Clinical and microbiological characterization of carbapenem-resistant Acinetobacter baumannii bloodstream infections

The incidence of carbapenem-resistant Acinetobacter baumannii infection is increasing, which might be associated with high morbidity and mortality among critically ill patients with limited therapeutic options. This study was conducted to evaluate the clinical and microbiological features of carbapenem-resistant A. baumannii bacteraemia. The medical records of 28 adult patients with this bacteraemia admitted to Korea University Guro Hospital; from January 2005 through December 2010, were reviewed. Using the 28 bloodstream isolates, we intended to detect genes encoding carbapenemases, and investigate the inoculum effect on each of the antimicrobial agents rifampicin, imipenem, colistin and tigecycline. With one blood isolate from a patient with pneumonia, rifampicin-inducible resistance was examined using the experimental mouse pneumonia model. Out of 28 carbapenem-resistant A. baumannii bloodstream infections (BIs), the most common primary focus was the central venous catheter (35.7%) and then the lung (32.1%). The 30 day overall mortality was 53.6%; in most cases (80%) the patients died within 10 days after the onset of the bacteraemia. By univariate analysis, inappropriate antimicrobial therapy (73.3 vs 30.8%, P=0.02), mechanical ventilation (53.3 vs 15.4%, P=0.04) and a high Pitt bacteraemia score (4.9 +/- 1.9 vs 2.2 +/- 1.2, P<0.01) were statistically significant risk factors for mortality, while only a high Pitt bacteraemia score (odds ratio 2.6; 95% confidence interval 1.1-6.5) was independently associated with 30 day mortality by multivariate analysis. All 28 isolates had the bla(OXA-51)-like gene with upstream ISAbal, 2 of which additionally had the bla(OXA-58)-like gene and the bla(OXA-23)-like gene. Inoculum effect and rifampicin inducible resistance were not detected. Considering the rapid progression to death in carbapenem-resistant A. baumannii BIs, early empirical antibiotic therapy would be warranted based on the local microbiological data in each hospital.