Putting the placebo out to pasture

I’ve long had this kind of fascination with the placebo effect, it’s like a kind of magic that even grownups are allowed to believe in. In fact, it’s a magic trick so good that you’d considered a bit off if you didn’t believe in it. This being the case I found it a bit traumatic when I first read a review of placebo effects in clinical trials that found that placebo effects were, if real at all, really not very powerful (Hrobjartsson and Gotzsche 2001). This was at a time when I’d recently started down the research road – all inspired and convinced by the ultimate truth-divining machine that is science. It was like my new favourite toy fighting with its immediate predecessor. I didn’t know who’s side to take.

What this experience did was set me down a path of looking a bit more deeply at research into placebo effects. The problem was, that the more I read and thought about all things placebo, the less I felt I could actually nail down what it is that placebos actually are. Perhaps the biggest source of misinformation regarding placebo effects historically has been the tendency to attribute any change after a placebo intervention to the placebo effect. We know however that this is not the case, at least part of this change is due to natural recovery (especially in acute conditions), part will be due to statistical effects, and part is probably also due to various types of bias. The greatest contribution of Hrobjartsson and Gotzsche’s review was to clearly point this issue out. What this did at the same time, was carve a big chunk off what had previously thought of as the placebo effect, and in certain cases not just a big chunk, but the whole lot.

Even if we accept that perhaps placebo effects are smaller than originally thought though, there is still something a bit wrong, and to me this stems from something very fundamental to the whole placebo concept. The issue is that there exists a logical paradox at the very heart of the way placebo effects are conceptualised. We have an intervention that is, by definition, inert (a placebo intervention) which produces an effect which is real (a placebo effect). Now maybe this reflects my own lack of imagination, but I just can’t get my head around an effect that has no mechanism. Surely there must either be no effect (i.e. there is no placebo effect), or the intervention must have a mechanism (i.e. placebos are not inert, but real treatments).

So if there is a placebo effect? Well, then let’s go looking for a mechanism. We already have several leads; there are ideas about physiological pathways via patient expectations, classical conditioning and reduction of anxiety, if there is an effect due to the nature of patient-practitioner interaction then it must be due to some specific psychological mechanism. If interventions, or parts of interventions work via these pathways, then we should investigate how they work so we can use this information to design more effective treatments. It serves no one to consign effects due to these factors to a black box, and call them placebo. For a start, doing so opens up an ethical can of worms for clinicians; is it OK to deliver a treatment you know is inert? It also serves as a conceptual barrier to trying to understand how, and for what, certain treatments might work. From a research perspective, simply assigning the label ‘placebo’ to the control arm of a study also serves no useful purpose because doing so doesn’t tell the reader anything about what contextual factors are being controlled for.

In my opinion, we can leave the placebo effect behind us. We can look back in some years time and see it as a useful crutch we used at a time when our understanding of biology was transitioning from the mystical to the scientific. There comes a time though when crutches no longer provide support, but rather become just another thing we’re lugging around.

So in the end, my life lost a bit of magic and for a time I felt a bit like I’d killed the Easter Bunny. But then I realised that even though I stopped believing in him years ago, it hasn’t stopped me getting chocolate eggs every year.

About Steve Kamper

Having completed Physiotherapy at USyd and a PhD at the George Institute in Sydney, Steve is currently “working” in Amsterdam at the EMGO+ Institute on an NHMRC fellowship. The thing Steve likes most about being funded by a government fellowship are the endless opportunities to remind mates that they are, in fact paying for every beer he has. Work involves research into the influence of patient expectations on outcome, back and neck pain, outcome measurement and the ongoing search for European conferences to ensure all holidays are tax deductible. Steve likes to spend his spare-time running around next to canals, riding his bike, giving blank looks to people who ask questions in Dutch and making sure he gets at least twice the recommended daily dose of ICECReam (www.theicecream.org/).


Kamper SJ, & Williams CM (2012). The placebo effect: powerful, powerless or redundant? British journal of sports medicine PMID: 22893511

Hróbjartsson, A and Gøtzsche PC (2001). Is the Placebo Powerless? An analysis of clinical trials comparing placebo with no treatment. NEJM, 345 (17), 1276-1279 DOI: 10.1056/NEJM200110253451712


  1. At last, a view on placebo that says it’s time to take it seriously. I agree that scientifically, and clinically, we need to move beyond saying “it’s a mysterious something that some people have and others don’t” – like one of your other correspondents, I also like Dan Moerman’s “meaning effect” as the term for the effects we can obtain from inert treatments. It certainly saves on the philosophical entanglements of “placebo = nothing” and “effects = something”!
    My problem is that I am still struggling to help some biomedical practitioners recognise the value of psychological and social aspects of pain, let alone add in this new set of factors!

    I also think the meaning effect is more than what “we” do as treatment providers – it’s an interaction in which both treatment provider and patient bring something to the mix. I can be as well groomed, have a smart office with lots of lovely certificates on the wall, but if the person coming to see me “wants” or “expects” a laidback hippy type in tie-dyed clothing and dreadlocks and crystals hanging from my windows and talk of mystical “energies”, the meaning effect will be negligible.

    As the treatment provider, my responsibility is to listen carefully to what the person says (and doesn’t say), to establish what it is he or she expects, wants, or needs. And that patient may not be aware of his/her expectations, or beliefs and preferences.

    We have a great deal to learn about “nonspecific effects” or “meaning response” – whatever we call it, it has a profound effect on outcomes, and we need to know more about it to use it wisely.

  2. Evert Jan Das says

    could there be a placebo effect in this one?


    Greetings from the Netherlands

  3. Massage Therapy is dead. And we killed it. Just like Physiotherapy. I am trying to get the same discussion er brawl going with Massage Therapists at R.M.T.A.O. Facebook. What is the theory behind manual therapy. Is it ‘tooth fairy’ science. I am another human primate social groomer and proud of it. Sunny Maya, a massage therapist.

  4. Fascinating that the analogies that crop up on these fora have a parallel evolution. On SomaSimple this debate draws up the image of ‘stabbing Tinkerbelle’ here the Easter Bunny is under threat. I hope that these myths of childhood and our mourning their passing are symbolic of our professions transition to adolescence. I thought adulthood but then thought that was a bit presumptious:)
    Another recurring theme on SS is that you can ‘cross the chasm’ from tissue based methodologies and clinical reasoning to neuroscience methodologies and reasoning and bring your manual techniques with you if you re-examine your clinical reasoning and your application of them. Do no harm, cognitively as well as physically with your intervention. Solid BPS concepts. So you could still use your ‘core stability’ just as graded exposure or behavioural experiments with caveats that it really is nothing special but still borrow the glamour of Pilates. Work to change the output of the neuromatrix. You can still lay hands on just know that you are working through layers of uncertainty and probably interacting with a patients expectations, hope and belief, not their facet joints or myofascia! The personal disclocation caused by this abrupt change in professional vector is well distressing at times. To paraphrase a philosopher “Physiotherapy is dead. And we killed it.” But we remain physiotherapists with people in distress to give care to.
    What have we learned from ‘the decade of the brain’? How do we position the profession within the industry of care giving to be true to the science as it is revealed and to the patients who want the best help now? And how can we position ourselves within the profession to ease the transition?
    The death of childhood is to be mourned but the adult world beckons…the teenage years will, doubtless, be challenging.

  5. Steve
    I guess this is where the debate is heading? “Is there any merit in differential biomechanical input (treatments) to accompany all the honey and roses of other multi-sensory input”?
    There’s been a hell of a ding dong over here….. http://www.linkedin.com/groupItem?view=&gid=2387209&type=member&item=180840941&qid=de79df5a-86a9-48fc-9bb3-374f653f2997&trk=group_most_popular-0-b-ttl&goback=%2Egmp_2387209
    for months now on this topic (it’s like a bar room brawl scene from The Quiet Man -a 1950’S rural Ireland tribal movie, for the under 40’s).

    I did pose the question (lost in one of the 360 thread comments) on this very issue of biomechanical derived sensory input with therapeutic neurophysiological benefits as a putative mechanism but didn’t get much response – but I still think it’s a critical question for the profession.

    But I think the point you raise Steve, exposes a far more significant issue in terms of research design methodology – the dismissal of treatments with lack of evidence when selected variables are controlled for and the reality being that we don’t even know what all the variables are!!

    I’m sure the more accomplished statistician’s (I’m thinking of you Neil?) here can educate me but it does seem a fundamental discrepancy.
    If we acknowledge the complexity of the pain experience and then except “evidence” which was specifically gathered to eliminate these variables it must surely raise questions about the validity of conclusions?

    I feel like a dying wasp clinging on to twigg in the toilet -can all my beloved manual therapy be a sham?? Maybe the Easter Bunny can save me?

  6. Thoughts so far.
    It is going to be tricky to design trials that look at levels of empathy, compassion, trust, hope (although our collegues in other areas have done some great work on belief) and there effect on outcomes. Hard enough to define a mobilisation, harder to measure ‘engagement’.
    Moving from a profession that bases its identity on its physical interactions to one which sees its interactions as applied psychology is going to be fraught. It will be hard not to be nihilistic change. There will be lots of resistance, there already is. Cognitive neurological PT.
    We could be the ‘light’ side of marketing as what we are exploiting should actually help improve peoples experiences.
    To be a human primate social groomer (hat tip Diane Jacobs) – that is all. As much as we would all like it to be more in our hubris. With our thoughts of physical interaction we are at root socially grooming our fellow apes. Passing on some memes at the same time that we hope may be more helpful than those they add on to. Applying contact to optimally stimulate let down of oxytocin and all the other neurotransmitters. The rest is emporer’s clothing.

  7. I remember a great keynote Pat Wall gave on placebo some years ago which was both informative and earth shattering at the same time. The possibility that “finely tuned” clinical skills may not be the solution to all Ill’s was challenging at the time but has certainly come to pass as time goes by.
    It’s not a comforting thought to think that one’s clinical efficacy may be the result of being well dressed, presentable and compassionate in a professional environment….. but maybe so. However, the anecdotal evidence would suggest that without these ingredients the chances of delivering satisfactory outcomes diminishes considerably.
    I wonder if we extend this analogy to draw from the fields of sales and marketing – whether patients are really “qualified leads (have raised their hands to volunteer interest), have unsatisfied needs (emotional hot buttons) and present to healthcare providers in seach of meeting these needs (purveyors of percieved solutions)?
    A rather vulgar analogy in such a forum but maybe healthcare bonding is like friendship – there are many intangibles?

    Steve Kamper Reply:

    This is a great perspective David. As you say it is not necessarily comforting, unless you take a wider view and see it as bringing us closer to working out exactly what it is that we need to do for patients.

    I also love the marketing analogy, and I think it has legs. My thinking is; at least part of what clinicians do is try to convince patients to behave in a certain way – self-management, activity advice, physical exercises, cognitive exercises, ergonomics etc etc. Getting people to behave in a certain way is pretty much the whole point of the marketing industry. Obviously there are differences, which I assume is where you are coming from re: the vulgarity, but surely there must be something we can learn?

  8. I’m surprised a bit to see questioning the power and biology of placebo – the research has exploded in this area with the “decade of the brain”. While I don’t have the time tonight to pull the references, there is (to my understanding of the science) very strong evidence of the biology of placebo. In particular, it involves areas of the ACC and DLPFC (anterior cingulate cortex and dorso lateral pre frontal cortex) and activation in these areas determines the strength and efficacy of how active placebo is – and this (in my opinion) also correlates to the conditions (for example, chronic pain) that a person may be dealing with, therefore teaching activation of these areas to strengthen the placebo effect (the biology of belief) is critical. Placebo is the elephant in the room – heck, it has to be controlled for and ruled OUT, it is so powerful – the bane of drug companies everywhere. Anyway, I would like to post some references when I get the chance to pull them.

  9. Hmmm.

    That will be ‘wear’ a uniform and do ‘no’ harm then.

  10. Does Fabrizzio Benedetti not write a good book on this in “The Patient’s Brain: The neuroscience behind the doctor-patient relationship”? He makes a good fist of presenting an evolutionary justification for how western medicine is practiced and the cultural evolution of the placebo effect within it.
    How do we remove social conditioning, trust and belief, the trinity of placebo, from treatment and research? Tricky. Three arm trials of no treatment, sham and treatment?
    Is the best we can do really to not exploit them? If we are doing trials to not select believers by screening them out. Acupuncturists and pharma take note. In treatment to not create artificial trust and belief. Or exploit hope. To avoid conforming to social stereotypes to magnify our cultural effect on patients? Its a tough one. Where a uniform! – be more effective? Practice within a hospital? A clinical space that ‘creates the right atmosphere’? For what? Accentuating social conditioning? Develop ‘personal gravitas’? To accentuate trust? Believe in your own skills by going on more courses? Communicate your self-belief with confidence. Exploit traditional cultural roles? Choose wisely.
    The more I think I know the more worried I become about what I think the placebo effect is. The less I want to be a part of it. I chose this cultural role (PT) and now I think it is less about me and more about the mantle. I just hold on to do know harm.
    My broader concern is of a profession that has in the past subconsciously honed its placebo treatment skills choosing to consciously hone them.

    Steve Kamper Reply:

    Steve; I’m not for a minute suggesting we try and remove any of the factors you mention. All I’m saying is that we should address them head-on, explicitly consider them as factors that may (or may not) influence treatment effect. All the things you mention; social conditioning, features of the treatment environment, trust, cultural roles etc seem plausible as far as explaining (at least part of) what is currently considered the placebo effect. So why not try and investigate these directly? Obviously some will be easier than others.

    Additionally, this conceptualization should also offer a soothing balm to the disquiet expressed in your last paragraph. Your job is to administer whatever treatments offer the best balance of benefits and harms, if we understand how all the contextual/meaning factors impact on the patient then a) you can use these to maximize your treatment effectiveness, and b) you’ve no need to worry about deceiving the patient by administering a treatment you know is a inert.

  11. ian stevens says


    Geoff , Diane Jacobs (as usual!) alerted me to the common factors model and its applicability to Physiotherapy . I talked about it last night at the Physiotherapy Pain Association meeting –together with Moermans meaning response . I think there was some interest. This to my mind is the most fascinating topic which really needs to be aired and taught early on now there are plausible models and mechanisms . I still think there are black box factors at work which are related to cultural values for example and other more tangible factors such as the design of treatment centres (the big curtained corridors don’t exactly encourage confidence in some stressed individuals ).

  12. Hi Steve,
    Great post. In many ways I agree with you. However, much of this placebo research seems to have provided a bit of biological plausibility to what the psychologists have been talking about for years: Common Factors. Like Psych, many PT interventions tested head to head are not better than than the other despite differing proposed pathways (e.g. Macedo et al review comparing motor control vs graded activity). Common factors theory essentially states that a big chunk of the variance in effectiveness of psychotherapy has to do with non-specific effects (some variance due to specific effects as well) related to provider-patient relationship, empathy etc. I realize PT is much different than Psychotherapy, but Common Factors might have an important role in PT as well – to me the placebo literature provides some biological plausibility. It also raises the possibility of avenues for effectiveness for manual therapy, for example, in the face of evidence questioning the tenability of biomechanical models for manual therapy effectiveness. Hrobjartsson’s and your methodological arguments are well taken and certainly valid. I am curious to learn your opinions/critiques of the famous (or infamous) Kaptchuk et al augmented placebo study where the issue of regression to the mean appears to be mitigated. Finally, I agree with the oxymoron of the term placebo effect – I like Moerman’s reconceptualization of placebo as a meaning response as it highlights that interventions do not occur within vacuums.
    I too no longer believe in the magic of Santa or the Easter bunny, but go to great lengths to keep the magic alive for my daughter. In some ways I’m doing the same with meaning responses for my students as a metaphor to broaden perspectives on how some things seem to ‘work’ and the biological viability of context. Later this week I am getting fitted for a Santa suit – I hope you don’t end up compelling me to put it away!

    Steve Kamper Reply:

    I can’t prove he/she doesn’t exist Neil – you know better than most that you can’t prove a negative – all I’m saying is that if you want to find out where the delicious chocolate comes from, there are better places to go looking

    Steve Kamper Reply:

    Thanks for the comments Geoff (and Ian below), I think we’re well and truly on the same page. The common factors idea holds a fair bit of water in my opinion, and taken a step further begs the question of whether we have been looking at RCTs with the wrong focus. What I mean is that perhaps we have been so caught up with what is different between tested interventions that we’re missing out on what is the same about them. I don’t here mean to question the value of RCTs for answering a particular type of question, but maybe there’s more we can learn.

    I very much like Kaptchuk’s study, and think it offers an entrée to what could be a very productive line of research. The point I try to make here and in the editorial is that maybe we should think about the added components as real interventions rather than placebos or components of placebos. For example, if we think showing empathy has an effect, why not conceptualise showing empathy as a treatment component. Let’s then go further and try to explore how, for what outcomes and for whom showing empathy has an effect. I think wrapping these factors up in a cloak called placebo only serves to confuse issues and potentially serves as a barrier to understanding. This is of course in addition to the issues with definition and logical problems with the concept as it stands. I don’t think there is too much doubt – amongst the readers of this forum especially – about the (neuro)physiological consequences of all sorts of what we might call contextual factors, so why hide them behind an archaic device.

    I have no dramas with the Santa suit, but presumably one day your daughter will want to delve a bit deeper into the mystery of the presents that appear once a year, and then she’ll find out that it isn’t magic at all. I reckon we’re ready to delve deeper too.

    ian stevens Reply:


    Steve, thanks for comments . You probably have this already but its probably a good place to post the resource for Ted Kaptchuk.

  13. Neil O'Connell says

    Great post Steve, couldn’t agree more. One the other factors that generates a larger supposed placebo effect in trials that often gets ignored is that old resentful demoralisation. Acupuncture is a rpime example. We hear loads about the large effects seen compared to no treatment or usual care in chronic conditions. But then you take a bunch of people who pretty much love the idea of acupuncture into your acupuncture trial (hence they volunteered), then randomise a bunch to get nothing or the same old treatments they have already failed with, and hey presto! there is a big effect for acupuncture. Its amazing how little it is discussed in that literature that a large part of that effect is likely to be created by the general annoyance/ frustration of the no-treatment/ usual care group. Another chip of the overhyped effect?

    One point of contention. The Easter Bunny does exist. You prove that he doesn’t.