Exploring educational disparities in breast cancer dynamics: a comprehensive analysis of incidence, death within 5 years of diagnosis, and mortality in the Belgian context.

Background: Breast cancer is the most prevalent cancer worldwide. Belgium shows high age-standardized incidence rates, but also high survival rates. Like many health outcomes, breast cancer has been associated with multiple factors of socioeconomic status. This paper aims to (a) map educational differences in breast cancer incidence, mortality and death rates within 5 years of diagnosis, (b) update earlier trends in breast cancer mortality rates in Belgium for the 2004-2013 period and (c) investigate the role of fertility indicators as mediating factors in the association between education and breast cancer outcomes.

Methods: Data consisted of a linkage between the 2001 Belgian Census, register data on mortality and cancer incidence data (2004-2013) from the Belgian Cancer Registry. We calculated age standardized rates, rate ratios (Poisson regression) and hazard ratios (Cox regression) and furthermore also applied the method of Excess Portion Eliminated (EPE) in a mediation analysis of the fertility indicators. We stratified our analysis by age: younger than 50 (premenopausal) and 50 or older (postmenopausal).

Results: We observed striking differences in breast cancer incidence, all-cause and cause-specific death rates 5-years after diagnosis by educational level. Higher educated women had higher breast cancer incidence, but also lower all-cause and lower cause-specific death rates; adding up to zero differences in breast cancer mortality in the postmenopausal group and lower breast cancer mortality in the premenopausal group.

Conclusion: A notable shift in the social gradient occurred since the 1990's, favouring higher-educated women in recent years. Especially, with regards to survival after diagnosis there is potential for policy intervention. Stage at diagnosis played a crucial role, but differences between socioeconomic groups remained significant after including this parameter. While fertility indicators played a role, the impact was less pronounced than expected.