P3, re very large trials.
This has been a harping point of BSR David, and I quite agree.
If you need to run a trial of X thousand to get stat sig, there obviously is a questionable benifit at best.
But if you have n=100 and p=.02 you likely have a much better treatment effect.
There is a real fundlemental problem here in the FDA law/regs. The drug must be proved effective, but the same standard of proof holds regardless of how effective it is.
Drug 1: Another NSAID that shows pain relief N=2000 , p=.05
Drug 2: Improves OS (HR=1.8) in stage 3 NSLC N=100, p=.06
Which one should be on market?
Market Data 
Markets 
AI
