Multivitamins: What the Large Randomized Trials Found

By The Kasivit Editorial Team · Library · October 2026

Multivitamins are among the most widely used supplements. In US survey data cited by the US Preventive Services Task Force (USPSTF), 52% of adults reported using at least one dietary supplement in the prior 30 days and 31% reported using a multivitamin-mineral supplement.^1. Multivitamins are also one of the few supplement types that have been tested in very large, long-running randomized trials. This article walks through what those trials reported, what they did not test, and how to read a multivitamin label.

General information, not medical advice.

Why look at randomized trials?

Observational studies compare people who already choose to take multivitamins with people who do not, and those groups often differ in diet, exercise, and access to care. The Physicians' Health Study II authors noted that such studies had shown inconsistent associations between multivitamin use and cardiovascular disease.² In a randomized, double-blind, placebo-controlled trial, chance decides who gets the multivitamin, and neither participants nor researchers know who is taking which. That is the most direct way to ask whether the supplement itself changes an outcome, although it still has limits.

The main trials at a glance

HR = hazard ratio. An HR of 1.0 means no difference; below 1.0 means a lower event rate with the multivitamin. If the 95% confidence interval (CI) crosses 1.0, the difference is not statistically significant.

Cancer, heart events and deaths

Physicians' Health Study II

The Physicians' Health Study II enrolled 14,641 male US physicians who were 50 or older at the start (mean age 64.3). They were randomly assigned to a daily multivitamin or placebo, starting in 1997 with follow-up through June 2011.³

Over a median of 11.2 years, the rate of total cancer was 17.0 events per 1,000 person-years in the multivitamin group and 18.3 in the placebo group (HR 0.92; 95% CI 0.86 to 0.998; P = .04). The authors reported this difference as modest but statistically significant. There was no significant difference for prostate cancer, colorectal cancer, other individual cancer sites, or cancer deaths (HR 0.88; 95% CI 0.77 to 1.01).³

For cardiovascular outcomes in the same trial, there was no significant difference in major cardiovascular events: 11.0 versus 10.8 events per 1,000 person-years (HR 1.01; 95% CI 0.91 to 1.10). Heart attack, stroke and cardiovascular deaths also did not differ significantly, and total mortality was not significantly different (HR 0.94; 95% CI 0.88 to 1.02).²

Note that the upper end of the cancer confidence interval sits just below 1.0, close to the line of no difference, and that every participant was a male physician, a group that may differ from the general population in diet and access to care.

COSMOS

The COcoa Supplement and Multivitamin Outcomes Study (COSMOS) was a randomized, double-blind, placebo-controlled trial of 21,442 US adults: 12,666 women aged 65 or older and 8,776 men aged 60 or older, free of major cardiovascular disease and recently diagnosed cancer. It used a 2-by-2 design that tested a daily multivitamin-mineral and a cocoa extract in the same trial. The intervention phase ran from June 2015 to December 2020.⁴

Over a median of 3.6 years, invasive cancer occurred in 518 participants in the multivitamin group and 535 in the placebo group (HR 0.97; 95% CI 0.86 to 1.09; P = 0.57). There was no significant difference for breast or colorectal cancer. Composite cardiovascular events occurred in 429 and 437 participants, respectively (HR 0.98; 95% CI 0.86 to 1.12), and all-cause mortality was not significantly different (HR 0.93; 95% CI 0.81 to 1.08). The authors reported no safety concerns.⁴

Among the secondary outcomes, the multivitamin group had a lower rate of lung cancer (HR 0.62; 95% CI 0.42 to 0.92).⁴ Secondary results are read with more caution, because many outcomes were examined and the trial was designed around total cancer. The authors concluded that a daily multivitamin, compared with placebo, did not significantly reduce total cancer among older men and women.

What the evidence reviews concluded

A 2013 review for the USPSTF, funded by the Agency for Healthcare Research and Quality, pooled two large trials (27,658 people) and reported lower cancer incidence in men taking a multivitamin for more than 10 years (pooled relative risk 0.93; 95% CI 0.87 to 0.99). The trial that included women showed no effect in women, and the reviewers described the result as a small, borderline benefit in men only.⁹

The 2022 update included 84 studies. Multivitamin use was associated with a lower incidence of any cancer (odds ratio 0.93; 95% CI 0.87 to 0.99; 4 trials, 48,859 people), but the reviewers stated that this evidence had important limitations. Overall, they concluded that vitamin and mineral supplements were associated with little or no benefit for cancer, cardiovascular disease and death, apart from that small multivitamin cancer finding.¹⁰

The USPSTF's 2022 statement, which covers community-dwelling, nonpregnant adults, concluded that the evidence is insufficient to assess the balance of benefits and harms of multivitamins for cardiovascular disease or cancer (an "I statement"). It recommended against beta-carotene or vitamin E supplements for those purposes.¹

Memory and thinking

The cognition findings point in different directions, so each study is worth looking at separately.

PHS II cognition substudy

In 5,947 male physicians aged 65 or older within the Physicians' Health Study II, up to four telephone cognitive assessments over 12 years showed no difference in cognitive change between the multivitamin and placebo groups. For the global composite score, the mean difference was −0.01 standard units (95% CI −0.04 to 0.02). The authors noted that the vitamin doses may have been too low or the population too well-nourished to show a benefit. PubMed lists funding from the National Institutes of Health, BASF, Pfizer and DSM Nutritional Products.⁵

COSMOS-Mind

COSMOS-Mind enrolled 2,262 COSMOS participants (mean age 73; 60% women; 89% non-Hispanic White) and assessed cognition by telephone each year for 3 years. The multivitamin comparison was a pre-specified secondary endpoint; cocoa extract was the primary one. Compared with placebo, the multivitamin group had a statistically significant difference on a global cognition score (mean z-score 0.07; 95% CI 0.02 to 0.12; P = .007). The authors called for further work in a more diverse group of people.⁶ The paper states that the National Institutes of Health, Mars, and Pfizer did not contribute to the trial's design, conduct, analysis, or writing.

COSMOS-Web

COSMOS-Web included 3,562 older adults randomly assigned to a daily multivitamin or placebo, with annual internet-based tests for 3 years. On the main outcome, immediate recall on a word-list test after 1 year, the multivitamin group scored significantly better (P = 0.025), and also across the 3 years on average. There were no significant effects on the secondary tests of object recognition and executive function. Using a cross-sectional analysis of age and test scores, the authors estimated the difference as equivalent to 3.1 years of age-related memory change.⁷

COSMOS-Clinic and the pooled analysis

COSMOS-Clinic tested 573 participants in person with more detailed assessments over 2 years. The difference in global cognition was 0.06 standard units, and its confidence interval crossed zero (−0.003 to 0.13), so it was not statistically significant. Episodic memory differed by 0.12 standard units (95% CI 0.002 to 0.23).⁸

The same paper pooled three COSMOS substudies, counting only participants who did not overlap between them (573 from COSMOS-Clinic, 2,158 from COSMOS-Mind and 2,472 from COSMOS-Web). The pooled differences were 0.07 standard units for global cognition (95% CI 0.03 to 0.11) and 0.06 for episodic memory (95% CI 0.03 to 0.10). The authors estimated the global cognition difference as equivalent to 2 years of cognitive aging.⁸

How to read these numbers

Differences of 0.06 to 0.07 standard units are small.^{6, 8} The "years of aging" figures are the authors' statistical estimates, not direct measurements. These studies measured test scores, not dementia diagnoses or day-to-day function. All three COSMOS substudies come from the same parent trial, so pooling them is not the same as confirmation from separate trials, and the longer PHS II substudy found no difference.⁵

Who paid for these studies?

Funding does not decide whether a study is right, but it is worth knowing, especially when the tested product was supplied by its manufacturer. This is what the papers report.

• Physicians’ Health Study II. Supported by National Institutes of Health grants and an investigator-initiated grant from BASF Corporation. The study agents and packaging were provided by BASF and Pfizer, and the multivitamin tested was Centrum Silver. Packaging was also provided by DSM Nutritional Products. The papers state that NIH, BASF, Pfizer and DSM had no role in the study design, conduct, data collection, analysis or interpretation.^{2, 3, 5}
• COSMOS. Supported by National Institutes of Health grants and an investigator-initiated grant from Mars Edge, the nutrition research segment of Mars Inc., which supplied the cocoa extract tested alongside the multivitamin. The multivitamin tested was Centrum Silver, supplied in part by Pfizer Consumer Healthcare (now Haleon).^{4, 7} The COSMOS-Mind substudy was supported by an award from the National Institute on Aging, and its paper states that NIH, Mars and Pfizer did not contribute to the trial’s design, conduct, analysis or writing.⁶
• Author disclosures. The pooled COSMOS cognition paper lists research support to some authors from Mars Edge, Pure Encapsulations and Pfizer, and honoraria or travel from the Council for Responsible Nutrition, a dietary supplement industry trade association.⁸

In short, the manufacturers supplied the pills and packaging, and company grants supported the work, while the papers state the companies had no role in the design or analysis. Readers can weigh that for themselves. The reviews for the US Preventive Services Task Force, which looked at all the trials together, were a separate, publicly funded process.⁹

Limits of the research

• Who was studied. PHS II enrolled only male physicians.³ COSMOS enrolled US adults aged 60 or 65 and over,⁴ and the COSMOS-Mind sample was 89% non-Hispanic White.⁶ Results may not apply to younger adults, to people with diagnosed deficiencies, or to other populations.
• Which product. Each trial tested one specific commercial formula.⁷ A different multivitamin with different forms and amounts is not the product that was tested.
• Duration. COSMOS ran for a median of 3.6 years,⁴ which is short for outcomes such as cancer that develop over many years.
• Effect sizes. Where differences were statistically significant, they were small, and some confidence intervals came close to the line of no difference.
• Many outcomes. When many secondary outcomes are examined, some may reach significance by chance, so findings such as the COSMOS lung cancer result need confirmation.
• Funding. The tested multivitamin was supplied by its manufacturer in both trials, and company grants helped fund them; see “Who paid for these studies?” above.
• What was not tested. These trials did not test pregnancy, childhood, people with absorption problems, or how a multivitamin compares with improving the diet.

Food first

Food remains the main route to vitamins and minerals. A varied diet of vegetables, fruit, whole grains, legumes, nuts and protein foods also brings fiber and many other compounds that a capsule does not.

A multivitamin label tells you what is in the product, not whether you need it. Whether any nutrient is low for you depends on your diet, life stage, health conditions and medicines, and a clinician can help you look at that.

How to read a multivitamin label

The Supplement Facts panel is the most useful part of the package. The same checks apply to any product, including our Complete Multivitamin.

Serving size

Every amount is per serving, and a serving may be one capsule or several. If the serving is two capsules and you take one, you get half of each listed amount.

% Daily Value

The % Daily Value (%DV) compares the amount per serving with a reference amount used on labels. It is a comparison tool, not a personal target. Nutrients without a Daily Value are marked with a symbol instead.

Amounts above upper limits

A %DV above 100 is not the same as exceeding a safe limit. The relevant figure is the Tolerable Upper Intake Level (UL), the highest daily intake considered unlikely to cause harm for most people. ULs are not printed on the label, so check them separately and add up all sources: the multivitamin, single-nutrient supplements, and fortified foods and drinks. Vitamin A, vitamin D, iron, zinc and folic acid often appear in more than one product.

Forms

The label lists the form of each nutrient, for example cholecalciferol (vitamin D3), or retinyl acetate or beta-carotene for vitamin A. Beta-carotene is worth noticing: the 2022 evidence review associated it with a higher risk of lung cancer and of cardiovascular death, and the 2013 review noted the lung cancer finding in smokers.^9,10

Other ingredients

This line lists the capsule shell, fillers and flow agents. Check it for gelatin or plant-based shells, colorings and allergens. Third-party testing marks relate to quality checks on contents, not to health effects.

Safety and who should check first

• Count every source. Add up nutrients from all supplements and fortified foods and compare the total with the upper intake levels.
• Keep supplements away from children. Store all supplements out of reach and sight of children. In the US, products containing iron typically carry a label warning about accidental overdose in young children.
• Be aware of possible harms. The 2022 evidence review found limited evidence that some single supplements may be associated with serious harms, including hip fracture (vitamin A), hemorrhagic stroke (vitamin E) and kidney stones (vitamin C and calcium).¹⁰

The short version

In the two largest trials, a daily multivitamin made no significant difference to cardiovascular events or deaths. PHS II reported a small, borderline difference in total cancer in older men; COSMOS found none over a shorter period. The USPSTF judged the evidence insufficient for cancer or cardiovascular disease. COSMOS substudies reported small differences on some memory scores, while a longer PHS II substudy found none. Food comes first, and a label lists contents, not whether you need them.