Articles

Homeopathy and Plausibility

The fundamental concept of science-based medicine (SBM) is that medical practice should be based upon the best available science. This may seem obvious, but there are many important details to its application, such as the relationship between clinical and basic science. Clinical claims require clinical evidence, but clinical evidence can be tricky and is often preliminary. It is therefore helpful (I would say essential) to view the clinical evidence in light of all of the rest of science.

A thorough basic and clinical science analysis of a medical claim can be summarized by the term “plausibility,” or “prior probability” if you want to put it into statistical terms. When we say a certain belief is plausible we mean it is consistent with what we know from the rest of science. In other words, because of the many weaknesses of clinical evidence, in order for a therapy to be generally accepted as part of SBM it should have a certain minimal supporting clinical evidence and overall scientific plausibility.

These can exist in different proportions – for example one therapy may be highly plausible (it would be shocking if it were not true) and have modest supporting clinical evidence, while another may have unknown plausibility but with solid clinical evidence of efficacy. But no therapy should have clinical evidence that suggests lack of efficacy, nor extreme implausibility (not just an unknown mechanism, but no possible mechanism).

The poster child for extreme scientific implausibility within medicine is, arguable, homeopathy. Its “law of similars” is little more than sympathetic magic, and its “law of infinitessimals” leads to concoctions diluted to the point that they have zero active ingredient and only the magical memory left behind. It is this reality which inspired a recent review to characterize homeopathy as “witchcraft.”

What about the clinical evidence? The clinical evidence, when taken as a whole, and even if we set aside the question of plausibility, shows that homeopathy does not work. Decades of research have failed to provide supporting evidence for any application of homeopathy. It has failed to reject the null hypothesis, to show efficacy, to allow for the recommendation of homeopathy for any indication, to differentiate homeopathy from a placebo – in other words, homeopathy does not work.

If we consider the SBM criteria above, we can summarize homeopathy by saying it has extreme implausibility and the clinical evidence shows lack of efficacy. It should not work, and it does not work. There is no legitimate controversy about this. The only think keeping homeopathy alive are delusional proponents and a public (including many regulators) who do not know what homeopathy truly is.

Proponents, however, are desperately trying to keep their pseudoscience alive by misrepresenting the evidence and the arguments of homeopathy critics. A recent example of this is a paper by Peter Fisher, the Clinical Director Royal London Hospital for Integrated Medicine, and a prominent homeopath. He is trying to coin the phrase “negative plausibility bias.” His argument, essentially, is that the evidence shows homeopathy works (at least as well as medical treatments that do not work, which is an odd argument), but there is a negative plausibility bias against homeopathy which motivates scientists to reject this evidence. Fisher is both wrong and irrelevant in this position.

Fisher is wrong in arguing that the clinical evidence supports the efficacy of homeopathy. He does this by cherry picking positive studies (which are part of the noise of any clinical research), a common strategy. Meanwhile, systematic reviews do not show supporting evidence for homeopathy. Worse for homeopathy, there is a clear pattern in the research. The better designed and controlled the study, the more negative the results – a clear pattern of lack of efficacy. Even reviews that desperately try to spin the results in a positive way in the end show there is no evidence to support the efficacy of homeopathy.

The main point of the article, however, is to dismiss the scientific assessment of homeopathy as a “bias.” It seems like it is news to Fisher that plausibility is not a bias – it’s science.

Interestingly, Fisher claims that the law of similars part of homeopathy is not even controversial – and then he makes the same, tired analogies to hormesis and paradoxical drug effects, which actually have no relevance to homeopathy. Homeopathy “ingredients” are chosen for magical reasons that have no relationship to actual biology or chemistry. Further, the high dilutions of homeopathic preparations render such analogies invalid. Fisher and other homeopaths are just fishing for any possible hand-waving justification for homeopathy, but they lend nothing to its plausibility because they fail to make a scientifically coherent argument.

Fisher then essentially argues that the clinical evidence shows homeopathy works (wrong) but is rejected because of the “negative plausibility bias.” There is a kernel of truth to his view in that, in the face of extreme scientific implausibility, even modestly positive clinical evidence is looked upon as insufficient and not definitive. We can look at it this way – what are the odds that a mountain of solid basic science is wrong vs some sloppy and tricky clinical research is wrong? It would take overwhelmingly rigorous and positive clinical evidence to call into question long established principles of basic science. Homeopathy does not come anywhere close – even if we take the distorted and incorrect view of the clinical evidence Fisher is pushing.

The article is essentially Fisher whining that the scientific community is not ignoring the extreme scientific implausibility of homeopathy.

Fisher also tries to make a tu quoque argument – recycling yet another old ploy of medical pseudoscientists. He says that the evidence for antibiotic use in upper respiratory tract infections (URTI) is no different than homeopathy but practitioners use antibiotics and not homeopathy. The evidence base for any other practice is irrelevant to homeopathy, but even that aside his argument is a curious one. I agree that systematic reviews fail to show efficacy for routine antibiotic use in URTI. Therefore – they should not be used. In fact there are efforts within mainstream medicine to reduce the use of antibiotics in URTI, and to eliminate their routine use.

The story with URTI is more complex, however, because some people do get bacterial interference with URTI and there may be a role for antibiotics in selected cases – the trick is in knowing how to select those cases. More research is legitimately needed to explore these issues.

The only consistent position, therefore, is to favor the elimination of routine antibiotic use in URTI, based upon current evidence, and also to favor the complete elimination of homeopathy as a practice. Meanwhile, it is reasonable to do more research into a possible limited role of antibiotics in selected cases (based partly on plausibility). It is also reasonable to favor the abandonment of any further research into homeopathy, based on its extreme scientific implausibility.

You can call it a “negative plausibility bias” or you can call it science, based upon your perspective.

Posted in: Homeopathy

Leave a Comment (47) ↓

47 thoughts on “Homeopathy and Plausibility

  1. cervantes says:

    The problem, of course, is that this is not an argument in which reason has any role to begin with. By the same standard, all religion is nonsensical and should be abandoned, but you aren’t going to win that battle any time soon by engaging in reason-based discourse. (Otherwise Richard Dawkins would long ago have been free to move on to other concerns.) What we are up against is faith, and a vast financial interest. The only useful strategy is to attack the financial interest by getting the authorities to recognize homeopathy as fraud. Homeopaths are relieving people of their money by making demonstrably false claims. That should be illegal. Prosecute them. That’s the only way to put a stop to it. Arguing with them is a waste of time.

  2. rork says:

    I suppose my plausibility for homeopathy is smaller than the mean of the population, since there are crazy folks with positive plausibility bias for it. For practitioners there’s the money. It’s the users I feel bad for.

    I was surprised by the insult to witchcraft. The stuff I smear on the coven’s broom handles ain’t no placebo. The analogy was lost on me.

  3. Josie says:

    Witchcraft as a general term for magic (not Wicca, not the practice of rubbing hallucinogens on sticks and rubbing a a certain mucous membrane against them).

    If someone did something that seemed not to follow everyday logic or plausibility they could be accused of being a witch –someone who had access to things that broke the rules of everyday existence. That someone was to be shunned and cast out.

    Pretty sure it’s the shunning and casting out that we’re focusing on here.

    I would like to see homeopaths shunned and cast out.

  4. Zetetic says:

    Where’s Dana Ullman when you need him! ;-)

  5. Purenoiz says:

    Above is my argument in favor of Homeopathy. It has been diluted to 23C, it is to strong for your comprehension, since your eyes are not yet evolved enough to see it. In the future my argument will be proven correct.

    quack quck

  6. pmoran says:

    I am a little surprised that Fisher is still trying to rationalize homeopathy. Hasn’t he been involved in some of the negative studies?

    The fraud approach won’t work, either politically or legally, because if Fisher’s actions demonstrate anything it is that he (along with many of his clients) truly believes homeopathy does work.

    Various illusions within medical practice, probably including placebo responses, are more powerful than I think many lay people recognize.

  7. David Gorski says:

    Various illusions within medical practice, probably including placebo responses, are more powerful than I think many lay people recognize.

    I used to think that. The more I look into it, though, the less I think that and the more I drift closer to Mark Crislip’s view of placebos.

  8. jre says:

    Various illusions … are more powerful than I think many lay people recognize.

    But the fact that someone fails to recognize an illusion as such is precisely what makes it so powerful, isn’t it? Anyone with a working knowledge of confirmation bias, motivated reasoning, the halo effect, and the other tricks we play on ourselves, is less likely to be fooled, and so less likely to benefit (if that’s the word) from the placebo response than someone without that knowledge.

    As for me, I can’t believe in it unless it works even when I don’t believe in it.

    And it’s worth noting that for most alternative therapies, the effectiveness (or lack thereof) of the placebo response is of mostly academic interest, because — pace David Freedman — most alternative practitioners do not accept that their therapies work no better than placebos. Instead, like Peter Fisher they clutch to any slender reed of argument that allows some possibility that the woo in question has objective effectiveness.

  9. Pete says:

    I wonder if Peter Fisher really believes what he is arguing. I have no medical training and am able to understand that there is no scientific basis whatsoever for homeopathy to work. I feel that he is just being intellectually dishonest just to make money. I wish there were some accurate way to test these homeopaths to see if they really believe what they are saying. I wish there were something like a lie detector test machine (but that actually works) . I wonder how many of them would be willing to be hooked up to the machine and answer the question “Do you rest believe that homeopathy works?”

    I don’t understand how a trained medical professional can really believe that homeopathy works when a lay person like myself can plainly see that there is no way homeopathy can work because there is no basis for it to work in the first place. Because of my ignorance with respect to medical treatments I always defer to the judgement of the medical community, not one doctor but multiple doctors.

  10. pmoran says:

    David:
    “Various illusions within medical practice, probably including placebo responses, are more powerful than I think many lay people recognize.”

    I used to think that. The more I look into it, though, the less I think that and the more I drift closer to Mark Crislip’s view of placebos.

    There is no question about the illusions that strongly suggest intrinsic treatment efficacy. We have a long tradition of susceptibility to that within the mainstream.

    As I have admitted, I also waver as to the exact role of placebo, but there is at least as much evidence and plausibility for as against.

    I mainly react when there is unwarranted “scientific” certainty on this and other matters, especially when there is also ad hominem against people who are not holding unreasonable positions, or who at most may be understandably misled.

    I have just heard of a cancer sufferer, a friend of my wife’s, who went to an acupuncturist for the relief of pain. He was advised to stop his medications and that it would take six months to “get the toxins out of his system”.

    As you know I have utterly no tolerance for this kind of thing. The patient made his own decision not to go back.

  11. Davdoodles says:

    But isn’t that how the ‘law of infinitessimals’ works?

    The less plausible, and the less efficatious, homeopathic nostrums are, the more powerful they become!
    .

  12. Nescio says:

    The popular misconception about placebos is that they engage a mysterious innate healing capacity. The evidence, disappointingly, suggests that they don’t. I used to have high hopes for psychoneuroimmunology, but not so much any more. The most interesting effects of placebos, in my opinion, are in conditions with a large psychological component, such as pain.

    David and Peter appear to agree that the placebo effect is an illusion, but in a sense pain is an illusion, being an interpretation of signals received by the brain. Every illness has some psychological component (there is the illness, and there is our experience of the illness), and placebos can affect this, sometimes to a surprising extent as we saw in that recent study of placebos and asthma.

    Perhaps SBM should support placebos that do not have any supernatural belief systems attached to them such as, for example, medical hypnosis, massage therapy, psychotherapy and counseling. Would that satisfy the need to be ethical and not lie to the patient but to also address the patient’s subjective experience of an illness?

  13. woo-fu says:

    @jre

    As for me, I can’t believe in it unless it works even when I don’t believe in it.

    Thank you for that. It pretty much sums up my attitude, too.

  14. cervantes says:

    Nescio — psychotherapy and counseling are not placebos. There is a substantial evidence base that specific forms of counseling — e.g. motivational interviewing, cognitive behavioral therapy — are more effective than non-specific forms of attention in alleviating depression, reducing substance abuse, improving sleep, and addressing other emotional and behavioral problems.

    True enough, a lot of psychotherapy is not evidence based or even grounded in one or another form of mysticism, but that is not true as a generalization.

  15. Nescio says:

    Cervantes,

    Nescio — psychotherapy and counseling are not placebos.

    I agree that they may not be placebos for emotional and behavioral problems. What I meant was they would be placebos for a non-psychological, objective, physical illness, but they could effectively address the psychological elements of that illness.

    Science based medicine is good at providing the best available treatment for objective illnesses, but perhaps not as good as CAM at dealing with the subjective elements of an illness. I think this is what Dr. Moran is getting at when he appears to support the prescription of placebos. I’m groping towards a way of addressing this shortcoming that doesn’t feed into nonsensical belief systems.

  16. daedalus2u says:

    cervantes and Nescio, whether a particular treatment is a “placebo” or not depends on the definition of “placebo” that you are using.

    If your “definition” is “something not effective”, then psychotherapy is not a placebo because at times it is effective. But then so is acupuncture, not a placebo because at times acupuncture is effective. Homeopathy is sometimes effective, so it to is not a placebo according to the definition of “something not effective”.

    I think that is a wrong and not useful definition of “placebo”. The definition I like is a treatment that has positive therapeutic effects which are mediated not through chemistry like active drugs or physical effects like surgery.

    By that definition psychotherapy is a placebo because the positive therapeutic effects are not mediated through chemical or physical effects.

  17. Nescio says:

    daedalus2u,

    I would characterize some of the interventions you mention as not having any positive therapeutic effects at all, as compared to placebo, which is often the definition of therapeutic efficacy.

    It also depends on the condition you are treating. Antibiotics (effective in some conditions) are ineffective for viral infections, yet a patient with a viral infection may feel better after treatment with an antibiotic; in that context the antibiotic is a placebo.

    Effective treatments have a placebo effect in addition to their specific therapeutic action. By ‘placebo’ we generally mean a treatment that has only placebo effects on the condition it is used to treat.

    I think we might need some better definitions or possibly new terms for some of these phenomena.

    For example, perhaps the term ‘placebo’ should be restricted to a therapeutically inactive intervention when used in a RCT.

    ‘Placebo effect’ is a much misused and misunderstood term that is used to mean different things by different people, and even when not misused covers a number of probably unrelated phenomena: suggestion, confirmation bias, regression to the mean etc.. Perhaps the term should be abandoned by SBM altogether, and these phenomena referred to separately and specifically, or collectively as ‘non-specific effects’. It might avoid some misunderstandings.

  18. rork says:

    I’m also not quite happy with daedalus’s “because the positive therapeutic effects are not mediated through chemical or physical effects”. Just saying something or doing something can change the other person’s brain – it’s chemistry, it’s structure. I’m not claiming I have got all the terms and their definitions worked out, just that this student had trouble there.

  19. tpr007 says:

    Woo and pseudoscience are the bane of modern medicine. The biggest problem is the general public saying “What’s the harm?”

    “A Lot” I say – http://coffeelovingskeptic.com/?p=711

  20. daedalus2u says:

    Nescio, if you don’t have a definition of placebo, you can’t use “placebo” as part of the definition.

    Defining “effective” as “something that works as well as something known to be effective” is equivalent to defining “non-effective” as “something that works as well as something that is a placebo”. It is the lack of a definition of “placebo” that allows acupuncturists to declare that “toothpicks work too!” because they work as well as real needles which acupuncturists “know” are effective. A circular definition allows both interpretations.

    Sometimes even medicines that are well known to be effective don’t work. Does that make them “placebos”? Or does it make them “ineffective”?

    What is the mechanism by which psychosocial stress makes some disorders worse? Is that a drug effect? I would say it is a placebo (actually nocebo because the effects are adverse) effect because the effects are not mediated through drugs or surgery.

    You can’t simply declare that psychosocial stress has no effects because it isn’t true.

    I appreciate that defining placebo causes people angst. If you want to have a discussion about placebos, you have to define your terms, and then use those terms consistently.

  21. ConspicuousCarl says:

    daedalus2u on 25 Aug 2011 at 6:22 pm

    What is the mechanism by which psychosocial stress makes some disorders worse? Is that a drug effect? I would say it is a placebo (actually nocebo because the effects are adverse) effect because the effects are not mediated through drugs or surgery.

    Stress hormones directly and physically affect neurons and can even cause permanent damage. I don’t know about the exact temporary effects of stress on specific symptoms of any particular disorder, but stress hormones are definitely more like drugs than like subjective experiences.

  22. Nescio says:

    daedalus2u,

    I have read your comment several times and I still don’t really understand what you are getting at. Perhaps this is an example of the misunderstandings I was referring to earlier.

    if you don’t have a definition of placebo, you can’t use “placebo” as part of the definition.

    I wasn’t aware that anyone was suggesting that you can. I certainly didn’t intend to.

    Defining “effective” as “something that works as well as something known to be effective” is equivalent to defining “non-effective” as “something that works as well as something that is a placebo”.

    No one defines ‘effective’ like that do they? ‘Effective’ is defined as working better than something that is known to be ineffective, like the fake asthma inhalers used in the study discussed here recently, or sugar tablets, or saline injections.

    It is the lack of a definition of “placebo” that allows acupuncturists to declare that “toothpicks work too!” because they work as well as real needles which acupuncturists “know” are effective. A circular definition allows both interpretations.

    I certainly don’t buy that interpretation, and I don’t think many people here would either. Since toothpicks work as well as needles, and both inserted randomly work as well as those inserted in the prescribed places, I think most people would conclude that acupuncture doesn’t work the way acupuncturists claim it does. Claiming that toothpicks work as well as acupuncture is surely just special pleading. In any case, acupuncture has only shown a ‘super placebo effect’ in conditions like pain or nausea that have a large psychological component. Saline injections work really well for these as well, and I don’t think anyone claims that saline injections have any specific effects.

    Sometimes even medicines that are well known to be effective don’t work. Does that make them “placebos”? Or does it make them “ineffective”?

    Neither. No one claims that any science-based treatment is always effective; RCTs give us an idea of how often they are. A good placebo for a specific condition would never have any effect on that condition but would be otherwise indistinguishable from the intervention being tested.

    What is the mechanism by which psychosocial stress makes some disorders worse?

    I would say it is the result of a combination of various factors including hormones, chronic muscular tension, poor sleeping patterns and the unhealthy lifestyle (smoking, alcohol, diet, lack of exercise) associated with psychosocial stress.

    I am not aware of much convincing evidence that the converse is true; that stress reduction, relaxation or indeed euphoria have any great effect on healing (I have been reading about this recently). This is certainly believed by many people and it’s an attractive idea, but when you start looking at the evidence, there really isn’t much of it, except perhaps for heart disease.

    Some studies have even found that mildly depressed people outlive those who are not depressed at all, or that optimism may lead to early death (though that may be because of risk-taking behaviors). Despite popular belief, breast cancer patients do not live longer if they acquire a positive outlook through psychotherapy or support groups, though I’m sure quality of life is improved.

    Is that a drug effect?

    No, I wouldn’t say so. A drug is a foreign substance introduced to the body from outside.

    I would say it is a placebo (actually nocebo because the effects are adverse) effect because the effects are not mediated through drugs or surgery.

    Is that a useful way to define a placebo/nocebo? As I wrote above, I think perhaps the term ‘placebo’ should be restricted to inert interventions used in RCTs. Otherwise you risk conflating non-specific effects of placebos, such as regression to the mean, natural variation in an illness, confirmation bias, expectation etc. and in the case of psychosocial stress the other factors I mentioned, with the more interesting effects and potentially useful effects I suspect you are hinting at.

    You can’t simply declare that psychosocial stress has no effects because it isn’t true.

    Has anyone declared such a thing? How can you tease out the effects of the different factors associated with psychosocial stress? How much of it is due to cortisol for example, and how much is due to not eating properly, drinking too much and not getting enough sleep?

    I appreciate that defining placebo causes people angst. If you want to have a discussion about placebos, you have to define your terms, and then use those terms consistently.

    As I said, I’m not sure that the term ‘placebo’ is useful outside the context of a RCT or that ‘placebo effect’ is a useful term at all. I’m also not convinced that physiological placebo effects are as therapeutically useful as I think you are suggesting. I see the positive effects of placebo over no treatment as having an effect on the patient’s subjective experience, but not directly affecting their objective physical condition. I suspect you disagree.

  23. pmoran says:

    David and Peter appear to agree that the placebo effect is an illusion, —

    No, I was careful to refer to the “illusion of intrinsic treatment efficacy” which can certainly be contributed towards by placebo responses and other “incidental” components of medical interactions e.g. simple reassurance, or distraction from symptoms .

    It seems to me to be a matter of taste which of an enormous variety of influences are included under “placebo” and I confess I try to avoid getting too entangled in that knotty task.

    I agree with you that the vocabulary is a problem. The word “placebo” is now so laden with extra meanings that it is difficult to have a rational discussion about the potential of these influences.

    Even the phrase “a treatment works/ doesn’t work” needs qualification for its exact meaning to be conveyed.

  24. evilrobotxoxo says:

    @daedalus:

    I think you’re correct in that the definition of “placebo” is highly context dependent, but I disagree with the underlying philosophy you’re using to try to define it more rigorously. In my opinion, “placebo” is just another word for a negative control. When you say that psychotherapy is a placebo because it does not involve physical or chemical mechanisms, that is a demonstrably false statement premised on dualistic notions of brain function. Psychotherapy is basically a set of techniques for producing patterned sensory stimuli that evoke specific patterns of action potential firing in ascending cranial nerves (and, in the case of certain CBT techniques, spinal afferents as well). There is no magic at work, and the mechanism of action is purely physical in nature.

    Similarly, imagine a clinical trial for a neurological disorder where viral vectors are injected into the some nucleus to decrease activity there (e.g. the subthalamic nucleus in Parkinson’s disease). The “placebo” (i.e. negative control) group undergoes cannulation and saline injection into the structure, while the “treatment” group gets the virus. The “placebo” group improves beyond what’s expected because the very act of cannulation and saline injection introduces a small lesion that has a therapeutic effect, which, BTW, is mediated solely by “physical” mechanisms, not reporting bias, patient expectations, etc.

    This is why I think the notion of the “placebo” group should be abandoned and replaced with the more consistent notion of a “negative control.”

  25. daedalus2u says:

    How do you know something is effective or ineffective? Only by testing it against something. How do you know if what you are testing it against is truly effective or truly ineffective?

    If you are trying to test a treatment that you “know” works, and you test it against a second treatment and you find out that the second treatment works as well as the first treatment that you “know” works, then you have achieved a finding of non-inferiority. This happens all the time.

    The definition of “placebo” that I am trying to get is any treatment that doesn’t have effects mediated through physical reality external to the patient. Essentially if you put the patient inside a bubble and nothing was exchanged through that bubble, no drugs, no effective surgical intervention. The only things that can pass through the bubble are communication, and magic.

    Of course all effect within the patient are mediated through physical effects, physiology is only physical effects. What I am trying to do is draw a control boundary around the patient and when only sensory effects pass through that boundary, then any positive therapeutic effects are due to the placebo effect.

    CAM posits that the effects of their interventions are due to their magical effects (but the CAM practitioners don’t appreciate that it is magic), I posit that any effects are due to the placebo effect mediated through communication (because there are no magical effects).

    Evilrobot, Is psychotherapy a placebo? Is injection of saline a placebo? Is injection of saline at 0.8% a placebo? Saline at 0.9%? If you had the opportunity to do near infinite numbers of trials at differentially different conditions, I would expect that isotonic saline would have slightly better therapeutic effects than saline that was hypotonic because the hypotonic saline would trigger some osmotic compensatory pathways and that would trigger a stress response. I think that the triggering of a stress response by needles puncturing the skin is why toothpicks actually seem to work a little bit better than needles for acupuncture.

    What I am trying to get at is an unambiguous definition of what is a “placebo” so we can compare placebos and non-placebos. I think that what the other comment authors here want is a definition like that of pornography; “I know it when I see it”. That is the problem, people don’t know “placebo” when they see it. What we need is a non-subjective definition of “placebo”, or “placebo effect”, or “placebo response”. To me they are all the same, a positive therapeutic effect mediated though communication or something other than physical reality.

    If homeopathy actually worked, it would still be a placebo under my definition, but it would be a placebo that worked, the way that psychotherapy is a placebo that works.

    If you want to discuss whether a placebo “works” or not, you can’t have the definition of what a placebo is contingent on whether it works or not.

    What is a negative control for psychotherapy? What is a negative control for isotonic saline?

  26. pmoran says:

    How do you know something is effective or ineffective? Only by testing it against something. How do you know if what you are testing it against is truly effective or truly ineffective?

    I am not sure that those questions make sense when applied to placebo responses. The placebo “does” nothing. It cannot be “effective” in and of itself and independently of patient expectations and everything else to do with the therapeutic encounter…

    Thus the same placebo may elicit nothing in one patient , but a strong response in another, few responses with one practitioner, many with another.

    So what does “effective” mean in such a context? If the usual placebo theory is correct then anything at all can have be “effective” in the sense of assisting with therapeutic responses under the right conditions. That is what we seem to observe in practice, most spectacularly with CAM, if interprested a certain way.

    The only uncertainty is the extent to which patients are lying about being improved by certain methods. There are obvious concerns if we are to base a scientific judgement on that assumption.

  27. evilrobotxoxo says:

    @daedalus:

    When you explain it that way, what you’re saying makes a little more sense. However, I still think there are practical problems with your definition. First, as you acknowledge, everything is physical, including the effects of communication. So your definition depends on something that you admit is arbitrary, and in practice, different people would disagree about where the draw the line of what constitutes “physical” for different cases. I don’t think an unambiguous definition can be based on that. I think the example of psychotherapy points out where your concept breaks down. For example, I used to work at a VA hospital that has a virtual reality combat simulator for exposure therapy for combat veterans with PTSD. They put veterans in there and gradually expose them to more and more frightening combat-like situations, while the veterans know they’re safe, and they eventually desensitize them to environmental triggers for their symptoms. It’s very effective, and it’s a specific treatment targeted at specific problems in the brain based on decades of work in pavlovian conditioning, fear extinction learning, etc. It’s not a nonspecific effect by any conceivable definition. Do you really think it makes sense to call that a “placebo that works?”

    A second issue is that in practice it’s unclear what percentage of the “placebo response” in trials is actually due to the placebo effect per se. For example, in antidepressant trials, most of the improvement in symptoms in the “placebo” group is actually due just to the passage of time, essentially a regression to the mean. I’m basing that statement on small studies where they compare symptom changes in people on the wait list to be enrolled in a trial vs. people who were enrolled in the placebo arm of a trial. So most of the “placebo effect” in that case is actually due to a nonspecific effect that is distinct from the placebo effect as you’re defining it. I think it’s a valid question whether it’s worth separating out “the placebo effect” from other nonspecific effects, trying to define it rigorously, and trying to measure it as you seem to be advocating. In my opinion, it’s not worth doing, IF your goal is just to determine whether a given treatment is effective or not. All you have to do is negative controls. The definition of negative controls is context dependent, as it should be, and this is fine as long as it’s clear what is and is not being controlled for.

  28. Nescio says:

    daedalus2u,

    If I understand you correctly, you are suggesting that there will be a physiological response to some placebo interventions, for example an injection of isotonic saline will induce a mild stress response which will induce secretion of various endocrine hormones and neurotransmitters which have a physiological effect. Something similar may occur when a patient sees a CAM practitioner; for example an acupuncturist inserting needles.

    But is this physiological response qualitatively or quantitatively different to the way the patient responds to one of the many stressors or pleasures that he/she frequently faces in everyday life?

    We are all subjected to up and downs in the course of living our lives. Our cortisol levels rise and fall diurnally, as well as in response to daily stresses and annoyances, as do other hormones and neurotransmitters. Does a placebo really have any measurable physiological effects that compete with the normal background noise of daily life?

    Short version – does any CAM intervention (or placebo) have greater physiological effects than taking a pleasant walk in the sunshine?

  29. daedalus2u says:

    evilrobot, I would call the virtual reality sessions a placebo because the effects are not mediated through some external physical mechanism but by the response to the visual images and sounds that are perceived. I appreciate this is an arbitrary definition.

    Nescio, How much response a specific stimulus will have depends on the state of the organism before the stimulus happens. All sensory systems have automatic gain control. All physiological control pathways have automatic regulation. When it is very dark, the eye becomes more sensitive. When it is very bright, less sensitive. When you anticipate being attacked, you will have a hair-trigger to enter the fight-or-flight state and then won’t leave it after the danger is gone.

    The “desensitization” that evilrobot talked about is actually normal regulation. When that regulation gets screwed up due to physiology being out of whack because of low NO (my hypothesis for just about everything), the automatic gain control doesn’t work as well. It still kind-of-sort-of works, so if you carefully enter the fight-or-flight state in a controlled manner and then leave it, and do this repeatedly, the normal non-fight-or-flight state can be re-entrained.

    The fight-or-flight state is entered by lowering the NO level. Your body can do that in a heartbeat by releasing catecholamines like epinephrine. They make ROS, lower the NO level, disinhibit mitochondria, cause vasoconstriction in places that don’t need blood, accelerate the heart and dump sugar into the blood stream. Coming down from a fight-or-flight state is a lot more difficult. There are many ways to enter, but only one way to leave, you have to feel safe, your brain gives the “all clear” and the various systems can start standing down. But if the NO level doesn’t go back up, they can’t “stand down” and you stay in the fight-or-flight state. In the long term, that causes damage that isn’t repaired, and “tunes” the system to enter the fight-or-flight state even faster. Being in a chronic fight or flight state is what is commonly called PTSD.

  30. evilrobotxoxo says:

    @daedalus:

    It’s not just that your definition is arbitrary, it’s ambiguous and fundamentally nonrigorous. I understand that there is a difference between psychotherapy and putting a substance into a person’s body, but to say that psychotherapy is not a physical intervention is simply incorrect, as you are well aware. The way you define psychotherapy in your most recent post as “not mediated through some external physical mechanism but by the response” would also apply to an intervention like cardiac defibrillation. I’m actually not entirely sure what you mean when you say “external.” No treatment acts externally, right?

    My more important point, which you have not addressed, is why you think it’s important to try so hard to come up with a definition of a subset of nonspecific effects in clinical trials. All we have to do is be able to isolate the nonspecific effects experimentally, not to come up with rigorous definitions of subsets of them. Why is a rigorously defined placebo concept necessary?

  31. daedalus2u says:

    I would say that cardiac defibrillation is not a placebo, that it is an externally acting physical thing, a couple hundred Joules of electricity.

    What I am trying to get at is “placebo” as things mediated through internal physiologically regulated processes as opposed to external effects that directly impinge on those internal processes.

    The problem is that we can’t separate the non-specific effects except arbitrarily and a priori. If we use arbitrary criteria, then CAM will use arbitrary criteria and there is a stalemate.

    The problem is that we don’t know what those internal processes are, so it is difficult to define them except by default which is what I am trying to do, define them by what they are not.

    A second problem is that most of those things are under automatic and autonomic control. They are not subject to conscious control usually, but some people have learned to trigger some of them sometimes. Entering the fight or flight state is something that people can do spontaneously just by observing something (a bear crashing through the door and running after you). Some people can trigger the fight-or-flight state spontaneously by “psyching” themselves up. An injection of epinephrine can trigger the fight-or-flight state as can physical trauma and trauma induced pain. These last 3 have external triggers, so they are not placebos, the first two are triggered by the CNS and cognition so they are placebos.

    I appreciate that the distinction I am making between sensory detection of an image of a bear on the retina as being a placebo but a sensory detection of pain from a whip striking skin being not a placebo is arbitrary.

    What I am trying to address is the coupling between what dualists call “the mind”, (but which we know is just an illusion, there is no “mind”, there is only the brain and stuff that the brain does) and the state of health of the body. We know there is coupling, we know the coupling goes both ways. The state of health of the body affects the operation of the brain and vice versa.

    Any positive effects of placebos are mediated through the coupling between the health of the body as directed by the operation of the brain, ANS and CNS. I think it is easier to demonstrate that the positive effects of acupuncture, which can be observed both with needles and with toothpicks are effects mediated through the placebo effect and not effects mediated through qi or meridians or some magical hocus pocus.

    The reason I want so very much to do this is because I have a treatment that will trigger those non-specific effects pharmacologically. What it actually does is increase the gain on the process that triggers the transition out of the fight-or-flight state. If you are already not in a fight-or-flight state it won’t do much, but if you are in one, and have been in one long enough that you are getting bad health from it, things like PTSD, depression, CFS. Then increasing the gain on the system that turns off fight-or-flight would be a miracle cure, even if it had no specific effects.

    Everything that stress makes worse, is made worse by stress because the internal control systems sense the state of stress and divert resources away from processes that are low priority during stress (things like healing) and toward processes that are high priority during stress (things like muscles to run from a bear). If your internal resource allocation paradigm is diverting resources away from healing and saving them in case you need to run from a bear, your healing capacity is going to be crappy. Divert those resources to healing and you will be able to heal like gangbusters. If you are already not in a state where healing rate is compromised, then switching resources to healing won’t do anything. If you are in a state where healing rate is compromised because of internal allocation of resources away from healing, there is absolutely nothing that will improve healing unless and until that internal allocation is changed.

    I have been trying to get this point across for years and no one wants to acknowledge that it is even possible that there is such a thing as differential resource allocation for healing. But there must be. We observe differential rates of healing depending on psychosocial stress. Therefore there is differential regulation of the rate of healing depending on differential levels of psychosocial stress. People acknowledge this but then say “but that is what stress hormones do”. Yes, exactly, that is the function of those stress hormones to turn off healing so that resources are made available for things other than healing. Things which must be more important than healing or physiology would not have turned off healing to divert resources to them. But what can be more important than healing? Escaping from a bear is more important than healing.

    If you are in a state where resources are allocated away from healing, and have been in that state for a long time, what would you call a psychogenic stimulus that changed your physiological state from one where healing is turned off to one where healing is turned on? We know there is the opposite switch, an injection of epinephrine can turn on the fight-or-flight state in a heart beat. An image of a bear chasing you on your retina will turn on the fight-or-flight state in a heart beat.

    What do we call a psychogenic trigger that turns off the fight-or-flight state and turn on the healing state? That is what I want to call a “placebo”.

  32. pmoran says:

    Any positive effects of placebos are mediated through the coupling between the health of the body as directed by the operation of the brain, ANS and CNS.

    and

    I have been trying to get this point across for years and no one wants to acknowledge that it is even possible that there is such a thing as differential resource allocation for healing.

    The recently discussed NEJM article should be a warning to you that placebo responses don’t fit your model that well, just as it is a shot across the bows of all those having exaggerated notions concerning the range and potency of mind-body influences in so-called “healing”.

    Here we have one of the most placebo responsive conditions known to man (asthma), involving an organ under considerable autonomic nervous system influence (the bronchus). Placebo had a strong effect upon reported symptoms even while having a negligible effect on the ability to expel air from the lungs as measured by FEV1.

    If the patients truly felt so much better, it was at the level of their perceptions, their experiencing of symptoms, not from anything being “healed”.

    Not that NO levels might not have an effect in some placebo responsive conditions, for example the IBS.

    Could circulating levels be elevated enough to reduce bowel smooth muscle activity in the irritable bowled syndrome (while not also inducing cardiovascular collapse?)? I don’t know enough to even speculate , but this would in any case be a conventional pharmacological effect similar to that once sought with serotonin antagonists in this condition.

    We could not relate that to placebo and I think the same could be said of any direct physiological effect of NO mediated treatment no matter where it is exerted.

    So I would join others advising that you skip your theory of placebo. Concentrate on the pharmacology.

  33. weing says:

    It appears that the placebo effect is the effect of lying to the patient. Telling a patient that something is a placebo and that placebos can have powerful effects is also lying. Putting a label from a $1000 bottle of wine on a $5 wine bottle is also lying, and results in increased enjoyment of the $5 wine and detectable on fMRI. So, it is the lie that counts. The effect of lying and not lying to subjects on line length perception, has been studied in the conformity experiments of Solomon Asch back in the 1950s.

  34. daedalus2u says:

    Placebo responses do fit my model well, the NEJM article didn’t present the data in a way that showed a good placebo effect and the data presented in the paper is rather limited. It was a highly contrived test, they measured air flow every 20 minutes over 2 hours and took the maximum objective and subjective response over that period as their single data point. Why didn’t they show the high, low, average and standard deviation? My suspicion is because then there would be an even greater and even more obvious difference between the albuterol treatment and all the others. In the 7 breath tests, if the albuterol treatments had 7 of 7 with improvement and the non-albuterol tests had 1 of 7 with improvement that would not be evident from the data they presented (they presented only the maximum FEV improvement and only the maximum subjective improvement).

    One subject did experience a treatment response greater than 12% for each of the treatment legs, albuterol, placebo, sham acupuncture and no treatment. Of the possible 3*39 = 117 responses for each of the non-albuterol legs, there were 28 for placebo inhaler, 23 for sham acupuncture and 21 for no treatment. All we are told about these non-albuterol responses is the number that were greater than 12%, and the average of all of the responses.

    What is the actual physiological mechanism for the 12% increase in FEV in the non-albuterol legs? “Regression to the mean” is not a mechanism unless you are considering it to only be instrumental error.

  35. daedalus2u says:

    NO should have an effect on IBS. Inflammation is a low NO state. The respiratory burst turns on the inflammatory state by lowering the NO level, uncoupling nitric oxide synthase with peroxynitrite and turning xanthine oxidoreductase into xanthine oxidase. The low NO then potentiates mast cells which perpetuates the inflammatory and irritable state.

    The NO effect on IBS might not be mediated through NO per se. S-nitrosothiols also have NO signaling properties and these can occur without release of free NO by transnitrosation. Most of the gut is nitrergic and is influenced by NO donors. There is a literature report where a MD who took nitroglycerin on a regular basis noticed a temporal correlation with flatus. The gut also tends to be pretty anoxic and so nitrite could be reduced to NO.

    It may not be possible to administer NO species in a pharmacological manner and get drug-like therapeutic effects on IBS without unacceptable side effects. It may be that IBS is completely caused by insufficient NO-species and that administered the “right” way (topically with my bacteria) will be effective and without adverse side effects. The problem is in the control of the delivery mechanism(s).

    As an analogy, consider O2 delivery. We know that cells need oxygen, and that oxygen has to be delivered to those cells. Would it be possible to deliver O2 to cells without using the normal physiology of O2 delivery? In vivo? Humans evolved to take O2 out of air with the lungs, put it in red blood cells by coordinating it to hemoglobin, circulate those red blood cells throughout the body, deliver O2 and then take up CO2. Administering O2 as an O2 containing gas is pretty trivial because physiology evolved to extract O2 from air. Could you simply administer O2 intravenously? Not really because O2 needs to be exchanged for CO2. There is no O2 species that could be administered intravenously and deliver sufficient O2. There was some research using concentrated H2O2, but there were problems with that. Getting enough O2 in was difficult, getting the CO2 out was problematic and there were adverse side effects.

    There are devices called heart-lung machines. They are large and cumbersome and work very poorly compared to a real heart and real lungs. They can only be used for short periods of time on an in-patient basis and even then there are adverse side effects. The blood cells get damaged pretty rapidly and the leaking hemoglobin has adverse effects.

    If you didn’t know that O2 could be delivered by an O2 containing gas to the lungs and tried to do it a different way, you would not be successful. The physiology of NO is more complex than that of O2.

    Consider delivery of serotonin at synapses. Is there any way to artificially deliver more of any neurotransmitter to each synapse that needs a different level? No, there isn’t. There are things like SSRIs, which modify the physiology of serotonin a little bit. There is still no actual measurement of serotonin levels at the synapse in vivo, so the idea that SSRIs work via changing the serotonin concentration at the synapse remains speculative.

  36. pmoran says:

    Sorry, D2, IBS =irritable bowel syndrome (as clarified in my next paragraph), which is not an inflammatory state. But thanks for the further information.
    —————————–
    D2: Placebo responses do fit my model well, the NEJM article didn’t present the data in a way that showed a good placebo effect and the data presented in the paper is rather limited.
    ——————————-
    Reply: we might look at that if there was any other reasonably strong evidence for placebo having “healing” effects on disease processes i.e. results not within the range expected from experimental artifact, randomness, or the Hawthorne effect that almost certainly explains the small objective benefits on FEV1 from placebos in the NEJM study.

    But there isn’t. That is one of many gaps in the chain of evidence needed to sustain your hypothesis.

    So your resistance to that evidence just looks bad, even more so because what we know about the authors does not support the cherry-picking of end-points you postulate. It is fairly certain that the principal authors (at least Kaptchuk) expected stronger objective responses to placebo.

    We can do no more than warn you that your hypothesis seems to have a much stronger grip on your thinking than the available evidence would support. It is a pity, as you are obviously a very smart fellow, and a sound scientist in all other respects. Are you a Linus Pauling?

  37. daedalus2u says:

    PM, no, that is simply not correct.

    You accept that psychosocial stress can exacerbate asthma. If a person is in a asthmatic state which asthma has been exacerbated by psychosocial stress, and that psychosocial stress is removed, what happens? The asthma gets better. What do we call the agent that removes the psychosocial stress? I call it a placebo and say it acts by removing psychosocial stress and the adverse effects of psychosocial stress.

    You want to call it something other than a placebo. You want to call it experimental artifact, error, regression to the mean, delusion, fantasy, Hawthorne effect, bias, recall error, something, anything other than the opposite of what ever adverse effect psychosocial stress is having on asthma.

    If psychosocial stress causes an exacerbation of symptoms, what do we call the opposite of that?

    If psychosocial stress causes an exacerbation of asthma, then there is coupling between the physiology of psychosocial stress and the physiology of asthma such that psychosocial stress exacerbates the symptoms of asthma, subjective symptoms and objective symptoms.

    If there is physiology by which psychosocial stress exacerbates asthma symptoms, by what basis do you claim there is no physiology that is the opposite of that?

  38. pmoran says:

    D2, are you interested in seriously looking at other viewpoints? I am talking to the evidence in as precise terms as possible and you want to refute me with more unsupported, highly simplistic theory of how psychogenic influences including inert “treatments” might affect disease symptoms and some illnesses.

    You are aware, D2, that stress hormones i.e. cortisol and adrenalin have a beneficial effect upon asthma clinically?

    If stress produces bronchospasm it is presumably via vagal overactivity in which case, on the D2 theory, atropine might logically reverse it, but that is not the case.

    I wouldn’t know how often psychosocial stress is a triggering factor in asthma compared to allergies, infections etc, but it doesn’t matter for the relevance of the NEJM findings.

    Even if all of the symptoms examined therein were due to stress, placebos were relatively ineffective in terms of dealing with the underlying bronchospasm. Why so?

    This is all evidence against your theory having any general or major application in understanding placebo responses.

  39. evilrobotxoxo says:

    I would say that cardiac defibrillation is not a placebo, that it is an externally acting physical thing, a couple hundred Joules of electricity.
    What I am trying to get at is “placebo” as things mediated through internal physiologically regulated processes as opposed to external effects that directly impinge on those internal processes.

    What I’m saying is that defibrillation works because it is a physical stimulus that interacts with internal physiological processes. Psychotherapy is exactly the same thing; the only difference is that defibrillation is mediated by an electric field, while psychotherapy is mediated by electromagnetic radiation (i.e. photons in the visible range) and compression/rarefaction waves (i.e. in the part of the spectrum humans can hear). Otherwise, the mechanism is essentially identical; applying a stimulus to a spontaneously active electrical organ to reorganize the dynamics of internal activity. Do you remember that pokemon episode where a flashing light stimulus evoked seizures in children? Is that a placebo? I’m not just trying to engage in sophistry here, I’m trying to point out that because your definition is arbitrary, as you admit, then different people can have legitimate disagreements about what constitutes a “placebo.”

    The problem is that we can’t separate the non-specific effects except arbitrarily and a priori. If we use arbitrary criteria, then CAM will use arbitrary criteria and there is a stalemate.

    I have no problem with using arbitrary criteria, as long as it is clear in each case what those criteria are. There are multiple possible negative controls for any given experiment, and we really don’t need to separate the non-specific effects if our goal is to measure how effective the specific treatments are.

    What I am trying to address is the coupling between what dualists call “the mind”, (but which we know is just an illusion, there is no “mind”, there is only the brain and stuff that the brain does) and the state of health of the body. We know there is coupling, we know the coupling goes both ways. The state of health of the body affects the operation of the brain and vice versa.
    Any positive effects of placebos are mediated through the coupling between the health of the body as directed by the operation of the brain, ANS and CNS.

    I think we’re making progress here. So the reason you want to define the placebo effect rigorously is because you think that it is a phenomenon worthy of study independent of its role as an experimental confound. In that case, I completely agree with you that it is necessary. However, I think I still disagree with you on a couple of points. First, your criteria of what constitute a placebo are arbitrary, as you acknowledge, but despite your intentions they’re arbitrary in a tacitly dualistic way, which I think is ultimately the issue that bothers me. Second, you’re trying to redefine an existing word to take on a meaning that you think is important for some word to have; maybe this is part of the reason you’re encountering so much resistance, and you might be better off using a different term. Your definition of “placebo” seems to me to be a subset of the nonspecific effects that we call “the placebo effect,” which also includes things like biased reporting of symptoms.

    What do we call a psychogenic trigger that turns off the fight-or-flight state and turn on the healing state? That is what I want to call a “placebo”.

    I think that’s a subset of what you want to call placebos. For example, in exposure therapy for PTSD, which you consider a placebo, the goal is to use psychogenic triggers to trigger fight-or-flight states, such that those states become more difficult to trigger by random environmental stimuli later on.

    BTW, I completely agree with you that mood/anxiety/stress states influence physical health. However, I would also argue that the effect of “mental” states on “physical” health is mediated primarily through behaviors that effect health, not through a direct effect on healing ability. The data supporting this is certainly much stronger.

  40. daedalus2u says:

    PM, yes I am and always have looked at other viewpoints. It is you who are not considering even the possibility that there is any physiology behind what is called the placebo effect. You cherry pick data and arbitrarily decide when you can disregard what ever data doesn’t fit your conceptualization of the placebo effect and when you decide what ever effect is observed isn’t enough.

    I don’t know the detailed physiological pathways that trigger and anti-trigger asthma. One of the fundamental problems is hypersensitivity of mast cells to degranulation, and that is mediated through low NO.

    http://www.ncbi.nlm.nih.gov/pubmed/12100016

    Because the actual physiological control pathway involves authentic NO, it will not be possible to achieve normalized regulation unless normal levels of authentic NO are restored. The background levels of NO that are important are difficult to measure but are usually in the sub nM/L range (depending on the tissue compartment).

    I don’t really know what “vagal overactivity” actually means, and I don’t think anyone else does in terms of what physiological pathways are triggered and anti-triggered, to what degree, in what tissue compartments and via what mechanisms and for what duration and with what side effects.

    There is evidence that vagal stimulation does cause the release of NO.

    http://www.ncbi.nlm.nih.gov/pubmed/21154273

    The “beneficial” effects of adrenaline and cortisol on asthma are only acute. What would happen if you tried to treat asthma chronically with adrenaline and/or cortisol as monotherapy? Would the patient have a satisfactory outcome? No, the patient would not. The average patient would fare worse than being treated with a placebo. There might be improvement for a few days, but there would be chronic adverse side effects.

  41. daedalus2u says:

    Evilrobot, I don’t think there is anything in your last comment that I disagree with except that I think the therapeutic effect of desensitization treatment also includes learning how to anti-trigger the fight-or-flight state (i.e. come “down” after it has been triggered), maybe more so than in retarding its triggering or in improving the discrimination of triggering (but there are obviously components of all three). Triggering fight-or-flight has to be so fast that interfering with the triggering might be difficult, but anti-triggering it a few seconds later could have just as valuable therapeutic effects and might be easier.

    I have PTSD, and I still “jump” when things happen, but I can deescalate fast enough that I don’t do anything during the prompt triggering.

    I completely agree with you that a large component of how mental states affect health is via behaviors, but it is not the only component. It is the component that is mediated through physiology that I am trying to get at. That component is thought to be small because there are no well characterized ways of triggering it so it can’t be studied. I think that raising NO levels with my bacteria is one of the few ways that can work.

    The term placebo is not well and uniquely defined. The sense that I am using it is within the broader sense of the term. I think that making up a new term would be more problematic than using the term placebo in the sense that I am using it.

    The original sense of the term was an inert treatment that had apparent positive effects. It is only when placebos were used in clinical trials that the term came to mean all non-specific effects including observer error and bias. Observer error and bias was never part of the original definition. Maybe the original definition was naïve, but it was coined to give a name to inert treatments that were given, often with seemingly positive effects.

    http://en.wikipedia.org/wiki/Placebo_in_history

  42. pmoran says:

    PM, yes I am and always have looked at other viewpoints. It is you who are not considering even the possibility that there is any physiology behind what is called the placebo effect.

    Actually I am contending that placebo responses, to the extent that they are “real” at all, are enabled by human neurophysiology, but also markedly limited by it.

    The evidence suggests that much of any response to placebo occurs at the cerebral level involving altered symptom perception, which your theories seem not to recognise.

    It also means that any peripheral, or systemic, “effects” of placebo, i.e. those involving biochemistry, physiology and actual tissue damage will be limited by neurological anatomy and neuro-humoral influences – i.e. they are specific to a few organs and functions and rather weak in the main. This is what is found in practice.

    For all I know NO may be involved in some of these pathways, as might electrical impulses along neurones, but we don’t regard electricity as explanatory of placebo responses in any meaningful way either, do we?

    This is how I, and I suspect all clinicians, read the evidence. I don’t think I have anything more to add.

Comments are closed.