Exploring spatially varying relationships between Lyme disease and land cover with geographically weighted regression

Understanding environmental variables responsible for the spatial distribution of Lyme disease is essential for determining disease risk and directing control and prevention efforts. This study provides a novel application of geographically weighted regression to explore how the relationship between Lyme disease and land cover varies across the Midwest and Northeast regions of the United States. Results revealed that specific land cover types, namely deciduous forest, evergreen forest and agricultural land, are significant explanatory variables for predicting the location of Lyme disease incidence. However, contrary to previous studies, we show how these relationships vary within each region. The results from this study are important for informing Lyme disease mitigation efforts that have typically treated Lyme disease and land cover relationships as spatially static across this region. As such, we recommend that Lyme disease mitigation efforts not associate a high risk of Lyme disease with specific land cover types without understanding the larger geographic context influencing the presence and spread of the disease.