Homeopathy, placebos, and the protests of Bruce Barwell
He has protested a comment of mine in the following blog entry:
Lancet research labelled biased
After quoting from published excerpt from a broadcast:
"Homeopathic Society President Bruce Barwell says the researchers used a very small sample, probably too small to base a sound judgement on, and there was also bias in the selection of homeopathic remedies they researched." http://www.newstalkzb.co.nz/newsdetail1.asp?storyID=79005
"Not only does Bruce Barwell speak for the Homeopathic Society of New Zealand as its President, he doesn't hesitate to misrepresent the facts (dare I say "lie"?). I'll let readers make their own decision. Denial is a personal attitude, where one lies to oneself, but he goes further.....hmmph!"
His reponse to my blog entry
He chose to comment on the blog entry in question:
Bruce Barwell (email@example.com) wrote the following:
"I have the distiction of being described as a liar in your blog on the subject of the Lancet and homoeopathy. You err in saying the assesssment was of 110 trials, it was in fact 8, of poor quality. It would have helped your case if you had actually read the Lancet piece. If you give me a postal address I can send you evidence supporting my claim.
Best wishes for your pursuit of truth,
As always, if I am mistaken, I appreciate being notified of my error, and will apologize if I am convinced of my error.
In this case I did not intend to actually call Barwell a liar, but to leave that decision up to readers. Now that I have read my remarks with new spectacles, I can see that my remarks could still be interpreted otherwise, for which I am sorry.
From the available press reports it appeared that he was claiming that only eight reports were studied, rather than 110. Since 110 studies were included, and only the best were chosen for the final evaluation, it ended up being eight homeopathic and six traditional medicine studies that ended up being compared to each other and to placebos.
If that's really what he was referring to, then I owe him an apology.
It certainly appeared that he was being deceptive by denying that it was 110 studies that were included in the meta-analysis. It also appeared that he was claiming that the authors of the study actually did original research themselves, where "the researchers used a very small sample,..... and there was also bias in the selection of homeopathic remedies they researched."
Well, it wasn't original research by the authors, nor of any specific remedies, but a meta-analysis of existing studies of all levels of quality. There is a big difference between researching "remedies" and analyzing "studies." That distinction is definitely not clear in Barwell's statement. He actually did use the word "researched."
To avoid confusion, Barwell should have said "studies" instead of "remedies," if that is what he really meant, for example:
"... there was also bias in the selection of homeopathic *studies* they researched."
Now if the written notice on the website has misquoted Barwell, then he should take that up with the editors who wrote it, because they have then done him a misservice, leading to misunderstanding.
Sorting of studies in meta-analyses
I certainly hope the authors used some bias in their selection of studies to include in the final phase!
Naturally the authors excluded studies of poor methodological quality in the final phase of their study. That's the way such studies are normally done. Studies are analyzed for methodological quality - not by results - and poor quality studies (without blinding, few participants, clearly prejudiced, one-sided, sales arguments, etc.) aren't used in the final analysis. Only those of high methodological quality - regardless of results - are included in the final analysis.
Serious allegations of research fraud
It's interesting that he claims the eight studies were "of poor quality."
If he can prove that the authors of the Lancet study actually had the audacity to compare homeopathic studies of "low quality" with traditional medicine studies of high quality, then we have a clear case of research fraud being perpetrated by the authors of the study. I haven't heard of any such allegations as yet (merely charges of "bias"). This would be enough to hit the front page of the New York Times, the Indian Express, and Times of India. The study was widely referred in media all over the world, and any serious charges against it would likewise be spread far and wide, especially in the homeopathic literature.
A challenge for Barwell
I challenge Bruce Barwell to:
1. Prove his allegations that the eight homeopathic studies were "of poor quality."
2. Prove that homeopathy has any significant effect above the placebo effect.
3. Prove that there is any *properly blinded* method of establishing the difference between
a. a homeopathically diluted and potentiated solution (water or alcohol) with an active ingredient, and
b. an identical solution that is likewise diluted and potentiated, but without the original active ingredient.
(I'm referring to truly homeopathic dilutions, not merely slightly diluted preparations which still have measurable quantities of molecules from the original "active" substance. Homeopathic "active ingredients" are oxymorons, since they must be diluted to the point of non existence, before they are truly "homeopathic".....;-)
I'd really like to see what he comes up with! I'm certainly willing to learn.
James Randi would certainly like to see his answer as well, considering he has put up one million dollars in his Million Dollar Challenge:
I'll let readers come to their own conclusions about the depth of Barwell's understanding of what constitutes "quality" studies and research.
If he's representative at all of typical homeopathic thinking - after all, he's the President of the Homeopathic Society of New Zealand - then his idea of quality homeopathic research is likely based on the final (favorable) results, not the methodological quality of the process leading to those results.
By this "reasoning" (not!), homeopaths automatically assume and claim that if research doesn't conclude that homeopathy "works", it must be of poor quality.
I contest that it is their "reasoning" that is of poor quality!
History has shown that homeopaths and legitimate scientists have very different understandings regarding research "quality," with the progress of scientific knowledge being conditioned on the use of research that seeks to distinguish the difference between *real* effects and *placebo* effects.
Homeopathy and placebo effects
In stark contrast, homeopathy continues its stagnation (and stubborn refusal to accept irrefutable evidence in large quantities and of high quality) by refusing to accept that mere placebo effects aren't good enough, especially since they produce no long-lasting physiologic effects of real worth on real biological illnesses, and only temporarily fool the mind with subjective feelings of improvement. Using them can thus have potentially dangerous consequences. The current methods of promoting and selling them is seriously unethical and should be illegal. One should not be allowed to deceive the public in this way.
Wikipedia's article on the Placebo Effect contains the following, to which I have contributed:
Objective or subjective effects?
An alternate opinion attributes the false perception of a placebo effect to the fact that patients who have been given a placebo report improvement earlier and more eagerly in order to please and thank the care giver. These patients may even do this when there is no real physical improvement attained. One quoted figure is that about one third of patients improve on a placebo, but a recent study has called that number into question (A. Hróbjartsson & P. C. Götzsche, 2001), claiming that the effect is much smaller, if it exists at all. The 30 percent figure derives from a paper by Henry Beecher, published in 1955 (H. Beecher, 1955). Beecher was one of the leading advocates of the need to evaluate treatments by means of double-blind trials and this helps to explain why it has been so widely quoted.
The Hróbjartsson & Götzsche study demonstrated that in many studies where a control group was used that did not get any treatment at all, the effects in the no-treatment group were almost equal to the effects in the placebo group for studies with binary outcomes (e.g. well treated or poorly treated). The authors concluded that the placebo effect does not have "powerful clinical effects," and conceded that placebos have "possible small benefits in studies with continuous subjective outcomes and for the treatment of pain." They therefore concluded that there was no justification for its use outside of clinical trials. 
In a follow-up study in 2004, the same authors were able to confirm their previous results and concluded: "We found no evidence of a generally large effect of placebo interventions. A possible small effect on patient-reported continuous outcomes, especially pain, could not be clearly distinguished from bias". 
If their conclusions are correct, the placebo effect is reduced to a subjective placebo illusion, while retaining its importance as a statistical research tool. As such it is imperative to use it in research, but unethical to use it in normal clinical treatment of patients.
These conclusions contradict what some would now consider to be a great deal of folklore that has evolved around the whole idea of the placebo effect. That folklore has evolved in a "research vacuum" of ignorance about the true nature of the placebo effect.
What is new about these conclusions is an emphasis on the key words "subjective" and "pain". This explains the well-established fact that the placebo effect is most "effective" in conditions where subjective factors are very prominent or significant parts of the problem. Some of these conditions are: headache, stomach ache, asthma, allergy, tension, and especially the most subjective of them all - pain, which is a significant part of most serious (and many mild) illnesses.
It also explains why there is no conclusive documentation for placebos causing significant healing effects in serious biological pathologies.
Practical implications and consequences
According to these findings, a placebo can make you think you are better, and even temporarily feel that you are better, but it can't actually make you any better. It will not cause any significant physiological change in a serious disease. In short, it will fool you (which is its intended function in double blind experiments).
To most scientists, these conclusions aren't revolutionary, since they have been using placebos in research for years, based on assumptions that this was the case. If they had believed otherwise, they would have been acting against better knowledge.
Much quackery achieves temporary "success" by a conscious or unconscious misuse of this placebo illusion. To the patient, such misuse of placebos can be expensive and ultimately fatal. To the quack, it will fool the patient long enough to keep the scam rolling.
Reciprocal Links: An Invitation