Adaptive vs. Group Sequential Designs in Survival Analysis

Posted by Esha Senchaudhuri

Mar 5, 2015 11:00:00 AM

View Slides

The Mehta-Pocock promising zone is often used to carry out unblinded sample size re-estimation during interim analysis. However, according to Jin Wang of Abbott Vascular, it can also be used to re-estimate follow-up times in the interim look of a survival analysis study.

According to simulations run by Jin, the Mehta-Pocock promising zone design offers the highest power and lowest Type 1 Error when compared to other common group sequential and adaptive designs. However, Jin questions whether this necessarily makes it the design of choice. 

Writes Jin: “For each adaptive strategy, there is always a GSD version that is more efficient.”

According to Jin, despite the statistical benefits of the promising zone, applying the design to re-estimation of follow-up time has the potential to increase the maximum follow-up time to an extent where a two-stage group sequential design might be preferable. However, this increase only occurs under the alternative (i.e. when interim results fall within the promising zone.)

Jin took part in Statistical Innovations in Clinical Development, an event co-sponsored by Cytel and the American Statistical Association (Bay Area Chapter). During his address, he revealed simulations of two common adaptive strategies for survival settings (Mehta-Pocock and Li), and compared them to the performance of group sequential designs. (The figure above shows Jin's results). 

Please enjoy the video of Jin Wang’s talk and find the slides by pressing the green button.





The Cytel blog keeps you up to speed with the latest developments in biostatistics and clinical biometrics.  Sign up for updates direct to your inbox. You can unsubscribe at any time.


Posts by Topic

see all

Recent Posts