The subject of placebo is a complex but fascinating one, particularly for those interested in alternative medicine. Most sceptics believe that alternative therapies rely heavily, if not entirely, on the placebo effect. Some alternative practitioners, when unable to produce convincing evidence that their treatment is effective, seem to have now settled to admitting that their therapy works (mostly or entirely) via a placebo effect. They the hasten to add that this is perfectly fine, because it is just an explanation as to how it works – a mechanism of action, in other words. Causing benefit via a placebo effect still means, they insist, that their therapy is effective.
In a previous post, I have tried to demonstrate that this belief is erroneous and where the notion comes from. It originates, I believe, from a mistaken definition of ‘effectiveness’: for many alternative practitioners ‘effectiveness’ encompasses the specific plus the non-specific (e. g. placebo) effects of their therapy. In real medicine, ‘effectiveness’ is the degree to which a treatment works under real life conditions.
The ‘alternative’ definition is, of course, incorrect but alternative practitioners stubbornly refuse to acknowledge this fact. Here are just two reasons why it cannot be right:
- If it were correct, it would be hardly conceivable to think of a treatment that is NOT effective. Applied with empathy and compassion, virtually all treatments – however devoid of specific effects – will produce a placebo effect. Thus they will all be effective, and the term would be superfluous because ‘treatment’ would automatically mean ‘effective’. An ineffective treatment would, in other words, be a contradiction in terms.
- If it were correct, any pharmaceutical or devices company could legally market ineffective drugs or gadgets and rightly claim (or even prove) that they are effective. Any such therapy could very easily be shown to generate a placebo-effect under the right circumstances; and as long as this is the case, it would be certifiably effective.
I do sympathise with alt med enthusiasts who find this hard or even impossible to accept. They see almost every day how their placebo-therapy benefits their patients. (It seems worth remembering that not just the placebo phenomenon but several other factors are involved in such outcomes – take, for instance, the natural history of the disease and the regression towards the mean.) And they might think that my arguments are nothing but a devious attempt do away with the beneficial power of the placebo.
The truth, however, is that nobody wants to do anything of the sort; we all want to help patients as much as possible, and that does, of course, include the use of the placebo effect. In clinical practice, we usually want to maximise the placebo effect where possible. But for this goal, we do not require placebo therapies. If we administer a specifically effective therapy with compassion, we undoubtedly also generate a placebo response. In addition, our patients would benefit from the specific effects of the prescribed therapy. Both elements are essential for an optimal therapeutic response, and I don’t know any conventional healthcare professionals who do not aim at this optimal outcome.
Giving just placebos will not normally generate an optimal outcome, and therefore it cannot truly be in the interest of the patient. It is also ethically problematic because it usually entails a degree of deception of the patient. Moreover, placebo effects are unreliable and usually of short duration. Foremost, they do not normally cure a disease; they may alleviate symptoms but they almost never tackle their causes. These characteristics hardly make placebos an acceptable choice for routine clinical practice.
The bottom line is clear and simple: a drug that is not better than placebo can only be classified as being ineffective. The same applies to all non-drug therapies. Double standards are not acceptable in healthcare. And the demonstration of a placebo effect does not turn an ineffective therapy into an effective one.
I know that many alternative practitioners do not agree with this line of thought – so, let’s hear their counter-arguments.
21 Responses to Placebo effects do not turn an ineffective therapy into an effective one
Leave a Reply to Doc Dale Cancel reply
This site uses Akismet to reduce spam. Learn how your comment data is processed.
Dear Professor Ernst,
A couple of points. “It originates, I believe, from a mistaken definition of ‘effectiveness’: for many alternative practitioners ‘effectiveness’ encompasses the specific plus the non-specific (e. g. placebo) effects of their therapy. In real medicine, ‘effectiveness’ is the degree to which a treatment works under real life conditions.”
Let’s talk about “real medicine” in “real life conditions” for a moment. Are you really trying to argue that effectiveness studies of pharmaceuticals control for non-specific effects? Consensus (and the definition of the term and the basics of trial design) is against you there, so would you grace us with a reference? Can you provide me with a single trial of any drug in real world conditions that controls for all non-specific effects and is only measuring treatment effects?
I think that’s going to be pretty difficult since it’s impossible by definition. As your computational biologist friend pointed out in another post: “if you measure something consisting of two variables . . . you have to estimate at least one to be able to determine the second. . . . That means the Efficacy/Efficiency is confounded with the placebo effect in a way no mathematical trick whatsoever could untangle.” In real world studies, you can never measure specific treatment effects, only overall observed effects that include treatment and non-treatment effect. I’m not sure why you’re hitting yourself so hard against this.
“If it were correct, it would be hardly conceivable to think of a treatment that is NOT effective. Applied with empathy and compassion, virtually all treatments – however devoid of specific effects – will produce a placebo effect. Thus they will all be effective, and the term would be superfluous because ‘treatment’ would automatically mean ‘effective’. An ineffective treatment would, in other words, be a contradiction in terms.”
And here’s where you’ve forgotten one of the most basic parts of trial design: the trusty control group. To say a treatment is “effective” is not absolute but a short-hand for “effective compared to” whatever the control was. So when a Cochrane Systematic Review finds that acupuncture is more effective than usual care in the treatment of tension type headache, it means that “acupuncture is effective for the treatment of tension type headache” compared to pharmaceutical usual care. A treatment may be effective compared to one control but not compared to another. But most often in pragmatic studies, the control is usual care which is why researchers are able to shorten this and people usually (present company excepted) understand what they mean.
In order to make your point (such as it is), you seem to be interpreting effectiveness as showing equivalency to usual care. Depending on what’s being studied and how the researchers set up the trial (or review), they can be looking at inferiority, equivalency or superiority. In the case of the review mentioned above, acupuncture was found to be superior to usual care. That’s really not something that every treatment can do. Indeed, if you are trying to argue that acupuncture’s effectiveness in this area is meaningless because any treatment achieves it, can you provide some examples?
Finally, your argument rests on there being no evidence of efficacy for acupuncture, which is simply not the case and something that you’ve repeatedly failed to demonstrate when asked to support this point of view. In trials of acupuncture that are adequately powered, acupuncture outperforms active pseudo-acupuncture controls.
To Quote: To say a treatment is “effective” is not absolute but a short-hand for “effective compared to” whatever the control was.
That’s very true – which is why in most cases (I won’t say all – there’s plenty of room for bad science out there) – the control for a drug provides a treatment that is taken and looks identical to the drug being tested. Remember that in double-blind studies, they have to be close enough that even those administering the treatment don’t know.
As there is no difference in delivery of the treatment and ideally there are no major differences in the populations of your control and treatment groups, the only remaining variable is the placebo effect – which gives you the basis of comparison. You can clearly show “Effective vs Placebo” because treatment was otherwise identical.
There is a larger margin of error when the study can’t compare placebo vs treatment quite as effectively due to the nature of the treatment. Ideally, you’ve got to figure out some form of acupuncture or chiropractic placebo that still makes your test subjects believe they’ve been treated with the real thing, in order to give the most accurate results possible.
You know I feel sorry for you Ernst, everyday you have the unenviable job of dealing with an ever increasing bunch of assholes who simply WON’T accept they FACTS .. No amount of overwhelmingly evidence to the contrary is ever enough ..😥
I am a registered medical practitioner, but also a magician, and am interested in how to create illusions and deceive people.
I draw a distinction between the therapy and the therapist; the practice and the practitioner – as I make clear in ‘Real Secrets of Alternative Medicine’ (Amazon) pages 27-34:
Type I effects are due to a constructive therapeutic relationship between the practitioner and patient (TLC, context, non-specific or placebo effects). Type II effects are due to the specific action of the therapy aside from type I. (And aside from regression, natural history, resolution which happen in any event).
Any claim that placebo effects represent an effect on a specific disease process is false. There is no plausible evidence of any such effect (by definition) – but of course, many patients do ‘feel better’. Being nice to people is nice. It’s akin to a magic trick.
Additionally, I am an active medical polititian (a member of the BMA’s Representative Body) and I realise patients want the consolation. hope and loving care which placebos can offer. Simply denying patients any such benefits does not help develop modern health care policies, nor help resit corrosive proposals for ‘integration’.
If placebos help make patients ‘feel better’, well and good – but, with one very important proviso:
Patients must give fully informed consent and understand the nature of placebo effects.
As Edzard points out, when using placebos, deception is usually employed (and to be deprecated), but I advise that practitioners need not and should not be deceptive.
Camists (who practice camistry – CAM) are disinclined to obtain fully informed consent, possibly feeling, and fearing, that patients will experience less benefit from placebos if their practices are openly declared as such.
So be it. Intellectual and professional integrity demand this honesty. Any other practice is unethical.
All patients should read my book!
The issue as to whether ‘needling patients’ skin’ (as opposed to caring for patients) has a specific effect is considered in other posts.
EE: “Applied with empathy and compassion, virtually all treatments – however devoid of specific effects – will produce a placebo effect.”
Curious, what research shows that applying empathy and compassion to a placebo results in any added effect?
“Curious, what research shows that applying empathy and compassion to a placebo results in any added effect?”
I believe the implication was ’empathy and compassion’ produced a placebo effect in and of themselves, not any added effect.
There is some evidence of a compassionate approach giving ‘added value’ to both placebo acupuncture and verum acupuncture though:
Kaptchuk, T.J., et al (2008) Components of placebo effect: randomised controlled trial in patients with irritable bowel syndrome British Medical Journal Vol. 336 pp. 999-1003
Niall
Yes, if I recall from that study, they found that the main driving force of the placebo response was the patient personality of extraversion.
“Yes, if I recall from that study, they found that the main driving force of the placebo response was the patient personality of extraversion.”
No, the authors made no comment about the ‘driving force of the placebo response’ and there was no mention of patient personality types.
Apologies, I had misremembered the details slightly, there was infact no acupuncture verum group. The trial was a comparison of three groups of patients with IBD – one was an observation only group, one had placebo acupuncture carried out in an impersonal way and the third had placebo acupuncture carried out with warmth and empathy. The second group did better than the first. The third group though, scored much higher in all the end points than the second, indicating warmth, empathy etc can enhance the placebo response.
Niall
I’ve been looking for a similar study only using a placebo pill…still looking.
Why is the application of compassion taken to be an enhancement of the placebo response? Why should it not be regarded as having an effect in itself? What if compassion, including presumably counselling, were given alone? Must a patient be provided with some physical therapy in order to feel treated? If so, why not “massage”, a sanctioned way of allowing bodily contact between therapist and patient?
No research is necessary — for every pupil who actually bothered to learn, and to understand, the vector mathematics that they were taught during high school maths and science lessons.
Interesting that folks from Harvard are looking into this subject considering you think it can be determined from what was taught in high school math and science.
What do you mean by “looking into this subject”? Is is: whether or not applying empathy and compassion results in any added effect; or how much effect it adds to each one of a variety of placebos? The former doesn’t need “looking into”; the latter would be interesting, if properly researched. NB: The term “placebo response” is *not* the same as the frequently bandied about word “placebo”.
Pete: I’m just trying to understand this comment:
“Applied with empathy and compassion, virtually all treatments – however devoid of specific effects – will produce a placebo effect.”
Your responses aren’t helping, not that that is surprising.
Doc Dale,
Let T = a treatment/modality that is totally useless — not even a ‘placebo’,
E&C = empathy and compassion,
PR = the placebo response (aka placebo reaction).
Now, the PR from E&C combined with T will always be greater than the PR from T alone. Why? Because E&C / TLC / sharing a problem is beneficial to everyone who is suffering an acute or a chronic illness. I think it unlikely that “folks from Harvard are looking into this subject”.
Let P = a treatment/modality that is a physiologically inefficacious placebo for the disease, but it has been scientifically shown to produce a placebo response.
Now, the PR from E&C combined with P will always be greater than the PR from P alone. Why? Because if E&C and P are unrelated variables (they are orthogonal variables), then a reasonable approximation to their combined effect = √(E&C² + P²) — the magnitude of this vector quantity. (The phase angle of this vector is well beyond the scope of this discussion, at this point in time). I think it unlikely that “folks from Harvard are looking into this subject” because I sincerely hope they are more than smart enough to already understand it.
Let M = a medical treatment that is medically efficacious for the acute or chronic illness.
Now, the PR from E&C combined with M will always be greater than the PR from M alone. However, the overall patient response to this treatment consists mainly of its physiological medical efficacy, and secondarily, the subjective (psychological) quality of life improvement from both the M plus the PR from E&C. This is an entirely different vector quantity: it resides in an entirely different domain from the preceding ‘T’ and ‘P’ cases. It is only the vendors of, and the apologists for, quackery who endlessly attempt to sweep their fundamental domain error under the carpet.
The above explains why A + B versus B trials are so frequently used by sCAM researchers, and espoused by sCAM apologists.
Your responses to me are, likewise, not helping and not at all surprising.
Pete, that doesn’t reflect what Ernst wrote. But if that is what he meant, well, I won’t split hairs over it.
Quite a good discussion by Jean Brissonnet, translation by Harriet Hall :
https://www.sciencebasedmedicine.org/placebo-are-you-there/
Covers the topic of placebo and medicine very well!
Placebo effects are often recognised as unworthy by all professionals practicing traditional medicine and it is easy to see why – everyone has the freedom to believe into the philosophy they recognize as true and that applies to their genuine believes the best. However, placebo effects, even unscientific in many cases, they are still a real and worthy part of medicine, because medicine’s main goal is to heal, no matter the way, while it still can offer cure, control of the disease, or at least the comfort of symptom relief.
Of course, one of the best placebos masquerading as medicine for the treatment of low back pain is Paracetemol.
http://www.dailymail.co.uk/health/article-3020504/Paracetamol-pain-s-no-better-placebo-Experts-say-treatment-does-improve-recovery-time-sleep-quality-life.html
Considering that the mechanism for low back pain is the same as most other neuro-musculo-sketelal complaints, I.e. biomechanical dysfunction, then Paracetemol will also be ineffective for the treatment of neck pain, knee pain, shoulder pain etc.etc.
Why? Does paracetamol only affect one kind of pain/discomfort/suffering? I reckon it even works as a mild mood lifter.
You may be interested in reading Jo Marchant’s exploration of placebo effects:
http://jomarchant.com
https://chriskresser.com/the-healing-power-of-the-placebo-effect-with-jo-marchant/