A validated model for predicting outcome after liver transplantation: implications on transplanting the extremely sick.

Given the organ shortage, there is a need to optimize outcome after liver transplantation (LT). We defined posttransplant hospital length of stay > 60 days (LOS > 60) as a surrogate of suboptimal outcome. In the first phase of the study, a 'Study cohort' (SC) of 643 patients was used to identify risk factors and construct a mathematical model to identify recipients with anticipated inferior results. In the second phase, a cohort of 417 patients was used for validation of the model ['Validation Cohort' (VC)]. In the SC, 65 patients (10.1%) had LOS > 60 days. One- and 3-year patient/graft survival rates were 81.9%/76.1% and 73.4%/67.4%, respectively. Patient and graft survival rates of those with LOS > 60 days were inferior (P < 0.0001), while transplant cost was greater [3.42 relative units (RU) vs. 1 RU, P < 0.0001]. In a multivariable analysis, pretransplant dialysis (P < 0.001), mechanical ventilation (P < 0.015), MELD (P < 0.003), and age (P < 0.009) were predictors of LOS > 60 days [ROC curve - 0.75 (95% CI 0.70, 0.81)]. In the VC, 53 patients (12.7%) were expected to have adverse outcome by the model. These patients had longer LOS (P < 0.0001), higher cost (<0.0001), and inferior patient and graft survival (P < 0.007).