Environmental and demographic risk factors for campylobacteriosis: do various geographical scales tell the same story?

Abstract Background Campylobacter is a common cause of bacterial gastro-enteritis characterized by multiple environmental sources and transmission pathways. Ecological studies can be used to reveal important regional characteristics linked to campylobacteriosis risk, but their results can be influenced by the choice of geographical units of analysis. This study was undertaken to compare the associations between the incidence of campylobacteriosis in Quebec, Canada and various environmental characteristics using seven different sets of geographical units. Methods For each set of geographical unit, a conditional autoregressive model was used to model the incidence of reported cases of campylobacteriosis according to environmental (poultry density, ruminant density, slaughterhouse presence, temperature, and precipitation) and demographic (population density, level of education) characteristics. Models were compared in terms of number of significant predictors, differences in direction and magnitude of predictors, and fit of the models. Results In general, the number of significant predictors was reduced as the aggregation level increased. More aggregated scales tend to show larger but less precise estimates for all variables, with the exception of slaughterhouse presence. Regional characteristics associated with an increased regional risk of campylobacteriosis, for at least some geographical units, were high ruminant density, high poultry density, high population density, and presence of a large poultry slaughterhouse, whereas a reduction in risk was associated with a lower percentage of people with diplomas, a lower level of precipitation, and warmer temperature. Two clusters of elevated residual risk were observed, with different location and size depending on the geographical unit used. Conclusions Overall, our results suggest that the use of municipality or census consolidated subdivision were the most optimal scales for studying environmental determinants of campylobacteriosis at a regional level. This study highlights the need for careful selection and analysis of geographical units when using ecological study designs.