A risk factor study of coccidioidomycosis by controlling differential misclassifications of exposure and susceptibility using a landscape ecology approach

State-reported coccidioidomycosis cases in Arizona have dramatically increased since 1997, raising concerns about a possible epidemic, its cause, and associated risk factors, including spatio-temporal differences in susceptibility and exposure. This stratified, two-stage, cross-sectional study evaluates inherent, socio-economic, and environmental risk factors of coccidioidomycosis from information collected during an address-based telephone survey of 5460 households containing 14,105 individuals in greater Tucson, Arizona. Three geomorphic and two demographic strata controlled for differences in group-level exposures and susceptibility, and assured recruitment of a minority population. Logistic regression of self-reported cases indicates that location of residence by geomorphic and demographic strata was a risk factor that confounded the associations of coccidioidomycosis with age, race-ethnicity, and educational attainment. The risk due to age is more evenly distributed across the population than bivariate results when individual- and group-level exposure and susceptibility factors are controlled. Similarly the association for being Hispanic decreased from strong bivariate 0.28 odds ratio to a weak multivariate 0.75. Location of residence confounded the risk due to race-ethnicity and was an effect modifier of risk due to age. Differential misclassification of exposure to Coccidioides spores and susceptibility to coccidioidomycosis was reduced through landscape stratification by demographics and geomorphic types. Landscape epidemiological studies of diseases with strong environmental and demographic determinants can reduce residual confounding and account for spatial and temporal differences between neighborhoods and at broader scales.