Few thoughts if I follow your logic.
Should the ad hoc results be more stat significant than what we saw in the non memantine population, especially since a 2 tailed test will be used in the confirmatory trial...and since the statistical analysis that was done comparing the results at 5 weeks held through the entire trial even the follow up at an incredible probability, was that what Alkon was referring to with his "black and white" statement?
First thing I noted when I saw our results was that the non memantine placebo group which included a few patients on no drugs at all and a couple other drugs not even specifically approved for severe alzheimer's but still being taken since their earlier stages actually performed much better than the placebo group as a whole. That fact seems illogical but it is a small sample size. In fact that group tested at almost even (slightly below baseline) at the end whereas the placebo as a whole was roughly 1.5 points below that. In other words if the placebo in the ad hoc study acted as expected historically and logically even, then our delta from placebo would have been nearly 8 pts. How much would that add to our stat significance I'm not sure? There will be a much larger placebo group in the upcoming trial. Let's hope it shows a normal downward slope for statistical reasons.
Another point, is the number just so small in non memantine population (14) that based on the standard deviations in such a population that no matter how consistent the improvement may or may not have been, would it still harm the stat significance. I'm not 100% sure of that answer. I just know they are confident with the numbers in the upcoming trial and they really aren't that much higher than what we've seen. Yes 3 or 4 times higher but you get the drift.
So what we don't know currently is if some number short of 14 patients benefited more than the 6+ pt average improvement with a couple patients harming the results who didn't show an improvement at week 5 or any other time...or if all patients improved at least some, with others peaking at 10 pt improvements. We also don't really know if patients with a 2 mmse score improved the most or the least or if levels of synaptic damage plays a part in this at all. IMO that would be the biggest potential factor IF anyone didn't show improvement in the Ad hoc group. Lots we don't know yet from that data.
What I think you are alluding to is could a trial be set up where test scores/plasma levels/pkce levels, etc be checked at week 5 and if they aren't showing an improvement then be dropped and the other 90% or whatever number that did show improvement and with a huge statistical probability that would continue, move on with the trial. That would be one heck of a great confirmation trial if what we've seen so far plays out with this current trial and could be incredibly convincing. JMHO and sorry for the ramble and any misunderstandings on my part.
