Determining affected factors on survival of kidney transplant in living donor patients using a random survival forest

The common method for estimating survival indices is the Cox model. In the data with high volume, different order of interaction in the model is predictable. In that case, performance of the Cox model is not properly. Correspondingly, random survival forest (RSF) model is an alternative to the Cox model in this situation. The aim of this study is to determine factors that affected survival of patients with kidney transplant using RSF.
This historical cohort study carried on 459 kidney transplant recipients of living donor during of 1993- 2011 in Hamadan (Iran). Time between kidney transplant and irreversible transplant rejection was considered as a response. Modeling of determinants of survival was performed using Cox and RSF models and they were compared.
The survival rate of 5, 10 and 15 years were 91.6%, 85.3% and 74.9% respectively. Important variables selected based on various criteria in RSF were age of transplant recipient, the patient's condition at discharge, hemoglobin of receptor, the last Creatinine and the use of immunosuppressive drugs (inhibitors drug) in RSF model. Age of recipient, the patient's condition at discharge and the use of immunosuppressive drugs in the Cox model were significant.
Age of recipient, the patient's condition at discharge and the use of immunosuppressive drugs are determined as risk factors in Cox regression while the RSF method with less limited assumptions is able to determine risk factors of survival rate more precisely.