Predicting Early Extubation After Liver Transplantation: External Validation and Improved Generalizability of a Proposed Fast-track Score

Background. Early extubation of liver transplantation recipients is a cornerstone of fast-track (FT) pathways. Identifying suitable candidates has previously been accomplished using perioperative variables to develop a FT probability score. The objective of this study was to externally validate a proposed FT score. Methods. Following Research Ethics Board approval, data were extracted on liver transplants conducted at a single center from 2009 to 2017. Data extracted included patient characteristics, intraoperative variables, and postoperative outcome variables. The proposed FT score utilized 9 variables: age, gender, body mass index, model of end-stage liver disease, retransplant, preoperative hospital admission, blood transfusion, operative time, and vasopressor use. We calculated the FT score in our cohort, and assessed the discrimination and calibration of the model. Score performance was explored by subgroup analyses, customization and altering the outcome definition. Results. The FT score was found to predict higher rates of successful FT than was observed in the external cohort (n = 1385) and had reduced discrimination (area under the receiver operating curve, 0.711; 95% confidence interval, 0.682-0.741) compared with the original internal validation cohort (area under the receiver operating curve, 0.830; 95% confidence interval, 0.789-0.871; P < 0.0001). Discrimination was improved by customizing the transfusion (P < 0.0001) components of the simplified score or by level 1 customization of all regression model coefficients (P < 0.0001). A time-based definition of FT (early extubation) did not alter the accuracy of the prediction score (P = 0.914), improving the model's generalizability. Conclusions. The proposed FT score may help identify patients suitable for early extubation and FT pathways after liver transplantation in conjunction with clinical judgment.

Automated summary

The study externally validated a proposed score for identifying liver transplantation recipients suitable for early extubation and fast-track pathways. The score showed lower discrimination compared to the original validation cohort, but customization of the transfusion components or regression model coefficients improved its predictive accuracy. Time-based definition of fast-track pathways did not affect the prediction score's accuracy, enhancing the model's generalizability.