“This means that if the 6m MPC group has any % higher than the placebo group even if it isn't as strong as the results above it will successfully meet the primary endpoint.”
No that’s not how it works.
The treatment success rate on the primary outcome has to be statistically and clinically significant.
So what superiority is required?
Very roughly – reverse engineer from the sample size of 404 – three groups of 134 each.
Assume a control group responder rate of 12% as in the P2 study.
You will have around 80% power to detect a difference of 14% (absolute). So treatment group has to be better than 26% assuming a control group responder rate of 12%.
Its probably a little bit more complicated than this because there is the MPC+HA treatment group so in theory there will be some sort of alpha correction for the multiple primaries. But very roughly this will be around the mark.
Now this will be all according to the “official” statistical analysis plan. If the primaries (assuming the two treatment groups are multiple primaries) fail MSB will then do as it likes. Which will be to revert to finding something statistically significant on one of the secondaries, a post hoc sub-group analysis, or just make up a new endpoint.
View attachment 2850749
(20min delay)