Commenter Martin has started his own blog, and I want to be sure to link to it before I end my tenure here. He's got an interesting post up about a recent NEJM series debating the various conduits to improved drug safety. What's clear from his post, and the expert's writings he excerpts, is that there's no easy answer for reforming the FDA.
In my mind, a few things need to happen. Last year the NIH introduced a strict set of new ethics requirements. These rules were not extended to the FDA, to its detriment, as former commissioner Lester Crawford proves quite well.
But we do have to think seriously about the question of improving drug safety in practical ways. Many have called for more extensive/longer clinical trials. That will delay necessary treatments for years. A better method of ensuring safety is to simply forbid the pharmaceutical companies from operating their own clinical trials. Study after study has shown that when corporations run their own trials the drug is much more likely to come out positively.
These results should be public as well; it's ridiculous that we let these corporations put potentially dangerous products on the market but keep questionable safety information to themselves.