Search results
The database search yielded 240 articles that were selected for further review based on title. The abstract of each article was reviewed and the inclusion/exclusion criteria were checked. From these 240 articles, 70 studies met the inclusion criteria of the current review (7 HD, 32 PD, and 31 AD studies). The majority of the studies described on-road driving performances (n = 45), 21 studies involved driving simulation, and 51 articles investigated the relationship between cognitive performances and driving outcomes. A summary of the included literature and the methods that were used is given per group in Tables 1, 2, and 3. When applicable, we will use the driving model of Michon et al. [10]. According to this model, driving errors can be sorted in three categories: (a) strategic errors that occur before actual driving, such as route planning; (b) tactical errors consisting of errors in speed adaptations, changing lanes, and keeping distance; (c) operational errors such as incorrect responses to changing driving environments and vehicle control [11, 12]. An overview of the committed driving errors by patient group per category is given in Table 4.
Table 1 Study details of included studies on Huntington’s disease
Table 2 Study details of included studies on Parkinson’s disease
Table 3 Study details of included studies on Alzheimer’s disease
Table 4 Types of driving errors categorized by group
Driving and Huntington’s disease
Huntington’s disease (HD) is a hereditary neurodegenerative disorder characterized by choreatic movements, cognitive dysfunction, and psychiatric symptoms [13]. It is caused by a gene mutation located on chromosome 4 [14]. The mean age at onset is between 30 and 50 years, with a mean disease duration of 17–20 years [13]. The earliest cognitive symptoms are characterized by executive dysfunctions, such as difficulties in planning, cognitive inflexibility, and lack of awareness [13, 15]. The cognitive symptoms gradually worsen and eventually result in dementia. Due to the progressive nature of the disease, patients become more dependent in their daily life activities. With the onset of HD during midlife, a lot of patients rely on their car for work and social activities, so patients might find it difficult to decide when to stop driving. However, concern about driving safely is one of the first issues reported by HD patients (33.5%) and has been associated with motor, cognitive, and depressive symptoms [16, 17]. The influence of other psychiatric symptoms, such as aggression and impulsivity, has not yet been investigated.
Only seven studies were found that investigated driving in HD patients [16–22]. Four of these studies used formal driving assessments, either on-road or simulated, to investigate driving competence [18, 20–22]. Due to the limited amount of studies available on HD and driving, the studies that did not investigate driving with formal driving assessments but with questionnaires or retrospective data analyses are also discussed [16, 17, 19]. An observational study investigating the association between different disease aspects of HD with functional changes showed that motor functioning and the Stroop task, measuring cognitive flexibility and information processing, were significantly associated with driving safety [16]. Increased motor impairment was related to a lower likelihood of being able to drive safely as rated by a professional. This study did not include a formal driving assessment. During a semi-structured interview, 11 out of 16 HD participants reported changes in their driving behavior [17]. They reported lower reaction times, had concerns about their safety, and had difficulties multi-tasking. A study that investigated clinical predictors of driving by retrospective patient file reviews showed that cognitive impairment, especially a reduction of psychomotor speed and attention, is a strong risk factor for driving cessation in HD [19]. Increased motor impairments were also associated with not driving a car, but were not a risk factor affecting the decision to cease driving [19]. An early study investigating driving in HD with a driving simulator showed that HD patients committed errors on the operational and tactical level [18]. They were less accurate and had longer reaction times compared to controls [18]. HD patients also had higher error rates in signaling, steering, braking, maintaining speed, and accelerator use. They were more likely to be involved in accidents compared to healthy individuals (58 and 11% respectively) [18]. Still, most of the HD patients in this study continued driving after onset of the disease (53/73). In addition, half of the HD patients that still drive failed an on-road driving assessment [20]. This confirms a limited insight regarding their own driving skills and emphasizes the importance of early evaluation [23–25]. In one study, 14 of the 30 HD patients (47%) failed the on-road driving test [21]. HD patients committed most errors on the operational and tactical levels, including errors in lane positioning, speed adaptations, keeping distance, turning left, and lane changing [21]. They also made more errors in perception of road signs, reflecting errors on the strategic level. Selective attention and disease stage were highly correlated with on-road driving failure in manifest HD [21]. A combination of neuropsychological tasks measuring visual processing speed, visual scanning, and attentional shifting best predicted the pass/fail rate of an on-road driving assessment, instead of a model that also included motor functioning [20]. More recently, it has been reported that some neuropsychological assessments focusing on speed of processing, cognitive flexibility, and visual attentional control seem to be good predictors for driving competence in manifest HD [22].
The results of the reviewed studies showed that driving competence is impaired in patients with HD and that concerns about driving safely are one of the earliest symptoms reported by both patients and families. Especially executive functioning and visuospatial abilities have been related to driving competence in HD. However, due to the limited amount of data, no conclusions can be drawn regarding which cognitive battery is most predictive of driving impairment in HD. None of the studies to date have focused on evaluating driving competency in the earliest stages of HD or in gene mutation carriers without a clinical diagnosis (i.e., premanifest gene carriers), while they often have questions for their physician regarding their driving skills and are most likely in need of a driving evaluation in the near future. Furthermore, no longitudinal studies have been performed investigating driving in HD, so there are no results available about the potential decline in driving competence during the course of the disease. Follow-up measurements are important to determine when driving-related issues become apparent and when to discuss potential driving cessation. It also provides an opportunity to monitor driving from early to more advanced stages of the disease.
Driving and Parkinson’s and Alzheimer’s disease
Contrary to driving studies in HD, a large number of studies have been performed evaluating driving competence in Parkinson’s disease (PD; n = 32) and Alzheimer’s disease (AD; n = 31). Three studies compared the driving competence of patients with PD and AD. In the following sections, we will discuss the on-road driving studies first, followed by the studies using driving simulators, and last the studies that also incorporated cognitive functioning in relation to driving performance.
Parkinson’s disease
Studies using on-road driving assessments (n = 22) to evaluate driving competence showed that 12–56% of the PD patients failed an on-road driving test [1, 26–34]. PD patients had a higher number of total driving safety errors compared to control participants. Studies that focused on identifying specific driving errors showed that PD patients are most likely to make errors on a tactical level including difficulties with yielding at intersections [29] and lane changing [1]. They were less likely to check their blind spot, and used their rear view and side mirrors less frequently than controls [1, 35]. Patients with PD also showed a decreased awareness of others, hesitated longer before making a turn, did not accelerate to a proper speed, and were less concentrated [26]. They made more errors in adjusting to different driving situations compared to controls [29] and showed difficulties driving in traffic flow [3]. PD patients made more errors in reversing and car parking [1]. Drivers with PD also had more difficulties with road positioning and driving on roundabouts compared to controls [33]. Most of the errors were present while driving in an urban environment [3]. Errors in the lateral position on the road at low speed and turning left [3] were the best predictors of overall pass/fail driving outcome [32]. Overall, PD patients had an unsteady car speed and tended to drive slower [35–37], especially during distraction [38]. However, it has also been reported that they drove faster on highways compared to controls [37], and had more difficulties adapting their speed at a higher speed [32]. They also identified fewer traffic signs and landmarks compared to controls [39].
On the operational level, PD patients made more incorrect turns and did not signal appropriately compared to controls [26, 35, 36]. They also made more errors in lane maintenance [1, 29, 40]. Strategically, PD patients made fewer driving trips [37, 41], drove less distance, and shorter durations [1, 41] compared to controls. PD patients had a higher preference for driving with a passenger [1, 37], which reported less nighttime driving [29, 37] and more often used alternative transportation [29]. Driving simulator studies (n = 12) showed that patients with PD had lower reaction times [42, 43], missed more red lights, and showed impaired accuracy compared to control subjects [42]. Furthermore, they had a higher number of traffic offences [43], more accidents [43, 44], and a worse overall simulator score compared to controls [43]. Patients who passed an on-road driving assessment also performed better on the simulator tests compared to patients who failed the on-road assessment [31]. Patients with PD tended to drive faster than controls and had poorer vehicle control, especially during low contrast visibility conditions [45]. PD patients were found to brake later during incongruent driving conditions [46]. They waited for external cues before they responded, while control subjects initiated a response prior to the cue. This result is similar to another study which found that PD patients relied more on external than internal cues to regulate their driving behavior [47].
A number of studies have incorporated cognitive assessments in an attempt to determine which test performances are associated with the driving competence of patients with PD. Most studies reported an association between cognitive functioning and driving competence [3, 12, 26–28, 31, 32, 36, 38–40, 43, 46, 48–52]. However, some studies also reported no associations between cognition and driving in PD patients [1, 33, 53], so results are inconsistent. Driving errors were particularly associated with lower performances in cognitive flexibility [26, 27, 38, 39, 49, 52], visuoconstructional abilities [26, 36, 39], attention [12, 27, 32, 36, 40, 46], psychomotor speed [46, 51], working memory [12, 49], set shifting [12, 48], information processing [12, 49], contrast sensitivity [27, 31, 43, 48, 51], visual scanning [32], visual acuity [32, 40], speed of visual processing [3, 27, 28, 40], and visual memory [3, 36].
Alzheimer’s disease
Twenty-three studies were included in this review that investigated driving competence in AD using on-road driving tests. Between 15 and 65% of the AD patients failed an on-road driving assessment [54–64]. They had lower overall driving performance scores compared to controls and committed more overall driving errors [62, 65–71], even in situations that were not considered challenging [54]. Driving performance scores tended to decrease with increasing dementia [57, 63, 72]. The largest decline in driving performance was reported in mild AD patients [57].
On a tactical level, AD patients committed more errors compared to controls in lane positioning [54, 67, 73], lane changing [57, 74], and checking their blind spot [74], and they tended to drive slower [68, 75]. They also had a higher inability to stop the vehicle appropriately [54, 76], and more difficulties avoiding potential collisions compared to controls [76]. Errors in turning [54, 70, 73, 75, 77], signaling [57, 74], and lane maintenance [54, 67, 73] were the most reported errors on the operational level. In contrast, some studies showed no differences between AD patients and healthy individuals in vehicle control [54, 70]. Strategic errors included less attention while driving, slower decision-making, and difficulties with road rules compared to controls [54]. AD patients also had more planning difficulties [75], identified fewer landmarks and traffic signs compared to controls [71], and showed more problems with route following [70].
Comparing driving competence of patients with PD and AD using on-road driving assessments showed that both patient groups committed more overall driving errors compared to controls [73]. These driving errors increased when a concurrent task was included [73]. There are also differences reported between both groups in types of driving errors [74]. Both AD and PD patients committed most errors on the tactical level, but patients with AD also made errors on the operational and strategic levels. Patients with PD committed relatively few operational and strategic errors compared to AD patients [74]. AD patients reported fewer driving trips and drove less miles compared to patients with PD and controls [62, 74]. Contrary, minimal differences between both groups have also been reported [53, 73].
The nine simulator studies reviewed showed that AD patients committed more errors in lane keeping (i.e., more lane deviations) [64, 78–81], turning left [78], and vehicle control [80] compared to controls. AD patients also tended to drive slower [64, 78, 80], took longer to complete the driving tests [78, 79], had less brake pressure [78], and made more judgmental errors (e.g., accidents, collisions) [80]. They failed to stop at traffic lights [80, 81] and exceeded the speed limit more often than controls [81]. Six out of eighteen AD patients crashed during a simulator test [82]. Cognitive and visual tests were predictive of the number of crashes [81–83]. Contrary, no differences in number of crashes between AD patients and controls have also been reported [83]. AD patients performed best when single, simple auditory-only driving navigation instructions were used compared to visual plus audio or visual-only instructions [84].
Drivers with increased cognitive impairments were more likely to be unsafe drivers compared to control subjects [74]. AD patients who failed an on-road assessment performed worse on neuropsychological tasks compared to AD patients who passed the on-road test [64]. Decreased performances on cognitive tests measuring speed of processing [62, 67, 73, 85], executive functioning [56, 74], attention [56, 70–72, 76], memory [67, 68, 70, 71, 73, 76], set shifting [62, 71, 73], visuoconstructional and visuospatial abilities [56, 67, 70, 71, 73, 74, 76], visual searching [56, 67, 72], and visual tracking [68] have been associated with worse scores on driving outcome variables and increased error rates in patients with AD. A composite battery of tests was more predictive of driving than individual tests [60, 67], and cognitive performance was more predictive of driving ability than AD diagnosis alone [85]. However, no correlations between neuropsychological outcome measures and on-road evaluations have also been reported [58, 77].
Self-assessment of driving performances
In addition to differences in driving performances, there are also differences reported in the evaluation of driving ability performed by patients, caregivers, and physicians. One study reported that PD patients rated their own driving performances lower than controls [50]. Contrary results showed that about 20% of the PD and AD patients misjudged their own driving ability [3, 43]. In addition, the rating performed by a neurologist (M = 8.0) was more optimistic compared to the rating performed by a driving instructor (M = 5.1) and psychologist (M = 5.7) [3]. Spouses tended to overestimate the driving ability of AD patients [86]. Ratings performed by an adult child were more related to driving outcome variables than ratings performed by spouses [86]. Self-ratings of driving ability performed by AD patients and ratings by spouses were significantly higher than ratings by an independent evaluator and physician [65, 87]. Ratings by a clinician were poorly associated with an on-road driving test, but not with naturalistic driving [86]. However, these clinician ratings were still more associated with driving performance compared to the self-evaluation by patients and the evaluation by spouses [65]. Caregivers did acknowledge general problems with driving, but still rated the AD patients driving competence significantly higher than an independent rater [87].
Driving simulator use
Since on-road driving assessments in patients with neurodegenerative disorders might be unsafe, an alternative is to evaluate driving competence with a simulator. Driving simulators provide the opportunity to present challenging situations and events in a standardized setting, with a high reproducibility compared to on-road driving assessments where situations cannot be manipulated [88]. Simulators are also used to train novice drivers before they start their on-road driving lessons [89]. Results of a concurrent and discriminant validity study comparing an on-road driving assessment with driving simulator tasks revealed that a driving simulator is a valid measure of driving performance for research purposes [90]. The driving simulator outcomes were able to discriminate between drivers with different levels of experience. In a study with elderly drivers, over 65% of the variability in the on-road assessments could be explained by driving simulator outcomes [91]. Adding a driving simulator increased the total variance explained by a potential screening battery to 60 and 94% [31, 43], suggesting that a driving simulator might be a useful screening tool to evaluate driving fitness. Studies that described the use of simulators for rehabilitation and training purposes in various disorders showed promising results, with more patients passing an on-road assessment after training with a simulator [92]. The lower ecological validity of a simulator, however, could be a disadvantage, because participants may prefer driving a real vehicle. It is also important to keep in mind that a reduction of driving performance measured with a simulator might reflect the adaption to the simulator itself and not actual driving ability. Therefore, it is necessary to further investigate the differences between disease groups and healthy individuals to minimize the effects of simulator use. In addition, the relationship between on-road performances and simulator driving should be further explored to determine whether simulator outcome measures are, indeed, consistent with on-road driving performance.
A common issue in simulator research is the existence of simulator sickness, which is comparable to motion sickness [93, 94]. It includes dizziness, nausea, vomiting, and sweating. The symptoms of simulator sickness are typically less severe than motion sickness and tend to decrease with multiple exposure and time [94, 95]. Dropouts in simulator studies have been related to simulator sickness, with up to one-third of the participants experiencing signs of simulator sickness [64, 84, 91]. The duration and configuration of the driving scenario influence this dropout rate [96]. For example, scenarios including more turns and sudden stops increase the risk for simulator sickness. Older age, female gender, and prior history of motion sickness have also been associated with higher susceptibility of experiencing simulator sickness [97, 98]. However, dropouts are not necessarily those subjects with the poorest performances [98, 99]. Several theories have been proposed to explain the occurrence of simulator sickness [94]. A conflict between structures within the sensory and vestibular systems has been the most widely excepted theory [94, 100]. When using a simulator to evaluate driving competence, this side-effect should be taken into consideration by excluding patients who experience simulator sickness or by screening beforehand. However, this might result in selection bias that should be accounted for. For more information regarding the topic of simulator sickness, we refer to the systematic review by Classen et al. [97].