Using Placebo Controls
Lead Editor: Bernard Lo, M.D.
A placebo control is acceptable if no effective therapy exists for the condition being studied. However, if researchers intentionally withhold effective therapies from participants in a clinical trial, it would violate the ethical principle of "do no harm." Nonetheless, under some circumstances, a placebo arm may be ethically acceptable even though an effective treatment exists and is feasible to administer.
A placebo contol is acceptable in the following circumstances:
Participants in the placebo arm are very unlikely to develop serious, irreversible harms.
Placebo controls are acceptable in trials for mild hypertension, mild pain, insomnia, etc. Patients also commonly forego treatment in these situations.
Practice guidelines do not recommend active treatment.
Clinical trial of antiviral suppression of herpes simplex in pregnant women near term used a placebo control because practice guidelines at the time did not recommend such suppression, although guidelines stated that it should be "considered" and that "some specialists recommended" it.
The effective intervention is commonly not prescribed.
Clinical trials of acellular pertussis vaccine in Sweden and Italy used placebo controls, because whole-cell pertussis vaccine was not required in those countries and not widely used.
To determine whether placebo control was acceptable in trial of new platelet inhibitors after percutaneous coronary stent placement, investigators surveyed physicians regarding the percentage of cases in which platelet inhibitors were used.
Participants have decided to forego established treatment (for example because of side effects or inconvenience).
Researchers should ask participants questions to assure that they understand that standard treatment is available outside the trial.
In placebo-controlled trials of medications for schizophrenia, patients need to be carefully monitored for adverse effects. Protocol must have procedures to declare failure of investigational drug and start standard therapy.
Participants are refractory to established treatments.
This is a common design for trials of new cancer drugs.
Designs that do not use placebo will not yield valid results to the research question.
Many clinical trials of analgesics, antidepressants, and anti-emetics fail to show that the active drug has an effect. Effect size varies from study to study, due to such factors as variation in populations, spontaneous improvement, and a waxing and waning course.
In clinical trials in resource-poor countries, the pertinent research question may be whether a simplified treatment regimen for HIV infection, tuberculosis, or malaria is more effective than current treatment. A trial comparing such an intervention to the regimen used in developed countries would not be relevant to public health decisions in the host country.
Researchers should consider an add on placebo design.
In clinical trials of new agents to prevent osteoporotic fractures, all participants might receive current best therapy, plus either new agent or placebo. New drug should have different mechanism of action than base therapy. This design will require a larger sample size.
The investigator who plans to use a placebo control in a clinical trial needs to provide the Institutional Review Board enough information to justify its use. As in presenting a study for peer review, the investigator should anticipate objections and concerns the reviewers are likely to raise, particularly concerns that have already been raised in the literature. The investigator should present alternatives to the proposed methodology and the reasons for rejecting them.