Little White Lies: A discussion of the placebo effect
Aidan O’Donnell, Consultant Anaesthetist, St. John’s Hospital, Livingston.
Introduction
The “placebo effect” is relevant to all doctors. Daily in our professional lives we read of placebo-controlled trials, and use them as an evidence base for our practice. On the other hand, there has been a recent increase in the number of patients seeking complementary and alternative medicine (CAM), which are proven to show no benefit compared to placebo.
The temptation might be to discard the placebo effect, as an inconvenient and uncomfortable confounding variable which gets in the way of good medicine. However, I believe that it is worth having a closer look at the placebo effect, and in this discussion I propose to describe the history of the placebo effect, and look closely at what it can and cannot do, together with some theories about how it might work. Along the way, I intend to touch on complementary and alternative medicine (CAM), and to examine the ethics of prescribing placebo.
I firmly believe in scientific principles. If there is indeed a demonstrable placebo effect, I believe it should be studied and elucidated using scientific methods we already possess.
History
The term placebo (Lat: I will please) came about due to a mistranslation of the Bible in about the 4th century, when St. Jerome, translating from Hebrew into Latin, rendered a particular phrase as “I will please the Lord”, instead of “I will walk before the Lord” (Psalm 116). This phrase was included as a prayer over the recently deceased, which was recited by monks who would charge bereaved relatives money to carry out this service. The monks felt no grief themselves. The term “placebo” became a shorthand for a form of words which were insincere but intended to bring comfort[1].
By the 18th century, the term had been adopted by the medical profession, as a remedy which was “adopted to please, rather than to benefit, the patient”[2]. In 1954, the Lancet published an editorial entitled “The Humble Humbug”, in which it described placebo as “a means of reinforcing a patient’s confidences in his recovery, when the diagnosis is undoubted and no more effective treatment is possible”. It went on “for some unintelligent or inadequate patients life is made easier by a bottle of medicine to comfort their ego”[3].
By this time, placebo was looked upon by doctors as a means of fobbing off desperate patients. There was no expectation of benefit, but likewise none of harm. The patient was spared the risks of taking real medicine with side effects, and the patient felt that at least the doctor was doing something. However, the juxtaposition between deception (an inactive treatment) and consolation (given with good intentions) remained unchanged.
The turning point came with Henry Knowles Beecher (1904-1976)[4], an American anesthesiologist (later a Harvard professor), who saw military service in field hospitals in Italy and North Africa. Sometimes the medical supplies, including morphine, would run out. On one particular occasion, Beecher was preparing to treat a young soldier with severe injuries, who was in great pain. No morphine was available. In desperation, a nurse gave the patient an intravenous injection of saline, along with some comforting words. Immediately the patient reacted as if he had been given a powerful analgesic.
Beecher was astonished. Whenever supplies ran out, he tried the technique again, and found it sometimes, but not always, had a beneficial effect. This one event was to colour Beecher’s view of medicine for the rest of his life.
When he returned to the USA, Beecher began to study the placebo effect. He performed one of the first ever meta-analyses of the few (fifteen) trials which had looked at the placebo effect. He found an impressive placebo response in many conditions, and suggested that the placebo effect could sometimes exceed the response to conventional treatment. He suggested that the placebo effect could produce objective evidence of disease improvement. He claimed that a third of patients (“a fairly constant 35%”) improved after placebo treatment[5].
Beecher argued that, to prove benefit, any new treatment must prove itself to be even better than placebo. He is therefore credited as the father of the placebo-controlled trial. However, Beecher’s position was that the placebo was a powerful remedy; the opposite of our modern view that the placebo is an inert control substance.
Modern analysis of Beecher’s work shows that it was deeply flawed. Of the 15 trials he quoted in his seminal paper, he misquoted figures from ten (including at least one he had co-authored). In addition, of the trials themselves, none had a control group, and many of the original authors proposed alternative explanations for observed improvements, without invoking the placebo effect. It is likely that if Beecher had quoted his figures accurately, his work would have had much less impact.
Despite this, Beecher’s figure of “one third” of all patients responding to placebo (alternatively, that “one third” of all therapeutic benefit is due to the placebo effect[6]) is widely quoted[7], but without any foundation.
A modern investigation
To study the placebo effect directly, it is necessary to compare it with something. The ideal trial has three arms: a treatment group, a placebo group, and a no-treatment, no placebo group. In the 1990s, two Copenhagen researchers, Asbjørn Hróbjartsson and Peter Gøtzsche, performed an enormous meta-analysis of all trials which they could find which fitted these criteria[8]. They found 114 trials, including over 8000 patients, and encompassing a wide variety of medical conditions (e.g. asthma, obesity, hypertension, anaemia, common cold, epilepsy, hypercholesterolaemia, menopausal symptoms, schizophrenia, chronic pain, carpal tunnel syndrome). Hróbjartsson and Gøtzsche were interested in placebo as a treatment of disease, so studies involving healthy volunteers were rejected.
Their conclusions were clear. They found “little evidence in general that placebos had powerful clinical effects”, and went on: “Outside the setting of clinical trials, there is no justification for the use of placebos.”
Closer analysis of their findings uncovers some interesting results. In those conditions where any change could be measured by the investigator (objective results), there was no statistical difference (e.g. serum cholesterol, blood pressure, blood sugar, peak flow, etc). However, where any change needed to be reported by the patient (subjective results, e.g. pain, depression, insomnia, anxiety, marital discord), several conditions did show a response to placebo, including 27 trials looking at different kinds of pain.
The consensus of evidence is that placebo does not produce improvement in organic disease. In other words, placebo can make you feel better, even if it doesn’t help you get better.
Unravelling the placebo effect
Some doctors consider the placebo effect to be an illusion. Natural history tells us that many new conditions get better without treatment (although the placebo may get the credit). In addition, the phenomenon of regression to the mean applies to chronic conditions, which often wax and wane. Patients are more likely to seek medical help (and enrol in clinical trials) when symptoms are worst, which then improve spontaneously (again, the placebo may get the credit).
For those (including me) who believe the placebo effect exists, there are three main ways to approach the phenomenon. The pharmacological approach, I believe, is unhelpful: there is no known mechanism whereby it might be considered to work. The other two approaches are the psychological (top-down) approach, and the neuroscientific (bottom-up) approach.
Psychologists point to various psychological effects, such as the expectancy effect, which is that the doctor subtly manipulates the patient to report a non-existent improvement. Hróbjartsson and Gøtzsche themselves wrote “a patient may tend to try to please the investigator and report improvement where none has occurred”. As an offshoot of the expectancy effect, patients may come to actually believe they are improving as a result of the care and attention they receive. The third psychological model is that of classical conditioning (Pavlov’s dogs): patients become so used to feeling better after visiting the doctor that they become conditioned to feel better even when the treatment is ineffectual.
Other important psychological factors include empowerment (the satisfaction from deliberately tackling the disease), reduced anxiety, and increased knowledge about the disease. It is clear that the placebo effect touches on many aspects of the doctor-patient relationship, not simply the handing over of a bottle of pills.
Critics (such as the author of the Lancet editorial above) argue that to respond to placebo must indicate gullibility or cognitive weakness: “it’s all in the mind”. However, response to placebo shows no correlation with “age, gender, ethnicity, educational level, intelligence, locus of control, extraversion, introversion, neuroticism or suggestibility”[9].
Neuroscientists have two main theories. The first is that placebo works by suppressing the function of the immune system[10]. This is supported by evidence of immune conditioning in animal models, which seems to suggest that the nervous system can induce changes in the function of the immune system. This work currently seems very far removed from the human situation.
The second theory is that placebo works by stimulating the production of endorphins, which fits with the observation that placebo is effective in the treatment of pain. This is supported by the evidence that some types of placebo analgesia (but not all) can be reversed by naloxone[11]. In addition, work with functional MRI suggests there is some activation of the midbrain, around the peri-aqueductal grey matter, by placebo[12]. Although promising, this work is still not entirely convincing.
Patient factors
There are several patient factors which seem to be important for placebo to work. It has been shown that sham acupuncture works better than placebo tablets[13]. Branded tablets work better than plain tablets. Coloured tablets work better than white tablets[14]. “Expensive” tablets (where the patient is told the tablets are expensive) work better than “cheap” tablets. Four tablets daily work better than two tablets daily. These findings suggest that patients respond better to dramatic or elaborate remedies than simple ones. These effects are deliberately maximised by pharmaceutical companies to make their products more desirable.
The patient must believe in the treatment given. If the patient believes he or she is receiving a placebo, it has much less effect. In clinical trials, patients are explicitly told they may receive an inactive treatment, and can often guess correctly (from the therapeutic effect or the side effects) which group they belong to. This may account for an apparent reduced placebo effect in clinical trials[15].
The opposite of the placebo effect is called the nocebo effect: the effect that a patient may come to believe he or she is being harmed when no harm is taking place. Nocebo means “I will harm” and is a modern term. There is a variety of evidence which supports the existence of the nocebo effect[16].
Complementary Medicine
Many types of CAM are proven to be no better than placebo. The CAM industry is worth £1.6 billion in the UK, and there is an ongoing increase in the number of practitioners. Despite the ineffectual remedies offered, satisfaction with CAM remains generally high[17].
CAM practitioners offer plenty of time, seldom break bad news, and are usually optimistic and enthusiastic about their treatments. The treatments themselves are often pleasant, involving aromatic smells, touch, music or relaxation. The ritual or mystery element may be compelling for some patients. The patient feels like a participant in the process, rather than a victim. Finally, there may be an element of transaction involved: patients value something which is expensive. By paying for it, they cement their own perception of that value.
By contrast, orthodox doctors are often short of time, and often bring bad news, and may be pessimistic (overtly or covertly) about their remedies. Hospitalisation and treatments may be unpleasant or dehumanising, rendering the patient a helpless victim. The patient may feel powerless. Increasingly, doctors have lost our own aura of mystery.
To elucidate whether the placebo effect lies in the treatment itself, or the practitioner, Ted Kaptchuk and colleagues in Harvard performed an elegant experiment[18]. Irritable bowel syndrome (IBS), typified by patient discomfort in the presence of normal investigations, shows many of the hallmarks of a condition which would be likely to be placebo responsive. 208 patients with irritable bowel syndrome (IBS) were randomised into three groups. Group 1 was placed only on a waiting list. Group 2 received sham acupuncture from a therapist who was instructed to remain silent. Group 3 received sham acupuncture plus “warmth, attention and confidence” from the practitioner. All groups were followed up for 3 and 6 weeks using commonly-applied symptom severity scores for IBS.
All the tests revealed similar results. Group 2 fared slightly better than Group 1, but Group 3 was substantially better than Group 2 in all categories, and all results achieved statistical significance.
Kaptchuk wrote “non-specific effects can produce statistically and clinically significant outcomes and the patient-practitioner relationship is the most robust component”.
Ethics of using placebo
Critics argue that to deliberately prescribe a treatment which you know to be ineffectual is quackery. Prescribing a placebo as if it were a real treatment involves a deception, which is unethical, and is likely to result in the patient losing trust in doctors if revealed. Use of placebo might delay appropriate investigation and treatment of an underlying, potentially serious, problem. Placebo treatments may be misused as a diagnostic test, to see if pain is “real”[19]. Finally, regular “reminders of illness” may make the patient less comfortable in the long run, not more.
Defenders[20] argue that placebo is harmless; it is effectively impossible to overdose; and there are very few side effects (but not none). If the patient is placebo-responsive, you save them the risks of taking more powerful treatments. Placebo works extremely well for a small number of patients: do we have a duty of care to treat those patients with placebo? Placebo works better outside clinical trials. What does it matter what we use, so long as the patient feels better? Most doctors would only prescribe placebo where they had ruled out serious illness or had no useful treatment to offer.
Defenders argue that there need not be deception involved in the use of placebo. The patient could be told: “I believe this treatment will help you, even although nobody quite understands how it works,” which is a statement of truth[21].
Placebo and the law
The GMC offers no explicit guidance about prescribing placebo. However, Good Practice in Prescribing Medicines (2006) says: “You should only prescribe drugs to meet identified needs of patients and never for your own convenience or because patients demand them”, and “You can prescribe unlicensed medicines, but, if you decide to do so, you must…be satisfied that there is a sufficient evidence base and/or experience of using the medicine to demonstrate safety and efficacy.” Deception violates informed consent, which is a core GMC principle.
The Medicines Act (1968) states that it is “…an offence for a person to supply a medicinal product in a container or package which is labelled or marked in such a way that it is likely to mislead as to the nature of the product or as to its uses or the effects of the medicinal products of that description”. Therefore a pure placebo must by law be labelled as such. To counter this restriction, some doctors use vitamin or mineral supplements instead.
Summary remarks
There is good evidence that placebo produces no objective improvement in organic disease.
There is good evidence that placebo produces some improvement in certain subjective conditions, such as pain.
I believe that the bulk of the evidence suggests that the essence of the placebo effect lies in the relationship between the patient and a warm, caring doctor. I believe that CAM teaches us that patients can be satisfied with an ineffectual treatment provided by a caring practitioner, as much, or perhaps even more than, an orthodox treatment provided by a neutral or unsympathetic practitioner.
I believe that we can improve our effectiveness as doctors by harnessing the “placebo effect” to treatments which are proven to be beneficial.
[1] Evans D (2004). Placebo: Mind over matter in modern medicine.Harper Collins, London.
[2] Hooper’s Medical Dictionary, 1811.
[3] Lasagna LC (1954). The humble humbug. Lancet 14th Aug 1954: 321.
[4] Kopp VJ (1999). Henry K. Beecher MD: Contrarian (1904-1976). ASA Newsletter 63:9 (available free online).
[5] Beecher HK (1955). The powerful placebo. JAMA 159:1602-6.
[6] For an excellent summary, see Placebo (2003), in Bandolier’s Little Book of Pain. Moore A et al. Oxford University Press.
[7] See Spinney L (2006). Purveyors of mystery. New Scientist 192 (2582) for a recent example.
[8] Hrobjartsson A, Gotzsche PC (2001). Is the placebo powerless? An analysis of clinical trials comparing placebo with no treatment. NEJM 344:21;1594-1602.
[9] Turner JA (2003). Non-specific Treatment Effects. in Bonica’s Management of Pain (3rd edn). JD Loeser (ed). Lippincott, Philadelphia.
[10] Evans D (2004). Placebo: Mind over matter in modern medicine. Harper Collins, London.
[11] Levine JD et al (1978). The mechanism of placebo analgesia. Lancet 2(8091):654-7.
[12] Wager TD (2006). The neural bases of placebo effects in pain. Curr Dir Psych Sci 14(4):175.
[13] Kaptchuk TJ, Stason WB et al. (2006). Sham device v inert pill: randomised controlled trial of two placebo treatments. BMJ , doi:10.1136/bmj.38726.603310.55.
[14] Schapira K, McClelland HA et al (1970). Study on the effects of tablet colour in the treatment of anxiety states. BMJ 1 (707):446-449.
[15] Fisher S, Greenberg RP (1993). How sound is the double-blind design for evaluating psychotropic drugs? Journal of Nervous and Mental Disease. 181(6): 345-350.
[16] Turner JA (2003). Non-specific Treatment Effects. in Bonica’s Management of Pain (3rd edn). JD Loeser (ed). Lippincott, Philadelphia.
[17] See Vickers A (1996). Research paradigms in mainstream and complementary medicine. In Ernst E (ed) Complementary Medicine: An Objective Appraisal. Oxford, Butterworth-Heinemann.
[18] Kaptchuk TJ, Kelley JM, et al. (2008). Components of placebo effect: randomised controlled trial in patients with irritable bowel syndrome. BMJ doi:10.1136/bmj.39524.439618.25.
[19] According to this paper, some Israeli doctors use placebo in this way: Nitzan U, Lichtenberg P (2004). Questionnaire survey on use of placebo. BMJ 329: 944-6. See also: Tilburt JC et al (2008). Prescribing “placebo treatments”: results of national survey of US internists and rheumatologists. BMJ 337: 1938.
[20] Hyland ME (2003). Using the placebo response in clinical practice. Clin Med 3 (4): 347-50.
[21] Lichtenberg P et al (2004). The ethics of the placebo in clinical practice. J Med Ethics 30: 551-554.
Copyright © Aidan O’Donnell 2010.
This article first appeared in The Annals of the Scottish Society of Anaesthetists, 2010.
Unauthorised reproduction prohibited.
Aidan O'Donnell 