Can a Daily Multivitamin Slow Aging? What the COSMOS Epigenetic-Clock Trial Found

Why a multivitamin result is suddenly interesting

For decades the multivitamin has been the establishment's least exciting supplement — cheap, ubiquitous, and repeatedly shown to do little for hard endpoints like heart attacks or cancer in healthy, well-fed adults. So a 2026 headline that a daily multivitamin might slow biological aging is worth reading carefully, because the claim rests on a relatively new kind of measurement: the epigenetic clock. The question the new trial actually asked was narrow and specific — does a daily multivitamin change the rate at which these clocks tick? — and the honest answer is a qualified, well-measured "a little."

What an epigenetic clock measures

As cells age, chemical tags called methyl groups are added to and removed from DNA in surprisingly patterned ways. An epigenetic clock is a statistical model that reads hundreds of these methylation sites and outputs an estimate of biological age. The first-generation clocks built by Steve Horvath and Gregory Hannum were trained to predict chronological age and are remarkably accurate at it. The clocks that matter for longevity research, though, are the second generation — PhenoAge, GrimAge and the pace-of-aging measure DunedinPACE — which were trained not on the calendar but on clinical biomarkers, mortality and the speed of physiological decline.

That distinction is the whole game. Second-generation clocks are consistently the stronger predictors of disease and death. A 2026 analysis in GeroScience comparing ten second-generation clocks confirmed they show tighter associations with lifespan and mortality than most first-generation models, and found that accelerated PhenoAge in particular tracked with smaller hippocampal volume — a brain-aging signal — across two independent cohorts.^[3] When a trial reports moving GrimAge or PhenoAge, it is claiming to nudge the version of the clock that is actually tied to health, which is exactly why the COSMOS investigators built their study around them.

The COSMOS trial: design and numbers

The new data come from a prespecified ancillary study of COSMOS — the COcoa Supplement and Multivitamin Outcomes Study — published in Nature Medicine in 2026 by Sidong Li, Howard Sesso and colleagues at Brigham and Women's Hospital and Harvard.^[1] COSMOS was a large randomized, placebo-controlled trial, and this sub-study analysed blood from 958 participants (482 women and 476 men) who provided samples at baseline and again after two years. Everyone was randomized in a factorial design to a daily multivitamin-multimineral (Centrum Silver), a cocoa extract (500 mg cocoa flavanols per day, including 80 mg of (−)-epicatechin), both, or matching placebos.

The researchers ran five DNA-methylation clocks on each sample — PCHannum and PCHorvath (first generation), plus PCPhenoAge, PCGrimAge and DunedinPACE (second generation). The headline finding was clean. Compared with placebo, daily multivitamin supplementation reduced the yearly rate of increase of the second-generation clocks: a between-group difference in annual change of −0.113 years for PCGrimAge (95% CI −0.205 to −0.020; P = 0.017) and −0.214 years for PCPhenoAge (95% CI −0.410 to −0.019; P = 0.032).^[1] In plain terms: over the two years, the multivitamin group's GrimAge crept up about a tenth of a year less per year than the placebo group's.

Two further details sharpen the picture. First, the benefit was not evenly spread. Among participants who were already showing accelerated biological aging at baseline, the effect on PCGrimAge was roughly −0.236 years per year; among those aging normally or slowly it was a negligible −0.013, and the difference between those groups was itself statistically significant (P = 0.018 for interaction).^[1] The multivitamin, in other words, seemed to do most of its work in the people furthest from where they should be. Second, the cocoa extract — despite cocoa flavanols' cardiovascular reputation — moved none of the five clocks at all.^[1]

How this compares to caloric restriction

To judge whether a tenth of a year is impressive or trivial, it helps to compare it with the most rigorously studied longevity intervention in humans: caloric restriction. The CALERIE trial randomized 220 healthy adults to either 25% caloric restriction or their normal diet for two years, then applied the same family of methylation clocks.^[2] Cutting a quarter of daily calories for two years slowed the DunedinPACE pace-of-aging measure — but, notably, did not significantly move PhenoAge or GrimAge, and the authors described the treatment effect sizes as small.^[2]

Set side by side, the two trials tell a consistent and sobering story. A demanding two-year dietary intervention and a daily pill both register on epigenetic clocks, both do so modestly, and they even move different clocks — which is a reminder that these algorithms are not interchangeable and each captures a slightly different slice of aging biology. The multivitamin result is real; it is not larger than what serious caloric restriction achieves. Similar small, clock-level signals have now been reported for other interventions covered here, from GLP-1 drugs like semaglutide to established NAD+ precursors.

What might explain the effect — and what it isn't

Mechanistically, the finding is plausible without being mysterious. Many older adults have marginal intakes of micronutrients — folate, B12, vitamin D, zinc — that feed directly into one-carbon metabolism and DNA methylation, the very machinery the clocks read out. Topping up a genuine shortfall could reasonably tidy the methylation landscape, which would also explain why the benefit clustered in those aging fastest: they may simply have had the most room to correct. This is a nutritional-repletion story far more than a "longevity drug" story.

It is equally important to say what the trial does not show. It does not show that multivitamins make people live longer, prevent a single disease, or benefit young, well-nourished adults who already get these nutrients from food. Epigenetic clocks are validated predictors of population-level risk, but a change in a clock is a surrogate marker — it has not been proven that pushing the marker down with a supplement translates into fewer heart attacks or extra years of life. The authors of the COSMOS analysis are careful on exactly this point, writing that the effects, though encouraging, are small and that additional studies are needed to determine their clinical relevance.^[1]

Honest caveats

Four limits deserve to be stated plainly. First, effect size: a difference of one or two tenths of a year, accumulated over two years, is at the very edge of what these assays can reliably detect, and small absolute numbers can be swamped by measurement noise and by the assumptions baked into each clock. Second, surrogate versus outcome: no one has yet shown that nudging GrimAge with a pill lowers real-world mortality — that is the study that has not been done. Third, generalisability: COSMOS enrolled older adults (the multivitamin was Centrum Silver), so the result speaks to that group, not to a 30-year-old with a good diet. Fourth, the clocks disagree with each other — the multivitamin moved PhenoAge and GrimAge but not DunedinPACE, caloric restriction did the reverse, and that lack of overlap is a caution against treating any single clock reading as a verdict on your biology.

There is also a wider context worth holding in mind: the same clocks are pushed around by things entirely outside the supplement aisle. A 2025 study in Science Advances found that cumulative exposure to extreme outdoor heat was associated with measurably accelerated epigenetic aging in older Americans — on the order of one to two years on PhenoAge over the longest exposure windows.^[4] Environment, in that analysis, moved the needle further than a daily multivitamin did in COSMOS. It is a useful sense of scale: the fundamentals with the hardest human evidence — a high VO₂ max, preserved muscle, sleep, and not being chronically nutrient-deficient — still dwarf what any single pill contributes.

The bottom line

The COSMOS result is a rare thing in supplement science: a positive finding from a large, randomized, placebo-controlled trial with a prespecified analysis and a credible biological rationale. Taken at face value, a daily multivitamin modestly slowed two clocks tied to health and mortality, especially in older adults who were aging quickly. That is genuinely worth knowing. It is also, on the numbers, a small effect on a surrogate marker — a reason to make sure you are not micronutrient-deficient, not a reason to expect a multivitamin to add years to your life. As a floor for people with marginal diets it is cheap and defensible. As a longevity intervention, it remains a modest supporting actor, not the lead.