Reprinted from BMJ 1996;313:1008-9
Placebo controlled trials are needed to provide data on effectiveness of active treatment
Kenneth J Rothman questions whether there is any point in using a placebo group if blind assessment can be achieved in a comparative trial of two active treatments.1 He minimises both the practical advantages of placebo controlled trials and their advantages for hypothesis testing. The limitations of statistical hypothesis testing mean that the probability of an inferential error is known if one is rejecting the null hypothesis but not if one is accepting it.2 The probability of a type II error in equivalence trials is unknown and can be high with small sample sizes. The conclusion that a difference exists is therefore made on a clearer basis in placebo controlled trials. Equivalence trials are rarely large enough to give the same degree of certainty, and the practical difficulties of obtaining larger sample sizes are more inhibiting than Rothman suggests.
The effect of active treatment is greater in equivalence trials than in placebo controlled trials. For example, in antidepressant trials, placebo controlled trials show a lower efficacy of active drugs than do studies that use no control or an active drug.3 The context of the trial is important, and the placebo component of the response to treatment seems to be greater in equivalence trials than in placebo controlled trials. Evaluating a new drug only in equivalence trials can lead to an inflated estimate of its effect size.
Blinding may not be as critical in equivalence trials as in placebo controlled trials. Unblinding does occur in clinical trials, leading to the potential for bias.4 Subjects who have never received any similar treatment are likely to have fewer cues to help them determine whether they are receiving the new or standard treatment in an equivalence trial, but side effects or other cues may help them determine whether they are receiving an active or a placebo drug. Such factors are likely to increase the chance of finding that two active treatments are equivalent, as the degree of unblinding correlates with measured efficacy in a clinical trial.5 Restricting the evaluation of a new drug to equivalence studies does not give the opportunity to estimate the bias introduced through unblinding.
Too much of the effectiveness of clinical practice is due to the placebo effect for placebo controlled trials to be abandoned. Placebos are required, if only to provide more data about the effectiveness of active treatment. If blind assessments could be achieved in clinical trials and scientific rigour did lead to the accumulation of medical knowledge then more confidence could be placed in equivalence trials.
1. Rothman KJ. Placebo mania. BMJ 1996;313:3-4 (6 July).
2. Rozeboom AW. The fallacy of the null-hypothesis significance test. Psychol Bull 1960;57:416-28.
3. Wechsler H, Grosser GH, Greenblatt M. Research evaluating antidepressant medications on hospitalised mental patients: a survey of published reports during a five-year period. J Nerv Ment Dis 1965;141:231-9.
4. Fisher S, Greenberg RP. How sound is the double-blind design for evaluating psychotropic drugs? J Nerv Ment Dis 1993;181:345-50.
5. Double DB. Unblinding in trials of the withdrawal of anticholinergic agents in patients maintained on neuroleptics. J Nerv Ment Dis 1995;183:599-602.