SIMULATED CAR CRASHES AND CRASH PREDICTORS IN DRIVERS WITH ALZHEIMER-DISEASE

Background: Alzheimer disease (AD) is the most common cause of dementi a and can impair cognitive abilities crucial to the task of driving. R ational decisions about whether such impaired individuals should conti nue to drive require objective assessments of driver performance. Obje ctive: To measure relevant performance factors using high-fidelity dri ving simulation. Design: We examined the effect of AD on driver collis ion avoidance using the Iowa Driving Simulator, which provided a high- fidelity, closely controlled environment in which to observe serious e rrors by at-risk drivers. We determined how such unsafe events are pre dicted by visual and cognitive factors sensitive to decline in aging a nd AD. Setting: The University of Iowa Hospitals and Clinics, Iowa Cit y, and the Iowa Driving Simulator. Participants: Thirty-nine licensed drivers: 21 with AD and 18 controls without dementia. Main Outcome Mea sures: We determined the number of crashes and related performance err ors and analyzed how these occurrences were predicted by visual and co gnitive factors. Results: Six participants (29%) with AD experienced c rashes vs 0 of 18 control participants (P=.022). Drivers with AD were more than twice as likely to experience close calls (P=.042). Plots of critical control factors in the moments preceding a crash revealed pa tterns of driver inattention and error. Strong predictors of crashes i ncluded visuospatial impairment, reduction in the useful field of view , and reduced perception of 3-dimensional structure-from-motion. Concl usions: High-fidelity driving simulation provides a unique new source of performance parameters to standardize the assessment of driver fitn ess. Detailed observations of crashes and other safety errors provide unbiased evidence to aid in the difficult clinical decision of whether older or medically impaired individuals should continue to drive. The findings are complementary to evidence currently being gathered using techniques from epidemiology and cognitive neuroscience.