Biased Cochrane study

Not only are the calculations based on generalizations leading to nonsense; they also presume a couple of adroit misrepresentations. The worst thing, however, is that what was actually written in The Lancet is completely different from what the public was told.
Four scientists at the acknowledged Cochrane institution have startled the world: In The Lancet, they claim that antioxidants are dangerous and that an estimated 9000 out of one million people taking antioxidants will die! How long before death ensues is not stated, but presumably it will be relatively quickly. Many more will die, therefore, if this abuse of vitamins continues - this even includes completely ordinary dosages.
Unfortunately, this is not only a frighteningly incorrect message. In order to reach this result, the scientists - who actually only statistically combine a number of old studies - have had to make so many dodges that it looks as though they would rather twist the facts rather than create enlightenment. If this is the case, it is incredibly disgraceful.
The statement below only involves the statement of excessive mortality. In the article, this statement is based on first disregarding two large studies (including the largest one of all) in which mortality was reduced - the reason for this being that they were of "low quality". Then, the best of the remaining seven high quality studies was disregarded because it is relatively small and "makes the material heterogenous". This study also indicated reduced mortality. However, another even smaller study was not removed. In this one, mortality was increased!
Misrepresentations on TV
After these comfortable adjustments, speaking with great precaution in The Lancet is understandable: "These antioxidants seem to be able to increase the total mortality", but "selenium could be an exception which can potentially reduce the risk of gastro-intestinal cancer", they say.
Oddly enough, this exception was totally forgotten when the public was informed. The information was: "There is an excessive mortality in the people who were given antioxidant vitamins. And this excessive mortality was particularly observed in the best trials." Head doctor Christian Gluud, one of the four authors, said on the news on TV that even normal doses of vitamin E and vitamin C are dangerous - a statement that is utterly unfounded. It would seem that all reservations had been completely forgotten.
What are the objections to the statement? Let us start by looking at the objections in the seemingly courteous but really completely scathing official commentary in the same issue of The Lancet. This was written by two statisticians, Altman and Forman, one of which (David Forman) is the manager of the British Cochrane institute.
Abuse of the Cochrane institution?
First of all, the results must be regarded as temporary, it says in the commentary. There are a whole number of other vitamin studies with information on mortality that are not represented here. Secondly, it is vital to a Cochrane study that the total knowledge is being taken into consideration without leaving anything out. This is their raison d'être.
In other words, they are surprised that the excessive mortality calculations have been made at all - they do not belong in the context of Cochrane studies!
Without miscalculations, no statistical certainty would exist
It gets even worse, though. According to the commentary, the result is not even statistically certain if another more correct statistical method is used instead of the one they used in the article! A so-called random effects model that considers the differences between the compared studies should have been used instead of the fixed effects model.
All results are based on a single abnormal study
Moreover, the result can be exclusively attributed to a single study that differed from the other five used in the calculations in several ways. The aim of the study was to treat heavy smokers and people who had been exposed to asbestos with monstrously large doses of vitamin A (25,000 I.U. a day) which are regarded as toxic in itself, plus large amounts of betacarotene which is dangerous to smokers if it is not supplemented with antioxidants such as selenium, vitamin E, etc. (see below)
If a study were to be disregarded, this should be the one as it so clearly separates itself from the rest. If this had been done, no excessive mortality would have been found.
The two commentators also stated that the claim of excessive mortality "is not convincing". In plain English, the meaning of this is that it is simply wrong!
The head doctor exaggerated
So - not only did the four scientists have to manipulate the numbers, including removing a couple of valuable results and including another, in order to achieve the (desired?) negative result. They also had to evade the actual Cochrane idea which is to make conclusions based on all available and sober knowledge. Yet, they could only make conclusions with reservations.
It can be no surprise, therefore, that the result is "not convincing" according to the official commentary. Colloquially, this means that it is unreliable!
Head doctor Christian Gluud was aware of these fierce commentaries when interviewed on TV. He knew that, at best, the results were only preliminary; or exploring as the article says. He was not the least bit bothered by this, however, as he did not only disregard the criticism but also went far beyond the conclusions that The Lancet had permitted him to make in the article.
Everything about this case is indicative of what is known as scientific dishonesty, so what made him go this far?
Good becomes bad
Unfortunately, Gluud were also aware of many other things when he so eagerly sent for the media - he just did not mention them! Here are a few examples:
A small study in the high quality group involved people infected with helicobacter pylori (the so-called gastric ulcer bacteria) in their stomach. These bacteria are known for being able to cause stomach cancer, that is if they do not cause gastric ulcer instead. In this relatively small study (485 participants), the mortality rate was quadrupled among the ones who were given antioxidants! This sounds bad, but read on:
For some reason, nobody in the placebo group died during this trial while two people died in each of the three different antioxidant groups. Technically, this can be calculated into the mentioned quadrupled excessive mortality.
There was another group, however, with which the comparison could have been made. In this group, nobody were given antioxidants, but a strong antibiotic cure during the first two weeks of the six year study. For the rest of the six years, they were not given any treatment, and now to the important part: In this group, two people also died! Had this been the group to which the antioxidant groups had been compared, then no excessive mortality would have existed. This option was not chosen, however.
Furthermore, the special thing about this study is that the antioxidants seemed to be absolutely brilliant at protecting against pre-cancerous changes in the stomach. The study was actually initiated in order to examine this very thing.
It was not at all geared to examine mortality, but owing to the particular use of the numbers in the Cochrane study, this good message was transformed into an excessive mortality of 400%. In The Lancet, the study was turned into something negative although, in reality, it was a study with a very positive outcome.
Questionable exclusion of trials
It was different picture with regards to the low quality studies - they were excluded from the calculations but they displayed a strong tendency to reduced mortality. There was one including almost 30,000 Chinese people who were given selenium, vitamin E, and betacarotene in order to see if it would prevent cancer and improve survival. This trial was the largest one of all.
However, the authors had neglected to explain how they had randomly distributed the participants into the different groups who received various supplements. For this reason, the study was rejected as being "low quality". It is noteworthy that no doubt was made as to the distribution being random - as it should be. They had just not described how this distribution was made! All other rules had been followed and reported.
The study lasted for five years. After a couple of years, the mortality rate started to decline - gradually it settled on 9% while the cancer risk was reduced by 13%. In the other groups nothing happened. This large study and its result is surely reliable as it reflects reality. However, as a consequence of the rigid Cochrane method, it was banished from Society and the results were not included. Had the results been included, the conclusion would have been different. Bureaucratically, it may have been fair enough to classify the study as "low quality", but what about the truth?
Even by using the accepted Cochrane method, there is the risk of achieving the opposite of what you want - that is if it is the truth you are after! We cannot do without a combined, rational estimate in which numbers are set off and scientists of all people should know this. Numbers from positive studies can be made to seem negative just as good studies can be made so seem bad.
But it has obviously not been taken into account that someone has been somewhat manipulating the results in order to get the calculations to fit.
Basic confusion of ideas
As if we have not heard enough, another problem has presented itself that is possibly even more disturbing: The generalized use of the word antioxidant! On TV, the distinct impression was given that Christian Gluud was hesitant about the term. He did not return a negative answer when asked whether the antioxidants in fruit and vegetables are as toxic as the ones in pills, but he added that fruit and vegetables contain "a large number of other substances that are likely to either repair the damages caused by the antioxidants or to completely neutralize them". Vitamin C, then, according to head doctor Christian Gluud, is a poison to be neutralized!
There were a number of other nasty gaps in Gluud's knowledge that we will overlook as the purpose of this article is only to point out that a Cochrane study is quite a bureaucratic enterprise. It is apparently just a matter of finding articles in the databases, finding certain number in these articles and then inserting the numbers into a spreadsheet, after which the program will do the rest. It is not necessary, therefore, to possess any knowledge of the subject as long as you can insert numbers into a statistics programme. If you want to interpret the numbers, however, you will naturally go wrong when lacking the necessary expert knowledge.
The mean of apples and pears
A biochemist would never refer to antioxidants as being a uniform group. He would know that there are various forms of vitamin E, that vitamin C has completely different effects from betacarotene, and that all three of these substances are far different from selenium, etc. You cannot compare treatments with vitamin E to those of betacarotene just as you cannot calculate the mean of apples and pears. This ought to be elementary, but still, the Cochrane study makes this very mistake.
Had there not been other points of complaint, this alone would have been decisive. Mixing up all antioxidant studies in one pile and then extracting a mean from this corresponds to mixing up all medical treatments of cardiac disease and describing them as pill treatments - in other words, pure nonsense.
We can now return to the commentary in The Lancet. Its closing statement says that if the four authors ever succeed in establishing an excessive mortality as a result of "antioxidants" they will have to find out which antioxidant(s) is the villain. It cannot possibly be all of them!
Ignoring betacarotene
In six of the seven studies called high quality, treatment consisted of betacarotene in combination with one or several other antioxidants. Smokers who take betacarotene are at risk because the tobacco smoke can transform the yellow pigment into free radicals in the lungs - which is the exact thing people use antioxidants to prevent.
In a Chinese study (Linxian), however, a 50% reduction in the risk of lung cancer was found when combining betacarotene with other antioxidants (selenium and vitamin E). Since 1994, it has been known that betacarotene taken as a single supplement (or in combination with a small supplement of synthetic vitamin E) can cause lung cancer in smokers and increase mortality.
For this reason, it would have been far more reasonable if the four authors had concluded that betacarotene seems to be able to increase mortality (in smokers who do not get an adequate amount of other antioxidants). The downside to this conclusion is only that it is old news and that slating all antioxidants simultaneously would not have been possible.
Deliberate omissions
The muddling of all antioxidants is suspicious for other reasons as well. When generally throwing down the gauntlet to all antioxidants, the reader will get the impression that all antioxidants have been examined. This, however, is not the case.
In the official commentary in The Lancet, attention is drawn to the fact that there are hardly any studies of the therapeutic correlation between vitamin C and cancer. That is surprising, they say, considering the many indications that vitamin C has a prophylactic effect in this area. When the four authors speak of "antioxidants" being harmful, they generalize about something of which they are completely ignorant. There is also highly inadequate knowledge of natural vitamin E (unlike the synthetic variant), flavonoids, lycopene, other carotenes than betacarotene, N-acetyl cysteine, allopurinol, etc.
However, one thing is speaking about something of which you do not know anything. Another thing is deliberately keeping something hidden. Despite the Lancet article expressly stating (with regret?) that selenium seems to be able to reduce both mortality and the risk of cancer quite substantially - the latter with great statistical certainty - this was not mentioned at all by Gluud - neither on TV nor in the more substantial interview in the Danish newspaper Information.
Why has it not been mentioned that such an important antioxidant seems to be able to reduce the risk of several forms of cancer by 50%? We can only venture conjectures.
Conclusion
The negative statements about antioxidants which are scientifically completely meaningless can only be traced back to studies of betacarotene and - as mentioned above - particularly to a single study that in almost every way separates itself from the rest. Furthermore, they have capered about the handling of the numbers and the selection of worthy studies and the rejection of others.
This triggers a serious suspicion about lacking scientific neutrality. And this is to put it mildly. It is even worse, however, that this suspicion is so violently strengthened by not just the absurdly incompetent but also the selective and tendentious way in which the results were presented to the public.
This study, and particularly its public presentation, casts an unhappy reflection on the Cochrane institution.