The latest post over at Bayesian Spectacles, entitled “A Bayesian Perspective on the Proposed FDA Guidelines for Adaptive Clinical Trials,” is of special interest to the medical field. Eric-Jan Wagenmakers, Quentin F. Gronau, Angelika Stefan, Gilles Dutilh, and Felix Schönbrodt comment extensively on the recent draft version of FDA guidelines for this category of more-efficient trial designs.
A major hook for this read is available from the beginning:
Unfortunately, the FDA document approaches the adaptive design almost exclusively from a frequentist perspective. The use of Bayesian methods is mentioned, but it appears unlikely that any Bayesians were involved in drafting the document. This is a serious omission, particularly because the Bayesian and frequentist perspectives on sequential designs are so radically different.
This is key. Medical professionals who wish to be “statistically sophisticated” (as I believe Leonhard Held says) should recognize that the frequentist bias within medical research is strong, and authoritative documents from regulatory organizations such as the FDA are one means for it to manifest. Give it a read!
*Post title attributed to Wagenmakers et al. in the linked article.