Dear all,
I am looking for guidance on covariate adjustment in non-inferiority trials with a binary outcome. In a superiority trial, covariate adjustment leads to gain in power. The treatment coefficient moves away from 0. But this appears to lead to loss in power in non-inferiority trials. Does this mean we should not adjust for covariates in non-inferiority trials? I am wondering if there any guidance on this, and what are your recommendations for this situation. Many thanks.
The only reason for choosing a less powerful method is to not be seeking the truth. Not a good strategy IMHO.
1 Like
Many thanks Frank. I agree that one should chose the strategy that maximizes power.
1 Like
covariates would have to be prespecified and i think ich e9 says you shouldnt nominate too many of them
1 Like