Yes, we need to do it at the front end, but also we need to understand that things will creep up. Rare reactions will creep up.
The other reality is that because we don't require head-to-head--that is, comparisons between drugs and interventions--before we approve a drug, the other thing post-market surveillance does is tell us whether or not the product really works. I know it sounds silly even to say that, but we need to realize that taking something that doesn't work, for whatever reason you were given it, is not just an adverse event; it's an incredible waste of money, time, and resources.