JAMA Antioxidant supplement study was very poorly done-hiding bias

[Guest guest]

Dear Arnold,

By the ANH team

5 March 2007

On 28th February we saw headlines around the world once again condemning

vitamin supplements. The stimulus? A Serbian doctor, Goran Bjelakovic, who was

involved in an earlier canning job on vitamins – in 2004 on vitamin

supplements for reducing risk of gastrointestinal cancers (Lancet 2004; 364:

1219–28) –

somehow found himself doing it all over again. This time he published in

the US-based Journal of the American Medical Association (JAMA). On both

occasions, his papers triggered headlines around the world which appeared to

have just one purpose: getting people to stop taking vitamin supplements.

Interestingly, in a bout of apparent schizophrenia for JAMA, Bjelakovic and

colleagues’ views are fundamentally opposed to those of Fairfield and

Fletcher published in the same journal some two years earlier (JAMA 2002;

287:3116-3126). Fairfield and Fletcher broke the long-standing anti-supplement

agenda

of the JAMA by supporting supplementation as a means of reducing risk of key

chronic diseases. But it seems it’s now back to business as usual for JAMA.

On the top of the Serbian’s hit list were the favourites: vitamin E and

beta-carotene. Dr Bjelakovic, given his previous work, appears to have a bit of

thing for antioxidant supplements which he considers to be beta-carotene,

vitamin A, vitamin C, vitamin E and selenium. From the point of view of any

informed scientist, this is a peculiarly narrow perspective on what is meant by

‘

antioxidant supplements’. First of all the vitamins in question are not

automatically antioxidants. In certain forms (especially as synthetic, isolated

forms) and dosages, they can actually have the reverse effect – and act as

pro-oxidants. Secondly, the dose and form of the vitamin are critical to

determining how the vitamin will behave in the body, as are the other nutrients

which

are consumed at the same time.

To give an example, high doses of vitamin E (the very forms that have swayed

Bjelakovic’s analysis) actually reduce the body’s absorption of the more

important antioxidant form of vitamin E, gamma-tocopherol, the predominant form

in foods and high quality vitamin E supplements.

It was seriously remiss of Bjelakovic and his team to not emphasise that; a)

the studies they used to condemn these vitamins were nearly all performed

using synthetic forms of the vitamins that behave in the body in remarkably

different ways to the natural forms and b) to not make clear the effects their

study selection approach would have on the final results.

Looking further at this second, crucially important point, Bjelakovic’s team

found 815 trials that were potentially relevant. But they culled out a

massive 747 (yes – a jumbo jet load!) of these trials, leaving just 8% of the

total number for the number crunching! The most important reason given by the

authors themselves for the exclusion of studies (responsible for 50% of the

exclusions – a total of 405 trials) was “morta_lity was 0 in both study

groups.â€

Consider what effect this might have on bias. If you remove 50 % of the

studies because they didn’t cause any increased risk of death – how can you

say

that vitamin supplements overall cause a 5% increase in risk of death….it

simply beggars belief that the JAMA can tolerate this type of science.

This is just one of the stunning problems with Bjelakovic’s study. Dr Steve

Hickey (of the ANH Expert Committee) and colleagues reveal more reasons for

the flawed nature of the Bjelakovic’s meta-analysis – click here _

(http://www.alliance-natural-health.org/index.cfm?action=news & ID=273) to view

their

rebuttal submitted and approved for publication in the Journal of

Orthomolecular

Medicine.

Bjelakovic’s specialty appears to be meta-analyses (statistical study of

studies). The application of a lot of statistics to a given data set does not

change the quality of the data set. In fact it can often magnify inherent

problems in the data. Bjelakovic’s data set is poorly selected and severely

compromised – and the results of his analyses do not provide any reflection

on the

risk of supplementing with vitamin A, beta-carotene, vitamin C, vitamin E or

selenium, the targets of the meta-analysis. In fact, the results tell you

absolutely nothing about taking, either the natural forms of these supplements,

or the effects of taking all these nutrients together, the common way they

are taken by most people – as multivitamin/mineral supplements.

Newspaper headlines on 28 February were as dire as “Supplements ‘raise

death rate by 5%’†(The Times, UK), “Vitamins ‘could shorten

lifespan’†(BBC

News, UK), “Des vitamines dangereuses pour la santé?†(Le Soir, France)

and “

Another knock on antioxidants†(Los Angeles Times, USA). But they may not

have had the impact that was hoped for by certain interest groups. Judging by

the vehement and often irrational hatred for vitamin supplements shown by

elements of the orthodox medical profession closely aligned with the

pharmaceutical industry, it seems likely that the main objective both of

Bjelakovic’s

meta-analysis and the resultant articles was to stimulate a turn-around in the

increasing numbers of people who are side-stepping pharmaceutical medicine in

their quest for health.

Just like all the other factors Bjelakovic and his colleagues failed to

consider, clarify or include in their meta-analysis, it seems that another

important factor has been ignored. And that’s the millions of people who have

derived benefits from taking supplements, combined with other aspects of a

healthy

lifestyle. These people will need a little more than a computer-generated,

reductionist, flawed analysis of past studies of pharmaceutical forms of

vitamins to put them off.

 

So will we.

Links:

_Journal of the American Medical Association_

(http://jama.ama-assn.org/cgi/content/abstract/297/8/842?maxtoshow= & HITS=10 & hits\

=10 & RESULTFORMAT= & fulltext=Bje

lakovic & searchid=1 & FIRSTINDEX=0 & resourcetype=HWCIT) , Bjelakovic et al

(http://www.timesonline.co.uk/tol/life_and_style/health/healthy_eating/article14\

4981

3.ece)

_Times online article, 28 Feb 2007, Supplements 'raise dealth rates by 5%' " _

(http://www.timesonline.co.uk/tol/life_and_style/health/healthy_eating/article

1449813.ece)

_BBC News_ (http://news.bbc.co.uk/1/hi/health/6399773.stm) , Vitamins 'could

shorten lifespan', 28 Feb 2007

(http://www.alliance-natural-health.org/_docs/ANHwebsiteDoc_270.pdf)

_Rebuttal by Dr Steve Hickey, ANH Scientific Expert Committee _

(http://www.alliance-natural-health.org/_docs/ANHwebsiteDoc_270.pdf)

To , please go to

_www.alliance-natural-health.org/index.cfm?action=unsub_

(http://www.alliance-natural-health.org/index.cfm?action=unsub)

<BR><BR><BR>**************************************<BR> AOL now offers free

email to everyone. Find out more about what's free from AOL at

http://www.aol.com.

 

 

 

[Guest guest]

There is a real problem with the study you mention. This study is a

meta-analysis it takes 63 studies and groups them together. When these studies

are all analyzed they show no benefit or harm for vitamin e or vitamin a . It

shows some benefit foe selenium and a little benefit for vitamin c. All these

benefits are in terms of mortality. Some of the studies they used where in

terminally ill people most others where on very sick people. Still after hand

selecting the studies they did not get the answer they apparently were looking

for. Then they divided the studies into low bias and high bias. How they did

this is not clear. what is clear is that the studies that these researchers

found high quality showed an increase in mortality.for vitamin a a just barely

significant increase in over all mortality for vitamin e and neutral result for

vitamin c The studies that these authors found low quality showed just the

opposite . It is most likely that this study shows the authors

bias . Most studies find what they are supposed to find.This one seemed to but

only after the data was " distilled " This may be the first step in the battle to

get government control for these thinks . Research codex