Paul Glasziou: Still no evidence for homeopathy

When the National Health and Medical Research Council report on homeopathy concluded that “There was no reliable evidence from research in humans that homeopathy was effective for treating the range of health conditions considered” few in conventional medicine were surprised, but the homeopathy community were outraged. As chair of the working party which produced the report I was simply relieved that the arduous journey of sifting and synthesising the evidence was at an end. I had begun the journey with an “I don’t know” attitude, curious about whether this unlikely treatment could ever work. Still, who would have believed that bacteria caused peptic ulcers, or that vaccines for cancers would become routine. So just maybe . . . but I lost interest after looking at the 57 systematic reviews (on 68 conditions), which contained 176 individual studies, and finding no discernible convincing effects beyond placebo.

Of course, with 176 trials we would expect a few p-values under 5% just by chance: 1/20 of 176 is about 9, which luck would class as “statistically significant.” So we relied on replication and systematic reviews to avoid such false positives. The NHMRC did not redo all 63 systematic reviews (which at say $50,000 each would have cost over $3M), but appraised the existing reviews and used them as a window on the body of evidence. Though that body was mixed in size and quality, no clear signal of effectiveness emerged from the higher quality studies.

One surprise to me was the range of conditions that homeopathy had been evaluated in, including rheumatoid arthritis, radiodermatitis, stomatitis (inflammation of the mouth) due to chemotherapy, and human immunodeficiency virus (HIV) infection. What subsequently shocked me more was that organizations promote homeopathy for infectious conditions, such as AIDS in Africa or malaria. Given the current effective treatments, that seems a very dubious activity, and is another example that justifies the NHMRC statement that “People who choose homeopathy may put their health at risk if they reject or delay treatments for which there is good evidence for safety and effectiveness.”

Unsurprisingly, there has been considerable pushback from those who use or sell homeopathic remedies. Indeed, the International Council for Homeopathy is currently leading a fundraising effort: not to fund better research, but to attack the NHMRC document. I can well understand why Samuel Hahnemann—the founder of homeopathy—was dissatisfied with the state of 18th century medicine’s practices, such as blood-letting and purging, and tried to find a better alternative. But I would guess he would be disappointed by the collective failure of homeopathy to carry on his innovative investigations, but instead continue to pursue a therapeutic dead-end.

Competing interests: While I chaired the NHMRC report on homeopathy, the contents of this blog are my own thoughts, and not those of the NHMRC or other committee members.

Paul Glasziou is professor of evidence based medicine at Bond University and a part time general practitioner.

  • Acleron

    The homeopaths have had long enough to produce reliable, convincing evidence. Should more trials of it be permitted? Treating patients with nothing appears to be unethical.

  • Dana Ullman

    What this BMJ article conveniently failed to report and what the Australian government’s press material failed to acknowledge was that ANY study that was less than 150 subjects was deemed “inadequate” by this report, and thus, the dozens of studies that have shown the efficacy of homeopathy in treating many ailments were totally thrown out and ignored, including many “high quality” randomized double-blind, placebo controlled trials that have been published in The Lancet, BMJ, Cancer, Pediatrics, Chest, Rheumatology, Pediatrics Infectious Disease Journal, British Journal of Clinical Pharmacology, European Journal of Pediatrics, and many others!

    Further, unless there were at least three studies conducted by three separate groups of researchers, with each study having over 150 subjects, the results were deemed to be “unreliable.”

    By not acknowledging these arbitrary guidelines, the BMJ and the Australian government are showing “bad faith” and are purposefully seeking to misinform the medical community and the general public.

    Based on these definitions of what “adequate” and “reliable” research, the vast majority of conventional drugs on the market today would also be deemed to be INEFFECTIVE and UNPROVEN.

    In fact, when the BMJ’s “Clinical Evidence” analyzed common medical treatments to evaluate which are supported by sufficient reliable evidence, they reviewed approximately 3,000 treatments and found only 11% were found to be beneficial (1). It should be noted and emphasized that the BMJ deemed 20 subjects to be a more reasonable guideline (2). If using the similar guidelines as the Australian government, only between 1% and 5% of medical treatments would be deemed to be “effective,” and virtually every surgical procedure would be consider “unproven.” Is Glasziou or the BMJ asserting that virtually all of medical treatment and surgical procedure be deemed unproven and ineffective? If not, then why use unrealistic and arbitrary guidelines for evaluating homeopathy? Are some extremely serious biases in evidence here instead of good science?

    Also, Paul Glasziou doesn’t seem to understand the real implications of his assertion that any study that has a P-value of .05 would suggest that this treatment had a 5% chance of occurring by sheer random chance. Based on the BMJ’s review of clinical research in the entire field of medicine having less than 5% efficacy, it could easily be assumed that many of these studies may have happened by chance, thereby suggesting that there is virtually no evidence for the entire field of medical treatment.

    For the record, Dr. Glasziou has conveniently ignored the many studies testing homeopathic treatment that has significantly better than a p-value of .05. Chest published a study on the homeopathic treatment of people with COPD with a p-value of 0.0001 (3). David Reilly and his team at the University of Glasgow conducted a series of four studies on patients with various types of respiratory allergies, two of which were published in the BMJ and one in the Lancet. Although their studies included over 200 patients, no single study included more than 150 patients, and therefore, ALL of the evidence from these high-quality trials were completely ignored, even though a review of the four trials found a p-value of 0.0007 (4). Even an editorial reviewing Reilly’s research has acknowledged that it is highly unlikely that these results are due to random happenstance (5).

    Or wasn’t it convenient that the Australian government’s report ignored a study on the homeopathic treatment of people with pancreatic cancer that showed that 39% of patients with this extremely serious chronic illness survived five years (6), even though no other study has ever found a five-year survival rate of greater than 1%. For the record, this study was not even considered by Dr. Glasziou’s report because it reviewed only 44 patients and was not a placebo-controlled trial, and yet, I challenge Dr. Glasziou or anyone to report results anywhere that can be comparable.

    Courts of law do not determine guilty or innocence only based on double-blind or placebo controlled trials. They report on all evidence.


    (1) What conclusions have Clinical Evidence drawn about what works, what doesn’t based on randomised controlled trial evidence? BMJ, 2015.

    (2) Nuts, bolts, and tiny little screws: how Clinical Evidence works. BMJ, 2015.

    (3) Frass, M, Dielacher, C, Linkesch, M, et al. Influence of potassium dichromate on tracheal secretions in critically ill patients, Chest, March, 2005;127:936-941.

    (4) Taylor, MA, Reilly, D, Llewellyn-Jones, RH, et al., Randomised controlled trial of homoeopathy versus placebo in perennial allergic rhinitis with overview of four trial Series, BMJ, August 19, 2000, 321:471-476.

    (5) This week in the BMJ. Homoeopathic dilutions may be better than placebo. BMJ 2000;321:0.

    (6) Chatterjee A, Biswas J, Chatterjee A, Bhattacharya S, Mukhopadhyay B, Mandal S. Psorinum therapy in treating stomach, gall bladder, pancreatic, and liver cancers: a prospective clinical study. Evid Based Complement Alternat Med. 2011;2011:724743. An abstract of the above study was published in the Journal of Clinical Oncology

  • Simon_Chapman

    Prof Glaziou, I’ve been washing a homeopathy bottle every day for the last month, but the residue just keeps on getting stronger. Any advice?

  • Permitted, yes. Funded by serious sources? No.

  • Dana, this comes across as you arguing for poor quality studies to be included.

    Is that actually what you’re saying? If not, could you explain your reasoning for including the poor quality studies a bit better.

  • Geeta Arora

    these are facts which Dana is brought out and should be considered. innumerable results from homeopathy, the work by dr jayesh bellare, the comments by nobel laurette Dr. Luc Montagnier all speak in favour of homeopathy ..

  • Yet all are overwhelmed by the evidence against it actually working.

  • Scientists still need to discuss this because there’s a vocal minority arguing very loudly for this to be integrated into actual medicine.

  • From that point of view homeopathy should be awash in funding.

  • There are ethical issues with *using* placebos for treatment.

  • They can’t. They can only work on an honesty system.

  • Kylie Turner

    I find comments like yours Guy Chapman quite disturbing as a representation of the level of commentary in discussions on a reputable medium like BMJ. Rather than actually address the issues which Dana has raised in his comment using reasonable scientific and statistical argument, (which he himself does), you, (and many others here I note) resort to personalised and emotive opinion statements. Most sadly though, is that you think these are sufficient enough to refute his arguments. How scientific is that? It seems more like web astro turfing than valid scientific discussion to me, and I think there should be more moderation regarding comments such as yours. “Dunghill of useless studies”? Now thats a scientific statement if I’ve ever heard one.However I am prepared to have an open mind and if you could point me to this dunghill that homeopaths use to bolster their beliefs, and also provide me with why you believe them to be useless I would be happy to consider your opinion. As should all with an open, enquiring scientific mind.

  • Acleron

    The idea that only alternatives to medicine are holistic are ludicrous and only promoted by non medics.

    Protocols to test homeopathic materials prescribed on an individual basis are readily available. Perhaps you can explain why homeopaths fail to use them.

    Many of the poor quality trials positive for homeopathy use the same prescription for all verum patients. This point only appears to bother homeopaths when the result is negative. This is cherry picking.

    Arguments of popularity and antiquity are known to be fallacious and misleading, yet these are the main plank of homeopathy marketing.

  • Acleron

    Appointing a skeptic who requires evidence and logic is a bad thing?

  • Anežka Marie Sokol

    There is work done in this field. This research shows that one can discriminate a homeopathic remedy (Stan. Met. D30) from control:

    Yet unpublished results shows that this experiment can be repeated under changing outer conditions.

  • Everyone who has had a nice drink of clear water has tried homeopathy.

  • chays

    One point no one has seemed to make is that, in homeopathy, for any one symptom, there can be anywhere from two to two hundred possible remedies. The only way to find the right one for a given patient and his/ her ailment is to cross reference many varying symptoms having to do with anything from how the patient sleeps to what he/she is afraid of or where his/her acne typically appears. No serious homeopath gives a remedy systematically for a given ailment. Therefor there’s no way a classic scietific study could demonstrate much of anything when it comes the efficiency of homeopathy. It would take finding a whole new way of approaching and understanding health.

  • Puddin’Head

    How do I face palm on this thing? Dana, your tirade of logical fallacies, special pleading, and misrepresentation of your so called “science” got old several years ago. The problems with the “research” you present are endless. The vast majority of studies into homeopathy are statistically under-powered and prone to false positive outcomes as per a simple Bayesian analysis. Even reviews such as Linde’s(1), which look at “high quality” research, end up including mainly under-powered studies (due to the lack of anything better) and still find no clear evidence of any effect. Let’s also not forget that, even when studies into homeopathy manage to look like actual science – enough so to get published by journals such as Nature – a simple inquiry into the methods reveals that it’s all a matter of smoke, mirrors, and gross incompetence (ahem…Benveniste)(2).

    As per your standard approach of an argument by bamboozlement, you reference a “study” supposedly treating pancreatic cancer which was actually not a study at all, but rather an open label trial with no control group. You challenge anyone to present comparable findings using any other intervention, but ignore the fact that this paper does not establish any therapeutic effect of homeopathy since it isn’t a study of anything. Your challenge is no different than estimating the average height of a man by measuring the LA Lakers and then challenging statistics that are actually representative of the general population to match that number.
    Given no plausible mechanism of action for homeopathy, it is just as likely that the outcomes of that paper can be attributed to the color of the patients’ hair or the intervention of magic fairies as they can to your dilutions and shaking.

    You reference the reviews of “Clinical Evidence” and come to the conclusion that most standard medical interventions don’t work, yet you ignore that your linked article states that the reviews “…are not an audit of the extent to which treatments are used in practice.”, meaning that these are not all mainstays of standard medical treatment and, in fact, are often studies of interventions which do not get adopted into standard medical practice. What’s more, many of the failed studies cited by the Clinical Evidence article are of “alternative” medicine treatments and not real medical interventions at all!

    I have dug through your tirades on HuffPo and elsewhere and come up with little more than appeals to authority, appeals to
    celebrity, misrepresentation of pseudoscience and garbage, and flat out fits of pouting. I have dug through just about
    every article you have ever promoted and have found glaring flaws in nearly every case. At the end of the day, 1000
    x 0 = 0, which is to say it doesn’t matter how many garbage studies posing as science you dig up, the aggregate of a flawed process is a flawed outcome. It leaves one to wonder – are you trying to be intentionally deceptive, or do you simply not understand what you are reading? You do the world a great disservice, Dana, by your persistent promotion of this absurd nonsense. Either learn how to read and assess the
    evidence or just give it a rest man.

    (1) Linde et al, Impact of Study Quality on Outcome in Placebo-Controlled Trials of Homeopathy, J. Clin. Epidem., Volume 52, Issue 7, July 1999, p. 631

    (2) Maddox, J. et al, “High-dilution” experiments a delusion., Nature, Volume 334, July 1988, p. 287

  • Puddin’Head

    Sorry Helen, but that Systematic Reviews paper used the same tired old pool of trials that several other reviews and meta-analyses have found to be replete with low quality data. The bulk are under-powered or otherwise grossly flawed. The fact that Linde and others have pointed out that controlling for study quality makes positive outcomes in homeopathy research disappear supplants any review that uses that same low quality pool of studies.

    You suggest that “Thousands of people are successfully treated by homeopathy”, but the evidence from properly designed and controlled studies says otherwise. A more accurate statement would be that “several thousand people report feeling better after being exposed to homeopathy”. Without any plausible mechanism of action or convincing evidence to suggest any sort of actual effect from homeopathy, this more appropriate observation is explained by the all mighty placebo effect.

    Finally, regardless of the cost, spending money on patent nonsense makes absolutely no economic sense.

  • Puddin’Head

    The explanation is that, along with statistical noise, the rest of that 41% you cite are poorly designed, improperly analyzed, statistically under-powered, or otherwise defined by such glaring flaws as to not qualify as science and to be utterly unconvincing.

  • lxmzhg

    Where is the evidence to prove that homeopathy was debunked time and time again? Don’t overwhelm me with a barrage of studies, show me just one, the best among them. If I can debunk that study or whatever it is you propose, then everything else goes down like dominoes… because if the best can’t cut the mustard, then how can anything else… right? Let’s see what you have?

  • Acleron

    Clinical trials to test the individualised treatment can and have been devised, there is nothing special about homeopathy in this respect.

    It is strange that homeopaths promote as proof those trials which are not individualised. Can we conclude that all the authors of such trials are not serious homeopaths?

  • Acleron

    Less than 6% of Indians trust homeopathy.

  • Dr. Shaji Kudiyat

    How do you know that? If so, it is confirmed that 6% trust homeopathy? How many percentage of Australians trust homeopathy?

  • Can you let us all know when these results are published?

  • “Large studies don’t ensure lower likelihood of bias or error. ”

    Agreed. That’s why we go for good quality studies. Good quality large studies do however carry more weight and do tend more towards the homeopathy is placebo conclusion.

  • Acleron

    A common complaint of homeopaths is that clinical trials are not individualised as is their supposed practice. This ignores the manufacturers who are prepared to sell water/sugar for disease states, the homeopaths who do likewise and the promoted clinical trials that do support homeopathy although they are generally of appallingly low quality.

    However, there is an area of homeopathy that is not individualised, that of ‘proving’. The new product is given to usually 6-8 individuals who are ‘normal’ and the symptoms that are produced are used to assess which patients should receive it. A study by homeopaths should be able to identify which materials are used in a blinded study. This has been suggested several times. It would be relatively cheap, easy to set up and quite definitive. This would answer all of the homeopaths objections.

  • This is why it is important that the NHS stops funding homeopathy: people (not just you, although you’ve made the point very clearly) seem to think it’s a popularity contest, not evidence-based science.

    The NHS needs to lead the way in dispensing with the dispensing of placebos!

  • Anežka Marie Sokol

    Sure, I hope it will start some constructive debate 🙂 It is a rather comprehensive study, so there will be published 3-5 articles, some on the methodological developments, some on the repetition and some on new results.

  • Vasiliy Vlassov

    Yes. In a honest system, when people are trusted, homepaths can declare any thing

  • Sebastian Muñiz

    Purification in the sun core is the only option, as I see it. 😛

  • Baptiste ROUSSEL

    “So we relied on replication and systematic reviews to avoid such false positives.” What is replication?

  • Jody Deane

    Hi Dana. Thanks for your positive contribution to the debate. Could you cite the urls of these studies? Many will find this invaluable. Thanks in advance.

  • Subbarao Nandikolla

    We don’t find even a single molecule in the preparations of homeopathic drugs. These drug preparations cannot be measured with the latest equipments now available.German Homeopathic Pharmacopoeia (HAB). The dilution 1:100 is
    containing a maximum of 1% of the original plant material. even in the total extract many other non pharmacological ingredients like chlorophill plant waxes and other plant pigments etc. were present..where the is a minute quantity of active ingredients ., of Atropine or Hyoscyamine etc. were found. So it can be imagined how much active principles were found. All experimental data on animal studies gives no results (in vitro and in vivo). Double blind clinical studies in humans there is some placebo effects. According to the laws of chemistry they will not practically work as therapeutic agents.
    N. Subba Rao M.Pharm.,
    Asst. Professor in Pharmacology( Retd.)
    Kakatiya Medical College , WARANGAL(Telangana State) , INDIA.