The major purpose of this research is to develop new methods for the design and analysis of time to event data that are encountered in cancer clinical trials. The research will focus on four topics. 1. Methods will be developed for jointly modeling and estimating the relationship of longitudinally measured covariates to censored survival data using a proportional hazards regression model. 2. A comprehensive approach will be developed for estimating and testing relationships regarding a primary outcome variable that is missing on some individuals due to incomplete follow-up. 3. Since the difficulty with incomplete follow-up is most pronounced during interim monitoring, improper inferential procedures on the primary outcome can severely bias stopping rules for early termination of a clinical trial. We will show how to use the tests derived in topic (2) above to build group-sequential stopping rules which have the appropriate operating characteristics. 4. We will show how multiple imputation can be used to estimate the parameters and the standard error of these estimates in a proportional hazards model with missing covariates.
Showing the most recent 10 out of 41 publications