A multivitamin nudges two biological-age clocks, but has not proved longer life

A randomized trial analysis found modest changes in two DNA-methylation measures after two years of daily multivitamin use in older adults. Clock readings are not the same as years gained, and the result is not a prescription for everyone. Evidence checked 3 November 2026.

At a glance

An ancillary analysis of the large COSMOS randomized clinical trial examined blood-based epigenetic ageing measures in 958 older adults assigned to a daily multivitamin–multimineral supplement or placebo for two years. The supplement modestly slowed change in two of five ageing clocks, especially clocks designed around mortality risk. The study did not test whether participants lived longer, avoided disease, or felt younger. It is an update about biomarkers, not proof of an anti-ageing pill.

What is a biological-age clock?

Ageing is not a single dial inside the body. Different organs and systems change at different rates. Epigenetic clocks measure patterns of DNA methylation—chemical marks on DNA—in blood or other tissue and use statistical models to estimate age-related processes. Some clocks are trained to predict chronological age; others are designed to correlate with health risks or mortality.

A clock is a measuring instrument, not the ageing process itself. If a lab value shifts in a favorable direction, that can be a useful signal. It does not automatically mean a person will live longer. For that, researchers need validated links between the biomarker and outcomes that matter, plus follow-up long enough to observe those outcomes.

What did COSMOS find?

The analysis included 958 participants, average age around 70, from COSMOS, a placebo-controlled, double-blind, 2-by-2 factorial trial in U.S. older adults. Researchers measured five blood-based epigenetic clocks at baseline and after two years. Daily multivitamin–multimineral supplementation was associated with a slower rate of change on two clocks, including measures that had been developed in relation to mortality. The other clocks did not show the same clear effect. Cocoa extract was also part of the parent factorial trial but was not the central positive finding in this analysis.

The study was published in Nature Medicine: Li et al. (2026). The trial registry describes the parent COSMOS design. These results are statistically interesting, but the effect was modest and based on a subset with usable blood samples. A biomarker endpoint is not a clinical endpoint.

What does the result not say?

It does not mean two years of vitamins add years to life. It does not show that everyone benefits, that megadoses work better, or that supplements replace food, activity, medical care, or correction of a documented deficiency. It also does not answer whether the result would hold in younger adults, people with different diets, or populations outside the U.S.

Clock studies can be affected by measurement noise, the choice of clock, and the characteristics of the people enrolled. A trial may be randomized while a subgroup biomarker analysis still has less power than the primary clinical outcomes. Researchers need replication and longer-term follow-up to determine whether a small clock shift predicts a real change in disease or function.

Why does this still matter?

The study adds a randomized-data point to a field often filled with dramatic claims from observational research or animal experiments. It suggests that a low-cost, widely used intervention may alter some measurable ageing signatures. That is a useful lead for science and a reason to ask what the markers mean biologically.

It also demonstrates the difference between “slower biomarker change” and “slower ageing” in everyday language. The former is what the study measured. The latter is a much broader claim. A well-chosen headline should preserve this distance rather than erase it.

For African readers, the question is not simply whether a supplement is available. Micronutrient needs vary, deficiencies are real, and access to diverse diets and healthcare varies widely. A U.S. study of older adults cannot establish benefit across African populations. Local nutrition research and clinical guidance matter; marketing claims should not substitute for them.

What next?

Researchers can test whether clock changes persist, whether they connect to functional outcomes, and whether effects vary by baseline nutrition or age. Independent groups should reproduce the result using different cohorts and measurement methods. Trials should prioritize outcomes people notice—mobility, cognition, disease, quality of life, and survival—rather than assuming a biomarker is a valid stand-in.

Ageing research often promises a fountain where the evidence currently offers a careful spring. This study brings a measured observation: a daily multivitamin shifted two of five molecular clocks over two years in a subset of older trial participants. That is worth investigating. It is not permission to sell immortality in a bottle.

Questions readers ask

Should everyone take a multivitamin to slow ageing?

This study does not establish that recommendation. Individual supplement decisions depend on diet, health, medications, and clinical guidance.

Did the supplement make people younger?

No. Researchers observed changes in selected DNA-methylation markers, not a reversal of chronological age or proven clinical ageing.

Does the cocoa arm matter?

COSMOS used a factorial design that also studied cocoa extract. This ageing-clock analysis primarily reported the multivitamin signal; read the paper for all treatment comparisons.

Sources

• COSMOS epigenetic-age analysis, Nature Medicine (2026)
• COSMOS trial record
• Nature news coverage of the result

Continue exploring

A new immune partner for stage III colon cancer. The brain may begin as two developmental neighborhoods.

Cover: Illustrative portrait of an older woman reading, by Jaroslav A. Polák, CC0 1.0. She is not a COSMOS trial participant. Credit: Jaroslav A. Polák. CC0 1.0. Original source.