With an increasingly aged population the number of older drivers is on the increase. Driving is the most common form of transport for older people, with only 3% of journeys by older people in the US undertaken by public transport. Transport and access to transport is an important factor in maintaining social inclusion and participation for older people: there is also considerable evidence that limiting this access (i.e. by driving cessation) is associated with poor health and depression (Marottoli 2000). Exclusion of the elderly from driving has been demonstrated to have many negative effects; loneliness, lower life satisfaction and lower activity levels. Drivers with dementia both limit their driving and eventually withdraw, with profound negative effects on their mobility (Taylor 2001).
We do not know what interventions will maintain safe mobility for the longest period for people with dementia. A full assessment (often including an on‐road assessment) may have beneficial effects for drivers with many illnesses (Stutts 2003), and restricted driving licences following assessment are associated with a lower crash risk (Vernon 2002). An assessment in dementia also allows for advance planning of driving withdrawal and substitution of other transportation modalities, by way of the so‐called modified Ulysses pact (Robinson 2004). It also allow for regular re‐testing to guide this process (Duchek 2003).
Although there is no standardized neuropsychological assessment battery, most driving assessment protocols include medical, occupational therapy and/or neuropsychology assessment and an on‐road driving assessment as required. We have no evidence that this intervention enhances either mobility or safety.

We will seek evidence that formalized driving assessments are effective in maintaining safety and mobility in drivers with dementia rather than merely excluding those who are unfit to drive. Continued driving and outdoor mobility is important to older people and is a sufficient outcome measure in its own right. Safety is a secondary end‐point, as crashes are relatively infrequent events, and fatal crashes even more so, 1 per 50 million km. RCTs of driving assessment should involve a careful design, randomizing patients who drive to either formal testing or usual care and assessment, with longitudinal follow‐up of satisfaction with transport (Rosenbloom 2003) and crashes or violations. Ethically, this should not pose a public health hazard in view of memory clinic studies which suggest no increase in crashes in those with dementia (Carr 2000; Drachman 1993; Trobe 1996), but due care will need to be given to advice to participating drivers on informing driver licensing authorities and insurance companies, depending on the jurisdiction(s) within which the study is taking place.