No wonder FDA is grabbing all the data - the HGEN
Post# of 148187
Add this to the recent Merck "FDA wants more data" failure after having an "anticipated EUA" for their fusion protein.
In trying to get in sync with TechGuru on power, I spent some time with the HGEN PR. Tricky stuff - none of which should make the FDA happy.
1) They are trying to get away with a 1.29 relative risk, but (1/1.29) = .775, only a 22.5% reduction in fails.
2) They ask the question in reverse: The data set is only survivors. They ask, if I'm a survivor, which arm did I come from? The null hypothesis is just that P(come from treat) = 0.5. There is no intuitive link to calculate actual people benefitted. Also, the step from old sample size to new doesn't compute.
Again, no wonder the FDA is grabbing all the data. And no way NP should or should be expected to open his mouth.