Estimating Relative Driver Fatality and Injury Risk According to Some Characteristics of Cars and Drivers Using Matched-pair Multivariate Analysis

Characteristics of cars, such as mass and age, play a significant role in crash severity, but their effect are difficult to quantify because of the great number of factors altering the outcome of a real world accident. In order to focus on the crashworthiness of cars, we examine risk factors of severity suffered by the drivers involved in two-car accidents recorded by the police between 1996 and 2000 in France. From them, we build three matched casecontrol studies where both drivers have different severity levels: killed or injured vs. uninjured; severely injured vs. slightly injured; and killed vs. injured. Odds-Ratios are estimated by conditional logistic regression. The risk of being injured decreases with the weight of the car, coming to six times lower when driving a 1200 kg or more car compared to a 800 kg or less car. The risk of being killed rather than injured increases with the age of the car, reaching the highest value of height when comparing 1990 or before cars to the most recent ones. As expected, highest risks of death or injury are shown for sideimpacted cars, seat belt wearing is confirmed as being very protective and drivers of vehicles with frontal airbag are less often injured. The risk of being injured or killed increases with the age of the driver, and is higher for women. These characteristics of drivers are associated with both their way of driving and their capacity to withstand an impact, and it was then necessary to adjust our estimates on them. Our results show that recent cars provide a better protection, but confirm that the compatibility of cars with each other according to their weight is a big issue. It also corroborates the necessity of adaptative safety devices for taking the characteristics of a car occupant into account.