BMJ. 1994 (Jul 9); 309 (6947): 69–70
The placebo effect: can we use it better?
Placebos work best for pain, disorders of autonomic sensation, and disorders of factors under neurohumoral control
Shapiro defined a placebo as any treatment deliberately used for non-specific psychological or psychophysiological effect.1 That the placeo effect, a classic example of the mind-body relation, is as clinically undeveloped as it is pervasive may reflect the dominance of modern chemotherapy. The placebo depends on largely subconscious interactions between the doctor, the treatment process, and the patient1 2; it is the form of a treatment without its substance.
In practice, placebo treatment usually consists of a dummy medication or an intervention, which ranges from surgery to history taking. Placebo medication commonly operates through the administration of a substance, either pharmacologically active (a drug) or inert. The net effect of a given drug is thus the sum of the drug's pharmacological effects and the placebo effect associated with the act of treatment.3
The use of placebos raises important ethical questions3 4; if these can be answered can we exploit the placebo effect to benefit patients? We first need to consider the neurophysiology of the placebo effect. In the psychoneurophysiology of pain the model of cognitive versus somatic pain survives.5 In this paradigm somatic pain is linked to the source of nociception while the patient's awareness and cognition, probably residing in the thalamus,6 determine the perceived pain.2 7 Anxiety activates the hypothalamic-pituitary-adrenal axis and increases perceived pain; likewise, the removal of anxiety decreases pain. This two component model is supported by "negative placebo effects" such as "clinic hypertension," in which anxiety increases blood pressure in a conditioned response.8
In 1964 Lasagna et al first showed the hyperalgesic effect of the partial opioid antagonist naloxone.9 Animal and clinical experiments later showed that a family of opioid peptides in the brain, the endorphins, mediates some types of somatic pain.7 9 10 The endorphins originate from proopiomelano-corticotrophin and are thus linked, through ß lipotrophin,11 with the regulation of the hypothalamicpituitary-adrenal axis.12 Neither animal nor human experiments using opioid antagonists have shown the precise functional links between the opioid, anti-opioid,7 and ß lipotrophin systems of neuroendocrine peptides13 in placebo analgesia. The neurotransmitter (gamma)-aminobutyric acid increases the secretion of both ß endorphin and ß lipotrophin. The endorphins may also modulate the stimulation by (gamma)-aminobutyric acid of secretion of endorphins in a negative feedback loop.
How endorphins interact with brain opioid receptors is poorly defined. The endorphins might behave as pure opioid µ receptor agonists, such as codeine or morphine,14 or as µ receptor partial agonists, since they produce only limited pain relief.5 9 A system of anti-opioid neurotransmitters seems to modulate the opioid system in the rat brain.15 Endogenous anti-opioids may confer supersensitivity to endorphins by increasing the number of opioid receptors.16
Although an integrated model including opioids, antiopioids, and (gamma)-aminobutyric acid may explain some aspects of placebo action, we do not know how a thought releases neural peptides. The neuropeptide hypothesis holds for placebo analgesia, but little is known about how placebo promotes wellbeing in other ways. The placebo mechanism thus seems to be both multidimensional and selfregulating. In evolutionary terms the placebo effect might compensate for overexcitation of the hypothalamicpituitary-adrenal axis by environmental threats. Is the placeo effect mimicked or enhanced by anxiolytic drugs such as the benzodiazepines? An unrecognised placebo response could, indeed, partly explain the success, in subjective terms, of anxiolytic drugs across cultures. Specific placebo treatment appeals both because it resonates with the holistic view and because it is unlikely to cause harm.
Can we select patients accurately for placebo treatment? Two factors are necessary for placebo action: a suitable disease and a dynamic relationship between patient and doctor. Often, however, neither patient nor doctor is aware of the placebo effect. Experience also colours the response to placebo. A placebo works in about one third of subjects.5 Placebos are more effective for clinical than experimental pain and for severe than mild pain,2 but mildly depressed patients respond better than severely depressed ones.17 Gender, suggestibility, and intelligence quotients do not affect respon-siveness to placebos. "Placebo reactors" who regularly respond to placebos do not exist.1 2 5 Selecting those patients most likely to benefit from placebo is therefore difficult.
In which diseases is a consistant response most likely? A placebo works best and most commonly in pain and disorders of autonomic sensation, such as nausea2 9: psychoneuroses, phobias, and depression2 5; and disorders of factors under neurohumoral control, such as blood pressure2 8 and bronchial airflow. Placebos will not work if the disease is hyperacute (for example, cardiac arrest), or when vital functions degenerate (for example, in severe metabolic acidosis). Also, the placebo effect usually fails in unremitting disease, such as hereditary syndromes.
Adjuvant placebo treatment is useful in some neuroses and mild depression and may be useful in some chronic non-cancer pain syndromes and in hypersensitivity or hyper-reactivity conditions in which psychic factors play a part (for example, dermatitis and bronchial asthma). Adjuvant placebo seems unlikely to affect the course of cancer. Could the placebo effect be boosted long term? Prolonged administration of opioids in mice inhibits the expression of cell surface markers on T lymphocytes in a dose related manner.18 This suggests an impairment of immune function, which may outweigh the advantage of a sustained placebo effect.
What about combining placebo treatment with unorthodox techniques such as acupuncture, hypnosis, and homoeopathy? These are likely to potentiate any placebo effect only if they work through an independent mechanism. Hypnosis and acupuncture may not depend on endorphins,19 but more information is needed. Even greater is the need to determine the placebo component of medicines--both traditional and conventional. The extent to which homoeopathy works through the placebo effect is unresolved, as discussed elsewhere in this issue (p 103).20 Another study in this week's journal reports that an impressive number of cancer patients experienced benefit such as increased optimism from complementary treatments (p 86).21 Interestingly, because a reduction in anxiety is a marker of placebo action, patients in this study who used complementary treatments were more anxious than those who did not.
Insights have come from recent studies of the placebo effect, but there is no easy way around the poor specificity and predictability of placebo treatment. Moreover, the self regulating nature of the placebo mechanism may limit the treatment gain. Nevertheless, in appropriate patients, doctors might consider giving a placebo when active treatment is both costly and likely to confer only marginal or transient benefit.
Vernon M S Oh
1 Shapiro AK. A historic and heuristic definition of the placebo. Psychiatry 1964;27:52-8.
2 Bourne HR. Rational use of placebo. In: Melmon KL, Morrelli HF, eds. Clinical pharmacology: basic principles in therapeutics. 2nd ed. New York: Macmillan, 1978:1052-62.
3 Oh VMS. Magic or medicine? Clinical pharmacological basis of placebo medication. Ann Acad Med Singapore 1991;20:31-7.
4 Stanley B. An integration of ethical and clinical considerations in the use of placebos. Psychopharmacol Bull 1988;24:18-20.
5 Beecher HK. The powerful placebo. JAMA 1955;159:1602-6.
6 Kinney HC, Korein J, Panigrahy A, Dikkes P, Goode P. Neuropathological findings in the brain of Karen Ann Quinlan. N Engl J Med 1994;330:1469-75.
7 Watkins LR, Meyer DJ. Organization of endogenous opiate and nonopiate pain control systems. Science 1982;216:1185-92.
8 Pickering TG, James GD, Boddie C, Harshfield GA, Blank S, Laragh JH. How common is white coat hypertension? JAMA 1988;259:225-8.
9 Lasagna L, De Kornfeld TJ, Pearson JW. The analgesic efficacy and respiratory effects in man of a benzomorphan narcotic antagonist. J Pharmacol Exp Ther 1964;144:2-6.
10 Levine JD, Gordon NC, Bornstein JC, Fields HL. Role of pain in placebo analgesia. Pro Natl Acad Sci USA 1979;76:3528-31.
11 Petraglia F, Facchinetti F, Martignoni E, Nappi G, Volpe A, Genazzani AR. Serotoninergic agonists increase plasma level of beta-endorphin and beta-lipotrophin in humans. J Clin Endocrinol Metab 1984;59:1138-42.
12 Howlett TA, Rees LH. Endogenous opioid peptides and hypothalamo-pituitary function. Annu Rev Physiol 1986;48:527-36.
13 Graceley RH, Dubner R, Wolskee PJ, Deeter WR. Placebo and naloxone can alter postsurgical pain by separate mechanisms. Nature 1983;306:264-5.
14 Rang HP, Bevan S, Dray A. Chemical activation of nociceptive peripheral neurones. Br Med Bull 1991;47:534-8.
15 Galina ZH, Kastin AL. Existence of antiopiate systems as illustrated by MIF-1/Tyr-MIF-1. Life Sci 1986;39:2153-9.
16 Chang SC, Lutfy K, Sierra V, Yoburn BC. Dissociation of opioid receptor upregulation and functional supersensitivity. Pharmacol Biochem Behav 1991;38:853-9.
17 Horvath P. Placebos and common factors in two decades of psychotherapeutic research. Psychol Bull 1988;104:214-25.
18 Kimes AS, Smith MJ, Winchell CJ, London ED. Effects of morphine on the phenotypic expression of cell surface markers on murine lymphocytes. Life Sci 1992;51:807-15.
19 Moret V, Forster A, Laverriere MC, Lambert H, Gaillard RC, Bourgeois P, et al. Pain 1991;45: 135-40.
20 Buckman R, Lewith G. What does homoeopathy do--and how? BMJ 1994;309:103-6.
21 Downer JM, Cody MM, McClusky P, Wilson PD, Arnott SJ, Lister TA, et al. Pursuit and practice of complementary therapies by cancer patients receiving conventional treatment. BMJ 1994;309:86-9.
© 2003 BMJ Publishing Group Ltd