When you're doing preliminary studies, you get a sense of how common the side effects are, and if a rare one shows up you know your statistical power on boundin

When you're doing preliminary studies, you get a sense of how common the side effects are, and if a rare one shows up you know your statistical power on bounding its likelihood. Multiplying by your expected market will tell you how well you've bounded total deaths. Consequently, you'll know when you need to extend the study to get a better detection - but to tell you the truth, for the issue to have been bad enough for them to choose to hide it entirely instead of disclosing it a list of terrible rare side effects that all medicines have, it must have been common enough in their sample for the large number of total deaths to have been predicted.