Multivitamins do not extend life. The one thing they may do is smaller than the pitch.

A daily multivitamin does not reduce the risk of dying. In three large US cohorts covering 390,124 adults with no history of cancer or chronic disease at baseline and up to 27 years of follow-up, during which 164,762 people died, multivitamin use was not associated with lower all-cause mortality in either the first half (hazard ratio 1.04, 95% CI 1.02 to 1.07) or the second half (1.04, 0.99 to 1.08) of follow-up [s2]. The US Preventive Services Task Force reached a complementary verdict from the trial literature in 2022: the evidence is insufficient to determine whether multivitamins help or harm in preventing cardiovascular disease or cancer, an I statement [s1]. But the flat dismissal that usually follows — that multivitamins do nothing at all — has itself been overtaken by randomised evidence on one outcome.

What the task force actually said, and about what

The USPSTF distinguished sharply between three things. On beta carotene, it concluded with moderate certainty that the harms of supplementation outweigh the benefits for preventing cardiovascular disease or cancer, and recommended against it — a D recommendation [s1]. On vitamin E, it concluded with moderate certainty that there is no net benefit [s1]. On multivitamins, and on single or paired nutrients other than those two, it concluded that evidence is lacking and the balance of benefits and harms cannot be determined [s1].

"Insufficient evidence" is not the same finding as "no effect", and the distinction is doing real work here. The task force also recorded the scale of the behaviour it was assessing: in NHANES data, 52% of surveyed US adults reported using at least one dietary supplement in the prior 30 days and 31% a multivitamin-mineral supplement, most commonly for overall health and wellness and to fill dietary nutrient gaps [s1].

Why the cohort study was run, and what it fixed

The 2024 cohort analysis was designed to address the two objections that sink most observational supplement research: healthy-user confounding, and reverse causation when people in poor health start taking vitamins [s2]. It pooled the NIH-AARP Diet and Health Study (327,732 participants), the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial (42,732) and the Agricultural Health Study (19,660), with baseline multivitamin use assessed between 1993 and 2001 and follow-up use between 1998 and 2004 [s2].

The healthy-user pattern was plainly visible in the data: among daily users, 49.3% were female and 42.0% college-educated, against 39.3% and 37.9% among non-users, while 11.0% of daily users were current smokers versus 13.0% of non-users [s2]. After multivariable adjustment, and in time-varying analyses, no mortality benefit appeared, and hazard ratios were similar for the major causes of death [s2].

The one signal that survives randomisation

COSMOS was a two-by-two factorial randomised trial of cocoa extract and a daily multivitamin-mineral supplement for the prevention of cardiovascular disease and cancer in 21,442 US adults aged 60 and over [s3]. Its cognitive substudies produced the most-cited positive result in this literature.

COSMOS-Mind, a three-year pragmatic trial conducted by mail and telephone in 2,262 participants (mean age 73, 60% women), found no effect of cocoa extract on global cognition (mean z = 0.03, 95% CI −0.02 to 0.08, P = .28), but a statistically significant benefit of the multivitamin (mean z = 0.07, 0.02 to 0.12, P = .007), with benefits also seen for memory and executive function [s4].

A later analysis in the in-person clinic subcohort — 573 participants given detailed neuropsychological testing rather than telephone assessment — found a modest benefit on global cognition over two years that did not reach significance (0.06 SD units, −0.003 to 0.13), and a significant favourable change in episodic memory (0.12, 0.002 to 0.23), with nothing in executive function or attention (0.04, −0.04 to 0.11) [s3]. Meta-analysing the three non-overlapping COSMOS cognitive substudies — COSMOS-Clinic (573), COSMOS-Mind (2,158) and COSMOS-Web (2,472) — produced 0.07 SD units for global cognition (0.03 to 0.11, P = .0009) and 0.06 for episodic memory (0.03 to 0.10, P = .0007) [s3]. The authors describe the global-cognition effect as equivalent to reducing cognitive ageing by two years [s3].

How much weight that finding will bear

Rather less than the two-years-of-cognitive-ageing framing suggests. The effect sizes are small in absolute terms — 0.07 standard deviation units on a composite score — and the confidence interval around the clinic subcohort's global cognition result crosses zero [s3]. All three substudies come from a single parent trial with a single supplement formulation in adults aged 60 and over, mostly non-Hispanic White; COSMOS-Mind's own authors called for confirmation in a more diverse cohort and for work to identify a mechanism [s4]. Nothing in this evidence bears on younger adults, and none of it measured dementia incidence.

Cognition is also the exception rather than the pattern. In a New England subcohort of 603 COSMOS participants (mean age 69.7, 49.3% women) who underwent grip strength, walking speed, standing balance, chair-stand and timed-up-and-go testing at baseline and two years, the multivitamin, compared with its placebo, did not affect any physical performance measure [s5]. The cohort evidence found none on death [s2].

Where that leaves the bottle on the shelf

The strongest claims on both sides fail. There is no good evidence that a daily multivitamin extends life or prevents cardiovascular disease or cancer in a well-nourished adult [s1] [s2]. There is randomised evidence of a small favourable effect on memory and global cognition in older adults within one trial programme [s3] [s4]. And there is a specific, moderate-certainty finding of net harm for one common supplement ingredient, beta carotene, which the task force recommends against [s1]. None of that is an argument about correcting a diagnosed deficiency, which is a different clinical question entirely and one the USPSTF statement does not cover.

This article is informational and is not medical advice. Supplement decisions, including any involving specific nutrient doses or a diagnosed deficiency, belong with a clinician.

Sources

• Vitamin, Mineral, and Multivitamin Supplementation to Prevent Cardiovascular Disease and Cancer: USPSTF Recommendation Statement — JAMA, 2022-06-21
• Multivitamin Use and Mortality Risk in 3 Prospective US Cohorts — JAMA Network Open, 2024-06-26
• Effect of multivitamin-mineral supplementation versus placebo on cognitive function: COSMOS clinic subcohort and meta-analysis — The American Journal of Clinical Nutrition, 2024-01-18
• Effects of cocoa extract and a multivitamin on cognitive function: A randomized clinical trial — Alzheimer's & Dementia, 2022-09-14
• Effects of cocoa extract supplementation on physical performance measures: results from the randomized controlled COSMOS study — JBMR Plus, 2026-03-18