Brickman · Proceedings of the National Academy of Sciences of the United States of America 2023 · randomized controlled trial · n=3562

Dietary flavanols restore hippocampal-dependent memory in older adults with lower diet quality and lower habitual flavanol consumption.

Cited 55 times in the scientific literature.

Level 2 - randomized trial

Individual randomized controlled trial

PubMed 37252983 · doi:10.1073/pnas.2216932120 · record verified 2026-08-26

What was done

In the COSMOS-Web randomized controlled trial (NCT04582617), 3,562 older adults were assigned to receive either 500 mg per day of cocoa flavanols (cocoa extract) or placebo for 3 years. Habitual diet quality was assessed using the alternative Healthy Eating Index in all participants, and a urine biomarker of flavanol intake was measured in a subset of 1,361 participants to evaluate associations and intervention effects on hippocampal-dependent memory.

What was found

The prespecified primary endpoint of memory improvement in the overall cohort after 1 year was not statistically significant. Baseline flavanol consumption and diet quality positively correlated with hippocampal-dependent memory. Flavanol supplementation improved memory specifically among participants in the lower tertiles of habitual diet quality or flavanol intake. Increases in the flavanol biomarker over the trial period correlated with memory improvement. The abstract did not report exact numerical effect sizes, test statistics, or confidence intervals.

Why it matters

The findings suggest dietary flavanols operate under a depletion-repletion model, where supplementation may reverse memory decline specifically in older adults with insufficient baseline intake rather than acting as a universal cognitive enhancer.

Limits

The trial failed its prespecified primary endpoint in the overall population, making positive findings dependent on subgroup analyses. Biomarker validation was restricted to a subset of participants (n = 1,361), and the abstract provides no exact numerical values, effect sizes, or p-values.

Cited by