Parish classification or dwelling coordinate for exposure assessment in environmental epidemiology — A comparative study using Geographical Information System

Background: Two previous epidemiological studies on the incidence of total malignancies in Sweden after the Chernobyl accident have shown consistently increased risks. The first study used an analogue map on 137Cs from 1986 to classify individuals in terms of the parish they lived in. In the second study, dwelling coordinateswerematched to a digitalmap fromthe year 2000 to assess the individual exposure. To establish the accuracy of the exposure assessment using the larger unit of parish, instead of coordinates, we decided to compare the methods. Methods: On the analogue map eleven isolines on the deposition of 137Cs (kBq/m2) were used to classify all individuals in each of the 450 parishes. Using the digital map, by contrast, each dwelling with its inhabitants could be matched to 137Cs deposition at a coordinate level. A population-weighted average of 137Cs deposition was calculated for each parish. In total, 1,126,960 individuals and 450 parishes were included and analysed into six different exposure categories. Results: Using the new parish exposure index, 111 out of the 450 parishes were reclassified as a result of the increased resolution of the digital map (86 parishes) or unequal distribution of the population compared with the deposition (25 parishes). Seventy-five per cent of the parishes remained in the same exposure category as on the analogue map. Conclusion: Using dwelling coordinates for exposure assessment may not always be superior to parish classification. Nor is it always a cost-effective way of estimating the exposure, especially if the exposure in a parish is relatively homogenous or if parishes can be merged into broader categories with little intra-parish difference.