Soil and climactic predictors of canine coccidioidomycosis seroprevalence in Washington State: An ecological cross-sectional study.

Coccidioidomycosis is a predominantly respiratory infection of animals and humans caused by soil-dwelling fungi. Long known to be endemic to North American deserts, locally acquired human cases first emerged in Washington State in 2010. To inform development of an environmental niche map, we conducted an ecological cross-sectional study of the association between soil and climactic variables and canine seroprevalence in Washington State, at the zip code tabulation area level. Soil predictors included soil texture (per cent sand, silt and clay), pH, electrical conductivity and water storage capacity; climactic predictors included mean annual air temperature and liquid precipitation. Clustering was evaluated with Moran's I, and four modelling approaches were adopted a priori: two models without any smoothing (logistic regression and quasi-logistic regression) and two smoothing models (non-spatial and spatial smoothing). No evidence was found for clustering, and both smoothing models resulted in marked attenuation of all coefficients. Temperature was found to have a positive effect in the non-smoothing models (prevalence odds ratio, logistic model: 1.70, 95% confidence interval 1.02, 2.84). While no other significant associations were found, there was suggestive evidence of a positive effect for pH. Despite the limitations inherent to the ecological and cross-sectional nature of these data, these findings provide insight for the development of an environmental niche map in Washington State and demonstrate the utility of using data from an animal sentinel to predict human disease risk.