Colonization with broad-spectrum cephalosporin-resistant Gram-negative bacilli in intensive care units during a nonoutbreak period

Publisher Website
Google Scholar
Abstract
To define the epidemiology of broad-spectrum cephalosporin-resistant gram-negative bacilli in intensive care units (ICUs) during a nonoutbreak period, including the prevalence, the risk factors for colonization, the frequency of acquisition, and the rate of infection. Prospective cohort study. Tertiary care hospital. Consecutive patients admitted to two surgical ICUs. Serial patient surveillance cultures screened for ceftazidime (CAZ) resistance, antibiotic and hospital exposure, and infections. Of the 333 patients enrolled, 60 (18%) were colonized with CAZ-resistant gram-negative bacilli (CAZ-RGN) at admission. Clinical cultures detected CAZ-RGN in only 5% (3/60) of these patients. By using logistic regression, CAZ-RGN colonization was associated with duration of exposure to cefazolin (odds ratio, 10.3; p < or = .006) and to broad-spectrum cephalosporins/penicillins (odds ratio, 2; p < or = .03), Acute Physiology and Chronic Health Evaluation III score (odds ratio, 1.2; p < or = .008), and previous hospitalization (odds ratio, 3.1; p < or = .006). Of the 100 patients who remained in the surgical ICU for > or = 3 days, 26% acquired a CAZ-RGN. Of the 14 infections caused by CAZ-RGN, 11 (79%) were attributable to the same species present in surveillance cultures at admission to the surgical ICU. Colonization with CAZ-RGN was common and was usually not recognized by clinical cultures. Most patients colonized or infected with CAZ-RGN had positive surveillance cultures at the time of admission to the surgical ICU, suggesting that acquisition frequently occurred in other wards and institutions. Patients exposed to first-generation cephalosporins, as well as broad-spectrum cephalosporins/penicillins, were at high risk of colonization with CAZ-RGN. Empirical treatment of nosocomial gram-negative infections with broad-spectrum cephalosporins, especially in the critically ill patient, should be reconsidered.

This publication has 28 references indexed in Scilit: