A network meta-analysis published in Molecular Psychiatry on 1 September 2026 reports that the larger a supplement's measured effect on depressive symptoms, the weaker the study measuring it tended to be. The team, led by Michele Fornaro and Chiara Di Lorenzo at the University of Naples Federico II, pooled 163 randomised controlled trials covering 137 distinct compounds and compound combinations in 15,757 adults with major depressive disorder, searching the databases up to 1 July 2025. Several molecules beat placebo with what the authors call "very large effect sizes". When the analysis was restricted to compounds with at least two trials, low risk of bias, no commercial sponsorship and no statistical outliers, three still produced results the authors judged unrealistic: eicosapentaenoic acid, or EPA (20 trials, 8,483 participants), vitamin D supplements (three trials, 344 participants) and the St John's wort extract ZE117 (three trials, 207 participants).
Why the biggest numbers were the least believable
The authors ran a meta-regression of effect size against adapted AMSTAR-Plus quality scores and found that larger standardised mean differences went with lower scores, which they read as an indicator of lower study quality. Heterogeneity across the network was high, at 85.3 per cent, and the analysis showed global inconsistency, which the authors say was explained by their subgroup and sensitivity analyses, although they report no evidence of publication bias. Their stated conclusion is that "upon filtering low-quality evidence, only a few nutraceutical/phytoceuticals showed potential for depressive symptoms, warranting more rigorous trials". Confidence in the evidence is rated low to very low. The paper is open access and lists academic funding, with no commercial sponsor.
What this means in practice
A single striking trial result is the raw material for most supplement marketing, and this paper is a measurement of how far that material can be trusted. The three compounds that survived the authors' filters did so on very small evidence bases, three trials and a few hundred participants in two of the three cases, and the researchers still described the size of the effects as unrealistic rather than promising. Nothing in the paper establishes that any product treats or improves a mood disorder, and its authors say plainly that more rigorous trials are needed. What it does change is how much weight a headline built on one nutrient study deserves. Perqen lists ingredient and amount information on every product page across its full supplement catalogue, which is what lets a reader check a claim in a headline against what is actually in a tub.
How Perqen members save
Perqen membership costs £4.99 a month or £29.99 a year and unlocks member pricing across the whole catalogue, rather than hiding a markup in the product price. The terms are set out on the membership page.
This article reports published news and research and is for information only. Food supplements are not a substitute for a varied diet or medical care, and no product mentioned is intended to diagnose, treat, cure or prevent any disease. Speak to your GP or pharmacist before changing what you take, especially if you are pregnant, breastfeeding, or taking prescribed medicines.