1. Obesity and Risk of Death Among Drivers in Motor Vehicle Crashes Shankuan Zhu njury Research Center, and Injury Research Center, and Department of Family and Community Medicine Medical College of Wisconsin Milwaukee, WI

2. Special Report of the United Nations Secretary General to the 58 th session of the General Assembly. (Epi Monitor. 2003;October:3&9) Road traffic injuries have became a global public health crisis and call for urgent action at the national and international levels  More than 1.2 million people are killed annually or more than 3,200 persons every day.  Road crashes rank 9 th among the leading causes of disease burden worldwide and will climb to the 3 rd by 2020. The Epidemiology of Road Injury Deaths variations by region, age, gender, and road user type Causes Causes Speeding and alcohol.

3. A New Traffic Safety Vision for the United States (L Evans, Am J Pub Health, 2003;93:1984-6) In the US, over 42, 000 people died on road in 2002  Prior to 1970s, US was number one in the world in traffic safety.  Compared to other developed countries, traffic fatalities from 1979 to 2000 only declined by 18% in the US, while 50% in Canada and Finland*, 46-48% in British and Australia. These countries view traffic deaths and injuries as much more of a public health problem than does the US. US mistakenly believes that the main way to reduce these deaths is to make every crash marginally more survivable while largely accept crashes as inevitable. The 2 Most Important Factors The individual’s behavior The behavior of other road users

4. Does obesity have any relationship with motor vehicle crashes ?

5. 1) Gender difference; 2) Drivers and passengers; 3) Covariates selection; 4) Interaction of BMI with covariates; Studies have suggested that obesity could cause an increased risk of death in motor vehicle crashes

6. The Aim of the Study To quantify the magnitude of the effects of BMI on motor vehicle fatality among drivers To assess the interactions of BMI with some important factors that may influence the relationship of BMI with motor vehicle crash fatality such as gender, age, seatbelt use, airbag deployment, type of collision and changes in velocity during the crash.

7. Subjects & Methods Data Source Crashworthiness Data System (CDS) of National Automotive Sampling System (NASS) Available at: ftp://ftp.nhtsa.dot.gov/nass/ Accessed September 17 2003. The NASS CDS provides a comprehensive national crash database each year from 1979 to 2001. These data were weighted to represent all police reported motor vehicle crashes occurring in the USA during the year involving passenger cars, light trucks and vans that were towed due to damage. Subject Selection Survey time 1997 – 2001 Age of subjects (yrs) 16 or over Number of eligibility 30, 667 drivers Exclusion 8, 230 1) Driving motorcycles or vehicle type unknown 2,154 2) Pregnant female drivers 303 3) Lack height or weight information 5,713 4) Death due to causes other than crash 60 5) BMI 45.0 330 Subjects used in analyses 22, 107 (12 million) Men drivers 13007; Women drivers 9100

8. Variable Selection Variable Selection Outcome variable: Motor vehicle death (within 30 days of crash) 1) Other variables: 1) Main explanatory variables: BMI (Body Mass Index, kg/m2) and BMI 2 2) Model controlled variables: Characteristics of subject and vehicle: Age, and vehicle age and weight. Driving behavior: Alcohol and drug intake, and seatbelt use. Collision: Airbag deployment, manner and type of collisions, road speed limit, and delta V.

9. Statistical Analysis 1). Model without delta-V (Model 1): Model A: BMI; Model B: add BMI 2 into Model A; Model C: test interactions of BMI and BMI 2 with gender, age, seatbelt use, airbag deployment, and type of collision. 2). Model with delta-V (Model 2): Re-run models (Model A – C) with delta-V Model D: test interaction between BMI and delta-V. All models were adjusted for controlled variables that related to drivers, vehicles or collisions Logistic Regression Model

10. Men Women Sample size 13007 9100 Weighted size (million) 6.49 5.06 Mean95% CI Mean95% CI Age (yr) 35.9 34.7 – 37.1 35.234.3 – 36.2 Height (cm) 177.9177.0 – 178.8 164.3164.0 – 164.6 Weight (kg) 81.981.3 – 82.5 66.865.1 – 68.4 Body Mass Index (kg/m 2 ) 25.825.6 – 26.1 24.724.2 – 25.3 Vehicle Age (yr) 7.47.0 – 7.8 6.05.7 – 6.4 Vehicle Weight (kg) 1435.41412.4 – 1458.3 1338.91312.4 – 1365.3 Road Speed Limit (km/hr) 67.364.8 – 69.7 66.164.5 – 67.8 Delta V (km/hr) 21.719.9 – 22.1 19.919.3 – 20.6 Table 1 (A). Characteristics of driver, vehicle, behaviors and collision by driver’s sex Percent (%) 95% CI Percent (%) 95% CI Motor vehicle fatality 0.87 0.50 – 1.24 0.43 0.31 – 0.56

11. Table 1 (B). Characteristics of driver, vehicle, behaviors and collision by driver’s sex Percent % 95% CI Percent % 95% CI Race-ethnicity White61.2 51.7 – 70.8 66.155.5 – 76.7 Black12.3 8.2 – 16.4 14.07.5 – 20.4 Hispanic7.6 5.4 – 9.8 4.63.1 – 6.1 Other3.4 2.0 – 4.8 2.10.7 – 3.6 Unknown15.5 9.3 – 21.6 13.18.7 – 17.5 Type of Collision Front58.456.3 – 60.6 56.853.1 – 60.5 Left side14.912.2 – 17.5 15.213.1 – 17.2 Right side14.813.2 – 16.4 18.417.0 – 19.8 Other11.79.6 – 13.9 9.36.6 – 12.0 Unknown0.20.0 – 0.4 0.30.0 – 0.6 Airbag Deployment Deployed17.815.7 – 19.8 22.821.2 – 24.4 Not-deployed79.777.8 – 81.6 74.972.9 – 76.9 Unknown2.61.7 – 3.4 2.31.5 – 3.1 Seat Belt Use No12.07.5 – 16.4 6.13.7 – 8.4 Yes81.075.4 – 86.7 87.983.1– 92.6 Unknown7.02.6– 11.4 6.11.2 – 10.9

12. Table 1 (C). Characteristics of driver, vehicle, behaviors and collision by driver’s sex Percent %95% CI Percent % 95% CI Manner of Collision Single vehicle29.824.8 – 34.8 23.217.9 – 28.4 Two or more68.964.0 – 73.8 75.570.2 – 80.8 Unknown1.30.7 – 2.0 1.40.4 – 2.3 Alcohol Involved Positive12.69.8 – 15.4 4.92.9 – 6.8 Negative75.472.3 – 78.5 83.581.1 – 85.9 Unknown12.110.3 – 13.9 11.78.2 – 15.2 Drug Involved Positive2.60.9 – 4.2 1.60.7 – 2.5 Negative71.662.2 – 81.0 77.470.6 – 84.3 Unknown25.816.9 – 34.8 21.013.8 – 28.1 Light Condition Daytime, light62.660.7 – 64.4 72.468.7 – 76.1 Night, light18.214.7 – 21.6 14.611.6 – 17.6 Dark19.215.8 – 22.6 12.911.0 – 14.8 Unknown0.060.00 – 0.12 0.050.00 – 0.11

13. Table 2. Adjusted OR for motor vehicle fatality by sex (Model 1 all subjects; Model 2 with delta-V) Men drivers Women drivers Model 1Model 2Model 1Model 2 Sample size 12122690586235493 Weighted 6.1 million3.2 million4.8 million2.8 million OR 95% CIP OR 95% CIP OR 95% CIP OR 95% CIP BMI (kg/m 2 ) 0.67 0.52 – 0.86 0.0040.53 0.37 – 0.740.001 0.99 0.78 – 1.27 0.9790.69 0.48 – 1.00 0.051 BMI 2 1.01 1.00 – 1.01 0.0031.01 1.01 – 1.02<.001 1.00 1.00 – 1.00 0.868 1.01 1.00 – 1.01 0.044 Nadir of BMI §2 28.3 (CI, 26.0 – 30.7, P<0.001) 28.4 (CI, 25.9 - 3 1.0, P<0.001) - 27.8 Age (yr) 1.03 1.02 – 1.04 <.0011.04 1.02 – 1.06<.001 1.05 1.04 – 1.06 <.0011.06 1.03 – 1.08 <.001 Delta V Delta-V 2 1.15 1.00 1.11 – 1.19 1.00 – 1.00 <.001 0.001 1.35 1.00 1.26 – 1.46 1.00 – 1.00 <.001 Seatbelt Yes 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) No Airbag 10.3 6.08 – 17.6 <.0017.17 3.05 – 19.09<.001 12.2 6.96 – 21.2 <.0016.23 3.79 – 10.2 <.001 Not deployed 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) Deployed1.98 1.33 – 2.94 0.0031.40 0.79 – 2.470.224 1.96 1.19 – 3.21 0.012 0.48 0.28 – 0.84 0.015 Type collision <.001 0.006 <.001 0.001 Front1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) Left2.05 1.41 – 2.96 <.0014.26 2.91 – 6.24<.001 3.12 1.83 – 5.38 0.0015.79 4.17 – 8.05 <.001 Right2.27 1.62 – 3.18 <.0011.80 0.90 – 3.600.087 1.77 0.70 – 4.25 0.0711.40 0.58 – 3.38 0.427 Other0.93 0.44 – 1.98 0.8420.43 0.13 – 1.380.141 2.04 1.41 – 2.94 0.0011.08 0.26 – 4.45 0.913

14. Fig 1A. Adjusted ORs of BMI for fatality from Model 1 by sex (Reference BMI: 25 kg/m 2 )

15. Fig 1B. Adjusted ORs of BMI for fatality stratified by type of collision in men drivers from Model 1 (Reference BMI: 25 kg/m 2 on same condition)

16. Fig 2A. Adjusted ORs of BMI for fatality from Model 2 by sex (Reference BMI: 25 kg/m 2 )

17. Fig 2B. Adjusted ORs of BMI for fatality stratified by type of collision in men drivers from Model 2 (Reference BMI: 25 kg/m 2 on same condition)

18. Fig 3A. Adjusted ORs of BMI for fatality stratified by change of velocity during crashes in men drivers involved in frontal collision (Reference BMI: 25 kg/m 2 ) drivers involved in frontal collision (Reference BMI: 25 kg/m 2 )

19. Fig 3B. Ratio scale of BMI for fatality stratified by change of velocity during crashes in men drivers involved in frontal collision (Reference BMI: 25 kg/m 2 ) drivers involved in frontal collision (Reference BMI: 25 kg/m 2 )

20. Discussion Our study suggests that men drivers have significantly increased risk of death in both ends of the BMI continuum than do women drivers. The relationship between BMI and motor vehicle fatality was modified by the type of collision but largely differed very little with age, airbag deployment and seatbelt use. The magnitude of the increasing risk for fatality observed at the high end of the BMI continuum in men drivers was determined mainly by the magnitude of ∆V.

21. Discussion Possible Mechanism It is likely that the increased risk of dying in motor vehicle crashes observed at the high end of BMI continuum may be due, in part, to the momentum effect caused by body weight; the obesity co-morbidities such as cardiovascular disease or dysfunction (CVD) and diabetes; sleep apnea, a known factor in motor vehicle crash injuries in the obese; and emergency, operative, and postoperative treatment problems in obese.

22. Discussion Gender Difference We detect a significant interaction of BMI with gender in all crashes as well as crashes with ∆V information available. The reasons for this gender difference are unknown. However, body shape may be a factor. Previous studies have shown that body fat distribution affect all-cause mortality and cardiovascular risk in women more than in men. While fat distribution is usually cited as a negative factor in women’s mortality rates, certain distributions and levels of body fat may have a protective factor in some crashes.

23. Discussion Vehicle Design Our observation of an optimal BMI in men of approximately 28 kg/m 2, which is considered overweight, was in agreement with a study showing a ‘cushion effect’ among overweight people. Current vehicle cabin design accords with the Federal Motor Vehicle Safety Standard using the 50 th percentile men Hybrid III Crash Dummy (H3CD, 1.78 m, 77.11 kg in the driver’s position, BMI = 24.3 kg/m 2 ). These standards may not be optimal for drivers with a different body habitus and may contribute to the higher fatality seen at both ends of the BMI continuum. Future dummy crash simulations and other studies should account for individual and gender-related variations in body mass and fat distribution in tests of velocity and vehicle design.

24. Discussion Limitations A potential limitation was that approximately 29% of the drivers who met the initial inclusion criteria were excluded from analyses. This might cause bias in either direction in the relationships of BMI with motor vehicle fatality.

25. Discussion Strengths Using nationally representative NASS CDS data enabled us to describe the relationships in the US population, while the large number of subjects increased statistical power in analyses. Focusing only on drivers and using separate analyses for men and women allowed us to eliminate the potential gender differences as well as differences in causal pathways and confounding factors between drivers and passengers. Including and excluding squared terms for BMI enabled us to obtain a better data fit for regression models. Testing interaction terms of BMI with some important covariates told us how these factors influence the relationship between BMI and motor vehicle fatality. And simultaneously adjusting for potentially confounding factors allowed us to assess the relationship of BMI with fatality independent of other factors.

26. Discussion Conclusions This study extends and improves on earlier studies that showed increased mortality in obese drivers by 1)quantifying the magnitude of the associations of BMI with motor vehicle fatality and 2)providing a clear graphic description of how type of collision and ∆V modify the likelihood of death due to BMI in motor vehicle crashes. These gender-specific findings could have an impact on high-risk cohort detection, motor vehicle design and motor vehicle injury prevention and treatment.

27. Acknowledgement Injury Research Center Department of Family and Community Medicine Medical College of Wisconsin Peter Layde MD, MSc Clare Guse MS Purushottam Laud PhD Frank Pintar PhD Stephen Hargarten MD, MPH Carol Cameron (Injury Research Center) Mary Czinner (Injury Research Center) Chris McLaughlin (Dept Family and Community Medicine)

28. THANK YOU!

29. Fig 1C. Adjusted ORs of BMI for fatality stratified by type of collision in women drivers from Model 1 (Reference BMI: 25 kg/m 2 on same condition)

30. Fig 2B. Adjusted ORs of BMI for fatality stratified by type of collision in men drivers from Model 2 (Reference BMI: 25 kg/m 2 in front collision)

31. Fig 2C. Adjusted ORs of BMI for fatality stratified by seatbelt use in men drivers from Model 2 (Reference BMI: 25 kg/m 2 on same condition)

32. Fig 2D. Adjusted ORs of BMI for fatality stratified by airbag deployment in men drivers from Model 2 (Reference BMI: 25 kg/m 2 on same condition)

33. Fig 3A. Adjusted ORs of BMI for fatality stratified by type of collision and seatbelt use in men drivers from Model 2 (Reference BMI: 25 kg/m 2 on same condition)

34. Fig 3B. Adjusted ORs of BMI for fatality stratified by type of collision and airbag deployment in men drivers from Model 2 (Reference BMI: 25 kg/m 2 on same condition)

35. Fig 4. Adjusted ORs of BMI for fatality in front collision in men and women drivers with delta-V in the model (Model 2) (Reference BMI: 25 kg/m 2 )