9.3 SCHOEMAKER J. H. (1995). Benefits of placebos. Nature 377: 98

SIR - Recent correspondence shows that placebo-controlled trials have their drawbacks. Beatrice Golomb suggests that studies should be carried out to test for possible specific effects of placebo agents so as to eliminate an eventual confounding factor in the interpreting of double-blind study results. Justifiable as it seems from a scientific point of view, this does not appear to me a cost-effective approach. Recent developments show that the costs of new drugs that come to the market will not be automatically reimbursed by health assurance systems. In many countries, the market prices of drugs for one and the same indication are compared an drugs with the lowest daily costs are privileged. The need for extra studies to test for possible specific effects of placebos will increase drug development costs and thus market price. An alternative would be not to use placebos for the control of clinical studies, or to accept small effects of a placebo where its use cannot be circumvented. There may be good reasons to use no placebos for the control of clinical studies-. Usually. however, two or more placebo-controlled (pivotal) studies are required to convince the medical profession and health authorities of the therapeutic value of a drug for a readily identifiable category of patients.

As Golomb rightly states, an apparent positive, negative or null effect of a drug may be the consequence of a negative, positive or same-direction effect of a placebo with which it is compared. Because it is not realistic to expect a pharmaceutical company to invest a huge amount of money in the development of a drug that is not viable, false negative effects arc unlikely to occur and, if they occur, are taken for granted. Similarly, a lack of significant difference between active and placebo treatment is always a reason to abandon a project and not to invest more money just to confirm whether the placebo is as (in)effective as the (in) active drug. In the worst case, a drug is credited for effects that it does not produce because by chance there had been negative effects of the placebo(s) used in the pivotal studies. Even then, this may not be harmful. It is generally assumed that when there is a moderate difference between two treatments in their effects on some specific disease outcome, this difference might be larger or smaller in other patients (studies), but it is unlikely to be reversed. Besides, a compound with no specific effect (a placebo), but efficacious in the treatment of a disease when adequately administered and free of any toxicity, may be the best 'drug' that one can imagine.

Department of International Clinical Research.
Roche Nederland BV, 3642 RR Mijdrecht, The Netherlands

1. Golomb, B Nature 375. 530 (1995).

2. Rothman. K.J. & Michels, K.B. N. Engl. J Med 331161,394-398 (1994).

3. Peto R Collins R & Gray R J elm. Epidem 48111. 230 (1995)

NATURE - VOL 377 - 14 SEPTEBMER 1995