Most physicians make at least some use of the placebo effect to enhance treatments, whether they realize it or not. This article examines the extent of the placebo effect in patients with psychiatric illness, and reviews what is known about how placebos work. It then discusses the application of the placebo response in enhancing the effects of treatment modalities in psychiatry.

It is instructive to note a feature of virtually every antidepressant efficacy trial: in outpatients with moderately severe depression whose symptoms are assessed using the Hamilton Rating Scale, most of the improvement occurs during the first 2 weeks of treatment, and during this time, no differences in outcome between active drug and placebo are evident. After about 2 weeks, the placebo response reaches a plateau. Response to medication may continue to increase for a time, but during the period when most of the improvement occurs, placebo and medication produce similar responses.

In fact, the strength of the placebo response can be quite significant in several psychiatric disorders.¹ Panic disorder is highly responsive to placebo, with a nearly 50% improvement in symptoms among patients assigned to that treatment strategy. In patients with posttraumatic stress disorder or depression, the placebo response is greater than 30%, and a response in the 30% range is also seen in generalized anxiety disorder.

Patients with other conditions, including obsessive-compulsive disorder (OCD) and psychosis, are less likely to exhibit a placebo response. Clearly, the placebo response is not uniformly strong in all psychiatric conditions. We have used these findings as a basis for recommending that clinical trials of agents being tested in patients with psychiatric disorders always include a placebo arm.¹

Placebo vs pharmacotherapy or psychotherapy

How do responses to placebo compare with those achieved with medication? When treated with clonazepam or an SSRI, 60% to 70% of patients with panic disorder become panic-free during a 10-week treatment period. During the same period, half or more of patients with panic disorder who are treated with placebo become panic-free, underscoring the magnitude and importance of the placebo effect in this common condition.²Active drug treatment for depression leads to improvement in about 60% to 70% of patients, while placebo administration is followed by improvement in 30% to 40%, again an impressive contribution.³

Conditions that are poorly responsive to psychotherapy are also unlikely to respond to placebo. Patients with attention-deficit/hyperactivity disorder, for example, are more likely to improve with medication than with behavior therapy. We would not expect a robust response to placebo in this patient group.

The results of a collaborative National Institute of Mental Health study on the treatment of depression are informative.⁴ Most of the patients had unipolar disease and were mildly, moderately, or severely ill. Some were randomized to clinical management and placebo; they attended a clinic, were given a placebo pill, and asked how they were doing. Patients in other groups were randomly assigned to cognitive-behavioral therapy, interpersonal psychotherapy, or imipramine. When the results were analyzed, no differences in outcome were evident according to the various treatments. None showed any advantage over the others.

The analysis was also conducted according to whether patients were mildly or moderately/severely ill, and there the effects of some of the treatments did separate. In the less severely ill patients, no statistically significant difference was present among the treatments. In the more severely ill patients, imipramine, which was a popular antidepressant at the time the study was conducted, had an advantage over the 2 psychotherapies and placebo. Although there was a difference between the psychotherapies and placebo, it was not statistically significant. In the severely ill patients, imipramine worked better than any other treatment modality, but placebo was about as good as the psychotherapeutic strategies.⁴

Although cognitive therapy for depression is widely used and taught in residency programs, the overwhelming majority of studies that have compared cognitive therapy with a pill placebo in the treatment of moderately severe depression show no difference in efficacy.^5,6 Less severely ill patients may benefit from any one of several different types of treatment, including psychotherapy, alternative therapies, and placebo.

The benefits of placebo treatment are, of course, not limited to psychiatry. In a review of hypertension therapy, it was documented that active pharmacologic agents were associated with a 40% to 60% rate of reducing blood pressure to the normal range, while placebo was associated with a 25% success rate.⁷ Some neurologic conditions, asthma, and certain pain problems are also responsive to a placebo effect.

What is a placebo response?

In my opinion, the term placebo is misleading. In fact, patients who receive "placebo" treatment get much more than a sugar pill, whether they are enrolled in a placebo-controlled clinical trial or are the recipients of ordinary medical care. "Treatment situation" is a more evocative term than "placebo treatment," and it includes several essential elements associated with healing, as shown in Table 1.

All these elements, including healing symbols and rituals, evaluations, and discussions of diagnosis and prognosis, have an important role and are part of the placebo treatment or treatment situation. Hippocrates said that physicians should make frequent visits with their patients, inquire into all particulars of their situations, and learn about prognosis. The goal was to inspire confidence among patients so that they would feel comfortable entrusting their care to the physician.

As an example, consider how patients respond to diagnostic tests. In a 1981 paper that appeared in the Annals of Internal Medicine, clinical outcomes were assessed in a group of 176 patients who had chest pain that was considered nonspecific.⁸ Participants in this study were randomized to have either routine ECG and serum creatine phosphokinase tests or to have no diagnostic tests. Short-term disability was reported by fewer patients (20%) in the test group than in the patients who received no tests (46%); this difference was statistically significant. Diagnostic tests were an independent predictor of recovery. More than half (57%) of the patients in the test group felt that the care they received was "better than usual." Fewer than one third of the patients who did not receive tests felt this way.⁸ This is evidence that doing some simple testing often has an impact on how people feel.

Similarly, a study conducted in New Zealand suggests the importance of a formality such as writing on a prescription pad. A group of 456 sedentary patients were given verbal recommendations to increase their physical activity levels. Then they were randomized to a written exercise prescription or to a verbal advice group. The number of people in both groups who engaged in physical activity increased markedly after 6 weeks, but more participants in the written prescription than the verbal advice group were active. The investigators concluded that a written exercise prescription was an important adjunct in getting their patients to increase their physical activity levels.⁹ In our culture, the written prescription has a large symbolic impact.

Placebo response vs placebo effect

No doubt some of the improvements observed in placebo-treated patients occur simply because of the passage of time. Many of the conditions that are most placebo-responsive, including insomnia, pain, asthma, hypertension, and depression, are by nature fluctuating illnesses. Some of the improvement we see associated with placebo is a result of regression to the mean. Nonethe- less, considerable evidence suggests that placebo treatment actually adds something that goes beyond the mere passage of time. Some would propose that the placebo effect should be defined as the placebo response minus the no-treatment response, or minus the changes that occurred simply as the result of waiting or the passage of time. Just as we say that the drug effect is the response that you obtain with a drug minus the placebo response, we can also say that the placebo effect is what we obtain from a placebo response minus what you would have gotten if no treatment of any kind were attempted.

Mechanism of action

The conditions that seem to be most likely to respond to placebo are those in which psychological distress plays an important role either in the exacerbation or expression of symptoms. Examples include depression, anxiety disorders, asthma, and painful conditions. Much of what we offer in the treatment situation for these conditions reduces distress. Obtaining a diagnosis and getting help from a recognized healing authority, an explanation for symptoms, and a plausible treatment are factors that alleviate distress. It is important to recall the distinction between illness and disease. Illness refers to the distress experienced by the patient, including symptoms, discouragement, and anxiety. Disease refers to objective evidence--for example, that which the physician observes in test results and imaging studies.

Placebos tend to help illness--but how? Expectation has been subjected to the most study. Conditioning, endorphin release, and distress relief have also been considered, as shown in Table 2. With regard to expectation, it is clear that people experience what they expect to experience. Double-blind studies have demonstrated that when people receive a drink that they are told contains alcohol, they experience some of the effects of alcohol ingestion, including some of the physiologic effects.¹⁰ Test subjects who are given decaffeinated coffee but are told that it is fully caffeinated exhibit improved reaction time and alertness.¹⁰

Expectation has a profound effect on what people experience as a result of treatment. Until recently, the prevailing opinion held that these findings could be explained by patients telling physicians what they thought they wanted to hear or by patients imagining that they felt better. However, the results of several brain imaging studies suggest that expectation has a neuroanatomic and neurophysiologic basis.^11,12 Conditioning also has a strong impact. Responses to chemotherapy are an example. After patients have had 1 or 2 infusions of a chemotherapeutic agent that caused nausea, they often begin to feel the onset of nausea as soon as they enter the facility where the medications are administered. We are aware of patients with asthma whose airways seem to open as soon as they see their nebulizers.

Conditioned responses are also seen in animals. In rats that have been the recipients of long-term amphetamine administration, extreme locomotor responses persist even when placebo is substituted for the amphetamine. In some pain experiments, placebos may increase endorphin levels, and placebo analgesia can sometimes be blocked by naloxone.

An intriguing study conducted in patients with idiopathic Parkinson disease sheds more light on conditioning and expectation. These patients had been implanted with stimulating electrodes. Not surprisingly, motor performance diminished in these patients within 30 minutes of the time the stimulators were turned off. After this exercise was conducted, a sham turning-off procedure was conducted: Patients were told the stimulators were being shut off, but they actually remained on. Even so, motor velocity decreased. This is an example of the nocebo effect, in which a negative event happens as a result of a negative expectation. In another experiment, patients were told that the stimulators would be turned back on. Although the stimulators actually remained off, patients experienced better motor function. Although the expectation was not as powerful as the stimulator itself, the results reveal much about the ability of expectation to fuel treatment results.¹³

The power of conditioning is demonstrated by the results of a study of 24 patients with mild to moderate hypertension who were randomized to receive placebo, atenolol, or no treatment. Before drug treatment was administered, blood pressure was similar in patients taking placebo and patients taking nothing. After patients had taken atenolol for a week, use of placebo caused a significantly greater antihypertensive response than receiving no treatment. The investigators concluded that the placebo response obtained after atenolol had been administered was not a simple residual drug effect.¹⁴

In a study conducted among men with unipolar depression, changes in brain glucose metabolism were measured by using positron emission to-mography. Similarities and differences were assessed among patients receiving placebo and fluoxetine. A strong placebo response went hand-in-hand with regional metabolic increases involving the prefrontal, anterior cingulate, premotor, parietal, posterior insula, and posterior cingulate areas; metabolic decreases occurred in the subgenual cingulate, parahippocampus, and thalamus. These areas of metabolic change overlapped with those that were observed in patients who had responded to fluoxetine. Among patients who responded to fluoxetine, however, changes were also seen in the brainstem, striatum, anterior insula, and hippocampus. The authors concluded that changes in certain areas of the brain are probably essential to the remission of depression, whether active drug or placebo is administered. Patients who respond to fluoxetine experience brain changes that do not occur in patients who do not respond to placebo. Patients who respond to placebo experience brain changes that are similar to but less extensive than those that occur in response to fluoxetine. These changes are different from those seen in patients who do not respond to placebo.¹⁵