This article is rated C-class on Wikipedia's content assessment scale. It is of interest to the following WikiProjects:

Chemicals Low‑importance This article is within the scope of WikiProject Chemicals, a daughter project of WikiProject Chemistry, which aims to improve Wikipedia's coverage of chemicals. To participate, help improve this article or visit the project page for details on the project.ChemicalsWikipedia:WikiProject ChemicalsTemplate:WikiProject Chemicalschemicals Low This article has been rated as Low-importance on the project's importance scale.

The list at the end incorrectly classifies Phylloflavan as Acetylated. 69.72.92.204 (talk) 07:50, 29 September 2014 (UTC)[reply]

I made this edit and trimming because both reviews (although thorough as Cochrane reviews are) presented little evidence of clear effect on cardiovascular risk factors from consuming flavan-3-ols. Example, the effect on blood pressure in both studies was 2 mmHg or less, i.e., although statistically significant, such a decrease is of little biological significance. WP:NOTJOURNAL #6-8: we are not writing a research paper discussing all possible findings from publications. Neither review is very convincing for the encyclopedia, and both conclude that the design of clinical trials to study catechins has been generally of low quality. Zefr (talk) 22:54, 27 May 2021 (UTC)[reply]

I think 2 mmHg are quite relevant, especially for a dietary intervention (it's about what can be achieved with a Mediterranean diet or moderate salt reduction) and in a population context, 2 mmHg can have quite an impact.^[1] The data on blood pressure are more convincing than on lipids Ggux (talk) 07:02, 28 May 2021 (UTC)[reply]

This review on the relationship of a Mediterranean diet and blood pressure shows the difficulty of interpreting high-quality dietary studies on blood pressure, the absence of convincing effects, and the coinvolvement of dietary factors that may affect blood pressure, such as flavonoids, omega fatty acids, and dietary fiber, among others. Isolating a specific role of flavan-3-ols in such small effects is even more uncertain. To make an encyclopedic case for any relationship leads into WP:SYNTH and overinterpretation. Zefr (talk) 14:49, 28 May 2021 (UTC)[reply]

In that case, it is a dietary patter, but flavan-3-ols are a specific class of compounds. When using a biomarker of intake - which is specific for a compound and not a food - the 2 mmHg remain.^[1] So I think it is not really that uncertain, given that many of the RCTs used compounds (or extracts) and not foods.Ggux (talk) 15:07, 28 May 2021 (UTC)[reply]

This is primary research - an observational study primarily from tea consumption - based on use of biomarkers for flavan-3-ol in single-samples of urine, and again with only a minor, vague association with blood pressure. There are many, potentially-conflicting dietary factors in this and similar studies attempting to relate food consumption with physiological effects. It does not meet WP:MEDASSESS. We can keep the 2019 systematic review currently used in the article, but the overall association of flavan-3-ol intake with potential for lowered risk of cardiovascular diseases is too weak for further content. Zefr (talk) 18:00, 28 May 2021 (UTC)[reply]

I'm sorry, but I tend to disagree. The study does specifically refer to flavan-3-ols and not foods - they did adjust for dietary sources and the biomarkers are fairly specific for those compounds. Evidence for a beneficial effect on blood pressure is considerably stronger than for blood lipids. I disagree with WP:MEDASSESS as it ignores e.g. Bradford Hill and is a bit outdated (see e.g. this EFSA article ^[1]) but I appreciate that there are conventions one should follow. I would however suggest to mention "blood lipids and vascular function". Would it be sensible to refer to the COSMOS-Study, which will provide more reliable results (with n=18,000 over 5 years)?

1. the COSMOS study has not been completed and published under peer-review, as best as I can tell from a PubMed search today. 2. The EFSA source you provided is a draft of a review, far from completion. 3. MEDASSESS exists under consensus from numerous senior medical editors, formed over years of discussion. 4. We need consensus among several editors here on this talk page, WP:CON, for more specific discussion of effects by flavan-3-ols. Recruiting other editors to the discussion would be in order, but I feel the article is adequately stated as of today. Zefr (talk) 22:07, 1 June 2021 (UTC)[reply]

Results for COSMOS have not been published yet - but it is the largest study of flavan-3-ols and thus the only one that can provide reasonable data. I think it would be useful to reference it for that reason. Most data suggest that flavan-3-ols affect vascular function, so I would find it odd not to mention this here but focus on lipids as data on lipids are considerably weaker. Ggux (talk) 18:37, 2 June 2021 (UTC)[reply]

COSMOS has now been published, showing a significant reduction in risk (per protocol analysis). It should therefore remain in the article. Ggux (talk) 16:08, 26 March 2022 (UTC)[reply]

1. 1: Vascular function is clearly stated as being affected in the cited references - and it is scientific consensus that flavanols improve vascular function (see Cochrane review & EFSA health claim). Effect on blood lipids is generally seen as secondary
2. 2: Primary research of n=20,000 is supporting vascular function statement - I don't see why this should be excluded. Ggux (talk) 21:35, 3 April 2022 (UTC)[reply]

The COSMOS study here had only 410 subjects taking the cocoa extract, i.e., a preliminary study for which the conclusion was "Cocoa extract supplementation did not significantly reduce total cardiovascular events", even though the overall risk of cardiovascular death was reduced. This is one early-stage study, not a WP:MEDSCI review, nor is it a recommendation or approval by any government agency to consume more cocoa extract. Zefr (talk) 21:49, 3 April 2022 (UTC)[reply]

Please read the study correctly: it had 21,442 participants. 410 participants of the intervention group had an event, 456 of the control group. This is not a preliminary study but a fairly large trial. Ggux (talk) 21:57, 3 April 2022 (UTC)[reply]

Because the primary outcome of the study was a possible effect of the cocoa extract on "cardiovascular events", it is only the group of 410 people that provide the results - no effect on the primary outcome. This was a supplement/dietary study for which there were many uncontrolled life and disease factors in the subjects over the 3.6 years of data collection. That not only defines "primary research", but also an inherent weakness of experimental design, which is too diffuse with complications, and disallows any clear conclusion, leading to the authors' conclusion: "Additional research is warranted to clarify whether cocoa extract may reduce clinical cardiovascular events." It is far too preliminary and inconclusive, and does not meet the source quality expected for an encyclopedia's medical content discussed in MEDRS. Zefr (talk) 23:55, 3 April 2022 (UTC)[reply]

I'm sorry, but this is wrong. According to this, the licensing study of BNT162b2 (https://www.nejm.org/doi/full/10.1056/nejmoa2034577) was based on 8 people as only 8 people (out of 40,000) in the intervention group got COVID? Ggux (talk) 05:45, 4 April 2022 (UTC)[reply]

There are two aspects of the article, Zefr and I cannot agree on:

• Scientific consensus suggests that the primary mode of action of flavan-3-ols is via improving vascular function - that is stated in reference 13. They should therefore be included in the text at least as prominent as the effect on blood lipids.
• Flavan-3-ols have been investigated in a large (n=20,000 subjects, 3.5 yr) clinical trial and shown to have a beneficial effect on cardiovascular-diseases. While this is primary research, it is a large study that provides considerably more evidence for a potential health effect than all other cited evidence as it uses hard endpoints. I therefore belief it has sufficient relevance to be at least mentioned.

Ggux (talk) 21:57, 3 April 2022 (UTC)[reply]

Both these statements above are incorrect: 1) there is no WP:MEDSCI scientific consensus that flavan-3-ols have an effect on vascular function - this remains under preliminary research; and 2) there was no evidence of beneficial effect on cardiovascular diseases, as stated in the abstract of the source. There is also a conflict of interest concern with this study, as it was financed by Mars, Incorporated, purveyor of the commercialized cocoa supplement, CocoaVia, and chocolate manufacturer, critiqued here by Dr. Marion Nestle, calling into question the remarkable main finding of the study: a 27% decrease in risk of death by taking a cocoa supplement! Zefr (talk) 00:26, 4 April 2022 (UTC)[reply]

This isn't really an RFC, but I would say that a single study is not scientific consensus of anything. I think this could benefit from more discussion, some proposed wording, and a third opinion from a WP:MEDRS familiar editor. ScottishFinnishRadish (talk) 23:46, 3 April 2022 (UTC)[reply]

• It is scientific consensus - stated in the article cited in the current version - that flavanols improve vascular function. There is a wealth of literature on this and for example the EFSA health claim is based on vascular function, not blood lipids. This review [[1]] clearly states the consensus.
• Marion Nestle does not criticise the outcomes of the study, but makes an "ad hominem" attack without any scientific base. There should be a scientific critique (which would be fine) but it appears that Zefr confuses the concept of clinical trials. The trial has about 20,000 participants - which is neither small nor preliminary. In other subjects - e.g. Vitamin D - results from large studies are also included. Ggux (talk) 05:49, 4 April 2022 (UTC)[reply]

Consensus might be too strong of a characterization, but Ggux has provided a review article suggesting evidence for improved cardiometabolic functions. As for the COSMOS study, it's a bit of a mixed bag of beans, but pros (large sample size, multiyear endpoints, db RCT) seem slightly more than cons (partial support from corporations, even significant CIs awefully close to 1 HR). By the way, what does "censoring follow-up at nonadherence" mean exactly? CurryCity (talk) 05:38, 5 April 2022 (UTC)[reply]

"censoring follow-up at nonadherence" - I hope it is appropriate to explain here:

When studies are analysed as "intention to treat", the comparison is simply between intervention and control group - whether the intervention group take their pills or not. This is useful to understand whether a treatment works, as a GP or physician will prescribe a treatment but has to accept that not all patients are compliant. If the treatment is unpleasant (large pills, side effects etc), it will impact compliance and that will impact the decision whether a treatment works or not.

The alternative is "per-protocol analysis" which includes only patients who followed the intervention. That is useful to investigate whether the actual treatment (drug, bioactive etc) works as it compares those who have completed the treatment with those who have not. In that case, study participants are excluded from the study once they stop taking the intervention - so their follow up stops. That's what is meant by "censoring f'up at nonadherence".

One can have very long discussions about whether this is appropriate or not, what it means and whether one should use it, but I assume this is beyond the scope of what should be included here. The relevance of COSMOS is that it is the first (and probably only) large study on a food compound (and multivitamins) as it's unlikely to be repeated soon. Ggux (talk) 08:31, 5 April 2022 (UTC)[reply]

Ok thank you for explaining. CurryCity (talk) 08:41, 5 April 2022 (UTC)[reply]

@Ggux, has anyone (i.e., in the scientific literature, not in the business news) written more about this study? WhatamIdoing (talk) 16:11, 5 April 2022 (UTC)[reply]

Not yet - it's a new study which was published only recently. But it complements the Raman et al review (https://academic.oup.com/ajcn/article/110/5/1067/5554778) which reviews much smaller studies on the effect of flavan-3-ols on CVD risk. I don't think the study in itself would be appropriate to include - but it adds to the totality of evidence and, as it is likely to be the largest study for some time, is appropriate to cite.

The article includes the EFSA health claim which is based on much weaker endpoints and fewer data. Ggux (talk) 16:37, 5 April 2022 (UTC)[reply]

For reasons (mostly having to do with the limited competence of the average Wikipedia editor), we tend to focus less on the scientific strength and more on the source's strength. It sounds like you are interested in mentioning the existence of this study in line with WP:MEDPRI ("If conclusions are worth mentioning (such as large randomized clinical trials with surprising results)..."). WhatamIdoing (talk) 18:00, 5 April 2022 (UTC)[reply]

@WhatamIdoing That's correct. Whether the results are surprising or not depends probably on prior bias of the reader, but they are definitely important as they are the first (and only) "proper" clinical study. So I believe they are relevant to be mentioned. Ggux (talk) 21:35, 5 April 2022 (UTC)[reply]

I don't object to mentioning the study. I wouldn't present its results as a fact, but something like "A study happened" could be okay. WhatamIdoing (talk) 16:23, 13 April 2022 (UTC)[reply]

Thanks - my suggestion is:

A 2017 Cochrane review showed that cocoa flavan-3-ol intake (average of 55 mg per day) over about 9 weeks had a small effect on lowering blood pressure (2 mmHg). Currently, only one large clinical trial, the COcoa Supplement and Multivitamin Outcomes Study (COSMOS), investigated the long-term effect of flavan-3-ols on health. The results of this study support a potential beneficial effect of flavan-3-ols on cardiovascular health. Ggux (talk) 09:34, 24 April 2022 (UTC)[reply]

I've recently been thinking more about our advice atWikipedia:Manual of Style/Medicine-related articles#Cite sources, don't describe them. In line with that, perhaps it should just say "Some clinical trials suggest that flavan-3-ol intake may have a small beneficial effect on cardiovascular health^[1][2]", without providing details but keeping the tentative nature of the information. WhatamIdoing (talk) 15:15, 24 April 2022 (UTC)[reply]

Perfect - so what do I need to do to add it? Thank you! Ggux (talk) 18:08, 24 April 2022 (UTC)[reply]

The two implied sources are not equivalent. One is a Cochrane review currently in the article to support weak evidence of a small effect barely worth mentioning: a 2 mmHg BP decrease. The authors spend considerable explanation on the overall poor-moderate quality and caveats about interpreting such weak clinical studies. The disputed source - the COCOA-COSMOS study - is primary research with too many design weaknesses to be treated as an encyclopedic source. As editors, we do have to assess evidence quality, WP:MEDASSESS, which disqualifies this research as a source. There is no conclusion possible from the COSMOS study, as nothing about cardiovascular function was established, and it remains too preliminary, vague, and without the WP:WEIGHT of scientific agreement or clinical practice to include. And its only "conclusion" is too far-fetched to be believed: a 27% decrease in the hazard ratio for death. If the COCOA-COSMOS study has any significance in the field of cardiovascular health, it will be included in a future Cochrane review. We should observe MEDASSESS and wait for that publication. Zefr (talk) 19:56, 24 April 2022 (UTC)[reply]

I think that the Cochrane source alone could be fairly summarized as "some clinical trials suggest a small beneficial effect".

Also, MEDASSESS does not tell us to care about how many words the authors spend explaining quality and caveats. We are supposed to assess them at the level of "Is this a review article or a clinical trial? Was this in humans or in mice?" – not "I'm some guy on the internet, and I have decided that this study, which somehow passed peer review, has too many design weaknesses to be able to draw these far-fetched conclusions". You might be absolutely right in this instance, but that is not what MEDASSESS recommends, or even permits. WhatamIdoing (talk) 20:34, 24 April 2022 (UTC)[reply]

I respectfully disagree with @Zefr for a number of reasons (I have to declare an interest here, as this is my research area and I have worked in this field for about 20 years - so I will probably give it more importance than it deserves):

• The Cochrane review talks exclusively about cocoa-flavanols, the Raman [2] review uses a wider definition - both show very clearly a potentially beneficial effect on vascular function. It is also consensus [3] that the main effect of flavan-3-ols is via vascular function and not blood lipids
• A 2 mmHg blood pressure reduction has a significant impact on a population scale and is comparable e.g. with reducing sodium intake [4]. It is obviously very difficult to model the impact of such a reduction on a population scale (and there are many uncertainties), but it is not as trivial as it appears [5]. It might be worth to consider that National Dietary Guideline committees think the evidence for flavan-3-ols to be sufficient to consider developing guidelines for it [[6]]
• I have to admit that I am not clear what the "design weaknesses" of COSMOS are. It was a blinded RCT following all current guidelines, is of a scale (21,000 participants) and duration (~3.6 years) way beyond anything included in the Cochrane review and supports those findings. It is obviously not sufficient as a stand alone, although I noticed that e.g. the article on the Mediterranean diet cites PREDIMED (which is also a large trial with considerably more limitations as it relied on dietary intake modification, which is much more difficult to achieve). Ggux (talk) 22:13, 24 April 2022 (UTC)[reply]

The only specific detail I remember Zefr mentioning was about the study being funded by a multinational source that sells chocolate candy bars. Funding is mentioned in the MEDASSESS section: Do not reject a higher-level source (e.g., a meta-analysis) in favor of a lower one (e.g., any primary source) because of personal objections to the inclusion criteria, references, funding sources, or conclusions in the higher-level source. Editors should not perform detailed academic peer review. In this case, the complained-about study is a primary source. Wikipedia editors generally take an agnostic view of funding, except when that funding has been called out as a problem in specific instances (usually a problem for pesticide safety research, not considered a problem for pharmaceutical drugs safety research). WhatamIdoing (talk) 23:13, 28 April 2022 (UTC)[reply]

Thanks @WhatamIdoing - so would the original idea of "Some clinical trials suggest that flavan-3-ol intake may have a small beneficial effect on cardiovascular health", citing the Cochrane review & COSMOS be acceptable?

There is consensus in the research area that the trial is quite important and that an improvement of vascular function is the main mode of action. Ggux (talk) 09:20, 1 May 2022 (UTC)[reply]

Further POV-pushing by Ggux in this and a subsequent edit today. The Cochrane and AJCN reviews are valid to include as satisfying WP:MEDASSESS, despite acknowledgement in both articles of the suboptimal quality of dietary studies reviewed. There is no consensus in the above three talk page sections to include primary research like the COSMOS study - see the left pyramid of MEDASSESS, identifying characteristics of primary research: Ggux is pushing it, one editor expressed neutrality ("a study happened"; "the complained-about study is a primary source"), and one editor opposes, i.e., no consensus. This edit and source discussing DRIs for flavan-3-ols has no encyclopedic value - it is 8 years old, no federal regulatory agency since the 2014 EFSA opinion has established such guidance, and there is no evidence of using such DRIs on food products. Zefr (talk) 16:29, 4 May 2022 (UTC)[reply]

I think there is no point in continuing this discussion. Zefr clearly has decided that including a large RCT violates Wikipedia Terms whereas I believe that such a study is clearly relevant considering the limitations of previous studies. I also do believe that there is scientific consensus about the potential benefits as the development of guidelines for flavan-3-ols is considered to be likely (based on a conference contribution, which I assume is not appropriate here, they will be published in autumn). I have therefore asked for a 3rd opinion.20:13, 4 May 2022 (UTC)[reply]

Response to third opinion request (Disagreement about inclusion of large RCT as primary source in text):

I am responding to a third opinion request for this page. I have made no previous edits on Flavan-3-ol and have no known association with the editors involved in this discussion. The third opinion process is informal and I have no special powers or authority apart from being a fresh pair of eyes.

I'll start by saying that this is an interesting discussion. I understand both very well and have some experience with medical literature and interpretation of it, and I have also done (but not completed) a course in nursing, just to list background as a formality and disclaimer, although I don't wish to add credentials to back up my statements or anything. My opinion given here is not an expert opinion. It is certainly true that the authors of the study given have reported (in my opinion) quite significant conflicts of interest. There are some things to be noted regarding the study itself, namely the following part: "Finally, our per-protocol analyses estimated the effects of cocoa extract if all participants were ≥75% compliant with taking their pills. The apparent observed 15% reduction in total cardiovascular events should be interpreted with caution as this secondary analysis is based on additional assumptions." This means that the observed reduction is influenced by assumptions rather than clinical findings, which per WP:MEDASSESS and pretty much any bio-medical scientific standard decreases the reliability (but not necessarily notability) of the end results. The authors acknowledge this too which is why they say caution should be taken when interpreting the findings as such. This also, however, decreases the extraordinaire of the resulting claim of 15% reduction in total cardiovascular events. It is hard for me to render a final opinion regarding whether this should be included, but for now, having read what is given at this time and knowing what I know, I would lean towards not including broad statements based on this one study. I would personally be okay with including it as a mention of purely encyclopedic value as WP:MEDASSESS does say we are not to take into account funding sources. I would strongly encourage the two disagreeing editors to come to terms with a particular wording for a brief mention in the article, giving proper weight to the Manual of Style and Medical Manual of Style. In my opinion it would be okay to add a disclaimer or warning per WP:MEDPRI2, these primary sources are generally not meant for public (non-expert) view, nor can the average Wikipedian be expected to be able to understand all of the jargon and implications given in the source. There are also potential concerns over interpretation of statistical significance and if we don't give this proper attention, this can get derailed very easily very quickly, as we all know that readers tend to love taking given sources for fact or derive conclusions from it that are not reliable. I do not know exactly what the best way to phrase it would be but we can think about this together, and I'm willing to pitch in with my thoughts as well! Let me know if there's anything else I can do or if you have any questions, I'll subscribe to this thread to be kept up-to-date. ★Ama TALK CONTRIBS 18:32, 5 May 2022 (UTC)[reply]

Useful to have your thoughts, Ama, with thanks for the thorough and objective discussion. I provided a compromise edit yesterday, here. The section is on "research", so the COSMOS study is included, but its nature as a primary source with some questionable results and interpretations (one example, 27% lower risk of death by consuming a branded cocoa supplement! - how will MARS market that!?) indicates we should use it only as an example of ongoing research, rather than describing it as "supportive" or consistent with findings from proper systematic reviews. Zefr (talk) 19:03, 5 May 2022 (UTC)[reply]

I don't think I would quite agree with your 'compromise' either though. Even if a study publishes highly contested or seemingly extraordinary claims (at first glance, through the optics of a non-professional, not to discredit you), it does not make it 'preliminary'. The study should be taken as it is reported (and purportedly verified), as a randomized controlled trial (RCT). Unless the scientific community at large deems the study to be discredited or influenced by corporations, which would be scientific misconduct, for which administrative actions or other penalties may be imposed, we should not discredit or devalue the type or significance of a study based on our personal findings. As such, I would be against calling it preliminary or any such term. The study ran for a little over three and a half years with tens of thousands of participants. That's by no means preliminary, even if the findings were to be wrong. Also, as I said before, the claim of 27% reduced risk of death (and other statistical claims) should be evaluated and interpreted with the authors' caution that these percentages were established making assumptions about compliance. They never specify just by how much the percentages were 'upped' to match assumed 'poor compliance'. And this is exactly one of the reasons why Wikipedia prefers secondary or tertiary medical sources over primary ones, as the average reader may not be able to distinguish these subtleties and nuances. Hope this helps. ★Ama ^{TALK CONTRIBS} 19:17, 5 May 2022 (UTC)[reply]

It is a preliminary (primary) study per MEDASSESS, pyramids left ("primary studies") and right ("unfiltered information"), with no qualifications needed for number of subjects or duration - still primary research, nowhere near achieving significant scientific agreement that would make a health claim eligibile for regulatory consideration. That is where Cochrane and other rigorous reviews or meta-analyses have a role in guidance documentation for actual health claims. Unless a medical source reviews numerous RCTs or other MEDSCI sources, it falls below the standard for encyclopedic conclusion, except as an example. The way the article shows the COSMOS study now - as one example of 2022 ongoing primary research - is where it should be, if at all. I am fine with deleting it altogether. Zefr (talk) 19:46, 5 May 2022 (UTC)[reply]

Preliminary and primary are not the same thing and carry different connotations in common speech. ★Ama ^{TALK CONTRIBS} 20:09, 5 May 2022 (UTC)[reply]

I agree with Amadeus on this point.

Also, I disagree in principle with "Unless a medical source reviews numerous RCTs or other MEDSCI sources, it falls below the standard for encyclopedic conclusion, except as an example". Some RCTs are WP:Notable (=qualify for a separate, stand-alone article) even before the first participant is recruited. It would be strange if we could write whole articles about the existence of RCTs before there are any results, but we simultaneously couldn't write a single sentence saying that an RCT happened.

I do think, though, that once we get into comments about exactly how various numbers were adjusted, we're well beyond the level of what editors should be doing. Primary medical research should generally be cited for brief and fairly non-committal facts, like "This thing exists" or "A single study said something". WhatamIdoing (talk) 21:27, 5 May 2022 (UTC)[reply]

Yeah, to be honest, it was more just an added note from my own knowledge and experience, and not one related to Wikipedia standard editing and such. Perhaps I shouldn't have included it to begin with for that very reason. I just can't help trying to help everywhere haha. ★Ama ^{TALK CONTRIBS} 21:32, 5 May 2022 (UTC)[reply]

Thank you @Amadeus1999 for your opinion. While I might disagree in some aspects on the interpretation of the study from a scientific point of view, I agree with your recommendations for inclusions. The interpretation of the results is difficult and one can argue in many ways whether "intention to treat" or "per protocol" is more appropriate - but both interpretations suggest a benefit and thereby support the results of Cochrane and Raman review.

Both systematic reviews comment on the poor quality of studies conducted so far, hence a low quality of evidence. This is a study that - based on Cochrane or GRADE criteria - is rated as fairly high quality (study design, low attrition etc). So I would consider it appropriate to mention it as "supporting" the findings of the review, without mentioning any further details.

A conflict of interest, as mentioned by @Zefr, needs to be reported - but it does not automatically invalidate results. Almost all pharmaceutical studies are funded by industry - and this is one reason why there are fairly strict rules on how studies are conducted and published. The cocoa flavanols are also not branded: ". "2 × 2 factorial trial testing a cocoa extract supplement [2 capsules/d containing 500 mg cocoa flavanols/d, including 80 mg (–)-epicatechin; supplied by Mars Edge] and a multivitamin supplement (Centrum Silver®; supplied by Pfizer Consumer Healthcare, now a part of GSK Consumer Healthcare)."20:05, 5 May 2022 (UTC) Ggux (talk) 20:05, 5 May 2022 (UTC)[reply]

Addendum: The entry for Pfizer–BioNTech COVID-19 vaccine contains quite a lot of references to primary research, including the results of the Phase 3 study. Using the criteria stated here, this study should not be included as a) it was primary research, b) it was funded by Pfizer and c) it promoted a Pfizer product. There are obviously a lot of differences, but within the field of flavanol research, a study on 21,000 people is a step-change and unlikely to be repeated (studies of that size are uncommon in nutrition). Ggux (talk) 21:31, 5 May 2022 (UTC)[reply]

That's obviously not a fair comparison, as you said yourself haha. The COVID-19 vaccine stuff is an exception of its own kind due to the pandemic and rapidly evolving and changing situation demanding the use of primary sources since there's not enough time for secondary or tertiary sources to be produced in a timely fashion so people don't die. We don't have that kind of rush here, I'd say. ★Ama ^{TALK CONTRIBS} 21:34, 5 May 2022 (UTC)[reply]

Of course it is not - but using the argument that primary research must not be cited would have to apply there as well. But there are inconsistencies in the application of the "no primary data" rule. The Β-Carotene article doesn't mention the key trial that showed an increased risk first (which might be of historical interest), but cites a number of fairly small in vitro studies.

I appreciate that it's difficult to find a balance and not to promote one side on an argument - but I have to admit that I have difficulties understanding the objection to including a reference to a study that (within the area of flavanol research) is generally considered to be very important to move "the field" forward.

(And as I said above: I'm clearly biased as this is my area of work.) Ggux (talk) 21:43, 5 May 2022 (UTC)[reply]

Hm, don't get me wrong. It's not even like there's a 'no primary data' rule set in stone that we must always follow. That's not the primary qualm I have with the source anyway. Yes, there's occasions where using primary sources are okay, even for medical topics, but they're few and far-between I'd say. I don't think it's the case here. There's too much controversy when it comes to the conflict of interest, extraordinary claims (with regards to decreasing mortality for example) and other issues that primary sources sometimes/often suffer that I personally think we should not expose the average Wikipedia reader too. That's just my two cents though. ★Ama ^{TALK CONTRIBS} 23:42, 5 May 2022 (UTC)[reply]

I'm surprised by the claims of controversy - there isn't really that much outside of this discussion. The study design is fairly conservative and the outcomes are largely what would be expected from previous studies, i.e. a modest reduction in mortality. The controversy has mainly been created by @Zefr who initially claimed the study to be small and based on only a few 100 people. Ggux (talk) 05:31, 6 May 2022 (UTC)[reply]

If there's a controversy, it's from the study's own conclusion, quoting from the article results under Cocoa extract and total cardiovascular events: "Participants taking the cocoa extract supplement experienced no significant benefit on the primary outcome of total cardiovascular events" which was defined as the primary study objective, having 410 participants taking the cocoa extract and 456 taking the placebo - these are the subject numbers and results I quoted from the beginning of this discussion. Note also - Marion Nestle, an eminent scientist in the nutrition field, also found this study controversial and the results weird and mainly insignificant, concluding: "We now have the result of this trial. Even though cocoa flavanol supplements did not reduce cardiovascular events, Mars got its money’s worth from what must have been a very expensive study." Other than the literally unbelievable benefit of lowering death risk by 27% (statistically only), what cardiovascular effects are worth talking about from the study? If we're going to use the study as a source in the article, all we can say, as one editor said above, "a study happened," which is weak content for the encyclopedia and useless to the common user. Zefr (talk) 06:26, 6 May 2022 (UTC)[reply]

Please quote the conclusion in it's entirety: "Cocoa extract supplementation did not significantly reduce total cardiovascular events among older adults but reduced CVD death by 27%. Potential reductions in total cardiovascular events were supported in per-protocol analyses." So the data suggest a beneficial effect according to the authors.

I don't understand why you focus on the number of participants who met the endpoint - this is one aspect of the study, but the sample size of 21,442 is generally considered to be more important. Based on your assumptions, one would have to reject the (Pfizer-funded) Bnt162b2 study as it is based on only 8/162 people in the intervention/placebo group reaching the endpoint.

Why do you quote Marion Nestle? She is not an expert in flavanols or bioactives and has a known bias against industry funding. She does not comment on the scientific validity of the study but simply voices an opinion. As an expert opinion, it should be treated as less relevant than a primary study based on any evidence pyramid I am aware of.

I believe mentioning the study has encyclopaedic value at this moment as no better data are available and it is a very fundamental study in this field. Ggux (talk) 06:52, 6 May 2022 (UTC)[reply]

I agree it has encyclopedic value, just maybe not about how it should be construed/presented. I doubt Zefr, you, and I will reach any meaningful consensus about that though. ★Ama ^{TALK CONTRIBS} 14:02, 6 May 2022 (UTC)[reply]

The section is "Health claim and Research" - I think in that context one could argue that it is an (important) part of research. But perhaps it would be better to separate research and health claims? Incidentally, the COSMOS-Study has a Wikipedia entry - I did suggest many years ago to refer to this study, but this was rejected by @Zefr. Perhaps this would be an opportunity to do so, and add a statement about the results to the COSMOS-Study entry?

I have to admit that I find this discussion rather complicated - I don't understand how a single person can decide the appropriateness of entries. Ggux (talk) 14:27, 6 May 2022 (UTC)[reply]

I don't understand how a single person can decide the appropriateness of entries. They can't, that's the entire reason a 3O was requested. If we're able to reach a clear and well-defined consensus, the opposing side cannot undo edits on his own. That's the entire reason we're discussing it. Amadeus²² 🙋 🔔 15:33, 6 May 2022 (UTC)[reply]

Let's hope we come to a conclusion - but in a different entry, @Zefr has simply removed a consensus statement a few years after an RfC. Ggux (talk) 15:43, 6 May 2022 (UTC)[reply]

Diffs of this supposed removal? You shouldn't make these directed claims at people without backing them up. Also, do you have any reason to not assume good faith? It happens that people miss RfCs or make controversial edits without knowing it. Amadeus²² 🙋 🔔 15:57, 6 May 2022 (UTC)[reply]

It was a very long discussion (about the quality of the evidence) and agreement vis RfC that it should remain in the text as it is relevant. Five years later, @Zefr removed it due to missed WP:MEDRS. I assume sometimes editors are a bit too keen enforcing rules. Ggux (talk) 17:46, 6 May 2022 (UTC)[reply]

Assuming rarely does any good in my (albeit limited) experience. Amadeus²² 🙋 🔔 20:35, 6 May 2022 (UTC)[reply]

I assume that's very sensible. But for a newcomer, some of this is very difficult to understand. Having sensible rules on consensus when writing about medical topics being enforced by a single person seems to be a bit contradictory.

There appears to be a consensus here among editors that citing the study (without further elaborating on it) would be acceptable, and that this is not really the place to second-guess the quality of a study published in a leading journal. However, the discussion keeps going and we have now the second round of discussion about funding and the sample size. I don't understand why this keeps going on ... Ggux (talk) 05:52, 7 May 2022 (UTC)[reply]

10,719 participants took the cocoa extract, 10,723 were in the control group. While there were no statistically significant benefits for the primary endpoint, there were significant endpoints for the secondary endpoint - both of these were pre-specified before data-analysis. CONSORT requires for good reasons a very formalised approach to publishing results, including a clear emphasis of the primary endpoint. But this does not invalidate the results of secondary endpoints.

The numbers you quote are those that reached the relevant endpoints - but this is taken out if context: fewer than 200 people developed COVID in the trial of the Pfizer vaccine. Marion Nestle might be eminent (she is also controversial), but we should prioritise evidence over eminence. She does not provide any scientific criticism of the study, she only refers to the funding source. The 27% are based on the data - I don't know what you mean with "unbelievable" or "statistically only". Do you accuse the authors of fraud or that AJCN did not review the study properly?

This is not preliminary research - a 20,000 participant study is not conducted as "preliminary" research, and you already quote various reviews that contradict this idea. At various conferences (and in the papers I included) the possibility of including flavan-3-ols in future dietary guidelines - that would be unlikely if it was preliminary.

I suggest to implement what @WhatamIdoing suggested:

"Some clinical trials suggest that flavan-3-ol intake may have a small beneficial effect on cardiovascular health", citing the Cochrane review & COSMOS. Ggux (talk) 15:37, 7 May 2022 (UTC)[reply]

Addendum

There is a range of papers discussing the development of Dietary Reference Intake recommendations for these compounds, e.g. here: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8321833/

Joanne Lupton - at least as eminent as Marion Nestle - clearly considered there to be a case: https://pubmed.ncbi.nlm.nih.gov/24566766/ Ggux (talk) 15:44, 7 May 2022 (UTC)[reply]

I agree this would be an acceptable inclusion. That means the current consensus is to add it, and one person's opinion won't be enough to equal or overrule it. Feel free to implement it (for now), people have had enough time to respond. Amadeus²² 🙋 🔔 20:34, 7 May 2022 (UTC)[reply]

It's not clear what Amadeus1999 regards as a consensus and acceptable content - discussion of dietary recommended intakes (not approved by any regulatory agency; not useable) or use of the COSMOS study as a source, which was already completed on 4 May with this edit. Regarding consensus about how to discuss the COSMOS trial, I don't agree that two in favor vs. one against is a consensus, indicating the need for further editor input to the discussion. There is a consensus that the COSMOS study is primary research, not equal to a Cochrane review for source weight (3 out of 4 editors concur on this, consistent with MEDASSESS). I adjusted the text further today here, removing the word "preliminary", as this seems to be a significant irritant to Ggux, who would do well to read and abide by WP:NOWIN. Zefr (talk) 21:23, 7 May 2022 (UTC)[reply]

I meant inclusion of the COSMOS trial, yes, and I'm sorry, I wasn't aware it was already added. I think removing the word preliminary was appropriate. Thanks for this and once again my apologies. Amadeus²² 🙋 🔔 21:43, 7 May 2022 (UTC)[reply]

Thank you for your help - it appears that this discussion has become somewhat more heated than necessary. Ggux (talk) 21:46, 7 May 2022 (UTC)[reply]

I suggest the consensus:

"Some clinical trials suggest that flavan-3-ol intake may have a small beneficial effect on cardiovascular health", citing the Cochrane review & COSMOS

I find it rather disingenuous that @Zefr makes such accusation after making wild and unfunded claims about a study (and refusing to elaborate on them when I tried to explain apparent misunderstandings). I have mentioned dietary recommendations (which will probably be presented in autumn at FNCE 2022) to highlight that the assumption that there is weak evidence and a lack of scientific consensus is incorrect. My assumption was that an encyclopedia would welcome contributors with subject knowledge and expertise - not try to treat them like a nuisance. Ggux (talk) 21:44, 7 May 2022 (UTC)[reply]

Please don't take the actions or words of Zefr, me, both of us or any other limited amount of Wikipedians as representing the 'image' of Wikipedia at large. There's a lot of contributors, and we're not a homogenous mass or hivemind. Also, we simply have to be wary of editors acting with a potential conflict of interest as there's an ever-growing increase of paid editors or otherwise unproductive additions or removals from WP. I'd like if you could not take this as a personal matter. I've also never included your subject knowledge or expertise as a reason to disprove or 'go against' your ideas or proposals. Amadeus²² 🙋 🔔 21:47, 7 May 2022 (UTC)[reply]

I appreciate that it takes two sides - someone with subject knowledge and someone with editorial knowledge to ensure that an entry is both correct and appropriate. I lack the latter, so I welcome being directed in the right direction and understand why a primary source should not feature prominently. But what I struggle with is that contributions are rejected based on wrong assumptions or opinion pieces. And I don't understand the inconsistencies: why is it appropriate to give an entire section to a single (primary) paper talking about an imaging method, but mentioning a large RCT is not? Ggux (talk) 21:58, 7 May 2022 (UTC)[reply]

I don't know what imaging method you speak of and it probably doesn't belong being discussed here but if you're unsure about a source being relevant or adequate in proportion to the claim being made at the article you can do a few things:

1) Be bold and remove the entry entirely if you believe the claim is factually wrong (especially if it's obviously wrong, like vandalism)

2) Start a discussion on the respective Article's Talk page regarding the source if you think it may be right but it may also not be

3) You can also use the {{moresources}} tag to notify readers and editors that a claim or sentence may need more citations to be included by Wikipedia/encyclopedic standards. Keep in mind with this that this should always be proportional to the claim being made: A small mention of a relatively minor claim requires less substantiated evidence or sources than an extraordinary claim that goes against the established order.

4) There's also the reliable sources noticeboard which is watched and maintained by experienced editors and administrators alike, which may intervene or give advice regarding a source or whether a source needs more citations.

I'd be more than happy to look at the imaging method article or section that's only cited by one (primary) source though. Amadeus²² 🙋 🔔 22:12, 7 May 2022 (UTC)[reply]

Thank you - it is this section here: flavan-3-ol#Other uses. Ggux (talk) 05:47, 8 May 2022 (UTC)[reply]

This blog post says that cocoa flavonols "are destroyed by traditional chocolate processing". Maybe that fact should be briefly mentioned in Flavan-3-ol#Sources of catechins (with a proper source)? WhatamIdoing (talk) 21:21, 8 May 2022 (UTC)[reply]

There has been a lot of research in this field, as supplements need to be standardised. It is not just that manufacturing can change the flavanol content - the analytical methods have also not been standardised for some time, so results can be different to compare. Wouldn't it be better to separate "health" and "research"? It would make it easier to provide information on research without giving wrong impressions about health benefits. Ggux (talk) 21:42, 8 May 2022 (UTC)[reply]

I have added something on this - I hope it is acceptable. Ggux (talk) 06:21, 9 May 2022 (UTC)[reply]

I would like to propose to revise the research & health section completely. At the moment, it looks like designed by committee (which it is) and there are information of different weight listed in a rather random order. Is it possible to draft a revised version and prepare it before publishing it? Ggux (talk) 08:13, 9 May 2022 (UTC)[reply]

I have revised the page and tidied up some section. I believe @Zefr should accept that mentioning COSMOS "A study took place" is acceptable according to the discussion here. Ggux (talk) 18:10, 13 May 2022 (UTC)[reply]

There is no consensus for your edits or for including the COSMOS study, which is primary research. We do not cite primary research for content related to human health or diseases, WP:MEDPRI. Further warning about WP:DE. Zefr (talk) 18:13, 13 May 2022 (UTC)[reply]

Can you please tell me where the "Source violation" is? And I would really like to understand why the edit (apart from COSMOS) is wrong as it is largely the same content as before. Ggux (talk) 18:31, 13 May 2022 (UTC)[reply]

"If conclusions are worth mentioning (such as large randomized clinical trials with surprising results), they should be described appropriately as from a single study:"

That applies to COSMOS. Ggux (talk) 19:15, 13 May 2022 (UTC)[reply]

I have added the following section to this entry which @Zefr has reverted without discussion. As the inclusion of the COSMOS-Study was subject to more discussion, I put it in a separate RFC. It is difficult to place this in the talk page, but the different versions are visible in the history. Ggux (talk) 18:55, 13 May 2022 (UTC)[reply]

@Ggux No idea what either of these RfCs are asking. Can you boil it down to an exact block of text, link, or WP:DIFF that can be discussed in a neutral fashion? I see multiple things that you and Zefr have disagreed over. CaptainEek ^{Edits Ho Cap'n!}⚓ 16:13, 14 May 2022 (UTC)[reply]

Thank you. The problem is that for the first one, I do not understand what @Zefr objects to. He has reverted edits based on the fact that flavan-3-ols are not nutrients, but this has never been claimed. I asked for an explanation, but have not received one.

I will revise the second one. Ggux (talk) 17:52, 14 May 2022 (UTC)[reply]

The differences are here:

Special:Diff/1087646114/1087646387 Ggux (talk) 18:11, 14 May 2022 (UTC)[reply]

Well I don't see much of a practical difference between the edits, except for and thereby a possible reduction of cardiovascular disease risk. These results are mainly based on small-scale, short-term clinical trials, and so far the COSMOS-study is the only large scale study investigating the long-term effect. which is uncited and does need to be removed. CaptainEek ^{Edits Ho Cap'n!}⚓ 19:04, 14 May 2022 (UTC)[reply]

Thank you! That is RfC 3 - is it appropriate to cite a single large scale study? There have been extensive discussions, with a broad consensus that it is appropriate if the study is in some ways noteworthy, but unfortunately @Zefr does not agree and does not engage with this discussion.

@Zefr's concern - as far as I see - is that the text refers to "Physiological Effects" and that the heading is "Flavan-3-ols in Human Nutrition" as this would imply the status as nutrient. Unfortunately, there has been no discussion or attempt to find a solution. Ggux (talk) 19:31, 14 May 2022 (UTC)[reply]

@CaptainEek, I wonder: to the extent that the problem appears to be "one editor does not agree and does not engage", do you think this might be more appropriate for ANI than RFC? I hate to send people that way, but content RFCs don't necessarily solve behavioral problems. WhatamIdoing (talk) 19:20, 22 May 2022 (UTC)[reply]

What is the conclusion of the RfC? I believe that the current state of the article does not represent the scientific consensus.

My suggestion was to include a section "Flavanols in human nutrition" as they are unavoidably part of the human diet and I don't see how this could be misunderstood as meaning that they are nutrients. But they are of some interest and most of their interest is based on their role in the diet. Ggux (talk) 21:52, 7 July 2022 (UTC)[reply]

Flavan-3-ols have been investigated in a large (n=20,000 subjects, 3.5 yr) clinical trial (Women's Health Initiative#COcoa Supplement and Multivitamin Outcomes Study (COSMOS) and shown to have a beneficial effect on cardiovascular-disease risk. This is a single intervention study - but the largest ever undertaken in this field. The results of the study support the conclusions from previous systematic reviews on a much larger scale. Is mentioning this study (with a link to Women's Health Initiative#COcoa Supplement and Multivitamin Outcomes Study (COSMOS)) appropriate according to WP:MEDPRI?

The relevant section in my opinion is:

Determining weight of studies requires reliable secondary sources (not press releases or newspaper articles based on such sources). If conclusions are worth mentioning (such as large randomized clinical trials with surprising results), they should be described appropriately as from a single study.

Ggux (talk) 17:58, 14 May 2022 (UTC)[reply]

It bothers me that the COSOMS study is part funded by Mars, as well as that confectionary company providing practical support such as, and I quote from the COSMOS website, 'providing infrastructure support and the donation of study pills and packaging'. Moreover, where is the 'beneficial effect on cardiovascular risk? Again, quoting from the COSMOS report of findings:

"...this translates to a modest 10% reduction in cardiovascular events for those taking cocoa extract, but it was not statistically significant."

Fortnum (talk) 16:26, 15 May 2022 (UTC)[reply]

Thank you. I hope it is appropriate to offer an explanation:

• The primary endpoint is non-significant, the secondary (cardiovascular mortality is). But whatever the outcomes, it is a large trial that - in my opinion - is relevant for this group of compounds. I don't think it would be appropriate or necessary to discuss it any further, but neither do I think it should be ignored (as suggested in WP:MEDPRI)
• Funding sources do not appear to be part of WP:MEDASSESS. It is a difficult topic (and has been at least in nutrition for a long time - not so much in pharmaceutical studies), but where would one place a threshold? One of the reviews cited here without objection (Raman et al.) was funded by ILSI where there are also considerable controversies (see ILSI#Nutrition). Multivitamin#Cohort_studies refers to a study funded by Pfizer (manufacturer of Multivitamin Pills) - and expands on it well beyond what has been proposed here. There is therefore a precedent for mentioning studies with industry funding. Ggux (talk) 19:00, 15 May 2022 (UTC)[reply]

• Summarize many studies - I'm not super well acquainted with this topic, but at first glance, it seems like there are a whole slew of papers out there linking flavanols to positive cardiovascular outcomes. I don't think the average reader is going to be interested in specific studies as much they'll be interested in knowing there's a body of work supporting the idea that flavanols have been linked to minor positive outcomes in cardiovascular health. NickCT (talk) 13:13, 17 May 2022 (UTC)[reply]

  The studies are summarised in two reviews cited - but these are all very small/short-term studies. The reason why I believe it would be appropriate to mention COSMOS is it's scale and duration. While most other studies had at most a few 100 participants, it has around 20,000 and lasted for more than 3 years. I think referring to the wikipedia entry for COSMOS would be sufficient (it's explained there) - I don't think it should be highlighted in any way beyond that fact that it happened. Ggux (talk) 08:19, 18 May 2022 (UTC)[reply]

  I don't doubt that this might be a particularly meaningful research study, but I don't think the source of the information really matters to the reader. We generally don't provide sources of facts or research unless we think that adding a source somehow increases or decreases the credibility of information being transmitted. I can't imagine many readers are really going to know or care what COSMOS is, as much as they're going to be interested in knowing about the basic conclusions of the research. NickCT (talk) 23:50, 19 May 2022 (UTC)[reply]

• Regardless of the comprehensiveness of the study, reviews/meta-analyses are always preferred. Since other editors have found the study notable enough to mention in another context, it may be worth mentioning simply that a long-term study concluded in 2022, without mentioning results. That is because even if the results were something that might be surprising or amazing, the scope of any one study is always far narrower than what most people like to extrapolate, especially when it comes to personal health. That said, what is currently written may be more cautious than it needs to be, since studies have been consistent. You can summarize the results in Ried 2017: some evidence that flavanols cause a small blood pressure-lowering effect in the short term, or more cautiously, some evidence links flavinols to a small reduction in blood pressure in the short term. SamuelRiv (talk) 16:19, 19 May 2022 (UTC)[reply]
• Wait. WP:MEDPRI is clear on this one. Having looked at the study, there are many red flags: p-values, while not specified, must be very close to 0.05 considering CIs are generally very close to including 1. The method of "censoring" participants who were deemed to be non-compliant seems unusual to me, but may be state of the art (and it is why I can't calculate those p-values myself); it may or may not be appropriate for an intervention with a drug that causes noticeable side effects. I think WP:MEDPRI applies in full here, particularly the section about "surprising results": there weren't any. You had about 20 outcomes, one of them seemed to be significant: that's what you'd expect from placebo-vs-placebo trials. The outcome that people care about most (CVD death) wasn't pre-specified, and the one that people care about even more (overall death) was nonsignificant. I say "red flags" not to disparage the researchers in any way, but by considering how I would respond to a family member deciding to take drugs based on this study. (Which is, of course, not medical advice). IpseCustos (talk) 20:57, 12 June 2022 (UTC)[reply]

  If I might make a comment:

  1. 1: Whether a result is null or not doesn't really say anything about the quality of the study - so I don't think this should be used to determine whether a study is referred to (or not).
  2. 2: Per-protocol analyses (i.e. censoring at the end of known compliance) is quite common in medical research, especially when there is more interest in the physiological effect of an intervention and not the intervention itself (if we want to know whether a drug works, including those who didn't take them will attenuate outcomes; if we want to know whether a treatment works, we need to consider how many people are non-compliant)
  3. 3: CVD death was a pre-specified secondary endpoint.

  My reason for suggesting that it should be included is that it the trial in itself is very unusual in this field (very few large RCTs) and that in my opinion it would be appropriate for it to be mentioned (with a referrer to Women's Health Initiative#COcoa_Supplement_and_Multivitamin_Outcomes_Study_(COSMOS) (for transparency: I have added the results). I did not expect this to be so incredibly controversial as it is not within the scientific community - but I accept the decision not to do so. Ggux (talk) 20:19, 14 June 2022 (UTC)[reply]

  I'm sorry, you are absolutely correct about 3. I had misread one of the lengthy statements about the non-prespecified outcomes that were constructed. IpseCustos (talk) 20:45, 14 June 2022 (UTC)[reply]

This edit was made to 1) move two complex images to a gallery, giving more attention to content that may aid general user understanding of this topic; both images are complex and out of the mainstream of the article; 2) move two sources into the text per WP:INCITE; and 3) remove from the Research section an unnecessary, inconclusive, non-neutral, advocacy sentence ("Accumulating evidence suggests that flavan-3-ol consumption can be beneficial for cardiovascular and metabolic health"). The following sentence in the section is factual, sufficient, and current as of 2019. Zefr (talk) 19:27, 8 September 2022 (UTC)[reply]