Driving impairment and crash risk in Parkinson disease: A systematic review and meta-analysis.

Abstract

To provide the best possible evidence base for guiding driving decisions in Parkinson disease (PD), we performed a meta-analysis comparing patients with PD to healthy controls (HCs) on naturalistic, on-the-road, and simulator driving outcomes. Seven major databases were systematically searched (to January 2018) for studies comparing patients with PD to HCs on overall driving performance, with data analyzed using random-effects meta-analysis. Fifty studies comprising 5,410 participants (PD = 1,955, HC = 3,455) met eligibility criteria. Analysis found the odds of on-the-road test failure were 6.16 (95% confidence interval [CI] 3.79-10.03) times higher and the odds of simulator crashes 2.63 (95% CI 1.64-4.22) times higher for people with PD, with poorer overall driving ratings also observed (standardized mean differences from 0.50 to 0.67). However, self-reported real-life crash involvement did not differ between people with PD and HCs (odds ratio = 0.84, 95% CI 0.57-1.23, <i>p</i> = 0.38). Findings remained unchanged after accounting for any differences in age, sex, and driving exposure, and no moderating influence of disease severity was found. Our findings provide persuasive evidence for substantive driving impairment in PD, but offer little support for mandated PD-specific relicensure based on self-reported crash data alone, and highlight the need for objective measures of crash involvement.