Use of single vehicle collisions to model fatigue-related crashes on rural two-lane highways

Abstract

Fatigue-related crashes are believed to be more common on rural highways than on urban roads and on two-lane roads rather than on other rural road types. Thus an understanding of how design factors affect fatigue-related crashes on rural to-lane roads is vital. The problem is that fatigue is rarely reported as a cause of crashes, since is is rarely suspected by the police as a possible cause and since potential liability may motive the drivers not to reveal the real causes of the crash. Thus, getting a handle on these crashes thorough modeling is a formidable challenge. Fortunately, there is research to suggest that single-vehicle run-off-road crashes, particularly those during periods of low circadian rhythm, can be used as a reasonable surrogate in modeling fatigue--related crashes. The paper is based on research to examine how fatigue-related crashes rural on two-lane roads, as represented by single vehicle crashes, are affected by various engineering design factors. This study's goal is to explore the effects of fatigue on driving on rural two-lane roads in North America, and to consider how we can work towards mitigating the effects of fatigue on traffic safety. For this investigation, generalized linear and logistic regression modelling were used on US Highway Safety Information System (HSIS) data from Ohio. Models were developed separately and combined for periods of high and low circadian rhythm and for single-vehicle run-off-road and other crashes. The results show, for example, the after controlling for traffic volumes, increases in speed limit, average curvature and average gradient and decreases in surface width and average shoulder width were found to be associated with increased fatigue related crashes. Important differences were found in the effects of factors for period of low and high circadian rhythm.