Nonparametric Methods for Irregularly Sampled Censored Data with Applications to Liver Transplant Allocations

• Activity Number: 551 - New Innovations in Handling Incomplete Biomedical Data in the Era of Data Science
• Type: Invited
• Date/Time: Wednesday, August 2, 2017 : 2:00 PM to 3:50 PM
• Sponsor: International Indian Statistical Association
• Author(s): Sujit K Ghosh* and Bradley Turnbull
• Companies: North Carolina State University and Apple Inc
• Keywords: Competing risk ; Liver transplant ; Nonparametric models ; Ordinal predictors ; Bayesian inference ; Imputation methods

Abstract

Over the last fifteen years, there have been major concerns that the current system of organ allocation may not ensure that available organs reach the patients most in need of a transplant. In 2002, motivated by an evaluation by the Institute of Medicine, the liver allocation system was changed from a subjective status-based algorithm to one using a continuous Model for End-stage Liver Disease (MELD) severity score. Using data from the United Network for Organ Sharing (UNOS), we examine and discuss several statistical aspects of modeling the waiting time distribution. We use a competing risk analysis method to compare the risk of a waiting list death as a function of severity (MELD score) and relevant patient level characteristics. The UNOS data features a few data irregularities: censored response, ordinal predictors, and missing patient level characteristics. We develop Bayesian (proper) imputation methods based on a hierarchical semi-parametric model. Ordinal predictors are tackled using a latent continuous variable approach. Proposed methods are first validated using simulated data and then illustrated by analyzing the UNOS data.