Driving Errors in Persons with Dementia Peggy P. Barco, OTD,a Carolyn M. Baum, PhD,a Brian R. Ott, MD,b Steven Ice, MOT,c Ann Johnson,d Michael Wallendorf, PhD,e and David B. Carr, MDfg
OBJECTIVES: To differentiate driving errors in persons with dementia who fail a performance- based road test from errors in persons who pass. DESIGN: Cross-sectional. SETTING: Community. PARTICIPANTS: Active drivers diagnosed with dementia (n = 60) and older adult controls (n = 32). MEASUREMENT: All participants completed a standardized clinical and on-road driving assessment. The outcome variable was the number and types of driving errors according to the Record of Driving Errors (RODE), a standardized tool to record driving errors. RESULTS: Sixty-two percent (n = 37) of individuals with dementia and 3% (n = 1) of controls failed the road test. Based on the RODE, individuals with dementia made twice as many driving errors as healthy controls. Within the dementia sample, individuals who failed the road test had more difficulties driving straight and making left and right turns than during lane changes. Dangerous actions occurred most often while driving straight and making left turns. Specific driving behaviors associated with road test failure in the sample with dementia included difficulties in lane positioning and usage, stopping the vehicle appropriately, attention, decision-making, and following rules of the road. Informants of participants with dementia who failed the road test reported more impairment with cognitive functioning on the Assessing Dementia 8 Screening Interview (AD8). CONCLUSION: This report highlights the driving errors most common in people with dementia who fail a road test. The finding that most of the dangerous actions in the sample with dementia occurred while driving straight condition is novel. Driving on straight roads has not been considered a From the aProgram in Occupational Therapy, School of Medicine, Washington University, St. Louis, Missouri; bDepartment of Neurology, Warren Alpert Medical School, Brown University, Providence, Rhode Island; cIndependent Drivers, LLC; dCenter for Clinical Studies; eDivision of Biostatistics; and Departments of fMedicine and gNeurology, School of Medicine, Washington University, St. Louis, Missouri. Address correspondence to Peggy P. Barco, OTD, OTR/L, SCDCM, Washington University School of Medicine in St. Louis, Program in Occupational Therapy, 4444 Forest Park Blvd., St. Louis, MO 63108. E-mail: [email protected] DOI: 10.1111/jgs.13508
JAGS 63:1373–1380, 2015 © 2015, Copyright the Authors Journal compilation © 2015, The American Geriatrics Society
condition of “high challenge” in prior driving studies in individuals with dementia. This finding has potential implications for future interventions related to vehicle instrumentation and driving recommendations for people with dementia. J Am Geriatr Soc 63:1373–1380, 2015.
Key words: driving performance; dementia; older drivers; driving errors
T
he prevalence of Alzheimer’s disease (AD) is expected to increase from an estimated 4.7 million individuals in 2010 to 13.8 million in 2050.1 Although some experts recommend that all adults with a dementia diagnosis refrain from driving, most clinicians base the decision on dementia severity and impaired driving performance.2,3 Outpatient surveys indicate that approximately 30% of older adults with dementia are current drivers.4,5 As the ability to diagnose AD in the preclinical and early stages of disease (e.g., mild cognitive impairment) continues to improve, some individuals will retain the appropriate skill sets to maintain driving privileges for years after diagnosis,6–8 before eventual driving cessation. The methods most commonly used to measure driving ability in older adults include performance-based road tests, driving simulators, questionnaires, and documentation of motor vehicle crashes.9,10 Although costly, advanced in-car video technology in combination with naturalistic driving have also been explored in determining driving safety in older adults.11,12 Although driving simulators eliminate the safety hazards present in road tests, debate continues regarding whether simulators can predict actual driving performance.13,14 Driving questionnaires completed by family members have limitations because limited knowledge of the older adults’ driving abilities and the fear of increased family responsibility if the family member is unable to drive may bias results. When driving older adults complete questionnaires, they may not accurately reflect performance on the road because of insight deficits that can accompany dementia.15,16 Documentation of crashes does not always clearly indicate “at fault” relationships, and many smaller crashes go unreported.
0002-8614/15/$15.00
1374
BARCO ET AL.
The performance-based road test is the most widely accepted method for determining driving competence. The road test evaluates the older adult’s abilities while actually operating a motor vehicle in traffic2,7,17 and has been associated with a history of crashes.18–21 Limitations of the performance-based road test include lack of objectivity or interrater reliability, safety risk to the individual and community, differing road courses, variability in traffic conditions, lack of test–retest measurements (e.g., stability), and various scoring procedures.10,22 Traditionally, road test performance with older drivers has been evaluated according to global performance outcomes (e.g., pass, marginal, fail),23–25 although some recent studies have examined the number and type of driving errors made during road testing.12,18,21,26–29 Studies with individuals with AD and specific driving errors have indicated a variety of errors, with one of the more common errors being difficulty with lane observance and lane changes.18,27,28 Although these studies have found significant differences in number and type of driving errors between medical diagnoses (e.g. between AD or Parkinson’s disease and controls), less is known about the differences in driving errors between those within the diagnostic groups who pass and those who fail. The aims of this study were to determine the differences in driving errors between older adults with and without dementia and to compare in detail the number and types of driving errors that older adults with dementia who fail a road test and those who pass.
METHODS The Missouri Department of Transportation, Traffic and Highway Safety Division funded this study, which was conducted at Washington University Medical School in St. Louis, through the DRIVING Connections Clinic at the Rehabilitation Institute of St. Louis (TRISL). The human studies committee at Washington University approved the study.
Participants Individuals with dementia were recruited through the Memory Diagnostic Center of Washington University School of Medicine in St. Louis and Washington University Physicians (January 2008 to September 2009 and April to September 2012). Healthy older adults (controls) were recruited from the Volunteer for Health Data Base maintained by Washington University Medical School, local community centers, and word of mouth from January 2010 through June 2011. Inclusion criteria for the individuals with dementia was a primary diagnosis of dementia from the referring physician, an Assessing Dementia-8 Screening Interview (AD8)30 score of 2 or more by an informant, physician referral for a driving evaluation, and an informant available to accompany the participant to the driving evaluation. Participants for the healthy older adults (controls) were required to be aged 55 and older, have an informant willing to answer questionnaires, be cognitively intact as indicated by an AD8 score of less than 2 (rated by an informant), and have a Short Blessed Test score less than 9. Exclusion criteria for both groups
JULY 2015–VOL. 63, NO. 7
JAGS
included lack of an active driver’s license; visual acuity that did not meet the state guideline; non-English speaking; less than 10 years of driving experience; any major chronic unstable disease or condition (e.g., seizures); severe orthopedic, musculoskeletal, or neuromuscular impairments that would require adaptive equipment to drive; visual, hearing, or language impairments that would interfere with testing; sedating drugs (e.g., new use of narcotics or anxiolytics within the past month or chronic use that causes sedation); or having had a previous driving evaluation in the last 12 months. One hundred twenty-six of 202 individuals with dementia who were screened over the telephone met criteria and agreed to participate in this study. Reasons for nonparticipation included refusal or lack of interest in study (47), not currently driving or no license (16), concern about potential loss of license (7), prior driving evaluation within a year (4), no return call (3), diagnosis of dementia not confirmed by physician (2), not comfortable with the driving route in the city (2), and miscellaneous reasons (e.g., unexplained cancellation, no show) (9). A subset of the 126 participants (n = 60) was arbitrarily selected (based on availability of the second driving specialist to score errors and study funding limitations) to be in the driving error study. Forty-one of 54 healthy older adults who were screened over the telephone met inclusion and exclusion criteria and agreed to participate in the study. Reasons for nonparticipation included medical illness (2), stopped driving already (1) error scores were not performed because of unanticipated absence of scorer (2), transportation problems (1), unknown reason or other (3). The remaining 32 healthy elderly adults participated in the study.
Procedures Participants underwent a comprehensive driving evaluation that included a clinical assessment and an on-road evaluation. Because their physician referred the individuals with dementia for medically indicated driving assessments, a recommendation meeting (with the informant and participant) related to future driving capability was held at the conclusion of the evaluation process. A full driving evaluation report with recommendations was sent to the referring physician with permission from the driver and the informant. A trained occupational therapist (OT) or certified occupational therapy assistant (with supervision from the OT) administered the clinical portion of the driving assessment. This portion of the driving assessment included standardized tests of vision, cognitive, and motor function with procedures described previously.23 The cognitive screens and assessments included the AD8,30 Short Blessed Test31 Freund Clock Drawing Test,32 Trail-Making Test Parts A and B,33 Snellgrove Maze Test,34 and Driving Health Inventory—Useful Field of View and Visual Closure Tests.35 The performance-based road test, the Modified Washington University Road Test (mWURT), used common traffic situations and road maneuvers from the previous Washington University Road Test.25 The mWURT has been described in detail along with the approach for qualitative
JAGS
JULY 2015–VOL. 63, NO. 7
assessment used in the current study (pass, marginal, fail).23,36 The road test was initiated in a closed parking lot and then proceded to low traffic conditions (a park setting) and then moderate to high traffic conditions (with complex intersections, traffic lights, and more traffic) as safety considerations permitted. The course was 13 miles long and took approximately 1 hour to complete. Participants drove a midsize driving evaluation car (with dual brakes for safety); an OT Certified Driving Rehabilitation Specialist (OT/CDRS) sat in the front seat to provide directions and maintain safety. A second OT (OT/Driving Rehabilitation Specialist—OT/DRS) blinded to clinical evaluation data sat in the back seat to recorded driving errors. The road test was discontinued if both incar evaluators agreed that safety was in jeopardy. The point of discontinuation was recorded. A standardized quantitative rating of number and type of safety errors using The Record of Driving Errors (RODE)* (Figure 1) was based on error categorizations (driving situation and driving behavior errors) from the literature.21,29 Driving errors were first recorded according to traffic condition (closed parking lot, low traffic, moderate to high traffic). Next, within each traffic condition, errors were recorded according to the specific driving situation involved (parking or backing, left turn, right turn, straight driving, lane change). Finally, within each driving situation, various combinations of driving behaviors that resulted in the driving situation error (driver behavior errors) were categorized as operational, tactical, or information processing errors and recorded. Similar to other models, operational errors37 were those involving operation of the car (e.g., use of turn signals), tactical errors37 were those related to driving tactics and skills (e.g., maintaining lane position, visual scanning), and information processing errors were those related to higher processing skills (e.g., following directions, memory, decision-making, knowing rules of the road), and were similar to, but extended beyond, the strategic and planning driving errors described in previous studies.28,38 A portion of this road test included being able to find a destination and use strategic planning to locate, enter, park, and exit independently. Important to the RODE scoring protocol, one driving situation error could be the result of multiple driving behavior errors (e.g., operational, tactical, informational processing errors). For example, a driving situation error involving a lane change could include failure to scan, failure to yield to a vehicle, failure to signal, and failure to anticipate the need to change lanes yet would still be counted as “one” driving situation error. Finally, any necessary interventions to maintain safety were recorded (e.g., brake intervention). Errors resulting in a potentially dangerous action (an action that the driver made posed an immediate potential threat to safety, e.g., failure to yield to a vehicle resulting in the instructor needing to make an intervention to avoid a collision, stopping in mid-traffic on a busy road causing other drivers to maneuver quickly to avoid a collision) were also recorded. Interrater reliability on scoring of the RODE was found to be strong across *For more information about The Record of Driving Errors (RODE), Barco, 2007, contact Peggy P. Barco at Washington University School of Medicine in St. Louis—Program in Occupational Therapy.
DRIVING ERRORS IN PERSONS WITH DEMENTIA
1375
error types,36 and both in-car instructors rated driving performance qualitatively as pass, marginal, or fail using an objective format described previously for which kappa for interrater reliability was found to be high (j = 0.84).23 The qualitative ratings were collapsed into two categories (pass (pass and marginal) and fail), and the OT/CDRS rating was used for analysis.
Statistical Analyses Differences in demographic characteristics, cognitive test scores, and driving errors between healthy older adults and individuals with dementia were analyzed using t-tests for continuous variables and the Fisher exact test for categorical variables. Similar analyses to determine demographic and cognitive differences were subsequently performed between individuals with dementia who passed and those who failed the road test to explore the differences further. The comparisons between individuals with dementia who passed the road test and those who failed were made through analysis of covariance (age adjusted). Analysis was to determine whether there was a difference in the number of total driving situation errors made and number of specific driving situation errors in moderate to high traffic (right turn, left turn, straight, lane changes), as well as driving behavior errors (operational, tactical, information processing). Because many of the specific driving behavior errors within the operational, tactical, and information processing error categories occurred only once or twice per person (if they occurred at all), the specific driving behavior errors were dichotomized (0 vs >1) and analyzed using logistic regression.
RESULTS Sixty-two percent (n = 37) of individuals with dementia and 3% (n = 1) of older adults without dementia failed the road test. There were significant differences in age (P = .03), race (P = .002), and all areas of cognitive testing between individuals with dementia and controls (Table 1). Participants with dementia (n = 53) were found to make twice as many driving situation errors as controls (n = 32) (dementia group mean errors 12.0 7.5, control group mean errors 6.0 4.9; P < .001). Although controls made considerably fewer driving errors than participants with dementia, they made errors in all driving situations. Individuals with dementia made errors more often than controls when driving straight (dementia group mean 2.06 2.42, control group mean 1.16 1.51; P = .04) and when making right turns (dementia group mean 1.04 1.54, control group mean 0.38 0.83; P = .03). In contrast to the dementia group, none of the driving errors in the healthy control group involved dangerous actions or brake interventions. Errors in the cognitively intact control group were most common in (recorded as percentage of participants with at least one error) decreased visual scanning (59%), difficulty maintaining lane positioning or lane usage (56%), lack of use of turn signals (44%), limited knowledge of rules of the road (44%), and difficulty with memory or following instructions (41%). Thus, although the control group had driving errors, they had errors significantly less frequently
1376
BARCO ET AL.
JULY 2015–VOL. 63, NO. 7
JAGS
Level One: Traffic Condition Closed Parking Lot
Moderate High Traffic
Low Traffic
Level Two: Driving Situation Errors Parking Backing
Right Turn
Left Turn
Straight away Driving
Lane Change
Level Three: Driving Behavior Errors Operational: Use of controls, gears, signals, foot pedals, steering Tactical: Visual Scanning, Distance Judging, Lane Usage/position, stopping, speed, yielding Information processing: Anticipates, Attention, Decision Making, Following Directions/Memory, Rules of The Road, Speed of Processing
Driving Interventions Wheel/Brake Interventions Dangerous Actions
Figure 1. Record of Driving Errors (RODE) Classification Levels. Adapted with permission from reference 39. Copyright 2015 American Occupational Therapy Association.
than the dementia sample, and the errors were rarely significant enough that controls failed the road test. Subsequently, the focus was directed to understanding the differences in the driving errors within the dementia sample—between those with dementia that passed a road test compared to those who failed. Of the 60 participants with dementia, 16 (27%) were not allowed to continue the road test because of serious safety concerns that both in-car evaluators had. These discontinuations occurred at every level of traffic, with 12% (n = 7) being discontinued so early in the road test that they never made it into moderate to high traffic. These seven individuals were not found to be significantly different from those who failed at the more-challenging levels of traffic in age or AD8 score, although most standardized cognitive tests given during the clinical assessment showed them to be significantly more impaired. Only individuals with dementia who were able to progress to the moderate to high traffic condition (n = 53) were included in this portion of the study (because they had the potential to make errors at all levels of traffic). Within the dementia group, 43% (n = 23) passed the road test, and 57% (n = 30) failed. There were no significant differences between those who passed and those who failed in age, sex, education, race, or years of driving experience (Table 2). Those who failed the road test had higher AD8 scores (indicating greater cognitive impairment) than those who passed, with other areas of cognitive assessment showing no significant differences between the two groups (Table 2). Significant differences were observed in the number of total driving situation errors (P < .001) between those with dementia who passed the road test and those who
failed. Nearly all traffic situations (in moderate to high traffic) with the exception of making lane changes were noted to result in more errors in those who failed than those who passed, including making turns (P = .03) and left (P = .001) turns and driving straight (not making a left turn, right turn, or lane change) (P = .01) (Table 3). Dangerous actions occurred most often when driving straight (32% of participants ≥1 occurrences) and turning left (21% ≥1 occurrences) as opposed to turning right (2% ≥1 occurrences) and changing lanes (8% ≥1 occurrences). Driving behavior errors were also noted to be significantly different between individuals with dementia who passed and those who failed, including total tactical errors (P = .001) and total information processing errors (P = .001) but not total operational errors (Table 3). The tactical errors that differentiated those who failed the road test from those who passed were difficulty with lane positioning and usage (P = .01) and inability to stop the vehicle appropriately (P = .04). The information processing errors that reflected significant differences between the groups included difficulty with attention while driving (P = .001), decision-making during driving (P = .001), and knowing the rules of the road (P = .04) (Table 4).
DISCUSSION Similar to other studies examining older adults with dementia on performance-based road tests, participants with dementia had higher levels of impaired driving performance, resulting in more failed road tests and driving errors than healthy older adult controls.27,28 Participants
JAGS
JULY 2015–VOL. 63, NO. 7
DRIVING ERRORS IN PERSONS WITH DEMENTIA
Table 1. Demographic and Cognitive Characteristics of Older Adults with Dementia and Older Adult Controls (N = 92) Characteristic
Age, meanSD Male, n (%) Education, years, mean sd White, n (%) Driving experience, years, meanSD Assessing Dementia 8 Screening Interview score, mean sd (range 0–8) Short Blessed Test score, meanSD (range 0–28) Clock Drawing: Freund Score, mean sd (range 0–7) Trail-Making Test Part A, seconds, meanSD Trail-Making Test Part B, seconds, mean sd Visual closure (Motor-Free Visual Perception Test) errors, meanSD (range 0–11 errors) Useful field of vision, ms, mean sd (range 100–500 ms) Snellgrove Maze Test, seconds, meanSD
Dementia (n = 60)
Controls (n = 32)
74.7 8.5 40 (66.7) 15.3 3.4
70.7 8.1 16 (50.0) 14.9 3.0
.03 .17 .64
49 (86.0) 57.5 9.0
18 (56.3) 52.6 9.4
.004 .02
5.5 1.6
0.28 0.46
.001
P-Value
9.8 6.8
6.4 2.5
.001
4.7 2.2
6.4 0.91
.001
68.3 42.5
41.3 13.9
.001
191.3 83.1
109.4 48.7
.001
4.3 2.8
2.2 1.5
.001
330.4 168.9
183.8 109.5
.001
58.3 46.1
36.2 13.7
.009
1377
Table 2. Demographic and Cognitive Characteristics of Older Adults with Dementia Who Passed and Failed the Road Test (n = 53) Characteristic
Pass, n = 23
Fail, n = 30
P-Value
Age, mean sd Male, n (%) Education, meanSD White, n (%) Assessing Dementia 8 Screening Interview score, mean sd Driving experience, years, meanSD Short Blessed Test score, mean sd, Clock Drawing: Freund score, meanSD Trail-Making Test Part A, seconds, mean sd Trail-Making Test Part B, seconds, meanSD Visual closure (Motor-Free Visual Perception Test) errors, mean sd Useful field of vision, ms, meanSD Snellgrove Maze Test, seconds, mean sd
73.6 8.8 18 (78.3) 15.3 3.5 17 (81.0) 4.8 1.5
76.4 7.5 18 (60) 15.1 3.5 26 (86.7) 5.9 1.5
.22 .24 .80 .70 .009
57.1 9.2
58.5 8.6
.57
7.2 5.3
10.5 7.6
.08
5.6 1.53
4.8 2.1
.11
50.7 18.9
66.2 36.2
.07
176.0 77.2
192.2 85.0
.50
3.7 2.8
4.1 2.7
.64
286.4 172.6
312.7 166.0
.72
43.6 16.1
51.4 24.2
.19
SD = standard deviation.
SD = standard deviation.
with dementia made approximately twice as many driving errors as the control group and made the errors during driving situations that are generally considered the least challenging (right turns and driving straight). This suggests that, although it is not unusual for older adults to make driving errors, it may be the most common and least challenging of driving conditions that serve to differentiate older adults with medical impairment (dementia) from those without. In addition, none of the errors in the control group, in contrast to the dementia group, resulted in a potentially dangerous action or brake intervention. Thus, although older adults without dementia make driving errors, they appear to plan responses in advance and react to avoid potentially dangerous actions, unlike those with dementia. Some individuals with dementia (43%) in this study demonstrated safe driving and passed the road test. This is consistent with studies from other sites8 and previous studies on the WURT25 indicating that some individuals with dementia retain their ability to drive safely and can pass an on-road test. The current findings indicate that the AD8, an informant-based rating scale of functional performance, was the only measure of cognitive function that significantly differentiated individuals with dementia who passed the road test from those who failed. This finding
Table 3. Differences in Driving Behavior and Driving Situation Errors in Drivers with Dementia Who Passed and Failed the Road Test (n = 53) Passed, n = 23 Error
Driving behavior Operational Tactical Information processing Total situation Right turn moderate to high traffic Left turn moderate to high traffic Straight driving Lane changes
Failed, n = 30
P-Value
Least Square Mean (Standard Error)
2.1 (0.7) 4.6 (0.8) 3.1 (1.0)
3.3 (0.6) 8.6 (0.7) 9.0 (0.9)
.23
OBJECTIVES: To differentiate driving errors in persons with dementia who fail a performance- based road test from errors in persons who pass. DESIGN: Cross-sectional. SETTING: Community. PARTICIPANTS: Active drivers diagnosed with dementia (n = 60) and older adult controls (n = 32). MEASUREMENT: All participants completed a standardized clinical and on-road driving assessment. The outcome variable was the number and types of driving errors according to the Record of Driving Errors (RODE), a standardized tool to record driving errors. RESULTS: Sixty-two percent (n = 37) of individuals with dementia and 3% (n = 1) of controls failed the road test. Based on the RODE, individuals with dementia made twice as many driving errors as healthy controls. Within the dementia sample, individuals who failed the road test had more difficulties driving straight and making left and right turns than during lane changes. Dangerous actions occurred most often while driving straight and making left turns. Specific driving behaviors associated with road test failure in the sample with dementia included difficulties in lane positioning and usage, stopping the vehicle appropriately, attention, decision-making, and following rules of the road. Informants of participants with dementia who failed the road test reported more impairment with cognitive functioning on the Assessing Dementia 8 Screening Interview (AD8). CONCLUSION: This report highlights the driving errors most common in people with dementia who fail a road test. The finding that most of the dangerous actions in the sample with dementia occurred while driving straight condition is novel. Driving on straight roads has not been considered a From the aProgram in Occupational Therapy, School of Medicine, Washington University, St. Louis, Missouri; bDepartment of Neurology, Warren Alpert Medical School, Brown University, Providence, Rhode Island; cIndependent Drivers, LLC; dCenter for Clinical Studies; eDivision of Biostatistics; and Departments of fMedicine and gNeurology, School of Medicine, Washington University, St. Louis, Missouri. Address correspondence to Peggy P. Barco, OTD, OTR/L, SCDCM, Washington University School of Medicine in St. Louis, Program in Occupational Therapy, 4444 Forest Park Blvd., St. Louis, MO 63108. E-mail: [email protected] DOI: 10.1111/jgs.13508
JAGS 63:1373–1380, 2015 © 2015, Copyright the Authors Journal compilation © 2015, The American Geriatrics Society
condition of “high challenge” in prior driving studies in individuals with dementia. This finding has potential implications for future interventions related to vehicle instrumentation and driving recommendations for people with dementia. J Am Geriatr Soc 63:1373–1380, 2015.
Key words: driving performance; dementia; older drivers; driving errors
T
he prevalence of Alzheimer’s disease (AD) is expected to increase from an estimated 4.7 million individuals in 2010 to 13.8 million in 2050.1 Although some experts recommend that all adults with a dementia diagnosis refrain from driving, most clinicians base the decision on dementia severity and impaired driving performance.2,3 Outpatient surveys indicate that approximately 30% of older adults with dementia are current drivers.4,5 As the ability to diagnose AD in the preclinical and early stages of disease (e.g., mild cognitive impairment) continues to improve, some individuals will retain the appropriate skill sets to maintain driving privileges for years after diagnosis,6–8 before eventual driving cessation. The methods most commonly used to measure driving ability in older adults include performance-based road tests, driving simulators, questionnaires, and documentation of motor vehicle crashes.9,10 Although costly, advanced in-car video technology in combination with naturalistic driving have also been explored in determining driving safety in older adults.11,12 Although driving simulators eliminate the safety hazards present in road tests, debate continues regarding whether simulators can predict actual driving performance.13,14 Driving questionnaires completed by family members have limitations because limited knowledge of the older adults’ driving abilities and the fear of increased family responsibility if the family member is unable to drive may bias results. When driving older adults complete questionnaires, they may not accurately reflect performance on the road because of insight deficits that can accompany dementia.15,16 Documentation of crashes does not always clearly indicate “at fault” relationships, and many smaller crashes go unreported.
0002-8614/15/$15.00
1374
BARCO ET AL.
The performance-based road test is the most widely accepted method for determining driving competence. The road test evaluates the older adult’s abilities while actually operating a motor vehicle in traffic2,7,17 and has been associated with a history of crashes.18–21 Limitations of the performance-based road test include lack of objectivity or interrater reliability, safety risk to the individual and community, differing road courses, variability in traffic conditions, lack of test–retest measurements (e.g., stability), and various scoring procedures.10,22 Traditionally, road test performance with older drivers has been evaluated according to global performance outcomes (e.g., pass, marginal, fail),23–25 although some recent studies have examined the number and type of driving errors made during road testing.12,18,21,26–29 Studies with individuals with AD and specific driving errors have indicated a variety of errors, with one of the more common errors being difficulty with lane observance and lane changes.18,27,28 Although these studies have found significant differences in number and type of driving errors between medical diagnoses (e.g. between AD or Parkinson’s disease and controls), less is known about the differences in driving errors between those within the diagnostic groups who pass and those who fail. The aims of this study were to determine the differences in driving errors between older adults with and without dementia and to compare in detail the number and types of driving errors that older adults with dementia who fail a road test and those who pass.
METHODS The Missouri Department of Transportation, Traffic and Highway Safety Division funded this study, which was conducted at Washington University Medical School in St. Louis, through the DRIVING Connections Clinic at the Rehabilitation Institute of St. Louis (TRISL). The human studies committee at Washington University approved the study.
Participants Individuals with dementia were recruited through the Memory Diagnostic Center of Washington University School of Medicine in St. Louis and Washington University Physicians (January 2008 to September 2009 and April to September 2012). Healthy older adults (controls) were recruited from the Volunteer for Health Data Base maintained by Washington University Medical School, local community centers, and word of mouth from January 2010 through June 2011. Inclusion criteria for the individuals with dementia was a primary diagnosis of dementia from the referring physician, an Assessing Dementia-8 Screening Interview (AD8)30 score of 2 or more by an informant, physician referral for a driving evaluation, and an informant available to accompany the participant to the driving evaluation. Participants for the healthy older adults (controls) were required to be aged 55 and older, have an informant willing to answer questionnaires, be cognitively intact as indicated by an AD8 score of less than 2 (rated by an informant), and have a Short Blessed Test score less than 9. Exclusion criteria for both groups
JULY 2015–VOL. 63, NO. 7
JAGS
included lack of an active driver’s license; visual acuity that did not meet the state guideline; non-English speaking; less than 10 years of driving experience; any major chronic unstable disease or condition (e.g., seizures); severe orthopedic, musculoskeletal, or neuromuscular impairments that would require adaptive equipment to drive; visual, hearing, or language impairments that would interfere with testing; sedating drugs (e.g., new use of narcotics or anxiolytics within the past month or chronic use that causes sedation); or having had a previous driving evaluation in the last 12 months. One hundred twenty-six of 202 individuals with dementia who were screened over the telephone met criteria and agreed to participate in this study. Reasons for nonparticipation included refusal or lack of interest in study (47), not currently driving or no license (16), concern about potential loss of license (7), prior driving evaluation within a year (4), no return call (3), diagnosis of dementia not confirmed by physician (2), not comfortable with the driving route in the city (2), and miscellaneous reasons (e.g., unexplained cancellation, no show) (9). A subset of the 126 participants (n = 60) was arbitrarily selected (based on availability of the second driving specialist to score errors and study funding limitations) to be in the driving error study. Forty-one of 54 healthy older adults who were screened over the telephone met inclusion and exclusion criteria and agreed to participate in the study. Reasons for nonparticipation included medical illness (2), stopped driving already (1) error scores were not performed because of unanticipated absence of scorer (2), transportation problems (1), unknown reason or other (3). The remaining 32 healthy elderly adults participated in the study.
Procedures Participants underwent a comprehensive driving evaluation that included a clinical assessment and an on-road evaluation. Because their physician referred the individuals with dementia for medically indicated driving assessments, a recommendation meeting (with the informant and participant) related to future driving capability was held at the conclusion of the evaluation process. A full driving evaluation report with recommendations was sent to the referring physician with permission from the driver and the informant. A trained occupational therapist (OT) or certified occupational therapy assistant (with supervision from the OT) administered the clinical portion of the driving assessment. This portion of the driving assessment included standardized tests of vision, cognitive, and motor function with procedures described previously.23 The cognitive screens and assessments included the AD8,30 Short Blessed Test31 Freund Clock Drawing Test,32 Trail-Making Test Parts A and B,33 Snellgrove Maze Test,34 and Driving Health Inventory—Useful Field of View and Visual Closure Tests.35 The performance-based road test, the Modified Washington University Road Test (mWURT), used common traffic situations and road maneuvers from the previous Washington University Road Test.25 The mWURT has been described in detail along with the approach for qualitative
JAGS
JULY 2015–VOL. 63, NO. 7
assessment used in the current study (pass, marginal, fail).23,36 The road test was initiated in a closed parking lot and then proceded to low traffic conditions (a park setting) and then moderate to high traffic conditions (with complex intersections, traffic lights, and more traffic) as safety considerations permitted. The course was 13 miles long and took approximately 1 hour to complete. Participants drove a midsize driving evaluation car (with dual brakes for safety); an OT Certified Driving Rehabilitation Specialist (OT/CDRS) sat in the front seat to provide directions and maintain safety. A second OT (OT/Driving Rehabilitation Specialist—OT/DRS) blinded to clinical evaluation data sat in the back seat to recorded driving errors. The road test was discontinued if both incar evaluators agreed that safety was in jeopardy. The point of discontinuation was recorded. A standardized quantitative rating of number and type of safety errors using The Record of Driving Errors (RODE)* (Figure 1) was based on error categorizations (driving situation and driving behavior errors) from the literature.21,29 Driving errors were first recorded according to traffic condition (closed parking lot, low traffic, moderate to high traffic). Next, within each traffic condition, errors were recorded according to the specific driving situation involved (parking or backing, left turn, right turn, straight driving, lane change). Finally, within each driving situation, various combinations of driving behaviors that resulted in the driving situation error (driver behavior errors) were categorized as operational, tactical, or information processing errors and recorded. Similar to other models, operational errors37 were those involving operation of the car (e.g., use of turn signals), tactical errors37 were those related to driving tactics and skills (e.g., maintaining lane position, visual scanning), and information processing errors were those related to higher processing skills (e.g., following directions, memory, decision-making, knowing rules of the road), and were similar to, but extended beyond, the strategic and planning driving errors described in previous studies.28,38 A portion of this road test included being able to find a destination and use strategic planning to locate, enter, park, and exit independently. Important to the RODE scoring protocol, one driving situation error could be the result of multiple driving behavior errors (e.g., operational, tactical, informational processing errors). For example, a driving situation error involving a lane change could include failure to scan, failure to yield to a vehicle, failure to signal, and failure to anticipate the need to change lanes yet would still be counted as “one” driving situation error. Finally, any necessary interventions to maintain safety were recorded (e.g., brake intervention). Errors resulting in a potentially dangerous action (an action that the driver made posed an immediate potential threat to safety, e.g., failure to yield to a vehicle resulting in the instructor needing to make an intervention to avoid a collision, stopping in mid-traffic on a busy road causing other drivers to maneuver quickly to avoid a collision) were also recorded. Interrater reliability on scoring of the RODE was found to be strong across *For more information about The Record of Driving Errors (RODE), Barco, 2007, contact Peggy P. Barco at Washington University School of Medicine in St. Louis—Program in Occupational Therapy.
DRIVING ERRORS IN PERSONS WITH DEMENTIA
1375
error types,36 and both in-car instructors rated driving performance qualitatively as pass, marginal, or fail using an objective format described previously for which kappa for interrater reliability was found to be high (j = 0.84).23 The qualitative ratings were collapsed into two categories (pass (pass and marginal) and fail), and the OT/CDRS rating was used for analysis.
Statistical Analyses Differences in demographic characteristics, cognitive test scores, and driving errors between healthy older adults and individuals with dementia were analyzed using t-tests for continuous variables and the Fisher exact test for categorical variables. Similar analyses to determine demographic and cognitive differences were subsequently performed between individuals with dementia who passed and those who failed the road test to explore the differences further. The comparisons between individuals with dementia who passed the road test and those who failed were made through analysis of covariance (age adjusted). Analysis was to determine whether there was a difference in the number of total driving situation errors made and number of specific driving situation errors in moderate to high traffic (right turn, left turn, straight, lane changes), as well as driving behavior errors (operational, tactical, information processing). Because many of the specific driving behavior errors within the operational, tactical, and information processing error categories occurred only once or twice per person (if they occurred at all), the specific driving behavior errors were dichotomized (0 vs >1) and analyzed using logistic regression.
RESULTS Sixty-two percent (n = 37) of individuals with dementia and 3% (n = 1) of older adults without dementia failed the road test. There were significant differences in age (P = .03), race (P = .002), and all areas of cognitive testing between individuals with dementia and controls (Table 1). Participants with dementia (n = 53) were found to make twice as many driving situation errors as controls (n = 32) (dementia group mean errors 12.0 7.5, control group mean errors 6.0 4.9; P < .001). Although controls made considerably fewer driving errors than participants with dementia, they made errors in all driving situations. Individuals with dementia made errors more often than controls when driving straight (dementia group mean 2.06 2.42, control group mean 1.16 1.51; P = .04) and when making right turns (dementia group mean 1.04 1.54, control group mean 0.38 0.83; P = .03). In contrast to the dementia group, none of the driving errors in the healthy control group involved dangerous actions or brake interventions. Errors in the cognitively intact control group were most common in (recorded as percentage of participants with at least one error) decreased visual scanning (59%), difficulty maintaining lane positioning or lane usage (56%), lack of use of turn signals (44%), limited knowledge of rules of the road (44%), and difficulty with memory or following instructions (41%). Thus, although the control group had driving errors, they had errors significantly less frequently
1376
BARCO ET AL.
JULY 2015–VOL. 63, NO. 7
JAGS
Level One: Traffic Condition Closed Parking Lot
Moderate High Traffic
Low Traffic
Level Two: Driving Situation Errors Parking Backing
Right Turn
Left Turn
Straight away Driving
Lane Change
Level Three: Driving Behavior Errors Operational: Use of controls, gears, signals, foot pedals, steering Tactical: Visual Scanning, Distance Judging, Lane Usage/position, stopping, speed, yielding Information processing: Anticipates, Attention, Decision Making, Following Directions/Memory, Rules of The Road, Speed of Processing
Driving Interventions Wheel/Brake Interventions Dangerous Actions
Figure 1. Record of Driving Errors (RODE) Classification Levels. Adapted with permission from reference 39. Copyright 2015 American Occupational Therapy Association.
than the dementia sample, and the errors were rarely significant enough that controls failed the road test. Subsequently, the focus was directed to understanding the differences in the driving errors within the dementia sample—between those with dementia that passed a road test compared to those who failed. Of the 60 participants with dementia, 16 (27%) were not allowed to continue the road test because of serious safety concerns that both in-car evaluators had. These discontinuations occurred at every level of traffic, with 12% (n = 7) being discontinued so early in the road test that they never made it into moderate to high traffic. These seven individuals were not found to be significantly different from those who failed at the more-challenging levels of traffic in age or AD8 score, although most standardized cognitive tests given during the clinical assessment showed them to be significantly more impaired. Only individuals with dementia who were able to progress to the moderate to high traffic condition (n = 53) were included in this portion of the study (because they had the potential to make errors at all levels of traffic). Within the dementia group, 43% (n = 23) passed the road test, and 57% (n = 30) failed. There were no significant differences between those who passed and those who failed in age, sex, education, race, or years of driving experience (Table 2). Those who failed the road test had higher AD8 scores (indicating greater cognitive impairment) than those who passed, with other areas of cognitive assessment showing no significant differences between the two groups (Table 2). Significant differences were observed in the number of total driving situation errors (P < .001) between those with dementia who passed the road test and those who
failed. Nearly all traffic situations (in moderate to high traffic) with the exception of making lane changes were noted to result in more errors in those who failed than those who passed, including making turns (P = .03) and left (P = .001) turns and driving straight (not making a left turn, right turn, or lane change) (P = .01) (Table 3). Dangerous actions occurred most often when driving straight (32% of participants ≥1 occurrences) and turning left (21% ≥1 occurrences) as opposed to turning right (2% ≥1 occurrences) and changing lanes (8% ≥1 occurrences). Driving behavior errors were also noted to be significantly different between individuals with dementia who passed and those who failed, including total tactical errors (P = .001) and total information processing errors (P = .001) but not total operational errors (Table 3). The tactical errors that differentiated those who failed the road test from those who passed were difficulty with lane positioning and usage (P = .01) and inability to stop the vehicle appropriately (P = .04). The information processing errors that reflected significant differences between the groups included difficulty with attention while driving (P = .001), decision-making during driving (P = .001), and knowing the rules of the road (P = .04) (Table 4).
DISCUSSION Similar to other studies examining older adults with dementia on performance-based road tests, participants with dementia had higher levels of impaired driving performance, resulting in more failed road tests and driving errors than healthy older adult controls.27,28 Participants
JAGS
JULY 2015–VOL. 63, NO. 7
DRIVING ERRORS IN PERSONS WITH DEMENTIA
Table 1. Demographic and Cognitive Characteristics of Older Adults with Dementia and Older Adult Controls (N = 92) Characteristic
Age, meanSD Male, n (%) Education, years, mean sd White, n (%) Driving experience, years, meanSD Assessing Dementia 8 Screening Interview score, mean sd (range 0–8) Short Blessed Test score, meanSD (range 0–28) Clock Drawing: Freund Score, mean sd (range 0–7) Trail-Making Test Part A, seconds, meanSD Trail-Making Test Part B, seconds, mean sd Visual closure (Motor-Free Visual Perception Test) errors, meanSD (range 0–11 errors) Useful field of vision, ms, mean sd (range 100–500 ms) Snellgrove Maze Test, seconds, meanSD
Dementia (n = 60)
Controls (n = 32)
74.7 8.5 40 (66.7) 15.3 3.4
70.7 8.1 16 (50.0) 14.9 3.0
.03 .17 .64
49 (86.0) 57.5 9.0
18 (56.3) 52.6 9.4
.004 .02
5.5 1.6
0.28 0.46
.001
P-Value
9.8 6.8
6.4 2.5
.001
4.7 2.2
6.4 0.91
.001
68.3 42.5
41.3 13.9
.001
191.3 83.1
109.4 48.7
.001
4.3 2.8
2.2 1.5
.001
330.4 168.9
183.8 109.5
.001
58.3 46.1
36.2 13.7
.009
1377
Table 2. Demographic and Cognitive Characteristics of Older Adults with Dementia Who Passed and Failed the Road Test (n = 53) Characteristic
Pass, n = 23
Fail, n = 30
P-Value
Age, mean sd Male, n (%) Education, meanSD White, n (%) Assessing Dementia 8 Screening Interview score, mean sd Driving experience, years, meanSD Short Blessed Test score, mean sd, Clock Drawing: Freund score, meanSD Trail-Making Test Part A, seconds, mean sd Trail-Making Test Part B, seconds, meanSD Visual closure (Motor-Free Visual Perception Test) errors, mean sd Useful field of vision, ms, meanSD Snellgrove Maze Test, seconds, mean sd
73.6 8.8 18 (78.3) 15.3 3.5 17 (81.0) 4.8 1.5
76.4 7.5 18 (60) 15.1 3.5 26 (86.7) 5.9 1.5
.22 .24 .80 .70 .009
57.1 9.2
58.5 8.6
.57
7.2 5.3
10.5 7.6
.08
5.6 1.53
4.8 2.1
.11
50.7 18.9
66.2 36.2
.07
176.0 77.2
192.2 85.0
.50
3.7 2.8
4.1 2.7
.64
286.4 172.6
312.7 166.0
.72
43.6 16.1
51.4 24.2
.19
SD = standard deviation.
SD = standard deviation.
with dementia made approximately twice as many driving errors as the control group and made the errors during driving situations that are generally considered the least challenging (right turns and driving straight). This suggests that, although it is not unusual for older adults to make driving errors, it may be the most common and least challenging of driving conditions that serve to differentiate older adults with medical impairment (dementia) from those without. In addition, none of the errors in the control group, in contrast to the dementia group, resulted in a potentially dangerous action or brake intervention. Thus, although older adults without dementia make driving errors, they appear to plan responses in advance and react to avoid potentially dangerous actions, unlike those with dementia. Some individuals with dementia (43%) in this study demonstrated safe driving and passed the road test. This is consistent with studies from other sites8 and previous studies on the WURT25 indicating that some individuals with dementia retain their ability to drive safely and can pass an on-road test. The current findings indicate that the AD8, an informant-based rating scale of functional performance, was the only measure of cognitive function that significantly differentiated individuals with dementia who passed the road test from those who failed. This finding
Table 3. Differences in Driving Behavior and Driving Situation Errors in Drivers with Dementia Who Passed and Failed the Road Test (n = 53) Passed, n = 23 Error
Driving behavior Operational Tactical Information processing Total situation Right turn moderate to high traffic Left turn moderate to high traffic Straight driving Lane changes
Failed, n = 30
P-Value
Least Square Mean (Standard Error)
2.1 (0.7) 4.6 (0.8) 3.1 (1.0)
3.3 (0.6) 8.6 (0.7) 9.0 (0.9)
.23
