How we've ended up regulating medicine to the nth degree, but when it's teeth we're like "oh well, lol", continues to mystify me.
I see what you mean.
But I'm a computer programmer, and if someone asked me to find a top quality academic study proving, beyond a shadow of a doubt, that it's a good idea to indent your code - I couldn't point you to one.
Not only is bias introduced from accidental collection flaws, it’s also often tampered with intentionally cherry picking data, choosing interesting data or in some cases flat out falsifying data. In addition, evidence based reasoning often suffers from there being a lack of evidence to make a decision from. Or in some cases some critical aspect surrounding the decision is very niche to the case so the data may not take that into account unless it’s highly tailored data (evidence based reasoning tends to focus on breadth of applicability because gathering evidence is a long and often expensive process).
There’s still a lot of place for using theory and reasoning in conjunction with or in absence of data. Things like experience, professional opinion, etc. Medicine should be no different in that regard to any profession. The key is of course to always strive for sound empirical evidence/data where possible, but to use sound documented reasoning and theory in its absence if you want the best objective results.
I've personally been on the receiving end of "the data we collected shows...x" (in a non-medical setting), but when I asked to have a look at it, it turned out that while this was true for a large part of the population sampled, there was a material difference between that population and a smaller population that can be clearly identified and for the latter, the data showed the exact opposite conclusion.
(think 100 men and 30 women, kind of scenario, except the difference wasn't gender, but job role).