Background: Antibiotic exposures have been described as risk factors for carbapenem-resistant Acinetobacter baumannii (CRAB) clinical infections. We aim to better understand this association by exploring novel ways to handle the exposure variables.
Methods: This retrospective cohort from 11/1/10 to 11/30/11 included consecutive patients admitted to the trauma intensive care unit (TICU), whom underwent rectal [and tracheal if intubated] surveillance cultures upon unit admission and weekly thereafter. Patients positive on admission surveillance were excluded. Hazard ratios for the development of CRAB infections were calculated using time dependent and time-independent antibiotic exposures. Time independent variables were: 1. binary variables (received versus not) determined for each antibiotic and for all antibiotics (any antibiotic received versus none) and 2. total sum of grams and 3. total sum of DDDs received during the observation period (for each antibiotic and for all [DDDs]). Time dependent antibiotic exposures included: 1. time after the first day of each antibiotic (0 vs. 1), 2. daily cumulative exposure (daily sum in grams or DDDs), and 3. immediate hazard ratios after a each discrete day of antibiotics. APACHE on admission, age, and gender were also analyzed using Cox models.
Results: A total of 364 patients were analyzed. Sixty patients developed CRAB infections, and 49 patients had positive surveillance cultures. Using binary variables, only quinolones had a [protective] association. When total antibiotics received were analyzed, piperacillin-tazobactam, cefazolin, vancomycin, and the sum of all antibiotics received (in DDDs) were independently associated with a protective effect. Using time-dependent exposures, carbapenems were associated with development of clinical infections. However, this association disappeared in the multivariable model. The impact of carbapenem exposures on the immediate hazard of developing clinical infections after each single day of antibiotic exposure is shown in Figure 1.
Conclusion: Antibiotic exposures are associated with the development of CRAB clinical infections, but these associations vary based on the exposure variables used in the models.
R. Latibeaudiere, None
S. Tarima, None
L. S. Munoz-Price, None