Wikipedia talk:Mediation Cabal/Cases/2007-03-28 Antioxidant

From Wikipedia, the free encyclopedia

Contents

[edit] Opening Comments from the Mediator

Okay. I'd like to open by mentioning that it looks, from what I'm seeing right now, as though Wikipedia policy would agree with Paul on the subject of his source. I don't want anyone to think I've lost my neutrality, as I will happily change that position if it comes to light that I am mistaken. I'm basing my conclusion on the following:

  • Wikipedia policy does not state that a peer-reviewed source is MORE reliable than a non-peer-reviewed source, so long as the non-peer-reviewed source is still reliable.
  • Even if Wikipedia policy did state that a peer-reviewed source were more reliable, it can be reasonably argued that Paul's source is peer-reviewed to the best of its ability.
  • Wikipedia policy DOES state that all positions must be given due weight if they carry proper sourcing.
  • It is unfair to remove a plainly valid position on the basis that if it's valid, another source will eventually pick it up. Paul's source, while not exactly the New American Dictionary of Medicinal Studies, appears to me (and I am admittedly not terribly knowledgeable regarding medicine) to be written by individuals familiar with the field and it appears to be a valid source under Wikipedia policy, so Paul's position must be given due weight.
  • Due to the limited availability of published material on the subject, I would argue that under Wikipedia policy, the study and those criticizing it should be given equal weight and the decision should be left, ultimately, to the reader of the article.

However. It also appears to me that the real problem here isn't in the source, it's in the amount of weight being given to the position associated with it. After reading the article's talk page, it looks like both of your versions of the paragraph give undue weight to your own positions. So, unless one of you disagrees with me, I think what we should really focus on is coming up with a version of that paragraph that gives each position more equal weight. --Moralis (talk) 18:08, 12 April 2007 (UTC)

[edit] Comments on policy

I'd disagree with that interpretation of the policy outlined in WP:RS and WP:UNDUE.

The parts that are especially relevant are:

Caution should be used when using company or organization websites as sources. Although the company or organization is a good source of information on itself, it has an obvious bias. link

NPOV says that the article should fairly represent all significant viewpoints that have been published by a verifiable source, and should do so in proportion to the prominence of each. link

As the industry news websites proposed as sources are of lesser quality (no peer-review etc) and suffer a bias problem, as outlined above. Then to give them equal weight is a clear breach of the NPOV policy. TimVickers 18:54, 12 April 2007 (UTC)

The policy in company/organization websites refers to using them as a source for an article on them. For example, it cautions against using nike.com as a source for Nike, or a source for a competitor, like Adidas. The source being used is not pushing an agenda- they don't have products to sell, and they don't have a vested interest in skewing public perception in their favor. They're writing an educated opinion on the report.
As far as policy is concerned, the report and the sources Paul's provided that challenge the report should be considered of equal prominence for the purposes of this article, in order to ensure that the report is not cast in an unfairly positive or negative light.
npicenter.com is not, simply because it is an industry publication, a "company or organization website." That policy, to rephrase, refers to websites of a commercial nature. The fact that it's an industry publication supports its case as a reputable source- it is by definition authored by individuals knowledgeable about the field. If the company had a reason to disparage the report in question, it would no longer be a valid source, but since it doesn't, that particular piece of policy doesn't apply.
I think we should focus on the more pressing issue, which is authoring a version of that paragraph that's more neutral. --Moralis (talk) 19:12, 12 April 2007 (UTC)

I'm not sure why you think companies that sell antioxidant supplements have no reason to disparage a report that suggests these supplements are harmful to health. This is a website directly associated with people selling these products. A report has suggested these products are harmful and the companies have published a press release disparaging the report. Personally, I think there is a clear and obvious bias problem! TimVickers 19:24, 12 April 2007 (UTC)

See, I've not heard anyone mention this company selling such supplements until now. I'm going to reserve my opinion until hearing Paul's response. --Moralis (talk) 19:39, 12 April 2007 (UTC)
www.npicenter.com "NPIcenter is the leading global online information resource for professionals in the Nutraceutical, nutritional, dietary supplement, cosmetic, and food industries, NPIcenter designs highly targeted programs that increase exposure and visibility and generate qualified leads for our clients - product and service companies involved in our target industries." One of the authors who's opinions kept being added was Dr. Andrew Shao, who is "Vice President of Scientific and Regulatory Affairs at the Council for Responsible Nutrition USA" link The Council for Responsible Nutrition is "a Washington-based trade association representing ingredient suppliers and manufacturers in the dietary supplement industry." link. I'm sure you can see why I think this source might be biased. I would be happy however with excluding these industry sources and instead using the opinions of independent academics with no financial interest. The Oregon State news release link does not suffer the obvious drawbacks of the NPIcenter and Neutracuticals world sources. However, it should be recognised that opinion is divided on this, with many positive reviews of the meta-analysis and scating critiques of the people who are trying to cast doubt on its conclusions. link 1, and link 2 link 3. I feel it would be wiser not to step into this cauldron of claim and counter-claim and simply let the facts speak for themselves. TimVickers 23:08, 12 April 2007 (UTC)

Returning to policy, I disagree with your statement that "the sources Paul's provided that challenge the report should be considered of equal prominence for the purposes of this article" as the policy states that "NPOV says that the article should fairly represent all significant viewpoints that have been published by a verifiable source, and should do so in proportion to the prominence of each." The JAMA is an international peer-reviewed scientific journal - while news releases and blog entries are far less prominent and reliable, consequently they cannot be given equal weight. TimVickers 23:35, 12 April 2007 (UTC)

For the purposes of the policy, I would say there's a point at which you simply see reliable vs. unreliable, rather than reliable vs. MORE reliable. However, if you're correct that the source in question is in fact a commercial web site, they do have a bias problem, and are not a reliable source. I'm waiting on Paul's response to comment on the source vs. policy again. If you're right, I think Wikipedia policy is clear- if not, I think we've got a different issue, as discussed above. --Moralis (talk) 06:13, 13 April 2007 (UTC)

Before we continue, I would like to hear Paul's reply to Tim's comment that the author of the source in question is employed by a company which sells the supplements in question. --Moralis (talk) 15:12, 13 April 2007 (UTC)

I provided comments earlier this morning at http://en.wikipedia.org/wiki/Wikipedia:Mediation_Cabal/Cases/2007-03-28_Antioxidant#Mediator_response
Are we consolidating these two discussion locations? Thanks.
Just as an example among several critical assessments of the JAMA article is this one coauthored by two MDs, http://www.lef.org/featured-articles/consumer_alert_020307.htm -- there are several reports like this by scientific parties critical of the JAMA meta-analysis but who would not be considered biased by industry association.
Also, my position need not be a matter of "equal weight", as I debated with Tim previously. A counter meta-analysis and getting an article published in a professional journal takes minimally a year to provide "equal weight", whereas balanced, scientific review and a different opinion would be perfectly valid counterpoints to the published article in JAMA -- perhaps not "equal", but that's not the point of disputing the JAMA meta-analysis. The point is providing balance in interpretation of the article. --Paul144 15:31, 13 April 2007 (UTC)
This article does appear to be heavily sourced and scientifically valid- which puts us back at authoring a mutually acceptable version of the paragraph. I do think it should be kept to a paragraph, or two, at most. The subject of the article is antioxidants, and the JAMA study is only a very narrow piece of that topic.
If both of you would like to post what you consider an "acceptable" version of the article here (this can be new or something you'd already written) for easy comparison, I think we could make some headway by trying to combine the two. --Moralis (talk) 15:40, 13 April 2007 (UTC)

[edit] Proposed versions

Counterpoint

These harmful effects may also be seen in non-smokers, as a recent meta-analysis including data from 68 separate clinical trials of 232,606 patients showed that β-carotene, vitamin A or vitamin E supplementation was associated with increased mortality[1]. However, this meta-analysis found no significant effect from vitamin C supplementation on mortality. Furthermore, its conclusions are questionable due to the wide heterogeneity of patients already ill with varied diseases studied in different trial designs, treatment dosages and durations[2]. --Paul144 16:37, 13 April 2007 (UTC)

These harmful effects may also be seen in non-smokers, as a recent meta-analysis including data from approximately 230,000 patients suggested that β-carotene, vitamin A or vitamin E supplementation is associated with increased mortality, but saw no significant effect from vitamin C.[132] These results are consistent with some previous meta-analyses that also suggested that Vitamin E supplementation increased mortality,[133] and that antioxidant supplements increased the risk of colon cancer.[134] However, the results of this meta-analysis are inconsistent with other large studies such as the SU.VI.MAX trial, which suggested that antioxidants have no positive or negative effects on cause-all mortality.[112][135] Overall, the large number of clinical trials carried out on antioxidant supplementation suggest that either these products have no effect on health, or that they cause a slight increase in the risk of disease.[109][ref] --TimVickers 17:12, 13 April 2007 (UTC)

I think Tim's version is definitely leaning toward the NPOV. How do you guys feel about:
These harmful effects may also be seen in non-smokers, as a recent meta-analysis including data from approximately 230,000 patients suggested that β-carotene, vitamin A or vitamin E supplementation is associated with increased mortality, but saw no significant effect from vitamin C.[132] These results are consistent with some previous meta-analyses that also suggested that Vitamin E supplementation increased mortality,[133] and that antioxidant supplements increased the risk of colon cancer.[134] However, the results of this meta-analysis are inconsistent with other large studies such as the SU.VI.MAX trial, which suggested that antioxidants have no positive or negative effects on cause-all mortality.[112][135] Furthermore, its conclusions have been questioned due to the wide heterogeneity of patients already ill with varied diseases studied in different trial designs, treatment dosages and durations[2].
Overall, the large number of clinical trials carried out on antioxidant supplementation suggest that either these products have no effect on health, or that they cause a slight increase in the risk of disease.[109][110]
Let me know if this is acceptable. If not, what changes would you make? If so, the next step is for you both to review the way that passage is sourced and decide whether you are satisfied with the references used. --Moralis (talk) 17:33, 13 April 2007 (UTC)

If you are going to include the "criticism" section (which I think is deeply unwise), we will also need to add that some welcomed the results of the trial as highly significant, otherwise this is giving undue weight to one side of the argument. This Times article and this article in NSNBC and this article in CNN give balanced and neutral reviews of this area for non-specialists. TimVickers 17:46, 13 April 2007 (UTC)

How about:
These harmful effects may also be seen in non-smokers, as a recent meta-analysis including data from approximately 230,000 patients suggested that β-carotene, vitamin A or vitamin E supplementation is associated with increased mortality, but saw no significant effect from vitamin C.[132] These results are consistent with some previous meta-analyses that also suggested that Vitamin E supplementation increased mortality,[133] and that antioxidant supplements increased the risk of colon cancer.[134] However, the results of this meta-analysis are inconsistent with other large studies such as the SU.VI.MAX trial, which suggested that antioxidants have no positive or negative effects on cause-all mortality.[112][135] While some have questioned this meta-analysis's conclusions and criticized its methods,[2] others see this as a significant study.link 1 link 2 link 3 Overall, the large number of clinical trials carried out on antioxidant supplementation suggest that either these products have no effect on health, or that they cause a slight increase in the risk of disease.[109][110]


The references 109, 100, 112 and 135 ("...have no effect or cause a slight increase in the risk of disease...") are outweighed by this evidence to the contrary:

From Faloon et al., 2007[1]

  1. A study involving over 29,092 male smokers aged 50-69 years followed prospectively for 19 years showed that men with the highest serum alpha-tocopherol levels had a 28% lower risk of total and cause-specific mortality than did those with the lowest levels, and a 21%, 29%, and 30% lower risk of deaths due to cancer, cardiovascular disease, and other causes;11
  2. A study in 3,254 people (1,260 males and 1,994 females) aged from 39 to 85 years followed from 1989 to 1995 showed that higher serum levels of carotenoids with pro-vitamin A activity significantly reduces the risk of mortality from cardiovascular disease and colorectal cancer;12
  3. A study in aging women that showed those with the lowest levels of alpha- and beta-carotene, lutein/zeaxanthin, and total carotenoids were significantly more likely to have increasing IL-6 levels over a period of 2 years, and those aging women with the lowest selenium levels had a significantly higher 54% risk of death over a 5-year period;13
  4. A study in patients with aggressive, small cell lung cancer showed a clinically significant 35% decreased risk of death associated with antioxidant supplement use after adjustment for tumor stage and other risk factors;14
  5. A study in 1,168 elderly men and women followed for 10 years showed that plasma carotene concentrations were associated with a 21% lower mortality risk for every 0.39 micromol/L increase in plasma carotene, a 41% lower mortality risk for cancer, and a 17% lower risk of mortality due to cardiovascular disease;15
  6. A study that evaluated the effect of Vitamin E, beta carotene, and vitamin C on prostate cancer risk in over 29,000 men during 8 years of follow-up showed that supplemental beta-carotene intake at a dose level of at least 2000 micrograms per day was associated with a highly significant 52% decreased prostate cancer risk in men with low dietary beta-carotene intake as well as a dramatic, 71% decreased risk of advanced prostate cancer with increasing dose and duration of supplemental vitamin E;16
  7. A study in 1,214 persons age 75-84 studied for over 4 years showed that those people with the lowest vitamin C plasma levels (< 17 micromol/L) had the highest mortality, whereas those aging people with the highest vitamin C plasma levels (> 66 micromol/L) had a mortality risk nearly 50% less;17
  8. A study that examined vitamin E and vitamin C supplement use in relation to mortality risk in 11,178 persons aged 67-105 years (Established Populations for Epidemiologic Studies of the Elderly) in 1984-1993 showed that vitamin E reduced the risk of all-cause mortality by 34%, reduced the risk of coronary disease mortality by 47%, and the simultaneous use of vitamins E and C was associated with a 42% lower risk of total mortality and 53% lower risk of coronary mortality;18
  9. A study (Chicago Western Electric Study) that followed over 1,800 middle-aged men over a 30-year period showed that during 46,102 person-years of follow-up the risk of fatal stroke was 29% lower in the group taking the highest amount of vitamin C and beta-carotene;19
  10. Two studies with different designs conducted in Linxian, an area of north central China with some of the world's highest rates of esophageal and stomach cancer and a population with a chronically low intake of several nutrients, showed significant reductions in mortality associated with antioxidant intake
  11. One study showed that in 3,318 persons with esophageal dysplasia, a precursor to esophageal cancer, significantly lower total and cancer mortality risk was observed in those Chinese receiving beta-carotene, vitamin E, and selenium, and a whopping 55% decrease in mortality due to cerebrovascular disease;20
  12. A second study in 29,584 adult Chinese followed from March 1986-May 1991 showed a significantly lower total mortality among those receiving supplementation with beta carotene, vitamin E, and selenium, with a significant 23% reduction in stomach cancer in this high-risk population;21
  13. A study in 1,078 pregnant women infected with HIV given daily multivitamin supplements including vitamins A, C, and E showed reductions in risk of death, reduction in risk of HIV progression, and reduction in viral load;22
  14. A study involving 15,419 children over one year showed the risk of death in the group supplemented with synthetic vitamin A (8,333 IU daily) was 54% less;23
  15. A study with lung cancer patients over age 60 showed that those patients taking supplements including antioxidant vitamins like A, C, and E had a dramatic 68% increase in survival, from only 11 months in non-users to an astounding 41 months for the vitamin users (median survival);24
  16. A study that showed daily oral administration of high-dose vitamin A (300,000 IU daily) was effective in reducing the number of lung cancers related to tobacco consumption and improved disease-free interval in patients surgically-treated for stage I lung cancer;25
  17. A study in 595 critically-ill ICU patients showed that supplemental vitamin C and vitamin E reduced the risk of multiple organ system failure by an amazing, statistically significant 57% along with a shorter duration of mechanical ventilation and length of ICU stay.26

These harmful effects may also be seen in non-smokers, as a recent meta-analysis including data from approximately 230,000 patients suggested that β-carotene, vitamin A or vitamin E supplementation was associated with increased mortality, but saw no significant effect from vitamin C.[132] Conclusions from this study, however, have been questioned due to the wide heterogeneity of patients already ill with varied diseases studied in different trial designs, treatment dosages and durations[2].

Also, the results of this meta-analysis are inconsistent with other large studies such as the SU.VI.MAX trial showing no positive or negative effects of antioxidant supplementation on cause-all mortality[112][135] and with at least 15 other clinical trials demonstrating improved mortality or morbidity resulting from antioxidant supplementation (literature summarized in [3]) --Paul144 01:19, 14 April 2007 (UTC)

I am very glad you have now reconsidered your position that "The natural products industry is too young -- just a few years of formal business activity -- to establish itself in academia where most peer-reviewed journals begin." and you have now chosen to cite studies on antioxidants from as long ago as 1995, over ten years old. However, as I'm sure you are aware, every time you change your story on this you lose credibility. Is your position now that sufficient clinical trials exist to make a simple statement on the risks and benefits of antioxidants supplements using data published in scientific journals? I would be happy to produce a summary of such data. TimVickers 06:56, 14 April 2007 (UTC)
Paul's changing his position is necessary in order to come to a compromise, as is some alteration of your own. Asserting that he loses his credibility by doing so isn't fair, and I think it undermines this dicussion to an extent. --Moralis (talk) 08:47, 14 April 2007 (UTC)

Where's the "change of position"? This is a process of revision with the benefit to Wikipedia and its public readers as a goal. I aspire to be open-minded and simply provide facts so there's balance preventing the entrenched snowjob Tim wishes to state. It doesn't matter whether relevant studies were done last year or last decade: they were clinical trials approved and published through peer-scrutiny and rigors of FDA review for describing study conclusions for the public.

The two paragraphs I provided above i) acknowledge the results of the JAMA meta-analysis and its apparent weaknesses, and ii) give the fair perspective that some antioxidant studies have also found no effects or significant effects. It is what it is.

We should agree this field takes only a snapshot over a relatively short time of clinical studies on antioxidants, a period of about 15 years. My bet is we'll need another 20 years of clinical trials to nail down specific effects for just a few antioxidant vitamins, minerals or phytochemicals. Meanwhile for Wikipedia, our position for readers should be a resource balanced with what the field is presently -- a mix of positive, negative and no-effect results, i.e., the fingerprint of a young research field and industry. --Paul144 13:40, 14 April 2007 (UTC)

I'm sorry, Paul. I of course welcome your introduction of high-quality peer-reviewed sources. As I'm sure we all agree "Disagreements over whether something is approached with a Neutral Point Of View (NPOV) can usually be avoided through the practice of good research."
With the vast amount of clinical trials that have been preformed on antioxidants, summarising the data is a challenge. As you can appreciate, merely choosing 15 positive trials from the thousands of examples on the grounds that these trials gave positive results would give a misleading slant to the summary. Fortunately, this work has been done for us through academic reviews of this mass of clinical trial data. As this may not be something you are familiar with Moralis, you can get an idea of how much is available to us here by going to PubMed and typing "antioxidant clinical trial" into the search engine. The search gives us over 8,000 results.
If you repeat the search, you can narrow it down a little - click on the "limits" tab below the search box and scroll down to "Type of Article" and tick "review". This narrows down the field to about 1,000 sources! Doing the same, but ticking "meta-analysis" gives you 59 studies that pooled together clinical trials and examined them together. I propose using just the most up-to-date of these meta-analyses (from the last five years?), rather than picking a few out of the thousands of primary sources. This would let us compare the JAMA meta-analysis with other recent meta-analyses, ie comparing like with like. Does this seem reasonable? TimVickers 17:45, 14 April 2007 (UTC)


I don't agree with comparing "like with like" since that is not possible due to the extensive flaws in the JAMA meta-analysis. "Like with like" would only be in name of "meta-analysis" since it is not possible to replicate trial designs precisely from analysis to analysis.

I have left the reference numbers intact as the article's reference list is well-organized -- references provided here would have to be incorporated into the reference list. It is likely more helpful to public users of this article that we provide a summary of evidence for benefit against major diseases and allow the Faloon article to summarize criticisms of the JAMA meta-analysis with examples of other clinical studies showing benefits.

Revised section

These harmful effects may also be seen in non-smokers, as a recent meta-analysis including data from approximately 230,000 patients indicated that β-carotene, vitamin A or vitamin E supplementation was associated with increased mortality, but saw no significant effect from vitamin C.[132] Conclusions from this study, however, have been questioned due to the wide heterogeneity of patients already ill with varied diseases studied in different trial designs, treatment dosages and durations[4].

Also, the results of this meta-analysis are inconsistent with large studies such as the SU.VI.MAX trial showing no positive or negative effects of antioxidant supplementation on cause-all mortality[112][135] and with numerous clinical trials demonstrating that antioxidant supplementation improved mortality or morbidity against

Other literature showing lowered disease risk resulting from antioxidant supplementation is summarized in [13]). --Paul144 15:45, 15 April 2007 (UTC)

What is the basis for your selection of these 7 clinical trials, from the hundreds available? Meta-analyses make this choice for us, and also assess the trials for quality and sample size. Moreover, if the JAMA meta-analysis is methodologically-flawed then the many other recent meta-analyses on this subject would give different results. I don't understand your statement that we shouldn't compare meta-analyses as it is "not possible to replicate trial designs" If you think the JAMA design is flawed, you would not wish to replicate its trial design. The best approach is to use the resources available that make a balanced and rigorous assessment of the large number of clinical trials. I would suggest comparing the results of the JAMA trial to the other large meta-analyses published in the last five years. See link 1 link 2 link 3 link 4 link 5 link 6 TimVickers 19:53, 15 April 2007 (UTC)


I feel it's easier to understand beneficial results of antioxidants by giving examples of clinical trials that showed positive effects against specific diseases the public would recognize. The 7 provided are just a sampling for major specific diseases studied.

This statement makes no sense: "...if the JAMA meta-analysis is methodologically-flawed then the many other recent meta-analyses on this subject would give different results." The JAMA meta-analysis is flawed because the authors biased the results through their choices of what to include and exclude[14]. Let's write and edit this article for the general consumer of average intelligence for whom meta-analysis is an esoteric term and specific major diseases are commonly known.

Look at where we were just on Friday when Tim proposed a near-final version by writing: Overall, the large number of clinical trials carried out on antioxidant supplementation suggest that either these products have no effect on health, or that they cause a slight increase in the risk of disease. That statement is simply false and misses an opportunity to state the benefit of antioxidants shown in a host of clinical work.

FYI - nowhere in the article is much emphasis given to the large number of disease-specific clinical trials showing positive effects of antioxidants. Both the section on Disease prevention and Adverse effects should be considered for revision to include this evidence.

Let's let Moralis now have a say. --Paul144 20:14, 15 April 2007 (UTC)

[edit] Arbitrary Break in Discussion

First off, I'm liking where this is going. With each successive reply you guys are getting closer to a version of the passage which is both neutral and complete. I agree with Paul that studies showing the benefits of antioxidant supplements should be included, considering the amount of space you're giving to the individual study that spurred this discussion. I also agree that even a layperson such as myself can see why the JAMA study is flawed- the results seem, based on this conversation, to have been tailored via the movement of samples, similar to the way that all toothpastes are recommended by nine out of ten doctors. In fact the study on its own could not be cited as a source- see the bits about replicability, corroboration, and recognition at Wikipedia:Reliable sources- but I think we can all understand why the study is being included. It's important to continue to write about the study, rather than citing it as a source and stating its conclusions as fact. --Moralis (talk) 20:37, 15 April 2007 (UTC)

OK, Moralis To summarise progress so far. We have agreed that industry-associated websites such as npicenter and industry spokesmen such as Andrew Shao are not reliable sources. We are now stuck on producing a version summarising how the conclusions of the report fit in with current research. Paul disagrees with the conclusions of reviews such as this, this and this and wishes to include a handful of clinical trials from the past 20 years, selected because they gave positive results. I do not think this selective presentation of positive data gives an accurate reflection of current research. Instead, I would advise comparing the JAMA meta-analysis with other recent meta-analyses, which combine all the clinical trials together and assess them as a whole.
I would recommend you read the JAMA study (if you have not already link I can send you a Pdf if you e-mail me) to understand why some studies were selected and others discarded. This is standard in these type of studies as some types of trials are more reliable than others. For example, studies comparing an antioxidant to a placebo are more reliable then those using drug alone. Studies where patients were assigned randomly to the placebo/antioxidant groups are even more reliable and the most reliable of all do this in a blind fashion where the people carrying out the trial do not know which group is which until they finish the trial. The fact that the JAMA study examined only low-bias randomised controlled trials and excluded poorly-designed uncontrolled trials is a mark of its quality, not a source of bias. TimVickers 20:51, 15 April 2007 (UTC)
I've looked it over. I think there are equally valid arguments for and against it as a reliable source. Regardless, it's my opinion that this dispute could have ended by now if you were each willing to give equal weight to the others' arguments. It's unlikely that an extra sentence or two in one direction or the other is likely to sway reader opinion, and the fact that we're having such difficulty determining which argument is more prominent indicates to me that neither argument is more prominent- especially when the sources being used are inaccessible to Joe Reader, anyway. --Moralis (talk) 20:03, 16 April 2007 (UTC)
I'm not sure what arguments there are for this top-quality peer-reviewed article published in an international scientific journal failing WP:Reliable sources, I realise some people have criticised it, but this comes with the territory - it stepped on a lot of toes. The problem is that I have looked over the recent publications in this area, and think giving equal weight to the "antioxidant supplements are good" and "antioxidant supplements are neutral/bad" sides of the argument is giving undue weight to a set of old and poorly-designed association studies over the more recent randomised controlled trials. I realise this is a difficult position for you as a mediator, as both Paul and I claim expertise you do not have. Paul is somebody actively involved in the research and sale of antioxidant supplements, and I am somebody who did their PhD on antioxidant metabolism. My preferred way out of this quandary would be to simply compare like with like by comparing the conclusions of this meta-analysis with other recent meta-analyses on the same subject. TimVickers 20:23, 16 April 2007 (UTC)

[edit] Definition of terms

As this discussion is getting more technical, it might help if we defined what we are talking about. TimVickers 02:49, 16 April 2007 (UTC)

Association study - A study using epidemiology to test if two things are associated. This type of study can say if two things occur together eg if saturated fat intake and heart disease, blue eyes and lung cancer, or cigarette smoking and watching NASCAR - commonly occur together. However, these studies cannot prove that A causes B, they only show if A and B are often seen at the same time.
Randomised placebo-controlled trial - A study where a set of people are divided randomly into two groups. One is given a drug and the other a placebo. The two groups are then followed over time to see if the drug has any effect different from the placebo. This type of trial can prove that A causes B.
Meta-analysis - A study that takes the results of several trials and combines them together to get a more accurate picture of the effects of a drug. Here, studies are assessed for the number of participants and how well they were designed - their trial quality. As you would expect, large, well-designed trials such as randomised placebo-controlled trials are given more weight than small badly-designed trials.

[edit] Compromise version

These harmful effects may also be seen in non-smokers, as a recent meta-analysis including data from approximately 230,000 patients suggested that β-carotene, vitamin A or vitamin E supplementation is associated with increased mortality, but saw no significant effect from vitamin C.[132] No health risk was seen when all the randomized controlled studies were examined together, but an increase in mortality was detected only when the high-quality and low-bias risk trials were examined separately. However, as the majority of these low-bias trials dealt with either elderly people, or people already suffering disease, these results may not apply to the general population.[2] On the other hand, these results are consistent with some previous meta-analyses that also suggested that Vitamin E supplementation increased mortality,[133] and that antioxidant supplements increased the risk of colon cancer.[134] Conversely, these results are inconsistent with other recent meta-analyses, which suggested that antioxidants have no positive or negative effects on all-cause mortality.link link. Overall, the large number of clinical trials carried out on antioxidant supplements suggest that either these products have no effect on health, or that they cause a small increase in mortality in elderly or vulnerable populations.[109][110] TimVickers 22:42, 16 April 2007 (UTC)

[edit] Alternative studies

  • malnourishment in children[15]
Yes, vitamin supplementation can treat vitamin deficiency, but this isn't what we're talking about.
From the first study cited "Our results do not provide strong support for population-wide implementation of high-dose antioxidant supplementation for the prevention of prostate cancer. However, vitamin E supplementation in male smokers and beta-carotene supplementation in men with low dietary beta-carotene intakes were associated with reduced risk of this disease." Second study is an associative study, which (as outlined above) does not prove A causes B
  • vascular disease[18]
Associative study.
  • immune deficiency[19]
Studied a multivitamin mix including B vitamins in HIV patients, cannot be described as a trial of antioxidants as B vitamins are not antioxidants. Moreover, HIV patients were not part of the JAMA meta-analysis so this study is not relevant to it's conclusions. The 2005 meta-analysis of these clinical trials concluded "There is no conclusive evidence at present to show that micronutrient supplementation effectively reduces morbidity and mortality among HIV-infected adults." link
Associative study.
  • age-related dementia[21]
Associative study.
Associative study.

These results have been reviewed in one of the reviews I linked to earlier:

Observational cohort studies have demonstrated an association between high intakes of micronutrients such as vitamin E, vitamin C, folic acid and beta-carotene, and lower risk of CHD, stroke and cancer at various sites. However, randomised intervention trials of micronutrient supplements have, to date, largely failed to show an improvement in clinical end points. The discordance between data from cohort studies and the results so far available from clinical trials remains to be explained. One reason may be that the complex mixture of micronutrients found, for example, in a diet high in fruit and vegetables may be more effective than large doses of a small number of micronutrients, and therefore that intervention studies that use single micronutrient supplements are unlikely to produce a lowering of disease risk. link

As the JAMA study examined the randomised controlled trials, rather than these cohort and associative studies, its conclusions are entirely consistent with the data above. It would be misleading to use data from associative studies, which cannot prove anything apart from A and B commonly appearing together, to dispute the results of the more reliable randomised controlled trials. TimVickers 22:44, 17 April 2007 (UTC)

[edit] Further negotiation?

Hi Paul, since you haven't edited this page for a week and said on Moralis' talk page that you were stepping back from this process and making your final points to him personally, I thought that you had decided to end this mediation. If you are still interested in discussing this, please say so. All the best, TimVickers 14:39, 20 April 2007 (UTC)

This is Moralis' mediation, not yours. I'll respond to his evaluation and recommendations as the mediation process would require. Your comment above is further evidence of your pathetic ego getting in the way of constructive collaboration.
Further, the sections you've written above (Definition of terms, Alternative studies) to dissect "associative" studies is conspicuous manipulation so you can make your independent point. The publications I cited are FDA-approved, government-financed, peer-assessed clinical trials published in respected journals -- in other words, the highest scientific standards. There's nothing you can say to detract from their conclusions or importance in the antioxidant field.
Your statement returned to the article without input from Moralis -- Overall, the large number of clinical trials carried out on antioxidant supplements suggest that either these products have no effect on health, or that they cause a small increase in mortality in elderly or vulnerable populations. -- is blatantly false and an ignorant disservice to the published record. --Paul144 14:08, 21 April 2007 (UTC)

Paul, that was completely out of line. I refuse to mediate a dispute where you aren't interested in assuming good faith. Tim's ego doesn't come into play here in any way. Your disappearance from this page led to a perfectly legitimate concern that you'd given up on the process, so of course Tim had to ask. It was exactly the right response.

This is not "my" mediation. It's your mediation. My job is to facilitate, not to offer a judgement, and my input is not required to make changes in proposals. If this were successful, there'd be a point where I would no longer be necessary at all- instead you're asking me to be a crutch, and that's not what I'm here for.

If you're not interested in reaching a conclusion, we can't move forward, period. --Moralis (talk) 16:10, 21 April 2007 (UTC)

To dissect associative studies from random placebo-controlled studies is a simple statement about methodology. Associative studies suggest that A and B might be connected, placebo-controlled trials test these predictions. Are you arguing that correlation proves causation? The sentence you criticise is the summary of three sources, link link and the JAMA paper, so can hardly be described as contrary to the public record.
Rather than making personal attacks, it would be more productive if we concentrated on producing a compromise version of the section in question. I have made several large changes from my previous versions in the proposal above, perhaps you could show some flexibility in return? TimVickers 16:14, 21 April 2007 (UTC)