Background: On-road driving behavior can be impaired in older drivers and particularly in drivers with mild cognitive impairment (MCI). Objective: To determine whether cognitive and non-cognitive risk factors for driving safety may allow an accurate and economic prediction of on-road driving skills, fitness to drive, and prospective accident risk in healthy older drivers and drivers with MCI, we examined a representative combined sample of older drivers with and without MCI (N = 74) in an observational on-road study. In particular, we examined whether non-cognitive risk factors improve predictive accuracy provided by cognitive factors alone. Methods: Multiple and logistic hierarchical regression analyses were utilized to predict different driving outcomes. In all regression models, we included cognitive predictors alone in a first step and added non-cognitive predictors in a second step. Results: Results revealed that the combination of cognitive and non-cognitive risk factors significantly predicted driving skills (R2adjusted = 0.30) and fitness to drive (81.2% accuracy) as well as the number (R2adjusted = 0.21) and occurrence (88.3% accuracy) of prospective minor at-fault accidents within the next 12 months. In all analyses, the inclusion of non-cognitive risk factors led to a significant increase of explained variance in the different outcome variables. Conclusion: Our findings suggest that a combination of the most robust cognitive and non-cognitive risk factors may allow an economic and accurate prediction of on-road driving performance and prospective accident risk in healthy older drivers and drivers with MCI. Therefore, non-cognitive risk factors appear to play an important role.
Bayesian network analysis of accident risk in information-deficient scenarios
