Predicting regional variations in mortality from motor vehicle crashes

Area of Science:

• Epidemiology
• Transportation Safety
• Mathematical Modeling

Background:

• An inverse relationship exists between population density and per-capita mortality from motor vehicle crashes.
• Existing models require refinement for accurate prediction of traffic-related fatalities.

Purpose of the Study:

• To derive the observed inverse relationship between population density and per-capita motor vehicle crash mortality using a predictive mathematical model.
• To validate a mathematical model for estimating regional per-capita mortality from vehicle crashes.

Main Methods:

• Proposed models where fatal crashes correlate with population and mean distance to hospitals, parameterized as Weibull survival models.
• Utilized U.S. Census county and state data, fitting linear regression equations on a logarithmic scale.
• Incorporated an indicator variable to differentiate regional models, specifically southern states.

Main Results:

• A model distinguishing southern states accounted for 74% of state-to-state variation in mortality (excluding Alaska).
• Southern states exhibited 1.37 times higher per-capita mortality compared to other states, even after controlling for mean inter-hospital distance.
• Mean inter-hospital distance proved to be a significant predictor of per-capita motor vehicle crash mortality.

Conclusions:

• Mean inter-hospital distance enables a reasonably accurate estimation of per-capita motor vehicle crash mortality.
• Per-capita vehicle crash mortality is elevated in southern states, independent of inter-hospital distance, suggesting un identified contributing factors.
• The developed mathematical model offers a valuable tool for predicting regional traffic-related fatalities.