The more I read about the AZ trial, the most screwed up it looks. Because the half dose was just at one site, it really messed up the results. Patients who got the half dose were apparently not remotely balanced demographically by any of the controls that they normally use. They appear to have cherry picked their data (they are only reporting roughly half of the participants in the Brazil arm, for reasons that are not at all clear). My Twitter feed of epidemiologists seem to doubt that they will meet the standard required for FDA approval at all. The
Wired story has a pretty good overview of the problems that have been pointed out so far.