Abstract: Objective To investigate the validity of the model for end-stage liver disease (MELD) score in predicting the prognosis of patients with severe chronic hepatitis B (CHB). OMethods A total of 561 patients with severe CHB collected from January 2006 to December 2008 in 5 hospitals were divided into the survivor group and the death group. MELD score of all the cases was calculated and its predictive value for severe CHB was evaluated by t-test, logistic regression analysis and the area under the receiver operating characteristic curve (AUROC). Results The average MELD score in the death group was statistically higher than that in the survivor group (28.94±8.76 vs. 22.25±4.35, P<0.001). The mortality would be increased to 1.182 times every 1 score increased in MELD. The predictive accuracy rate of MELD score using logistic regression analysis was 68.4% (P<0.001). AUROC for prediction of the mortality in patients using MELD score was 0.759 with medium accuracy (95% confidence interval: 0.720~0.798, MELD threshold: 27).Conclusion MELD score may be useful in predicting the prognosis of patients with severe CHB, but the power is medium. Therefore, a better mathematical model should be developed to predict the prognosis of the patients with severe CHB in China.

Key words: Severe chronic hepatitis B; Prognosis; Liver failure; Logistic regression; MELD