I mean for one thing - the placebo effect exists. That's why double blind testing of new drugs is the standard

Agreed, but many drugs have a direct effect, not subtle and it's pretty much immediate to know, let say a user try that new sub-q injectable from a shady forum that discuss X compound that is supposed to cure acne, then follow protocols from other users, then do their report, often with photos, it's useful even if it's not standard, as you have a clear picture of before/after usage of that drug (it's probably fair to say that the acne clearing is that isn't placebo effect), it's useful even after a drug is authorized on the market as well as it gives additional user reports of efficiency. Realistically, some data is better than none.

Saying that wouldn't be useful is kinda like saying HN/Reddit isn't useful because it's all anecdotal, that applies to almost all topics in life.

Lastly, let's not pretend that most doctors do actually follow properly a patient blood before/during/after a course of a drug, in most cases, doctor will just ask How did it go, any side effects?... That's a new "data" for the doc, all anecdotal, almost the same value as a "trusted" rando on Reddit.