Mapping social health and dementia risk: A register-based study of older adults in Finland

This study investigates the role of individual and area-level social health factors in shaping geographic variation in dementia incidence among older adults in Finland.\r\nUsing nationwide register data on all individuals born between 1935 and 1939 and residing in Finland in 2015 (N = 185,712), we estimate cumulative dementia incidence over a four-year follow-up period (2016-2019). To reduce compositional bias in geographical comparison, we applied Matching on poset-based Average Rank for Multiple Treatments (MARMoT), a non-parametric matching approach that balances observed individual level characteristics across municipalities. Spatial scan statistics were then used to identify geographic clusters of excess dementia incidence considering municipality-level contextual measures after adjustment for individual-level characteristics.\r\nBefore MARMoT adjustment, several contiguous clusters of elevated dementia incidence were identified, particularly in eastern and southern Finland, with the highest risk cluster exhibiting a 36% higher incidence than the rest of the country. After balancing individual-level characteristics, some clusters attenuated, whereas others persisted or newly emerged, suggesting a confounding role of individual characteristics in the relationship between dementia incidence and place of residence. Excess incidence remained in parts of eastern Finland (21% - 51% excess risk) and emerged in west-central municipalities (27% excess risk). The inclusion of municipality-level indicators did not substantially alter these patterns.\r\nThese findings underscore the importance of accounting for social health and socio-demographic composition in spatial analysis of dementia and demonstrate the value of integrating matching-based and spatial methods to distinguish compositional from contextual disparities in ageing societies.