Placebo by proxy

The effect of placebo on the patient and the patient’s environment is often debated, but people other than the patient may also feel better when a patient receives placebo treatment. The anthropologist Claude Lévi-Strauss alluded to this when he described how medical treatments, including those where ritual alone provides the cure, occur in a social context and convey a “sense of security” to the social group.1

Clinicians and family members may have an emotional response to a patient’s treatment and think that the treatment is helping the patient even in the absence of any direct physiological benefit to the patient or indication from the patient that the treatment is working. These feelings and perceptions may arise when placebos, including “impure” placebos, such as active drugs or operations that have no effect on the disease process, are used in clinical practice and research settings. Because these feelings and perceptions are not accounted for in descriptions of the placebo effect and can exist independently of any placebo effect on the patient, they can be described as placebo effects by proxy, or placebo by proxy for short. Although placebo by proxy has important implications, the phenomenon is underappreciated and rarely discussed.

Placebo by proxy could be as ubiquitous as the placebo effect, and in some situations placebo by proxy can dominate clinical decision making. For example, the parents of a child with a viral upper respiratory illness may believe that the child needs antibiotics, and parental expectations or the doctor’s perception of parental expectations (or both) may influence prescribing patterns.2 3 Antibiotics are overprescribed in these situations and function as impure placebos4; the psychological benefit to the parent, such as a relief from worry, represents placebo by proxy.

In the absence of evidence to support long term use of antipsychotics in patients with dementia and a history of agitation, aggression, or psychosis,5 6placebo by proxy may help to explain why clinicians often maintain their patients on these drugs. In a controlled double blind study of discontinuation of psychotropics in a sample of patients with dementia who were taking these drugs because of a history of agitation (67.3% were taking antipsychotics), the authors found a lack of an effect of psychotropics when they compared mean scores on behaviour rating scales during the treatment and placebo phases of the trial.6 However, clinical workers were resistant to stopping treatment because they had “a great deal of faith in the utility of these drugs to control agitation in this population.”6 Psychotropics functioned as impure placebos for many of these patients, and the concern of clinicians that patients would decompensate without these drugs provides another example of placebo by proxy.

In addition to influencing clinical decision making, placebo by proxy could influence estimations of treatment outcome in trials, especially when the perceptions of clinicians or family members are favoured over objective markers of patient response.7 In placebo controlled clinical trials of secretin to reduce symptoms of autism, the patient’s response to secretin was primarily based on the opinions of parents, teachers, or clinicians (or a combination thereof).8 Ten randomised controlled trials and a meta-analysis showed robust and consistent reductions in core symptoms of autism as measured by standardised rating scales in both the treatment and placebo groups, but no significant differences were seen between the groups on these measures.8 9 The robust response to placebo may have been influenced by anecdotal reports, which generated positive press about secretin and gave rise to the need for clinical trials.8 9 Perception or misperception by parents may also partially explain a larger placebo response in children than in adults participating in clinical trials of treatment resistant epilepsy.10

Placebo by proxy may also explain findings from meta-analyses where improvement measured by observers is much higher than improvement reported by patients. For example, for depression, the mean effect size of observer rated improvement is 1.85 (95% confidence interval 1.69 to 2.01), compared with 0.67 (0.49 to 0.85) for patient reported improvement.11 For irritable bowel syndrome, the pooled placebo response rate for physician rated improvement is 53.0%, compared with 37.4% for patient rated improvement (P=0.005).12

The different mechanisms that underlie the placebo effect are likely to shape placebo by proxy also.12 These processes may include seeing other patients respond to the same drug, associative learning such as conditioning, a supportive physician-patient relationship, and reduced anxiety.

Placebo by proxy and the placebo effect may interact to create positive change. For example, if clinicians and family members feel empowered and optimistic about a disease, the patient’s environment can become less stressful and more supportive. Clinicians and family members may react to placebo by smiling more, paying more attention to the patient, promoting treatment adherence, encouraging the patient to engage in new activities, or creating other targets for behavioural change.8 9 In this way, placebo by proxy may elicit changes in the patient’s psychosocial context that mediate the placebo effect.

Although placebo by proxy may enhance clinical outcomes, it might also cause harm. If treatment related decisions are unduly weighted towards providing a psychological benefit to people other than the patient, the patient may not benefit from the treatment yet may still bear the risks. The treatments mentioned above cost money and are not benign: antibiotics cause diarrhoea and are associated with antibiotic resistance, antipsychotics are associated with an increased risk of death in elderly patients with dementia related psychosis, and secretin infusion may result in a life threatening allergic reaction. Placebo by proxy may create a false sense that a patient is getting better and thereby prevent more appropriate treatment,7 8 or it may produce changes in attitude and behaviour towards the patient that lead to neglect. Placebo by proxy is probably more likely to cause harm when decisions are made in the absence of clinical evidence and when patients cannot make decisions for themselves.

Treatment benefits could be maximised and harms mitigated if placebo by proxy were taken into account when making clinical decisions. For many of the above clinical situations, an appreciation of placebo by proxy may prompt clinicians to question attachments to treatment practices that are not supported by evidence. Placebo by proxy could be used as a potential clinical tool. For example, it may be possible to boost the psychological benefit of a patient’s treatment by generating appropriate enthusiasm for treatment among those involved in the patient’s care. Because almost every treatment related decision is shared and patients sometimes agree to treatment to make their clinicians and family members happy, clinicians should be aware of placebo by proxy when guiding clinical decisions and evaluating treatment response.