Background: For mortality prediction models of decompensated liver cirrhosis to be valid, re-evaluations of score performances are needed. This research aimed to assess the prognostic scores performance in the prediction of mortality among decompensated cirrhotic patients admitted to the ICU, involving Acute Physiology and Chronic Health Evaluation II (APACHE II), sequential organ failure assessment (SOFA), Chronic Liver Failure Consortium acute-on-chronic liver failure (CLIF-C ACLF), Child-Turcotte-Pugh (CTP), Model for End-Stage Liver Disease (MELD), MELD Sodium (MELD-Na), MELD lactate, Glasgow Coma scale (GCS).

Methods: In an observational descriptive cohort study, we recruited 80 patients with decompensated liver cirrhosis who were critically ill and admitted to the ICU. Scores of the APACHE II, Child-Pugh, MELD, MELD-Na, and MELD-lactate were used to evaluate the severity of the disease.

Results: Statistically significant differences were revealed between ICU outcomes and different prognostic scores; as the median of CTP score, MELD, MELD Na, MELD Lactate (MELD.L), S. Lactate, SOFA, APACHE II & CLIF-C ACLF were higher among deceased patients, while the median of GCS & MAP was lower among deceased patients. By Multivariate regression analysis, low GCS (Odd Ratio:0.4), high CTP (OR: 2.37), and high APACHE II (OR: 2.44) were independent predictors of mortality (p<0.001), whereas serum sodium, albumin, platelets, hemoglobin, and blood urea were not.

Conclusion: When compared to previous models, SOFA and APACHE scores were superior in the predicting the overall mortality among patients at the ICU with decompensated liver cirrhosis.