Background & aims: A staged prognostic model of cirrhosis based on varices, ascites, and bleeding has been proposed. We analyzed data on infections in patients with cirrhosis to determine whether it is also a prognostic factor.
Methods: Studies were identified by MEDLINE, EMBASE, COCHRANE, and ISI Web of Science searches (1978-2009); search terms included sepsis, infection, mortality, and cirrhosis. Studies (n = 178) reporting more than 10 patients and mortality data were evaluated (225 cohorts, 11,987 patients). Mortality after 1, 3, and 12 months was compared with severity, site, microbial cause of infection, etiology of cirrhosis, and publication year. Pooled odds ratio of death was compared for infected versus noninfected groups (18 cohorts, 2317 patients).
Results: Overall median mortality of infected patients was 38%: 30.3% at 1 month and 63% at 12 months. Pooled odds ratio for death of infected versus noninfected patients was 3.75 (95% confidence interval, 2.12-4.23). In 101 studies that reported spontaneous bacterial peritonitis (7062 patients), the median mortality was 43.7%: 31.5% at 1 month and 66.2% at 12 months. In 30 studies that reported bacteremia (1437 patients), the median mortality rate was 42.2%. Mortality before 2000 was 47.7% and after 2000 was 32.3% (P = .023); mortality was reduced only at 30 days after spontaneous bacterial peritonitis (49% vs 31.5%; P = .005).
Conclusions: In patients with cirrhosis, infections increase mortality 4-fold; 30% of patients die within 1 month after infection and another 30% die by 1 year. Prospective studies with prolonged follow-up evaluation and to evaluate preventative strategies are needed.
Copyright © 2010 AGA Institute. Published by Elsevier Inc. All rights reserved.