(Total Views: 549)
Posted On: 01/03/2021 9:35:23 AM
Post# of 153903
Mtruong,
FIrst of all let me state that the statistics are important (not mine but in general) for the FDA.
The the threshold value of 0.05 has been amply discussed and challenged. And there are many papers published in respect of the wisdom of having such a rigid criteria.
https://www.researchgate.net/publication/2629...itor_-_TAS
I agree that there should be some flexibility (unfortunately I have been invested in some companies that just missed it), however, with such enormous economical interest at stake the FDA has decided to "shield" themselves behind the 0.05 threshold.
In regards to the 34.7% this will comply with this p-value (type I error), however one has to take into consideration the Power of the test (type II error or false negative) as well. This is not an issue with trials with thousands of participants, it is with hundreds as ours.
As I mentioned before, our trial is somewhat underpowered by design (post-hoc 74.1% in the example you refer to), as there is need to obtain results rapidly in the middle of a pandemic.
Will the FDA approve with 35% reduction in mortality with a power slightly lower than, say 80% ??.
I don't know but I believe they should.
I suspect MESO's trial had a p-value way above the threshold and that is why it was stopped, not because was "just above" as continuing would had given the opportunity to demonstrate efficacy.
So, I think we will be OK with 34% Death reduction.
Just imho, of course
FIrst of all let me state that the statistics are important (not mine but in general) for the FDA.
The the threshold value of 0.05 has been amply discussed and challenged. And there are many papers published in respect of the wisdom of having such a rigid criteria.
https://www.researchgate.net/publication/2629...itor_-_TAS
I agree that there should be some flexibility (unfortunately I have been invested in some companies that just missed it), however, with such enormous economical interest at stake the FDA has decided to "shield" themselves behind the 0.05 threshold.
In regards to the 34.7% this will comply with this p-value (type I error), however one has to take into consideration the Power of the test (type II error or false negative) as well. This is not an issue with trials with thousands of participants, it is with hundreds as ours.
As I mentioned before, our trial is somewhat underpowered by design (post-hoc 74.1% in the example you refer to), as there is need to obtain results rapidly in the middle of a pandemic.
Will the FDA approve with 35% reduction in mortality with a power slightly lower than, say 80% ??.
I don't know but I believe they should.
I suspect MESO's trial had a p-value way above the threshold and that is why it was stopped, not because was "just above" as continuing would had given the opportunity to demonstrate efficacy.
So, I think we will be OK with 34% Death reduction.
Just imho, of course
(6)
(0)
