To create a clinical tool based on institution-specific risk factors to estimate the probability of carbapenem resistance among Pseudomonas aeruginosa isolates obtained from infected patients. By better estimating the probability of carbapenem resistance on the basis of patient-specific factors, clinicians can refine their empirical therapy for P. aeruginosa infections and potentially maximize clinical outcomes by increasing the likelihood of appropriate empirical antimicrobial therapy.

A retrospective, cross-sectional study.

Tertiary care academic hospital.

All adult inpatients who had a respiratory tract infection due to P. aeruginosa between January 2001 and June 2005.

Data on demographic characteristics, antibiotic history, and microbiology were collected. Log-binomial regression was employed to identify predictors of carbapenem resistance among P. aeruginosa isolates and to devise the clinical prediction tool.

Among 351 patients with P. aeruginosa infection, 44% were infected with carbapenem-resistant P. aeruginosa strains. Independent predictors of carbapenem resistance were prior receipt of mechanical ventilation for 11 days or more, prior exposure to fluoroquinolones for 3 days or more, and prior exposure to carbapenems for 3 days or more.

With carbapenem resistance rates among P. aeruginosa isolates on the rise at our institution, the challenge was to identify patients for whom carbapenems would remain an effective empirical agent, as well as the patients at greatest risk for infection with carbapenem-resistant strains. The clinical prediction tool accurately estimated carbapenem resistance among this risk-stratified cross-sectional study of patients with P. aeruginosa infection. This tool may be an effective way for clinicians to refine their selection of empirical antibiotic therapy and to maximize clinical outcomes by increasing the likelihood of appropriate antibiotic treatment.

Patients colonized or infected with vancomycin-resistant enterococcus and methicillin-resistant Staphylococcus aureus may be at risk of acquiring vancomycin-resistant S. aureus if the vanA gene is transferred from vancomycin-resistant enterococcus to methicillin-resistant S. aureus..

Our goal was to identify risk factors for cocolonization or coinfection (CC/CI) with vancomycin-resistant enterococcus and methicillin-resistant S. aureus.

We conducted a descriptive, epidemiologic study of all patients with CC/CI identified from January 1998 to May 2003 and a nested case-control study of a cohort of patients hospitalized in the burn and wound unit.

We conducted our study in a 813-bed tertiary care university teaching hospital.

The study population consisted of patients found to have CC/CI during the study period.

Descriptive epidemiologic data were collected from hospital records of all patients identified as having CC/CI. A subset of patients hospitalized in the burn and wound unit were included in a case-control study.

CC/CI was detected in 71% of the patients during a single hospital stay. The burn and wound unit, which does active surveillance for both organisms, and the general medicine unit, which does not do active surveillance for either organism, cared for more than one-half of these patients. Among patients being cared for in the burn and wound unit, having exposure to 2 or more invasive devices (central venous catheters, indwelling urinary catheters, and enteral feeding tubes) and renal insufficiency were independent risk factors for CC/CI.

Patients with CC/CI are the population at greatest risk for vancomycin-resistant S. aureus colonization or infection. The number of invasive devices to which patients are exposed and, thus, possibly the patients' underlying severity of illness, as well as renal insufficiency, appear to be risk factors for CC/CI.

To investigate an outbreak of Staphylococcus aureus surgical-site infections.

Case–control study.

Pediatric cardiothoracic surgery service of a tertiary-care university medical center.

Molecular typing was used to identify healthcare workers who carried the epidemic strain.

Three children acquired surgical-site infections caused by a single strain of S. aureus. Fourteen (25%) of the staff members in the operating room and 17 (11%) on nursing units carried the epidemic strain (P = 01). A case–control study identified 4 healthcare workers who were associated statistically with the outbreak, 2 of whom (a cardiothoracic surgeon and a perfusionist) carried the epidemic strain in their nares. The surgeon also carried the epidemic strain on his hands. Each staff member who carried the epidemic strain was treated with mupirocin; those carrying the strain on their hands were required to wash their hands with chlorhexidine. The surgeon was not allowed to perform surgery until 2 of his hand cultures did not grow S. aureus.

Only three children were infected with the epidemic strain, but it was disseminated widely among staff who cared for children who underwent cardiothoracic surgery. No additional cases were identified after staff members who carried the epidemic strain were decolonized. Both classic epidemiologic methods and molecular typing techniques were necessary to identify the source and extent of this outbreak.