Placebos and sham treatmentsBMJ 2015; 351 doi: https://doi.org/10.1136/bmj.h3755 (Published 10 July 2015) Cite this as: BMJ 2015;351:h3755
- Philip Sedgwick, reader in medical statistics and medical education,
- Carwyn Hooper, senior lecturer in medical ethics and law
- Correspondence to: P Sedgwick
The effectiveness of topical chloramphenicol in preventing wound infection after minor dermatological surgery was evaluated. A randomised placebo controlled superiority trial was performed.1 Participants were patients with high risk sutured wounds who had undergone minor surgery. The intervention was a single application of topical chloramphenicol ointment to the sutured wound immediately after suturing. Chloramphenicol ointment is an antibiotic indicated for the treatment of bacterial conjunctivitis, but it is often used as prophylaxis for, or treatment of, wound infection, although little evidence exists for its effectiveness. The control treatment was placebo ointment, which consisted of a mixture of soft white and liquid paraffin with no known anti-infective properties. In total, 972 patients were recruited and randomised to topical chloramphenicol ointment (n=488) or placebo (n=484). All participants were instructed to follow standard management, keeping their wound dry and covered for 24 hours after surgery.
The primary outcome was infection on the agreed day of removal of sutures or sooner if the patient re-presented with a perceived infection. The percentage of participants with an infection in the topical chloramphenicol group was significantly lower than in the placebo group (6.6% v 11.0%; difference −4.4%, 95% confidence interval −7.9% to −0.8%; P=0.010). Although the application of a single dose of topical chloramphenicol to high risk sutured wounds after minor surgery produced a statistically significant reduction in the infection rate, the researchers concluded that the reduction was not clinically significant because it was only a moderate reduction.
Which of the following statements, if any, are true?
a) The control treatment is referred to as a sham treatment
b) The control treatment is described as a positive control
c) The percentage of patients with an infection in the control group estimated the placebo response
d) The current version of the Declaration of Helsinki permits the use of placebos in randomised controlled trials
Statements c and d are true, whereas a and b are false.
The purpose of the trial was to investigate the effectiveness of topical chloramphenicol in preventing wound infection after minor dermatological surgery. A randomised controlled trial was performed. The control treatment was placebo ointment, which consisted of a mixture of soft white and liquid paraffin with no known anti-infective properties. The placebo was used in the trial above to facilitate double blinding and therefore minimise ascertainment bias.2
A placebo is similar in every respect to a treatment that has potential or proved effectiveness, except that it does not contain the active ingredient or component through which the treatment is assumed to exert its effectiveness. A placebo can be for a pharmacological product, such as an antidepressant drug; for a physical intervention, such as continuous positive airway pressure; or for a psychological intervention, such as psychotherapy. In general, the term “placebo” is used in pharmacological studies, whereas “sham treatment” is used for non-pharmacological studies, including those looking at devices and psychological and physical treatments. In the trial above, the placebo was for a pharmacological product and therefore would not be referred to as a sham treatment (a is false).
The above study was a superiority trial by design—the aim was to establish whether the intervention reduced the percentage of patients with an infection and was therefore superior in effectiveness to the control treatment, or whether the control treatment was superior. Superiority trials have been described in a previous question.3
If the control treatment in a trial has proved effectiveness—whether it is a drug, therapy, or medical device—it is referred to as an active or positive control. In the trial above, the control was a placebo with no known therapeutic effects and is therefore referred to as a negative control (b is false).
The percentage of patients in the placebo group who had an infection was 11.0%, and this estimated the placebo response (c is true).4 The placebo response consisted of non-specific treatment effects, including the placebo effect plus the natural course of wound infection. The placebo effect represents the participants’ response to investigation, including the response to a therapeutic ritual, subsequent response to observation and assessment, and response to the patient-doctor interaction. No doubt there is a complex association between these components. The natural course of wound infection represents what would happen if participants did not receive treatment—chloramphenicol ointment or placebo—or any investigation. It would not be straightforward to estimate separately the magnitude of the placebo effect or the natural course of wound infection.
The proportion of participants with an infection in the topical chloramphenicol group was 6.6%, and this estimated the treatment response. The treatment response is made up of two components—the treatment effect plus non-specific treatment effects not directly ascribed to the active treatment. The non-specific treatment effects not directly ascribed to the active treatment would be estimated by the placebo response, as described above. The treatment effect is the part of the treatment response that was caused directly by the chemical composition of the topical chloramphenicol ointment. In the above trial the treatment effect was estimated by the difference between the topical chloramphenicol and placebo groups in the proportion of participants with an infection—that is, the absolute reduction in the infection rate of 4.4% (95% confidence interval 0.8% to 7.9%).
The use of placebos and sham treatments in research is ethically controversial. This is especially true when there is good evidence that an effective intervention already exists. The World Medical Association (WMA) recognises this controversy and makes direct reference to the use of placebos in the most recent version of the Declaration of Helsinki. The Declaration of Helsinki (2013) states that placebos may be used “where no proven intervention exists” or if there are sound methodological and scientific reasons why the use of placebo is necessary to determine the efficacy or safety of an intervention and the patients who receive the placebo will not be subject to additional risks of serious or irreversible harms as a consequence of not receiving the best proved intervention. In effect, the WMA regards the use of placebos as ethically permissible, but only in specific circumstances (d is true). The precise meaning of “sound methodological and scientific reasons” is unclear. However, the Declaration of Helsinki includes a warning that extreme care must be taken to avoid abuse of this “option.” Researchers should, therefore, err on the side of caution when interpreting this clause.
In the above trial the control group was instructed to follow standard management in addition to receiving the placebo. The use of a placebo was probably ethically permissible from the perspective of the WMA because the control group was not denied the best proved intervention or exposed to serious or irreversible harms. If receiving the placebo put the control group at additional risk of infection then the acceptability of this research would have been more questionable. Presumably, the use of a soft white and liquid paraffin mixture as the placebo did not pose an additional risk.
Cite this as: BMJ 2015;351:h3755
Competing interests: None declared.