A half century after Henry Beecher's groundbreaking if controversial paper on ‘The Powerful Placebo’,1 the placebo effect has established its credentials as a force capable of altering a subject's experience and the actual physiology of pain.2 Formerly a handy little ruse symbolised by the sugar pill, its reputation has risen to the point that some now think of renovating medical practice by translating it from the realm of experiment into the clinic.3 The principal obstacle standing in the way is that subjects in experiments designed to elicit the placebo effect undergo manipulations, ranging from sham treatments to the covert deactivation of a pacemaker,4 that are difficult or impossible to reconcile with ethical clinical practice. If the placebo effect represents an untapped medical resource, and yet deception seems required to call it forth, then what is to be done?

Some, questioning the need for deception, look to the open placebo as a way out of this bind. The possibility of using placebos presented as placebos was first tested in a 1963 trial involving a handful of patients and no control group, among other methodological deficiencies.5 In a recent study with suggestive results, irritable bowel patients were treated with a placebo explicitly disclosed as such, though also credited with ‘healing’ potential.6 In all, however, there has been little study of open placebos, and the regular citation in the placebo literature of the few trials that have taken place attests more to the level of interest in the open placebo than the strength of the evidence for it. Certainly there followed from the 1963 trial nothing like the wave of studies that ensued from the establishment of the double-blind clinical trial as proposed by Beecher. Over and over again such trials have reported a placebo response even though they are not in fact an ideal way of evoking it. They are not ideal because the placebo effect is largely dependent on expectation, and the subjects in an randomised clinical trial (RCT) who consent to the 50/50 chance of getting placebo instead of an active treatment have correspondingly qualified expectations. It would seem that the optimal way to activate the placebo effect would be to present a placebo treatment as a definitely effective one and surround it with all the conventional pomp and circumstance of medical care—the theatre that researchers have in mind when they refer to the placebo effect as a ‘context effect’. So we are back to the issue of deception.

But if deception is defined narrowly, such that half-truths and the like do not qualify, then a potentially large field is opened to the cultivation of the placebo effect. In actuality, many uses of placebo have been legitimised or in any case covered by the use of half-truths or equivocations. Precisely because equivocations have some truth in them and are not as offensive as lies they may escape ethical scrutiny. In the case of placebo treatments, however, equivocations demand such scrutiny because they can undermine the vital ethical requirement of informed consent. The argument that follows, then, presumes and is predicated on the importance of informed consent.

Half-truths surrounding the use of placebos

When the doctor prescribed a sugar pill a century ago, he was in no doubt about what he was doing—he was using a placebo. But what if a doctor today prescribes an antibiotic when it is not indicated or a vitamin in the absence of any specific deficiency or a medication at a suboperative level? It is an open secret that doctors continue to employ placebos from time to time, albeit not in the stark form of the sugar pill,7 and some believe that by shunning inert substances and giving the placebo the cover of medical legitimacy, doctors do not so much enhance the theatrical effect as finesse their own as well as their profession's disapproval of the use of placebos.8 To open up territory in between the prohibited sugar pill and the accredited medical treatment, they exploit the ambiguity of the ‘impure placebo’—impure because not as completely deceptive as the sugar pill. The equivocation of the impure placebo (sometimes called a pseudo-placebo) serves to licence the use of a placebo in the face of general censure of deception. At this point I am concerned not with the ethics of these manoeuvres as such but the larger pattern into which they fit, that of getting around restrictions on the use of placebos by deploying half-truths.

In research on placebos (as opposed to RCT's using placebos as controls) some degree of deception appears to be endemic, although the studious ambiguity of published papers conceals its extent. Documenting this problem of compound deception—deceptive experimentation overlaid by misleading reports—Franklin Miller and Ted Kaptchuk pointed out in 2008 that descriptions of research methods in published papers ‘typically do not highlight the use of deception. Instead, the use of deception must be inferred by the reader.… More problematic is the typical boilerplate statement in published articles of deceptive studies that informed consent was obtained from research participants’.9 It seems that papers based on deceptive placebo-research resort not so much to plain lies as to tactical silences and standardised cover language—in effect, half-truths. Claims that informed consent was observed in these cases represent half of the truth in that the study subjects probably did sign consent documents, though the consent given was less than informed.

An example concerns a study which Kaptchuk singled out for censure in the 2008 paper, and which was pitched to subjects as an investigation of analgesia using traditional acupuncture, whereas the acupuncture was really a sham. According to the report of the study in question, ‘Experiments were conducted with the understanding and written consent of each subject and approval by the Human Subjects Committee at Massachusetts General Hospital. At the end of the experiment, all the subjects were told about the true nature of the experiment, because they had been recruited with the understanding that this was a study about acupuncture analgesia’.10 The first sentence conveys the impression that all is ethically in order, though if that were the case there would presumably be nothing to debrief the subjects about. The subjects’ ostensibly informed consent turns out not to have been informed. The field of placebo studies seems to lend itself to such equivocations.

A striking instance of tactical equivocation figures at the very centre of a study of the influence of verbal suggestion on pain relief in patients with irritable bowel syndrome. Upon entering the study ‘the patients were told that four drugs that reduced and increased pain in relation to IBS, respectively, were being tested, and that they had proved effective in previous studies’. One of the ‘drugs’, however, was in reality a placebo (and another a nocebo). Thus, when the subjects were informed, ‘The agent you have been given is known to significantly reduce pain in some patients’ just after a balloon was inserted into their rectum, they had reason to believe the ‘agent’ was a drug, even when it was not.11 Making the carefully constructed equivocation more misleading is that the person delivering it was none other than the subjects’ own doctor. We do not expect our doctor to talk to us in double meanings. Commenting on this study (to which the subjects reportedly gave informed consent), a paper on ‘Deception in Research on the Placebo Effect’ notes that ‘the participants were deceived by being informed that they would receive drugs that were in fact placebo interventions’.12 But because placebo treatment for irritable bowel syndrome (IBS) had previously been shown to yield a ‘somewhat large reduction’ in induced rectal pain, the experimenters contend that they did not really lie to the participants at all, and some accept this line of argument as unproblematic. A leading researcher of the placebo effect, Fabrizio Benedetti, has commented on the form of words used in this experiment on the effect of words: ‘No deception was present in this case, as indeed a placebo is known to reduce pain in some subjects’.2 By the same reasoning, one could present a saline injection as a painkiller.

Another example of the sort of equivocation seemingly authorised by placebo discourse is the argument that it is advisable to tell patients with lower-back pain that ‘studies indicate that acupuncture produces greater relief than usual medical care’ while suppressing the information that acupuncture is in fact no more effective than sham acupuncture—that is, a placebo.13 Clearly the potential for equivocation exists whenever the information to be presented is mixed or ambiguous, as medical information often is. Precisely because the possibility of equivocation is built into medical discourse, restricting deception to cases of outright lying seems artificially limiting in the context of medicine. It is thus appropriate that the 2008 paper on deception in neuroscience research9 broadens deception to include misleading though not utterly false statements, as well as artful silences. Regarding placebos in particular, it is only too easy to conceal their nature while describing them in ways not patently false. A placebo which is by definition inert, but nevertheless appears to induce a response, has plenty of the ambiguity that allows for and attracts equivocation. One could describe a placebo as an active treatment without mentioning that it is not a medication (equivocation by omission), and one could state or imply that, being an active treatment, it operates at the same level of efficacy as a medication (an exaggeration). The next section of this paper looks into exaggerated claims of placebo efficacy.

A case can be made that deception is ethically permissible in placebo research, provided that subjects are notified that it will take place and duly consent.12 But consent is only as good as the disclosure language presented to the subjects for their consent; if that language is designed to appear to say one thing but really mean another (like the statement, ‘This treatment has been shown to help others’, accompanying the administration of a placebo), informed consent can readily be compromised. Significantly, the authors of a proposal for ‘authorised deception’, recognising the potential for equivocation in disclosure language, caution against the use of inaccurate wording in the consent forms themselves.12 The issue of equivocation strikes even deeper than this, however. Deception cannot be authorised unless it is admitted to exist in the first place, and half-truths are often not admitted to be deceptive by those who employ them in placebo research, any more than Benedetti considers deceptive the form of words used in the IBS study just cited. The fact that a half-truth reads at face value like something other than a lie, even though it is designed to mislead and does mislead, does not make it less deceptive; it means it is a well-crafted instrument of deception.

In brief, half-truths are open to the same sorts of objections as lies, but may be more insidious precisely because they are less flagrant. Additionally, protocols for authorised deception in placebo research leave the question of placebo use in clinical practice. Yet there too it would be easy to cloak placebos with forms of words that mislead without technically being false. At this point I want to turn to a second-order claim made or implied in the placebo literature—namely, that descriptions of the placebo as a medication, while of course untrue, engender expectations of benefit that produce their own realisation. With so much equivocation about placebos already in place, such a defence of misrepresentation could readily serve to licence abuse of the placebo effect.

The placebo effect: not a self-fulfilling prophecy

The concept of a self-fulfilling prophecy was first laid out in an essay by the American sociologist Robert Merton on the stubborn, self-reinforcing nature of racial and religious prejudices. In the course of his analysis, though, Merton instances the case of a student who fails an exam precisely because he fears failing it. ‘Convinced that he is destined to fail, the anxious student devotes more time to worry than to study and then turns in a poor examination. The initially fallacious anxiety is transformed into an entirely justified fear’.14 In outline, the case resembles a nocebo event—one in which a treatment (say) produces a side effect not because it is harmful in itself but because of a ‘fallacious’ expectation held by the anxious person having it. In the same way, a medical benefit that we enjoy merely because we expect it constitutes a placebo event. But while expectations may well influence our medical experience, it is not the case that the expectation of medical benefit generates its own realisation in the manner of a self-fulfilling prophecy. According to Merton, racial and religious prejudices confirm themselves so automatically that nothing less than the power of the state is capable of breaking the vicious cycle. One doubts that anxious students condemn themselves to failure quite so inevitably. Certainly the placebo effect does not work so inevitably.

A study much cited in the placebo literature found that the branding of pills enhanced their efficacy, such that branded aspirin outperformed generic aspirin in relieving headache pain, as branded placebo outperformed its generic counterpart.15 It would seem that advertising boosts a pill's analgesic effect by proclaiming that effect loudly enough to establish a brand name. By the act of trumpeting a product's powers, the manufacturer calls them into effect. What if the manufacturer were to allege to the world, in so many words, that its aspirin has been proven to work better than other aspirin? We can only hope that would be disallowed as false advertising. But if doctors were to claim that the act of recommending (‘advertising’) placebo to the patient as a medication makes it perform like a medication, would this not be like the aspirin maker arguing in its own defence that advertising the superior efficacy of its product creates just that efficacy? Emphasising as it does the power of words to elicit therapeutic effects, the placebo literature nearly makes this claim. Consider a case reviewed in Benedetti's paper, ‘How the Doctor's Words Affect the Patient's Brain’.16

A number of postoperative patients, divided into three groups, are treated over 3 days with (a) a basal infusion of saline solution and (b) buprenorphine on request. The first group is told nothing, the second that the infusion might be either a potent painkiller or a placebo (as in trials where the chance of receiving an active treatment is 50/50) and the third that the infusion is a potent painkiller. How much buprenorphine did each group request? Group 2 requested 21% less than group 1 and group 3 34% less. This finding is in keeping with the consensus of placebo research that definite expectations are more powerful than qualified ones—‘it does work’ is more powerful than ‘it may work’, as Benedetti puts it—as well as with the principle that a treatment that would otherwise be inert proves not to be inert at all when used in a medical setting with appropriate words, ritual, theatrical props, and contextual touches and effects. Arguably, though, it is closer to the truth to say that saline solution may work, than that it does work, as a painkiller in this case. As Benedetti acknowledges by referring to the message given to group 3 as deceptive, nothing, not even the power of words, can in fact convert saline into a painkiller that does work. A painkiller that worked no more than a third of the time (the same fraction Beecher arrived at in his now-criticised estimate of the power of the placebo) would be a sad treatment of choice. We note that when patients in the experiment requested painkiller, they received a bona fide analgesic, not something that works one-third better than nothing.

By contrast, in the irritable-bowel experiment discussed above, placebo performed as effectively as lidocaine. How was this parity achieved? According to the authors, ‘in addition to the verbal suggestion, there may have been a number of interpersonal and contextual factors that contributed to the manipulation of pain levels’,11 including the hospital setting and the patients’ history with the doctor who performed the experiment. That the same doctor who cared for the patients experimented on them constitutes a blurring of ‘therapeutic and research roles’.12 It is also mentioned that the doctor ‘took time to talk with each patient before the experiment’,11 though what he or she said is not reported—a significant dark spot in a study of the contribution of verbal suggestion to the placebo effect. In all, then, pain carefully administered by the patients’ own doctor, who briefed them beforehand and vouched for the agent he or she treated them with, proved unusually responsive to placebo. Ordinarily placebo does not perform at so high a level. In a meta-analysis of pain studies involving 130 patients who received placebo (the IBS study had a total of 13), only 16% showed a placebo response of more than 50% of the ‘maximum possible pain relief score’.17 Do the words ‘This [placebo] will work’ act like a charm capable of boosting placebo to parity whenever it is recited? The buprenorphine experiment suggests not. Doctors are not mages whose words make it so, and despite the findings of the IBS study they do not have the power to dictate reality by the use of appropriate words and theatrical techniques, as another study performed by a doctor on his patients makes clear.

In the study I refer to, which has achieved classic status even though it was conducted on patients without their knowledge, those told positively that they would be better in a few days fared better than those told that the doctor could not be sure what was wrong with them, despite the fact that the two groups suffered from the same transient complaints and the only variable was the doctor's manner.18 By the use of certain potent words, the doctor relieved patients—but by no means all of them. In the event, 64% of the reassured patients improved, compared with 39% of the others. Thus, many of the patients reassured by the doctor in definite terms did not improve, while about the same number improved without being reassured at all. If the doctor is a therapeutic agent, the study suggests that he or she is an agent that may or may not work in the case of the symptoms presented by minor, self-limiting ailments, an agent whose power is therefore considerably more modest than implied by portrayals of the placebo effect as a self-fulfilling prophecy, as something that does work, or even—as will be seen—as an act of shamanism. (One doubts that a study where treatment consisted of telling patients they would soon get better, supplemented in half the cases by a placebo pill, would have been conducted with patients who actually required medical treatment.) Read as a demonstration of the power of expectation, the study supports the tendency to exaggerate the placebo effect. Insofar as it is not read as a case of breached informed consent, the study indicates an ethical blind spot in the literature.

Expectations can influence outcomes but by no means dictate them, although this distinction may be lost to the goal of phrasing the placebo effect more rather than less positively. It is very possible to describe the power of expectation misleadingly, in a way that receives reinforcement from many other half-truths that envelop the placebo effect, some documented above. Practices like these add up. However, unlike equivocations inspired by the desire not to reveal too much (as in accounts of consent procedures in the literature), exaggerations of the doctor's power to ‘heal’ by raising the expectation of healing are backed up by the widely held notion that the more strongly the placebo effect is described, the more strongly it operates. In a study that appeared 30 years ago but still reads as current, dental patients were given an inert pill along with either an ‘Oversell message’ (‘This is a recently developed pill that I've found to be very effective’) or an ‘Undersell message’ (‘This is a recently developed pill that reduces tension, anxiety and sensitivity to pain in some people.…I personally have not found it to be very effective’), before being asked to rate the pain of an injection. The ‘Oversell message’ won. Stronger phrasing, stronger expectation, stronger reported effect.19 No wonder there is an inbuilt tendency in placebo discourse to oversell the placebo effect itself. According to an often-cited book devoted to the placebo effect, for example, ‘Insofar as [the doctor's] convictions are somehow conveyed to patients and, in the process, convinces them of the doctor's power, then they are likely (within the bounds of our physical mortality) to be effective’ (emphasis in original).20 If the doctor's ability to make things so (by eliciting faith in his or her ability to make things so) is constrained only by our mortality, then this person's power goes far indeed.

If the placebo effect thrives on overstatement, by the same token modest recommendations of the power of the placebo run contrary to the thrust of the literature. In an effort to balance the demands of truthfulness and benevolence, Dr. Howard Spiro would tell some of his patients, ‘I'm going to have you get some B ₁₂ injections. They have helped many other patients, but I cannot explain to you why they work and I cannot promise you they will work. I can simply say that many patients tell me they feel better and stronger after such a course of therapy’.21 According to the placebo literature, this is the way not to mobilise the power of the placebo. It is an Undersell message. It says something about Dr. Spiro's uniqueness that he seems to have been able to rally the placebo effect (relieving and encouraging his patients) while doing much to quell it. He might be taken as the exception confirming the rule that the placebo effect calls for some degree of false advertising—though even he did not admit to his patients that the vitamin injections in all likelihood had no particular medical utility; they were impure placebos.

What's the harm of exaggerating the power of the placebo? Why does it matter?

Few at this moment would deny that placebos are not to be used covertly at the doctor's discretion, as in the days before the requirement of informed consent. If the ethical use of placebos calls for informed consent as most now agree, the potential for exaggeration that is part and parcel of placebo discourse makes such consent more difficult. As we have seen, the practice of informed consent in placebo experimentation has already been compromised by equivocation. The pressure exerted on informed consent by the use of placebos is suggested by a recent open-label placebo trial which arguably overstated the power of the placebo. Telling the enrolled patients with irritable bowel syndrome that ‘placebo pills, something like sugar pills, have been shown in rigorous clinical testing to produce significant mind-body self-healing processes’,6 the experimenters distinctly implied that placebos have shown the potential to heal a condition like IBS, as in fact they have not. IBS is not like ulcers, which heal in response to treatment and placebo (as well as nothing at all), and at the end of the study the subjects were not even asked if they had healed. Additionally, the reference to subjects who responded to sugar-like pills in past trials suppresses the fact that they didn't know the pills were placebo; in this sense an element of concealment was grandfathered into this study of placebos without concealment. Note too the inflation that raises clinical testing to ‘rigorous clinical testing’ and self-healing processes to ‘significant …self-healing processes’. The study is intended as a preliminary test of the possibility of using placebos in clinical practice, and if they were, we can be sure that there too they would be oversold. Nothing but salesmanship can make up for the loss of the advantages of deception that have been exploited so impressively in placebo research.

As noted, investigators generally agree that the use of placebos as controls in clinical trials under-represents the potential of the placebo in that study subjects know they have only a 50/50 chance of getting an active treatment; expectations are a driver of (but as we have seen, do not determine) the placebo effect, and their expectations are diminished. As two analysts put it, This situation creates uncertainty among the patients randomized to real or sham treatments about what they actually received. In contrast, in clinical practice when patients receive a placebo medication or a treatment such as acupuncture for osteoarthritis of the knee, which in fact may be no better than sham acupuncture, they are not uncertain about whether they have received an actual treatment. The significantly different informational context and associated patient expectations regarding treatment in placebo-controlled trials, as compared with routine clinical practice, may influence the therapeutic efficacy of treatments that work solely by virtue of a placebo effect.22If patients who receive placebo are in no doubt that they have obtained an actual treatment, this can only mean that the placebo has been sales-pitched to them as something that does work, just like a medication. Thus the authors conclude, ‘In sum, the effect of placebo analgesia is markedly greater when patients are told that a placebo treatment is a powerful painkiller than when they are told that they may receive either a powerful painkiller or placebo’. The advertising of placebo as a powerful painkiller constitutes, of course, a deception. We can readily imagine the sort of half-truths that the use of the ‘informational context’ to frame ‘expectations’ could lead to, as bland as such terminology is.

American television has sprouted all manner of ads for vitamins, energy drinks, immune-system boosters, system-balancers, ‘probiotics’—something like impure placebos—that are misleading but nevertheless not illegal, because they are covered by the disclaimer that the product has not been evaluated by the Food and Drug Administration and is not intended as a medication. It would be a pity if doctors were to stoop to this level of equivocation by exaggerating the power of the placebo in the ways I have highlighted.

However, the harms of overselling the placebo effect go beyond the use of questionable advertising and the impairment of informed consent. In the conclusion of his 2002 paper on the influence of the doctor's words, Benedetti did not hesitate to liken the doctor to a shaman. ‘One must bear in mind that the word “doctor” can be replaced with the more general term “healer”’, he wrote, to make it clear that the ‘shaman factor’ is always present in any medical treatment. In other words, the placebo effect and the mechanisms underlying the healer-patient relationship act irrespective of whether the therapeutic approach comes from conventional or unconventional treatments. What counts, at least for some symptoms, is the context and the interaction of patients with their healers, be they doctors, psychologists, or shamans.Ten years later, after ideas like this had gone into circulation (minus the qualifier ‘at least for some symptoms’), the author was less at ease with the implications of equating doctor and shaman. In a commentary on unscrupulous practice and ‘the vulnerability of the patient’, he wrote this time, If a syringe filled with distilled water and handled by a doctor may induce expectations of benefit, then the same expectations can be induced by talismans, mascots and bizarre rituals carried out by quacks and shamans. Particularly in very recent times …I personally was deluged with requests and proposals of new bizarre procedures, concoctions, talismans and mascots that could possibly enhance expectations, beliefs, trust and hope.23Benedetti concluded that ‘the new neurobiology of suggestion and placebo faces the ethical dilemma of what to do’, one so disturbing that it ‘presage[s] a worrisome future for medicine’.

Recommendations

An appropriate response to this dilemma is first of all to consider the placebo effect as a therapeutic benefit arising from the conscientious performance of the rituals of good medicine, and not as a resource to be tapped by the use of trickery (with equivocations counting as trickery) or dispensed in the form of pills. In connection with this last point, the cited study of the therapeutic effect of a positive manner conveys a suggestive message. It found no difference between the number of reassured patients treated with a placebo pill who improved, and the number receiving reassurance alone who improved; the pill added nothing.18

Second, it is just as important to recognise, that (a) medical expectations are not self-fulfilling (which does away with the given rationale for talismans and mascots); and (b) while placebos can change our experience of illness, they do not ‘heal’, if ‘heal’ means ‘cure’ (as my dictionary says). ‘There is little reliable evidence that the placebo effect can cure or control disease by modifying pathophysiology’.3 Like other statements I have cited, the suggestion that placebos ‘heal’ is at once ambiguous and misleading. To the doctor who recommends a placebo as a healing agent, this bit of phrasing may mean nothing more than that placebos can make you feel better. To lay ears, however, the claim that placebos heal may sound very much like a claim that they cure. After all, when we say, ‘Physician, heal thyself’, we mean something more than ‘Physician, feel better’. We mean, ‘Physician, cure your own malady’. A faith-healer, too, doesn't just help clients feel better. If they are lame, they are supposed to lose their lameness; if blind, their blindness; such is faith-healing. The statement that placebos contribute to healing (now common in the literature) thus resembles many claims in placebo discourse that employ a kind of informational double entendre, conveying a false impression under the guise of something uncontroversial. Placebo-champions are on especially risky ground when they allege that placebos can promote the ‘healing’ of cancer patients,24 which can easily be taken by the desperate to mean that placebos can, indeed, cure.

Ever since Beecher's ‘Powerful Placebo’, the placebo's potential to reduce pain has been emphasised in the literature. But while experiments in which subjects might or not might receive placebo are not the most effective way to elicit the placebo effect, in the case of pain, studies may in fact overstate it. If only because pain studied, controlled, even administered in the laboratory differs from pain of unknown origin capable of exciting anxiety and even terror, studies of placebo treatments of pain do not translate straightforwardly into clinical practice. Beecher himself noted in another paper that ‘It requires little imagination to suppose that the sickbed of the patient in pain, with its ominous threat against his happiness, his security, his very life, provides an entirely different milieu (and reaction) than the laboratory, with its dispassionate and unemotional atmosphere’ (emphasis in the original).25 Scaling down placebo rhetoric may, therefore, also necessitate qualifying our estimate of the actual pain-reducing potential of the placebo.

If the placebo effect represents a self-fulfilling prophecy, then in order to realise it, all medicine needs to do is induce the expectation that it will occur. Such a technique would have a boundless application. The portrayal of the placebo as a potentially all-transforming force—in effect, a panacea—recalls the ethereal fluid known as animal magnetism, summoned and channelled by the shamanistic healer Franz Anton Mesmer, which inspired the first investigation of the placebo effect in the late 18th century.26 (Mesmer did not merely claim that he made people feel better. He professed to cure them, and cited written testimonials to his power.) If we are to avert the worrisome future projected by an investigator who has done much to advance understanding of the placebo effect, it is necessary to root the placebo effect in the attentive practice of medicine itself and, finally, to repudiate the equivalence of doctor and shaman, and with it the notion that the doctor's power to heal is constrained only by his or her ability to raise the expectation of healing.