Keywords: Non-proportional hazard, treatment crossover, design and monitoring
Survival trials are often facing a lot of challenges in study design and monitoring when the proportional hazards assumption cannot be reasonably assumed. The common reasons for the non-proportional hazard ratio (HR) are delayed treatment effect and treatment crossover. A delayed treatment effect means the treatment will not have effect until enough exposure has been achieved therefore the HR is 1 in the beginning and becomes smaller at a later time. Treatment crossover means the study participants may cross from the assigned treatment to the other treatment arm therefore reducing the treatment effect. In this talk, we are presenting a design and monitoring tool to account for these complex situations.