Date of Award


Document Type

Doctoral Dissertation - Open Access

Degree Name

Doctor of Public Health


Public Health

First Advisor

Kenneth Knapp

Second Advisor

Elizabeth Drugge

Third Advisor

Adam Block


The allocation of organs is a constantly evolving area of transplantation. The latest iteration of this process is a move toward continuous distribution of organs. This process considers dynamic factors instead of static constraints that box patients into certain categories. The expected post-transplant survival of patients is one of these dynamic factors that will be included in the assessment of patients' place in a continuous distribution allocation process. However, the predictive value of this construct is questionable, given that it uses only four factors (age, diabetes diagnosis, years on dialysis, and previous transplant) to represent complex patients who are on our transplant lists. The predictive value based on the C statistic is only 67%. To improve the predictive value of this construct, we sought to determine the factors that are important in determining patient post-transplant survival and then incorporate these into a new predictive model that can be used in determining patients' post-transplant survival. This model was obtained using the Cox proportional-hazards model. United Network Organ Sharing (UNOS) database 2000-2020 was used to test the predictive value of the original model. Cox proportional-hazard testing was carried out using variables that were found to be predictors of mortality in renal transplant patients. The additional variables in the model included functional status, peripheral vascular disease, hepatitis C, and body mass index (BMI). These new variables were included in new models and compared with the original model. The Eta squared, C- statistic and Cox-proportional hazard regression were used to evaluate the model and goodness of fit compared. We noted that the new model had enhanced Eta squared for the variables, enhanced C-statistic or area under the curve, and adjusted R squared, indicating the newer model had an improved predictive value.

The conclusion is that the new model with the additional variables has enhanced predictive value. We propose that in the development of the continuous distribution paradigm, the expected post-transplant survival be re-evaluated to create a superior predictor of survival.