Omega-3 (EPA / DHA)

Verdict: Suggestive (general supplementation) / Probable (high-dose EPA in elevated triglycerides / CV risk per REDUCE-IT) Last reviewed: 2026-04-24 Triangulated against anchor: Rapamycin (Probable)

TL;DR

VITAL (n=25,871) was null on primary CV / mortality endpoints with 1g/day EPA+DHA. REDUCE-IT (n=8,179) showed clear hard-endpoint benefit with 4g/day high-dose EPA (icosapent ethyl) in patients with elevated triglycerides on statins. The popular framing of "fish oil for longevity" rests largely on observational data (DHA blood levels predict lower mortality, ~17% reduction in highest quintile) and meta-analyses showing dose-dependent effects. Verdict: Suggestive for general low-dose supplementation; Probable for the specific REDUCE-IT-equivalent population.

What it is

Marine-source long-chain omega-3 polyunsaturated fatty acids:

• EPA (eicosapentaenoic acid)
• DHA (docosahexaenoic acid)
• ALA (alpha-linolenic acid, plant-source) is precursor; conversion to EPA/DHA is poor in humans.

Sold as fish oil, krill oil, algae oil (DHA-only typical for vegan formulations), and prescription icosapent ethyl (purified EPA, REDUCE-IT formulation). Doses range from 250 mg/day combined (mainstream supplements) to 4 g/day pharmaceutical-grade EPA.

Proposed mechanism

• Reduced triglycerides (the dominant established cardiometabolic effect)
• Anti-inflammatory effects (reduced eicosanoid signaling, specialized pro-resolving mediators)
• Membrane fluidity and neuronal function (DHA is enriched in brain, retina)
• Reduced platelet aggregation
• Possible effects on vascular function and arrhythmia susceptibility

Confidence: Established for triglyceride and membrane effects.

Evidence ladder

Animal models (T3-T4)

EPA/DHA effects on inflammation, lipids, and cardiovascular biology are well-documented in rodents. Lifespan effects modest where reported.

Human (T1)

The story splits sharply by dose and population.

Low-to-moderate-dose general-population RCTs:

• VITAL — 1 g/day EPA+DHA (460 mg EPA, 380 mg DHA), n=25,871, ~5.3 years. Primary endpoints null (major CV events, invasive cancer). Some signals on MI specifically (secondary endpoint).
• ASCEND (diabetics), STRENGTH, OMEMI — all roughly null on primary endpoints with low-dose mixed EPA/DHA.

High-dose EPA in elevated-triglyceride / on-statin populations:

• REDUCE-IT (Bhatt 2019, NEJM) — n=8,179, icosapent ethyl 4 g/day vs placebo in CVD patients with elevated triglycerides on statins. 25% reduction in primary composite of CV events, including reduced cardiovascular death, MI, stroke. Strong, pre-specified, hard endpoints.
• STRENGTH (carboxylic acid mixed EPA/DHA, similar dose) — null. Suggests purified EPA may matter; or that mineral oil placebo in REDUCE-IT was not inert (a contested critique).

Meta-analyses (2021-2025):

• Dose-dependent effects: higher omega-3 dose associates with larger CV mortality reduction.
• EPA-only formulations show greater effects than EPA+DHA in pooled analyses.
• Bayesian analysis supports modest reduction in all-cause mortality (HR ~0.95-0.96) for EPA+DHA supplementation.

Observational (T2):

• Higher blood DHA/EPA quintiles associate with 15-20% lower all-cause mortality across multiple cohorts (Harris 2021, Nat Commun, n>40,000 across 17 studies).
• Effect size larger than most supplement-RCT effects, suggesting either residual confounding or population biology that's hard to capture in supplementation trials.

Confounds

• Mineral oil placebo controversy in REDUCE-IT — whether the placebo was truly inert is debated; some meta-analyses partially discount REDUCE-IT for this reason. Most cardiology consensus accepts the trial result.
• Background fish consumption in trial populations attenuates effect estimates.
• DHA vs EPA matter differently for different endpoints; pooling them obscures effects.
• Reverse causation in observational data — sicker people consume less fish.
• Statin background therapy changes the marginal benefit profile substantially.

Conflict of interest scan

• REDUCE-IT was Amarin-funded (manufacturer of icosapent ethyl). Pre-registered, FDA-grade. Apply 1-tier discount in principle, partially offset by pre-registration.
• VITAL was NIH-funded — no discount.
• General fish-oil industry is large but doesn't fund the major mortality RCTs.
• Net: minimal effective discount on the primary trial evidence.

Human translation

Honest decomposition:

1. For someone on statin therapy with elevated triglycerides: 4 g/day icosapent ethyl (prescription) is supported by REDUCE-IT and is likely beneficial. T1 evidence.
2. For general healthy adults supplementing fish oil for longevity: the 1 g/day VITAL formulation showed no primary-endpoint benefit. Observational signal exists but is modest and partly confounded. Suggestive at best.
3. For someone eating fatty fish 2-3x/week: no supplementation needed; dietary intake is adequate per most guidelines.
4. For DHA-specific benefits (brain, retinal aging): plausible but evidence-thin at the longevity-relevant level; observational signal larger than RCT signal.

Calibrated verdict

Suggestive for general low-dose omega-3 supplementation in healthy adults — VITAL was null and the observational signal is partly confounded. Probable for high-dose purified EPA (icosapent ethyl) in the REDUCE-IT population (elevated triglycerides, on-statin, established/high-risk CVD).

Compared to vitamin D (Mixed), omega-3 has a similar pattern (null primary endpoint in mega-trial, signals on secondaries) but benefits from REDUCE-IT's clear hard-endpoint result in a specific population. The dose-response gradient also gives omega-3 slightly stronger general support than vitamin D.

Compared to GLP-1 agonists (Probable in obese/CV), both are dose-dependent and population-specific. GLP-1s have larger population-relevant benefits and broader endpoint coverage in modern trials.

Confidence interval on verdict

• Could move to Probable (general) if dose-response trials at higher doses (>2 g/day) confirm benefit in non-CVD populations. STRENGTH-style failures argue against this.
• Could move to Mostly hype (general) if remaining mortality-endpoint trials read consistently null and the observational signal is shown to be confounded.
• Most likely 2-year trajectory: stable at Suggestive (general) / Probable (specific).

Open questions

• Q: Is the REDUCE-IT vs STRENGTH discrepancy real (purified EPA matters) or methodological (mineral oil placebo)?
• Q: At what dose does general-population omega-3 supplementation cross from null to beneficial on hard endpoints?
• Q: For populations not yet on statins but at moderate CV risk, what is the evidence?
• Q: Does DHA specifically support cognitive aging beyond what general dietary intake provides?

Sources

• Manson et al. 2019, NEJM — VITAL primary results
• Bhatt et al. 2019, NEJM — REDUCE-IT trial: cardiovascular risk reduction with icosapent ethyl
• Harris et al. 2021, Nat Commun — Blood n-3 fatty acid levels and mortality from 17 prospective studies
• Marine omega-3 supplementation and CVD: meta-analysis of 13 RCTs and 127,477 participants, JAHA 2019
• Meta-analysis of omega-3 effects on CV outcomes, eClinicalMedicine 2021
• N-3 Fatty Acids and CV Health updated review (PMC12628397, 2024-2025)
• Mayo Clinic Proceedings 2023 — Circulating DHA and risk of all-cause and cause-specific mortality

Produced under methodology locked 2026-04-24. Triangulated against rapamycin anchor.