Predictive validity of driving-simulator assessments following traumatic brain injury: a preliminary study

Objective: To evaluate whether driving simulator and road test evaluations can predict long- term driving performance, we conducted a prospective study on 11 patients with moderate to severe traumatic brain injury. Sixteen healthy subjects were also tested to provide normative values on the simulator at baseline. Method: At their initial evaluation ( time- 1), subjects' driving skills were measured during a 30- minute simulator trial using an automated 12- measure Simulator Performance Index ( SPI), while a trained observer also rated their performance using a Driving Performance Inventory ( DPI). In addition, patients were evaluated on the road by a certified driving evaluator. Ten months later ( time- 2), family members observed patients driving for at least 3 hours over 4 weeks and rated their driving performance using the DPI. Results: At time- 1, patients were significantly impaired on automated SPI measures of driving skill, including: speed and steering control, accidents, and vigilance to a divided- attention task. These simulator indices significantly predicted the following aspects of observed driving performance at time- 2: handling of automobile controls, regulation of vehicle speed and direction, higher- order judgment and self- control, as well as a trend- level association with car accidents. Automated measures of simulator skill ( SPI) were more sensitive and accurate than observational measures of simulator skill ( DPI) in predicting actual driving performance. To our surprise, the road test results at time- 1 showed no significant relation to driving performance at time- 2. Conclusion: Simulator- based assessment of patients with brain injuries can provide ecologically valid measures that, in some cases, may be more sensitive than a traditional road test as predictors of long- term driving performance in the community.