Survival Analysis for Truncated Data and Competing Risks

The purpose of this project is to consider the problems of left truncation and competing risks in analyzing censored survival data, and to compare and contrast various approaches for handling these problems. The motivation for this work comes from an analysis of data from the Cache County Memory Study. Study investigators were interested in the association between early-life psychologically stressful events (e.g., parental or sibling death, or parental divorce, among others) and late-life risk of Alzheimer\u27s disease (AD). While conventional methods for censored survival data can be applied, the presence of left truncation and competing risks (i.e., other adverse events such as death that may lead to censoring with respect to AD) may require some consideration in order to avoid potential bias in terms both of estimation and inference. In this paper we briefly summarize the issues of truncation and competing risks in the context of survival analysis, and apply and compare several approaches suggested in the literature to the Cache County Data