This study was performed at the Tel-Aviv Sourasky Medical Center, Israel, a 1,200–bed tertiary care teaching hospital with 70,000 admissions annually. Approximately 82,500 clinical microbiologic cultures are processed annually. We designed this as a matched case-control study to identify the individual risk factors for having MDR A. baumannii. We also performed epidemiologic investigations and genetic typing of the organisms to clarify the spread of this nosocomial pathogen.

Case-patients were defined as patients from whom MDR A. baumannii was isolated from any clinical culture (not surveillance cultures) during a 6–month period, from January 1, 2001, to June 30, 2001. A control patient was matched to each study patient on temporospatial factors as previously described (24). Briefly, controls were randomly chosen from the list of patients who stayed on the same ward in the same calendar month as the matched case-patient and who were hospitalized for at least the same number of days by the day the culture yielded MDR A. baumannii in the study patient. Controls were not MDR A. baumannii positive (i.e., the patient’s samples were cultured, and either non-MDR A. baumannii or no A. baumannii was isolated, or the patient’s samples were never cultured). Random control selection was performed by creating a list of all possible controls, assigning each candidate a random number, and choosing the highest random number (without replacement).

Case-patients and control patients were included only once in the study. Data were collected from the patients’ records and from hospital computerized databases into a pre-prepared electronic questionnaire (Microsoft Access, Microsoft Corp., Redmond, WA, USA). The parameters registered for each patient (case-patients and controls) were age, sex, habits of smoking and alcohol consumption before hospitalization, cause and ward of hospitalization, transfer from another institution or ward within our institution, intensive care unit (ICU) stay, underlying disorders, immunosuppressive therapy, severity of illness as defined by the McCabe score (25), functional capacity and neurologic condition at time of isolation of A. baumannii, Foley catheter, invasive devices, surgery, mechanical ventilation, dialysis, infection, and antimicrobial drug therapy. Only variables occurring before inclusion in the study (culture day for case-patients and match day for controls) were analyzed as possible risk factors. A. baumannii was isolated from clinical specimens submitted to the microbiology laboratory and identified by using the Gram-Negative Identification Panel (Microscan, Dade Behring Inc., Sacramento, CA, USA). This system may not distinguish between closely related genotypic strains of Acinetobacter, and thus, some of these organisms may belong to these closely related strains. Susceptibilities were determined by automated microdilution broth testing (Neg/Urine Combo panel, Dade Behring Inc.). Resistance to imipenem and meropenem was confirmed by using Kirby-Bauer disk diffusion, according to the National Council for Clinical Laboratory Standards (NCCLS) guidelines. A. baumannii isolates were collected prospectively and stored at –70°C for further work-up.

Isolates of our patients, when available, were kept frozen at –70ºC and genetically characterized with PFGE. DNA preparation and cleavage with 20 U of ApaI endonuclease (New England Biolabs, Beverly, MA, USA) were preformed as previously described (26). Electrophoresis was performed in a 1% agarose gel (BMA products) prepared and run in 0.5 x Tris-borate-EDTA buffer on a CHEF-DR III apparatus (Bio-Rad Laboratories, Hercules, CA, USA). The initial switch time was 5 s, the final switch time was 35 s, and the run time was 23 h at 6 V/cm. Gels were stained in ethidium bromide, destained in distilled water, and photographed by using a Bio-Rad GelDoc 2000 camera. DNA patterns were analyzed visually and by using Diversity software (Bio-Rad). PFGE DNA patterns were compared and interpreted according to the criteria of Tenover et al. (27). The obtained PFGE DNA patterns were used to cluster the clones of the A. baumannii clinical isolates that were included in the study.

Statistics were run in Stata version 7 (Stata Corp., College Station, TX). All analyses were matched to correspond to the study design. All variables were examined by univariate analysis with the McNemar test and paired Student t test. Variables with a p value <0.2 in the univariate analysis were included in the multivariate model. Risk factors were examined by using conditional logistic regression. A final model was built that included all the variables with a p value <0.2. Variables that were not retained in the model by this procedure were then tested for confounding by adding them 1 at a time to the model and examining their effects on the β coefficients. Variables which caused substantial confounding (change in β coefficient of >10%) were included in the final model. After constructing the explanatory model, the effect of exposure to antimicrobial agents (i.e., antimicrobial treatment before inclusion in the study) was examined by adding them to the model.

In addition to examining statistical significance and confounding, the effect modification between variables was evaluated by testing appropriate interaction terms for statistical significance. Colinearity was examined by replacing variables with each other and examining the effect on the model. All statistical tests were 2-tailed. A p value <0.05 was considered significant.

Among the 118 case-patients, the first MDR A. baumannii in 104 (88%) was isolated after more than 72 h of hospitalization (mean 17.5 ± 23.7 days). Among the other 14 case-patients, 12 were admitted from another institution or had been hospitalized recently. No nosocomial origin was documented in 2 cases. A. baumannii was initially isolated in 27 different wards. Figure 1 shows the case distribution among them. A higher concentration of patients was clearly evident in 3 wards: the general ICU (ward “I”, 16 cases), and two internal medicine wards (ward Q, 10 cases, and ward W, 9 cases).

The time distribution of new cases is presented in Figure 2. We did not find any aggregate of cases within a specific ward at any specific time. The occurrence during the months February, May, and June was lower than during January, March, and April. This circumstance is not explained by differences in infection control measures recommended; during the entire study period this included contact isolation of every patient from whom MDR A. baumannii was isolated and cohorting of case-patients if single patient rooms were not available. A statement was added to the culture result: “MDR organism; contact isolation is required.” Epidemiologic nurses checked with the ward to confirm that the patient was isolated. On certain floors surveillance (nose, forearms, armpits, and perirectal swabs) and environmental cultures (using swabs, contact plates, and direct culturing of fluids) were performed to try and identify a reservoir of organism. In this study, case-patients and controls were matched by ward and calendar time to focus on individual risk factors and not on differences between wards and temporal changes.

A total of 51 unique patient MDR strains were available for further study, and they were analyzed by using PFGE. We identified 10 distinct clones of A. baumannii. Figure 3 shows 6 different PFGE-defined clones, each having from 1 to 4 subtypes showing a 1- to 2-band difference. Two of the 10 different clones dominated: 22 case-patients had clone A and 10 case-patients had clone B, although no specific clone dominated in a specific ward but rather each clone was spread among several wards during the entire study period (Table 1). We also found various antimicrobial drug susceptibility phenotypes (all belonging to our definition of MDR) within each PFGE clone, but almost all cases of carbapenem resistance belonged to clone A.

The study patients’ characteristics are displayed in Table 2. Case-patients were similar to their matched control patients with respect to mean age (67.7 vs 64.4 years) and sex distribution (men, n = 71 [60%] vs controls, n = 59 [50%]). The groups were also similar in habits of smoking and alcohol consumption and in the occurrence of coexisting conditions of lung disease, diabetes, kidney, liver disease, malignancy, and posttransplantation condition. The groups differed in the prevalence of ischemic heart disease: study case-patients 69% vs. controls 52% (OR 2.33, p = 0.006).

Hospital events (before study entry) differed between case-patients and controls. Case-patients were more likely to have received mechanical ventilation (OR 2.9, p = 0.001), to be treated with metronidazole (OR 1.9, p = 0.038), and to have a Foley catheter (OR 2.42, p = 0.005). They were less likely to have had another bedside surgical procedure before the isolation of A. baumannii (OR 0.53, p = 0.035).

Several variables tended to be more associated with case-patients, but the values did not reach statistical significance: admission from another institute (OR 2.1, p = 0.06), unconsciousness (OR 0.706, p = 0.07), and previous use of third-generation cephalosporin (OR 1.63, p = 0.093) and of macrolides (OR 2.25, p = 0.056). A matched multivariate model, adjusted for the hospital length of stay, was developed by using conditional logistic regression (Table 3). The variables that were identified by this model as being significant risk factors for MDR A. baumannii were male sex (OR 3.8, p = 0.002), ischemic heart disease (OR 3.3, p = 0.005), mechanical ventilation (OR 6.2, p < 0.001), and home antimicrobial drug use (OR 4.7, p = 0.018). Two agents used in the hospital were associated with MDR A. baumannii: metronidazole was identified as a risk factor (OR 2.3, p = 0.018), and the penicillin group was identified as having a protective effect (OR = 0.38, p = 0.029).