Prof. Manfred Wilhelm, biostatistician, Hochschule Ulm (University of Applied Sciences), Germany
“Adequate intake of micronutrients, such as vitamins, minerals, and trace elements, is essential for human health. This has been demonstrated by a variety of studies. Meta-analyses summarize the results of many different studies and thus draw upon a broad base of data. However, a conclusive summary can only be compiled if similar conditions exist in the individual studies involved.
The recent review by Bjelakovic et al. (1) does not meet this fundamental require-ment for scientific validity. For example, of the 615 initial individual studies on antioxidants, the authors excluded 537 of these from their further analysis, including all the studies in which no deaths occurred. Only the remaining 78 studies (13 percent) were analyzed, i.e., the studies in which deaths were recorded. Yet only 25 of the 78 studies were designed as mortality studies. Nevertheless, studies in which mortality was not systematically examined – i.e., studies that should have only been used to generate hypotheses (to be tested in further studies) – were also included in the mortality considerations. This selection error represents a significant methodological flaw that has led to biased results.
Above all, each of the studies in the meta-analysis comprised extremely different dosages and observation periods (i.e., ranging between 28 days and 12 years), and are thus not comparable as such. For example, the vitamin E dosage varied from 10 to 5,000 IU in the individual studies. This is methodologically inadmis-sible and therefore does not constitute a basis for general conclusions. Despite being based on such dispa-rate, raw data, this meta-analysis derives a not more than 3% higher risk of death from antioxidant supple-mentation – and that using a statistical model with fixed effects. There are, however, no significant results derived using a model with random effects, which, given the heterogeneity of the studies, would be of greater significance. In this respect, drawing general statements about, or even basic warnings against, dietary supplementation with micronutrients from this study is not vindicable.”