Development of a pediatric end-stage liver disease score to predict poor outcome in children awaiting liver transplantation

Background. A pediatric end-stage liver disease (PELD) score for children with chronic liver disease using easily obtainable, objective, verifiable parameters, would be useful to prioritize children awaiting liver transplantation. Methods. Data from the Studies of Pediatric Liver Transplantation (SPLIT), a consortium of 29 U.S. and Canadian centers, were used to develop the PELD score. Two pretransplantation endpoints were evaluated: (1) death, n=884; and (2) death or moving to the intensive care unit (ICU), n=779. The analyses were restricted to children with chronic liver disease who were listed for a first transplant. Preliminary analyses of 17 possible factors yielded 6 parameters of interest: age < 1 year, total bilirubin, international normalized ratio (INR), albumin, growth failure (height or weight Z score < -2), and calculated glomerular filtration rate. In a univariate Cox regression analysis, age, bilirubin, INR, and albumin were significant (P < 0.01) predictors of both endpoints; glomerular filtration rate was not significant for either endpoint; and growth failure was significant for death/ICU but not death alone. In the multivariate analyses, age, bilirubin, and INR were significant for the death endpoint; and bilirubin, INR, growth failure, and albumin were significant for the death/ICU endpoint. From these results, three PELD models were evaluated to predict both outcomes at 3 and 6 months: PELD I (age, bilirubin, INR); PELD 2 (bilirubin, INR, albumin, growth failure); and PELD 3 (bilirubin, INR, albumin, growth failure, and age). The area under the receiver operating characteristic curve (AUC ROC) was used to compare models. For PELD 3, the most inclusive model, the AUC ROC at 3 months was 0.92 for death and 0.82 for, "death-moved to ICU." A comparison of the AUC ROCs for the other models and for the model of end-stage liver disease ([MELD], the adult end-stage liver disease severity score model), none of which performed better than PELD 3, are presented. Conclusion. A model using five objective parameters can accurately predict death or death-moved to ICU in children awaiting liver transplantation.