BACKGROUND
An increasing number of patients with alcoholic liver disease (ALD) are being referred for critical care support, but limited information is available on their short- and medium-term outcomes. This study aimed to determine mortality rates, identify optimal predictors of prognosis, and determine the appropriate time to apply these predictors in patients with ALD admitted to intensive care unit (ICU).

METHODS
In this retrospective study, patients admitted to ICU between 2009 and 2012 with a primary diagnosis of ALD were included. Survival was calculated using the Kaplan-Meier method, risk factors for death determined by logistic regression analysis, and discriminative capacity of models using receiver operating characteristic curves.

RESULTS
Of 170 patients admitted with liver disease, 62 met the inclusion criteria. Survival rates in the ICU, in hospital, and at 6 months were 40.3% (95% confidence interval [CI], 30.7%-49.9%), 35.5% (95% CI, 25.35%-45.65%), and 29% (95% CI, 19.4%-38.6%), respectively. Multiple linear regression analysis of day 1 variables produced an equation with Sequential Organ Failure Assessment score and lactate as significant predictors of mortality; this model had an area under the receiver operating characteristic curve of 0.93. A score greater than 12 in this model correlated with a mortality of more than 80% at all time points and was more accurate than any other score examined.

CONCLUSION
Patients admitted to ICU with ALD have a very high inhospital mortality. A combination of the established Sequential Organ Failure Assessment score and lactate provided the most accurate predictor of outcome on day of admission and at all subsequent time points.