Informed consentBMJ 1998; 317 doi: https://doi.org/10.1136/bmj.317.7163.947a (Published 03 October 1998) Cite this as: BMJ 1998;317:947
Numbers inform the debate
- Michael Baum, Professor of surgery
- University College London Medical School, London W1P 7LD
- Department of Anaesthesia, Western Infirmary, Glasgow G11 6NT
- 2 Shirehall Gardens, London NW4 2QS
- Information and Statistics Division, Edinburgh EH5 3SQ
- Department of Radiotherapy and Oncology, Ninewells Hospital and Medical School, Dundee DD1 9SY
- Department of Surgery, Western Infirmary, Glasgow G11 6NT
- Department of Histopathology, George Eliot Hospital NHS Trust, Nuneaton, Warwickshire CV10 7DJ
- Department of Genitourinary Medicine, Whittall St Clinic, Birmingham B4 6DH
- Great Ormond Street Children's Hospital, London WC1 3JH
EDITOR—Smith's editorial recognises the complexity of the issue of informed consent and states that the BMJ is prepared to relax its absolutism.1 At the risk of being misunderstood I would like to attempt to construct a decision theory model based on certain explicit assumptions that may allow us to compute numerical values better to inform the debate. At the outset I accept the ethical principle of non-exploitation so beautifully described by Mary Warnock2I also accept the importance of consumers' involvement (I was the founding father of the consumers' advisory group for clinical trials chaired b Mrs Hazel Thornton).
Let us anticipate 150 000 deaths from breast cancer in this country over the next 10 years and let us make the conservative assumption that we already possess a novel therapeutic adjuvant that in absolute terms would reduce the risk of death by 6% over this period—in other words, save 9000 lives.
Next let us assume that the UK Coordinating Committee for Cancer Research has approved three different clinical trials evaluating three promising new agents, any one of which might produce this desired 6% absolute reduction in mortality, which is equivalent to a relative risk reduction of about 25% for patients with an average prognosis. Each trial would need to recruit about 2000 patients to have adequate statistical power to detect this order of relative risk reduction. Let us now suggest that trial A will show no difference between best standard treatment and the new treatment. Trial B will show that the new treatment will produce the desired benefit, and trial C will show that the new treatment is worse by the same order of magnitude. In aggregate 6000 women will have been recruited to these three trials. Altogether 120 will be better off than if they had had best standard treatment, …