Abstract
BACKGROUND: Many studies observe a negative association between education and all-cause dementia, however, little is known about how the association develops. Current evidence regarding mediatory factors is limited, conflicted and suggests a complex relationship. In this study we used UK Biobank data to determine to what extent multiple factors mediate the association. METHODS: Data was sourced from UK Biobank and multiple imputation used to replace missing values. Education was measured at baseline assessment and dementia diagnosis determined from self-report or linked healthcare records. Five potential mediators were considered: Social isolation, income and occupational complexity were measured at baseline and health behaviour and health outcome scores derived. Logistic regression was used to examine the association between education and dementia with adjustment for potential mediators. Causal mediation analysis was then performed to determine the proportion of the dementia education association occurring via each mediatory pathway. RESULTS: In a sample of 384,284 participants, higher level of education was associated with reduced odds of dementia. When considered as a confounder, occupational complexity almost fully attenuated the association (OR: 0.94, CI: 0.89-0.99) and was found to mediate 73% of the association. Average income, health outcomes and health behaviours also acted as mediators, explaining 10%, 27% and 35% of the relationship respectively. CONCLUSIONS: Occupational complexity mediates a large proportion of the association between education and dementia. Intervention to improve access to cognitively stimulating work and leisure activities, particularly to those who are less educated, may reduce the number of people in the UK living with dementia.