Objective: To compare the capability in predicting the prognosis of chronic severe hepatitis among 3 prediction systems: Model for end-stage liver disease (MELD), Child-Turcotte-Pugh (CTP) system, and King's College Hospital (KCP) system, and to explore the capability of MELD in predicting the curative effect of artificial liver.
Methods: Scoring was made among 66 patients of chronic severe liver diseases with the underlying disease of hepatitis B, 11 in early stage, 14 in middle stage, 15 in late stage, and 26 unclassified, by MELD, CTP, and KCP systems. The accuracy of each system was evaluated by ROC curve, the differences between the systems was analyzed by Kaplan-Meier survival curve.
Results: The MELD score of the patients at admission predicted the mortality within 3 months with the c-statistic of 0.894, higher than those of the CTP and KCP systems (0.703 and 0.89 respectively). The MELD scores of the patients in the early stage was 24 +/- 4, significantly lower than those in the middle and late stages (31.11 +/- 2.90 and 41.38 +/- 9.98 respectively, all P < 0.01). The MELD score was positively correlated with the stage of disease (r = 0.737, P < 0.01). The mortally was 10.7% for the patients with an admission MELD score < or = 30, was 47% for the patients with an admission MELD score of 31 approximately 39, and was 60% the patients with an admission MELD score of > or = 40.
Conclusion: The predictive capability of MELD system is better than the KCP and CTP systems. Artificial liver support treatment is the best choice e for the patients with an admission MELD score of 31-39. An admission MELD score > 40 is the indication for liver transplantation.