The opening post is right about the mechanism and I think understates the practical bit.
Pregnancy and breastfeeding are the clearest cases where the answer is that the trials excluded them and the labelling reflects that exclusion.
Worth saying I have only my own numbers here, and n is small.