Continued Need for Placebo in Many Cases, Even When There Is Effective Therapy

In a recent commentary,¹ Miller and Shorr discuss generally the use of placebo controls when there is an existing effective treatment and then consider specifically 3 recent placebo-controlled studies in asthma, all of which they believe did not need to use a placebo. In at least 1 of the 3 studies, the issue is more complex than they allow.

The authors note the recent increased interest in the ethics of the use of placebo controls provoked by the 2000 revision of the Declaration of Helsinki,² which said such controls were not appropriate when there was existing effective therapy. Miller and Shorr clearly recognize, however, the interpretive problems often posed by active control equivalence or noninferiority trials^3- 5 and thus the continued need for placebo controls in many situations, even when there is existing therapy. Indeed, a commentary on the 2000 revision by the World Medical Association in 2001⁶ indicates similar recognition, so long as use of the placebo instead of available therapy does not expose patients to serious harm. Miller and Shorr divide placebo considerations into the following 3 categories: (1) when failure to use available therapy exposes patients to risk of death, irreversible harm, or intolerable discomfort (placebo is clearly unacceptable in such cases); (2) when the condition causes only mild to moderate discomfort (headache, allergic rhinitis, and/or heartburn) and patients frequently forgo treatment (there is no ethical impediment to use of placebo in such cases); (3) when the illnesses are more serious and discomfort associated with nontreatment is more severe, but methodological considerations call for use of a placebo and the study is designed to minimize discomfort. It is into this last category that the authors place asthma, along with angina, migraine headaches, and depression. Although studies of those conditions require attention to patient selection and study design (in case of worsening), that should not lead to the conclusion that placebo controls are not needed in those cases.