Edzard Ernst discovers flaws can lurk under the surface of even the most rigorous scientific studies.
I doubt many British GPs will know what homotoxocology is. Put simply, it is a form of homeopathy that currently seems to be gaining popularity.
Homotoxicology was developed by the German physician Hans Reckeweg (1905-1985) and is strongly influenced by homoeopathy; in fact, its remedies are diluted in the homeopathic way. Homotoxicology does not, however, subscribe to the ‘like cures like' principle, as homeopathy does.
According to the concepts of homotoxicology, human diseases are the result of toxins, which originate either from within the body itself or from its environment. Proponents of homotoxicology believe that each disease process is thought to run through six specific phases and is the expression of the body's attempt to cope with these toxins.
Diseases are thus viewed as biologically useful and necessary defence mechanisms. Health, on the other hand, is the expression of the absence of toxins in the body. These assumptions are not based on science and bear no relationship to accepted principles of toxicology or therapeutics.
Several years ago, we published a systematic review of all RCTs testing the efficacy homotoxicology [1]. We were able to include seven RCTs; three of which tested the efficacy of Traumeel. This remedy is given after major or minor trauma and is by far the best known homotoxicological remedy in the UK.
All RCTs were of good methodological quality and all suggested that the tested remedies are superior to placebo. But there were a number of caveats that made alarm bells ring for me.
In several RCTs, at least one author was an employee of the manufacturer; yet in most articles no conflicts of interest were declared. Many articles were published in a journal closely linked to, perhaps even owned by the manufacturer.
Our suspicion of sponsor bias was strengthened by correspondence with one manufacturer. They told us that financial support for our research project, i.e. the review [1], would only be given after they had inspected (and, by implication, agreed with) our results. None of the included RCTs reported having an independent monitor which, of course, would have been the normal and adequate measure to minimise this type of bias.
In several trials, de-blinding of the patient was a real possibility. Generally speaking, it is advisable to check blinding rather than to assume it. Other studies could have generated a false-positive result due to their small sample size. In two studies, it was possible that the ‘placebo' caused a worsening of symptoms thus producing significant inter-group differences in the absence of improvements in the verum groups.
This project has taught me two lessons: 1) Even formally rigorous studies, i.e. those that score highly on validated instruments for assessing bias, can be dodgy; 2) Sometimes one needs to do quite a bit of detective work to notice all the potential weaknesses of apparently rigorous trials.
Commentators to (not just) this blog have often stressed that trials sponsored by big pharma are sometimes less than reliable. I quite agree, we need to be vigilant. But this brief discussion suggests that conflicts of interest and other well-hidden flaws might also play an unfortunate role in some CAM research.
Click here for more from Edzard Ernst.Professor Edzard Ernst
