Citation

Dean JM, Reading JC, Nechodom PJ. Transp. Res. Rec. 1995; 1485: 186-191.

Copyright

(Copyright © 1995, Transportation Research Board, National Research Council, National Academy of Sciences USA, Publisher SAGE Publishing)

DOI

unavailable

PMID

unavailable

Abstract

Motor vehicle crashes are a leading cause of injury and mortality in the United States. The effectiveness of seat belt use has been difficult to study because of the unavailability of population-based crash data bases that include all noninjured occupants. The 1991 Utah Crash Outcome Data Evaluation System data base was developed to determine seat belt effectiveness. It includes occupants of all police-reported motor vehicle crashes. Seat belt effectiveness may be overestimated, however, because of self-reporting of seat belt use when crash occupants are questioned by police. The effect of misclassification of seat belt use on the calculated odds ratio associated with seat belt use was studied by using logistic regression models of four levels of injury. The odds ratio associated with seat belt use for any degree of injury was 0.448 [95% confidence interval (CI) 0.425-0.473]; the odds ratios associated with seat belt use for injuries requiring outpatient emergency care, hospitalization, or fatalities were 0.476 (95% CI 0.449-0.504), 0.203 (95% CI 0.170-0.241), and 0.148 (95% CI 0.097-0.226), respectively. Adjustment of the fraction of correct classification of seat belt use among reported belt users decreased the protective effect associated with seat belt use for all four levels of injury. This is consistent with overestimation of seat belt effectiveness associated with nondifferential misclassification. Based on the assumption that the 1991 observational use rate applies to the 1991 crash population, odds ratios were corrected for seat belt self reporting bias. The corrected odds ratio associated with protection from any degree of injury was 0.723 (95% CI 0.685-0.763); the corrected odds ratios associated with seat belt use and injuries requiring outpatient emergency care, hospitalization, or fatalities were 0.747 (95% CI 0.705-0.791), 0.505 (95% CI 0.421-0.606), and 0.455 (95% CI 0.296-0.697), respectively. The study results confirm the protective effect of seat belts in motor vehicle crashes and provide a methodology for correcting seat belt effectiveness estimates.

Language: en

Keywords

Accident prevention; Highway accidents; Mathematical models; Regression analysis; Crashworthiness; Motor transportation; Database systems; Automobile seat belts