Abstract:
|
The duration of response is an important endpoint in oncology studies. The conventional method estimates the distribution of the duration of response via the Kaplan Meier curve among responders only. However, such an approach can be biased due to informative censoring, and also violates the ITT principle. We will first introduce the restricted mean duration of responses as a population summary, and propose several estimators for the distribution of the duration of response in the entire patient population including a simple IPCW estimator, its efficiency-augmented counterpart, and a regression-based estimator. The performance of these estimators will be compared theoretically and empirically in numerical studies.
|