Prognostic Value of Model for End-Stage Liver Disease Score Measurements on a Daily Basis in Critically Ill Patients with Cirrhosis

Objective To determine whether daily measurement of Model for End-Stage Liver Disease (MELD) score adds prognostic value to the initial MELD score in predicting mortality among patients with cirrhosis admitted to the intensive care unit (ICU). Methods We included 830 consecutive patients with cirrhosis admitted to a tertiary care ICU from January 1, 2003, through December 31, 2013, who had MELD scores on admission day 1 (MELD-D1). Daily MELD score during the first 7 days of ICU admission were retrospectively abstracted. The performances of MELD-D1 to MELD-D7 and changes in MELD score on consecutive days (Δ-MELD) in predicting 90-day mortality were determined using logistic regression. Results MELD-D1 was an independent predictor of mortality (adjusted odds ratio, 1.07; 95% CI, 1.05-1.10; P<.001), with an area under the receiver operating characteristic curve (AUC) of 0.72. MELD-D2 to MELD-D7 yielded comparable performance to MELD-D1 with an approximately 10% increase in risk of death per each incremental unit of MELD score (odds ratios, 1.09-1.11; P<.001; AUCs, 0.68-0.72). Δ-MELD-D2 to Δ-MELD-D7 were not independently associated with mortality (P=.69, P=.42, P=.81, P=.94, P=.83 and P=.28, respectively) and did not increase the predictive performance (AUCs) when combined with MELD-D2 to MELD-D7. Conclusion Repeating MELD score assessment during the first 7 days after ICU admission does not improve the ability of the initial MELD score for predicting 90-day mortality among patients with cirrhosis. Our finding does not support the practice of routine daily measurement of the MELD score.