Authors' replyBMJ 1996; 312 doi: https://doi.org/10.1136/bmj.312.7029.509c (Published 24 February 1996) Cite this as: BMJ 1996;312:509
- Nicholas J Wald,
- Philip Murphy,
- Philippa Major,
- Carol Parkes,
- Chris Frost,
- Joy Townsend
- Professor Computer programmer Research assistant Lecturer Statistician Cancer Research Campaign Cancer Screening Research Group, Wolfson Institute of Preventive Medicine, St Bartholomew's Hospital Medical College, London EC1M 6BQ
- Senior scientist Medical Research Council Epidemiology and Medical Care Unit, Wolfson Institute of Preventive Medicine
EDITOR,—As A H S Lee and colleagues imply, cancers detected only by two view mammography would be expected to be smaller, but our data indicate that any difference is small. Tumour size was similar in women who had either one or two views. In the group which had two views (one interpreted by one reader and both by another) the median size of the tumours detected by one view only was 13 mm compared with 12 mm if two views were used.
The effectiveness of breast cancer screening has been well demonstrated in randomised trials, and our trial shows the advantage of two view mammography over one view. To perform a trial of two view mammography with mortality from breast cancer as the end point would be impractical and unnecessary. As randomised trials have shown that mammographic screening reduces mortality, the prevalence of cancer detected by screening is a sufficient end point in trials comparing screening methods. If two view mammography detects 24% more preclinical cancers than one view, given the evidence on tumour size, a similar proportionate effect on mortality would be expected. The absence of a simple relation between the prevalence of detected cancers and the proportionate reduction in mortality from breast cancer across different trials of screening confirms that the rates of breast cancer and the effect of treatment vary in different populations and at different ages. It does not mean that the effectiveness of screening in detecting cancers is unrelated to its effect in reducing mortality.
It is reasonable to seek consent to participate in research from individuals invited to have a treatment or procedure that departs from recommended practice. Because one view mammography was recommended practice, consent was obtained only from women receiving two views. This issue is unrelated to that of providing appropriate information to women attending for screening, which should be (and was) done routinely.
Sneh Bhargava and colleagues have misunderstood our economic calculations on the marginal cost of two view mammography. This was done in the standard way (the difference in total costs between one and two view mammography divided by the number of extra cancers detected) with the total costs based on the average costs of screening women by each method. The costs related to detecting a cancer at the time of screening, so discounting was not appropriate. The paper does, however, provide the necessary data for relating these costs to future benefits using any selected discount rate.