MMSE as a predictor of on-road driving performance in community dwelling older drivers.

Screening tools such as the MMSE have been used extensively in driving research studies to determine mild cognitive impairment or dementia. While some studies have shown the MMSE to correlate with driving performance, few studies have shown the predictive validity of the MMSE in determining on-road performance. In a sample of 168 community dwelling older adults, including 20 with Parkinson's disease (PD), the primary objective was to determine the validity of the MMSE to predict pass/fail outcomes of an on-road driving test using receiver operating characteristics curves. The area under the curve (AUC), an index of discriminability, for the total sample was .654, 95% CI=0.536-0.772, p=.009. Meanwhile, the AUC for the PD group was 0.791, 95% CI=0.587-0.996, p=.036. The total sample showed statistically significant yet poor predictive validity. However, the PD group showed statistically significant and good predictive validity of the MMSE to predict pass/fail outcomes on the road test, but caution is warranted as the confidence intervals are wide (due to small sample) and the positive and negative predictive values are less than desirable due to the associated error. The findings show that using the current cut-off point of ≤24 on the MMSE is not adequately sensitive to predict on-road performance in both community dwelling older drivers and in drivers with PD. This study offers strong evidence to support the current best practice of not using the MMSE in isolation to predict on-road performance.