Use of self-reporting

Studies looking at self-reports of crashes and abnormal driving behaviour have found problems with recall bias. Reference HollandHolland (1993) showed that most drivers perceived their own chance of having a road accident to be significantly lower, and their own skill to be greater, than that of their peers. When this is enhanced by the forgetfulness and lack of insight found in dementia, there are obvious shortcomings with using self-report alone as a method of risk assessment.

Use of carer reporting

A commonly used source of information on a patient's fitness to drive is an interview with carers, but Reference JohnsonJohnson (1998) noted that friends and relatives do not always give a true picture. This may be because they do not wish to take away the patient's independence, or because they do not wish to become burdened by a relative who can no longer drive but wishes to remain sociable. Partners may overestimate driving ability if they rely on their loved one to get around or have memory impairment themselves. Informants are also wary of being blamed for a patient's licence being taken away.

Reference Adler, Rottunda and BauerAdler et al (2000) found that drivers with dementia had a greater risk of crashes and getting lost than controls, and relatives often took over responsibility for the transportation needs of older family members. They also found that concerned relatives tried to stop individuals with dementia from driving, but long-standing family dynamics often interfered with the sensitive negotiation required. Older people are more likely to accept a recommendation to stop driving from a doctor or authority figure than from a relative or friend.

Carers' reports can be unreliable, as shown in a study by Reference Brown, Ott and PapandonatosBrown et al (2005). Participants (with mild, very mild or no dementia), informants and an experienced neurologist were asked to rate participants' driving ability as safe, marginal or unsafe. The participants then underwent an on-road driving assessment with a professional driving instructor. Only the neurologist's ratings were significantly related to the on-road assessment.

However, Reference Croston, Meuser and Berg-WegerCroston et al (2009) reported that informants often observed drivers with Alz-heimer's dementia having difficulty with traffic awareness, maintaining appropriate speeds and staying in their own lane (although in up to 25' of cases it took an on-road incident to prompt the decision to stop driving). This suggests that informants may be reliable sources of information if asked appropriate and relevant questions about driving safety

Use of other collateral information

It may be useful to enquire whether the patient has been involved in any recent accidents, or indeed a community psychiatric nurse may report seeing dents and scratches appearing on the car. Collecting data from police records and insurance companies poses difficulties as diagnoses of dementia may be unknown, especially in the early stages, and minor bumps and offences often go unnoticed or unreported.

Two US studies (Reference Trobe, Waller and Cook-FlannaganTrobe 1996; Reference Carr, Duchek and MorrisCarr 2000) have looked at state-recorded vehicle crash details and found that the crash and violation rates of individuals with dementia were not significantly different to those of age-matched controls. Both studies found, however, that patients with dementia did not drive as frequently or as far as controls, and the crash characteristics between the two groups differed, with the dementia group having more at-fault crashes.

The importance of the clinical presentation

It is really important to consider the nature of the symptom profile and natural history of the patient's specific dementia. Establishing the dementia subtype is important, as some subtypes are associated with higher risk than others. Dementia with Lewy bodies and frontotemporal dementia are the riskiest dementias in relation to driving (British Psychological Society 2001). The hallucinations associated with dementia with Lewy bodies are likely to have an impact on attention when driving, and any Parkinsonian symptoms will affect reaction times (Reference Wood, Worringham and KerrWood 2005). The changes in impulsivity, forward planning and shifting attention found in frontotemporal dementia mean driving skills can be affected very early in the disease process. Reference Zuin, Ortiz and BoromeiZuin et al (2002) found that patients with frontotemporal dementia had the highest number of accidents. They also noted that these patients had a stronger desire to continue driving and their families found it harder to try to stop them.

Studies looking at specific types of dementia have mostly focused on people with Alzheimer's disease. Most individuals with moderate to severe Alzheimer's disease are unsafe or unable to drive, so studies have focused on the mild to very mild end of the spectrum (Reference Dubinsky, Stein and LyonsDubinsky 2000; Reference Duchek, Carr and HuntDuchek 2003; Reference Brown, Ott and PapandonatosBrown 2005). The preferred diagnostic tool is the Clinical Dementia Rating scale, which can distinguish between very mild and mild Alzheimer's disease (Reference MorrisMorris 1993). This reliable and validated semi-structured interview for patients and informants takes 90 min to complete and gives scores of 0 (no dementia), 0.5 (very mild dementia), 1 (mild dementia), 2 (moderate dementia) and 3 (severe dementia). Very mild dementia is equivalent to mild cognitive impairment (Reference Duchek, Carr and HuntDuchek 2003).

Drivers with a Clinical Dementia Rating score of 1 were found to have an increased risk of crashes and abnormal driving behaviours than those with scores of 0.5 (Reference Dubinsky, Stein and LyonsDubinsky 2000). Overall suggestions are for individuals with mild Alzheimer's disease to have their driving status reviewed every 6 months, while those with very mild Alzheimer's disease to be reviewed every year.

The general consensus appears to be that it is usually safe to continue driving for about 3 years after the onset of Alzheimer's disease (Reference Trobe, Waller and Cook-FlannaganTrobe 1996; Reference Hopkins, Kilik and DayHopkins 2004; Reference Breen, Breen and MooreBreen 2007). Other dementias, such as the vascular type, may remain stable for longer, although the cognitive decline is far more unpredictable. As discussed above, frontotemporal dementia and dementia with Lewy bodies usually require individuals to stop driving much earlier. Sleep disturbance, anxiety and low mood associated with dementia may also affect driving ability by reducing concentration (Reference Gilley, Wilson and BennettGilley 1991).

Consider comorbidities

Older people with dementia often have multiple physical comorbidities which can affect driving ability and may require the individual to stop driving. Medications prescribed for physical comorbidities, as well as antidepressants and anti-psychotics, can affect sensorimotor and perceptual abilities that may have a further detrimental effect on driving ability (Reference Gilley, Wilson and BennettGilley 1991; Reference Zuin, Ortiz and BoromeiZuin 2002). The effect of cholinesterase inhibitors on driving performance has not yet been examined, but this is an important area given their increased use and benefit in milder stages of the illness.

A note on the MMSE

The Mini-Mental State Examination (MMSE; Reference Folstein, Folstein and McHughFolstein 1975) is a screening tool rather than a functional assessment and does not test functions specific to driving, i.e. executive function, judgement, divided attention and perception. Although an MMSE score of <20/30 would suggest moderate dementia and infer difficulty driving, the test is not sensitive to mild cognitive impairment and can be affected by the educational level of the patient, so risky drivers could still score >27/30 (Reference Dubinsky, Stein and LyonsDubinsky 2000; Reference Rees, McKenna and BellRees 2008). Although the MMSE is in common use, it is not helpful in determining whether someone with dementia is safe to drive and, indeed, its future use may be seriously curtailed owing to copyright issues.

Modification of driving behaviour

It is important to appreciate that many older people begin to modify their driving behaviour with time, avoiding busy roads at busy times, using familiar routes and not driving in bad weather or at night. When comparing driving safety statistics and age, this needs to be taken into account. Previous perceived wisdom that older people as a group have more crashes has been disproven, as studies have established that this is due to low mileage and not increasing age (Reference Langford, Methorst and Hakamies-BlomqvistLangford 2006). Evidence suggests that patients with Alzheimer's disease usually modify their driving behaviour, but this is often not enough to reduce their crash risk completely (Reference Adler, Rottunda and BauerAdler 2000; Reference Man-Son-Hing, Marshall and MolnarMan-Son-Hing 2007).

Driving cessation

A number of studies have looked at driving retirement (Reference Adler, Rottunda and BauerAdler 2000; Reference Croston, Meuser and Berg-WegerCroston 2009) and found that very few people even consider making plans for when they are no longer able to drive. This is an inevitable stage of life for most people; Reference Foley, Heimovitz and GuralnikFoley et al (2002) showed that men tend to outlive their driving ability by 6 years and women by 10 years. Even when given a diagnosis of dementia, many individuals do not make plans for complete driving cessation (Reference JohnsonJohnson 1998; Reference Breen, Breen and MooreBreen 2007). This may in part be due to individuals not recognising they have impairment as a result of their cognitive state, quite different to people with physical illnesses such as epilepsy or visual problems when insight is not impaired.

Making the decision to retire from driving will have significant psychosocial consequences that may have a negative impact on people with dementia. Reference Marottoli, Mendes de Leon and GlassMarottoli et al (1997, Reference Marottoli, Mendes de Leon and Glass2000) has shown that cessation of driving leads to a significant reduction in out-of-home activity levels, as well as an increase in depressive symptoms. It is increasingly clear that to stop driving is a life-altering decision, reducing independence and limiting access to family, friends and services (Reference Breen, Breen and MooreBreen 2007).

DVLA and the law

In the UK, driving licences are provided by the DVLA. Since the agency was set up in 1972, it has developed medical guidance for many illnesses that can adversely affect driving ability. Dementia has only been recognised as a specific condition in recent years. Prior to this, dementia and cognitive impairment were grouped under the generic heading of ‘neurological conditions’. Over time, guidelines for many medical conditions have become clearer and tighter; however, the current guidance for drivers with dementia remains necessarily vague (Box 1).

The responsibility for renewing or revoking driving licences in the UK ultimately falls to the DVLA, but it is recognised that decisions are usually based on medical reports from doctors who care for these patients. A Canadian report by Reference Hopkins, Kilik and DayHopkins et al (2004) noted that the responsibility for identifying drivers with dementia has fallen on the healthcare system, a role for which it was never designed or equipped to handle. They also noted that placing this responsibility on clinicians can be detrimental to the doctor–patient relationship.

Reference Naidu and McKeithNaidu & McKeith (2006) surveyed old age psychiatrists who completed DLVA medical reports at the time and concluded that the current system for determining driving ability in people with cognitive impairment was unsatisfactory. The DVLA addressed this by developing a medical questionnaire specifically for cognitive impairment and dementia (Driver and Vehicle Licensing Agency 2009), but this still relies heavily on psychiatrists' clinical opinions, requiring yes/no answers to statements such as ‘Does your patient lack insight and/or judgement to a degree that would make driving dangerous?’

All drivers are required to inform the DVLA of any diagnosis or disability which may affect their driving, and it is prudent of doctors to advise patients of this. With dementia, however, a definitive diagnosis may not be made until the illness is well established, especially if the patient does not recognise the symptoms, does not have insight or is in a state of denial. Once a diagnosis is made, the same factors may still be relevant, as the patient may dispute the diagnosis or may not understand or recall the advice (British Psychological Society 2001). The General Medical Council (GMC; 2009) provide specific guidance on confidentiality and the reasons to break this if the risks and interests of public safety outweigh the interests of the patient (Box 2).

Driving restrictions

Reference Dubinsky, Stein and LyonsDubinsky et al (2000) suggested that driving limitations could allow drivers with dementia to reduce their crash risk, but noted that restrictions to licences such as those suggested for young drivers may be inappropriate, as the older group's skills are expected to degenerate over time, whereas younger drivers' skills should improve with experience. Limited licences have also been rejected by Reference Man-Son-Hing, Marshall and MolnarMan-Son-Hing et al (2007), who suggest that drivers with dementia lack the insight to understand the rationale for the restrictions and would therefore be unlikely to adhere to them.

An alternative view was offered by Reference Marshall and GilbertMarshall & Gilbert (1999) who looked at physicians' attitudes and knowledge regarding assessment of fitness to drive. As previously noted, physicians felt that the need to drive was greater for rural residents than urban dwellers, and that the doctor–patient relationship was negatively affected by reporting unsafe drivers to the licensing authority. Consequently, 85.5% of physicians felt that restricted licences were a fair compromise between personal independence and public safety, and 60.3% felt that the availability of limited licences would positively influence their decision to report at-risk drivers.

On-road testing

The gold standard of fitness to drive is an on-road assessment, but this is not a readily available test. There are only 17 mobility test centres across the UK, and as well as providing driving assessments to people with dementia, they offer assessments to people of all ages with medical conditions or disabilities affecting driving skills. They also assess and advise carers who need to transport individuals with restricted mobility in their vehicles and make recommendations on vehicle modifications (see www.mobility-centres.org.uk).

Clearly, the centres do not have the capacity to offer even single tests to everyone with dementia, let alone repeated tests as the condition deteriorates. Even if on-road tests were readily available, the cost of repeated tests would be untenable, requiring reams of paperwork, which would be difficult to coordinate with 6-monthly follow-ups in memory clinics scattered around the region.

Regular testing may also be detrimental to the patient with early dementia, as the added stress and anxiety of getting to the centre (which could be up to 50 miles away) and being tested, could lead to false failures. For this reason, much research has focused on finding a simple test or battery of tests which can be carried out in memory clinics.

In-office testing

Researchers have been trying to develop batteries of so-called ‘in-office tests’ to better evaluate driving safety and ability in dementia, bringing together more specific psychometrics and effective probing collateral reports.

Reference Molnar, Patel and MarshallMolnar et al (2006) completed a systematic review of simple in-office cognitive tests which claimed to differentiate drivers with dementia as ‘safe’ or ‘unsafe’. They found that commonly recommended tests such as the MMSE or Trail Making Test B did not demonstrate robustly positive findings across studies and that the Clock Drawing Test was not evaluated in any study. Cut-off scores were only reported in one study, making any meaningful comparison of studies impossible and preventing the creation of evidence-based guidelines to allow clinicians to identify patients as safe or unsafe to drive. Molnar et al explored this further and suggested alternative solutions. The most commonly employed technique currently is stratification, where rather than a dichotomous pass/fail outcome, there is a third ‘uncertain’ or ‘indeterminate’ outcome, which can suggest the need for further assessment such as the on-road driving test.

Current tests that have been developed

Dementia Drivers Screening Assessment (DDSA)

The DDSA was published in 2006, following work by Reference RadfordRadford in 2001. It looked at nine cognitive assessments covering a range of cognitive domains likely to affect driving ability, such as attention, memory, executive functioning, reasoning and visuospatial skills. When validated against an on-road driving test, no single test was able to differentiate between safe and unsafe drivers, consistent with the finding of Reference Molnar, Patel and MarshallMolnar et al (2006). Further analysis identified a combination of six tests which correctly classified 92% of drivers with dementia as safe or unsafe.

Reference Lincoln, Radford and LeeLincoln et al (2006) noted that ex-professional drivers were likely to be identified as unsafe on the cognitive assessments, but safe on the road, whereas patients with spatial problems performed poorly on the on-road test, but were not identified by the cognitive assessments. A further study by Reference Lincoln, Taylor and VellaLincoln et al (2009) looked at the accuracy of the DDSA. The authors found that the assessment battery correctly classified 76.2% of participants as safe or unsafe to drive, and recommended that it be used to stratify patients into safe, unsafe and a third ‘uncertain’ group who might benefit from further testing such as an on-road assessment.

Rookwood Driving Battery (RDB)

The RDB was proposed in Reference McKenna1998 by McKenna, who noted that traditional psychometric tests used to assess fitness to drive were often IQ-related and bore little relation to the cognitive systems specifically involved in driving behaviour. McKenna aimed to construct a flexible test battery incorporating specific tests for ‘each identifiable function underlying movement and analysis of the visual world as they relate to driving’ and ‘a method of measuring the integrity of cognitive functioning as it relates to driving’.

The RDB was launched in 2004 (Reference McKenna, Jefferies and DobsonMcKenna 2004) and it comprised a battery of 12 tests. In a comparison with an on-road driving test the RDB had 92% accuracy in predicting those with brain injury or pathology who would fail the on-road test and a 91% predictive accuracy of those who would pass. The test was much less accurate for those aged 70 and above, with only 85% accuracy for predicting an on-road fail and 37% for a pass.

A further study (Reference Rees, McKenna and BellRees 2008) collected normative older adult data to enable more appropriate use of the battery with older participants. Significant differences were found in performance on the RDB for people over 70 years suggesting that the pass/fail scoring system developed for younger adults would give cognitively intact older adults a higher baseline error score. Rees and colleagues suggested several cut-offs for pass/fail, but proposed that a clear fail on the battery still predicts a high fail rate on the road, making the RDB a useful screening tool. As with the DDSA, the RDB allows for an indeterminate outcome requiring further assessment.