Socio-economic differences in diabetes mortality: a retrospective cohort study in the Brussels-Capital Region

Objective: The aim of this study is to identify socio-economic (SE) differences in premature diabetes mortality in men and women aged 25 to 74. To capture the actual burden of diabetes, both death certificates with diabetes as underlying (UCOD) and multiple (MCOD) cause of death (any mention of diabetes) are analysed.
Methods: The study population consists of all inhabitants of the Brussels-Capital Region aged 25-74 (N=598,578). Data are derived from record linkage between the Belgian 2001 census and registration records of all deaths and emigrations for the period 2001-2005. The total number of diabetes-related deaths is 720. Indicators of socio-economic status (SES) are educational attainment (immaterial dimension) and housing quality (material dimension). Age-adjusted mortality rates are calculated using direct standardisation. Rate ratios (RR) are obtained by Poisson regression models with diabetes mortality as the dependent variable.
Results: Overall, a lower SES is associated with higher diabetes mortality. According to the indicator of SES used and the extraction of diabetes as either UCOD or MCOD, the strength of the association differs. In men, an inverse educational gradient is observed. Men with primary or no formal education have a RR(MCOD) of 2.4 (95% CI 1.8-3.3) compared to those who attained higher education. In women, the lowest educated - primary or no formal education - have a 6 times higher RR of dying from diabetes compared to women with higher education (RR(MCOD)=6.0; 95% CI 3.2-11.3). However, the relation between education and diabetes mortality is not linear. Further analyses point in the direction of an exclusion mechanism. A linear gradient between housing quality and diabetes mortality is observed for both men and women. In women, the gradient is steeper than in men. Women with insufficient or basic housing quality have a RR(MCOD) of 7.9 (95 % CI 2.9-21.4) compared to those living in very comfortable circumstances.
Conclusion: The results of this study show considerable SE differences in diabetes mortality. Health interventions targeting both the gradient and the most disadvantaged groups are necessary to reduce premature diabetes mortality.