External validation and improvement of LiFe score as a prediction tool in critically ill cirrhosis patients.

ABSTRACT

AIM: The LiFe (liver, injury, failure, evaluation) score, calculated according to arterial lactate, total bilirubin, and international normalized ratio (INR), is a novel score for risk prediction in intensive care unit (ICU) patients with cirrhosis. The present study aimed to externally validate and optimize the LiFe score for predicting outcomes in critically ill cirrhosis patients. METHODS: The study used the single-center database Medical Information Mart for Intensive Care-III (MIMIC-III) for analysis. A total of 536 critically ill cirrhosis patients from the MIMIC-III database were analyzed. Routine clinical and laboratory variables were included to compare survivors with non-survivors. The LiFe score was then regraded into three groups to calculate the optimal cut-off values. RESULTS: In-ICU mortality occurred in 169 (31.5%) of the patients. Survivor and non-survivor cohorts were similar in age, gender, and etiology of cirrhosis. Multivariate analyses of in-ICU mortality identified four independent variables total bilirubin, creatinine, INR, and arterial lactate. An external validation of the LiFe score showed good accuracy for predicting in-ICU mortality with an area under the receiver operating characteristic curve of 0.708. In addition, a significant positive correlation exists between LiFe score and acute-on-chronic liver failure grade (r = 0.393, P < 0.001). A log-rank test comparing the strata of simplified LiFe scores found that in-ICU mortality rates were 16.8%, 27.7%, and 51.7%, respectively, among patients in the three simplified risk categories. CONCLUSIONS: The LiFe score, based on laboratory tests, can be useful as a preliminary and convenient scoring tool in a broad cohort of critically ill cirrhosis patients. Simplified risk categories to stratify patients into three groups improves its feasibility and generalizability for clinical application.