How homeopathy fails at science
After getting an opinion piece published online in the debate section of a major Norwegian newspaper (link, but in Norwegian), I’ve been having an interesting sort of dialog with the Twitter account of the Norwegian Association of Homeopaths (NHL). Now this is of course just one person, or at most a handful, but through the dialog and the links provided on their side of the discussion, I’ve made what I find are some interesting observations on the relationship the world of homeopathy has with science. It has ended up rather rambling, for which I apologise, but hey, it’s my personal blog, not the NEJM.
Let me first say that it’s entirely possible that any or all aspects of the practice of homeopathy are correct. “Like cures like” could actually be a near universal principle. Succussion could really be causing some sort of water memory. And the benefits perceived by homeopathic practitioners and satisfied patients could really be due to the treatment, and not due to a combination of various placebo effects and the stress-reducing effects of being listened to and taken seriously. Even in a strictly rational world view this is possible. It’s just not very likely. In fact, it barely registers on the “likely”-scale at all, so how do the homeopaths end up with such a strong conviction that it’s not only fact, but scientifically proven fact?
As I’ve written about before in this more narrow post, the father of homeopathy, Samuel Hahnemann, had good excuses for being wrong. He would have known very little about how hard it is to do accurate medical research, and his results would have been better than the competition at the time because many treatments in those day were actively harmful. Modern day homeopaths have no such excuse, and to hold on to their beliefs they operate with a rather bizarre relationship with modern science.
If you try to look into the scientific backing of homeopathy, you’ll probably find lists such as Medical conditions with positive conclusions for homeopathy from the European Committee for Homeopathy with a bunch of references to research. This was the first link offered me in my discussion with the NHL. It’s a hodgepodge of references of varying significance and validity, and even though it supposedly only includes positive conclusions, one of the links is to a Cochrane review on Oscillococcinum as a remedy against influenza which a) concludes that he evidence is insufficient to draw a conclusion, and b) is from 2006 and was updated in 2009, 2012 and 2015. Neither review really changes the conclusion, so it’s baffling that this list hasn’t been updated or that NHL would use the 2006 reference in work done in 2013 as they do here (link in Norwegian). But it does show that to the homeopaths, not having been proven wrong is as good as having been proven right.
Now I’m not going to discuss how systematic reviews done by homeopaths and the medical science community differs, except to mention that the former consider the body of research proof of the efficacy of homeopathy and that medical science have the opposite conclusion. When questioned by my twitter opponent what my qualifications were for criticising homeopathy, I referenced the British NHS as an example of expertise I trusted who’ve recently rejected homeopathy.
Instead of discussing the quality of such reviews, which are frankly beyond my level of expertise, my focus will instead be on how homeopathy appears to only be interested in science to confirm already existing assumptions, and lacks the system or inclination to take the consequences of negative results. Normal medical research regularly leads to both new and established practices being discarded as inefficient or harmful, while finding an example of this happening in homeopathy seems rather difficult. So I asked the NHL-twitter jockey, I asked the teeming millions on the Straight Dope message boards (Are there any homeopathic remedies that have been discontinued due to negative research results?) and I did some searching myself. Results: silch. So basically homeopathy appears immune to negative research results, admittedly based on a rather limited search. Or it could be that homeopathy is actually flawless and all negative and null results are invalid, but that seems rather unlikely in the face of so much research.
When confronted with this question my NHL opponent moved to a favoured argument of homeopaths. Homeopathy is individualized. Every treatment is different, based on how the patient responds, and testing that double blind is just impossible. Thus there doesn’t exist negative results that are relevant to changing practice. Never mind that they’re touting lots of studies without individualization as evidence, setting up a double blind test of individualized homeopathy is actually rather simple. I offered up this study on asthma as an example, but apparently conventional asthma medicines perfectly nulls out homeopathy in addition to actually working. And when I found a paper where that wasn’t an issue, a trial where patients with generalized anxiety disorder improved greatly and equally whether in the placebo group or receiving homepathic “drugs”, the response was that then they really should try homeopathy before SSRIs. In and of itself damning evidence of at least one homeopath’s understanding of science, but let’s keep going.
Curious as to whether this twitter account was really busy or I was the special guest of the week I checked its feed and immediately found an interesting reference. Touted as Homeopaths recognize homeopathic remedy in blinded proving my opponent had linked to this Protocol for a phase 1 homeopathic drug proving trial. Protocols being evidence of nothing I followed the one link to an actual trial citing the protocol. The result of that trial: no difference between homeopathic remedy and placebo. My opponent’s response was to dig into a single detail that could be interpreted as positive for homeopathy, totally ignoring that the results conclusively falsified the hypothesis or that the trial only had 29 participants. In their view the results were “interesting”.
Now in case you didn’t know it, homeopathic practice is built on a system of research that’s supremely vulnerable to all the pitfalls of bias that modern research protocols are designed to counter. Preliminary tests of a new remedy is called a proving, and the aforementioned trial was an adaptation of this to a more stringent double blind approach. And the intention of promoting the link in the first place had been to put forth the citations in the protocol suggestion, not the protocol itself or the resulting research. Which is no surprise as the research following the protocol were damning to homeopathy while the results inspiring it were spectacular.
In fact they were the kind of results that would convince me homeopathy was real, if they were independently verified and follow-up research hadn’t given null-results. As reported in Homeopathic Pathogenetic Trials Produce Specific Symptoms Different from Placebo:
- 8 participants on homeopathic remedy A showed an average of 6 (+/-2) symptoms specific to A, and 0 (+/- 2) specific to B or non-specific.
- 10 participants on B had a similar 5 (+/-2) symptoms specific to B, and 0 others
- 7 participants on placebo had 11 non-specific and zero belonging to A or B.
Ignoring the horror of including the negative side of the y-axis on a graph of the number of symptoms one can only see these results as amazing. (Can you really have an average of 0 +/- 2 symptoms? That seems like statistics malfeasance to me.) It’s a small trial, but if one judges from it alone, homeopathic remedies are not only perfectly aligned with those old not-so-rigid provings, they are also perfect protection against the nocebo effect. All you’d have to do to upturn physics as we know it would have been to replicate the findings. Which hasn’t been done, or at least not successfully. Which doesn’t matter to homeopaths. Which is proof positive that they are scientifically illiterate.
If one takes the homeopathic approach to science, homeopathy seems well founded. There’s is positive research out there, and research that is inconclusive, and, according to homepaths, very few negative results. But homeopathy and homeopathic research as a whole appears unable to take the inconclusive and negative results to hearth and see how damning they are to a practice that claims spectacular results in actual use. And when one looks into the cites and details of their reviews … Yes, there are some positive results included, often low quality studies, and if you bother you can easily find someone (The Research Evidence base for Homeopathy) claiming as “broadly positive” a review that concludes thusly:
There is some evidence that homeopathic treatments are more effective than placebo; however, the strength of this evidence is low because of the low methodological quality of the trials. Studies of high methodological quality were more likely to be negative than the lower quality studies. Further high quality studies are needed to confirm these results.
When discussing the evidence for homeopathy one can debate the weight given to different types of evidence and what research is selected to be included in reviews, and unless one is personally qualified to assess such things homeopathy can seem to have as much claim to have proven itself as the medical community has to the claim that they’ve disproven it. And so homeopathy does claim that the evidence is just as strong for homeopathy as for conventional medicine, or would be, if they had as much money for research as conventional medicince, as they do in this attempted rebuttal of the strawman “There is no scientific evidence that homeopathy works” (same link as the first link in the previous paragraph). But even if one accepted that claim, and after looking at the selection criteria I certainly don’t, there is one major difference homeopathy can’t escape. In actual medical science inconclusive and negative results lead to further research or to changes in practice, in theories, and in research avenues, and positive results are replicated, whereas in the bizarro homeopathy version results are only taken into account when they confirm the deeply flawed provings of the last 200+ years and otherwise mostly ignored.