PSDV - Actually they are using what they call the Full Analysis Set using the combined numbers from both trials which is likely to stand a much smaller chance than the ART analysis thereby improving the odds of you not having to eat your hat. They get p values of .002 for both doses using that sleight of hand.
The FAME study protocol provides that the primary assessment of efficacy will be based on the Modified ART dataset and that the other datasets will be considered secondary; the protocol did not specify the Full Analysis Set as a dataset for analyzing the study. However, we believe that the FDA will consider the Full Analysis Set to be the most relevant population for determining safety and efficacy in Trials A and B.