Predictive and explanatory assessments of traffic-related pedestrian injuries

Abstract

Pedestrians are an integral part of the modern transportation system but are often considered as the most vulnerable to severe traffic-related injury when compared to other road users. Between 1975 and 2009, annual pedestrian fatality counts in the United States have been decreasing steadily. However, after 2009, these counts have been increasing. Given that both the national demographic profile of the United States population and the physical makeup of transportation infrastructure show signs of aging, it is essential to understand how these affect pedestrian safety moving forward into future decades. This two-part thesis examines pedestrian safety trends from both of these perspectives. In the first part of this thesis, pedestrian fatality trends between 1975-2015, stratified by pedestrian age and sex, were analyzed and forecasted to the year 2035. Pedestrian fatality and exposure data were extracted from the NHTSA FARS and NHTS databases, respectively. Results showed that exposure-adjusted pedestrian fatality trends were consistently higher than observed pedestrian fatality counts across all ages and sexes, suggesting that interventions to reduce pedestrian fatalities have had a positive effect. Our fatality projection models indicated that traffic-related pedestrian deaths among children may continue to decrease, while pedestrian fatalities among adults aged 55 and older may increase significantly, which suggests that this cohort is at elevated risk. The second part of this thesis aimed at identifying factors that are significant in severe pedestrian injuries. Pedestrian injury data from the NHTSA GES database between 2011-2015 were examined. Odds ratios (ORs) of factors at the pedestrian, driver, crash, vehicle, environment and roadway levels were calculated. Results indicate that crashes at midblock had lower odds of fatal or serious pedestrian injury (OR = 0.79, 95% CI = 0.74 – 0.84) when compared to crashes at intersections with three or four approaches. Undivided roads (OR = 0.25, 95% CI = 0.23 – 0.27) and roads with painted medians (OR = 0.37, 95% CI = 0.35 – 0.40) had protective effects against severe pedestrian injuries compared to roads with physical medians. Compared with locations with signalization, unsignalized locations with signage (OR = 1.57, 95% CI = 1.44 – 1.71) or without signage (OR = 1.36, 95% CI = 1.27 – 1.45) were associated with higher odds of severe pedestrian injuries. Other factors such as light conditions and road surface conditions were also found to be significant in affecting the odds of a severe pedestrian injury. The findings presented in the two parts of this thesis provide further insight into the relationship between traffic-related pedestrian injury, human factors and the built environment. Further quantitative research is recommended to expand our understanding of pedestrian injury causality.