Objectives: We assessed the prognostic significance of infections in relation to current prognostic scores and explored if infection could be considered per se a distinct clinical stage in the natural history of cirrhosis.
Methods: We included consecutive patients with cirrhosis admitted to a tertiary referral liver unit for at least 48 h over a 2-year period. Diagnosis of infection was based on positive cultures or strict established criteria. We used competing risk analysis and propensity score matching for data analysis.
Results: 501 patients (63% male, 48% alcoholic liver disease, median Model of End-stage Liver Disease (MELD)=17) underwent 781 admissions over the study period. Portal hypertensive bleeding and complicated ascites were the commonest reasons of admission. The incidence of proven bacterial infection was 25.6% (60% community acquired and 40% nosocomial). Survival rates at 3, 6, 12, and 30 months were 83%, 77%, 71%, and 62% in patients without diagnosis of infection, vs. 50%, 46%, 41%, and 34% in patients with diagnosis of infection. Overall survival was independently associated with MELD score (hazards ratio (HR) 1.099), intensive care (ITU) stay (HR 1.967) and bacterial infection (HR 2.226). Bacterial infection was an independent predictor of survival even when patients who died within the first 30 days were excluded from the analysis in Cox regression (HR 2.013) and competing risk Cox models in all patients (HR 1.46) and propensity risk score-matched infected and non-infected patients (HR 1.67).
Conclusions: Infection most likely represents a distinct prognostic stage of cirrhosis, which affects survival irrespective of disease severity, even after recovery from the infective episode.