Sublette · The Journal of clinical psychiatry 2011 · systematic review and meta-analysis of randomized controlled trials · n=15 trials (916 participants)

Meta-analysis of the effects of eicosapentaenoic acid (EPA) in clinical trials in depression.

Cited 416 times in the scientific literature.

Level 1 - systematic review of randomized trials

Systematic review and meta-analysis of randomized controlled trials.

PubMed 21939614 · doi:10.4088/JCP.10m06634 · record verified 2026-08-29

What was done

A systematic search of PubMed (1960–June 2010) identified prospective, randomized, double-blind, placebo-controlled trials of omega-3 polyunsaturated fatty acid (PUFA) supplementation for major depressive episodes. Fifteen trials involving 916 participants met inclusion criteria. The outcome evaluated was the standardized mean difference in change from baseline to endpoint depression scores between PUFA and placebo groups. Mixed-effects models examined whether efficacy differed based on EPA percentage (EPA ≥ 60% vs. EPA < 60% of total EPA + DHA), treatment duration, age, and EPA dose.

What was found

Supplements with EPA ≥ 60% showed significant clinical benefit over placebo (effect size = 0.532; 95% CI, 0.277 to 0.733; t = 4.195; P < .001). Supplements with EPA < 60% were ineffective (effect size = -0.026; 95% CI, -0.200 to 0.148; t = -0.316; P = .756). Treatment duration and participant age did not moderate effects. Exploratory analyses indicated efficacy was driven nonlinearly by the dose of EPA in excess of DHA across a range of 200 to 2,200 mg/d.

Why it matters

This meta-analysis clarifies conflicting results across clinical trials by identifying supplement composition—specifically an EPA proportion of at least 60%—as the critical determinant of antidepressant efficacy.

Limits

The total sample size was modest (15 trials averaging ~61 participants per study). The search was restricted to English-language publications up to June 2010 from PubMed only. Dose-response conclusions were based on exploratory cross-study modeling rather than head-to-head randomized dose comparisons.

Cited by