among patients with carbapenem-resistant <i>Acinetobacter</i> bacteremia, breakthrough bacteremia during carbapenem therapy was associated with a significantly higher 14-day mortality (51.7% versus 37.4%, respectively; <i>P</i> = 0.025 by bivariate analysis) and a higher 30-day mortality (<i>P</i> = 0.037 by log rank test of survival analysis) than in the nonbreakthrough group. For the treatment of breakthrough <i>Acinetobacter</i> bacteremia during carbapenem therapy, tigecycline-based therapy was associated with a significantly higher 30-day mortality (80.0%) than those with continued carbapenem therapy (52.5%) and colistin-based therapy (57.9%) by survival analysis (<i>P</i> = 0.047 and 0.045 by log rank test, respectively). Cox regression controlling for confounders, including severity of illness indices, demonstrated that treatment with tigecycline-based therapy for breakthrough <i>Acinetobacter</i> bacteremia was an independent predictor of 30-day mortality (hazard ratio, 3.659; 95% confidence interval, 1.794 to 7.465; <i>P</i> < 0.001). Patients with breakthrough <i>Acinetobacter</i> bacteremia during carbapenem therapy posed a high mortality rate. Tigecycline should be used cautiously for the treatment of breakthrough <i>Acinetobacter</i> bacteremia that develops during carbapenem therapy.