Assessment of driver fitness: An alcohol calibration study in a high-fidelity simulation 26 April 2013 Dr. Yvonne Kaussner WIVW Wuerzburg Institute for Traffic Sciences
Methodological issues on the assessment of driver fitness Modern driving simulators combine the advantages of classical psychometric test batteries and on road trials and solve the shortcomings: - Systematic design, presentation and reproduction of scenarios (Kaussner & Krüger, 2005) - Neither patients nor other road users are endangered - High face validity, well-accepted by subjects - Practice of compensatory strategies possible as in real traffic (Kaussner, 2007) 2
Methodological issues on the assessment of driver fitness in driving simulation Software and scenarios: - Theoretical foundation - Representativeness and sensitivity - Validation Hardware configuration: - High quality of steering gears, visual system and pedals - Compatibility with scenarios Endpoints: - Assessment of driving performance as a whole - Profile of various theoretically well-founded parameters (e.g. according to Michon 1985) 3
Study Design Influence of 0.05%, and 0.08% blood alcohol concentration (BAC) on driver fitness in comparison to placebo Design: Double blind, randomized, crossover. Subjects: 24 healthy volunteers (11 females / 13 males) mean age 30 years (SD=8.3) mean BMI 24 (SD=3.2) mean consumption of alcohol per week 34 g (SD=16 g) simulator familiarization passed with very good tolerance Alcohol application: Computation of required alcohol quantum with a formula on total body water (depending on gender, weight, height; see Watson et al. 1980 and Vollrath 2000) 4
Driving Simulator Motion system: - Stewart platform with 6 degrees of freedom - 6 electrical actuators, 3 passive pneumatic actuators Visual system: - 180 frontal, horizontal field of view, projected onto a spherical screen - LCD-displays for rearview and two outside mirrors Mockup: completely equipped BMW 520i Driving simulation software SILAB by WIVW 5
Test course Three parallel versions of a representative set of driving scenarios from the SILAB scenario package DFA (Driver Fitness and Ability) Covering 80 km (about 60 min to drive) Designed to test safe driving in compliance with the German Traffic Regulations (Schubert & Wagner 2003) Representative for real driving according to different classifications and guidelines (e.g. Brenner-Hartmann 2002, Bukasa & Piringer 2001, Michon 1985, ICADTS Working Group 2009) Sensitive for the influence of psychoactive drugs and neurological conditions as empirically proven in prior studies (e.g. Kaussner et al. 2010) 6
Exemplary DFA scenarios Tracking scenarios of varying difficulty to assess lane keeping (operational level) Cognitive demanding scenarios (tactical level) 7
Endpoints 1. Expert raters overall assessment of driving performance 11 point rating scale according to Neukum & Krüger (2003) with 3 main categories (driving behavior was normal / noticeable / critical) 2. Driving errors registered and classified into errors in lateral control, errors in longitudinal control, cognitively based errors, number of critical situations and collisions 3. Well-established tactical and operational driving parameters recorded by the simulation Lateral: Standard deviation of lane position (SDLP), number of lane of departures Longitudinal: mean speed, time headway to leading vehicles (% of time <1s) Cognitive: size of accepted gaps, number of lane changes on highways 4. Subjective overall assessment of driving performance 8
mean z-values (+/-SE) Results 3,000 Global assessments & errors (z-standardization with placebo condition) 2,500 * *** *** ** *** ** 2,000 1,500 1,000 0.00% 0.05% 0.08%,500,000 * p <.050 ** p <.005 *** p <.001 9
mean z-values (+/-SE) Results 003 Lateral driving data (operational level) (z-standardization with placebo condition) 003 * ** ** 002 002 001 0.00% 0.05% 0.08% 001 000 * p <.050 ** p <.005 *** p <.001 10
mean z-values (+/-SE) Results 003 003 002 Longitudinal and cognitive data (tactical level) (z-standardization with placebo condition) * ** * 002 001 0.00% 0.05% 0.08% 001 000 * p <.050 ** p <.005 *** p <.001 11
Conclusions A comparison with the impairing effects of 0.05% and 0.08% BAC helps to assess the clinical relevance of findings of future studies. The study clearly showed alcohol-related deficits confirming the validity and sensitivity of the experimental set up. Parameters differentiated distinctly between the alcohol conditions depending on the underlying scenarios. Thus, the selection of driving scenarios and driving performance parameters is of crucial importance for an assessment of driver fitness. For further studies on drug and disease related effects on driver fitness, we recommend to evaluate driving performance as a whole by a profile-like analysis of various parameters at different levels of the driving task. 12
z-values Hypothetical profiles of a stimulating and a highly sedating drug for global assessments & errors 3,000 2,500 2,000 1,500 stimulating drug sedating drug 0.05% 0.08% 1,000,500,000 13
Dr. Yvonne Kaussner WIVW GmbH yvonne.kaussner@wivw.de