Yeung · The American journal of clinical nutrition 2023 · randomized controlled trial · n=3562

Multivitamin Supplementation Improves Memory in Older Adults: A Randomized Clinical Trial.

Level 2 - randomized trial

Individual randomized controlled trial

PubMed 37244291 · doi:10.1016/j.ajcnut.2023.05.011 · record verified 2026-08-26

What was done

In the COSMOS-Web ancillary randomized controlled trial (NCT04582617), 3,562 older adults were randomly assigned to receive either a daily multivitamin/multimineral supplement (Centrum Silver) or a placebo. Cognitive performance was evaluated annually over 3 years using an internet-based battery of neuropsychological tests. The prespecified primary outcome was change in episodic memory (immediate recall on the ModRey test) after 1 year. Secondary outcomes included episodic memory changes over 3 years, as well as novel object recognition and executive function over 3 years.

What was found

Participants assigned to the daily multivitamin had significantly better ModRey immediate recall at 1 year compared to placebo (t(5889) = 2.25, P = 0.025) and on average across the 3 years of follow-up (t(5889) = 2.54, P = 0.011). Based on cross-sectional age modeling, the authors estimated this improvement was equivalent to offsetting 3.1 years of age-related memory decline. Multivitamin supplementation showed no significant effects on secondary outcomes, including novel object recognition and executive function.

Why it matters

This large trial suggests that an accessible, low-cost daily multivitamin may help support episodic memory performance in older adults, though benefits did not generalize to other cognitive domains.

Limits

Cognitive testing was entirely internet-based and self-administered rather than conducted in person by trained psychometricians. The benefit was restricted to immediate episodic recall and did not extend to executive function or object recognition tasks. Specific baseline nutritional status, diet quality, and long-term clinical outcomes (such as incident mild cognitive impairment or dementia) were not reported in the abstract.