These inconsistencies led the reviewers from the University of Ioannina, Greece, to question why the benefits reported for the nutrients in observational studies were still being perpetuated in the literature despite trials contradicting the efficacy.
The analysis fans the flames of debate over the efficacy of antioxidant supplements for the prevention of chronic disease, based on the disagreement between epidemiological/ observational studies and randomised clinical trials (RCTs).
The analysis, published in the Journal of the American Medical Association, states: "Citations to the 2 highly cited observational studies proposing an association of vitamin E with reduced cardiovascular events became less favourable over time, as contradicting data from randomized trials accumulated "Nevertheless, despite the eventual accumulation of strongly refuting evidence, even in 2005, half of the articles citing these epidemiological studies were still favourable to the vitamin E claim."
The same situation was observed for beta-carotene, said the authors.
"In 2006 more than half of the articles citing the highly cited epidemiologic articles on beta-carotene for cancer prevention... remained favourable for these interventions. "For beta-carotene, after a decade had passed from the contradiction of its effectiveness, counterarguments were uncommon: citing articles simply did not mention the contradicting trials."
Pertaining to two highly-cited epidemiological studies proposing major heart health benefits for vitamin E in 1993, lead author Athina Tatsioni and co-workers reviewed 172 articles on vitamin E and cardiovascular health. The studies were rated as favourable, equivocal and unfavourable to the original 1993 studies. For beta-carotene, 16 articles relating to cancer prevention were reviewed. The sampled articles were classified as being published before, early and late after publication of the refuting RCTs.
Despite the publication of large, contradictory RCTs, the reviewers found that more than half the articles on vitamin E cited the observational studies favourably.
"Even among articles that cited the contradicting HOPE [RCT] trial, rather than the positive epidemiological studies, the majority in 2005 still could not conclude that vitamin E was ineffective," wrote the authors. Regarding beta-carotene, 62.5 per cent of the articles cited the observational findings favourably, while only 18.8 per cent viewed them unfavourably.
"Favourable citations to beta-carotene, long after evidence contradicted its effectiveness, did not consider the contradicting evidence," wrote Tatsioni.
The authors could not discern why the observational results still received positive citations, quoting "differential interpretation, inappropriate entrenchment of old information, lack of dissemination of newer data, or purposeful silencing of their existence," as possible reasons.
Offering an alternative view, Andrew Shao, vice president, scientific and regulatory affairs for the supplement industry trade association the Council for Responsible Nutrition, said yesterday: "The reality is that science doesn't always move forward - there is some back and forth - and while research may seem to contradict itself, that should not be interpreted to mean one type of study trumps another, particularly when different studies ask and answer different questions."
Shao continued: "Seemingly conflicting data can exist side by side, when one understands that not all studies are asking the same questions in the same populations. "The RCTs with negative results attempted to answer the question, 'can a supplemental nutrient treat or reverse a disease or a lifetime of unhealthy habits in patients who are also taking prescription medications?'
"The observational studies with positive results attempted to answer the question, 'if we start with a mostly healthy population generally free of disease, can we identify various diet/nutrient and lifestyle factors that make them more or less prone to disease?'
"These are very different questions," he said, "making the studies incompatible for direct comparison and demonstrating that one type of study can't necessarily be used to refute the other."
"We should put studies into the appropriate context and evaluate the total body of evidence, which includes RCTs and observational studies, and other types of research. We hope that the publication of this analysis will encourage all of those within the scientific community to become more involved in this discussion," concluded Shao.
Source: Journal of the American Medical Association
December 5, 2007, Volume 298, Number 21, Pages 2517-2526
"Persistence of Contradicted Claims in the Literature"
Authors: A. Tatsioni, N.G. Bonitsis, J.P.A. Ioannidis
Readers' comments "I would like to react on the statements from Shao in the latest NutraIngredients. I fully agree with his statement that we should put studies into the appropriate context. In my opinion this means that we have to take the important differences between different study designs into account. Shao seem to put cohort studies on the same quality level as RCT's. However, the major problem with cohort studies is the residual confounding. A beneficial effect of a diet rich in antioxidants, may be simply explained by other healthy aspects of that diet or lifestyle related to that diet for which you can't completely control. I do agree that some of the major trials were done in people with already developed risk. But in this respect, I would like to point out that the more recent SU.VI.MAX trial (Hercberg et al Arch Intern Med. 2004;164:2335-2342) using nutritional doses of antioxidants in a healthy population could not find a beneficial effect on heart disease either, and only a modest prevention of cancer in men, but not in women. The claimed beneficial effects of antioxidants therefore remain at the most still unproven." Dr Louise Mennen Mennen Training & Consultancy The Netherlands
"With great interest I follow the discussion about the usefulness of antioxidant vitamins. Unfortunately, most scientists to not realize that conclusions from RCI trials are on a different level than the predictions from epidemiological studies; as matter of fact, they do not necessarily contradict each other. Epidemiological studies predict a lower risk of disease if your vitamin intake is in the range of the RDA. RCI trials show that vitamins do not offer a benefit if taken on top of the RDA. [There are examples of] plasma levels of the placebo groups of some RCI trials that are… in the range where epidemiological studies predict a lower risk. Thus, epidemiology does by no means predict a further significant risk reduction if you belong already to the highest or second highest quintile. If somebody would like to prove the prediction of epidemiology, one should select people corresponding to the lowest quintile, but this has never been done." Dr. Ulrich Moser Basel
"Important to take into account while evaluating the efficacy of vitamin E is the genotyping of the subjects in RCT's. The HOPE-studies did not show significant results, because the study population was 'diluted' with subjects who were less prone to oxidative stress and therefore at less risk for cardiovascular disease (the homozygote Hp 1-1 genotype). This contrary to the Hp 2-2 genotype as has been shown by Levy AP and co-workers (Arterioscler Thromb Vasc Biol. 2007 Nov 21; [Epub ahead of print]). They concluded that vitamin E supplementation (400 IU) appears to reduce cardiovascular events in individuals with DM and the Hp 2-2 genotype. In my view, the current knowledge about genetics (and epigenetics) urges us to include the genotyping into study-designs as a very important variable and will most probably influence the outcomes of RCT's." Dr. Gert Schuitemaker Ortho Institute, Netherlands