Prospective study of bacterial infection in acute liver failure: an analysis of fifty patients.

Fifty consecutive patients admitted with acute liver failure, minimal grade II encephalopathy, were studied prospectively to determine the incidence, timing and cause of bacterial infection, the relationship to clinical criteria for infection; and the influence of early microbiological diagnosis on clinical outcome. There were 53 proven bacterial infections in 40 patients, whereas in 5 of the remaining 10 patients infection was suspected on clinical grounds in the absence of significant cultures. Seven patients (14%) had more than one bacterial infection, and four patients had simultaneous infections caused by different organisms at each site. Fourteen infections (26.4%) were associated with bacteremia, and in six of these no source was found. Twenty-five infections (47.1%) arose from the respiratory tract, 12 (22.6%) from the urinary tract and 2 (3.7%) from central venous cannulas. Thirty-seven (69.8%) of the 53 infections were due to gram-positive bacteria; Staphylococcus aureus accounted for 19 (35.8%) of all the infections. Thirty patients died (60%), 28 of whom had bacterial infection at some time; in 24 of these the infection was diagnosed less than 24 hr before death. All nine deaths that occurred more than 7 days after admission were directly attributable to microbial infection. Clinical features such as elevated temperature and elevated peripheral white blood cell count were poor indicators of bacterial infection because these were absent in 30.2% of cases. These data show that there is a high incidence of bacterial infection early in the course of acute liver failure and suggest that prophylactic antimicrobial therapy, although unproven, might be justified.