Letters

Randomised controlled trials for homoeopathy

Providers have much to gain from homoeopathy being accepted

Studies comparing homoeopathy and placebo are unhelpful

Language is being distorted

Study is in effect trying to compare apples with oranges

When is useful improvement a waste of time? Double positive paradox of negative trials

Authors' reply

Providers have much to gain from homoeopathy being accepted

EDITORDebates generated by the lack of evidence of benefit of screening for breast and prostate cancer are similar to the argument for lack of effectiveness of homoeopathy. How does one show a lack of benefit in the face of wishful thinking by both customers and providers, particularly when the providers have a financial interest in a positive outcome? And, as an anaesthetist and non-user of any of these services, why should I care?

The answer is given by Feder and Katz, who point out that patients can be referred to homoeopathic specialists in the NHS or general practitioners who prescribe homoeopathically.¹ Consumers of these resources should have a duty to show a benefit from any consultations or treatments provided, and the fundholders should have a duty to withhold funds in the absence of such evidence. If the customers then wish to purchase treatment that is not beneficial they will of course be free to do so and the homoeopaths, iridologists, colonic irrigationists, and the rest can flourish.

As the editorial points out, meta-analysis of results of trials of homoeopathic treatments suggest benefit greater than placebo in some cases, but this can easily be discounted by publication bias. Statistically, some trials of homoeopathic treatments will show some benefit even when the real effect of the treatment is zero, and these are the trials that are most likely to be published.

Homoeopaths argue that their treatments are not amenable to statistical analysis in the same way as conventional medicine is, in part because most trials of homoeopathic medicines do not individualise treatment. I can give an example from my own practice: anaesthesia for coronary artery surgery. In this, treatment is tailored to individuals, whose coronary anatomy is unique, yet statistical analysis shows an aggregate beneficial effect from treatment. This is how medicine works; patients are individuals yet patterns can be recognised, and this allows us to base our diagnostic and therapeutic interventions on these patterns.

Why should homoeopathy be allowed to hide behind a smokescreen that is denied to conventional medicine, and why should the NHS fund such treatment in the absence of any evidence of useful effect? The answer is that the providers have a lot to gain, in terms of financial reward and intellectual acceptance; quacks will always be with us.

Michael Foley, consultant anaesthetist. 
James Cook University Hospital, Middlesbrough TS4 3BW mikefoley{at}doctors.org.uk

Studies comparing homoeopathy and placebo are unhelpful

EDITORWhy do we need another study comparing homoeopathy with placebo?¹ The word placebo is a language construct for an apparently inert product, just as the word homoeopathy is a language construct for treatment with an apparently active product. We know that both may have positive effects on an illness, and occasionally one of the interventions will be more effective than the other. But when both are studied against an effective active treatment no difference is found between them. ^{2 3}

Clinicians must decide whether to prescribe effective active treatment that is significantly better than placebo or homoeopathy. For mild illness, where the active treatment leads to small benefits or high risks, a placebo or homoeopathy can be chosen. For severe illness, where the active treatment leads to large benefits, the active treatment can be chosen. There will be a grey zone.

Our debate should really be about the extent of this grey zone. It is certainly different in Australia from that in the United Kingdom, where five times more general practitioners are involved with homoeopathy. ^{4 5} Why is this? Differences in health funding? Availability and costs of active treatments? Cultural differences? These are some of the questions worth studying, rather than spending our resources on further trials of homoeopathy versus placebo.

John Weiner, visiting allergist. 
St Vincent's Hospital, Fitzroy, Victoria 3065, Australia jmweiner{at}allergynet.com.au

Language is being distorted

EDITORThe conclusion in the abstract to Lewith et al's paper states: "Homoeopathic immunotherapy is not effective in the treatment of patients with asthma."¹ The last sentence of the paper itself states: "In conclusion, in this double blind, randomised controlled trial of homoeopathic immunotherapy we have failed to confirm that this treatment is therapeutically efficacious in allergic asthma." These two conclusions are not the same.

The conclusion of the paper is a reasonable one to make. "In this trial we have failed to confirm" captures the point exactly. To extrapolate that to the conclusion quoted in the abstract suggests that no homoeopathic immunotherapy is ever effective in patients with asthma; this is not logical. What if a different potency or different frequency of dosing were to show a difference? This trial only used three doses of 30c over 24 hours. Many other different regimens are used in practice. This trial doesn't show that those other regimens don't work.

Sadly, this loose use of language then generates front page headlines like the one on the cover of the 2 March issue, which says: "Homoeopathy for dust mite allergies? No, it's a waste of time." This further extrapolates from the conclusion of the abstract to claim that any use of homoeopathy in treating dust mite allergies is a waste of time. This is an even less defensible position.

So, from article to conclusion, to conclusion of the abstract, to front page headline we lose the truth and develop generalisations that are not only wrong but unscientific.

Robert W Leckridge, associate specialist. 
Glasgow Homoeopathic Hospital, Glasgow G12 0XQ bob.leckridge{at}virgin.net

Study is in effect trying to compare apples with oranges

EDITORHomoeopathy is a system of healing that has been around for over 200 years. It does not operate in the way that allopathic medicine does.

Allopaths treat symptoms. They say that if patient A has a certain set of symptoms then a specific drug will be used for treatment. All patients are treated on the basis of their symptom picture. Sometimes a history is taken to ensure that patients are not allergic to a certain drug; if they are, then another one will be substituted. But, generally, allopathy will treat symptoms rather than the whole person.

Homoeopathy treats the person. If a person has asthma a homoeopath will generally spend at least one hour on a first visit, getting an idea of the person's history and lifestyle. What is his/her diet like? What other problems has s/he been treated for in the past? Are there any emotional issues? Then, and only then, will a remedy be prescribed, based not on the symptoms that the patient has but on what the homoeopath thinks is causing those symptoms. Only then can someone truly be cured.

Lewith et al's study does not compare homoeopathic and allopathic treatments of asthma.¹ It sets out with the intention of proving that homoeopathy does not work. To accomplish this it is using non-homoeopathic treatments and methods.

I have never heard of a homoeopath using potentised allergens to treat an allergy, although possibly in some cases this might be an effective treatment. Certainly, if you took 242 people with asthma and treated them homoeopathically you might well find that they would require 242 different remedies.

How is it possible that something as ill constructed and ill conceived as this study could be published in the BMJ? Where were the peer reviewers when the paper was submitted? Does the BMJ have any homoeopaths as reviewers, or was the paper looked at by people who are as ignorant of homoeopathy as the authors obviously are?

When I read something of this nature in the BMJ I truly wonder what other studies make it through the peer review process for absolutely no good reason.

Meryl Dorey, national president. 
Australian Vaccination Network Inc, PO Box 177, Bangalow NSW 2479, Australia meryl{at}avn.org.au

When is useful improvement a waste of time? Double positive paradox of negative trials

EDITORLewith et al report a pilot study of the use of homoeopathic immunotherapy,¹ complementing but differing from Reilly et al's earlier study.²

The table accompanying this letter is published on bmj.com, as is a rapid response from me.³ The table lists some of the fundamental differences between the two trials that may have determined their different results, and my rapid response may also clarify matters. Lewith et al's data show day to day effectiveness but leave a question mark over the efficacy issue of "more than placebo." Taylor et al's series suggested efficacy more than placebo in homoeopathy but did not tackle effectiveness.⁴

In Reilly et al's trial only the active group improved; in Lewith et al's both the placebo and active groups improved. The cover of the 2 March issue of the BMJ has an italic headline saying that homoeopathic immunotherapy is a waste of time. But the graphs of subjective and objective variables in both groups in Lewith et al's paper show the effectiveness of homoeopathy in an everyday context for a common and serious disease. We have created a curious intellectual and clinical absurdity hereuseful improvement that is a waste of time.

Perhaps it is now time to ask of a negative study whether it is a "double positive" study (both groups improving) or a "double negative" study (neither group improving). This question is important because each scenario calls for different conclusions, debate, and treatment in systematic reviews. If we reject double positive trials without due care we risk the paradox of throwing out treatments with efficacy exactly because they also show high effectivenesswhen the real problem is that our studies lack the power to see the baby of efficacy in its bath water of "context/non-specific/placebo" effectiveness. In a double negative study there is neither baby nor bath water.

Lewith et al's study may well have obtained false negative results, finding it hard to show additional activity over excellent context induced improvementsrather like the failure of sertraline and St John's wort (Hypericum perforatum) to better placebo facilitated improvement in depression⁵; that study was another double positive study (or triple positive in that case) that has been reported in the media as yielding a negative result. The BMJ's classification of homoeopathy as a negative treatment is premature.

David Reilly, lead consultant physician. 
Glasgow Homoeopathic Hospital, Glasgow G12 OXQ davidreilly1{at}compuserve.com

   The table is published on bmj.com

Authors' reply

EDITORHomoeopathy arouses strong emotions among believers and non-believers, as the extent of responses to our article suggests; our letter encompasses the points made electronically and above.

Homeopathy is a "waste of time," states the front cover of the BMJ. This type of response to our scientific paper is inaccurate and unhelpful. We tested the model of homoeopathic immunotherapy as suggested and defined by Reilly et al and observed no clinical effect.¹ However, as pointed out in our paper, we recognise that this model is not generalisable into current homoeopathic practice. It was used to investigate the different effects of ultramolecular potencies compared with placebo rather than pragmatic homoeopathy.

Feder and Katz suggest that the mechanisms underpinning homoeopathy are best understood through laboratory experimentation. We believe that clinical investigations in humans offer insights that are impossible to achieve in a laboratory. We have suggested some hypotheses that might explain our observed oscillatory effects of homoeopathy, but these require further work²; they may be a type I error. We hope that Feder and Katz's suggestion about pragmatic studies does not represent a retreat from rigour.

One reason for conducting our study was that we were concerned by the small size and potential for statistical error and the atypical lack of improvement over time for placebo in the asthma study of Reilly et al.¹ Our study owes much to the design of this earlier work, but it was not a mere duplication. We rigorously rediagnosed asthma in each patient, and before the study we communicated with Reilly to check that we were following the same protocol in relation to inclusion criteria, dosage regimen, and primary outcome (visual analogue scale).^1-3 Reilly suggests that he used repeated doses of homoeopathic immunotherapy but does not report this in his asthma study.¹ We were also not aware that each patient was subjected to an individual case conference with, possibly, unreproducible inclusion criteria.¹

It has been suggested that a hospital based study would select a different population of patients from those in general practice. There is no evidence that this is the case; many high quality asthma effectiveness studies have been conducted in primary care. We selected a group of asthmatic patients in two neighbouring counties who had variable asthma so we could measure change. We cannot understand Reilly's hypothetical differentiation between an effectiveness and efficacy study in this context.

Like us, Taylor et al avoided the pollen season and used potencies of house dust mite while not avoiding the house dust mite season.⁴ The use of Reilly's visual analogue scale on alternate weeks avoided patient data recording fatigue and produced greater compliance over a 20 week study. Our process of randomisation entailed minimisation, thus balancing the treatment groups. Both nurses and patients were unable to guess trial allocation, therefore we are clear that our blinding allocation was completely secure. Had we used a different time point for analysis we might have obtained different (false positive) results.

We analysed response to homoeopathic immunotherapy over 4 months and as a consequence our conclusions are less open to misinterpretation. The only major methodological difference between our study and that of Reilly et al is the lack of a placebo run-in period; an unlikely cause of a significant difference in outcome.¹

Homoeopathic immunotherapy may be effective in rhinitis, ^{3 4} but independent investigations have failed to confirm this.⁵ Reilly may have overinterpreted one small such study on asthma. We accept that no two clinical trials are exactly the same, but we believe our study is comparable.¹ It comes to different conclusions about the differences between homoeopathy and placebo and about treatment effect, reporting that homoeopathy and placebo are significantly different. The robustness of these observations requires careful further work rather than more debate.

George T Lewith, senior research fellow. 
Complementary Medicine Research Unit, Mail Point OPH, Royal South Hants Hospital, Southampton SO14 0YG

Michael Hyland, professor of health psychology. 
Department of Psychology, University of Plymouth, Plymouth PL4 8AA

Stephen Holgate, clinical professor of immunopharmacology. 
Respiratory, Cell and Molecular Biology Division, Mailpoint 810, Southampton General Hospital, Southampton SO16 6YD