We have a hard enough time keeping desperate people from trying things that we know for a fact are scams. Perhaps doctors could put the studies in clearer context when prescribing, but they are always clear that these drugs aren't perfectly effective and that it's kind of a crapshoot.
When you've got patients who will die anyway, it's no surprise that we end up treating them as a final stage of clinical trials. You need to be honest about that and get consent, but many will give it willingly.
And those Iowa farmers and shopkeepers and schoolteachers at the end of their lives sign the form. An unreasonable population of them. For no reason but to help.
That would allow outcome based decisions about keeping that approval or comparing new treatments to the actual efficiency of existing treatments.
I'm pretty sure we do this?
I mean I suppose we don't have laws which mandate this but medical researchers are constantly measuring and publishing how effective existing legal treatments are. I am sure researchers would love to find which treatments do not actually work and since publishing such work would affect treatment guidelines and have a huge impact.
Much of that would be addressed by having a single-payer healthcare system, which could mandate record standards and submission of specific portions of the data in ways that do not unduly risk identifying patients.