The Model for End Stage Liver (MELD) score is validated to predict pretransplant mortality. However, as a predictor of postoperative outcomes, its utility has proven inconsistent. Recently developed MELD-Lactate models better predict 30-day survival as compared with the MELD and MELD-Sodium scores. We compared the MELD-Lactate, original MELD, and MELD-Sodium formulae and the initial postoperative lactate as predictors of 30-day and in-hospital mortality following liver transplantation.
Adult patients (n = 989) undergoing orthotopic liver transplant between 2002 to 2013 were included. In addition to the previous models, the first postoperative lactate value and a newly derived Mount Sinai MELD-Lactate score and associated c-statistics were compared.
The Mount Sinai MELD-Lactate model yielded the highest c-statistic value (0.749), followed by the original MELD-Lactate (0.740), initial lactate value (0.729), postoperative MELD (0.653), and MELD-Sodium (0.641) models in predicting survival at 30 days following liver transplantation. For in-hospital mortality, the original MELD-Lactate model had slightly higher c-statistic (0.739) compared with the Mount Sinai MELD-Lactate model (0.734). Despite the distribution differences in the MELD-Lactate models, the model validation results, both from cross-validation and bootstrap methods, were similar.
Postoperative MELD-Lactate and isolated postoperative lactate values were moderately predictive of 30-day and in-hospital mortality following liver transplantation in this patient cohort.