Author(s): Aislin Mushquash, Sacha Dubois, Bruce Weaver, Michel Bédard
Slidedeck Presentation Only (no paper submitted):
Background/Aims: Elevated body mass index (BMI; an index of weight-for-height) is associated with the risk of several health conditions such as type 2 diabetes, high blood pressure, heart disease, and certain cancers (National Heart, Lung, and Blood Institute). Evidence also suggests that individuals with obesity have elevated odds, as high as 60%, of fatal crash injury (Jehle 2012). We further explored the association between BMI and crash injury severity after controlling for driver and vehicle factors known to independently contribute to injury severity (Bédard 2002).
Methods: Utilizing the Fatality Analysis Reporting System (FARS) we examined crashes between 1998-2009 for which the driver’s BMI data was reported. Our model included vehicle and driver factors previously shown to independently contribute to driver fatalities. These included: driver age, sex, seat-belt use, BMI, passenger vehicle type (car, light truck, or van), principle impact point, vehicle weight, and vehicle age. Both linear and quadratic terms were included for age and BMI. To control for environmental factors we examined drivers from two-car paired crashes only. We used General Estimating Equations to account for the correlated nature of paired crash-data while examining the independent contribution of these driver and vehicle related factors on injury severity as both a binary (fatal, non-fatal) and ordinal (fatal, incapacitating, non-incapacitating, no or possible injury) outcome. Finally, we conducted analyses while controlling for alcohol and drugs for those drivers who were also blood-tested for these substances.
Results: A total of 22,297 crashes (44,594 drivers) with complete data were extracted for the analysis. There were 1,235 (6%), 15,437 (69%), 5,625 (25%) crashes respectively where both, one, or neither driver was fatally injured. Mean BMI was 25.81 (SD=5.36) for those fatally injured and 25.44 (SD=5.06) for those not. Before including BMI in the binary logistic regression model, female sex, older driver age (both linear and quadratic terms), not using a seat-belt, driver-side impact, lighter vehicle weight, driving an older vehicle, and driving a car (versus van or light truck) were harmful (i.e. had increased odds of the driver being fatally injured). After including BMI in the model these associations remained. The odds ratio (95% CI) of a fatal injury for a normal BMI (centered at 21.75) for an 8-unit change was 0.95 (0.88, 1.01) after controlling for the quadratic BMI term (OR: 0.08; 95%CI: 1.04, 1.11) and independent contributors. For individuals with elevated BMIs, the odd ratios of a fatal injury were 1.21 (1.14, 1.29) and 1.33 (1.20, 1.46) respectively for individuals with Class II and Class III obesity. Results from the ordinal logistic regression model were comparable. Approximately one in four crashes included drivers blood-tested for both alcohol and drugs. After controlling for these substances the harmful effect of elevated BMI levels was reduced to about 4-6%.
Discussion/Conclusions: After controlling for several known contributors to injury severity, and using a paired crash design, individuals with Class II or Class III obesity had increased odds of being fatally injured. Potentially, these increased odds may be related to inadequate vehicle designs not appropriately engineered for individuals with higher BMIs.