Statistical modelling of breast cancer incidence and mortality rates in Scotland.

The interpretation of time trends in disease rates can be facilitated using estimable contrasts from age-period-cohort models. Cohort and period trends in breast cancer incidence and mortality rates in Scotland were investigated using contrasts that measure the changes in the linear trends. These contrasts were compared with estimates obtained from mortality rates in the USA and Japan. A significant moderation of both breast cancer incidence and mortality rates was observed in Scotland, associated with cohorts of women born after the Second World War compared with women born between the two world wars. The moderation of breast cancer mortality among cohorts born after 1925 compared with cohorts born before 1925 that was observed in the USA and Japan was also observed in this study. This moderation is not present in the incidence rates. The relative decline in the risk of breast cancer seen in younger cohorts seems to be contradictory to the temporal pattern present among breast cancer risk factors. It may well be that the alteration of eating patterns as a result of rationing in the wartime and immediate post-war period, and the subsequent influence on certain breast cancer risk factors probably produced by such changes, may have had some influence on the development of healthier girls and women. Such speculation could be addressed in a well-designed epidemiological study. There have been no changes in the mortality rate trends with period in Scotland, although the changes in the incidence rate trends with period are consistent with an increase in registration coverage.