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 1 (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.