Age-environment model for breast cancer

In the field of breast cancer study, it has become accepted that crucial exposures to environmental risks might have occurred years before a malignant tumor is evident in human breasts, while age factors such as ages at menstruation have been known as risks for the disease already. To project trends in two such kinds of risks for the disease, the concept of environment effects is introduced for (age, period)-specific breast cancer mortality rates. Also, a new model, named the age-environment (AE) model, which assumes that the logarithm of the expected rate is a linear function of environment effects and age effects, is proposed. It is shown that, although environment effects have different meanings from period effects or cohort effects, in the age-period-cohort (APC) model, the range space of the design matrix for the AE model is included in that for APC model. It is seen, however, that the AE model provides a better fit to the data for females in Japan and the four Nordic countries than does the APC model in terms of AIC. From the results of ML estimation of the parameters in the AE model based on the data obtained in Japan, we see high levels of environment effects associated with the Sino-Japanese war, World War II and the environmental pollution due to the economy in the recovery period from the defeat. Besides, from those based on the data obtained in the four Nordic countries, we see high levels of environment effects associated with the environment becoming worse after the year of Helsinki Olympics and low levels of them associated with the period including the year of 'Miracle of the Winter War' in Finland. Copyright (C) 2004 John Wiley Sons, Ltd.