I read with interest the article by Biggins et al. evaluating serum sodium as a predictor of mortality because I was hoping their study supported my clinical intuition that hyponatremia would add predictive value to the MELD score.1 However, I was disappointed on two accounts. First, although serum sodium was independently associated with mortality in multivariate analysis, the improvement in the receiver operator curve was statistically and clinically insignificant. The area under the ROC for 3-month mortality increased from 0.88 with MELD alone to 0.917 for MELD plus sodium <126, P = .207. Results for the area under the ROC for 6-month mortality were similar. The second reason for disappointment was that the conclusion reached by the authors, “Addition of serum sodium to MELD increases the ability to predict 3- and 6-month mortality in patients with cirrhosis” was not supported by their results.
A study published in the October 2004 issue of HEPATOLOGY was more encouraging and demonstrated that hyponatremia by itself may not improve upon MELD for predicting mortality, but hyponatremia in the presence of ascites is better at predicting mortality in patients with MELD scores less than 21.2 Biggins et al. may have been unaware of the results from the earlier study at the time they submitted their manuscript, but the editors should have had knowledge of both studies and asked them for an analysis of serum sodium in the subgroup of candidates with low MELD scores.