Omega-3 and ageing: what the trials measured when they measured telomeres

The largest randomized telomere trial ever run put 25,871 people on fish oil or placebo and found no effect on telomere length at two years or four. The meta-analyses that say omega-3 helps telomeres are pooling 337 people from five studies.

mixedStudies disagree, or the mechanism is contested.Revised 15 Oct 2026

An open tin of amber softgel capsules on a pale wooden table with three spilled beside it and a fork holding a piece of oily fish on a plate behind, cool north light

The supplement aisle sells a biomarker. The best trial of that biomarker with this supplement in it is one of the largest randomized studies ever conducted, and it returned nothing.

There is a trial of 25,871 people in this field, run for five years, with a telomere sub-study measured in 2,571 blood samples. It is one of the largest randomized studies ever conducted. And for the omega-3 arm, the result is in the abstract of the 2025 paper in the American Journal of Clinical Nutrition, in a single sentence:

Marine n-3 FAs supplementation had no significant effect on LTL at either year 2 or year 4.

Everything else you have read about omega-3 and ageing rests on studies two to three orders of magnitude smaller. This article starts there and works outward, because the structure of this evidence base is the story.

The largest trial, and the null

The VITAL trial — the VITamin D and OmegA-3 TriaL — is a two-by-two factorial, double-blind, placebo-controlled design in a representative sample of 25,871 US adults: women aged 55 and over, men aged 50 and over. Two arms got vitamin D3 at 2,000 IU a day, two got marine omega-3 fatty acids at one gram a day, and the factorial meant the others got both or neither.

The telomere sub-study took 1,054 participants who came in person to the Harvard Clinical and Translational Science Center. Leukocyte telomere length was determined by quantitative PCR at baseline, year 2 and year 4, giving 2,571 samples from 1,031 participants for analysis.

The vitamin D3 result: a reduction in telomere attrition of 0.14 kilo base pairs over four years (95% CI 0.007 to 0.27, p = 0.039), and a trend analysis showing telomere lengths about 0.035 kb higher per year of follow-up than placebo (95% CI 0.002 to 0.07, p = 0.037).

The omega-3 result: nothing, at two years, at four years.

Read the two results side by side. In the same trial, on the same assay, in the same people, the vitamin D arm produced a statistically significant effect and the fish oil arm did not. That is as clean a within-study comparison as this literature will ever produce, and it is not reported nearly as often as the vitamin D finding.

The meta-analyses that say otherwise, and how big they are

The 2022 mini meta-analysis in Biomolecular Concepts is the most-cited supporting evidence and it is small. From 573 records, 5 clinical trials, 337 participants. Overall effect on telomere length: mean difference 0.16, 95% CI 0.02 to 0.30, p = 0.02.

Compare the units. A mean difference of 0.16 in whatever unit the studies reported, from 337 people across five trials, against a null from 1,031 people measured three times over four years. The confidence interval in the small meta-analysis, 0.02 to 0.30, has its lower bound almost at zero. The authors themselves write that larger clinical trials are needed to confirm the finding and to clarify the mechanisms.

The 2022 systematic review in Prostaglandins, Leukotrienes and Essential Fatty Acids reviewed 6 clinical trials and reached a different reading of overlapping material: supplementation did not statistically affect telomere length in three out of three studies that measured it, and affected telomerase activity in two out of four. When a systematic review of six trials finds three of three nulls, and a meta-analysis of five trials finds a positive pooled effect, the difference is almost certainly in the selection and the pooling, not in the biology.

The 2022 review in Nutrients, which asks directly whether omega-3 is the elixir of youth, is candid about the state of it. Telomere shortening is regulated by genetic damage, cell division, ageing, oxidative stress and inflammation — a long list that does not obviously include fish oil. The review covers cross-sectional studies, randomized controlled trials and rodent work, and concludes that although the results of the studies performed to date are not consistent, the vast majority indicate a beneficial effect.

“The vast majority indicate a beneficial effect” is true in the sense that most of the studies are observational and small. It is not a statement about the randomized evidence, which is where the two largest studies are, and where both are null on telomere length.

The 2025 narrative review in Current Opinion in Clinical Nutrition and Metabolic Care gives the observational picture fairly: most but not all studies report beneficial associations for self-reported omega-3-rich food intake, for estimated intake, and for blood-based omega-3 biomarkers. Findings for linoleic acid are inconsistent, and inverse associations are reported for the omega-6 to omega-3 ratio. It also reports a Mendelian randomization finding monounsaturated fatty acids and polyunsaturated fatty acids have a positive effect on telomere length and saturated fatty acids the opposite.

The honest summary of the observational literature is that people who report eating oily fish have longer telomeres, and that people who report eating oily fish differ from other people in about fifty other ways.

The single observational study worth knowing

The 2022 study in the American Journal of Clinical Nutrition is the strongest of the observational data and it is worth knowing exactly what it did. 2,494 US males with telomere length measurement, drawn from four nested case-control studies within the Health Professionals Follow-Up Study. Blood was collected between 1993 and 1995; intake was collected by questionnaire in 1994. People with a history of cancer, diabetes or cardiovascular disease before blood collection were excluded.

DHA intake was positively associated with leukocyte telomere length. Compared with the lowest intake quartile, the percentage differences in telomere length across the second, third and fourth quartiles were −3.7 (−13.7 to 7.5), +7.0 (−4.3 to 19.7) and +8.2 (−3.5 to 21.3).

The fourth quartile confidence interval runs from minus 3.5% to plus 21.3%. That is an interval that includes no effect. A cross-sectional study where the association in the top quartile is not statistically distinguishable from zero is a hypothesis.

Cardiovascular outcomes: the same substance, a different answer

It would be wrong to leave the impression that omega-3 does nothing measurable. On cardiovascular endpoints the trial literature is large and the effects are small and real.

The 2022 amount-stratified meta-analysis in Clinical Nutrition is the most sceptical, and it is the most rigorous: 19 double-blind randomized trials of at least one year, 97,709 participants, with trial sequential analysis alongside the pooled estimates. Result: no statistically significant reduction in all-cause mortality, cardiac death, myocardial infarction or stroke — with one exception, reduced cardiac mortality at the equivalent of two capsules a day (RR 0.55, 95% CI 0.33 to 0.90, p = 0.0169, I² = 0%). Trial sequential analysis reached the required information size only for the lower amounts, where it showed no significant association.

The 2021 meta-analysis in Mayo Clinic Proceedings covers 40 studies with 135,267 participants and reports the effects in the way a clinician should read them, as numbers needed to treat:

myocardial infarction: RR 0.87, NNT 272, high certainty
coronary heart disease events: RR 0.90, NNT 192, high certainty
fatal myocardial infarction: RR 0.65, NNT 128, moderate certainty
coronary heart disease mortality: RR 0.91, NNT 431, low certainty
cardiovascular disease events: no significant effect

A number needed to treat of 272 is a real effect. It is also a much smaller effect than “cardiovascular health”, and the certainty ratings attached to them are not uniform.

The 2019 meta-analysis in the Journal of the American Heart Association covered 13 trials with 127,477 participants over a mean of 5.0 years. Excluding the REDUCE-IT trial, marine omega-3 was associated with myocardial infarction RR 0.92, coronary heart disease death 0.92, cardiovascular disease death 0.93 and total cardiovascular disease 0.97. And the authors note the inverse associations for all outcomes were strengthened after REDUCE-IT was included.

That last sentence is the entire controversy in one line. REDUCE-IT used icosapent ethyl, a specific high-concentration formulation, and the field has not resolved whether its result generalises to the fish oil capsules on shop shelves. Every meta-analysis in this paragraph moves depending on whether it is in.

What these trials do not show

The largest randomized telomere trial found nothing for omega-3. 1,031 participants, 2,571 samples, two and four years, no effect. That is the most important single sentence in this article and it is not a footnote in it.

The positive telomere meta-analysis pools 337 people from five trials, with a confidence interval whose lower bound is 0.02, and its own authors call for larger trials to confirm.

The systematic review of six trials found three of three nulls on telomere length. Where a review and a meta-analysis disagree, the likeliest explanation is the inclusion criteria, not the intervention.

Almost all the positive telomere data is observational and cross-sectional. A 2022 paper in Frontiers in Nutrition is a good example of how little the design constrains: an industry-linked analysis of omega-3 and telomere length in the NHANES sample reporting a positive association, in a cross-sectional design where the exposure and the outcome are measured at the same moment, with a correlation that cannot separate omega-3 from everything else correlated with oily fish consumption, and with a cohort whose mean age is in the late thirties — where telomere length differences of a fraction of a percent are the normal range between individuals.

No trial in this field has an ageing endpoint. Not a year of healthy life, not disability-free life expectancy, not mortality in a healthy population, not a biological age measure validated against outcomes. Cardiovascular events and telomere length are both surrogates, and the surrogates disagree with each other here.

The cardiovascular effect is small and contested. Relative risks of 0.87 to 0.97, numbers needed to treat between 128 and 431, certainty ratings from high to low, and a result set that changes materially depending on whether one specific trial of one specific formulation is included.

Nobody has shown that omega-3 changes any rate of ageing in humans. No trial has measured biological age by a validated composite, no trial has measured epigenetic age against an outcome, and no trial has run long enough for any of it to matter.

And the participants are adults over 50 in a US trial, men in the 1990s Health Professionals cohort, and mixed populations in the meta-analyses. The field has essentially no data in young adults, and the telomere biology that drives the hypothesis — cell division rate, oxidative stress, replicative history — is precisely the biology that differs between a 28-year-old and a 68-year-old.

The defensible version: omega-3 fatty acids have small, real, well-quantified effects on cardiovascular events in adults at risk, with numbers needed to treat in the low hundreds and disputed sensitivity to which formulation you mean. For the thing the supplement aisle actually sells — a slower rate of ageing — the largest randomized measurement is null, the supporting meta-analyses cover 337 people, and the observational associations that generate the hypothesis are in people who eat differently in fifty other ways.