Reprinted from British Journal of Psychiatry 1991, 158, 573-4.
I think the dispute between Newcombe (Journal, December 1990, 157, 934-935) and myself (Journal, August 1990, 157, 300) about the value of double-blind trials arises partly because of his idealistic view of the randomised controlled trial. The logic of the method is essentially unassailable, but I am more concerned with the real world of clinical practice. The problem of unblinding will not be solved merely by pleas for improved study design and execution.
Dr Newcombe's useful critique of the study by Karlowski et al (1975) fails to reach a practical conclusion about whether ascorbic acid is effective in the treatment of a common cold. Should we, therefore, take the advice of the Nobel Prize winner, Linus Pauling (1970), with his regime of vitamin C to combat the common cold, based on his assessment of the evidence as a scientist?
Have clinical trials produced firm conclusions about the effectiveness of any treatment in psychiatry? If so, which are the methodologically sound studies? If not, considering the number of studies that have been conducted, would it not suggest there is an inherent difficulty in the design of the 'double-blind' method itself?
As an example, consider the evidence for the use of tricyclic antidepressants in depression in general practice. Hollyman et al (1988) found Amitriptyline to be effective. By contrast, Porter (1970) found no difference between Imipramine and placebo. Interestingly, Porter did not pretend his trial was double-blind, because he recognised that no trial of this kind can be conducted under completely blind conditions. In fact, he openly declared his bias that tricyclic antidepressants probably had no specific action in depression illness, although they may suppress anxiety and agitation by their sedative effect. He argued that his attitude towards the effectiveness of the drug might neutralise the influence of the breaking of the blind. The bias of Hollyman et al (1988) is less clear. Has their use of double-blind methods eliminated potential expectancy effects? It is a legitimate question. I am not suggesting it is easy to answer, but some evaluation may be possible with evidence from participants' guesses about medication status. An insistence on statistical purity in the analysis might produce a lack of awareness of the fallacy of the method.
The problem is that the results of 'double-blind' studies tend to be automatically accepted as scientifically valid. A misleading self-deception is encouraged that trials can be conducted double-blind, and the role of expectancies is underestimated. I understand the wish for a scientific basis for psychiatric treatment, but professional status should not mean that the challenge to double-blind methodology goes unnoticed (Oxtoby et al, Journal 1989, 155, 700-701).
HOLLYMAN, J.A., FREELING, P., PAYKEL, E.S., et al (1988) Double-blind placebo-controlled trial of Amitriptyline among depressed patients in general practice. Journal of the Royal College of General Practitioners, 38, 393-7.
KARLOWSKI, T.R., CHALMERS, T.C., FRENKEL, L.D., et al (1975) Ascorbic acid for the common cold. Journal of the American Medical Association, 231, 1038-42.
PAULING, L. (1970) Vitamin C and the Common Cold. San Francisco: Freeman.
PORTER, A.M.W. (1970) Depressive illness in a general practice. A demographic study and a controlled trial of Imipramine.British Medical Journal, i, 773-8.