Environmental security involves the geo-referenced assessment of health risks. A straightforward estimation of relative risks can be based on aggregated health data, which are often routinely collected in administrative districts and, therefore, cheap and easily available. Disease maps based on raw data can be misleading, however. The value of the maps can be improved by applying a statistical model that, firstly, accounts for the spatial autocorrelation and, secondly, admits the inclusion of exposure data (e.g. from monitoring networks). A third important task is the inclusion of socioeconomic and further individual-specific confounders.

The paper presents a hierarchical model that facilitates the mapping of the relative risk and the adjustment for socio-demographic and individual-specific variables (ecological adjustment). While the basis of the model is the conditional autoregressive (CAR) approach, we extend this approach by including a model for the individuals at risk. Bayesian inference is used for the risk estimation.

As an example, the statistical model is applied to the analysis of the spatial epidemiology of Helicobacter pylori in Leipzig, Germany. So far, the epidemiology of this infection is not fully understood. The Helicobacter pylori [¹³C]urea breath test was offered to all school beginners in the urban area of Leipzig to determine the colonization prevalence and potential transmission pathways of the bacterium. A total of 2888 school-starters participated in the test and their parents completed detailed, self-administered questionnaires. Calculating the spatial distribution of H. pylori among school-starters in the urban area of Leipzig, we present maps of district-wise relative risks and demonstrate the usefulness of the ecological adjustment.