Most nutrition claims rest on observational data: watch a large group of people, record what they eat, see who gets sick. It’s useful, and it’s weak. The gold standard — randomise people to an intervention or a placebo and follow them for years — is expensive enough that it’s rarely done for a food compound.
COSMOS is the exception. The COcoa Supplement and Multivitamin Outcomes Study randomised 21,442 older adults and followed them for a median of 3.6 years. It is the largest randomised controlled trial of cocoa flavanols ever conducted, and it’s the reason we build products around real high-flavanol cacao rather than a flavouring.
It also did not cleanly succeed. That part matters, and we’re going to walk through it honestly.
How the trial was built
COSMOS used a two-by-two factorial design: participants received cocoa extract, a multivitamin, both, or neither. The cocoa extract delivered 500 mg of cocoa flavanols per day, including roughly 80 mg of epicatechin — a dose chosen to sit meaningfully above typical dietary intake.
The cohort was 12,666 women aged 65 and older and 8,776 men aged 60 and older. The primary endpoint was total cardiovascular events: a composite of heart attack, stroke, coronary revascularisation, cardiovascular death, and several related outcomes.
What it found
Over the follow-up period, 410 participants taking cocoa extract had a confirmed cardiovascular event, compared with 456 on placebo. That’s a 10% relative reduction — which did not reach statistical significance. By the strictest reading of a clinical trial, COSMOS missed its primary endpoint.
The pre-specified secondary endpoints were more striking. Cardiovascular death fell by 27% in the cocoa extract group. Heart attacks were reduced by 13%.
How to read a mixed result honestly
There are two bad ways to interpret COSMOS and one good one.
The first bad reading is the supplement-marketing version: “27% reduction in cardiovascular death!” — quoting the most flattering secondary endpoint as though it were the headline result. Secondary endpoints in a trial that missed its primary are hypothesis-generating. They tell you where to look next. They do not establish efficacy.
The second bad reading is the dismissive version: “the trial failed, cocoa flavanols do nothing.” A 10% reduction that misses significance is not the same as no effect. COSMOS was powered for a larger effect than it observed, and 3.6 years is short for cardiovascular prevention in a population that was, on average, already fairly healthy. Adherence in long nutrition trials also tends to drift, which biases results toward the null.
The honest reading sits between them: COSMOS produced a consistent directional signal across multiple endpoints, with the cardiovascular-death finding strong enough to merit a dedicated confirmatory trial. The investigators said as much — that the suggestive benefits “would optimally be reproduced in subsequent trials.”
COSMOS randomised 21,442 older adults to 500 mg/day of cocoa flavanols or placebo for a median 3.6 years. Total cardiovascular events fell 10% (not statistically significant). Cardiovascular death fell 27% and heart attacks 13% as secondary endpoints. It is the strongest randomised evidence available for cocoa flavanols and cardiovascular health — and it is suggestive rather than conclusive.
What COSMOS does not tell you
A few boundaries worth drawing, because they get blurred constantly:
- It tested an extract, not chocolate. A standardised 500 mg flavanol capsule is not a candy bar. Most commercial chocolate delivers a small fraction of that dose, especially if it has been Dutch-processed.
- It studied older adults. Whether the same effect applies at 35 is unknown.
- It does not license disease claims. No food or supplement can be sold as treating or preventing heart disease on this evidence, and we don’t claim otherwise.
What it does support is a modest, defensible position: cocoa flavanols are among the better-studied dietary compounds in cardiovascular research, the mechanistic story is coherent, and getting a meaningful daily dose is a reasonable thing to do.
Getting a real dose
The practical difficulty is that flavanol content is invisible on most packaging, and standard processing destroys most of it. A heavily alkalised cocoa can be 100% cocoa and still be a poor flavanol source — which is why we source high-flavanol cacao and skip alkalisation entirely.
If you want the mechanism behind these numbers, read how cocoa flavanols work in the body. If you want the practical dosing question, how much dark chocolate per day covers it.
Coffee + Cacao — 12 oz
Specialty coffee blended with 20% high-flavanol, non-alkalised cacao — a daily flavanol habit that fits the routine you already have.
Frequently asked questions
What was the COSMOS trial?
COSMOS — the COcoa Supplement and Multivitamin Outcomes Study — was a randomised, double-blind, placebo-controlled trial of 21,442 older adults testing 500 mg/day of cocoa flavanols and a multivitamin against placebo over a median 3.6 years. It is the largest randomised trial of cocoa flavanols conducted to date.
Did the COSMOS trial prove cocoa flavanols prevent heart disease?
No. The primary endpoint — total cardiovascular events — showed a 10% reduction that did not reach statistical significance. Pre-specified secondary endpoints were stronger, including a 27% reduction in cardiovascular death, but secondary endpoints from a trial that missed its primary are considered hypothesis-generating rather than confirmatory.
How much cocoa flavanol did COSMOS use?
500 mg of cocoa flavanols per day, including approximately 80 mg of epicatechin, delivered as a standardised extract in capsule form rather than as chocolate.
Can I get the COSMOS dose from eating chocolate?
It is difficult. Commercial chocolate varies enormously in flavanol content, and Dutch-process alkalisation removes most of it. Reaching 500 mg reliably from ordinary chocolate would mean a large daily sugar and calorie load.
Is more cocoa flavanol always better?
Not necessarily. Dose-response work on endothelial function suggests benefits plateau, with one meta-analysis identifying an optimal effect around 710 mg total flavanols. Higher intake has not been shown to produce proportionally greater benefit.
References
These statements have not been evaluated by the Food and Drug Administration. This product is not intended to diagnose, treat, cure, or prevent any disease. This article is for educational purposes and is not medical advice. Talk to your doctor before changing your diet, caffeine intake, or any medication — particularly if you are pregnant, nursing, managing blood pressure or a heart condition, or taking prescription drugs.