Background: Serum sodium predicts prognosis in chronic severe hepatitis B and may improve the prognostic accuracy of the model for end-stage liver disease (MELD) score, but the available information is limited. The present study was undertaken to study the clinical use of the serum sodium incorporated MELD (MELD-Na) and assess its validity by the concordance (c)-statistics in predicting the prognosis of the patient with chronic severe hepatitis B.

Methods: A total of 426 adult patients with a diagnosis of chronic severe hepatitis B between January 1, 2007, and December 31, 2007 at a single center were studied. The scores of serum sodium, MELD, MELD-Na, and DeltaMELD-Na (DeltaMELD-Na = MELD-Na at 14 days after medical treatment -MELD-Na score on admission) of the patients with chronic severe hepatitis B were calculated. The 3-month mortality in the patients was measured, and the validity of the models was determined by means of the concordance (c) statistics.

Results: The average MELD, MELD-Na scores of survival group were 25.70 +/- 5.08 and 26.60 +/- 6.90, and those of dead group were 35.60 +/- 6.78 and 42.80 +/- 9.57 on admission. There was a significant difference in MELD and MELD-Na between the survival and dead groups (P < 0.01). The average DeltaMELD-Na score of the survival group was -0.97 +/- 3.51, and that of the dead group was 3.45 +/- 2.38 at 2 weeks after the treatment. There was a significant difference in DeltaMELD-Na between the survival and dead groups (P < 0.01). The areas under the receiver-operating characteristic curves of Na, MELD and MELD-Na for the occurrence of death in 3 months were 0.742, 0.875 and 0.922. The 3-month mortality of the MELD-Na scores group < 25, 25-30, 31-34, 35-40 and > 40 were 2.0%, 5.4%, 35.4%, 53.8 % and 86.9%, respectively. There was a significant difference in the 3-month mortality between the five groups (P < 0.05). The 3-month mortality of the DeltaMELD-Na > 0 group was 65.9%, and that of the DeltaMELD-Na = 0 group was 15.8%; there was a significant difference in the 3-month mortality between the two groups (P < 0.05).

Conclusions: MELD-Na score is a valid model to predict the 3-month mortality in patients with chronic severe hepatitis B. DeltaMELD-Na is a clinically useful parameter for predicting the therapeutic effect of chronic severe hepatitis B.