Organ procurement organizations (OPOs) face increasing regulatory scrutiny, and the performance of predictive models used to assess OPO performance is critical. We sought to determine whether adding deceased donor physiological and critical care data to the existing Scientific Registry of Transplant Recipients (SRTR) heart yield model would improve the model's performance. Donor data and heart transplanted (yes/no), the outcome of interest, were obtained from the United Network for Organ Sharing Donor Management Goal (DMG) Registry for 19 141 donors after brain death, from 25 OPOs. The data were split into training and testing portions. Multivariable LASSO regression was used to develop a statistical model incorporating DMG data elements with the existing components of the SRTR model. The DMG + SRTR and SRTR models were applied to the test data to compare the predictive performance of the models. The sensitivity (84%-86%) and specificity (84%-86%) were higher for the DMG + SRTR model compared to the SRTR model (71%-75% and 76%-77%, respectively). For the DMG + SRTR model, the C-statistic was 0.92 to 0.93 compared to 0.80 to 0.81 for the SRTR model. DMG data elements improve the predictive performance of the heart yield model. The addition of DMG data elements to the Organ Procurement and Transplantation Network data collection requirements should be considered.
Keywords: clinical research; donor after brain death (DBD); donors and donation; heart transplantation; intensive care management; organ procurement organization (OPO).
Published by Elsevier Inc.