A prospective study of cognitive tests to predict performance on a standardised road test in people with dementia

Objective Previous work by Lincoln and colleagues produced a cognitive test battery for predicting safety to drive in people with dementia. The aim was to check the accuracy of this battery and assess whether it could be improved by shortening it, including additional cognitive tests, and a measure of previous driving. Methods Participants with dementia, who were driving, were recruited. They were assessed on cognitive tests including measures of concentration, executive function, visuospatial perception, verbal recognition memory, and speed of information processing. Patients were then assessed on the Nottingham Neurological Driving Assessment (NNDA) by an approved driving instructor (ADI), blind to cognitive test results. Results Seventy‐five patients were recruited and completed the cognitive tests. Of these, 65 were assessed on the road. These participants were aged 59–88 (mean = 75.2, SD = 6.8) and 49 were men. Time driving varied from 19 to 73 years (mean = 52.5, SD = 10.0). Thirteen participants were unsafe and 52 safe to drive. Using a cut‐off of > 0 to indicate safety to drive, the original predictive equations correctly classified 48 (76.2%) of 63 participants with complete data. Logistic regression including additional tests reduced misclassifications. Conclusions A lower proportion of participants were found to be unsafe on the road than in previous studies. Nevertheless, the previously identified equation predicted safety to drive in most patients. Including additional tests reduced the misclassification rate but requires independent validation. We suggest that the cognitive test battery might be used in clinical practice to identify patients with dementia who would benefit from on‐road assessment. Copyright © 2009 John Wiley & Sons, Ltd.