Prediction of Kidney Allograft Discard Before Procurement: The Kidney Discard Risk Index.Price, M. B., et al.
Experimental & Clinical Transplantation 2021; 19(3): 204-211.
The aim of this study was to establish predictors of kidney discard and to derive a kidney discard risk index (KDSRI) on the basis of these factors.
The study population was randomized to the training set and the validation set.
102,246 potential renal allograft donors.
The primary outcome was renal allograft discard.
mean follow-up time, 4.5 ± 3.7 years
This paper attempts to derive a risk index for kidney allograft discard using OPTN data from the US. The authors randomly split a cohort of 102,246 kidney offers into training and validation sets and used logistic regression to create an index to predict discard based upon factor available prior to procurement. The final model shows a C-statistic of 0.89, suggesting good performance in this cohort. The three most predictive factors for discard were age, serum creatinine and hepatitis C antibody status. The authors suggest that the risk index couple be used to identify organs at high risk of discard as a basis for interventions or allocation policies to improve utilization. One interesting aspect of this study is that the authors also developed machine learning models based on the same input variables and demonstrated that the logistic regression model achieved similar performance to machine learning approaches whilst maintaining better transparency and explainability. Clearly, the model is only validated in the US allocation system and may not have the same predictive performance in other settings – further research should explore generalizability.