Objective: To study the occurrence of multiple-drug-resistant pathogens in nosocomial bloodstream infection associated with pneumonia. To evaluate prediction of multiple drug resistance by systematic surveillance cultures.
Design: A retrospective study of a prospectively gathered cohort.
Setting: Fifty-four-bed adult medical-surgical intensive care unit of a tertiary hospital.
Patients: One hundred twelve intensive care unit patients with nosocomial bloodstream infection associated with pneumonia from 1992 through 2001.
Interventions: None.
Measurements and Main Results: Concordance of blood cultures with prior surveillance culture was assessed. Surveillance cultures were taken routinely as thrice weekly urinary cultures and oral swabs, once weekly anal swabs, and thrice weekly tracheal aspirates in intubated patients. Tracheal surveillance cultures from 48 to 96 hrs before bloodstream infection and surveillance cultures from any site during the same intensive care unit episode but ≥48 hrs before bloodstream infection were evaluated separately. Forty-four bloodstream infections (39%) were caused by a multiple-drug-resistant pathogen. Multiple-drug-resistant pathogens were predicted by tracheal surveillance culture in 70% (concordant); in 15%, tracheal surveillance culture grew a multiple-drug-resistant pathogen not found in blood cultures (discordant). Multiple-drug-resistant pathogens were predicted by any surveillance culture in 88%, but these surveillance cultures grew additional multiple-drug-resistant pathogens not causing bloodstream infection in up to 46% of patients. In 86% of bloodstream infections, early (i.e., within 48 hrs) antibiotic therapy was appropriate. Patients were divided into four risk categories for multiple-drug-resistant bloodstream infection based on length of prior intensive care unit stay and prior antibiotic exposure. In patients with two risk factors, knowledge of surveillance cultures increased appropriateness of early antibiotic therapy from 75-79% to 90% (p < .05) while limiting use of broad-spectrum antibiotics such as antipseudomonal betalactams, fluoroquinolones, and carbapenems.
Conclusions: In our intensive care unit, tracheal surveillance culture predicted multiple-drug-resistant etiology of bloodstream infection associated with pneumonia in 70% of patients but yielded discordant resistant pathogens in 15%. In the subgroup of patients with two risk factors for multiple-drug-resistant infection, incorporating results of surveillance cultures moderately contributed to adequacy of early antibiotic therapy while limiting antibiotic consumption.