06 July 2018

Let The Battle Begin

It should come as no surprise that I remain a Bayes skeptic. Imagine my amusement (of a macabre nature, naturally) reading this review of Biogen's latest Alzheimer's trial.
Study 201 used adaptive randomisation to enrol 856 subjects with mild Alzheimer's disease across five dose levels. It used a complex Bayesian statistical model to yield a faster result than traditional designs.

You should read the article through. There are issues, in the author's view, beyond the use of Bayes. One might infer that some or all of these other issues are, to steal a word, co-linear with the choice to leverage Bayes.

Alzheimer's is a dread, particularly for those of us past 40 or so. Finding clinical trial failure supportive is perverse, to say the least. On the other hand, sponsors shouldn't (and should really be prevented from) "torture" data to defend a compound. And in this kind of case where the history, even of this compound, with the target is universally negative. What's worse, Biogen isn't some pump&dump nano-cap San Diego bio, either.

Go to your corners and come out fighting!

No comments: