The only source for that I have seen assumed that if the optimal treatment was chosen that the outcome would always be positive which I suspect is not super accurate. I wish I remembered where that source was. Do you have one? Would love to look at it again.