Analyzing safety data for two-stage randomization designs

Ensuring patient safety is paramount in the development of new pharmaceutical products, which, while beneficial, can also pose potential risks. Traditional methods for analyzing safety data have been limited to single-stage randomization clinical trials. However, no existing methodology addresses the complexities of two-stage randomization designs with survival endpoints. This paper introduces a novel methodology utilizing inverse probability weights to analyze safety data in two-stage randomization designs. Our approach is applied to data from a leukemia clinical trial, the use of the weighted Aalen-Johansen estimator is recommended while the use of the weighted Kaplan-Meier is discouraged. This advancement provides a crucial tool for enhancing patient safety in complex clinical trial designs.