Part two of three in the series What Held Up.
Omega-3 is the only ingredient I know of that is simultaneously the most vindicated and the most debunked thing in the cabinet.
Search for it, and you will find headlines announcing that fish oil does nothing for your heart, built on enormous and well-run trials.
Search slightly differently, and you will find a Cochrane review, graded high certainty, reporting that it cuts early preterm birth by 42 percent.
Both are accurate.
They describe the same molecule tested in different places, and the gap between them is the entire subject of this article.
Asking whether omega-3 supplementation works turns out to be the wrong question. The useful ones are where, in whom, and at what dose.
What This Series Is Testing
Part one set out the standard. Weak evidence shrinks under replication. Strong evidence holds its size, which is necessary but not sufficient, because a stable effect can still be trivial.
The sharper test is predictive.
A compound with a real mechanism tells you in advance where it should work and where it should fail. When researchers follow it into new territory, the hits and the misses both land where the biology said they would.
Creatine passed that test by replication at scale.
Its most famous null result, no memory benefit in young adults, was the mechanism working: young brains are not short of phosphocreatine.
Omega-3 passes the same test.
But its corridor is far narrower than the supplement aisle implies, and its famous failures came from testing outside it.
Does Omega-3 Supplementation Work? Start With the Failures
Two trials did most of the damage to omega-3’s reputation, and both deserve respect.
VITAL randomized 25,871 adults to one gram a day of EPA and DHA for a median of 5.3 years. The primary composite endpoint of major cardiovascular events was not significantly reduced.
ASCEND gave a similar dose to 15,480 people with type 2 diabetes for over seven years. Its primary composite was also null.
Those are large, long, well-conducted trials. The headlines wrote themselves.
Look inside the composites, though, and something more specific appears.
In VITAL, total myocardial infarction fell by 28 percent, with a confidence interval running from 10 to 41 percent. Total coronary heart disease fell 17 percent. In ASCEND, cardiovascular death fell 19 percent.
A composite endpoint bundles heart attack, stroke and cardiovascular death into one number.
If a compound plausibly affects one of those and not the others, the composite dilutes it toward nothing.
That is not a scandal. It is arithmetic, and it is exactly what a mechanism-led prediction would have warned about in advance.
The Baseline Nobody Measured
Here the story rhymes with creatine, and the rhyme is not a coincidence.
EPA and DHA are structural materials.
They are built into cell membrane phospholipids, where they change membrane behaviour and serve as precursors for signalling molecules. Like any structural pool, it saturates. Once your membranes hold as much as they will hold, further intake has nowhere useful to go.
That pool is measurable.
The omega-3 index reports EPA plus DHA as a percentage of red blood cell fatty acids, with a proposed target range of 8 to 11 percent.
Most Western populations sit nowhere near it.
Reported averages include 4.4 percent in American college athletes, 4.97 percent in German world-class athletes, and 3.7 percent in vegans and vegetarians.
Now reconsider VITAL.
In participants eating less than the median 1.5 servings of fish per week, heart attacks fell significantly. In participants already eating more fish, they did not.
That is the creatine pattern again, in a different tissue. Fill a depleted pool, and you see something. Top up a full one, and you do not.
A trial that enrols both groups and reports the average will understate the first and overstate the second.
Treat the index itself with mild caution. The standardized version is a commercial product. The striking risk gradients attached to it, such as the Framingham finding of roughly 35 percent lower total mortality above 6.8 percent, come from observational data rather than randomization.
Where Omega-3 Supplementation Clearly Works
Two effects survive scrutiny without needing a subgroup to rescue them.
If you want the short answer to whether omega-3 supplementation work, it lives here.
The first is triglycerides.
A Cochrane review found that doses from 0.5 grams a day upward reduced serum triglycerides by around 15 percent, with larger effects at higher baseline levels. The mechanism is direct: omega-3 reduces hepatic VLDL production. This is not a contested finding and never has been.
The second is more complicated, and honesty requires presenting it unresolved.
REDUCE-IT gave 8,179 statin-treated patients with high triglycerides four grams a day of icosapent ethyl, a purified EPA, and reported a 25 percent reduction in major cardiovascular events. That is a large effect on a hard endpoint.
Then STRENGTH tested four grams of an EPA and DHA preparation in a similar population and found nothing.
The trials used different placebos. REDUCE-IT used mineral oil, which may have worsened outcomes in its control group and inflated the apparent benefit. STRENGTH used corn oil. The field has argued about this for years without resolving it.
Anyone who tells you confidently which trial was right is telling you about their priors, not the data.
The Best Evidence Is Where Nobody Is Looking
The strongest omega-3 supplementation evidence in existence is not cardiovascular at all, and almost nobody buys fish oil for it.
A Cochrane review of omega-3 supplementation in pregnancy pooled 26 randomized trials in 10,304 women. Preterm birth before 37 weeks fell, with a risk ratio of 0.89 and a confidence interval of 0.81 to 0.97. Early preterm birth before 34 weeks fell much further: a risk ratio of 0.58 across 9 trials and 5,204 women, a 42 percent reduction.
Both outcomes were graded high certainty. In the Cochrane system, that is as strong as a rating gets.
This is a hard endpoint. Babies are either born before 34 weeks, or they are not, and nobody is adjudicating a surrogate marker. It is also mechanistically coherent.
DHA is a structural component of fetal brain and retinal tissue, demand rises sharply in the third trimester, and omega-3 shifts the prostaglandin signalling involved in the onset of labour.
A compound with a real mechanism should work best where the deficit is largest and the biology most specific.
Pregnancy is that case, and the evidence is correspondingly strong.
Do Omega-3 Supplementation Work Better Than Eating Fish?
If baseline intake is the variable that matters, the obvious question is whether you need a capsule at all.
For most people, no.
Two portions of oily fish a week move the index substantially, and fish arrives with selenium, iodine, protein and vitamin D rather than EPA and DHA in isolation.
VITAL’s own subgroup result points the same way: the people who benefited were the ones not eating fish.
There is one genuine exception, and it is the case with the best evidence behind it.
Pregnancy is where omega-3 supplementation has high-certainty support, and it is also where fish consumption runs into mercury advice that limits intake of exactly the larger oily species.
A purified supplement sidesteps that conflict.
It is an unusual situation where the capsule is not a convenience substitute but the cleaner delivery route.
The other exception is dose.
Nobody reaches four grams a day of EPA through food, so the triglyceride and cardiovascular trials at that level describe a pharmaceutical intervention rather than a dietary one.
That distinction gets lost when a trial result is used to sell a one-gram capsule.
Where Omega-3 Supplementation Starts Hurting
A genuine dose-response runs in both directions, and this is where the article stops being good news.
A meta-analysis of 8 randomized cardiovascular outcome trials covering 83,112 participants found that omega-3 supplementation increased atrial fibrillation, with a risk ratio of 1.24.
The dose pattern is the part that matters.
At one gram a day or less, the risk ratio was 1.12. Above one gram a day, it rose to 1.51, with a significant test for interaction.
So the same high doses that produce the clearest triglyceride effects also carry the clearest arrhythmia signal. That is not a contradiction. It is what a real mechanism looks like when you push it past its useful range.
One more boundary worth naming.
The plant omega-3, ALA, converts to EPA and DHA at reported rates below 15 percent, and far lower for DHA specifically. Flaxseed is a fine food. It is not a reliable route to the molecules these trials were testing.
Bottom Line
So does omega-3 supplementation work?
Yes, inside a corridor with clearly marked walls.
- They lower triglycerides reliably and dose-dependently.
- They reduce preterm and early preterm birth with high-certainty evidence.
- They plausibly reduce heart attacks in people whose baseline intake is low, which the large trials support more strongly than their headline results suggested.
- They do not meaningfully help people whose membranes are already replete.
- And above a gram a day, they measurably raise atrial fibrillation risk.
The famous failures were not a refutation.
- They were a dose-and-population mismatch that a mechanism-first reading would have flagged beforehand.
- The trials asked whether topping up a largely sufficient population changed a bundled endpoint. The biology had already answered no.
Next in this series: folic acid, where there is no corridor to map and no subgroup to hunt for. One trial, one hard endpoint, an effect so large it ended the argument and rewrote the food supply of entire countries.
References
- Manson JE, Cook NR, Lee IM, et al. Marine n−3 fatty acids and prevention of cardiovascular disease and cancer (VITAL). New England Journal of Medicine. 2019;380:23–32.
- ASCEND Study Collaborative Group. Effects of n−3 fatty acid supplements in diabetes mellitus. New England Journal of Medicine. 2018;379:1540–1550.
- Bhatt DL, Steg PG, Miller M, et al. Cardiovascular risk reduction with icosapent ethyl for hypertriglyceridemia (REDUCE-IT). New England Journal of Medicine. 2019;380:11–22.
- Nicholls SJ, Lincoff AM, Garcia M, et al. Effect of high-dose omega-3 fatty acids vs corn oil on major adverse cardiovascular events (STRENGTH). JAMA. 2020;324(22):2268–2280.
- Middleton P, Gomersall JC, Gould JF, et al. Omega-3 fatty acid addition during pregnancy. Cochrane Database of Systematic Reviews. 2018;11:CD003402.
- Gencer B, Djousse L, Al-Ramady OT, et al. Association between omega-3 fatty acid treatment and atrial fibrillation in cardiovascular outcome trials: a systematic review and meta-analysis. Cardiovascular Drugs and Therapy. 2021.
- von Schacky C. Omega-3 index in 2018/19. Proceedings of the Nutrition Society. 2020;79(4):381–387.
- National Institutes of Health, Office of Dietary Supplements. Omega-3 fatty acids: fact sheet for health professionals.
- Bork CS, Lundbye-Christensen S, et al. Recent clinical trials shed new light on the cardiovascular benefits of omega-3 fatty acids. Methodist DeBakey Cardiovascular Journal. 2019;15(3):171–178.
