A Recalibrated MELD Score Predicts Mortality in Acute Variceal Hemorrhage

Atif Zaman, MD, MPH

This tool outperformed the Child-Pugh, Augustin, and D'Amico models in predicting 6-week mortality from bleeding.

Despite recent improvements in its management, acute variceal hemorrhage (VH) is associated with significant morbidity and mortality. Several prognostic models, including the model for end-stage liver disease (MELD) and Child-Pugh, have been shown to predict poor outcomes. However, they are seldom used, partly because of a lack of external validation.

The aims of the current study were to evaluate the performance of a few recently described prognostic models for VH (MELD, Child-Pugh, Augustin, and D'Amico), identify the best performing model in a prospective cohort of patients, calibrate this model to maximize performance, and validate the new model in two external cohorts. The initial cohort comprised 178 patients with cirrhosis and VH who received standard treatment. The validation cohorts were from Canada (240 patients) and Spain (221 patients). The main outcome variable was 6-week mortality after VH.

The overall 6-week mortality rate in the initial cohort was 16%. The MELD model had the best performance among the prognostic models evaluated. A recalibrated MELD score of ≥19 predicted a mortality rate of ≥20%, and a score ≤11 predicted a mortality rate of <5%. The MELD-based model performed well in the external validation cohorts: Discrimination was excellent in the Canadian cohort (area under the curve, 0.873) and good in the Spanish cohort (AUC, 0.794).

Citation(s):

Reverter E et al. A MELD-based model to determine risk of mortality among patients with acute variceal bleeding. Gastroenterology 2014 Feb; 146:412.