Well, you may know it, but it seems that the majority don't.
ERTs were widely considered drugs for the prevention of CV disease in post menopausal women based on huge retrospective studies that appeared conclusive. They even gained FDA approval to use the claim.
Only when prospective randomized studies showed this was not the case did people figure out that women taking ERTs in the prior datasets were inherently healthier than controls. it wasn't ERTs protecting them, it was their lifestyle choices and availability to access to proper health care.
The medical community all supported ERTs. They were surprised by the results of the controlled study. This is why, I personally, am very careful about these "public consensus" of interpretation of data. Cochrane reports and other similar ones are very careful in HOW they evaluate the data NOW. Expert physicians now are much more cautious about interpretation. They also read Kahnemann...
I still remember what a shock it was when the HERS trial came out. I think it was a reminder of the importance of large randomized trials before drawing definitive conclusions. And as you likely know, this wasn't the first time a large, randomized trials reversed the prevailing medical opinion at the time. Other examples include preoperative beta-blockers (POISE), preoperative cardiac revascularization (CARP), intensive glucose targets (NICE-SUGAR), antiarrhythmics for asymptomatic arrhythmias (CAST), and so on.
I agree that we should consider the evidence a recommendation is based on (best example: the notion that opioids are not addictive when used to treat pain is based on a decades-old letter-to-the-editor, not even a study). But this doesn't mean that we can ignore retrospective or non-randomized studies, particularly when they are independently reporting the same outcomes. How often does a large, authoritative, randomized trial actually CONFIRM the results of prior lower quality papers? Actually quite often. Sure, when "medical reversal" occurs, it is sensationalized, and people get understandably upset. Some people might even say "well I'm never trusting anything again unless it's a multi-center RCT published in NEJM."
But in reality, true medical reversal only occurs 15-20% of the time. What this means is that if we only act based on large, authoritative RCT's, and we disregard anything "lesser" (like a large retrospective/observational paper), then statistically, then an actual effective medical therapy/intervention is not being provided for fear of being wrong.
It's very possible that vitamin D has zero benefit for preventing or treating COVID-19 infections. But there is currently low-quality published evidence that it does carry a benefit. The risk of vitamin D supplementation, for me, is probably exceedingly low. Plus, they come in gummy vite form. So until there is more definitive evidence available, the risk/benefit consideration for me favors taking it. And I do.
None of this should be construed as medical advice, btw.