Read on Twitter
https://twitter.com/T_Allen1998/status/1323335976991756290?s=20
@statsepi
Part of the problem is using the same symbols for different things.
In a superiority trial delta is the minimum clinical difference, or as @stephensenn calls it "the treatment difference we would not like to miss." /1
@statsepi
Part of the problem is using the same symbols for different things.
In a superiority trial delta is the minimum clinical difference, or as @stephensenn calls it "the treatment difference we would not like to miss." /1
In a non-inferiority trial what the FDA document refers to as M or M1 is a property not of the test drug but of the reference drug - "the entire known effect of the active control relative to placebo". What the FDa prefers is "a smaller margin (M2) /2
this should reflect "the largest loss of effect that would be clinically acceptable. ". The problem of course is that this can me much smaller than M1 and therefore will require a large sample size if the study is powered at a zero treatment difference /3
This is the main reason that many clinical trials researchers dislike an NI design because they believe that a large sample size is always necessary. Not so. If you have developed the new drug to be "better" than the reference there is no need to sample size /4
for a zero treatment difference, and I would not recommend that this approach is taken routinely /5
