Cocoa Flavanol Research: 8 Surprising Findings From Chocolate Science (2026)

By BT Staff | October 27, 2026

Does chocolate’s cocoa content mean it improves heart or brain health? Not necessarily. Cocoa flavanol research has found modest effects on certain vascular measurements, but a major randomized trial did not establish a statistically significant reduction in its primary composite cardiovascular outcome. In the 2022 COSMOS trial, Howard Sesso, JoAnn Manson, and colleagues studied 21,442 older adults and reported a hazard ratio of 0.90 (95% confidence interval 0.78–1.02) for total cardiovascular events with cocoa extract versus placebo. The result concerns a standardized research extract, not ordinary chocolate bars.

National Chocolate Day offers a useful reason to separate enjoyable food culture from measurable scientific claims. This supporting article explains flavanols, cocoa processing, clinical evidence, and frequent misconceptions. It is intentionally distinct from the day’s bodytechpharma.ca flagship article on occupational therapy and recovery. The focus here is consumer-science literacy, not product promotion or dietary or medical instructions.

Table of contents

1. Eight scientific findings
2. What flavanols actually are
3. Why chocolate is not a standardized extract
4. Cardiovascular trial evidence
5. Blood pressure and vessel function
6. Cognition and diet quality
7. How to interpret headlines
8. Research limitations
9. Frequently asked questions
10. References

Eight findings from cocoa flavanol research

Flavanols are a class of plant polyphenols present in cocoa and other foods. Researchers study defined compounds and mixtures, while chocolate products differ considerably in composition. These eight findings are best understood as an evidence map, not an endorsement of any product.

Finding	Supporting evidence	Important boundary
1. Flavanols are measurable compounds	Food chemistry and analytical studies	Cocoa percentage is not a precise flavanol measurement.
2. Processing changes the mixture	Cocoa processing research	Two dark chocolates can have different flavanol profiles.
3. Vascular effects have been studied	Short-term trials and scientific assessments	A vessel-function finding is not a clinical-event guarantee.
4. Blood pressure effects are small on average	Cochrane review, 35 trials	Short trials do not establish lifelong prevention.
5. COSMOS tested a standardized extract	Randomized trial, 21,442 adults	Extract results cannot simply be assigned to chocolate bars.
6. The primary cardiovascular result was not significant	COSMOS primary analysis	Positive secondary signals require context.
7. Cognitive results differ by question and subgroup	COSMOS-Web and clinic studies	Subgroup observations do not prove universal memory benefits.
8. Commercial labels cannot replace trial methods	Measurement and evidence appraisal	No ordinary percentage-cocoa label proves efficacy.

These findings show why cocoa flavanol research is more nuanced than the familiar contrast between “healthy dark chocolate” and “unhealthy candy.” Scientific confidence depends on the compounds measured, trial design, comparator, and endpoint.

Honest cocoa flavanol research also distinguishes physiological outcomes from disease-prevention claims.

Finding 1: flavanols are not the same as cocoa solids

Cocoa contains several chemical classes, including polyphenols, methylxanthines such as theobromine, fats, carbohydrates, and minerals. Researchers interested in flavanols often characterize compounds such as epicatechin and catechin as well as larger procyanidins. These categories describe chemistry, not identical biological effects.

A chocolate label that lists a high percentage of cocoa ingredients does not necessarily report how many milligrams of flavanols survived production. Analytical composition depends on the starting beans, fermentation, roasting, alkalization, extraction, and other processing stages.

This basic distinction shapes cocoa flavanol research: a precise intervention must be described using actual analyzed constituent amounts, not just marketing terminology or cocoa colour.

Chemical characterization is the starting point for reproducible cocoa flavanol research.

Finding 2: chocolate processing matters

The flavour and texture of chocolate result from processing decisions that can also alter polyphenol content. Fermentation and roasting affect the chemical profile of cocoa. Alkalization, sometimes called Dutch processing, can produce a particular colour or flavour but also changes measurable flavanol compounds.

Not every cocoa product undergoes the same steps. Because manufacturers use different bean mixtures and recipes, two products with similar dark-chocolate percentages can differ in both flavanol content and other nutritional features. This is why researchers may standardize extracts or characterize their test foods analytically.

Good cocoa flavanol research identifies the measured intervention. A conclusion about one specially formulated preparation cannot be attributed automatically to all brown-coloured cocoa products.

Processing differences make cocoa flavanol research comparisons more demanding than their headlines suggest.

Finding 3: blood-vessel function is a testable mechanism

The endothelium lines the inside of blood vessels and participates in vascular signalling. Researchers have investigated whether cocoa flavanols influence endothelium-dependent vasodilation, which describes a vessel’s ability to change diameter through endothelial signalling mechanisms.

In 2012, the European Food Safety Authority’s expert panel assessed evidence related to cocoa flavanols and maintenance of normal endothelium-dependent vasodilation. Its favourable evaluation concerned that specified physiological endpoint and particular characterized flavanols. It did not establish that ordinary chocolate prevents heart attacks or strokes.

Cocoa flavanol research is strongest when it distinguishes a mechanistic or physiological measurement from a long-term clinical outcome. More favourable vessel-function readings do not automatically equal fewer cardiovascular events.

That endpoint boundary is a fundamental cocoa flavanol research safeguard.

Finding 4: the average blood-pressure difference is modest

A Cochrane review published in 2017 evaluated 35 short-term trials, incorporating 40 treatment comparisons and 1,804 participants. The meta-analysis found an average systolic blood-pressure difference of approximately −1.76 mmHg relative to control; the diastolic estimate was also approximately −1.76 mmHg.

Those are statistically detectable group averages but small numerical changes. The trials generally lasted two to 18 weeks, so the evidence cannot by itself demonstrate decades-long vascular protection. The review’s authors also discussed heterogeneity and graded the certainty of the evidence as moderate.

For cocoa flavanol research, magnitude and duration are as important as statistical significance. A one- or two-unit change in a physiological measurement should not be advertised as a dramatic clinical transformation.

Short follow-up remains a recurring limitation in cocoa flavanol research.

Finding 5: the largest outcome trial used a controlled extract

The COcoa Supplement and Multivitamin Outcomes Study, abbreviated COSMOS, was a randomized, double-blind, placebo-controlled 2 × 2 factorial trial. It enrolled 21,442 older United States adults. The cocoa-extract arm used a standardized preparation containing 500 mg of cocoa flavanols, including 80 mg of epicatechin, under the study protocol.

Those quantities describe what investigators tested; they are not recommendations to consumers. Study materials and funding disclosures are essential to evaluating independence and conflicts of interest. The research supplement was supplied by an industry partner, which the published paper disclosed; that fact does not make the randomized outcome numbers “company-reported.”

The principal lesson for cocoa flavanol research is simple: a capsule-based extract trial answers a different question from whether a particular supermarket chocolate bar improves health.

Intervention specificity must accompany every cocoa flavanol research summary.

Finding 6: the primary cardiovascular endpoint was not significant

COSMOS’s prespecified primary cardiovascular outcome combined several events, including myocardial infarction, stroke, cardiovascular death, coronary revascularization, and other vascular events. In the cocoa-extract comparison, the reported primary-outcome hazard ratio was 0.90, with 95% CI 0.78–1.02.

Because the interval included 1.0, the investigators did not find a statistically significant effect on that primary total-cardiovascular endpoint. The study also reported a favourable secondary signal for cardiovascular death, but secondary endpoints deserve their own interpretation, including the possibility of chance findings across multiple outcomes.

Balanced cocoa flavanol research reports both points together. Highlighting a secondary positive finding while hiding the primary nonsignificant result would distort the trial’s main conclusion.

Prespecified outcomes deserve priority in cocoa flavanol research reporting.

Finding 7: cognition is a separate research question

Scientists have examined cognitive performance using tests of memory and other domains. The COSMOS-Web ancillary trial involved 3,562 older participants followed for three years, examining a standardized cocoa-flavanol intervention and web-based memory assessments.

The published analysis explored whether effects varied according to habitual diet quality and baseline flavanol exposure. Certain results suggested the possibility of a subgroup-specific benefit among participants with lower baseline dietary flavanol intake or poorer diet quality. Those findings should not be rewritten as a universal claim that chocolate improves memory.

Cocoa flavanol research on cognition requires careful attention to primary versus secondary outcomes, subgroup definitions, practice effects in repeat testing, and whether findings replicate independently.

Independent replication is especially valuable for cocoa flavanol research on complex memory outcomes.

Finding 8: ordinary chocolate remains a distinct food

Chocolate products contain varying proportions of cocoa solids, cocoa butter, sweeteners, milk components, and other ingredients. Their nutritional profiles differ from standardized research capsules and from one another. A product’s taste, bitterness, colour, or advertised cocoa percentage is not a validated proxy for the amount of each flavanol.

Because preparations vary, consumers cannot translate trial extract quantities into equivalent chocolate-bar servings with confidence. Doing so would require product-specific chemical analysis and consideration of all other food constituents. It would also confuse the study intervention with an untested food exposure.

This is why cocoa flavanol research should be read as compound-specific or preparation-specific evidence, not as a blanket endorsement of a food category.

Product variability is therefore part of cocoa flavanol research interpretation.

What cocoa flavanols are

Plant polyphenols are a broad group of naturally occurring compounds, and flavanols are one subgroup. Cocoa is studied partly because it contains flavanol monomers and oligomeric procyanidins. Scientists investigate their chemical identities, how they change during processing, and what metabolites appear after consumption under controlled study conditions.

“Flavonoid” and “flavanol” are not interchangeable terms. Flavonoids encompass several chemical subclasses, whereas flavanols refer to a narrower category. Research summaries that merge all polyphenols into one ingredient risk obscuring which compound mixture an experiment tested.

Cocoa flavanol research often focuses on (−)-epicatechin as a characterized marker, but whole preparations can contain a variety of related compounds. Their biological availability and metabolism can differ, and reporting one marker does not fully describe the entire mixture.

How researchers measure flavanols

Analytical laboratories may use chromatographic separation and appropriate detectors to identify or quantify selected components. The method must be suitable for its intended sample matrix and concentration range. A research paper should identify the relevant compounds and provide enough procedural detail for the measured composition to be understood.

Studies can report total cocoa-flavanol amounts, individual compounds, or metabolites. Different assays and reference standards can produce values that should not be compared naively. Readers need to know what “total” includes before treating two numerical labels as equivalent.

Rigorous cocoa flavanol research also needs documentation of sample stability and method limitations. Chemical composition can change, and an appealing number on a product label is not a substitute for the test definition.

Clear analytical definitions improve the comparability of cocoa flavanol research trials.

Why the term “antioxidant” is incomplete

Cocoa polyphenols can participate in reactions studied in laboratory antioxidant assays. However, an in-vitro antioxidant measurement is not proof that a food or supplement prevents human disease. Biological pathways are complex, involving digestion, metabolites, tissue exposure, and many interacting processes.

The language of “antioxidants neutralizing damage” may be memorable, but it can conceal the gulf between a test-tube assay and clinical outcomes. Investigators need human studies that measure meaningful endpoints before broad health claims can be justified.

Cocoa flavanol research is better summarized with specific vascular, blood-pressure, or cognitive outcomes than with a generic claim that an ingredient “fights oxidation.”

Why chocolate is not a standardized extract

In clinical experiments, researchers try to control the quantity and character of the test material. A standardized cocoa extract may be characterized for particular flavanol compounds and used with an appropriate placebo. Chocolate foods vary much more in sweetness, fat content, processing, and potential participant expectations.

This difference changes what the results mean. A double-blind capsule study can reduce some biases that are difficult to eliminate in food comparisons where products have distinct tastes or textures. Neither design is inherently worthless; they simply answer different questions.

Cocoa flavanol research should state whether an experiment involved extract capsules, cocoa drinks, powdered cocoa, confectionery, or specially prepared high-flavanol chocolate. The intervention type belongs next to every quoted numerical result.

The percentage-cocoa trap

An 85% dark-chocolate label commonly refers to ingredient composition under the relevant labelling system. It does not certify a standardized flavanol dose, a defined epicatechin concentration, or equivalence to any published trial intervention. Higher cocoa percentages may correlate with some constituents in some products, but the relationship is not fixed.

Likewise, a dark chocolate can differ from cocoa powder because of formulation and processing. Product-specific testing would be needed to evaluate actual flavanol content. Even then, composition alone would not establish a clinical effect for that product.

In cocoa flavanol research communication, this is one of the most important consumer-science distinctions. Chemical content and proven outcome are separate questions.

Calories and other constituents complicate comparisons

Chocolate is a food containing energy and a mixture of macronutrients. In an ordinary food trial, changes in energy intake or substitution of foods may influence some outcomes independently of the flavanols under investigation. Researchers try to design controls that help isolate the mechanism of interest.

A study of flavanols alone cannot justify ignoring the rest of a food’s composition. Conversely, a nutritional analysis of an ordinary bar does not resolve the specific mechanistic questions being addressed in a controlled extract experiment.

Strong cocoa flavanol research reports enough about the comparator and overall intervention to allow readers to identify which difference between groups was deliberately tested.

What the major cardiovascular trial found

The 2022 COSMOS publication by Sesso, Manson, and colleagues is central because it examined clinical outcomes rather than only short-term physiology. The study’s randomized design and large enrollment addressed a question that small mechanistic trials could not settle on their own.

Participants were older adults, and the trial’s entry criteria excluded certain recent major events. The factorial design also investigated a separate multivitamin intervention. Interpreting its cocoa-extract findings requires keeping the two randomized factors distinct and considering prespecified analyses.

COSMOS detail	Published information	Meaning
Participants	21,442	Large study; not representative of every age or risk group
Study design	Double-blind, randomized, placebo-controlled, factorial	Designed to compare assigned interventions
Follow-up	Median approximately 3.6 years	Longer than most cocoa blood-pressure trials
Primary cardiovascular outcome	HR 0.90 (95% CI 0.78–1.02)	Did not meet statistical significance
Confirmed total cardiovascular events	410/10,719 assigned extract versus 456/10,723 placebo	Absolute event counts provide context
Secondary cardiovascular-death finding	Reported 27% relative reduction	Secondary result; not a substitute for primary endpoint

The sample size and primary result make this essential cocoa flavanol research. Yet even a large trial cannot establish all possible benefits or risks. The result pertains to the investigated formulation, participants, follow-up, and endpoint definitions.

Why a secondary outcome can look more encouraging

Trials often measure several events and subgroups. Some comparisons may meet conventional significance thresholds even when the primary endpoint does not. These secondary findings can generate hypotheses, but multiple testing and the original analysis plan influence how confidently they should be interpreted.

COSMOS found an encouraging cardiovascular-death signal while its prespecified total-cardiovascular primary outcome remained nonsignificant. The authors described possible benefit while calling for further research. Both statements belong in an honest summary.

Cocoa flavanol research becomes misleading when secondary results are converted into blanket statements such as “proven to prevent heart disease.” A smaller event category and a non-significant primary outcome demand calibrated language.

That calibration is essential to trustworthy cocoa flavanol research communication.

Funding and manufacturer involvement

Industry involvement is relevant to study appraisal. The published COSMOS report identified the supplier of the cocoa-extract intervention and disclosed funding and conflicts. These are study disclosures, not reasons to erase the findings or to treat independently reported outcome counts as marketing numbers.

If a company advertises a benefit using its own unpublished analysis, that claim should be clearly labelled company-reported and evaluated separately. Published peer-reviewed trial findings should be attributed to the investigators and journal, with financing disclosures retained.

For cocoa flavanol research, transparency is more helpful than either automatic dismissal or uncritical endorsement.

Blood pressure and vessel function

Blood pressure is frequently used in shorter nutrition studies because it can be measured repeatedly without waiting for rare clinical events. Trials can compare changes over several weeks, but the results are not equivalent to demonstrating differences in heart attack or stroke rates.

The 2017 Cochrane review, by Karin Ried and colleagues, synthesized 35 trials with 1,804 participants and 40 comparisons. It reported a mean systolic difference of −1.76 mmHg (95% CI −3.09 to −0.43) and a mean diastolic difference of −1.76 mmHg (95% CI −2.57 to −0.94).

Those figures represent pooled group-level changes under varied study conditions. They are not individual forecasts and should not be applied to a random chocolate bar. Cocoa flavanol research must distinguish the statistical result from its practical magnitude and the uncertainty introduced by different preparations.

Why study heterogeneity matters

The review combined trials using different flavanol amounts, cocoa formats, durations, and comparison products. Participants also differed in baseline blood pressure and other characteristics. Such diversity can help test whether effects appear across settings, but it complicates a single average estimate.

The authors conducted subgroup and sensitivity analyses and cautioned about unexplained heterogeneity. They also considered the potential influence of investigators employed by industry sponsors. Readers should not present one pooled number as if all studies produced precisely the same effect.

Good cocoa flavanol research distinguishes reproducible average signals from uncertainty about the most influential compounds, populations, and duration of exposure.

Heterogeneity should inform future cocoa flavanol research design rather than be concealed.

Regulatory assessment is endpoint-specific

The European expert assessment of cocoa flavanols and endothelium-dependent vasodilation focused on a particular physiological claim. Its scientific question concerned maintenance of vascular function under the evidence and characterization reviewed, not broad prevention of all cardiovascular disease.

Regulatory scientific assessments are specific to defined substances, evidence, and claims. An opinion concerning a characterized flavanol preparation should not be repurposed as a blanket authorization for unrelated products or sweeping treatment statements.

In cocoa flavanol research literacy, the crucial skill is identifying the measured endpoint and resisting the jump from mechanistic support to untested clinical conclusions.

Cognition and diet quality

Brain-health headlines have brought additional attention to cocoa flavanols. Some studies consider memory tasks, while others examine broader cognitive composites. Outcomes vary with participant characteristics, measurement instruments, follow-up time, and baseline nutritional status.

The 2023 COSMOS-Web publication, involving 3,562 older participants in a three-year randomized study, investigated memory and dietary-flavanol hypotheses. The researchers examined whether effects varied with habitual diet quality or flavanol intake, using repeated cognitive assessments and available biomarkers.

The intriguing subgroup findings are a reason for further investigation, not proof that all older adults gain memory benefits from chocolate. For cocoa flavanol research, precision about prespecified endpoints and secondary analyses is essential to explaining the findings fairly.

Why subgroup claims can be fragile

Suppose a study compares several dietary categories, cognitive tests, and time points. Some comparisons can look significant by chance alone, especially if researchers explore many possibilities after seeing the data. Preregistered hypotheses can strengthen interpretation, but replication remains valuable.

Baseline diet quality might modify response, or it might correlate with other unmeasured factors. Researchers can test such interactions using appropriate analysis, yet small subgroups may produce imprecise effect estimates. A positive subgroup does not mean the entire population benefited.

Responsible cocoa flavanol research identifies which question was primary, how groups were defined, and whether the overall finding was favourable, neutral, or uncertain.

Biomarkers are not memory outcomes

An objective biomarker can help assess exposure to flavanols or adherence under study conditions. That is useful because self-reported diet can be imperfect. However, a biomarker changing in the expected direction does not itself prove that memory, reasoning, or daily functioning improved.

Similarly, observing an association between biomarker changes and cognitive scores does not automatically demonstrate a causal pathway. The strongest claims require alignment among randomized comparisons, carefully measured outcomes, and reproducible findings.

Cocoa flavanol research should report exposure measures as supporting evidence while keeping meaningful human outcomes distinct.

How to read cocoa health headlines

The most common headline errors concern three substitutions: chocolate for extract, short-term biomarkers for clinical events, and secondary signals for established benefits. Each substitution expands a claim beyond what the original experiment measured.

A useful reading strategy begins with the full study title and methods. Identify what participants received, who was recruited, how long follow-up lasted, what the comparator was, and which outcome investigators designated as primary. Then examine the results and limitations.

The following questions can be used whenever cocoa flavanol research appears in a news article:

• Was the intervention ordinary chocolate, a drink, a specially prepared food, or a standardized extract?
• Did investigators measure actual flavanol compounds or rely on a general cocoa percentage?
• Was the design randomized, observational, or a small mechanistic experiment?
• What was the sample size and participant age range?
• Was the primary endpoint blood pressure, vessel function, memory, or a clinical event?
• Did the primary result meet statistical significance, and what did the confidence interval show?
• Were subgroup and secondary findings accurately labelled?
• Were industry funding, intervention supply, and potential conflicts disclosed?
• Do other independent studies support the same conclusion?

These questions help consumers enjoy chocolate culture while reading the research critically. Neither enthusiasm nor scepticism needs to replace measured interpretation.

Why “dark chocolate” is not a scientific outcome

The colour or percentage-cocoa category of a food is not itself a clinical endpoint. Scientific studies may use a standardized dark-chocolate preparation, but that does not authorize statements about all products carrying a similar description.

Reporting a precise flavanol amount without characterizing the intervention would also be incomplete. Investigators need confidence that the material remained consistent through the study and that the placebo or comparator was appropriately selected.

Cocoa flavanol research becomes useful to the public when it states the actual test material and refuses to substitute vague food categories for experimentally defined interventions.

What the research cannot establish

The current evidence does not establish that any ordinary chocolate bar prevents cardiovascular disease, reverses cognitive decline, or substitutes for established healthcare. Short trials of blood-pressure changes are not equivalent to long-term event prevention. Large randomized trials can be informative while still yielding nonsignificant primary outcomes.

It is also inappropriate to take extract quantities reported in scientific papers and convert them into personal consumption instructions. Product formulations differ, and the trial context cannot be reduced to a food-shopping rule. Research results should be interpreted as study findings rather than individualized plans.

Balanced cocoa flavanol research also recognizes that negative or mixed findings are valuable. They constrain claims, improve future study design, and prevent plausible mechanisms from becoming premature promises.

What future trials should clarify

Future investigations can improve by using well-characterized flavanol preparations, transparent methods, meaningful clinical endpoints, and independent replication. Comparisons involving ordinary foods should document their composition and account for other nutritional differences between groups.

Larger studies can examine whether effects vary by baseline diet quality, vascular risk, age, or other prespecified characteristics. Researchers should report absolute event rates, adverse events, statistical uncertainty, and conflicts rather than publishing only positive subgroup narratives.

High-quality cocoa flavanol research will also benefit from clear distinctions among chemistry, physiology, cognitive outcomes, and disease events. Each contributes knowledge, but no one measurement answers every question.

Meaningful clinical endpoints remain central to future cocoa flavanol research.

The most useful conclusion for National Chocolate Day

Chocolate can be enjoyed as a food without assigning it unproven medicinal powers. Scientific curiosity about cocoa compounds is warranted because researchers have identified measurable biological effects and have run substantial clinical studies. At the same time, important prevention and cognition claims remain uncertain or conditional.

The mature conclusion from cocoa flavanol research is neither “miracle food” nor “all studies are meaningless.” It is that evidence must be tied to specific preparations, study populations, outcomes, and limitations. That is how interesting food science becomes trustworthy public knowledge.

Frequently asked questions

Does cocoa flavanol research prove that chocolate prevents heart disease?

No. In the 21,442-participant COSMOS randomized trial, cocoa extract did not significantly reduce the prespecified total-cardiovascular primary outcome (HR 0.90; 95% CI 0.78–1.02). Ordinary chocolate products were not the standardized intervention tested.

Is higher-percentage dark chocolate always higher in flavanols?

Not necessarily. Cocoa percentage is not a validated measure of flavanol composition. Bean characteristics, fermentation, roasting, alkalization, and formulation can change the measured flavanol profile.

What did the 2017 Cochrane blood-pressure review find?

The review included 35 trials and 1,804 participants. It reported small average systolic and diastolic blood-pressure differences of about −1.76 mmHg relative to controls in short-term comparisons. Those results do not establish long-term prevention of clinical events.

Did cocoa flavanols improve memory in every older adult studied?

No. COSMOS-Web examined memory outcomes among 3,562 older adults and explored variation by diet quality and flavanol status. Subgroup signals do not justify a universal claim about chocolate and cognitive improvement.

Why can’t a chocolate bar be equated with a cocoa-extract trial?

Trials can use characterized preparations with specific compounds and comparison conditions. Chocolate bars vary in flavanol content and other ingredients, so an extract finding cannot automatically be transferred to an untested confectionery product.

References

1. Sesso HD, Manson JE, Aragaki AK, et al.; COSMOS Research Group. “Effect of cocoa flavanol supplementation for the prevention of cardiovascular disease events: the COcoa Supplement and Multivitamin Outcomes Study (COSMOS) randomized clinical trial.” American Journal of Clinical Nutrition. 2022. 21,442 randomized; primary cardiovascular HR 0.90 (95% CI 0.78–1.02). DOI: 10.1093/ajcn/nqac055. Full text.
2. Ried K, Fakler P, Stocks NP. “Effect of cocoa on blood pressure.” Cochrane Database of Systematic Reviews. 2017; Issue 4. 35 trials; 1,804 participants; pooled systolic −1.76 mmHg (95% CI −3.09 to −0.43). Cochrane review.
3. Brickman AM, Yeung LK, Alschuler DM, et al. “Dietary flavanols restore hippocampal-dependent memory in older adults with lower diet quality and lower habitual flavanol consumption.” Proceedings of the National Academy of Sciences. 2023. 3,562 randomized participants in COSMOS-Web. DOI: 10.1073/pnas.2216932120. Full text.
4. European Food Safety Authority Panel on Dietetic Products, Nutrition and Allergies. “Scientific Opinion on the substantiation of a health claim related to cocoa flavanols and maintenance of normal endothelium-dependent vasodilation.” EFSA Journal. 2012;10(7):2809. Physiological endpoint assessment, not a clinical-event trial. Opinion.
5. COSMOS Research Group. “Design and baseline characteristics of participants in the COcoa Supplement and Multivitamin Outcomes Study (COSMOS).” 2022. Describes the 21,442-person factorial trial and its primary outcomes. PubMed PMID 35288332.
6. COSMOS investigators. “Effect of cocoa extract supplementation on cognitive function: results from the clinic subcohort of the COSMOS trial.” 2023. Additional cognitive outcomes and diet-quality considerations. PubMed PMID 38070683.

For laboratory research use only. Not for human consumption.