News & Events
2026.10.13
2026.10 데이터사이언스융합연구소 Seminar 개최
.jpg){kind=link}
Joo Young Lee Department of Applied Statistics, Chung-Ang University
Statistical methods for addressing post-progression treatment switching
Treatment switching is common in randomized oncology trials, particularly when individuals assigned to the control arm are allowed to switch to an experimental intervention after disease progression. When the estimand of interest is the causal effect of treatment on overall survival, a common strategy is to censor control-arm individuals at the time of crossover to remove the subsequent effect of the experimental treatment. However, this artificial censoring is informative and time-dependent, therefore, standard survival analyses can yield biased estimates. This talk will present inverse probability of censoring weighting (IPCW) and g-computation approaches for estimating treatment effects on overall survival in the presence of treatment switching. The IPCW approach uses a weighted Aalen–Johansen estimator to estimate transition probabilities within a multistate framework representing disease progression, treatment switching, and death. Two g-computation approaches will also be considered, differing in whether post-switching data are incorporated into model estimation. Simulation studies were conducted to evaluate the finite-sample performance of these methods under varying degrees of treatment switching.