Abstract:
|
Perpetual platform trials, with a continuously enrolling control arm and the entry and exit of multiple therapies, present an interesting variation on the utilization of historical data. Ordinarily historical control information is valuable in a clinical trial but must be utilized with caution when the historical data arose from past trials under different conditions. Within a platform trial, we observe control data separated in time from concurrent controls, but data observed within the same trial and thus from similar sites with similar inclusion and exclusion criteria. In addition, the entire time course of the control arm can be modeled, allowing a direct estimate of whether past controls within the trial are similar to current controls. In this talk we will discuss the similarities and differences between utilizing past control data from a platform as opposed to historical data from separate trials, methods for utilizing this information specific to the perpetual platform trial structure, and illustrate the efficiency gains available from the use of this information.
|