I think that is that this article claims.LINK
What they ignore: a lot of these "Evidence" studies ignore real life, and many are biase.
often they are "Garbage in Garbage out".
Often they are small numbers. Or a scholarly article that takes a lot of GIGO articles and summarizes them.
Often the studies are of small numbers of people.
Often they ignore cost, or hman beehavior For exaple, studies taht show newer medicines that cost a lot more work better, but the patients know their older meds. Or the experts decide taking many pills several times a day give a better result (eg when they stopped all those lovely combinations of medicines for blood pressure, so our patients had to take three pills a day instead of one: They don't like to take pills or forget to take the twice a day pill in the evening).
And of course, they are assuming we are treating what we say we are. If a person comes in to complain about her husband for a half hour, we list her problem as blood pressure. But of course she isn't having a problem with her pills, so we don't chage them to the newest one. Or don't give any
pills at all.
And of cours no one wants to suspect a lot of drug companies bias the "evidence".