One of the problem areas is anachronistic regulatory models. The FDA drug approval process is onerous. The FDA acts as if the worst error that it can make is to approve a drug that it later regrets approving. Essentially, it treats every drug as snake oil unless proven otherwise. As scientific progress speeds up, the FDA process turns into a significant bottleneck.

Bartley J. Madden and Gregory Conko propose,

However, after making a preliminary demonstration of safety and efficacy by completing Phase I trials and at least one Phase II trial, drug manufacturers would be given the option to place an experimental product on a parallel Free To Choose track that would enable patients, advised by their doctors, to make an informed choice to use the experimental drug. Drug makers could opt to continue pursuing a standard FDA approval—with all the attendant clinical testing that would require—concurrent with placing a drug on the Free To Choose track. Or, they could put off standard FDA-regulated clinical trials indefinitely, using Free To Choose track experience to guide future development decisions and randomized control trial designs.

Thanks to Alex Tabarrok for pointing me to the article. However, I think that there may be other instances in which the social cost of denying access to a drug is high relative to the risk that the drug will cause harm.

Somebody who is dying or enduring great suffering might have little or nothing to lose from trying an unproven remedy even before Phase I trials are complete.

As medicine becomes more customized, to the genetic makeup of the patient (or, in the case of cancer, the genetic makeup of the tumor), a question arises as to what is the relevant population for a clinical trial.