Multivariate prediction model of kidney transplant success rates.

An excellent correlation of predicted and observed cadaver kidney graft survival rates was obtained using a nine-factor computer model. A total of 924 recipients for whom a greater than 80% 1-year survival rate was predicted had an observed rate of 84.6%, whereas 179 recipients with a predicted 1-year success rate of less than 40% had an observed rate of 38.2%. Our model gives improved results as compared with previously published methods. We anticipate that the model's predictive power can be further refined, and that patient selection and organ sharing will benefit from the application of sophisticated computer models for the prediction of transplant success.