The Accuracy of Nonstandardized MELD/PELD Score Exceptions in the Pediatric Liver Allocation System

Background. In the United States, over half of pediatric candidates receive exceptions and status upgrades that increase their allocation model of end-stage liver disease/pediatric end-stage liver disease (MELD/PELD) score above their laboratory MELD/PELD score. We determined whether these "nonstandardized" MELD/PELD exceptions accurately depict true pretransplant mortality risk. Methods. Using data from the Scientific Registry of Transplant Recipients, we identified pediatric candidates (<18 y of age) with chronic liver failure added to the waitlist between June 2016 and September 2021 and estimated all-cause pretransplant mortality with mixed-effects Cox proportional hazards models that treated allocation MELD/PELD and exception status as time-dependent covariates. We also estimated concordance statistics comparing the performance of laboratory MELD/PELD with allocation MELD/PELD. We then compared the proportion of candidates with exceptions before and after the establishment of the National Liver Review Board. Results. Out of 2026 pediatric candidates listed during our study period, 403 (19.9%) received an exception within a week of listing and 1182 (58.3%) received an exception before delisting. Candidates prioritized by their laboratory MELD/PELD scores had an almost 9 times greater risk of pretransplant mortality compared with candidates who received the same allocation score from an exception (hazard ratio 8.69; 95% confidence interval, 4.71-16.03; P < 0.001). The laboratory MELD/PELD score without exceptions was more accurate than the allocation MELD/PELD score with exceptions (Harrell's c-index 0.843 versus 0.763). The proportion of patients with an active exception at the time of transplant decreased significantly after the National Liver Review Board was implemented (67.4% versus 43.4%, P < 0.001). Conclusions. Nonstandardized exceptions undermine the rank ordering of pediatric candidates with chronic liver failure.