To the Editor: Life expectancy in developed countries increased by an average of about 20 years during the 20th century. An editorial in the Journal by Rodger referred to mortality in populations having screening mammography.1 Data quoted indicated that there had been only slight changes in breast-cancer mortality in Australia up to 1996.

Data for 1999 are available in the report of the Australian Bureau of Statistics Causes of death, published in December 2000.2 The standardised all-causes death rate per 100 000 for all persons in 1989 was 758.9 and in 1999 was 584.2, a reduction of 23.0%. For women, the standardised death rate attributable to breast cancer in 1989 was 27.2 and in 1999 was 22.1, a reduction of 18.75%. Recent decreases in breast cancer mortality of similar magnitude have also been observed in the United Kingdom and the United States.3 However, screening mammography could only be responsible for a small portion of these changes, because population screening has been in place for little more than a decade and the benefits of earlier detection and treatment would take more than five years to become evident. The causes of these dramatic reductions in death rates are not yet understood.

Regarding the effect of population screening mammography on mortality rates, this is limited to breast-cancer-specific mortality and cannot be expected to translate into a reduction in overall mortality. In a recent overview of the situation in Sweden,4 breast-cancer-specific mortality in the screened group was 22% lower than in the non-screened group. However, the age-adjusted relative risk for total mortality was 1.00 (95% CI, 0.98–1.02). In other words, the mammographically screened population died less frequently from breast cancer, but nevertheless died at the same rate as the non-screened population (from other causes such as heart disease and other cancers). If we consider that, in the age group 40–79 years, breast cancer accounts for about 3% of total mortality, a reduction in breast cancer mortality of 25% would be 25% of 3%, or 0.75%. This change is so small that it would probably never be possible to show an effect of breast-cancer screening on overall population mortality.

It is therefore realistic to regard the benefits of screening mammography as limited to early detection and treatment (possibly with improved quality of survival) and a reduction in breast-cancer-specific mortality.

In reply: I agree with Gough and welcome the more recent data showing up to an 18.75% reduction in breast cancer mortality in Australia in the 10 years from 1989.

Obviously, this cannot be attributed solely to the now 10-year-old National Mammographic Screening Program, but it may result from a combination of the screening program, ad-hoc screening before the program, and the more rigorous use of adjuvant therapies based on the results of clinical trials. That breast screening is unlikely to have an impact on overall population mortality gives the lie to the conclusions of Olsen and Gøtzsche's overview,1 which are based only on overall mortality.

Nevertheless, Gough and I agree that screening mammography is likely to deliver other benefits through detection of earlier-stage disease and a reduction in deaths from breast cancer.