Comment by nxobject
6 hours ago
Be careful - many studies in the Vitamin D meta-analysis *enrolled patients already taking antidepressants.* [1] Reporting effect sizes without specifying "on which population?" is misleading.
(As an aside, Cohen would be the person not to tell you to assign qualitative values to effect sizes. They are as arbitrary as any other threshold used by working statisticians.)
[1] https://www.cambridge.org/core/journals/psychological-medici...
EDIT – that is, please don't draw the conclusion that you can substitute supplements for antidepressants. The meta-analyses don't seem designed to examine that hypothesis, and I doubt anyone would ever participate in a such a trial. In general (and as a working biostatistician), I would be very, very, very cautious applying estimates of average effect to myself, you, or any other individual person in a field as murky as psychiatry. That's why even the stingiest American health insurance plans still have an incredibly large range of antidepressants in their formularies.
Hi, author of the blog post here! Thanks for bringing this up -- it wasn't my intention to say one should replace antidepressants with vitamins (the conclusion even says "(Don't quit your existing antidepressants if they're net-positive for you!)", but you're right that the intro may give that impression. I'll edit the intro to say you can stack, not substitute, regular antidepressants.
> many studies in the Vitamin D meta-analysis enrolled patients already taking antidepressants.
Yes, and that's even more encouraging, that there's still effects of Vitamin D on major depression even if already on antidepressants! This suggests we can "stack" the interventions.
Table 1 of the meta-analysis ( https://pmc.ncbi.nlm.nih.gov/articles/PMC11650176/ ) shows the raw sub-group analysis. There were 9 studies on patients using antidepressants, 13 on patients who weren't, the rest were Mixed or Not Reported (...how do 6 studies just not report that?) Anyway,
Effect size of Vit D for people on antidepressants: −0.54 (−0.85, −0.23)
Effect size of Vit D for people NOT on antidepressants: −0.28 (−0.40, −0.16)
Both negative. Weirdly, the effect of Vit D seems to be a bit stronger for people on antidepressants, but the difference isn't statistically significant at the p<0.05 level (P subgroup difference is 0.23)
(As for why those effect sizes, -0.54 & -0.28, are lower than what I (and that meta-analysis itself) report, -1.82, that's because the majority of RCTs for any group used far less than 5000 IU. Table 2 in that paper shows the effect (with 95% CI) for various dosages.)
I'll lightly edit my blog post to emphasize stack them, don't substitute. Thanks again for your comment!
Based on your own citation, this is untrue. The citation you have linked is a meta analysis of 31 trials, many of which specify no medication use at recruitment.
You are however correct about population being important (which is a big reason meta analyses can be very useful).
I said "many", not "all". (Many don't indicate, either.)