A GIS-based matched case-control study of road characteristics in farm vehicle crashes

Background: Farm vehicle-related crashes (crashes) are hazardous for farm and non-farm vehicle users; however, most studies examine risk factors of injury given a crash, and shed little light on risk factors of crashes. We evaluated the association of road sinuosity and gradient with crashes in nine Midwestern States from 2005 to 2010. Methods: We collected crash data from the state departments of transportation, and road segment data from the Environmental Sciences Research Institute. We measured gradient and sinuosity of road segments using ArcGIS. A road segment with a crash was defined as a case (n = 6,848), and that without a crash was defined as a control. Controls were matched to cases by ZIP code, road type, and length in 1:1 (controls = 6,808) matching scheme. In addition, a 1:many control matched scheme was employed such that all road segments adjacent to the case would serve as controls (n = 24,390). We computed odds ratios (OR) and 95% confidence intervals (CIs) using multivariable conditional logistic regression. Results: The adjusted OR of a crash on a road segment with 6%-10% gradient was 0.60 (95% CI: 0.49, 0.75) as compared with a leveled (<1% gradient) road segment. Compared with a straight (<1% sinuosity) road segment, the adjusted OR of a crash on a road segment with 6%-10% sinuosity was 0.38 (95% CI: 0.29, 0.52). Conclusions: Roads with increased gradient and sinuosity had fewer farm crashes. These associations may be due to cautious driving behaviors on curvy or steep roads and road side signage alerting drivers of impending curve or grade.