Prediction of delayed graft function in combined liver-kidney transplantation: An analysis using the UNOS registry
Recommended Citation
Chau LC, Safwan M, Moonka D, Kim D, Yoshida A, Abouljoud M, and Nagai S. Prediction of delayed graft function in combined liver-kidney transplantation: An analysis using the UNOS registry. Transplantation 2019; 103(8):379-380.
Document Type
Conference Proceeding
Publication Date
9-2019
Publication Title
Transplantation
Abstract
Background: Delayed graft function (DGF), defined by the need for dialysis within 7 days after transplantation, impacts graft and patient outcomes. This study aims to predict the occurrence of DGF after combined liver kidney transplantation (CLKT). Methods: We examined adult CLKT transplanted from Jan 01 2003 to Mar 31 2018 in the UNOS registry. Recursive feature elimination on random forest was used to select the top 20 out of 134 variables. Logistic regression models were fitted based on selected variables to predict the occurrence of DGF. Performance of the model was assessed by computing the area under the receiver operating characteristic curve (AUROC) after 10-fold stratified cross-validation. Results: 6934 adult CLKT were included in this study. The incidence of DGF was 22,14%. Risk factors for DGF after CLKT include donor BMI, cause of death, recipient MELD score, KDRI (Rao), dialysis prior to transplant, life support prior to transplant, days on the waiting list, re-transplant, distance to transplant center, creatinine at transplant, and cause of organ failure. The model AUROC was 0.7139 (95% CI: 0.6829, 0.7447). Conclusion: This study identified novel risk factors and is the first to date to estimate the development of DGF after CLKT using variable selection via a random forest-based method with subsequent logistic regression. Further study is required to confirm and explore these associations in other kidney transplant populations.
Volume
103
Issue
8
First Page
379
Last Page
380
