Omega-3 (EPA+DHA)

GISSI-Prevenzione Investigators, group-authored (Lancet, 1999; PMID PubMed ID: the permanent number the US National Library of Medicine gives a published paper. Each one links to that paper's record on PubMed. 10465168) reported reduced all-cause mortality after myocardial infarction at 1 g/day of mixed marine n-3, in 11,324 patients. All-cause mortality is a hard clinical endpoint The outcome a trial measures to decide whether the treatment worked, such as heart attacks, blood pressure or a symptom score. and the trial is large. The letter is B rather than A because the result has not replicated The same finding reproduced by a separate research group. A result seen once can be chance or bias; a replicated one is harder to explain away. in the current treatment era: the trial predates universal statin therapy, and ASCEND (New England Journal of Medicine, 2018; PMID 30146932) tested 1 g/day of marine n-3 in 15,480 patients on modern background therapy and found no reduction in serious vascular events. One large randomized trial on a hard endpoint supports B. Replication is what A requires. Provenance note: GISSI-Prevenzione carries two published errata (Lancet 2001;357(9256):642 and Lancet 2007;369(9556):106). Neither is a retraction or an expression of concern, and PubMed indexes no separate record for either. EBL discloses them rather than leave them unmentioned. The Grade B does not rest on GISSI alone: it reflects ASCEND (PMID 30146932) testing the common formulation and finding no benefit, so a correction to GISSI in either direction would not raise the letter.

REDUCE-IT (New England Journal of Medicine, 2019; PMID 30415628): 8,179 statin-treated patients with elevated triglycerides, prespecified primary endpoint The one outcome a trial names in advance as its main test. Results on other outcomes carry less weight., a 25 percent reduction in major adverse cardiovascular events. This grade attaches to prescription icosapent ethyl at 4 g/day, not to over-the-counter mixed EPA and DHA. The molecule, the dose and the population all differ. WHY THIS IS A B AND NOT AN A: Grade A on this site requires that an effect replicate across independent research groups. This record carried an A until September 2026 and cited no replication at all: its two PMIDs were REDUCE-IT and STRENGTH, and STRENGTH was null. The nearest genuine candidate is JELIS (Yokoyama et al., Lancet, 2007; PMID 17398308): 18,645 Japanese hypercholesterolemic patients randomized to purified EPA plus a statin or a statin alone, five years, a reduction in major coronary events. That is a real second randomized trial of purified EPA on a cardiovascular endpoint from an independent group. It is not a replication of what this record grades. JELIS used 1,800 mg/day, less than half the 4 g/day this grade attaches to, and EBL grades an intervention at the dose and protocol studied rather than by the name of the molecule. JELIS was open-label Everyone in the trial knows which treatment they are receiving, so expectations can shape the results, especially for symptoms people report themselves. with blinded endpoint assessment, against a statin-only control rather than a placebo An inactive substitute made to look like the real treatment, so participants cannot tell which one they received.. Its population was Japanese with high baseline fish intake and was not selected for elevated triglycerides, where REDUCE-IT was. A trial in a different population at a different dose narrows the question rather than closing it. WHY THIS IS A B AND NOT A C: Grade C describes a single positive trial nobody has independently repeated. That is not this. Two randomized trials in roughly 26,000 patients, run by independent groups on two continents, both tested purified EPA against cardiovascular endpoints and both were positive. The evidence is not scarce. It is abundant and imperfectly aligned, which is a different problem and a different letter. Grade B on this site is at least one well-designed randomized trial showing a meaningful effect, supported by further trials rather than standing alone. That is what this is. THE PUBLISHED CHALLENGE: STRENGTH (JAMA, 2020; PMID 33190147) was null on the same endpoint. It is named here rather than omitted, and it does not lower the grade further because it tested a different agent: omega-3 carboxylic acid combining EPA and DHA against corn oil, reaching substantially lower plasma EPA. The open question that remains is the mineral oil comparator used in REDUCE-IT, which may have raised LDL Low-density lipoprotein cholesterol, the cholesterol fraction most closely tied to plaque in the arteries. and CRP in the control arm and inflated the apparent effect. WHAT WOULD MOVE THIS TO A: a randomized, placebo-controlled trial of purified EPA at approximately 4 g/day, in a population selected for elevated triglycerides, run by an independent group, reporting a cardiovascular outcome. That trial does not currently exist.

VITAL (NEJM, 2019; PMID 30415637) tested the intervention most people actually buy, mixed marine n-3 at 1 g/day, in primary prevention and found no reduction in major cardiovascular events or cancer. This is the closest trial to general-population supplement use, and it was null.

Bischoff-Ferrari et al. (Nat Aging, 2025; PMID 39900648): post hoc analysis of 777 participants from the DO-HEALTH randomized trial. Omega-3 alone slowed three of four DNA methylation clocks (PhenoAge, GrimAge2, DunedinPACE An epigenetic test that estimates how fast a person is aging now, in years of biological aging per calendar year.) by a small amount, about 3 to 4 months over 3 years. A methylation clock is not an established stand-in for a clinical outcome, the analysis was post hoc, and it has not been repeated. One co-author founded a nonprofit that licenses patents on several of the clocks, including GrimAge. C.