The aim of the present study was to explore the use and understanding of the concepts ‘placebo’ and ‘placebo effect’ in 12 empirical studies that have addressed the prescription of placebos by doctors in clinical practice. There were great differences in the general methodology and in the definitions (or lack of any definition) of the basic concepts in these 12 studies. Therefore, the results reflect different things. They tell us a little about the use of ‘pure placebos’, more about the use of ‘impure placebos’, but most of all, they tell us about the conceptual confusion in this area.
Statistics from Altmetric.com
‘Half of doctors routinely prescribe placebos’1 wrote the New York Times in its headline on 23 October 2008. ‘Half of German doctors prescribe placebos, according to a new study for the German Medical Association’2 reported the Guardian in March 2011, and continued, ‘In Bavaria, it found, 88% of GPs have sent patients home with prescriptions for placebo drugs’.2
These puzzling headlines were based on empirical studies in the USA3 and Germany,4 respectively. The immediate questions they raise in the minds of lay people and professionals are: (1) do doctors around the world really use placebos—inert substances—deliberately and frequently without telling that to their patients? (2) If doctors report the use of placebos, but do not refer to inert substances, what exactly do they mean by placebos?
As usual, the original reports paint a much more complex and colourful picture. According to the first report, ‘US internists and rheumatologists commonly recommend ‘placebo treatments’’ but the most commonly prescribed such treatments were vitamins and over-the-counter analgesics.3 According to the second report, ‘Placebo interventions are a widely accepted part of medical treatment in German general practices and are used primarily for their psychological effects. Impure placebos are used much more frequently than pure placebos’.4
Thus, it seems that the placebo concept has been interpreted in ways that do not refer only to placebos as inert substances.
The concept ‘placebo effect’ has also been used with many different meanings. In the words of Benedetti,5 ‘The inappropriate use of the words ‘placebo effect’ and ‘placebo response’ has pervaded the medical literature for many years, and even today these terms are sometimes confused with other phenomena’. Because of the conceptual confusion, several alternative terms have been suggested,6 7 none of which has so far been accepted for wider use.
The aim of this study was to explore the use of placebos in clinical practice and the use and understanding of the concepts ‘placebo’ and ‘placebo effect’ in empirical studies that have addressed the prescription of placebos by doctors in clinical practice.
A Medline search was performed to identify studies that addressed the actual use of placebos by doctors in clinical practice. Studies about attitudes towards placebo use and studies older than 20 years were excluded. Details of the search are described in supplementary figure 1.
Studies on the use of placebos by doctors
In all, 12 studies that met the inclusion criteria were found. They were undertaken between 1993 and 2011 in Sweden,8 Denmark,9 Israel,10 Switzerland,11 Canada,12 New Zealand,13 Germany (two studies)4 14 and the US (four studies).3 15–17 A total of 11 studies were published as scientific articles in peer-reviewed journals, and 1 study13 was published as a letter. Only one study provided quantitative data on the supply of placebo drops and tablets from the hospital pharmacy.14 The studies are described in the supplementary appendix.
There was considerable variation in the definitions. It could be narrow and resemble the standard understanding of placebos in the research context: ‘… an inactive substance administered to a patient in place of a medication’.15 Or it could be wide, as in one of the US studies16 and the Canadian study,12 which provided several alternatives for the definition of placebo. The respondents could give their own definition or choose between the following alternatives, the second of which was exactly the same as in the Danish study9: ‘(1) an intervention, that is, not expected to have an effect through a known physiologic mechanism, (2) an intervention not considered to have a ‘specific’ effect on the condition treated, but with a possible ‘unspecific’ effect, and (3) an intervention that is inert or innocuous’.12 16
In one study an explicit distinction between ‘pure placebo’ and ‘impure placebo’ was made in the questionnaire: ‘‘Pure placebos’ refer to inert substances or methods, whereas ‘impure placebos’ include substances or methods which do have a known pharmacological or physical activity but which cannot be expected to have any direct therapeutic effects for the respective disease and in the chosen dosage’.11 In another study the questions about impure placebos were posed without using the term ‘placebo’. Instead, they were described as ‘drugs and other therapies with no intrinsic effects (eg, pharmacological or physical activity) on the current patient's symptoms’.4 One definition seems to have covered both meanings: ‘a substance with no known specific pharmacological activity against the condition being treated’.17
In some studies lists of impure placebos were provided already in the questionnaires. Fässler et al,11 for example, list the following interventions as potential placebo interventions: ‘Positive suggestions/Simple ointments and/or bandages for contusions without visible skin damage/‘Sugar pills’/Injections with saline solution/Therapies without pharmacological or physical efficacy for the patient's conditions (eg, vitamins or antibiotics without approved indication)/Diagnostic practices, maybe on patient's request or to calm the patient’. A statement in the questionnaire of Raz et al12 was formulated as follows: ‘I have prescribed or given the following form(s) of treatments in situations without demonstrated or expected clinical efficacy: Antibiotics (eg, for viral or other non-bacterial diagnoses)/Saline infusions or saline intramuscular injections (eg, for pain)/Ibuprofen/Vitamins/Sugar or sweetener pills/Herbal supplements/Prepared placebo tablets/Sub-therapeutic doses of medication/Other’.
Of the papers included in this study, the newer studies commonly referred to earlier studies without any critical discussion of the differences between the study settings and the definitions used in the questionnaires.
There was a large variation in the frequency of prescribing or administering placebos among doctors in the 12 studies included in this investigation. The lowest and highest reported proportions of doctors who confirmed that they prescribed or administered placebos in their clinical practice were 20%12 and 88%,4 respectively. However, due to the variations in the definitions and study approaches, these figures should not be interpreted as a measure of the same phenomenon.
In the studies that provided the respondents several options for the definition of placebo,12 16 the option ‘an intervention that is inert or innocuous’ was the least popular answer. This demonstrates the fact that clinicians often introduce some aspects of their understanding of ‘placebo effects’ into the placebo concept.
When it was possible to separate the use of pure and impure placebos, the latter was far more common. In one of the US studies, for example, 55% of the respondents reported having recommended at least one placebo treatment, which usually meant an impure placebo (or, according to the concept used in that particular study, ‘active placebo’). Only 2% had recommended ‘sugar pills’ and 3% saline.3
The difference between pure and impure placebos was also found in the Danish study in which ‘70% of general practitioners reported having used placebos in the form of antibiotics within the last year and roughly 50% in the form of B vitamins, physiotherapy, or sedatives’.9 This finding illustrates the heterogeneous understanding of the placebos among the respondents.
Only one study provided quantitative data on the supply of placebos from a hospital pharmacy. In 2007, for example, 1100 placebo tablets were prepared and delivered to the 24 clinics taking care of more than 45 000 inpatients every year.14 When the size of the hospital is considered, the amounts were thus small.
Actual administration of placebos
The studies referred to the use of placebos in different ways. Several questionnaires asked whether the doctor had ‘administered or prescribed a placebo’, while others mentioned having ‘used pure placebos’,4 ‘recommended placebos’,3 or ‘ordered placebos’.15
There was, however, hardly any discussion about the actual prescription or administration processes. It remains unknown how often the respondents had used written prescriptions of pure or impure placebos, how often they had recommended something that was otherwise available for the patient and how often they had simply agreed upon a patient's wish to use something that the doctors themselves considered to be a placebo.
The studies discussed in this article demonstrate that the concepts placebo and placebo effect are understood in such different ways that a direct comparison of usage statistics is not reasonable. They also demonstrate that the conceptual difference between placebo use and placebo effect is often not made. The latter represents a wide spectrum of phenomena, and only a small minority of them is related to the use of placebos. Sherman et al,16 for example, failed to differentiate between the two when they wrote: ‘In the broader ethics literature, some commentators on informed consent and nondeceptive therapeutics caution against the use of placebos in medical practice. Others propose that the placebo effect can be harnessed in various therapeutic contexts that do not pose ethical dilemmas’. Kermen et al17 concluded that ‘Many US family doctors use placebos and generally believe the placebo effect has both psychological and physical benefits’.
Some of the studies refer to the distinction between specific and unspecific effects. Hrobjartsson and Norup,9 for example, defined placebo ‘as an intervention not considered to have any ‘specific’ effect on the condition treated but with a possible ‘unspecific’ effect’. In the paper, the authors deliberated on the concepts as follows: ‘Whereas this distinction is relatively clear in connection with pharmacological interventions, it is less obvious in connection with non-pharmacological treatments such as physiotherapy’. Although common in the placebo literature, the concepts ‘specific’ and ‘unspecific’ are vague. What else could ‘unspecific’ refer to than the fact that we do not know the mechanism of its effect?18
Perhaps the most ambiguous concept in the studies was ‘impure placebo’. As the preceding examples and citations suggest, the concept covered a large variation of drugs and therapies, and it is highly debatable whether the reasons and ways they are used justify classification under one heading. In the systematic review by Fässler et al,19 the problematic nature of ‘impure placebo’ is pointed out: ‘The academic concept of an impure placebo might inappropriately reflect the complex situations and motivations in which healthcare professionals apply interventions which are not backed up by scientific evidence’.
Tilburt et al3 also noticed this: ‘Physicians might have multiple, possibly conflicting, beliefs and motivations for recommending active placebo treatments. They might believe that a given benign and safe treatment, such as some indicated by our respondents (eg, vitamins and over-the-counter analgesics), might have some finite chance of a beneficial pharmacological effect, despite remaining unproved for the symptoms or condition at hand; or, at the very least, such treatments might promote positive expectations in patients or help the patient to feel like they are being taken care of’. However, in their conclusions, the same authors ignored this comment when they stated: ‘Prescribing placebo treatments seems to be common and is viewed as ethically permissible among the surveyed US internists and rheumatologists’.
The case of antibiotics illuminates the ambiguity of the concept of ‘impure placebo’ particularly well. Respiratory infections, for example, are viral by origin in most cases, but particularly in general practice, laboratory tests are hardly ever made for confirmation. In prolonged or atypical cases doctors often prescribe antibiotics although no firm evidence of a bacterial origin exists. Furthermore, many patients have viruses and bacteria simultaneously, and it is not even possible to determine for sure whether antibiotics are necessary.18 20
Only one study14 included quantitative data about the actual provision of prepared placebo tablets from the hospital pharmacy. In light of the size of the hospital and the corresponding population, the use cannot be considered common.
The study reports hardly mentioned the actual administration of placebos. Writing a prescription for a placebo without telling the patient that the doctor believed it to be inert, suggesting the use of a placebo to a patient, and allowing the use of a placebo on the request of a patient are very different kinds of situations.
The availability of prepared pure placebos in outpatient care was mentioned in only one study: according to one of the German studies,4 21% of GP's stocked placebo tablets in their practice. The availability of pure placebos is not, however, obvious across the world. For this study I checked the situation in three countries and at least in the UK, Sweden and Finland it is not possible for a doctor to prescribe prepared placebo tablets (Iona Heath and Rolf Ahlzén, personal communication).
There was also very little discussion in the reports that compared the prescription of pure and impure placebos. Deliberate prescription of prepared placebo tablets differs in many aspects from the prescription of, for example, vitamins in situations in which the doctor believes that there might be a chance of a beneficial effect although there is no particular evidence supporting it.
There were great differences in the general methodology and in the definitions (or lack of any definition) of the basic concepts in these 12 studies. Therefore, the results reflect different things, and it is possible that the range of responses is an artefact of the problematic methodologies. The results tell a little about the use of pure placebos, more about the use of so-called impure placebos, but most of all, they tell about the conceptual confusion in this area.
More research is clearly needed. Questions that should be addressed are, eg, (1) how do doctors understand and explain the interactions in which they have administered or prescribed ‘pure’ or ‘impure’ placebos? (2) How are the patients informed in these situations? (3) If pure placebos are administered, how does that exactly take place?
It may well be that doctors have far more nuanced views than the studies have suggested so far. Quantitative surveys may simply be inadequate to address the complexity of the clinical reality here. Instead, qualitative approaches are probably needed to address these questions.
I wish to thank Dr Harri Hemilä for valuable advice and Dr Raimo Puustinen for general inspiration.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.