3 The Placebo Effect
The Massachusetts General Hospital Handbook of Pain Management
The Placebo Effect
Brian W. Dubois and Paul J. Christo
Impatient at being kept awake by pain, I availed myself of the stoical means of concentration upon some indifferent object of thought, such as for instance the name of “Cicero” with its multifarious associations; in this way I found it possible to divert my attention, so that the pain was soon dulled.
—Immanuel Kant, 1724–1804
I. Placebos and the natural course of illness
II. Active agents
III. Placebo characteristics
1. Placebo response rate
2. Placebos and procedures
3. Placebo sag in chronic pain
4. The active placebo
5. The nocebo
IV. Placebo mechanisms
1. Cognitive theory
2. Conditioning theory
3. Endogenous opioids
The word placebo was originally used to describe something that was pleasing, for example, a medical treatment that was used more to please the patient than to treat the medical condition. In present day medical terminology, a placebo is a drug or therapy that simulates medical treatment but has no specific action on the condition being treated. A placebo often provides a real therapeutic benefit even though it does not have a specific therapeutic action. It is well known that receiving medical treatment, in and of itself, often produces a “nonspecific” therapeutic benefit. It is unclear, however, how such a benefit arises and whether it occurs because patients expect relief from medical treatment or because their anxiety is reduced. Placebos have been administered deliberately (for example, in an attempt to satisfy patients who wish to pursue medical therapy), and they have been administered unintentionally, as in the use of therapies whose efficacy was later refuted. In one form or another, placebos have been used since the dawn of medical therapy.
Henry K. Beecher, the first chairman of anesthesia at the Massachusetts General Hospital, made some classic observations about pain plasticity, which have led to the present-day concept that pain and pain perception can be altered by a variety of central nervous system (CNS) factors. He first observed that after battle, soldiers experienced less pain and requested less morphine than did civilians hospitalized after surgery. He surmised that this effect was caused by lower levels of anxiety in the soldiers, who were relieved to be off the battlefield, in sharp contrast to the civilian patients whose concerns about surgery and hospitalization filled them with heightened anxiety. He had observed that a patient’s state of mind could alter pain perception, and he studied other possible ways that a subject could conjure up feelings that would suppress pain.
In 1955, he published a classic study, “The Powerful Placebo,” in which he analyzed the findings of 15 drug trials involving placebos in a thousand patients with various ailments. In the study, he noted a consistent therapeutic response to placebos for a variety of medical conditions. He surmised that patients’ expectations of benefit were sufficient to achieve therapeutic benefit. As a result, Beecher advocated the use of placebos, blinding, and controls in medical studies. He had not only altered the standard for medical studies, he had also introduced psychosomatic medicine into modern medicine.
Since the publication of Beecher’s classic study, placebos have been commonly used in studies in an attempt to separate out the “nonspecific” therapeutic benefits of medical treatment from true treatment efficacy. Many clinical and preclinical studies are conducted against a placebo-group baseline. The following are typical study designs involving placebos:
The double-blinded, placebo-controlled study , comparing a drug (or therapy) with a placebo when neither the patient nor the investigator knows what the patient is receiving.
The single-blinded study , comparing a drug (or therapy) with a placebo when the patient is blinded to what he is receiving but the investigator knows.
The open-label study , when both the patient and the investigator know what the patient is receiving.
The crossover study , when the patient sequentially receives both the placebo and the drug (or therapy), usually in a blinded fashion.
I. PLACEBOS AND THE NATURAL COURSE OF ILLNESS
Many diseases are characterized by periods of acute exacerbation followed by periods of remission or resolution. Common complaints such as backaches, headaches, muscle strains, earaches, and coughs usually improve spontaneously. The natural course of untreated illness must be appreciated to understand the true benefit of treatments.
The placebo effect, or the nonspecific therapeutic benefit derived from a placebo treatment, must be distinguished from the recovery or remission that occur during the natural course of disease. Most placebo-controlled studies do not include an untreated group, and consequently they cannot measure the magnitude of the true placebo effect in the study. By directly comparing the placebotreated group with the untreated group, the placebo effect can be quantified for a given study.
Since most placebo-controlled studies do not include an untreated group, the reported placebo effect probably overestimates the true placebo effect. In fact, a published search by Ernst and Resch of clinical trials containing both an untreated group and a placebo group from the Medline literature (from 1986 to 1994) yielded only 12 reports, of which six dealt with pain relief. In the pain trials, the authors noted that placebo treatments were more effective in alleviating pain than no treatment. The untreated group, however, did improve, and this improvement, the authors note, must be appreciated to quantify the true placebo effect.
II. ACTIVE AGENTS
Active drugs or therapies have an efficacy greater than that shown by a placebo. The difference in efficacy between an active drug and a placebo defines the specific therapeutic benefit of the active drug. The greater the difference between the active drug and the placebo, the greater the specific therapeutic benefit of the active drug. The overall therapeutic effect of an active drug is therefore composed of two components: (a) the specific therapeutic benefit and (b) the nonspecific (placebo) effects of treatment. As Beecher put it, “The power attributed to morphine is presumably its drug effect plus a placebo effect.” Any factor of treatment that potentially increases the nonspecific (placebo) component of therapy increases the overall perceived therapeutic effect of the active drug.
III. PLACEBO CHARACTERISTICS
1. The Placebo response rate
Beecher and others have examined the benefit of placebos in treating minor ailments such as headaches, nausea, anxiety, angina, backaches, and coughs. Beecher found that the number of patients given a placebo who had a response, defined as “a 50% or more relief of pain,” varied widely, ranging from 15% to 53%. (The myth that the placebo response rate is approximately 30% originates from the gross averaging of these numbers.) Other investigators have also observed a wide range of placebo response rates, and some report response rates that are much higher than the 15% to 53%. High reported placebo response rates might overestimate the true placebo response if the remitting nature of the ailment is not appreciated.
The placebo response is very much influenced by patient perception and expectations. In fact, patients often feel better simply because medical treatment has been initiated. An impressive medical setting instills patient confidence and expectation of good medical treatment. The placebo response can be further augmented by a good physician–patient interaction. Physician expectations, and patients’ perceptions of them, have also been shown to influence the placebo effect. For example, Gracely and colleagues, in a controlled double-blinded study of postoperative dental pain, showed that patients were influenced by subconscious signals received from their physicians. In the study, a first group of patients received a placebo, an opioid antagonist, or an opioid (an expected pain reliever), while a second group of patients received just the placebo or the opioid antagonist. Patients receiving the placebo in the second group had more pain than those receiving the placebo in the first group. It was argued that the clinicians’ knowledge of the range of treatment alternatives (no opioid in the second group) was unconsciously communicated and perceived by the patients.
The placebo response has not been linked to any particular personality trait or personality type. As yet, there is no reliable way to predict who will respond to a placebo. Beecher, in his early studies, for example, could not find any response-rate difference that was based on sex or intelligence. In a recent study in a depressed patient population, Wilcox and colleagues looked at age, sex, marital status, education, duration of illness, and severity of illness to predict placebo response rates. No significant prediction of placebo response could be made on the basis of sex, age, education, or duration of illness. They did, however, find a slightly increased placebo response rate when the depression was less severe and when the patient was married.
2. Placebos and procedures
The placebo response is also seen with nondrug therapies, including medical devices and invasive procedures. For example, Hashish and colleagues have shown that ultrasound therapy reduces pain and swelling after dental surgery, but that it is no more effective than mock ultrasound therapy. A powerful placebo response to ligation of the internal mammary arteries for the treatment of angina became apparent after the treatment was widely popularized. The procedure was thought to improve coronary blood flow, and it was associated with a dramatic improvement in anginal symptoms and exercise tolerance in multiple non-placebocontrolled trials. Subsequent double-blinded studies using sham skin incisions with no ligation showed a similarly dramatic (70%) response in anginal symptoms and exercise tolerance. The ligation procedure has since been abandoned.
3. Placebo sag in chronic pain
Placebo sag is a decrease in the placebo response rate seen in patients who have experienced a number of treatment failures. Positive treatment experiences tend to augment the placebo response, whereas negative treatment experiences tend to cause an “extinction of the placebo response.” Placebo sag has been described in patients with chronic pain, who frequently feel that previous pain therapies have failed them. Treatment of their pain becomes even more difficult, because not only are they less likely to respond to placebo but also they are less likely to respond to active medications (all of which have a placebo or nonspecific therapeutic treatment component to them).
4. The active placebo
During double-blinded placebo-controlled studies, patients (and evaluators) can sometimes differentiate placebos from active drugs on the basis of side effects. This is particularly true with psychiatric and pain treatments, when drugs have significant neurologic and cognitive effects. To enhance blindability in placebo-controlled trials, some investigators have proposed using active placebos. An active placebo is a drug that simulates medical treatment (through side effects) but has no specific action on the condition being treated. For example, certain antidepressants have significant anticholinergic properties. An active placebo for an antidepressant trial could be a substance with anticholinergic properties that does not have specific antidepressant properties.
5. The nocebo
Another observation made by Beecher was that “toxic” side effects may result from placebo drug administration, including “nausea, drowsiness, headache, fatigue, and dry mouth.” These “noxious” side effects have subsequently been termed nocebo effects. Other investigators have noted these effects and have even described allergic-type reactions after placebo administration. Such side effects are augmented by heightened patient expectations of possible negative effects. Moreover, when patients expect little therapeutic benefit from therapy, they are at an increased risk of side effects. For example, it is reported that young healthy volunteers who feel they have little to gain from a treatment tend to experience more side effects.
IV. PLACEBO MECHANISMS
Many investigators feel that the placebo effect is caused by a reduction in patient anxiety and its consequent reduction in pain perception. Beecher, in fact, observed that placebos “are most effective when stress, anxiety, or pain is the greatest,” and he noted the importance of a patient’s “perception” and “reaction” to pain.
1. Cognitive theory
The cognitive theory states that the expectations of patients play an important role in the placebo response. Clinically, it seems clear that patients who expect a good response from treatment look for signs of a good response and try to dismiss any negative effects. Investigators have shown that patients’ expectations of a drug’s effects will alter their perception of those effects. For example, when patients were blindly given a psychostimulant, investigators showed that the stimulatory effects could be either exaggerated or diminished depending on whether the patient was told that the drug was a stimulant or a sedative.
2. Conditioning theory
The conditioning theory states that learning through association is important in the placebo response. Further, this theory proposes that the placebo response is a conditioned response that can be elicited by stimuli that, through prior conditioning, produce a reduction in symptoms. Evidence of this exists in both animals and humans. Ivan Pavlov, who described classical conditioning, demonstrated a conditioned placebo response in dogs. He reported that dogs, which previously received morphine when placed in an experimental chamber, displayed morphine-like effects when again placed in the experimental chamber. Since then, a number of investigators have published studies demonstrating conditioned responses to placebos in animals.
In humans, a learned placebo response to the analgesic effects of propoxyphene for the treatment of pain was demonstrated by Laska and Sunshine. In their study, patients were given the analgesic at different strengths. The patients who received the higher strength received greater analgesia. All patients were then given a placebo. The patients who had received the higher-strength analgesic thought the placebo was much more effective than the patients who had received the lower-strength analgesic. Here, the previous analgesic experience predicted the efficacy of the placebo.
3. Endogenous opioids
Endogenous opioids may be responsible for placebo analgesia, because naloxone, an opioid antagonist, has been shown to reverse placebo analgesia. Levine and colleagues, for example, examined the effects of naloxone on placebo analgesia in postoperative dental pain. They termed patients whose pain responded to placebo “placebo responders,” and those patients whose pain worsened after placebo administration, “placebo nonresponders.” When naloxone was given after placebo to patients in both groups, the placebo responders had a much greater increase in pain than the placebo nonresponders. This suggested that placebo-induced analgesia in the placebo-responder group was mediated by the release of endogenous opioids.
Ter Riet and colleagues recently reviewed the literature and found studies supporting naloxone reversibility of placebo-induced analgesia in both postoperative and experimentally induced pain. Another study that supports the role of endogenous opioids in the placebo response was conducted by Lipman and colleagues. These investigators measured endogenous opioid “peak B” fraction endorphin levels in cerebrospinal fluid after placebo administration in chronic pain patients. After placebo administration, they found that peak-B endorphin levels were significantly higher in placebo responders than in placebo nonresponders.
Because the placebo effect is a real therapeutic benefit, it is clear that the use of placebos to differentiate “real” pain and illness from “imagined” pain or illness is incorrect and inappropriate. The benefit of the placebo response, as well as a patient’s trust, can easily be lost if placebos are used inappropriately. Even during the conduct of a trial, patients should be informed that they might receive a placebo. Although the placebo effect may be a confounding factor when determining new drug efficacy, it is of considerable benefit to the practicing clinician, and this should not be sacrificed. The effect is not simply a response to a dummy drug or procedure but is a response that can be triggered by many factors, such as the smell of a doctor’s office, the sight of a needle, or the soothing words of a caregiver. By working to build good patient relationships, the placebo effect can be maximized. Maximizing the nonspecific (placebo) component of treatment increases the perceived therapeutic effect of both active and inactive treatments.
Beecher HK. The powerful placebo. JAMA 1955;159:1602–1606.
Ernst E, Resch KL. Concept of true and perceived placebo effects. Br Med J 1995;311:551–553.
Gracely RH, Duloner R, Wolskee PT, et al. Placebo and naloxone can alter post-surgical pain by separate mechanisms. Nature 1983;306:264–265.
Hashish I, Hai HK, Harvey W, et al. Reduction of postoperative pain and swelling by ultrasound treatment: A placebo effect. Pain 1988;33:303–311.
Heeg MF, Deutsch KF, Deutsch E. The placebo effect. Eur J Nucl Med 1997;24:1433–1440.
Laska E, Sunshine A. Anticipation of analgesia: A placebo effect. Headache 1973;13:1–11.
Levine JD, Gordon NC, Fields HL. The mechanism of placebo analgesia. Lancet 1978;2:654–657.
Lipmann JJ, Miller BE, Mays KS, et al. Peak B endorphin concentration in cerebrospinal fluid: Reduced in chronic pain patients and increased during the placebo effect. Psychopharmacology 1990;102:112–116.
Margo CE. The placebo effect. Surv Ophthalmol 1999;44:31–43.
Pavlov I. Conditioned reflexes. London: Oxford Press, 1927.
ter Riet G, de Craen AJM, deBoer A, Kessels AGH. Is placebo analgesia mediated by endogenous opioids? A systematic review. Pain 1998;76:273–275.
Turner JA, Deyo RA, Loeser JD, et al. The importance of placebo effects in pain treatment and research. JAMA 1994;271:1609–1614.
Wilcox CS, Cohn JB, Linden RD, et a. Predictors of placebo response: A retrospective study. Psychopharmacol Bull 1992;28: 157–162.