>>unlikely given the small difference in the means<< I don't remember when I last looked at a non-cancer trial. Yes it would be unlikely each arm if the variance were big. But keep in mind we also have a large n of 450 in each arm.