Cropland associated with risk of Parkinson's disease in the northern Great Plains

Introduction: We sought to identify regional associations between cropland density and crop types and PD in the U.S. Methods: We conducted a population-based study of 21,639,190 Medicare beneficiaries, 89,790 with incident PD in 2009. We used county-level geographic weighted regression (GWR) to identify region(s) of the U.S. where the association between PD RR and cropland density was strongest. In a broad region identified by GWR in which cropland density was associated with PD, we performed logistic regression using individual-level beneficiary data (2733 cases and 805,984 non-cases) with high-resolution cropland density data. We adjusted for age, sex, race, smoking, healthcare utilization, and PM_2.5 (particulate matter <2.5 μm). We then explored PD-cropland associations for each type of crop within a subregion, in which the association was the strongest. Results: GWR identified a 9-state region in the Great Plains in which county-level cropland density and PD RR were associated. Within this region, the strongest GWR coefficients centered around the Williston Basin. High-resolution analysis demonstrated an association between cropland density within a 5-mile radius of residential zip+4 and PD. When comparing the highest to lowest quartile of cropland density, the odds ratio (OR) for PD was 1.14 (95 % confidence interval [CI] 1.01–1.27) in the 9-state region and 1.99 (95 % CI 1.09–3.61) in the Williston Basin. In the Williston Basin, percentage of sunflowers, winter wheat, and alfalfa within 5 miles of a beneficiary's zip+4 was associated with PD. Conclusion: We identified a region-specific association between cropland and crop type and PD in the Williston Basin.