Henry Beecher’s classic 1955 study “The Powerful Placebo” first attracted attention to the therapeutic effectiveness of placebo treatments in different painful conditions.¹ Although the impressive effect size he attributed to the placebo effect was subsequently questioned given a lack of controls for natural history and regression to the mean, his article initiated a paradigm shift and opened up a new field of research. Further groundbreaking work demonstrated that placebo analgesia could be blocked by naloxone, which implied that endorphins were involved in this context.² This work opened the path to a progressively greater understanding of what is now considered a psychobiological phenomenon. This chapter aims to review briefly the current knowledge of the psychology and neurophysiology underlying the placebo effect and then to focus on three main questions: (1) why the placebo effect is important in the practice of pain management; (2) what ethical considerations are raised by the clinical use of the placebo effect; and (3) how best to use the placebo effect for therapeutic purposes.

The term placebo is derived from the Latin for “I shall please” and originates from its use as an intervention aimed primarily at pleasing, rather than treating, the patient. Today, the word placebo is used in multiple contexts to refer to different but related concepts. Consequently, it has been extremely difficult for researchers to reach consensus on terminology in placebo studies, with even the term “the placebo effect” being raised as problematic (i.e., the placebo effect is the effect of something that has no effect). In this chapter, specific definitions are proposed and used (Box 18-1).

Placebo effects are recruited through different mechanisms, which can essentially be grouped into two categories: (1) creating positive expectations and (2) learning through classical conditioning is an important form.^3,4 These processes are detailed below. However, it should be emphasized that the placebo effect is often elicited by a combination of the two mechanisms, and disentangling them has been a topic of debate.⁵ Considerable evidence suggests that conditioning procedures lead to more robust placebo analgesia than verbally induced expectancy.^6,7 Yet it is almost impossible to condition a human being without some manipulation of expectations.^8,9 In fact, it has been suggested the two processes could lead to a “virtuous cycle”: a patient who experiences a positive first encounter has increased expectations of therapeutic benefit, which leads to enhanced conditioning effects associated with treatment and subsequent heightened overall placebo effect.³

EXPECTATIONS

Expectancy theory is based on the notion that the expectation of a positive outcome may elicit certain cognitive, emotional, and behavioral changes that increase the likelihood of that outcome occurring.^10,11 Positive expectations in the context of the therapeutic encounter may lead to a reduction in anxiety,¹² a decrease in self-defeating thoughts,⁵ or the resumption of a normal daily schedule.¹³ In fact, when expecting a painful stimulation, the brain shows patterns of activity in areas relevant to pain inhibition, such as the anterior cingulate cortex (ACC) and brainstem.^14,15 Furthermore, when participants in an experimental pain task are told to expect a higher or a lower painful stimulus, not only does this lead to increased or decreased reports of pain for stimuli that are actually of identical intensity, but it also elicits corresponding changes in the intensity of cerebral activity in areas relevant to pain perception.^16,17 Hence, the degree to which a patient expects a procedure to be painful can have an important impact on the actual perception of pain.¹⁸ Finally, expectations about treatment benefits can also alter the experience of a given stimulus in patient populations.¹⁹

Expectations regarding a particular treatment may exist prior to the initial contact with a health care provider (e.g., expectancy induced by past experience, advertisement, media, social interaction). These expectations can then be modified by a health care provider through information; suggestions; and other forms of communication, including nonverbal cues.²⁰

LEARNING

Classical conditioning is an important form of implicit learning in which an initially neutral stimulus is repeatedly paired with a biologically active stimulus (the unconditioned stimulus), which reliably elicits a response. After repeating pairings of the two stimuli, when the initially neutral stimulus is presented alone, it evokes a response (the conditioned response) by itself. At this point, the originally neutral stimulus is now referred to as a conditioned stimulus. For example, morphine is an unconditioned stimulus that reliably elicits analgesia (the unconditioned response). If morphine is repeatedly delivered via a blue pill (the neutral stimulus), eventually a blue pill containing no morphine (the conditioned stimulus) will elicit an analgesic effect (the conditioned response). In clinical settings, conditioning may arise from repeated positive experiences with a specific ritual or attribution of relief to a specific yet potentially noncausal factor. For example, patients frequently attribute a reduction in pain to resting. This attribution can lead certain patients to become fully sedentary over time or develop specific fears of movement,²¹ with an important negative impact on their general health.

The study of immune functioning in rodents provided early evidence for a link between classical conditioning and the placebo effect. After a conditioning procedure in which a saccharin-flavored beverage was paired with the immunosuppressant cyclophosphamide, rats who continued to receive a saccharin-flavored placebo beverage showed persistent suppression of immune functioning.²² Similar effects have been obtained in humans.²³ In the case of experimentally induced placebo analgesia, conditioning mostly consists of pairing a placebo treatment with the surreptitious reduction of stimuli. This procedure leads to placebo analgesia that can last up to 4-7 days, with some reduction in analgesia over time.²⁴ Interestingly, when first treated with a placebo, subsequent conditioning with an active medication is less efficient than if conditioning with active medication is performed without an initial placebo treatment.²⁴

Although classical conditioning is an important form of learning that can produce placebo effects, other types of learning are starting to be investigated. For example, social observation, has been shown to lead to placebo effects: individuals experienced placebo analgesia after simply observing a confederate reporting relief from the same placebo treatment.²⁵

The current understanding of the neurobiological underpinnings of placebo analgesia was initiated by a pharmacological study of postoperative pain, which demonstrated that the opioid antagonist naloxone could dampen the analgesic response elicited by a placebo intervention.² In parallel, animal models have shown that spinal excitability can be modulated by descending information originating from the brain.²⁶ This led to the description of a descending pain inhibition system also in humans and set the stage for an understanding of the neuroanatomical correlates of placebo analgesia.²⁷

NEUROTRANSMITTERS INVOLVED IN PLACEBO ANALGESIA

Opioids

Following the groundbreaking work of Levine et al.,² subsequent studies, including more recent neuroimaging research, have further underlined the role of endogenous opioids in placebo analgesia. For example, a study using positron emission tomography (PET) has shown that placebo analgesia is associated with increased µ-opioid–mediated neurotransmission in prefrontal and subcortical regions.²⁸ Furthermore, PET imaging has revealed similar patterns of activity in the ACC and the brainstem—two key structures of the descending inhibitory pain system—during both placebo analgesia and opioid induced analgesia.²⁹ These findings suggest that these two routes to pain reduction share overlapping neural mechanisms. Interestingly, activity in areas known for µ-opioid activity, such as the prefrontal cortex (PFC), periacqueductal gray matter (PAG), and amygdala, is correlated with change in pain perception in placebo analgesia trials.^28,30 Finally, high doses of naloxone, which inhibits endorphin binding, suppress the functional correlation between the ACC and the PAG, with a correlated reduction in the magnitude of the placebo effect.³¹

Non-opioid Neurotransmitters

Cholecystokinin

The neurohormone cholecystokinin (CCK) has anti-opioid properties and can both inhibit placebo-induced analgesia and promote nocebo-induced hyperalgesia.³² Congruently, the blockade of CCK with the antagonist proglumide can enhance the magnitude of placebo analgesia³³ and reduce the magnitude of nocebo-induced hyperalgesia.³⁴

Dopamine

Dopamine is a neurotransmitter involved in reward, motivation, and goal-directed behavior. It is released in the nucleus accumbens and the striatum upon expectation of analgesia and during the experience of pain relief, both of which can be conceptualized as inherent rewards.³⁵ The release of dopamine by the nucleus accumbens during placebo analgesia has been demonstrated by PET imaging, with a correlation between the magnitude of neurotransmitter activity and the degree of both the expected and actual analgesia.³⁶

Endocannabinoids

As noted previously, conditioning with opioids elicits placebo effects that are reversible by naloxone. In contrast, conditioning with nonsteroidal anti-inflammatory drugs results in placebo effects that are not reversible by naloxone, suggesting a different mechanism for these effects.⁷ Yet, the endocannabinoid CB1 receptor antagonist rimonabant specifically blocks placebo analgesia produced by ketorolac conditioning,³⁷ suggesting that endocannabinoids act as mediators of placebo analgesia in this context.

NEUROANATOMICAL UNDERPINNINGS OF PLACEBO ANALGESIA

The descending pain inhibitory system has central relays, including the ventrolateral PFC, dorsolateral PFC, ACC, PAG, and thalamus, and it exerts its effect on the posterior horn of the spinal cord, partially through endorphins.²⁷ Imaging studies have demonstrated that key areas of this system are involved in placebo analgesia, starting at the PFC and descending to the spinal cord, establishing that the descending pain inhibitory system is the likely neuroanatomical underpinning of placebo analgesia.^29,31,38,39 In fact, there is overlap between the neural correlates of placebo analgesia and those underlying other cognitive modulations of pain perception.^40,41,42 Interestingly, cognitive interventions that show potential for the treatment of chronic pain, such as mindfulness meditation and cognitive-behavioral therapy, also recruit similar prefrontal areas.^41,43

Pain is defined as a subjective symptom (International Association for the Study of Pain), which is particularly susceptible to the placebo effect and to the nonspecific treatment effects. A large literature suggests that these effects are quite important in the field of pain medicine, with studies performed using several different designs (Fig. 18-1).

Figure 18-1A illustrates how a placebo treatment can be used as a control for an active treatment for example in the context of a randomized controlled trial (RCT) of an analgesic medication. The placebo group can demonstrate a significant amount of analgesia. In a meta-analysis of placebo-controlled RCTs of analgesia for postoperative pain, 7% to 37% of patients in the placebo condition reported a 50% reduction in pain versus 5% to 63% of patients in the active treatment conditions.⁴⁴ It is interesting to note that some RCTs for analgesic interventions report no difference between the placebo and active treatment conditions, with significant improvement noted in both groups. For example, two trials of vertebroplasty versus a sham procedure showed equal clinical improvements with both interventions. However, because the studies did not include a ‘no intervention’ control group, it is unclear whether the benefits from vertebroplasty are attributable to a placebo effect or to the natural course of the illness or regression to the mean.^45,46

In an open versus hidden trial design (Fig. 18-1B), active medication can be delivered with or without the subject’s awareness. This allows for the nonspecific treatment effects to be teased apart from the overall treatment benefits in both experimentally induced pain in healthy subjects,^47,48 as well as in clinically relevant pain in patients.^49,50 This research suggests that an open injection of a saline solution with the verbal suggestion of analgesia may be as potent as a hidden injection of 6 to 8 mg of morphine.^51,52

More generally, when compared with the absence of treatment (Fig. 18-1C), placebo analgesia is estimated to produce a two point improvement on a 10-point visual analogue scale.^47,53 This effect size is considered clinically relevant.⁵⁴

Finally, Fig. 18-1D illustrates how placebo treatments can be used to study the response to nonspecific treatment effects. For example, patients with irritable bowel syndrome (IBS) were randomized to one of three conditions: (1) a wait list, (2) sham acupuncture delivered by a practitioner who treated the patient in a business-like way with very little interpersonal interaction, or (3) sham acupuncture delivered by a practitioner who treated the patient in a warm and empathic manner.⁵⁵ Interestingly, the patients with the placebo treatment and limited interaction had more improvement than the natural course group yet less than the group with placebo treatment and enhanced relationship. In parallel, the importance of medication labeling was highlighted in a clinical trial with oral medication for patients with migraine. Labeling a placebo as an active medication (“rizatriptan”) increased the placebo analgesic effect compared with uncertain labeling (“placebo or rizatriptan”), and in turn, uncertain labeling produced a greater placebo effect than open labeling (“placebo”). Finally, the open label placebo treatment was significantly more effective than no treatment at all.⁴⁹ In addition, a group of studies examined the impact of offering participants the choice between two different placebo treatments (presented as different analgesic creams) compared with being randomly allocated to one of the treatments. Volunteers who preferred a high degree of control (compared with those with a lesser need for control) experienced greater placebo analgesia when they were able to choose their treatment.⁵⁶ A final example of such designs is a study in which a placebo pill presented as an expensive analgesic drug elicited higher placebo analgesia than the same pill presented as an inexpensive analgesic.⁵⁷

Although clinicians have a number of treatments at their disposal, current therapies provide inadequate relief for many patients with chronic pain and can have significant, use-limiting side effects. Enhancing placebo-related effects is therefore a very interesting target for future research. Furthermore, patients with chronic pain can demonstrate problematic solution seeking,⁵⁸ which can lead to requests for further interventional or surgical approaches. These additional interventions can sometimes be harmful. Relief of symptoms and suffering might be better indicated in these situations, which can be achieved through cognitive or behavioral change.⁵⁹ Therefore, harnessing the power of the placebo effect might decrease suffering and improve current treatments for patients with chronic pain.