Comparison of driving simulator performance with real driving after alcohol intake: a randomised, single blind, placebo-controlled, cross-over trial

Abstract

The purpose of this study was to establish and validate a driving simulator method for assessing drug effects on driving. To achieve this, we used ethanol as a positive control, and examined whether ethanol affects driving performance in the simulator, and whether these effects are consistent with performance during real driving on a test track, also under the influence of ethanol. Twenty healthy male volunteers underwent a total of six driving trials of 1h duration; three in an instrumented vehicle on a closed-circuit test track that closely resembled rural Norwegian road conditions, and three in the simulator with a driving scenario modelled after the test track. Test subjects were either sober or titrated to blood alcohol concentration (BAC) levels of 0.5g/L and 0.9g/L. The study was conducted in a randomised, cross-over, single-blind fashion, using placebo drinks and placebo pills as confounders. The primary outcome measure was standard deviation of lateral position (SDLP; "weaving"). Eighteen test subjects completed all six driving trials, and complete data were acquired from 18 subjects in the simulator and 10 subjects on the test track, respectively. There was a positive dose-response relationship between higher ethanol concentrations and increases in SDLP in both the simulator and on the test track (p<0.001 for both). In the simulator, this dose-response was evident already after 15min of driving. SDLP values were higher and showed a larger inter-individual variability in the simulator than on the test track. Most subjects displayed a similar relationship between BAC and SDLP in the simulator and on the test track; however, a few subjects showed striking dissimilarities, with very high SDLP values in the simulator. This may reflect the lack of perceived danger in the simulator, causing reckless driving in a few test subjects. Overall, the results suggest that SDLP in the driving simulator is a sensitive measure of ethanol impaired driving. The comparison with real driving implies relative external validity of the simulator.