ABSTRACT
Geographical health studies have been carried out previously using administrative areas as study units. Instead of administrative units, a 10 km by 10 km grid has now been used. A spatial hierarchical Conditional Autoregressive (CAR) model was fitted within a Bayesian framework. This chapter describes the Bayesian model formulation for the small area analyses. The association between Ca and Mg in groundwater and the incidence of acute coronary syndrome, a subclass of coronary heart disease, was estimated. In total 93 205 acronym ACS cases (67 755 men and 25 450 women) aged 35-74 years living in the rural Finland (excluding Lapland, Åland and the Turku Archipelago), were identified from the Finnish Cardiovascular Disease Register during 1991-2003. The data on the corresponding population-at-risk were obtained from the Finnish Population Register Centre. Ca and Mg concentrations were obtained from the groundwater database of the Geological Survey of Finland. The geographic pattern of acute coronary syndrome incidence was quite similar in men and women diagnosed in 1991-2003, suggesting common risk factors for both sexes and the data on men and women were, therefore, pooled. A 1 mg/l increment in Mg concentration was associated on average with a 2% (95% highest density regions: 0.3% to 3.9%) decrease in ACS incidence. Ca concentration in groundwater did not have any marked association with the incidence of ACS. Low natural concentrations of Mg in local groundwater were associated with an increased incidence of acute coronary syndrome in rural Finland.
