Predictive Value of Surveillance Cultures and Subsequent Bacteremia with Extended-Spectrum  -Lactamase-Producing Enterobacteriaceae

To the Editor—We read with great interest the article by Reddy et al. [1] about screening for extended-spectrum β-lactamase (ESBL)–producing Enterobacteriaceae by rectal surveillance cultures in high-risk patients and the subsequent acquisition of bacteremia. Their study revealed that 54.9% of the bloodstream infections caused by ESBL-producing Enterobacteriaceae occurred in patients who were not screened, whereas in patients with surveillance cultures that were positive for ESBL-producing Enterobacteriaceae, the risk of developing bloodstream infections with this multidrug-resistant (MDR) pathogen was substantial, and bloodstream infections with these MDR pathogens in patients with negative surveillance culture were more rare. On the basis of their database, it would be interesting to calculate the positive and negative predictive values of presence of ESBL-producing pathogens in rectal surveillance cultures to assess the risk of subsequent bloodstream infection being caused by this MDR pathogen. Involvement of MDR pathogens in bloodstream infection is a known risk factor for inappropriate empirical therapy and, as such, for adverse outcome. Depending on the test characteristics of surveillance cultures, knowledge of colonization by ESBL-producing pathogens by systematic surveillance cultures would identify a subset of patients who would benefit from broad-spectrum empirical coverage—for example, with use of a carbapenem as core antibiotic—as well as identify a subset of patients who could be treated with narrower spectrum antimicrobial therapy [2, 3]. In our intensive care unit, surveillance cultures are obtained for all patients at the time of admission to and during stay in the intensive care unit. These surveillance cultures consist of endotracheal aspirates (at intubation and thrice weekly, as long as patients undergo mechanical ventilation), urinary culture (on admission to the unit and thrice weekly), and nasal, oral, and rectal swabs (on admission to the unit and once weekly). This strategy was introduced in the mid-1980s to control an outbreak of infection with ESBL-producing Klebsiella pneumoniae, and it is still used for infection control purposes. Through the years, however, the data provided by surveillance strategy have been progressively used to assist the empirical prescription of antibiotic therapy in case of infection. This approach allows achievement of high rates of appropriate therapy while minimizing antibiotic selection pressure because last-line, broad-spectrum antibiotics, such as carbapenems or glycopeptides, are generally used only for patients who are colonized with MDR pathogens. This strategy has been proven of value in patients in the intensive care unit who have either nosocomial pneumonia or bacteremia [4, 5]. For example, tracheal surveillance cultures predicted the causative etiology in ∼70% of the bacteremic pneumonia episodes [5], resulting in higher rates of appropriate therapy and survival as well [5, 6]. Previously, we failed to demonstrate a significant attributable mortality among patients with bacteremia caused by gram-negative bacteria. We believe that this is due to the high rate of appropriate therapy that was achieved through this surveillance-assisted prescription strategy [7–10]. Obviously, the efficacy of such a surveillance-assisted approach depends on the positive and negative predictive value of surveillance culture. Therefore, we wonder whether Reddy et al. [1] can provide these on the basis of information included in their database. Potential conflicts of interest.All authors: no conflicts.