Content uploaded by Kaarin J. Anstey
Author content
All content in this area was uploaded by Kaarin J. Anstey on Oct 15, 2020
Content may be subject to copyright.
Content uploaded by Kaarin J. Anstey
Author content
All content in this area was uploaded by Kaarin J. Anstey on Oct 15, 2020
Content may be subject to copyright.
Original Study
On-Road Behavior in Older Drivers With Mild Cognitive Impairment
Ranmalee Eramudugolla DPsych
a
,
b
, Md Hamidul Huque PhD
a
,
b
, Joanne Wood PhD
c
,
Kaarin J. Anstey PhD
a
,
b
,
*
a
School of Psychology, University of New South Wales, Randwick, NSW, Australia
b
Neuroscience Research Australia (NeuRA), Randwick, NSW, Australia
c
Queensland University of Technology (QUT), Centre for Vision and Eye Research, Institute of Health and Biomedical Innovation, Brisbane, QLD,
Australia
Keywords:
Mild cognitive impairment
fitness to drive
older driver
cognitive decline
abstract
Objectives: Dementia increases the risk of unsafe driving, but this is less apparent in preclinical stages
such as mild cognitive impairment (MCI). There is, however, limited detailed data on the patterns of
driving errors associated with MCI. Here, we examined whether drivers with MCI exhibited different on-
road error profiles compared with cognitively normal (CN) older drivers.
Design: Observational.
Setting and Participants: A total of 296 licensed older drivers [mean age 75.5 (SD ¼6.2) years, 120 (40.5%)
women] recruited from the community.
Method: Participants completed a health and driving history survey, a neuropsychological test battery,
and an on-road driving assessment including driver-instructed and self-navigation components. Driving
assessors were blind to participant cognitive status. Participants were categorized as safe or unsafe based
on a validated on-road safety scale, and as having MCI based on International Working Group diagnostic
criteria. Proportion of errors incurred as a function of error type and traffic context were compared across
safe and unsafe MCI and CN drivers.
Results: Compared with safe CN drivers (n ¼225), safe MCI drivers (n ¼45) showed a similar pattern of
errors in different traffic contexts. Compared with safe CN drivers, unsafe CN drivers (n ¼17) were more
likely to make errors in observation, speed control, lane position, and approach, and at stop/give-way
signs, lane changes, and curved driving. Unsafe MCI drivers (n ¼9) had additional difficulties at in-
tersections, roundabouts, parking, straight driving, and under self-navigation conditions. A higher pro-
portion of unsafe MCI drivers had multidomain subtype [n ¼6 (67%)] than safe MCI drivers [n ¼11
(25%)], odds ratio 6.2 (95% confidence interval, 1.4e29.6).
Conclusion and Implications: Among safe drivers, MCI and CN drivers exhibit similar on-road error pro-
files, suggesting driver restrictions based on MCI status alone are unwarranted. However, formal eval-
uation is recommended in such cases, as there is evidence drivers with multiple domains of cognitive
impairment may require additional interventions to support safe driving.
Ó2020 AMDA eThe Society for Post-Acute and Long-Term Care Medicine.
Cognitive impairment is a key risk factor for unsafe driving and
crashes in older adults,
1,2
and drivers with dementia have approxi-
mately 10 times the risk of failing an on-road driving test relative to
healthy older drivers.
2,3
The progressive nature of dementias, how-
ever, means that it is not always clear when and how the transition to
unsafe driving occurs. In fact, the evidence for driving impairment in
preclinical and early stages of dementia are mixed, with some studies
reporting a comparable range of performance to healthy drivers,
4,5
whereas others report higher fail rates, although less than that for
dementia.
2,3
A similar pattern of findings is evident for performance
on driving simulators, with reports of significant impairment in
drivers with dementia,
6e8
but little or no impairment in drivers with
mild cognitive impairment (MCI).
6,8e10
The variability in findings may
in part reflect the use of different diagnostic criteria across studies, but
also the heterogeneity in level of impairment and function in this
group. Furthermore, MCI increases the risk of progression to demen-
tia, but the annual conversion rate is only approximately 5% to 10%,
with most remaining stable or reverting to age-normative cognition.
11
The authors declare no conflicts of interest.
This study was funded by the National Health and Medical Research Council of
Australia (NHMRC grant #1045024). Kaarin Anstey is funded by NHMRC Research
Fellowship #1102694.
* Address correspondence to Kaarin J. Anstey, PhD, School of Psychology, Uni-
versity of New South Wales, Randwick, 2031 NSW, Australia.
E-mail address: k.anstey@unsw.edu.au (K.J. Anstey).
https://doi.org/10.1016/j.jamda.2020.05.046
1525-8610/Ó2020 AMDA eThe Society for Post-Acute and Long-Term Care Medicine.
JAMDA
journal homepage: www.jamda.com
JAMDA xxx (2020) 1e7
This group is therefore an important target for the development of
interventions and tools to support driving safety and independence.
To achieve this, detailed data on the type of driving behaviors and
difficulties experienced by drivers with MCI is needed, particularly for
driving under on-road, in-traffic conditions.
Few studies have examined the pattern of driving difficulties in MCI,
with most data derived from simulated driving tests.
6e9,12
On-road
errors and crash situations typically associated with increasing age in
healthy drivers include negotiating complex intersections, turning
against traffic, maintaining appropriate speed, and the ability to
maintain vehicle position within a trafficlane.
13e16
Relative to healthy
older drivers, simulator data indicate MCI is associated with more errors
in lane positioning,
8,12
maintaining appropriate speed,
8
delayed
breaking speed,
6
and errors at stop-signecontrolled intersections.
9,12
One study using on-road assessment also reported greater errors in
lane positioning and executing turns against traffic in drivers with MCI
relative to healthy older drivers.
4
Although it is difficult to compare
across studies, the broad pattern of errors associated with MCI appears
similar to that reported for healthy older adults, albeit more severe.
The broad categories of MCI and dementia in the preceding studies
do not distinguish between different etiologies and patterns of neu-
rocognitive impairment. However, emerging data indicate this is an
important factor in on-road safety.
17
Despite variations in on-road
assessment across different jurisdictions, the reported fail rate for
healthy older drivers typically ranges from 0% to 11%,
2,3,18
whereas
that for early or mild Alzheimer disease ranges between 10% and
50%,
2,3
and for vascular dementia, fronto-temporal dementia, and
dementia with Lewy-bodies the fail rate is between 35% and 70%.
17e19
Consistent with this, 1 simulator study reported that drivers with
single-domain amnestic-type MCI showed less impairment than those
with multidomain amnestic-type MCI.
12
Studies that have examined
cognitive skills associated with on-road driving in healthy older adults
also suggest a pattern in which reduced skills in selective attention,
task switching, response inhibition, and visual perceptual abilities are
the most predictive of driving errors.
13,20e22
A recent study found a
similar range of overall driving safety levels among older drivers who
were cognitively healthy or had MCI,
5
but this study did not examine
whether specific driving skills or contexts were more affected by MCI.
Thus, there is some evidence that drivers with MCI may be hetero-
geneous with respect to on-road safety,
4e6,8e10
but data on the nature
of MCI impacts on-road driving skills is still lacking. Given the po-
tential for this group to benefit from tailored interventions, this article
seeks to compare the pattern of errors incurred during a standardized
on-road driving assessment between safe and unsafe older drivers
who were identified as either cognitively healthy or MCI. We hy-
pothesized that MCI as a group would be associated with a greater
number of errors, but a similar pattern of errors, to cognitively normal
(CN) older drivers.
Methods
Participants
Participants were involved in the Driving Ageing Safety and Health
(DASH) study
5
that was undertaken in the Australian city of Canberra
from 2013 to 2016. Drivers aged 65 years and older with a current
valid license were recruited from the community through advertise-
ments in newspapers, community groups, and primary care providers,
as well as those referred to the Australian Capital Territory Health
Disability and Rehabilitation Service due to concerns about driving
safety. For the purposes of this study, data from participants with
known diagnoses of dementia or MMSE scores <23 were excluded
(Figure 1).
Standard Protocol Approvals and Consents
The study protocol was approved by the University Human
Research Ethics Committees (2012/643), and informed, written con-
sent was obtained from all participants before study involvement.
Procedures
Following consent, participants completed a general health and
driving survey to obtain information on basic demographics, self-
reported health conditions, self-rated health,
23
subjective memory
concerns,
24
instrumental activities of daily living,
25
and driving
Fig. 1. Participant flow in the DASH study and selection for current sample.
R. Eramudugolla et al. / JAMDA xxx (2020) 1e72
history. Participants then completed a laboratory assessment
including vision, hearing, and balance tests, and a neurocognitive
battery. Within 3 months of the laboratory assessment, participants
took part in an on-road driving assessment with a driver qualified
occupational therapist on a standard urban route in a dual-brake
vehicle.
Cognitive Measures
A neurocognitive test battery designed to sample performance
from a range of cognitive domains was administered by a trained
research assistant. The test battery took approximately 30 to 45 mi-
nutes within a longer laboratory session that lasted up to 2 hours
including rest breaks. Learning and memory function was assessed
using the California Verbal Learning Test,
26
language was assessed
using verbal fluency (Controlled Oral Word Association Test
27
) and the
15-item Boston Naming Test,
28
visuospatial skills were assessed with
the Copy test from the Benton Visual Retention Test,
29
complex
attention was assessed with the Victoria Stroop Part D (colored dots)
and W (nonconflict words)
30
; and executive function was assessed
using the Victoria Stroop Test Part C (conflict color-words),
30
Digit
Span Backwards,
31
Trail Making Test Part B,
32,33
and the Game of Dice
Test.
34
All scores were converted to z-scores for each participant based
on published age, gender, and education level stratified normative
data. For each cognitive domain, the z-scores for the component tests
were averaged to create a domain z-score. Thus scores near 0.0
represent normal performance, above zero is better, and below zero is
worse than normal performance in units of standard deviation.
Classification of MCI
A psychometric approach to classification of MCI was taken to
categorize participants as either CN or MCI.
5
Briefly, participants were
classified as meeting International Working Group diagnostic criteria
for MCI
35
if they demonstrated (1) no evidence of dementia (ie, MMSE
>24 and no known diagnosis of dementia); (2) subjective memory
concern (score >24 on the Memory Complaints Questionnaire
24
); (3)
objective MCI (of between 1 to 2 SDs below published age and gender
stratified norms) on 1 or more of 5 neurocognitive domains: complex
attention, learning and memory, language, visuo-perceptual skills,
and executive function; and (4) preserved basic activities of daily
living or minimal impairment in complex instrumental activities of
daily living. To examine MCI by subtype, participants with MCI who
had amnestic impairment (ie, z-score less than 1.0 for Memory
domain) were classed as amnestic-type. A second category was
created for multidomain type, in which participants were categorized
as such if there were more than 1 domain with a z-score less
than 1.0.
On-Road Driving Assessment
Each participant undertook a 50-minute open-road assessment in
an automatic vehicle with dual-brake controls fitted. The route was
predetermined and located in an urban area and included a range of
controlled and uncontrolled intersections, highway driving, residen-
tial areas, shopping strips, school zones, car parks, stop/give-way in-
tersections, and roundabouts. The route was approximately 20 km and
was divided into a total of 164 possible locations or maneuvers to
assist with scoring. Driving performance was scored following a
standard, validated protocol.
1,13,36
Scoring was conducted by a driver
trained occupational therapist who sat in the rear passenger seat of
the vehicle and scored performance using a Driver Safety Rating (DSR)
on a scale of 1 to 10 (where higher scores indicated higher safety), as
well as recording turn-by-turn errors in observation, lane position,
blind-spot checking, indication, speed control, gap selection, and
approach. A driving instructor sat in the front passenger seat and
provided turn-by-turn directions for 80% of the route, and for the
remaining 20%, the participant was asked to self-navigate to the
nearest hospital using the available road signs. A pass/fail safety cutoff
was used that was previously validated for this DSR scale,
37
given that
this is the typical outcome of driver licensing decisions.
38
Unsafe
driving was identified by participants whose DSR ranged from 1 to 3,
demonstrating multiple serious errors requiring either intervention
by the driving instructor to prevent a crash, or errors that would lead
to failing a local licensing test. Some measures of on-road errors varied
depending on opportunity (eg, lane change, indication) and route
changes. To ensure comparability across participants, errors of each
type (eg, observation, lane position), and errors in each traffic context
(eg, roundabout, straight driving) were converted to proportion of
errors per opportunity as reported previously.
5
Statistical Analysis
Participants were categorized according to on-road safety (Safe:
DSR of 4e10; Unsafe: DSR of 1e3), and MCI status (ie, MCI or CN),
producing 4 levels: CN-safe, CN-unsafe, MCI-safe, and MCI-unsafe.
Demographics, self-reported health, and cognitive measures and on-
road performance were reported in numbers with proportions or
median with interquartile ranges, as appropriate, according to the
preceding 4 levels of on-road safety and MCI status. The traffic context
of errors and different error types across various levels of on-road
safety and MCI status were compared with the CN-safe category us-
ing multinomial logistic regression models. All models were adjusted
for age, gender, and years of education. All the analysis was conducted
using Stata version 16.0 (StataCorp, College Station, TX).
Results
Of the 296 participants, 242 were classified as CN and 54 as MCI.
Within these categories, 93% of CN and 83% of MCI were classed as safe
on the on-road assessment. Sample characteristics for each of the 4
groups are presented in Table 1. Drivers in the CN-unsafe group tended
to be older than the other groups, and those in the MCI-unsafe group
tended to report lower driving distance and frequency per week.
Physical health in the MCI-unsafe group was generally similar to that in
the CN-safe and other groups. In terms of cognition, MCI participants
had lower domain z-scores relative to the CN groups. Although MCI-safe
drivers had greater levels of impairment in attention, memory, and
language, MCI-unsafe drivers had greater levels of impairment in vi-
suospatial, executive function, and attention. MMSE scores were close
to ceiling for all groups, consistent with this test’slowsensitivitytomild
impairment, and in highly educated samples.
39
Unsurprisingly, unsafe
drivers incurred approximately twice as many errors as safe drivers.
Approximately 75% of drivers in each category rated their own on-road
performance as being comparable to a typical driver, whereas the self-
ratings among MCI-unsafe drivers included greater numbers reporting
excellent performance, suggestive of poor insight.
Traffic Contexts of Errors
Table 2 presents the regression coefficients for the association
between safety category and errors at different traffic contexts.
Overall, there was no statistically significant difference in errors in any
of the traffic contexts between CN-safe and MCI-safe drivers. Relative
to CN-safe drivers, both MCI-unsafe and CN-unsafe drivers were prone
to make more errors at stop/give-way signs, when executing a lane
change, and driving on curved roads. For CN drivers, unsafe driving
was also significantly associated with errors in school zones and at
traffic pedestrian crossing. In comparison, unsafe driving for MCI
drivers was significantly associated with errors at both traffic-light
R. Eramudugolla et al. / JAMDA xxx (2020) 1e73
and nonetraffic-light controlled intersections, roundabouts, parking,
and driving on straight roads when compared with CN-safe drivers
(Table 2,Supplementary Figure 1).
Error Types
Figure 2 displays the group (grouped by safety category and MCI
status) mean error proportions of each error type, for the driver-
instructed component (Figure 2A) and the self-navigation compo-
nent (Figure 2B). Unsafe CN drivers had the highest proportion of
errors in checking their blind-spot, for both driver-instructed and
self-navigation conditions (Figure 2C). Table 3 displays numerical
comparison to aid interpretation. Relative to CN-safe drivers,
MCI-safe drivers had a similar likelihood of making errors, except for
marginally higher brake/acceleration errors. Relative to CN-safe
drivers, CN-unsafe drivers were more likely to make errors (Table 3),
Table 1
Sample Characteristics by MCI and Driving Safety Category
CN (n ¼242) MCI (n ¼54)
Safe Unsafe Safe Unsafe
Sample size, n (%) 225 (93) 17 (7) 45 (83) 9 (17)
Demographics
Age, mdn (p25, p75) 75 (70,78) 83 (81,85) 75 (70,79) 75 (74,81)
Females, n (%) 92 (40.9) 6 (35.3) 17 (37.8) 5 (55.6)
Years of education, mdn (p25, p75) 16 (13,18) 17 (14,18) 16 (13,17) 16 (10,17)
Days driven weekly, mdn (p25, p75) 6 (5,7) 6 (3,6) 6 (5,7) 4 (3,7)
Distance driven weekly (km) 200 (100,250) 100 (50,200) 120 (74,200) 80 (40,100)
Years of driving experience 55 (51,60) 65 (60,67) 56 (50,59) 60 (50,66)
Self-reported health
SF-12 Physical Scale, mdn (p25, p75) 42 (35,50) 40 (37,46) 42 (25,48) 55 (52,56)
SF-12 Mental Scale, mdn (p25, p75) 59 (52,61) 54 (46,61) 57 (53,62) 53 (44,60)
Number of health conditions 4 (2,5) 5 (4,5) 3 (2,5) 4 (2,5)
Vision impairment, n (%) 35 (15.5) 1 (5.9) 6 (13.3) 1 (11.1)
Hearing impairment, n (%) 60 (26.7) 6 (35.3) 12 (26.4) 3 (33.3)
Arthritis, n (%) 122 (54.2) 9 (52.9) 19 (42.2) 4 (44.4)
Heart disease/Heart attack, n (%) 36 (16) 3 (17.6) 12 (26.7) 0
Diabetes, n (%) 19 (8.4) 4 (23.5) 3 (6.7) 1 (11.1)
Parkinson disease, n (%) 3 (1.3) 0 1 (2.2) 0
Stroke/transient ischemic attack, n (%) 26 (11.6) 2 (11.8) 6 (13.3) 1 (11.1)
Cognition, mdn (p25, p75)
MMSE score (0e30) 29 (28,30) 29 (27,30) 29 (29,30) 28 (28,30)
Complex attention z-score 0.25 (0.2,0.6) 0.27 (0.2,0.5) 0.84 (1.4,0.2) 0.41 (1.1,0.0)
Memory z-score 0.37 (0.3,0.9) 0.05 (1.0,1.0) 0.48 (1.0,0.2) 0.13 (0.3,0.1)
Language z-score 0.24 (0.1,0.6) 0.10 (0.3,0.9) 0.18 (0.8,0.2) 0.03 (1.0,0.3)
Visuospatial z-score 0.57 (0.3,0.7) 0.62 (0.4,0.7) 0.31 (2.0,0.7) 1.17 (2.0,0.3)
Executive function z-score 0.20 (0.2,0.5) 0.00 (0.4,0.4) 0.13 (0.4,0.1) 0.62 (1.0,0.3)
On-road errors, median (25th and 75th percentile)
Total number of errors 21 (12,31) 42 (33,52) 19 (12,28) 50 (38,59)
Total number of locations 161 (159,160) 156 (152,161) 161 (160,163) 159 (154,163)
Self-rated on-road performance, n (%)
Excellent 2 (1) 0 0 2 (22)
Above average 38 (17) 1 (6) 8 (18) 2 (22)
Average (typical driver) 168 (75) 13 (77) 33 (73) 3 (33)
Below average 14 (6) 3 (18) 4 (9) 2 (22)
Note: MMSE: Mini-mental Status Exam ehigher scores are better with scores above 24 indicating no dementia. Cognitive domain z-scores ehigher scores are better with
mean normal performance at 0.0 and negative values indicating below normal performance in standard deviation units.
Table 2
Association Between MCI-Driving Safety Status and Errors Occurring in Different Traffic Contexts Using Multinomial Logistic Regression Adjusting for Age, Gender, and Years of
Education
Traffic Contexts of Errors CN-safe CN-unsafe MCI-safe MCI-unsafe
(Ref) Coef (95% CI) PCoef (95% CI) PCoef (95% CI) P
Proportion of total errors - 10.34 (0.3 to 20.37) .04 1.42 (4.68 to 7.52) .65 12.45 (0.8 to 24.1) .04
Traffic-light controlled intersection - 2.52 (0.05 to 5.09) .05 0.39 (2.04 to 1.26) .64 5.46 (1.85 to 9.08) <.01
Nonetraffic-light controlled intersection - 2.33 (0.45 to 5.1) .10 0.55 (1.63 to 2.74) .62 5.38 (1.81 to 8.95) <.01
Stop give-way - 12.75 (7.24 to 18.26) <.001 1.38 (2.3 to 5.06) .46 16.96 (9.64 to 24.27) <.001
Roundabout - 2.46 (0.34 to 5.26) .08 0.4 (1.3 to 2.09) .65 5.46 (1.81 to 9.12) <.001
Lane change - 5.1 (2.23 to 7.98) <.01 0.15 (1.79 to 2.09) .88 5.44 (2.16 to 8.72) <.01
Merging - 1.21 (1.1 to 3.51) .31 0.6 (1.12 to 2.31) .50 -*-*
Curve driving 1 way - 4.76 (0.82 to 8.7) .02 1.54 (1.19 to 4.27) .27 11.01 (5.68 to 16.33) <.001
Curve driving dual-carriage way - 3.93 (1.95 to 5.92) <.001 1.18 (0.31 to 2.66) .12 4.6 (2.26 to 6.94) <.001
Straight driving one-way - 8.14 (0.37 to 16.66) .06 4.42 (1.11 to 9.94) .12 29.24 (15.38 to 43.11) <.001
Straight driving dual-carriageway - 2.54 (0.12 to 5.21) .06 0.18 (2.54 to 2.19) .88 5.01 (2.05 to 7.97) <.001
School zone - 4.94 (2.14 to 7.74) <.01 0.09 (1.99 to 2.17) .93 3.14 (0.18 to 6.47) .06
Traffic calmer/Pedestrian crossing - 5.71 (1.04 to 10.37) .02 0.19 (3.16 to 3.54) .91 4.29 (1.37 to 9.95) .14
Parking - 2.36 (0.77 to 5.49) .14 1.62 (4.53 to 1.29) .27 4.44 (1.06 to 7.82) .01
Pull in out - 1.79 (0.49 to 3.1) .01 0.38 (0.42 to 1.18) .35 -*-*
Reversing - 0.61 (0.65 to 1.87) .34 0.08 (0.64 to 0.8) .83 0.53 (2.2 to 1.14) .54
Turnaround maneuvers - 0.14 (1.36 to 1.08) .83 0.35 (1.09 to 0.39) .35 0.54 (3.03 to 1.95) .67
*Estimation not possible as all participants had error proportion of 0.0.
R. Eramudugolla et al. / JAMDA xxx (2020) 1e74
particularly under driver-instructed conditions: in observation, brake/
accelerator use, lane position, gap selection, and approach. Under self-
navigation conditions, CN-unsafe drivers were more likely to make
lane position and approach errors compared with CN-safe drivers.
Compared with CN-safe drivers, MCI-unsafe drivers were also
impaired on observation, brake/accelerator, lane position, and gap
selection under driver-instructed conditions. However, under self-
navigation conditions, MCI-unsafe drivers were more likely to make
errors in observation, brake/accelerator, lane position, and gap selec-
tion, when compared with CN-safe drivers, whereas CN-unsafe drivers
did not show this pattern.
Analysis of MCI Subtype and Safety
To examine the association between MCI subtypes and on-road
safety, the MCI group was further categorized in terms of whether
impairment (domain z-score less than 1.0) was evident in more
than 1 of the 5 domains (multidomain type), and whether amnestic
impairment was apparent (Memory domain z less than 1.0 )
(amnestic type). One participant was excluded due to insufficient
cognitive data. Of the 53 MCI cases analyzed, frequencies for the 4
MCI subtypes were as follows: amnestic single domain 8 (15.1%),
nonamnestic single domain 28 (53%), amnestic multidomain 4
(7.5%), and nonamnestic multidomain 13 (25%). Due to small
numbers in each of these subtypes, the sample was categorized into
the characteristics of interest: presence of any amnestic deficits, and
presence of deficits of any type across multiple domains. Here, 12
(22.6%) had amnestic impairment, and 17 (32.1%) had impairment
across multiple domains. Of the 44 MCI drivers classified as safe, 11
(25%) were amnestic, and 11 (25%) were multidomain, and of the 9
MCI drivers classified as unsafe, 1 (11%) was amnestic and 6 (67%)
were multidomain. Binary logistic regression was used to examine
amnestic category and multidomain category as predictors of unsafe
driving. There was no association with amnestic subtype [odds ratio
(OR) 0.33; 95% confidence interval (CI), 0.04e2.80; P¼.307], but
there was a higher odds of unsafe driving in those with multidomain
subtype (OR 6.32; 95% CI 1.35e29.62; P¼.019). This remained after
adjustment for age, gender, and education (OR 7.69; 95% CI,
1.2 1e48.77; P¼.03), although the CI remains large because of the
small sample size.
Discussion
We sought to characterize the on-road error profile of drivers
with MCI, and to identify unsafe on-road behaviors associated with
MCI. We found no differences between safe drivers with and without
MCI. In contrast, unsafe drivers with MCI demonstrated additional
difficulties at intersections, roundabouts, parking, and driving on
straight roads than unsafe CN drivers. Unsafe MCI drivers also
demonstrated additional errors during self-navigation. This pattern
of additional errors among unsafe drivers with MCI, but not CN-
unsafe drivers, suggests that the impact of MCI is mostly revealed
under cognitively demanding traffic contexts and driving conditions.
Our findings are consistent with simulator studies suggesting MCI is
associated with more errors in lane position,
8,12
speed control,
8
and
at intersections.
4,9,12
Fig. 2. (A) Mean proportion of errors to opportunity under driver-instructedconditions as a function of MCI edriver safety category and error type. (B) Mean proportion of errors to
opportunity under self-navigation conditions as a function of MCI edriver safety category and error type. (C) Mean proportion of errors to opportunity for errors in blind-spot
checking as a function of driving condition (self-navigation vs driver-instructed) and MCI-driver safety category.
R. Eramudugolla et al. / JAMDA xxx (2020) 1e75
We found that all of the drivers in our sample had the highest
proportion of errors to opportunity when required to reverse,
conduct turnaround maneuvers, and blind-spot checks. Errors in
checking the blind-spot were also more likely under self-navigation
conditions for all driver categories, suggesting that the increased
cognitive load during self-navigation compromised this activity, and
confirms prior findings.
13,40
We also found that MCI-unsafe drivers
had similar levels of performance to CN-safe drivers in school zones,
at pedestrian crossings, indicator use, and in their approach to
traffic situations. Despite a small MCI sample, it could be speculated
that some aspects of safe driving may be preserved in this group,
possibly relating to the low speeds typically adopted in these
situations.
Although most drivers with MCI were classified as safe (83%), those
who were unsafe were also more likelyto have impairment across more
than 1 cognitive domain, and, as a group, had poorer visuospatial,
attention, and executive function than memory and language. This
cognitive profile is consistent with the finding that unsafe drivers with
MCI had greater difficulties with spatially demanding tasks such as
parking, roundabouts, and intersections, as well as executively
demanding conditions such as self-navigation. Further studies are
needed on driving errors and MCI subtype to confirm these findings,
given the low prevalence of unsafe driving and multidomain MCI in our
sample. Prior studies have reported that visuospatial impairment, along
with attentional deficits, are predictive of crashes in simulated
driving.
22,41
Our findings are also consistent with evidence that drivers
with non-Alzheimer's dementia may present with earlier and more
severe driving impairment than those with predominantly amnestic
decline.
17
Importantly, for safe drivers, the presence of MCI alone is not
associated with a different on-road error profile when compared with
CN drivers. To date, most studies that have examined the impact of
MCI on driving have used Clinical Dementia Rating (CDR) of 0.5 as a
definition of MCI.
2,3,6,9,17
However, CDR is functionally defined,
whereas MCI incorporates neuropsychological as well as subjective
decline and functional impairment.
35
As a result, the 2 approaches
often classify different groups
42
: CDR 0.5 tends to have a higher
prevalence than MCI,
42,43
and captures a wider range of functional
impairment, including mild dementia,
42e44
which may account for the
mixed findings in the driving literature. MCI-safe drivers tended to
have impairments in only a single domain, which supports the notion
that driving deficits emerge with increasing burden of cognitive
impairment.
18
Future studies will need to examine the longitudinal
changes in cognitive profile as well as on-road driving errors to better
understand the progression from safe MCI to unsafe MCI and
dementia.
Limitations of the present study include the small sample of MCI
drivers who were categorized as unsafe, and the nonerandom sam-
pling method used to obtain the cohort. Strengths of the study include
the use of a standard open-road driving assessment with detailed
characterization of error types and inclusion of a self-navigation
component. This allowed for precise analysis of the specific areas
where drivers with MCI had difficulties as well areas where perfor-
mance was not affected. Further strengths include detailed neuro-
psychological testing and the application of established MCI criteria
with greater sensitivity than CDR 0.5.
Conclusions and Implications
We found no differences in the on-road driving errors of safe
drivers with and without MCI. Unsafe drivers with MCI had greater
difficulties under some conditions. Drivers with impairment across
multiple domains, particularly visuospatial and executive, may need
more tailored advice, support, and re-skilling on driving under
cognitively demanding conditions. In terms of implications for prac-
tice, our data confirm previous findings that suggest that driver re-
strictions are unwarranted based on MCI status alone, but that these
drivers may require formal driving evaluation.
Acknowledgments
We thank Ally Gunn, Stephanie Sabadas, Emily Wilford, Sidhant
Chopra, Lily O’Donoughue-Jenkins, and Elizabeth Parkes who assisted
with data collection; Jasmine Price and Dereck Crooke for on-road
assessments; Rebecca Lawrence and Morgan Laird for data cleaning;
and Ray Wondal for assistance with manuscript preparation. We are
grateful to the ACT Health Driver Assessment and Rehabilitation Ser-
vice who assisted with recruitment.
References
1. Anstey KJ, Wood J, Lord S, et al. Cognitive, sensory and physical factors enabling
driving safety in older adults. Clin Psychol Rev 2005;25:45e65.
2. Chee JN, Rapoport MJ, Molnar F, et al. Update on the risk of motor vehicle
collision or driving impairment with dementia: a collaborative international
systematic review and meta-analysis. Am J Geriatr Pschiatry 2017;25:
1376e1390.
3. Hird MA, Egeto P, Fischer CE, et al. A systematic review and meta-analysis of
on-road simulator and cognitive driving assessment in Alzheimer’s disease and
mild cognitive impairment. J Alzheimers Dis 2016;53:713e729.
Table 3
Association Between MCI-Driving Safety Status and Error Type During Assessment Using Multinomial Logistic Regression Adjusted for Age, Gender, and Years of Education
Error Types CN-safe CN-unsafe MCI-safe MCI-unsafe
(Ref) Coef (95% CI) PCoef (95% CI) PCoef (95% CI) P
Instructed driving component
Observation - 22.75 (10.15 to 35.36) <.001 1.17 (9.2 to 6.87) .78 27.71 (12.52 to 42.9) <.001
Blind-spot - 0.34 (1.66 to 2.33) .74 0.11 (1.31 to 1.09) .86 1.04 (4.07 to 1.98) .5
Indication - 10.65 (0.63 to 21.93) .06 3.7 (3.7 to 11.1) .33 9.25 (4.51 to 23.02) .19
Brake/accelerator - 30.17 (17.09 to 43.26) <.001 11.29 (0.68 to 21.89) .04 35.34 (20.5 to 50.19) <.001
Lane position - 45.54 (27.12 to 63.95) <.001 8.39 (6.15 to 22.93) .26 43.79 (24.67 to 62.91) <.001
Gap selection - 61.28 (11.13 to 111.42) .02 3.37 (40.37 to 33.63) .86 101.91 (46.71 to 157.11) <.001
Approach - 19.31 (8.82 to 29.79) <.001 3.15 (12.61 to 6.3) .51 12.02 (0.93 to 24.97) .07
Self-navigation component
Observation - 2.34 (4.22 to 8.89) .48 1.06 (3.09 to 5.21) .62 10.23 (2.84 to 17.62) .01
Blind-Spot - -*-*0.5 (1.46 to 2.46) .62 1.17 (3.4 to 1.06) .3
Indication - 1.89 (3.19 to 6.97) .47 0.71 (2.62 to 4.04) .68 0.05 (7.37 to 7.46) .99
Brake/accelerator - 5.62 (3.15 to 14.39) .21 1.34 (6.1 to 8.78) .72 16.51 (6.87 to 26.15) <.001
Lane position - 16.82 (7.72 to 25.92) <.001 5.06 (1.63 to 11.75) .14 18.55 (8.83 to 28.28) <.001
Gap selection - 4.09 (28.69 to 36.87) .81 1.87 (15.47 to 19.2) .83 25.76 (3.72 to 47.81) .02
Approach - 10.93 (1.51 to 20.36) .02 2.78 (4.07 to 9.63) .43 9.58 (1.59 to 20.75) .09
*Estimation not possible as all participants had error proportion of 1.0.
R. Eramudugolla et al. / JAMDA xxx (2020) 1e76
4. Wadley VG, Okonkwo O, Crowe M, et al. Mild cognitive impairment and
everyday function: an investigation of driving performance. J Geriatr Pschiatry
Nerol 2009;22:87e94.
5. Anstey KJ, Eramudugolla R, Chopra S, et al. Assessment of driving safety in older
adults with mild cognitive impairment. J Alzheimers Dis 2017;57:1197e1205.
6. Frittelli C, Borghetti D, Iudice G, et al. Effects of Alzheimer’s disease and mild
cognitive impairment on driving ability: a controlled clinical study by simu-
lated driving test. Int J Geriatr Psychiatry 2009;24:232e238.
7. Stinchcombe A, Paquet S, Yamin S, et al. Assessment of drivers with Alz-
heimer’s disease in high demand driving situations: coping with intersections
in a driving simulator. Geriatrics 2016;1:21.
8. Pavlou D, Papadimitriou E, Antoniou C, et al. Comparative assessment of the
behaviour of drivers with Mild Cognitive Impairment or Alzheimer’s disease in
different road and traffic conditions. Transportation Research Part F: Traffic
Psychology and Behaviour 2017;47:122e131.
9. Devlin A, McGillivray J, Charlton J, et al. Investigating driving behaviour of older
drivers with mild cognitive impairment using a portable driving simulator.
Accid Anal Prev 2012;49:300e307.
10. Kawano N, Iwamoto K, Ebe K, et al. Effects of mild cognitive impairment on
driving performance in older drivers. J Am Geriatr Soc 2012;60:1379e1381.
11. Mitchell AJ, Shiri-Feshki M. Rate of progression of mild cognitive impairment to
dementiaemetaeanalysis of 41 robust inception cohort studies. Acta Psychiatr
Scand 2009;119:252e265.
12. Hird MA, Vesely KA, Fischer CE, et al. Investigating simulated driving errors in
amnestic single-and multiple-domain mild cognitive impairment. J Alzheimers
Dis 2017;56:447e452.
13. Anstey KJ, Wood J. Chronological age and age-related cognitive deficits are
associated with an increase in multiple types of driving errors in late life.
Neuropsychology 2011;25:613.
14. Mayhew DR, Simpson HM, Ferguson SA. Collisions involving senior drivers:
high-risk conditions and locations. Traffic Inj Prev 2006;7:117e124.
15. Di Stefano M, Macdonald W. Assessment of older drivers: relationships among
on-road errors, medical conditions and test outcome. J Safety Res 2003;34:
415e429.
16. Boufous S, Finch C, Hayen A, et al. The impact of environmental, vehicle and
driver characteristics on injury severity in older drivers hospitalized as a result
of a traffic crash. J Safety Res 2008;39:65e72.
17. Toepper M, Falkenstein M. Driving fitness in different forms of dementia: an
update. J Am Geriatr Soc 2019;67:2186e2192.
18. Fuermaier A, Piersma D, de Waard D, et al. Driving difficulties among patients
with Alzheimer’s disease and other neurodegenerative disorders. J Alzheimers
Dis 2019;69:1019e1030.
19. Piersma D, Fuermaier AB, De Waard D, et al. Assessing fitness to drive in patients
with different types of dementia. Alzheimer Dis Assoc Disord 2018;32:70.
20. Jeong H-G, Jeong H, Kim TH, et al. Association between driving performance
and neuropsychological deficits in elderly people with mild cognitive impair-
ment. Alzheimers Dement 2012;8:P359.
21. Karthaus M, Wascher E, Falkenstein M, et al. The ability of young, middle-aged
and older drivers to inhibit visual and auditory distraction in a driving simu-
lator task. Transportation Research Part F: Traffic Psychology and Behaviour
2020;68:272e284.
22. Dawson JD, Uc EY, Anderson SW, et al. Neuropsychological predictors of
driving errors in older adults. J Am Geriatr Soc 2010;58:1090e1096.
23. Ware JE Jr, Kosinski M, Keller SD. A 12-Item Short-Form Health Survey: con-
struction of scales and preliminary tests of reliability and validity. Med Care
1996;34:220e233.
24. Crook TH, Feher EP, Larrabee GJ. Assessment of memory complaint in age-
associated memory impairment: the MAC-Q. Int Psychogeriatr 1992;4:
165e176.
25. Sonnega A, Faul JD, Ofstedal MB, et al. Cohort profile: the health and retirement
study (HRS). Int J Epidemiol 2014;43:576e585.
26. Delis DC, Massman PJ, Kaplan E, et al. Alternate form of the California Verbal
Learning Test: development and reliability. Clin Neuropsychol 1991;5:
154e162.
27. Borkowski JG, Benton AL, Spreen O. Word fluency and brain damage. Neuro-
psychologia 1967;5:135e140.
28. Mack WJ, Freed DM, Williams BW, et al. Boston Naming Test: shortened ver-
sions for use in Alzheimer’s disease. J Gerontol 1992;47:P154eP158.
29. Benton AL. Benton visual retention test - Revised. 4th ed. San Antonio, TX:
Psychological Corporation; 1983.
30. Strauss E, Sherman EM, Spreen O. A compendium of neuropsychological tests:
Administration, norms, and commentary. New York, NY: Oxford University
Press; 2006.
31. Wechsler D. Wechsler adult intelligence scaleeFourth Edition (WAISeIV). San
Antonio, TX: NCS Pearson; 2008. p. 816e827.
32. Reitan RM. Validity of the Trail Making Test as an indicator of organic brain
damage. Percept Mot Skills 1958;8:271e276.
33. Bowie CR, Harvey PD. Administration and interpretation of the Trail Making
Test. Nat Protoc 2006;1:2277.
34. Brand M, Pawlikowski M, Labudda K, et al. Do amnesic patients with Korsak-
off’s syndrome use feedback when making decisions under risky conditions?
An experimental investigation with the Game of Dice Task with and without
feedback. Brain Cogn 2009;69:279e290.
35. Winblad B, Palmer K, Kivipelto M, et al. Mild cognitive impairmentebeyond
controversies, towards a consensus: report of the International Working
Group on Mild Cognitive Impairment. J Intern Med 2004;256:240e246.
36. Mallon K, Wood JM. Occupational therapy assessment of open-road driving
performance: Validity of directed and self-directed navigational instructional
components. Am J Occup Ther 2004;58:279e286.
37. Anstey KJ, Wood J, Caldwell H, et al. Comparison of self-reported crashes, state
crash records and an on-road driving assessment in a population-based sample
of drivers aged 69e95 years. Traffic Inj Prev 2009;10:84e90.
38. Salmi LR, Leproust S, Helmer C, et al. Assessing fitness to drive in the elderly
and those with medical conditions: guidelines should specify methods and
evidence. Inj Prev 2014;20:210e212.
39. Mitchell AJ. A meta-analysis of the accuracy of the mini-mental state exami-
nation in the detection of dementia and mild cognitive impairment. J Psychiatr
Res 2009;43:411e431.
40. Wood JM, Anstey KJ, Lacherez PF, et al. The oneroad difficulties of older drivers
and their relationship with selfereported motor vehicle crashes. J Am Geriatr
Soc 2009;57:2062e2069.
41. Rizzo M, McGehee DV, Dawson JD, et al. Simulated car crashes at intersections
in drivers with Alzheimer disease. Alzheimer Dis Assoc Disord 2001;15:10e20.
42. Wada-Isoe K, Kikuchi T, Umeda-Kameyama Y, et al. Global Clinical Dementia
Rating score of 0.5 may not be an accurate criterion to identify individuals with
mild cognitive impairment. J Alzheimers Dis Reports 2019;3:233e239.
43. Ganguli M, Chang C-CH, Snitz BE, et al. Prevalence of mild cognitive impair-
ment by multiple classifications: The Monongahela-Youghiogheny Healthy
Aging Team (MYHAT) project. Am J Geriatr Pschiatry 2010;18:674e683.
44. Chang Y-L, Bondi M, McEvoy L, et al. Global clinical dementia rating of 0.5 in
MCI masks variability related to level of function. Neurology 2011;76:
652e659.
R. Eramudugolla et al. / JAMDA xxx (2020) 1e77
Appendix
Supplementary Fig. 1. Mean proportion of errors to opportunity as a function of traffic context and MCI-safety category.
R. Eramudugolla et al. / JAMDA xxx (2020) 1e77.e1
