The thinking doctor’s guide to placebosBMJ 2008; 336 doi: https://doi.org/10.1136/bmj.39564.454502.C2 (Published 01 May 2008) Cite this as: BMJ 2008;336:1020
- Rudiger Pittrof, consultant in integrated sexual health and HIV1,
- Ian Rubenstein, general practitioner2
- 1Department of Reproductive and Sexual Health Care, Town Clinic, Enfield, Middlesex
- 2Eagle House Surgery, Ponders End, Enfield
- Correspondence to: R Pittrof
Discussion of the planned use on patients of the placebo effect is long overdue. For many years it has been medicine’s “dirty little secret.” Clinicians have found it to be an embarrassment that needs to be factored in when planning randomised controlled trials. Ethically its use has always been considered dubious. Recent high profile publications suggesting the lack of efficacy of selective serotonin reuptake inhibitors (SSRIs) in mild to moderate depressive illness have shown that clinicians routinely make use of the placebo effect—but from a position of ignorance rather than intention.1
Whatever one’s ethical stance, this state of affairs cannot be a good thing for our patients, our relationship with our patients, or the public purse. We outline a third way whereby clinicians can use the placebo knowingly yet still stay within the ethical constraints of modern medical practice.
Drugs given to conscious patients have at least three effects: one mediated through the biomedical activity of the drug; a nocebo effect2 (the opposite of placebo: negative non-biomedical effects); or a placebo effect.3 The placebo effect is usually expected to be synergistic with the biomedical effect and is often thought to be of less importance. Indeed a Cochrane review of placebo versus no treatment concluded: “There was no evidence that placebo interventions in general have clinically important effects. A possible small effect on continuous patient-reported outcomes, especially pain, could not be clearly distinguished from bias.”4 Other researchers have come to different conclusions: “The placebo effect appears in those instances where it is expected.”5 And others found neurobiological mechanisms of the placebo or nocebo effect.6 7
Hróbjartsson and Gøtzsche found that placebos are no panacea and do not work in all clinical conditions.4 This cannot imply that placebos should not be used, as interventions known to be highly effective are not panaceas either. There is strong observational evidence that placebo treatment can lead to measurable and lasting benefits.
Good treatment must be based on good evidence, and the gold standard of medical evidence is the randomised controlled trial. However, for two main reasons placebo controlled randomised trials are not the best way to assess the placebo effect. Firstly, the placebo effect depends on the belief of patients that they will receive an effective treatment that will improve their condition. This scenario differs from that in placebo controlled trials, which are ethical only if no effective treatment is known.8 Informed consent ensures that participants are aware they might receive a placebo or a drug that might not be effective. It is likely that this will reduce the placebo effect. By their very design randomised controlled trials are an imperfect tool to assess placebo effects. Secondly, studies where a placebo has significantly better results than an active drug may fall victim to publication bias.
Despite this, Hróbjartsson and Gøtzsche’s meta-analysis found significant improvement in placebo over treatment for some of the most common medical conditions (common cold, insomnia, asthma, and phobia).4 Clinical use of placebo treatment may be seen as unethical if more effective interventions are available; there are, however, at least four strong reasons why this is not always the case:
The evidence base—Placebo treatment is extremely well researched. In many studies, improvement against baseline has been reported in the group of patients who receive the placebo treatment.
Cost—Placebos are not protected by patent laws and are cheaper than active drugs. Where placebo treatment results in perceived better health, placebo treatment is likely to be the most cost effective intervention. In a resource restricted environment, cost becomes an ethical issue.
Safety—Although side effects occur with placebo treatment, serious side effects (such as teratogenicity or death from overdose) of the type that occur with drug treatments must be extremely rare.
Individualisation of treatment—Clinical trials usually show a range of effects of active and placebo treatment on the study population. Many clinicians believe that they can select those patients who are most likely to respond to placebo treatment.
For some chronic, distressing, fully evaluated, and relatively harmless conditions such as common colds, insomnia, phobias, most cases of premature ejaculation, and some pain or mood disorders a trial of placebo treatment may well be indicated. Despite the low incidence of dangerous side effects, placebo treatment is not risk free. Placebo treatment may result in:
Medicalisation of harmless, self limiting conditions
Insufficient clinical improvement
Ridicule by peers
Perfunctory use of placebo intervention by the doctor as a means of pacifying the patient
An ethically ambiguous position for the clinician, and
Deterioration of the patient-clinician relationship.
Many conditions that can respond to placebo treatment require an ongoing patient-clinician partnership that is based on transparent care and participation of the patient in decision making. Here, placebos can work only if the clinician and patient see them as the most appropriate management; and counselling of the patient and appropriate documentation will be paramount.
The primary ethical objection to the use of placebo is that deception is involved on the part of the clinician. For example, as patients have a right to see their medical records, placebo treatment cannot be recorded as such. Certainly this issue needs to be dealt with. Such deception can be overcome by practising evidence based medicine.
Published evidence only applies to patients if they have similar characteristics to patients in the study population. Even if this is the case, their response can rarely be accurately predicted. This is one of the problems with evidence based medicine: often its application to the individual is under less than ideal conditions. However, where the study shows that placebo results in a response, when compared with baseline, then that is precisely what the study shows. It is possible to state with confidence that, were the conditions of treatment to be replicated as in the study, the patient would have a statistical likelihood of responding to the placebo, just as we can say that the patient is statistically likely to respond to the active intervention. There is thus a response whose cause is debatable but the response is genuine. On strictly scientific grounds no deception would be involved in referring to this response as an evidence base.
We think that doctors can use placebos ethically while remaining within the spirit of scientific, evidence based medicine. To be evidence based, they should prescribe placebos similar to those used in randomised controlled trials that show an improvement over baseline. Clinicians could then be confident of the strength of the evidence and could then tell patients that treatment X has been shown in trials to result in improvement in similar patients. Publishers will have to facilitate the appropriate use of placebo treatment by insisting on a complete description of the placebo in question. This would also offer some protection against ridicule and damage to the patient-clinician relationship when the patient’s details are communicated to other healthcare providers and copied to the patient.
The chief ethical objection to the current use of placebos is that it involves less disclosure of facts to the patient. However, we consider that the ethical use of placebo would in fact involve more disclosure to the patient. This disclosure would involve discussion of the side effect profile and toxicity versus expected effectiveness of the usual treatment or placebo. For example, when advising a patient about treatment for mild to moderate depression clinicians might explain that use of SSRIs may be associated with a higher risk of suicide attempts9 and that about 80% of the benefits of SSRIs might be attributable to a placebo effect.1 In men with erectile dysfunction we can adequately predict that, in comparison with baseline, their confidence, self esteem, and erectile function are likely to improve when they take a placebo.10 Although the improvement will be less than if they took the active drug, there would be no risk of drug interactions, cardiovascular illness, or dangerous side effects.
For many patients a low risk of side effects is more important than a high probability of relief of symptoms. Indeed, where an effective placebo treatment exists for a condition, not offering it may be unethical, as we cannot second guess the importance that patients attach to different effects (or side effects) of the treatment. Anecdotally, we have seen many risk averse patients who would happily accept reduced benefits for much reduced risks.
Given that the use of placebo treatments is not risk free, clinicians need to adhere to a clear strategy for their safe use. Our recommendations are similar to those of Lichtenberg and colleagues.11 A placebo may be used where:
The clinical evidence indicates that a placebo may be useful in a given clinical situation
A response to placebo treatment does not preclude the possibility of the patient having a dangerous condition
Placebo treatment, as a formal trial of treatment, is regularly reviewed (ineffective placebo treatment should be discontinued immediately)
There is no agreed “gold standard” treatment
Placebo treatment is initiated after full discussion with the patient of the treatment options
The patient is made aware that other treatments are available that may be more effective, if this is the case
The patient has given informedconsent and this has been documented, and
The patient is given full disclosure of the nature of the placebo treatment, including the evidence base, if they ask about it.
If these principles are adhered to, placebo treatments would not only be evidence based and cost effective but would offer the best option for respecting patients’ autonomy. Paradoxically, where evidence of a benefit of placebo treatment exists, not discussing it and, where indicated, not prescribing it should be seen as unethical.
We have seen many risk averse patients who would happily accept reduced benefits for much reduced risks