Objectives: To determine what physiological and biochemical factors predict development of bacteremia in nontransplanted patients with acute on chronic liver failure and, on diagnosis of bacteremia, what is the natural history of bacteremic patients versus control subjects (acute on chronic liver failure).
Interventions: None.
Design: Retrospective analysis of data collected prospectively and entered into a dedicated physiology database.
Setting: Specialist liver intensive therapy unit.
Patients: Critically ill non-transplanted patients with acute on chronic liver failure admitted between January 2003 and July 2005.
Measurements and main results: One hundred eighty-four patients were defined with acute on chronic liver failure; 67 (36%) had bacteremia. One hundred seventeen (64%) patients did not (acute on chronic liver failure). Fifty-eight percent of isolates were Gram-negative organisms, 36% were Gram-positives, and 6% fungemia. Median time to first bacteremia was 8 days (range, 3-12 days). On admission (univariate), bacteremic patients had significantly higher Modified End Stage Liver Disease scores (27 vs. 24, p = .037), Acute Physiology and Chronic Health Evaluation II scores (23 vs. 21, p = .049), and greater degrees of encephalopathy (Glasgow Coma Scale score 10 vs. 12, p = .001). During their liver intensive therapy unit course, bacteremic patients had significantly greater requirements for renal replacement therapy (64% vs. 49%, p = .043), mechanical ventilation (88% vs. 68%, p = .002), and a longer median liver intensive therapy unit stay (16 vs. 5 days, p < .001). Survival to hospital discharge was worse in the bacteremic group (25% vs. 56%, p < .001). Multivariate analysis (logistic regression) was performed separately modeling with Acute Physiology and Chronic Health Evaluation II and Modified End Stage Liver Disease. In the first model, Acute Physiology and Chronic Health Evaluation II (odds ratio 1.24) and bacteremia (2.24) were independent predictors of mortality. In the later model, Modified End Stage Liver Disease (odds ratio, 1.06), requirement for renal replacement therapy (3.08), Glasgow Coma Scale (0.72), and bacteremia (2.30) were significant. Both models performed similarly (Modified End Stage Liver Disease area under the receiver operating characteristic curve, 0.864; Acute Physiology and Chronic Health Evaluation II, 0.862).
Conclusions: In nontransplanted patients with acute on chronic liver failure, bacteremia was associated with increased severity of illness on admission, greater requirements for organ support, and independently adversely impacted on survival. Higher Acute Physiology and Chronic Health Evaluation II and Modified End Stage Liver Disease scores were also independently predictive of mortality.