Abstract:
|
In clinical trial setting, time-to-event analysis is often carried out under the proportional hazards (PH) assumption. If this assumption does not hold biased results may be obtained. For example, in immune oncology studies with presence of lag time to clinical response, assuming PH will be no longer valid. As an alternative we explore the Restricted Mean Survival Time (RMST) method and propose a new model based on survival change-point approach to quantify clinical benefit and make inferences appropriately. It can be shown that the proposed model can be viewed as a segmented proportional hazards model which can be readily implemented by using currently available software. The methods are illustrated through an example.
|