Underlying characteristics of patients harboring highly resistant Acinetobacter baumannii.

Background: Highly resistant (HR) Acinetobacter baumannii (AB) are frequently hospital-acquired and may be important causes of severe nosocomial infections.

Objective: Determine risk factors associated with such colonization/infection.

Method: Retrospective review in 2000 of all AB isolates from sterile (blood, cerebrospinal fluid [CSF]) and nonsterile (respiratory, urine, and miscellaneous) sites. HR was defined as resistance to amikacin and/or imipenem and/or ampicillin-sulbactam. Isolates were analyzed as representing infection or colonization. A database including prior hospitalization, prior antibiotic use, nursing home residency, and procedures undergone was compiled.

Results: One hundred twenty-two cases of AB were identified. Eighty-four met the definition of HR; 6 (4.9%) were completely resistant to all antibiotics. Four (4.7%) isolates were from sterile body sites (3 blood, 1 CSF); 43 (51.2%) were from respiratory sites; 20 (23.8%) were from urinary sites; and 17 (20.2%) were from "other" sites. Only 4 (20%) of the urinary, 6 (35.2%) of the miscellaneous, and 23 (53.4%) of the respiratory isolates were deemed true pathogens; all blood/CSF isolates were considered pathogens. Associated risk factors included prior antibiotic usage (71%); prior hospitalization (24%); prior nursing home residency (34%); ventilator use (77%); tracheostomy placement (56%); and Foley catheterization (85%). Twenty-seven (63%) of 43 respiratory, 8 (40%) of 20 urinary, and 6 (35%) of 17 "other" body isolates were treated. Outcome was not statistically significant in treated versus untreated patients. All patients with CSF/blood isolates underwent successful microbiologic eradication with 50% survival. The overall mortality rate was 10%.

Conclusion: Antibiotics, Foley catheters, and tracheostomy/ventilator usage were strongly associated with AB isolation. Prior hospitalization and nursing home residency were less common risk factors. Outcome was not different in treated versus untreated patients, indicating colonization is a marker of severe illness but is not necessarily causal.