Supplements for depression: big effect sizes the authors call unrealistic

A network meta-analysis in Molecular Psychiatry, published 1 September, pulls together the randomised evidence on "nutraceuticals" and "phytoceuticals" for major depressive disorder — omega-3 fatty acids, vitamins, herbal extracts and the like. Its most important contribution is not a ranking of which supplement works best. It is a caution about how to read effect sizes that look too good to be true [s1].

The scope

The authors searched the major databases through 1 July 2025 for randomised trials of nutraceuticals or phytoceuticals in major depressive disorder, compared against placebo or against treatment as usual with antidepressants [s1]. They assembled 163 studies covering 15,757 participants and 137 distinct compound combinations, with change in depressive symptoms, treatment response and all-cause discontinuation as co-primary outcomes [s1]. Across that whole set, several molecules and classes outperformed placebo with what the paper calls "very large effect sizes" [s1].

That phrase is the tell. In depression research, effect sizes this large are a red flag, because they exceed what even standard antidepressants achieve — and when a cheap, over-the-counter supplement appears to crush placebo, the more likely explanation is a problem with the trials than a miracle in the bottle.

The numbers, and why they invite suspicion

To probe that, the authors restricted the analysis to compounds with at least two trials, tested against something other than antidepressants, at low risk of bias, from non-sponsored trials, and excluding statistical outliers [s1]. Even after that filtering, several effect sizes remained "higher-than-expected" or, in the authors' word, "unrealistic" [s1]:

• Eicosapentaenoic acid (EPA), an omega-3: 20 trials, 8,483 participants, standardised mean difference −1.67 (95% CI −2.38 to −0.97) [s1].
• Vitamin D3: 3 trials, 344 participants, SMD −1.59 (95% CI −2.71 to −0.46) [s1].
• St John's wort extract ZE117: 3 trials, 207 participants, SMD −1.02 (95% CI −1.90 to −0.15) [s1].

A standardised mean difference of −1.67 is enormous; for comparison, most antidepressants land around −0.3. The analysis also found a more modest, and more believable, effect: curcumin reduced anxiety symptoms with an SMD of −0.58 (95% CI −0.99 to −0.18) in the low-risk-of-bias analysis [s1].

Why the authors distrust their own top results

The paper reports substantial statistical heterogeneity (I² = 85.3%; 95% CI 81.9%–88.0%) and significant global inconsistency across the network, even as it found no evidence of publication bias [s1]. High heterogeneity means the individual trials disagree with one another far more than chance would predict — a sign that they differ in patients, doses, or quality in ways that make a single pooled number misleading [s1]. In plain terms: the very large effect sizes are most plausibly artefacts of small, heterogeneous, sometimes low-quality trials rather than a true measure of how much these compounds help.

This is not the first meta-level analysis to land in cautious territory. A 2019 meta-review in World Psychiatry found the strongest supplement evidence in mental disorders was for omega-3s in depression, while judging most other nutrient claims weak or unproven — a more tempered picture than the raw effect sizes here might suggest [s2]. The convergence is telling: two independent syntheses, seven years apart, both point to EPA as the compound with the least shaky case and treat most of the rest as unsettled.

The heterogeneity figure is the technical heart of the caution, and it is worth translating. An I² of 85.3% means roughly that most of the variation between the trials' results reflects real differences among the studies rather than sampling noise — different patients, doses, formulations and study quality — so averaging them into one number risks producing a figure that describes no actual trial [s1]. The absence of detectable publication bias makes the story more interesting, not less: the inflated effects do not appear to come simply from unflattering trials going unpublished, which pushes the explanation toward the design and conduct of the small studies themselves [s1]. That is why the authors' own verdict — that the standout effect sizes are "unrealistic" — is the most important sentence in the paper, and the one a reader should carry away [s1].

How to read it

The honest summary is uncomfortable for both camps. Supplement sceptics cannot claim the trials show nothing — EPA in particular has repeatedly separated from placebo. But supplement enthusiasts cannot take the eye-popping effect sizes at face value, because the authors who calculated them explicitly warn that they are not credible [s1]. The useful signal buried in the noise is narrow: EPA has the most data behind it, curcumin's anxiety effect is of a believable magnitude, and everything else needs larger, better, independent trials before any confidence is warranted.

Some of these compounds also carry real interaction risks — St John's wort notably interferes with many prescription drugs — which is one reason decisions about them belong with a clinician rather than a supplement aisle.

This article is informational and does not constitute medical advice.

Sources

• [s1] Nutraceuticals and phytoceuticals for the treatment of major depressive disorder: a systematic review and network meta-analysis. Molecular Psychiatry, 1 September 2026. https://doi.org/10.1038/s41380-026-03759-1
• [s2] The efficacy and safety of nutrient supplements in the treatment of mental disorders: a meta-review of meta-analyses of randomized controlled trials. World Psychiatry, 9 September 2019. https://doi.org/10.1002/wps.20672