> You’re cherry picking papers.

I just picked the most recent meta-analysis I could find, which also specifically estimates the dose-response curve. (Since averaging the effect at 400 IU and 4000 IU doesn't make sense.)

> Others have already shared other studies showing no significant effects of Vitamin D intervention.

Yes, and the Ghaemi et al 2024 meta-analysis addresses the methodological problems in those earlier meta-analyses. (For example, they average the effects at vastly varying doses from 400 IU and 4000 IU)

> According your numbers, taking Tylenol would be worse than placebo alone! 0.4 vs 0.7

No, I understand this fine. Taking Tylenol would give you active medication + placebo + time, which is 0.4 + 0.7 + X > *1.1.* Taking open-label placebo is just placebo + time = *0.7* + X.

(Edit: Also, these aren't "my" numbers. They're from a major peer-reviewed study published in Nature, the highest-impact journal. I don't like "hey look at the credentials here", but I bring it up to note I'm not anti-science, see below paragraph)

===

Stepping back, I suspect the broader concern you have is you (correctly!) see that supplement/nutrition research is sketchy & full of grifters. And at the current moment, it seems to play into the hands of anti-establishment anti-science types. I agree, and I'll try to edit the tone of the article to avoid that.

That said, there still is some good science (among the crap), and I think the better evidence is accumulating (at least for Vitamin D) that it's on par with traditional antidepressants, possibly more. I agree that much larger trials are required.

> They're from a major peer-reviewed study published in Nature, the highest-impact journal.

No, the domain name is nature.com because it's in a Nature Publishing Group journal, Scientific Reports, which is their least prestigious journal.

It's a common mistake, and they do that on purpose, of course, to leverage the Nature brand.

It's also a mistake that implies a complete lack of familiarity with scientific publishing, unfortunately, which makes it a bit difficult to take your judgements regarding plausibility very seriously.