The continuing unethical use of placebo controls
KJ Rothman, KB Michels
N Engl J Med. 1994;331:394-398Is There a Scientific Rationale for Placebos?
The FDA is not alone in pushing for placebo controls. For example, a recent textbook on clinical drug trials advocates using them because "if a new drug has only been compared to an active control (without a placebo-controlled trial), this is not a convincing proof of efficacy (even if equivalence can be demonstrated)."37 Without justification, such statements confer on placebo control a stature that ranks it with double blinding and randomization as a hallmark of good science.
The randomized, controlled trial is well recognized as the most desirable type of study in which to evaluate a new treatment. This recognition acknowledges the essential role of comparison and the importance of randomization in enhancing the comparability of two or more treatment groups. Using a placebo for comparison controls for the psychological effects of receiving some treatment and also permits blinding. No scientific principle, however, requires the comparison in a trial to involve a placebo instead of, or in addition to, an active treatment.
Why, then, are placebo controls considered important? Three arguments have been advanced, none of which withstands scrutiny.Establishing a Reference Point
By allowing the investigator to determine whether a new treatment is better than nothing (beyond the psychological benefits of treatment), a placebo control offers a clear benchmark. After all, even if a new treatment is worse than an existing one, it may still be "effective" in that it is better than no treatment. On the other hand, as Hill pointed out in 1963, the essential medical question at issue is how the new treatment compares with the old one, not whether the new treatment is better than nothing1.Avoiding Difficult Decisions about Comparison Treatments
Determining whether one treatment is better than another is not always a straightforward matter. Beyond the question of efficacy, one can and should take into account unintended effects, interactions, costs, routes of administration, and other factors. Thus, it may appear simplistic to demand that the best proven treatment be chosen as the standard for comparison, if "best proven" refers only to efficacy. For some patients there may be advantages to a treatment that is inferior to a current standard with regard to efficacy but better with respect to cost or quality of life. For example, the adverse effects of some accepted treatments might offset the therapeutic benefits for some patients sufficiently that a placebo control would be ethically justified. This reasoning involves a complex decision that should be defended in submitted research proposals and published reports. It is not justifiable, however, to assign placebo controls simply to avoid the complex decision of which treatment should be used as a standard. Investigators are ethically obliged to make such decisions.Bolstering Statistical Significance
One FDA scientist contends that placebo-controlled trials are superior to studies using an active treatment as the control because it is much easier to demonstrate a statistically significant effect in the former case36. The FDA relies heavily on statistical significance in judging the efficacy of new drugs36. Despite its popularity, however, this tool is not a good one for measuring efficacy38,39,40,41,42. The significance of an association depends on two characteristics -- the strength of the association and its statistical variability. A weak effect can be "significant" if there is little statistical variability in its measurement, whereas a strong effect may not be "significant" if there is substantial variability in its measurement. Of the two characteristics, only the strength of the effect should be fundamental to the decision about approval of the drug. Ideally, statistical variability should be reduced nearly to zero when the magnitude of a drug effect is assessed, so that random error does not influence the assessment.
Unfortunately, the main way to reduce statistical variability is to conduct large studies, which are expensive. Statistical significance, on the other hand, can be obtained even in small studies, if the effect estimate is strong enough. When a placebo control is used instead of an effective treatment, the effect of a new drug appears large and may be statistically significant even in a small study. The scientific benefit, however, is illusory. Because the study is small, the measurement of the effect is subject to considerable statistical error. Thus, the actual size of the effect, even when a new drug is compared with placebo, remains obscure, and the study does not address the question of the effectiveness of the new treatment as compared with currently accepted treatments.The small placebo-controlled studies fostered by the FDA benefit drug companies, which can more easily obtain approval of an inferior drug by comparing it with placebo than they can by testing it against a serious competitor.
Smaller studies are also cheaper. Unfortunately, the costs saved by the drug company are borne by patients, who receive placebos instead of effective treatments, and by the public at large, which is supplied with a drug of undetermined efficacy.There is no sound scientific basis for these arguments on behalf of placebo controls.
Furthermore, regardless of any apparent merit these arguments have, scientific considerations should not take precedence over ethical ones, even if the use of active controls requires more difficult decisions about study design, more costly studies, and more complicated analyses.