It’s Evidence Jim, But Not As We Know It!
Posted by apgaylard on November 8, 2007
A Commentary On In Pursuit Of Evidence By Kate Chatfield RSHom
The main reasons for my interest were that when I first came across this piece I was immediately struck by it’s misrepresentation of clinical evidence and the philosophy of science.
I claim no special skill in philosophy, but a second-year undergraduate course sparked an interest which has stayed with me. I may not be an expert, but I’ve read Popper and Lakatos. I also have a passing acquaintance with the works of Kuhn. So I think I know when someone is not telling it how it is.
Having engaged my attention, the commentary on this piece has ended up being rather long, so I’ve provided an index.
Philosophical Straw Men
Allopathic Medicine is Mean
Good Quality Evidence is Bad
You Mean We Have To Be Accountable?
Oh No, We Don’t Have Any Real Evidence
The ‘Whole Person’ Gambit
The Cherry Picking Gambit
It’s Evidence Jim, But Not As We Know It!
A Lamentable Lack of Lakatos
Clutching at Kuhn
We’re Not Wrong We’re Underdetermined
You Can’t Prove Anything, So We Are Right
The High-Dilution Problem Ignored
8 Out of 10 Cats Prefer it
So Let’s Have Customer Satisfaction Surveys Instead
We’re Not Wrong, We’re Incommensurable
We’re Taking Our Ball But Not Going Home
In Pursuit of Evidence seeks to justify a lower standard for evidence of homeopathy’s effectiveness than would be the case for conventional medicine. The justification for this is both pragmatic and philosophical.
The pragmatism comes from a perception that high quality evidence for homeopathic treatments is dismissed by prejudiced advocates of scientific medicine. Through the development of this line of thought the author’s lack of understanding of the statistical nature of these data emerges.
Philosophical arguments are used to make the case that the ‘paradigms’ of scientific medicine and homeopathy cannot be compared and homeopathy cannot be shown to be in error.
At various points you will see the author proclaim that homeopathy has been found to ‘work’ (better than a placebo) in a range of studies. This is, of course, true. However, Chatfield is sampling from one extreme end of the distribution of outcomes reported in the literature. This is what is commonly called cherry picking.
If a medical intervention of any sort is tested a number of times against a placebo and is actually ineffective the play of uncontrolled variables and random chance will lead to a distribution of results. If they are all reported and the trials have been fair, the average result will be that there is no difference between the intervention and placebo.
However, some will likely show the intervention to be less effective than the placebo, others more. A truly effective treatment will have an average effect that is in excess of that shown by the placebo and the distribution of trial outcomes will be shifted towards efficacy by a significant margin.
Now, not all trials are well designed; these are more likely to be biased. Some are small and subject to more statistical variation as a result. Not all negative results are published, adding a so-called publication bias to the literature. Many investigators may be testing the same treatment, further increasing the chances of finding some positive results. Given this, some positive trials of homeopathy are inevitable.
Ioannidis has neatly demonstrated that looking for statistical associations where the probability of their being a real relationship is low leads to an increased risk of spurious results.
This is why any honest examination of this topic must review the literature, not pick on a relatively small number of anomalous papers.
A corollary of this is that reviews need to be discriminating. Performing a meta-analysis on a good number of studies will provide an opportunity to reduce the play of random chance. It will not necessarily overcome biases built into the individual studies by poor trial design. It is no good complaining, as Chatfield does, that certain meta-analyses have removed poor quality studies from their ambit.
Next philosophy. Here Chatfield adopts a number of approaches. The first is to set up scientists as creatures slavishly bound by two limiting philosophies: Positivism and Inductivism. These straw men are pursued, ignoring the more appropriate descriptions of the philosophy of science. The idea of falsification, as advanced by Popper, is then presented as somehow undermining the notion that the progress of science can sort the wheat from the chaff. Finally Feyerabend and Kuhn are invoked to support a retreat to Relativism which is then re-labelled as Rationalism.
By following these rhetorical lines the author seeks to withdraw homeopathy from the scrutiny of science, whilst at the same time invoking some of its symbols to provide a cloak of respectability.
This culminates in the advocacy of a new kind of evidence: the sort that is not subject to scrutiny and is merely the servant of homeopathic advocates.
In the following commentary Chatfield’s writing is italicised.
“… Anyone who has been working with homeopathy in the UK, particularly in the past five years, must have noticed a steady increase in both the extent and depth of adverse publicity that homeopathy is receiving. Repeatedly the critics ask the same question: “Where is your evidence?” Consideration of this question first led me to become involved with homeopathy research, but increasingly I find that my background in philosophy compels me to challenge the very concepts of evidence, proof and science that drive the research agenda. What follows is an account of evidence-based medicine, some of the implications for homeopathy, and philosophical challenges for the foundations of our assumptions.
The practice of allopathic medicine has been subject to many trends, driven not only by developing scientific knowledge, but also in reaction to social, political and financial pressures. The introduction of evidence-based medicine (EBM) has been heralded as the most recent revolutionary phase, described as a ‘paradigm shift’ that will change medical practice for years ahead (Guyatt et al 2002).
Here we are introduced to the homeopath’s favourite pejorative term for scientific medicine: allopathy. Invented by Samuel Hahnemann, the originator of homeopathy, it’s used to perpetuate the slander that scientific medicine treats, or suppresses, symptoms whereas homeopathy treats the whole person and addresses underlying causes. To quote William T. Jarvis (emphasis is mine):
“… Opponents of medicine claim that they treat the underlying causes of disease, while [medical doctors] treat only the symptoms. Further, they claim that medicine suppresses the symptoms, thus interfering with the body’s inherent healing processes. A close examination reveals that this line of reasoning is only clever rhetoric. When they say the are treating the underlying causes, these vitalistic ideologists refer to a metaphysical life force rather than actual causes of disease such as viruses, bacteria, protozoa, genetic defects, radiation, chemical insult, and so forth … Homeopathy has always been based upon symptomatic relief …”
The author’s implication is that homeopathy is somehow immune to societal pressures. Odd for a ‘discipline’ that is clearly both a social and a commercial enterprise. This lazy implication runs through the tract; only scientific medicine has social pressures and philosophical problems!
In 1992 a group of physician-researchers, otherwise known as the Evidence Based Medicine Working Group (EBMWG), published an article urging physicians to base clinical decisions purely on the basis of evidence. In this way unreliable, intuitive judgements would be replaced with rational calculation and the use of research (EBMWG, 1992). Gone are the times when it was acceptable to base clinical decisions upon personal experience, discussions with peers, case reports, information from textbooks or intuition. Now we have a more reliable scientific method for decision-making that involves study and critical assessment of all available research that is relevant to our question.
Well that is certainly a good description of a laudable aim. Who really wants opinion-based medicine?
This notion of EBM is predicated upon the assumption that there is a hierarchy of evidence for medical interventions, such that certain types of evidence are deemed of greater value than others. At the top of this hierarchy are results from randomised controlled trials (RCTs) and systematic reviews or meta-analyses of these trials. A systematic review is a means of examining results from more than one trial to look for trends. A meta-analysis takes this overall view further by performing statistical analysis on the combined results to look at the statistical significance of these trends.
Certain types of evidence are not “deemed“ to be better than others, they are better. RCT‘s are only better because they reduce the effect of bias by randomising the allocation of patients to the treatment or control group. This works towards ensuring that these two groups are as similar as possible to enable fair comparisons to be made.
RCT’s are also better for the use of control groups. Without some patients receiving a placebo how is the effect of the specific intervention to be distinguished from the ritual of delivery? Control groups also, where ethically possible, allow the effect of the intervention to be separated from the natural powers of recovery of the human body.
Well executed Systematic reviews are better still because they look at many individual trails, thus reducing the influence of random chance.
The synthesis of large amounts of clinical trial data into manageable systematic reviews or meta-analyses is supposed to revolutionise medical practice and offer objective and politically transparent criteria for treatment choice and funding decisions. Proponents pride themselves on their attempt to ‘realign medicine with science’, whilst critics object that EBM emphasises exclusively the science of medicine whilst denying completely the art of medical practice, (Miettinen, 2001).
The “art of medical practise” is an interesting term. Doesn’t that just mean a combination of individual experience and prejudices?
It’s a red herring anyway. What does homeopathy have to do with this art? Most of its practitioners are not medical practitioners. This is particularly the case of the membership of the SoH; These are non-medical homeopaths.
Implications for Homeopathy
The call for an evidence base is not limited to allopathic medicine and there is growing pressure on all forms of complementary and alternative medicine (CAM) to provide the kind of evidence that is deemed of high value from RCTs and reviews of these trials.
The House of Lords, in its report on CAM, suggested that it is desirable to obtain evidence about the efficacy, safety, cost-effectiveness and mechanisms of these therapies, before their use is advocated (House of Lords, 2000).
And why not?
The requirement for evidence has practical implications for the availability of CAM. NHS healthcare providers require evidence before considering CAM services (Prince of Wales’s Foundation for Integrated Health, 2003).
This seems very reasonable. I certainly would not want to be treated using therapies that were not supported by an evidence base. As a tax payer in the UK I certainly object to funding un-evidenced therapies. As a point of ethics shouldn’t CAM practitioners have a duty to make sure that what they sell works?
In 1997, a London health authority decided to stop paying for homeopathic treatment after concluding that there was not enough evidence to support its use. The Lambeth, Southwark and Lewisham Health Authority had been referring more than 500 patients per year to the Royal Homeopathic Hospital in London (Wise, 1997).
Common sense prevails. How come I have an uneasy feeling that the author is not too happy about this? Why should the public purse support ineffective therapies, especially when effective ones exist and sometimes struggle for funds?
Finding the Evidence
The two main resources for evidence-based practice in the UK are Bandolier (http://www.jr2.ox.ac.uk/Bandolier/), written by Oxford scientists, and the Cochrane Library (http://www.cochrane.co.uk/en/clibintro.htm). Both carry easily accessible information about an extensive range of medical conditions and the research results. A brief examination of the Cochrane Library quickly highlights the vast disparity between both the amount and the types of research evidence that are available for allopathic and homeopathic interventions. For example under ‘rheumatoid arthritis’ we find a large amount of research evidence for each particular kind of allopathic intervention reviewed separately: 13 different drugs, different forms of exercise, different forms of electrotherapy and so on. Whereas, across the whole site, homeopathy is only reviewed under four areas in total: asthma, dementia, induction of labour, and the use of Oscillococcinum for prevention and treatment of influenza.
Things have moved on a little, there are now five negative reviews available from the Cochrane Library. Here is a list of these reviews, along with their ‘plain English’ summaries. (emphasis is mine).
“… This review aimed to assess the evidence for homeopathy as an intervention for attention deficit/hyperactivity disorder. Four trials were retrieved and assessed with mixed results. Overall the results of this review found no evidence of effectiveness for homeopathy for the global symptoms, core symptoms or related outcomes of attention deficit/hyperactivity disorder …”
“… There are different types that may be used for asthma, such as classical homeopathy (tailored to an individual’s symptoms) or isopathy (for example using a dilution of an agent that causes an allergy, such as pollen). The review of trials found that the type of homeopathy varied between the studies, that the study designs used in the trials were varied and that no strong evidence existed that usual forms of homeopathy for asthma are effective. There has been only a limited attempt to measure a ‘package of care’ effect (i.e., the effect of the medication as well as the consultation, which is considered a vital part of individualised homeopathic practice). Until stronger evidence exists for the use of homeopathy in the treatment of asthma, we are unable to make recommendations about homeopathic treatment. …”
“… No evidence that homeopathy is effective in treating dementia … The researchers did not find any good quality trials and so cannot say whether it is or is not effective for treating this condition. As no information is available on how much homeopathy is used for dementia, it is difficult to say whether it is important to conduct more trials.”
“… Homoeopathic Oscillococcinum does not prevent influenza but might shorten the length of the illness … Oscillococcinum is a homoeopathic preparation manufactured from wild duck heart and liver (common sources of influenza). It is claimed that Oscillococcinum (or similar homeopathic medicines) can be taken either regularly over the winter months to prevent influenza or as a treatment. Trials do not show that homoeopathic Oscillococcinum can prevent influenza. However, taking homoeopathic Oscillococcinum once you have influenza might shorten the illness, but more research is needed. …”
“… There is not enough evidence to show the effect of homoeopathy for inducing labour.”
No wonder the author doesn’t like the evidence. Chatfield has missed other useful sources, namely The NHS’ “Complementary and Alternative Medicine Specialist Library” and the mostly historical James Lind Library.
Typically, where homeopathy has been subject to systematic review for a particular condition such as asthma, all kinds of homeopathy are lumped in together. Isopathy, therapeutic prescribing, combination-remedy treatment, individualised prescribing, and so on, are all considered under the label of homeopathy. Hence it is immediately apparent that homeopathy is not reviewed in an equivalent way to allopathy. We would not find a systematic review of allopathy, including each kind of allopathic intervention for the treatment of asthma, because it would be considered meaningless. Whereas there are so few trials of homeopathy for the treatment of asthma that they are all banded together regardless of whether or not they in any way reflect what happens in the real world.
What sort of homeopathy works then? Isn’t it up to homeopaths to tell us? Could the SoH let us know what the good and bad sorts of homeopathy are? If they all work, then lumping them together should not be a problem.
Also, isopathy is not really homeopathy is it? It breaks with the so-called ‘law of similars’ by using causative agents (i.e. not similar but the same).
There have of course been many more research trials than are mentioned in either the Cochrane Library or on Bandolier, but research in homeopathy is so badly under-resourced that there is still only a tiny fraction of the amount that would be necessary to make sensible comparisons with trials of allopathic interventions.
Homeopaths continue to proudly regale us with the 200 year history of their subject. I would have thought that in all this time they could have sorted out some decent research protocols. In fact real medical practitioners have been asking them to do this since the 1830′s.
Homeopathy also has sufficient resources to take forward its own research programme. Just note what the SoH has recently said about homeopathy.
“… with thousands of people seeing a practitioner and a £38 million pound spend on homeopathy predicted for this year …”
“… The homeopathy market for 2007 is estimated to be worth £38m and projected to reach £46m in 2012. “Homeopathy is showing steady growth and this is likely to continue“. (Mintel, Complementary Medicines, April 2007)”
Here’s a quote with a more European focus from Homeopathy Today (accessed 1/11/2007)
“… France, with over 300 million euros, is the largest homeopathy market in the world, followed by Germany (200 million euros). 40% of the French have already been treated with homeopathy*, and 74% of the patients stated that they are “inclined to follow a homeopathic treatment if prescribed by their doctor“. …”
It’s a bit much for homeopathy advocates to plead both poverty and prosperity, when the circumstances suit them.
Historically the CAM research community has attempted to address the call for evidence and many of its members have undertaken the challenge to provide the gold-standard RCT research that the EBM movement demands. Experience of this has taught us many important lessons including the following:
It’s not just the EBM movement. Mainstream medicine and fair minded individuals have been asking for good quality trials since 1835.
• That homeopathy, as a whole person treatment, does not lend itself well to testing through the placebo-controlled, randomised trial, and many researchers have begun to look for alternative methodologies that are more suited to researching this complex intervention (Weatherley-Jones et al, 2004).
This is a kop-out. It has been shown that placebos can be administered and patients assigned to test and control groups randomly and still have a nice chat with the homeopath.
Let’s also remember as the ‘whole person’ defence is trotted out what the guiding principles of homeopathy are. Hahnemann‘s homeopathy contends that:
Medicines are to be chosen on the basis of the patient’s symptoms, without reference to the supposed disease process underlying them. For Hahnemann, the symptoms are the disease, and once they have gone the disease is cured.
Medicines must be chosen for the similarity of their effects to the symptoms of the patient. This ‘similimum principle’ is of course the kernel of the homeopathic method.
Medicines are to be repeated only when recurrence of the patient’s symptoms indicates the need.
Anthony Campbell, Homeopathy In Perspective: Myth And Reality (2004) p22.
This, I think, dispels the myth of homeopathic holism. Homeopathy is a treatment of symptoms. Hahnemann was opposed to the search for causes. “Medicines” are selected and doses administered on the basis of symptoms.
• That even when trials have produced positive results they are dismissed as flukes, or subject to researcher bias. For example, David Taylor Reilly and colleagues conducted a series of trials, over a number of years, in patients with hay fever, asthma and perennial rhinitis that demonstrated a significant difference between the placebo and homeopathic groups. The important thing about these trials was that the results were reproducible and yet they are still not considered firm evidence. (Reilly et al, 1986; Reilly et al 1994; Taylor et al, 2000).
This is not just limited to homeopathy. If a good methodology is not followed any trial can be biased. Biases are reproducible. In fact true replication needs to be done by different investigators. Chatfield’s example is mere repetition.
A real replication was attempted by Lewith et al, the result? The authors concluded that “…Homoeopathic immunotherapy is not effective in the treatment of patients with asthma.” Why was this attempt at replication giving a different (negative) result? Feder and Katz suggested “…duration of treatment … although chance is another explanation …”
In contrast to what Chatfield would have us believe this work is not reproducible. This was demonstrated in 2002.
As previously discussed, when inferring statistical significance a standard is almost universally adopted that gives rise to a one-in-twenty chance that the result is a fluke. That’s why systematic reviews have been invented, to take a view on the body of evidence to allow the signal to be separated from the statistical noise.
What We Mean When We Speak of Evidence
The current notion of EBM is fundamentally based upon the assumption that evidence can provide us with ‘facts’ about the world and that scientific beliefs stand or fall in the light of this evidence. This is however an over-simplistic understanding of the nature of evidence that philosophically would be described as explicitly ‘positivist’.
Not necessarily. The Positivists were, perhaps, the first modern philosophical movement in science. There is no doubt that there is some Posivitist influence in science. However, there are other philosophical positions which accept the existence of an underlying reality but do not take an explicitly Positivist view (for example: Critical Realists, Objectivists).
Perhaps a practical point is important here. The tradition of measuring, comparing and evaluating has reaped large dividends in science in general and medicine in particular. To argue against it one really needs to provide something better, rather than just calling it “simplistic“.
Positivists believe that scientific questions can be settled in a purely objective manner and that experimental testing allows the scientist to compare his theory directly with the facts. They claim it is possible and essential for the inquirer to adopt a distant, non-interactive posture such that values and other biasing, and confounding factors are thereby automatically excluded from influencing the outcomes. In this way science provides us with a sure route to the truth.
Here the straw man steps forward. This is a description of a simplistic philosophical stance that does not characterise contemporary science.
Now, I’d like to paint a more realistic picture of science from a philosophical perspective. It clearly exposes the banality of Chatfield’s straw man.
Kuhn was perhaps the first to focus on science as a social enterprise. Lakatos built on this as well as Popper‘s falsificationism to show how scientists generally work within research programmes, rather than with individual hypotheses or theories. Within Lakatosian research programmes relatively immature, but key, ‘hard core’ theories are protected from premature falsification by a protective belt of hypotheses that can be modified or discarded without sacrificing the hard core. This gives research programmes, and their theoretical cores, the chance to become progressive (Like the Newtonian movement) or degenerate and fade away (or become disreputable). This analysis reveals the practise of science as a more subtle endeavour. It also demonstrates that science judge’s worth ultimately on the basis of the utility and success of theories, not on arbitrary considerations.
There are no claims here of automatically excluding bias or confounding factors. What we have, on one level, is a co-operative social endevour. However, there are also underlying principles, such as falsification. This, over the long term, makes science self-correcting.
Science is certainly not without fault; but it has proven relatively robust in the face of political influence, personal bias and error. I wonder what philosophical underpinnings homeopathy has that can give it a similar degree of robustness?
This model also shows how science is able to cope with experimental data that does not fit with the hard core of a new theoretical structure until it has been given the chance to mature and prove its worth.
This complex reality, however, does not fit in with Chatfield’s plan of attack. So a simplistic straw man is attacked instead.
However, over the past 50 years much of the philosophy of science has been preoccupied with challenging the positivist picture of scientific methodology. The challenges fall broadly into two main areas:
Well, in reality this whole debate has been and gone. Popper overturned the Positivists in the 1960′s. Kuhn, among others, raised the problem of observation. Lakatos, as we have seen, provided a sophisticated synthesis of these views. Feyerabend railed, somewhat inconsistantly (more on this later) against science.
The outcome of all this is: science has dealt with it (as we shall see) and moved on.
• First, observations are necessarily dependent upon the observer
This is a big simplification. We all know that observation is theory laden. We make observations, not from first principles, but through the cipher of systems of theory (it is said though that some areas of Chemistry do not have this difficulty). This a long way from the post-modern relativism that Chatfield is hinting at here namely: it may be true for you but not for me.
Most philosophy of science (as opposed to the philosophy of critics of science) still admits an objective reality; even Kuhn admits to progress and utility. Its success can be seen as evidence for this position.
• Secondly, theories are never determined exclusively by data (underdetermination)
Again, true to an extent. However I have the feeling that we are about to be invited to consider the inverse: that theories can be determined through lack of data.
Observations and Observers
The positivistic understanding of evidence does not take into consideration culture, contexts and the subjects of knowledge production. What an observer sees that is, the visual experience that an observer has when viewing an object, depends in part on past experience, knowledge and expectations. Individual experiences cannot be logically deduced to be identical or even the same. Observation is dependent upon and coloured by our individual senses and our background beliefs and assumptions, so that what we see is never what really is, even under the most highly controlled experimental settings (Kuhn, 1970, and Feyerabend, 1978).
This is extremely simplistic post-modernist posturing. Even the most optimistic scientists only hold that they are making increasingly better approximations to reality. All that the scientific process really needs to deliver are increasingly useful approximations.
It is worth remembering that Chatfield teaches on a Batchelor of Science (B.Sc.) programme, albeit in homeopathy. One should therefore be able to infer that the author, at the very least, accepts that there is a general utility in science.
The position adopted in this passage is what philosophers call the retreat to absolute relativism. The references attribute this analysis to Kuhn and Feyerabend. For a practising scientist both of these philosophers are controversial.
Kuhn has his critics among scientists. As a practicing scientist I have always found Kuhn’s description of the non-linear nature of scientific progress useful. Conversely, Kuhn is often portrayed as promoting the sort of universal relativism displayed by Chatfield. This does seem to flow reasonable naturally from his conjecture that rival paradigms are incommensurable (literally the impossibility or unavailability of a common system of measure between paradigms). However, there is some evidence to suggest that Kuhn was not comfortable with this kind of relativism, preferring a more closely defined local relativism.
“[There are] contexts in which I am wary about applying the label ‘truth’… . Members of a given scientific community will generally agree which consequences of a shared theory sustain the test of experiment and are therefore true, which are false as theory is currently applied, and which are as yet untested. Dealing with the comparison of theories designed to cover the same range of natural phenomena, I am more cautious. If they are historical theories…I can join with Sir Karl [Popper] in saying that each was believed in its time to be true but was later abandoned as false. In addition, I can say that the later theory was the better of the two as a tool for the practice of normal science… Being able to go that far, I do not myself feel that I am a relativist. Nevertheless, there is another step, or kind of step, which many philosophers of science wish to take and which I refuse. They wish, that is, to compare theories as representations of nature, as statements about ‘what is really out there’. Granting that neither theory of a historical pair is true, they nonetheless seek a sense in which the latter is a better approximation to the truth. I believe nothing of that sort can be found.”
Kuhn, T S., “Reflections on my critics“
Kuhn certainly does not deny the utility of science, or that one theory can be shown to be a better tool than another. His relativism is focussed on whether theories represent, or approximate, an underlying reality.
As an aside, one of the five foci of Kuhn’s normal science (the forgotten but dominant part of his thesis) is the determination of the value of natural constants. Thus the working out of Avogadro’s number falls well within Kuhn’s description of normal science. I’m afraid that misreading Kuhn does not let homeopathy off having to confront the Avogadro limit!
The author provides a good example of an absolute ‘anything goes’ relativism associated with Feyerabend. It is, however, interesting to note that a decade before he wrote his provocative treatise ‘Against Method‘ he wrote an article in which he sought to distinguish between the reasonable scientist and the crank (emphasis mine).
“… The distinction does not lie in the fact that the former ["respectable"] suggest what is plausible and promises success, whereas the latter ["cranks"] suggest what is implausible, absurd, and bound to fail. It cannot lie in this because we never know in advance which theory will be successful and which theory will fail. It takes a long time to decide this question and every single step leading to such a decision is open to revision … No, the distinction between the crank and the respectable thinker lies in the research that is done once a certain point of view is adopted. The crank is usually content with defending the point of view in its original, undeveloped, metaphysical form, and is not at all prepared to test its usefulness in all those cases which seem to favour the opponent, or even to admit that there exists a problem. It is this further investigation, the details of it, the knowledge of the difficulties, of the general state of knowledge, the recognition of objections, which distinguishes the “respectable thinker” from the crank. The original content does not. If he thinks that Aristotle should be given a further chance, let him do it and wait for the results …”
Feyerabend, P. “Realism and Instrumentalism: Comments on the Logic of Factual Support“, in The Critical Approach to Science and Philosophy, ed. M Bunge, New York Free Press, 1964, p.305.
In this passage Feyerabend admits the possibility of testing, the value of investigation and the worth of results. It is interesting to note that Feyerabend’s crank defends the original theory, is not prepared to test its usefulness if it appears that the test may favour the opponent or admit the existence of problems. Sounds like the position Chatfield is advancing; on these terms homeopaths become Feyerabend’s cranks.
Feyerabend was sometimes critical of those who quoted what he had said in the past. Quite reasonably, his ideas moved on. Perhaps he may have moved on in this view as well, though nothing in Against Method indicates this. However the views of philosophers, though interesting and useful, are not Holy Writ.
The Stanford Encyclopedia of Philosophy gives us this retrospective view on Feyerabend’s critique:
“… His critique of science (which gave him the reputation for being an “anti-science philosopher”, “the worst enemy of science”, etc.) is patchy. Its flaws stem directly from his scientific realism. It sets up a straight confrontation between science and other belief-systems as if they are all aiming to do the same thing (give us “knowledge of the world”) and must be compared for how well they deliver the goods. A better approach would be, in Gilbert Ryle’s words, “to draw uncompromising contrasts” between the businesses of science and those of other belief-systems. …”
In the end it’s results that count. Just look at what scientific medicine has delivered and make an “uncompromising contrast” with homeopathy.
James Le Fanu provides a provocative list of what he considers to be the twelve definitive moments of modern medicine. Perhaps we may be tempted to chose others, but consider: what has homeopathy provided in comparison?
Chlorpromazine and the revolution in Psychiatry
The Copenhagen Polio Epidemic and the Birth of Intensive Care
Control of hypertension
Treatment of childhood leukaemia
Identification of Helicobacter as the Cause of Peptic Ulcer’s
We may demur at some of the choices made. However, this list makes an effective point. If scientific medicine is so philosophically flawed how come it has delivered such success? If we accept the Relativist position, can we really dismiss all of this as a function of perspective?
Feyerabend provided some interesting insights; he does not provide an adequate justification for ignoring this reality, in its ordinary sense.
Our acceptance or rejection of evidence is also value-laden, and our natural tendency is to reject evidence that does not fit with currently held theory; hence, positive results from even the highest- standard RCTs in homeopathy will be rejected by those who do not believe in the possibility that ultra-high dilutions can have an effect.
Here Chatfield provides, inadvertently, a good justification for RCT’s. Because individuals have their biases, measures of outcome and methods of analysis are defined before a trial. Similarly, subjects are assigned to the placebo or medication at random. These precautions are taken in order to remove bias and limit the play of chance, reducing the chance of “value-laden” decisions.
As noted previously the reason why individual positive trails have not lead to the acceptance of homeopathy is that they will still contain some bias, plus chance outcomes, along with the ‘one-in-twenty lottery’ of significance testing. Therefore, as already discussed, it the picture painted by the totality of high quality trials that counts.
Otherwise known as the Dunhem-Quine [sic] thesis, this states that any given body of evidence may support numerous, even contradictory theories. Scientific theories are inevitably underdetermined by data, such that what counts as good scientific theory depends largely upon other factors: social and political agendas, preferences, biases and whims – not the evidence itself.
We can find clear examples of underdetermination in homeopathy. Many people have carried out meta-analyses of placebo-controlled homeopathy trials in an attempt to answer the question of whether or not homeopathy is more effective than placebo (NHS Centre for Reviews and Dissemination, 2002)
These meta-analyses have been drawing upon the same data sources, published trials of homeopathy, and yet they come up with conflicting and contradictory results. The same data can be used to support either the claim that homeopathy does have an effect over and above placebo and also that it does not have an effect over and above placebo. Both theories are underdetermined by the available data and hence the data can be used to support either one.
In August 2005 The Lancet published a damning meta-analysis of homeopathy trials, which concluded that their results were consistent with the theory that the clinical effects of homeopathy are placebo effects (Shang et al, 2005). An editorial in the same issue called for ‘the end of homeopathy’. This meta-analysis is an extreme example of underdetermination. The final statistical analysis was performed upon only eight trials of homeopathy and six trials of allopathy. The authors used certain criteria for the selection of ‘highest-quality’ trials and these criteria were based upon their own beliefs about what makes a trial high quality, and nothing to do with what makes a trial sound from the perspective of a homeopath. It would be an easy task to use different criteria for the selection of high-quality trials and generate results that support the exact opposite theory.
Here Chatfield mistakes decisions made on a transparent, pre-determined and methodological basis for bias, rather than a precaution against it. That there were only eight homeopathic trials of the highest quality tells its own story; as does the consistent reduction in reported treatment effect with increase in objective quality.
Under-determination might have a role in principle, however it is being used here as a smokescreen for poor quality. It is also special pleading for a reduced standard of evidence to be used in assessing homeopathy.
Now to the Duhem-Quine Thesis: It is more usually stated as positing “… that is it is impossible to test a scientific hypothesis in isolation, because an empirical test of the hypothesis requires one or more background assumptions …”
It is important to note that the Duhem-Quine Thesis is an unbalanced one. Duhem believed that only in the field of Physics can a single individual hypothesis not be isolated for testing. He specifically excluded certain branches of Chemistry from this assertion.
On a basic level it is chemistry that homeopathy contradicts. Hence at least one leg of this thesis is irrelevant.
Quine‘s theory of underdetermination is a much stronger version. Epistemologically, he sees all human knowledge as being of one piece.
In reality, rather than the wishful thinking of homeopaths, scientific theories are rarely, if ever, tested in isolation; even Popper talked about systems of theories. As we have seen Lakatos’ research programmes also see theories as part of a system: with a hard core and protective belt of hypotheses.
So the Duhem-Quine Thesis does not do what the author hopes, because the wrong target has been selected: single theories. Theoretical groups can still be falsified.
The wheat can still be separated from the chaff. In Lakatos’ philosophy: “… in a progressive research programme, theory leads to the discovery of hitherto unknown novel facts.” whilst “… In degenerating programmes, however, theories are fabricated only in order to accommodate known facts.”
So with Lakatos the demarcation between science and pseudoscience lies not in whether a theory can be ‘tuned’ to accommodate new data, but in its success at providing new knowledge.
James Randi, magician and self-proclaimed sceptic, has famously offered $1,000,000 to anyone who can prove that homeopathy is any more than placebo effect; but none of the above constitutes proof.
In actual fact Randi has made it easier than that. The demonstration of clinical effects is not required:
“… can the applicant differentiate between homeopathic and non-homeopathic materials? We’ll accept positive results and the determination can be done by any means: chemical, physical, optical, biological (in vivo or in vitro), using infrared, ultraviolet, polarized, high-intensity, or pulsed light, conductivity or electrochemical means, Tarot cards, or a crystal ball. …”
It is hard to believe that the author is unaware of this. A homeopathic claim was tested in the UK and covered on national television in 2002 (Horizon). This trial was inspired by Benveniste‘s notorious experiment on the effect of massively diluted anti-immunoglobulin E on the degranulation of human basophils. The result: “…the experiment was a total failure. The scientists were no better at deciding which samples were homeopathic than pure chance would have been.”
So, Chatfield’s comments on Randi and his challenge are just plain wrong.
• Randomised, placebo-controlled trials that demonstrate statistically significant results? We have those (Bell et al, 2004; Yakir et al 2001)
Of course you do. What matters is the average result for a therapy over a number of good quality trials.
• Meta-analyses that demonstrate effect over and above placebo? We have these (Kleijnen et al, 1991; Linde et al 1997)
This commentary is just plain wrong.
The meta-analysis of Kleijnen et al did indeed show a positive result for homeopathy. However, the author’s concluded:
“… At the moment the evidence of clinical trials is positive but not sufficient to draw definitive conclusions because most trials are of low methodological quality and because of the unknown role of publication bias. This indicates that there is a legitimate case for further evaluation of homoeopathy, but only by means of well performed trials. …”
Chatfield has misrepresented this reference.
Linde et al concluded that the data “were not compatible with the hypothesis that the clinical effects of homoeopathy are completely due to placebo“. Rejecting this null hypothesis does not prove the alternate.
It should also be noted that a detailed analysis reported by Bandolier showed the apparent positive outcome to be very weak indeed.
In the same paper Linde et al also reported that they “… found insufficient evidence from these studies that homoeopathy is clearly efficacious for any single clinical condition …” Not exactly the ringing endorsement that homeopaths portray the paper to be.
It is also important to note that a re-analysis of the data used in this paper in 1999 by Linde, with many of the same authors, concluded that “in the study set investigated, there was clear evidence that studies with better methodological quality tended to yield less positive results.”
Another review, by Cucherat et al came to the same conclusion. This is exactly the result that would be expected if homeopathy were, indeed, ineffective.
Chatfield also neglects to metion the unfavourable trials. There is a good list of reviews of homeopathy here.
I think that it is clear that Chatfield is being economical with the actualité.
• Trials involving children, demonstrating positive results? We have those too (Frei & Thurneysen, 2001; Jacobs et al 2000)
• Trials involving the treatment of animals, demonstrating positive results? Yes we also have these (Merck von et al 1989; Searcy et al 1995)
More individual trials are cherry picked. Homeopaths do like to infer that trials in children and animals are somehow free from bias. See the ‘animals & babies’ section of this BBC commentary for a swift dismissal of the claim.
• How about trials that demonstrate the effects of ultra-high dilutions in the test tube? We even have these (Belon et al, 2004)
It’ll be interesting to see if it can be independently replicated. It’s also important to note that this family of experiments inspired by Benveniste is purported to show the same reaction occuring at ultra-molecular dilutions. For this to be a homeopathic effect the reaction would need to be reversed.
But none of the above constitute proof…
That’s quite correct. Chatfield knows (see below) that science does not seek to prove, rather to falsify. The author seems to be making the case that it would be reasonable for homeopathy to be ‘proven‘ in a pseudo-inductive cherry picking exercise.
Anyway where do Chatfield’s comments leave homeopathic ‘provings‘? Some consistency would be nice.
So just what would serve as ‘scientific proof’ that homeopathy has a positive effect over and above placebo? Well, in fact, no amount of clinical trials could ever prove this because, as any philosopher of science can tell you, proof can never be attained through scientific endeavour.
As stated above, it is quite true that science does not seek to prove. It seeks the best explanations it can, attempts to falsify them and accepts them on a provisional basis in the interim. After a hypothesis (or more usually a system of hypotheses) has proven predicatively useful and survived serious attempts to falsify it, it may ascend to the rank of theory but remains provisional.
What would it take for homeopathy to be viewed as part of the body of scientific knowledge? If a large majority of high quality clinical trials of homeopathy produced results better than placebo then it would certainly be accepted. Some plausible mechanisms and successes with very serious conditions would also help no end.
Those who say that science is so conservative that it rejects the heterodox have just never studied its history. Newton, eventually, prevailed over Aristotle and Einstein supplanted Newton.
Quantisation of energy was discovered by Planck, who really did not like the idea at all. Einstein was deeply unhappy about certain aspects of quantum theory. Nonetheless quantisation and quantum mechanics prevailed over their critics; and in a remarkably quick time too.
If the theory and evidence are good enough, history tells us that they eventually prevail. That homeopathy has failed to so in 200 years, occupying the same historical space as quantum mechanics is a valuable insight into its true, rather than rhetorical, strength.
A large proportion of our beliefs are obtained by a process of projecting from observed (past or present) events to cases that are either unknown, unobserved or in the future. For example: “Every swan I have ever seen has been white. It is likely that the next swan I see will be white.”
Not in contemporary science. Here we have classic, or naïve, inductivism introduced by stealth. This old philosophy of science was swept away during the 20th century. This is just another straw man.
Scientific inquiry proceeds in exactly the same way, projecting from (repeated) observation to universal laws. David Hume was the first philosopher to highlight the problems with this form of reasoning, way back in the 18th century. Hume stated that what leads us to believe that the future will resemble the past is a matter of custom or habit and cannot be rationally justified.
This is a criticism of naïve inductivism, rather than science. It’s a shocking statement given Chatfield’s apparant familiarity with Popper. Again, real science is not an inductive enterprise.
It is also important not to misrepresent Hume‘s views. He did, quite rightly, conclude that inductivism was not driven by reason. However, he was very taken with the predictive accuracy of inductivism.
Seventeenth century Europeans only had access to white swans, so even a completely unbiased researcher would have had reason to conclude that all swans were white. When Europeans finally travelled to Australia, they discovered that it is not true that all swans are white. If someone had suggested in the 17th century that some swans might be black, this would have been dismissed as entirely fanciful; no one could have predicted that some swans are black.
Karl Popper, a leading influence in 20th-century philosophy of science, is most famous for his belief that theories can never be proven, however many pieces of evidence that we have to support them. Rather, they can only be disproven. It does not matter how many white swans we observe, we could never prove that all swans are white because we can never be sure that we have seen all swans. However the observation of just one black swan is enough to disprove the theory that all swans are white.
Hence ‘scientific proof’ is a misnomer and the best we can aim for are hypotheses that have not yet been refuted.
After impugning science by saying it is an inductivist endeavour, Chatfield introduces Popper, arguably the single most influential philosopher of science in the 20th century. Popper showed that although science cannot offer proof it can rule out that which is false (falsification).
Even though his philosophy has been, perhaps, supplanted (or modified) by more advanced notions, this influence lives on.
Since Popper scientists tend to define science as being about testable hypotheses that can be falsified. It is easy to frame the central doctrines of homeopathy in a manner that is falsifiable. Does like really cure like? Is all chronic disease, apart from that due to orthodox medicine or to faulty living habits, caused by one of three ‘miasms‘ – syphilis, sycosis, and psora? Does ultra-dilution increase the inverse effect of a substance?
We are never going to be able to prove the efficacy of homeopathy through clinical studies. All we can do is add to the number of observations that support the theory that homeopathy has an effect over and above placebo. Even when we have a very large number of these positive trials, they will not constitute proof. Randi’s million dollars are safe.
More cherry picking: if homeopathy really worked then the number of high quality clinical studies showing this would outweigh the negative ones, not vice versa.
I’m sure that a homeopath of Chatfield’s standing is aware that Randi is not asking for evidence that homeopathy is clinically effective; just that homeopathic preparations can be distinguished at ‘ultra-molecular’ dilutions.
The High-Dilution Problem
It’s hardly surprising that homeopathy is accused of being no more than placebo. This is a completely logical conclusion from the perspective of a person who thinks and works within the current prevailing scientific paradigm. Positive results from placebo-controlled trials of homeopathy are doubted, as critics claim that remedies are composed of nothing more than water. To accept that homeopathic remedies have an effect would entail revision of much of our basic scientific conception of chemistry and the body (Sehon & Stanley, 2003).
Now into paradigms. Having sought refuge in a bizarre interpretation of Popper the author runs to Kuhn with his allegedly anarchistic interpretation of science. It’s not a matter of paradigms; it’s a matter of the quality of the evidence.
In spite of this apparent irrationality, homeopathy is the most frequently used form of CAM in five out of 14 surveyed countries in Europe, and among the three most frequently used in 11 out of 14 surveyed countries (NOU 1998). Between one quarter and one fifth of all EU citizens use homeopathic medicines (EU commission statement, 1997).
It’s of no relevance that homeopathy is the most widely used form of CAM. This is just an example of the logical fallacy: appeal to common practise. Just because lots of people use homeopathic products doesn’t make any case for their efficacy.
In 2005 the results from a six-year outcome study at the Bristol Homeopathic Hospital were published (Spence et al 2005). This was an analysis of over 23,000 outpatient consultations from 1997 to 2003. Over 70% of follow-up patients reported clinical improvement following homeopathic treatment, clearly demonstrating the clinical effectiveness of the homeopathic intervention. Whilst this extensive study can tell us that homeopathy was largely effective for this particular group of people, what it cannot demonstrate is that this was not simply placebo effect.
This is a customer satisfaction survey. It cannot determine efficacy as there is no control group: there is no way to know how many of these ‘patients’ would have got well of their own accord. There is not even any data in this study to enable ‘placebo’ effects to be separated from the natural, unaided, powers of recovery of the human body.
This debacle has been covered in detail here.
However, if homeopaths can facilitate a placebo-induced healing response in over 70% of people who visit them, many of whom have previously not been helped by various types of allopathic intervention, then surely homeopaths should be highly revered and re-labelled ‘miracle workers’.
There is a grain of sense in here. If homeopathy is such a powerful placebo, then there is the possibility of recovering sensible medical interventions out of homeopathy’s wreckage of failed hypotheses and mysticism. Perhaps the author would be better occupied devoting her philosophical skill to overcoming the ethical impediments to this outcome.
Time for a New Paradigm?
The history of science teaches us that there have always been things that could not be explained according to currently held scientific opinion. Science is not static, it evolves as understanding increases and is occasionally subject to a complete revolution, bringing with it an entirely new scientific paradigm (Kuhn, 1970). Kuhn describes how, when we are close to this type of paradigm shift, there will be many things that appear to be incommensurable with the prevailing paradigm. These he terms ‘anomalies’ and they cannot be explained within the framework of current scientific opinion. Homeopathy might be considered an example of an anomaly.
Or it most likely is just false. New scientific paradigms cannot have arbitrary content. A new theoretical structure must embody the successes of the old as well as making bold new (testable) predictions.
If it were time for a new paradigm then homeopathy would have demonstrated not just its efficacy, but given us bold new predictions as well as encompassing the successes of scientific medicine and the relevant parts of chemistry.
It’s not a new paradigm that Chatfield is advocating. It’s a refuge where Kuhnian normal science is suspended.
It is normal for people who try to work with these anomalies to be marginalised, as the emergence of a new paradigm is painful for those with vested interests in the existing paradigm. However the emergence of a new paradigm that embraces holism, vitalism and homeopathy need not involve the rejection of rationalism. Rather we might expect a re-direction of the rationalistic tradition that encompasses better explanation of known anomalies.
We have already seen that homeopathy, being obsessed with symptoms, would need to change radically to embrace holism (in the sense of treating the whole person).
A paradigm including vitalism would be strange indeed. Anthony Campbell points out that “… The term Hahnemann used was ‘dynamis’, which is usually translated as ‘vital force’. By this he meant a spiritlike principle that gives life to the body.”
Now this may be fine as a metaphor, but it is now known that this is not how living things work.
Even Kuhn, as we have seen, was quite happy to consign failed theories to the past. It’s a shame that Chatfield is not so taken with this aspect of his philosophy. This dogged defence of Hahnemann’s original doctrine is in Feyerabend’s terms distinctly cranky.
There is no justification for returning to a falsified hypothesis. It would not be rational.
How can this new rationalist paradigm include both the pathology of disease and miasms? How can it reconcile ultra-molecular dilution and Avagadro? How can it have both vitalism and the organism?
Whilst the practice of homeopathy becomes ever more popular, in spite of its incommensurability with the prevailing scientific paradigm, homeopaths should not be at all surprised by resistance and even attack from those whose beliefs are firmly rooted within that paradigm. Until we have an accepted explanation about how ultra-high dilutions can have an effect, then no small amount of ‘scientific evidence’ through clinical trials will convince the sceptics.
What would convince the sceptics, and win Randi’s million dollars? I think a definitive experimental discrimination between the homeopathic preparation and the solvent would do nicely.
What should not convince anyone is cherry picking clinical data, dodging the issue and misrepresenting the content of papers.
Chatfield does seem much taken with the idea of incommensurability. As we have seen this literally means the impossibility or unavailability of a common system of measure. Now Kuhn clearly allows for partial translation between paradigms. This makes sense because there are common systems of measure between homeopathy and scientific medicine: clinical outcomes. Death, for example, is the same in either paradigm. Homeopaths talk about diseases that are recognised in medicine all the time. While homeopathy’s theoretical content may not always translate, the degree of commensurability available to us does allow for clinical evaluation of homeopathy. To quote a commentary on Kuhn:
“… even if we retain a holism about the sense of theoretical terms and allow that revolutions lead to shifts in sense, there is no direct inference from this to a shift in reference. Consequently, there is no inference to the inadmissibility of the comparison of theories with respect to their truth-nearness …”
Maybe it is time for researchers in homeopathy to focus upon what is of benefit to the development of homeopathic practice instead of trying to prove that homeopathy has an effect over and above placebo. Just exactly what is it that we do that enables more than 70% of our patients to say that they have felt benefit? Clearly we are doing some things well, but many of our traditions and beliefs in homeopathy have not been subject to rigorous testing, and research has a lot to offer us in this area. If through research we can improve upon that 70% figure, then just imagine the future for homeopathy.
Not a very convincing plea. The author has spent most of this tract in arguing against rigorous trails for homeopathy and particularly a rigorous assessment of the body of evidence.
This is not the basis for believing that the author has any real interest in rigorous testing of any of the other facets of homeopathy.
Given that the author has invoked the extreme relativism of Feyerabend, how would the traditions and beliefs of homeopathy tested? What yardstick would be used? If everything has the same worth as everything else how can anything be ruled out or in?
This tract is about justifying homeopaths “taking their ball”, but I don’t think that they want to “go home”.
It misrepresents and cherry picks the clinical data to build a sense of grievance. This is amplified through a misrepresentation of the philosophy of science.
Now, if they want to withdraw from the ‘paradigm’ of science, that’s fine with me. Just as long as they take this to its logical conclusion: that they don’t expect their nonsense to be supported via the public purse and they withdraw from university ‘medical’ departments.
So after all the references to the ‘evidence’ for homeopathy and the philosophysing, what do we have?
Mainly a plea to allow advocates of homeopathy to cherry pick from the available studies or use customer satisfaction surveys instead.
The number of reviews of homeopathy available was criticised, without mentioning that they were overwhelmingly negative. How many more negative reviews would the author like made available on line?
More worryingly, cited studies were badly mishandled. For instance Kleijnen et al cannot be cited to demonstrate an effect for homeopathy over placebo; the data were of such low quality that the authors stated that no definitive conclusions could be drawn.
The author neglects to mention that Linde et al (1999) contains the important caveat that they ”… found insufficient evidence from these studies that homoeopathy is clearly efficacious for any single clinical condition …”
The philosophical content is just a smokescreen for when studies don’t turn out the way the author wants. Popper is cited as part of an argument that says because science cannot prove it cannot judge. This is an abuse of Popper because what he really says is you cannot prove but you can disprove and thereby judge. In fact Popper draws a demarcation line between science and pseudoscience on this basis.
Kuhn is invoked for his talk of paradigms and aledged relativism. We have seen that even Kuhn believed that competing theories could be compared. His incommensurability is not an escape route for homeopathy.
Feyerabend was certainly a relativist. Even he, at least at one point in his development, thought that there were respectable and cranky ways of thinking. Holding on to doctrines in their original form was one of the cranky habits. The author’s paen for vitalism seems to fall foul of this.
Is it reasonable, on the authority of Feyerabend, to discard any sense that we can judge between theories and select one as more useful than the other? This would be incredibly rash.
Most disappointingly Lakatos does not figure in the authors philosophical treatise. This is amazing. He was Popper’s student, along with Feyerabend. Both were contemporary with Kuhn. Perhaps it’s because his sophisticated falisficationism and synthesis of Kuhn’s social analysis make for a much harder target?
This moves us on to the Duhem-Quine Thesis and underdetermination. This unbalanced thesis is defeated because science is about systems of theories, as explained by Popper and more thoroughly by Lakatos. Perhaps this is another reason for Lakatos’ omission.
All in all, this piece is a ‘dressed up’ version of an argument commonly advanced by advocates of nonsense therapies. Instead of the usual cry that the mumbo-jumbo is ”outside of science” Chatfield claims it’s in a different, incommensurable, paradigm.
As a final challenge to those who would argue that our judgements are relative to such a degree that ‘anything goes’ I would invite them to consider the salutory case of Thabo Mbeki and HIV/AIDS. For Mr Mbeki, if reports are to be believed, the idea that HIV causes AIDS is “… entrenched in “centuries-old white racist beliefs and concepts about Africans” …” For Mbeki, in essence, it’s a product of western European cultural biases. Now under universal relativism this view is as good a view as any. This has not, however, stopped Africans from developing AIDS due to HIV infection. It does nothing to stop the tide of devastation. Death and sickness are available as measures between any paradigms.
Regardless of what some people wish, or what philosophy they prefer: there is a reality out there that we cannot cheat, no matter how hard we try.
16 Responses to “It’s Evidence Jim, But Not As We Know It!”
Sorry, the comment form is closed at this time.