Among nearly 1,000 older adults randomly assigned to take either a daily multivitamin or a placebo for two years, those in the supplement group aged more slowly at the molecular level, according to findings published in Nature Medicine in March 2026. The difference showed up across all five epigenetic aging clocks the researchers measured. And the people who benefited most were those whose DNA already showed signs of accelerated aging when the study began.
The results come from a prespecified analysis nested inside COSMOS, the COcoa Supplement and Multivitamin Outcomes Study, one of the largest randomized supplement trials ever conducted. Led by investigators at Brigham and Women’s Hospital and Harvard, including epidemiologist JoAnn Manson, COSMOS enrolled more than 21,000 older Americans to test whether a daily Centrum Silver multivitamin, a cocoa flavanol extract, or both could prevent cancer and heart disease. The parent trial’s answer to those big questions was largely no. But the new epigenetic data suggest something subtler may be happening beneath the surface.
What the trial measured and found
Epigenetic clocks estimate biological age by reading patterns of chemical tags, called methyl groups, that accumulate on DNA over a lifetime. These tags don’t change the genetic code itself, but they influence which genes are active and which are silenced. Researchers have developed algorithms that translate thousands of these methylation marks into a single number: a person’s biological age. When that number runs ahead of the calendar, it predicts higher risks of chronic disease, cognitive decline, and earlier death.
For this analysis, the team examined blood samples from 958 COSMOS participants, all women 65 or older and men 60 or older, and scored each sample on five established clocks: PCHannum, PCHorvath, PCPhenoAge, PCGrimAge, and DunedinPACE. Each clock uses a different set of methylation sites and was built to capture a distinct dimension of aging, from cellular wear to predicted mortality risk.
Across all five, the multivitamin group aged more slowly than the placebo group. The two clocks most tightly linked to mortality, PCGrimAge and DunedinPACE, showed statistically significant separation. On average, the supplement group accumulated roughly four fewer months of biological aging over the two-year period. That gap is modest in absolute terms, but given that the intervention costs pennies a day and requires no prescription, even a small, real effect would have outsized public health implications.
The cocoa extract arm, by contrast, showed no significant effect on any of the five clocks, a result that mirrors cocoa’s limited impact on the trial’s clinical endpoints and serves as a useful internal comparison. If the multivitamin signal were simply noise or a placebo artifact, you would expect the cocoa arm to look similar. It didn’t.
The subgroup finding that matters most
Perhaps the most striking result involves the people who entered the trial already aging faster than their birth certificates suggested. Among participants whose baseline epigenetic age exceeded their chronological age, the multivitamin produced a larger and more consistent slowing effect. Those who were aging roughly on schedule saw a smaller benefit.
This pattern carries a practical implication. Rather than pushing already healthy biology into some enhanced state, the multivitamin appears to partially correct molecular deficits in people who are falling behind. Think of it less as a performance booster and more as a backstop: filling in gaps that may accumulate from marginal nutrient shortfalls, low-grade inflammation, or other stressors that accelerate biological wear.
If the finding holds up in future studies, it could eventually help clinicians identify which patients are most likely to benefit from supplementation, turning a blunt, one-size-fits-all recommendation into something more targeted.
Why the results carry weight
Several design features set this analysis apart from the typical supplement study. The epigenetic investigation was prespecified, meaning the researchers locked in their methods and primary comparisons before unblinding the data. That discipline sharply reduces the risk of cherry-picking favorable results after the fact.
The parent COSMOS trial used a 2×2 factorial design, independently randomizing participants to multivitamin or placebo and, separately, to cocoa extract or placebo. This structure allowed investigators to isolate the multivitamin signal cleanly. The consistency of the effect across five independent clocks, each built from different methylation sites in different cohorts, further strengthens the case that the finding reflects genuine biology rather than a statistical fluke in one particular algorithm.
Publication in Nature Medicine, with peer review and an accompanying expert commentary, adds another layer of scrutiny. The commentary, notably, urged caution about over-interpreting the results, but did not challenge the study’s methodology or internal validity.
It is also worth noting that COSMOS has already produced a related finding in a different domain. A separate ancillary study, COSMOS-Mind, found that the same daily multivitamin modestly slowed cognitive decline in older adults over three years. That result, published in Alzheimer’s & Dementia, doesn’t prove the epigenetic and cognitive findings share a mechanism, but it does suggest the supplement is producing detectable biological effects across more than one measurement system.
The gap between clocks and clinical outcomes
The most important caveat is one the researchers themselves emphasize: epigenetic clocks remain surrogate markers. They correlate with disease and death in large population studies, but no randomized trial has yet demonstrated that deliberately slowing a clock directly prevents illness, preserves physical function, or extends life.
COSMOS itself illustrates the disconnect. The parent trial’s primary outcomes paper found that daily multivitamins did not significantly reduce cancer or major cardiovascular events over the main follow-up period. (Some secondary cancer analyses hinted at a modest signal, but the primary result was null.) Slower ticking on an epigenetic clock over two years did not translate into fewer heart attacks, strokes, or cancer diagnoses within the timeframe studied.
That gap does not invalidate the epigenetic findings, but it does limit how far anyone should run with them. A favorable shift in a molecular marker is not the same thing as a proven health benefit. It is a reason to keep investigating, not a reason to declare victory.
Open questions and limitations
The two-year follow-up leaves durability unresolved. Would the roughly four-month biological aging advantage persist, widen, or plateau over five or ten years of continued use? Aging unfolds over decades, and a short-term molecular shift might represent lasting recalibration or a transient blip that fades.
The 958-person methylation subset, while large for a mechanistic study, is still a fraction of the full COSMOS cohort. Participants were generally healthy, predominantly white older adults who volunteered for a prevention trial, a group that tends to be more health-conscious and better resourced than the broader population. Whether similar effects would appear in younger adults, people managing multiple chronic conditions, or communities with different dietary patterns and healthcare access is unknown.
Mechanistically, the study cannot identify which ingredients in the Centrum Silver formulation are doing the work. The pill contains more than 20 vitamins and minerals. The epigenetic analysis does not distinguish whether the benefit comes from correcting subclinical deficiencies in B vitamins, zinc, or vitamin D, from antioxidant effects, from support for DNA repair enzymes, or from some combination. Baseline nutrient status and dietary habits, which varied across participants, likely modulate the response but were not the focus of this report.
Finally, the findings apply to one specific product tested against placebo in one demographic. They do not compare brands, doses, or targeted nutrient stacks, and they should not be extrapolated to high-dose supplements, which carry their own risk profiles.
What this means for supplement decisions right now
For clinicians and patients weighing whether a daily multivitamin is worth taking, the COSMOS epigenetic data add a new, cautiously positive data point to a mixed evidence base. The supplement appears safe for most older adults, is inexpensive, and now has randomized evidence linking it to slower molecular aging, particularly in people whose biology suggests they are aging faster than average.
But the data do not support treating multivitamins as proven anti-aging therapy. They do not replace exercise, blood pressure management, smoking cessation, or a diet built around whole foods. And they leave unanswered the question that matters most to any individual considering the pill: will this help me live longer or stay healthier?
What the COSMOS team has shown is that a cheap, widely available supplement can measurably shift molecular markers of aging in a rigorous trial setting. Whether that shift translates into years of better health is the next question, and answering it will require longer follow-up, larger methylation cohorts, and studies designed to connect clock changes to the outcomes people actually care about. Until that evidence arrives, the multivitamin occupies an unusual position in medicine: a low-risk intervention with a real biological signal and an incomplete story.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
