Using Serial Trichotomization to Determine Fitness to Drive in Medically At-Risk Drivers.

IMPORTANCE: Clinical tests that identify fit and unfit drivers with 100% sensitivity and specificity would reduce uncertainty and improve efficiency of occupational therapists performing comprehensive driving evaluations (CDEs). OBJECTIVE: To examine whether serial trichotomization of clinical tests predicts pass-fail outcomes with 100% sensitivity and specificity in a sample of medically at-risk drivers and in drivers with and without cognitive impairment (CI) referred for a CDE. DESIGN: Retrospective data collection and analysis of scores on the Montreal Cognitive Assessment; Trail Making Test, Part A and Part B; and the Useful Field of View® Subtests 1 to 3 and outcomes on the CDE (pass-fail or indeterminate requiring lessons and retesting). Receiver operating characteristic curves of clinical tests were performed to determine 100% sensitivity and specificity cut points in predicting CDE outcomes. Clinical tests were arranged in order from most to least predictive to identify pass-fail and indeterminate outcomes. SETTING: A driving assessment clinic. PARTICIPANTS: Among 142 medically at-risk drivers (M age = 69.2 yr, SD = 14.1), 66 with CI, 46 passed and 39 failed the CDE; 57 were indeterminate. OUTCOMES AND MEASURES: On-road pass-fail outcomes. RESULTS: Together, the six clinical tests predicted 62 pass and 49 fail outcomes in the total sample; 21 pass and 34 fail outcomes in participants with CI; and 58 pass and 14 fail outcomes in participants without CI. CONCLUSIONS AND RELEVANCE: Serial trichotomization of clinical tests increases the accuracy of making informed decisions and reduces the number of drivers undergoing unnecessary on-road assessments. Plain-Language Summary: Clinical tests and their cut points that identify fit and unfit drivers vary substantially across settings and research studies. Serial trichotomization is one method that could help control for this variation by combining clinical test scores showing 100% sensitivity and specificity to identify pass (fit drivers) and fail outcomes (unfit drivers) and to reduce the number of drivers undergoing unnecessary on-road assessments.

Driving Errors Predicting Pass/Fail On-Road Assessment Outcomes Among Cognitively Impaired Older Drivers.

Older drivers with cognitive impairment (CI)/dementia make significantly more driving errors than healthy controls; however, whether driving errors are predictive of pass/fail outcomes in older drivers with CI/dementia are unclear. This study determined the driving errors that predicted failing an on-road assessment in drivers with CI. We retrospectively collected comprehensive driving evaluation data of 80 participants (76.1 ± 9.3 years) from an Ontario driving assessment center. Adjustment to stimuli (area under the curve [AUC] = 0.88), lane maintenance (AUC = 0.84), and speed regulation errors (AUC = 0.85) strongly predicted pass/fail outcomes. Worse performance on the Trails B (time) and Useful Field of View® (Subtest 2, Subtest 3, and risk index) were significantly correlated with adjustment to stimuli (p < .05), lane maintenance (p < .05), and speed regulation errors (p < .05). Adjustment to stimuli, lane maintenance, and speed regulation errors may be critical indicators of failing an on-road assessment in older drivers with CI. Prioritizing these errors may help identify at-risk drivers.

Clinical Tests Predicting On-Road Performance in Older Drivers with Cognitive Impairment.

Background. The Trail Making Test Part B (Trails B) and Useful Field of View® (UFOV) can predict on-road outcomes in drivers with cognitive impairment (CI); however, studies have not included drivers referred for comprehensive driving evaluations (CDEs), who typically have more severe CI. Purpose. We determined the predictive ability of Trails B and UFOV on pass/fail on-road outcomes in drivers with CI (Montreal Cognitive Assessment <26) referred for CDEs. Method. Retrospective data collection from two driving assessments centers (N = 100, mean age = 76.2 ± 8.8 years). Findings. The Trails B (area under the curve [AUC] = .70) and UFOV subtests 2 (AUC = .73) and 3 (AUC = .76) predicted pass/fail outcomes. A cut-point ≥467 ms on UFOV subtest 3 better-predicted pass/fail outcomes with 78.9% sensitivity and 73.5% specificity. In comparison, a cut-point ≥3.58 min on Trails B had lower sensitivity (73.7%) and specificity (61.8%). Implications. The UFOV subtest 3 may be more useful than the Trails B for predicting pass/fail outcomes in drivers with more severe CI referred for CDEs.

Driving errors that predict simulated rear-end collisions in drivers with multiple sclerosis.

OBJECTIVE: Drivers with Multiple Sclerosis (MS) may have an increased crash risk. However, the driving performance deficits that contribute to crashes are not fully understood. Based on the extant literature, adjustment to stimuli errors indicate failing an on-road assessment. This study examines whether adjustment to stimuli errors can detect the occurrence of collisions in a driving simulator in drivers with MS. METHODS: As part of a quasi-experiment, 38 participants with MS and 21 participants without MS completed visual-cognitive and driving simulator assessments, which also recorded their adjustment to stimuli maneuvers. We quantified participants' adjustment to stimuli maneuvers via initial pedal reaction time (seconds), time to collision (seconds), mean speed (meters per second), and the occurrence of rear-end collisions (collide vs. did not collide) when a simulated vehicle cut across the lane in front of them. RESULTS: Logistic regression analyses indicated that, compared to drivers without MS, those with MS had a shorter time to collision (OR= .04, p= .001, 95% CI= [.006, .27]) and a faster mean speed (OR= 1.32, p= .04, 95% CI= [1.01, 1.74]) which increased the odds of experiencing a rear-end collision. Receiver operating characteristic curve analyses indicated that, for MS and control groups, time to collision (MS group = AUC= .94, p<.0001, Control group = AUC= .86, p<.0001) and mean speed (MS group = AUC= .76, p=.005, Control group = AUC = .78, p= .005) differentiated between participants who collided vs. did not collide. For drivers with MS, a time to collision of ≤1.81 seconds (85% sensitivity, 100% specificity, 15% error rate), and a mean speed of ≥7.83 meters per second (77% sensitivity, 76% specificity, 47% error rate) predicted the occurrence of collisions with the lowest error rate. CONCLUSIONS: During a driving simulator assessment, adjustment to stimuli errors predicted the occurrence of rear-end collisions in drivers with MS (vs. without MS). Driving assessors may target scenarios that measure participants' adjustment to stimuli, via time to collision and mean speed, to make decisions about their visual-cognitive deficits and driving performance.

Clinical predictors of driving simulator performance in drivers with multiple sclerosis.

BACKGROUND: Drivers with multiple sclerosis (MS) may experience visual-cognitive impairment that affects their fitness to drive. Due to limitations associated with the on-road assessment, an alternative assessment that measures driving performance is warranted. Whether clinical indicators of on-road outcomes can also predict driving performance outcomes on a driving simulator are not fully understood. OBJECTIVE: This study examined if deficits in immediate verbal/auditory recall (California Verbal Learning Test-Second Edition; CVLT2-IR) and/or slower divided attention (Useful Field of View™; UFOV2) predicted deficits in operational, tactical, or strategic maneuvers assessed on a driving simulator, in drivers with and without MS. METHODS: Participants completed the CVLT2-IR, UFOV2, and a driving simulator assessment of operational, tactical, and strategic maneuvers. RESULTS: Deficits in immediate verbal/auditory recall and slower divided attention predicted adjustment to stimuli errors, pertaining to tactical maneuvers only, in 36 drivers with MS (vs 20 drivers without MS; F(3, 51) = 6.1, p = 0.001, R2 = 0.3, Radj2=0.2). CONCLUSION: The CVLT2-IR and UFOV2 capture the visual and verbal/auditory recall, processing speed, and divided attention required to appropriately adjust to stimuli in a simulated driving environment. Clinicians may use the CVLT2-IR and UFOV2 as precursors to driving performance deficits in drivers with MS.

Insufficient sleep and fitness to drive in shift workers: A systematic literature review.

BACKGROUND: Insufficient sleep, <6.5 h per night, majorly affects shift workers, placing them at higher risk for motor vehicle crash related injury or fatality. While systematic reviews (SLRs) examine the effects of insufficient sleep and driving, to date, no SLR focuses on driver fitness or performance in shift workers. OBJECTIVES: Determine the class of evidence (Class I-highest to Class IV-lowest), and level of confidence (Level A-high, to Level U-insufficient) in the determinants of driver fitness and performance in shift workers. Next, consider evidence-based recommendations for clinical practice, research, and policy. METHODS: A protocol was registered on PROSPERO (#CRD42018052905) using an established SLR methodology: a comprehensive electronic database search, study selection, data extraction, critical appraisal, analysis, and interpretation using published guidelines. RESULTS: Searches identified 1226 unique records with 11(2 on-road, 9 simulator) meeting final inclusion criteria. Class III to IV evidence identified that exposure to overnight shift work possibly predicts (Level C confidence) drivers at risk for adverse on-road outcomes and likely predicts (Level B) drivers at risk for adverse driving simulator outcomes. Higher ratings of subjective sleepiness and extended time driving possibly predict (Level C) drivers at risk for adverse driving simulator outcomes. CONCLUSIONS: This study demonstrates a low to moderate level of confidence in the determinants of driving in shift workers. A critical need exists for gold-standard on-road assessments integrating complex driving environments representative of real-world demands, targeting tactical and strategic outcomes in a broad spectrum of shift workers.

Clinical Determinants of Fitness to Drive in Persons With Multiple Sclerosis: Systematic Review.

OBJECTIVE: To critically appraise the evidence for the clinical determinants of fitness to drive in adults with multiple sclerosis (MS). DATA SOURCES: The research librarian and lead author searched 7 databases for driving simulator and on-road studies for adults with MS published in the English language from 1991 to 2018. STUDY SELECTION: Three reviewers independently screened titles, abstracts, and full-texts for studies with: cohort, case-control, or cross-sectional designs; participants, 18 years or older, with relapsing or progressive MS; visual, cognitive, or motor clinical assessments as predictors; and driving performance through simulator or fitness to drive through on-road assessment as outcomes. DATA EXTRACTION: Using the 2017 American Academy of Neurology guidelines, reviewers independently classified each study from class I to class IV, or highest to lowest amount of rigor. For each clinical assessment, reviewers independently rated the level of confidence for predicting driving performance or fitness to drive from level A, highly probable; B, probable; C, possible; to level U, insufficient conclusions. DATA SYNTHESIS: Through qualitative synthesis, 2 class III and 4 class IV driving simulator studies employed 24 clinical assessments with level C (n=4) or level U (n=20) confidence for predicting driving performance. Six class II and 3 class IV on-road studies employed 35 clinical assessments with level B (n=9), level C (n=22), or level U (n=4) confidence for predicting fitness to drive. CONCLUSIONS: This systematic review identified mostly insufficient conclusions for predicting driving performance in driving simulator studies, and possible conclusions for predicting fitness to drive in on-road studies. The best available evidence suggests that the Stroke Driver Screening Assessment and Useful Field of View test probably predict fitness to drive in adults with MS (level B). Class I studies that compare predictors of fitness to drive with large prospective samples of adults with and without MS are necessary for highly probable conclusions.

Insufficient sleep and fitness to drive in shift workers: a systematic literature review protocol.

BACKGROUND: The majority of shift workers experience insufficient sleep as a result of their employment. Insufficient sleep is associated with impaired neurocognitive functioning, affecting key skills required for driving, resulting in shift workers experiencing a disproportionate burden of RTC injuries and fatalities. Yet, to our knowledge, no systematic literature review (SLR) exists to critically appraise and synthesise evidence on the determinants of fitness to drive (assessed on-road) and driving performance (assessed in a driving simulator) in shift workers with insufficient sleep. OBJECTIVES: A SLR protocol is established to conduct analysis and synthesis of the level of evidence and confidence in the determinants of fitness to drive and driving performance, among shift workers with insufficient sleep. METHODS: This study follows Cooper and Hedges' established SLR methodology: formulate the problem, locate and select studies, collect data, appraise critically, analyse and present data, interpret results and disseminate information. Critical appraisal and analysis follows the 2017 American Academy of Neurology guidelines determining the level of evidence and the level of confidence for each determinant identified in the literature. Protocol and results reporting adhere to the Preferred Reporting Items for Systematic reviews and Meta-Analyses Protocols guidelines. CONCLUSIONS: This SLR contributes to research evidence examining the impact of insufficient sleep and driver sleepiness on fitness to drive and driving performance. Analysis of the level of evidence and level of confidence in the existing literature will advance evidence-informed prevention strategies and critical decision-making, to mitigate adverse effects of insufficient sleep for improving road safety.

On-road assessment of fitness-to-drive in persons with MS with cognitive impairment: A prospective study.

BACKGROUND: Cognitive impairment is common in multiple sclerosis (MS). In other populations, cognitive impairment is known to affect fitness-to-drive. Few studies have focused on fitness-to-drive in MS and no studies have solely focused on the influence of cognitive impairment. OBJECTIVE: To assess fitness-to-drive in persons with MS with cognitive impairment and low physical disability. METHODS: Persons with MS, aged 18-59 years with EDSS ⩽ 4.0, impaired processing speed, and impairment on at least one measure of memory or executive function, were recruited. Cognition was assessed using the Minimal Assessment of Cognitive Function battery. A formal on-road driving assessment was conducted. Chi-square analysis examined the association between the fitness-to-drive (pass/fail) and the neuropsychological test results (normal/impaired). Bayesian statistics predicting failure of the on-road assessment were calculated. RESULTS: Of 36 subjects, eight (22.2%) were unfit to drive. Only the BVMTR-IR, measuring visual-spatial memory, predicted on-road driving assessment failure ( X2 ( df = 1, N = 36) = 3.956; p = 0.047) with a sensitivity of 100%, but low specificity (35.7%) due to false positives (18/25). CONCLUSION: In persons with MS and impaired processing speed, impairment on the BVMTR-IR should lead clinicians to address fitness-to-drive.

Visual Correlates of Fitness to Drive in Adults With Multiple Sclerosis.

The impact of visual and visual-cognitive impairments on fitness to drive in persons with multiple sclerosis (PwMS) are not well studied. We quantified visual correlates of fitness to drive in 30 PwMS. PwMS completed visual ability and visual attention assessments, and a standardized on-road assessment, and were compared with 145 older volunteer drivers. PwMS (vs. older volunteer drivers) made more total ( W = 12,139, p = .03) and critical driving errors (predictive of crashes) in adjustment to stimuli ( W = 11,352, p < .0001), vehicle positioning ( W = 11,449, p < .0001), and wide lane turns ( W = 9,932, p < .0001). PwMS who failed (vs. passed) made more total ( W = 325, p = .04), adjustment to stimuli ( W = 321.5, p = .02), and gap acceptance errors ( W = 333, p = .03). For PwMS, adjustment to stimuli errors moderately correlated with visual acuity (ρ = .50, p = .006), and gap acceptance errors moderately correlated with visual processing speed (ρ = .40, p = .03). Visual-cognitive impairments may be indicative of critical driving errors and help identify PwMS at-risk for fitness to drive.

Driving Errors That Predict On-Road Outcomes in Adults With Multiple Sclerosis.

Driving errors that predict on-road outcomes for persons with multiple sclerosis (PwMS) are not well studied. The objective of this study was to determine whether adjustment-to-stimuli and gap acceptance errors significantly predict passing/failing a standardized on-road assessment of PwMS. Thirty-seven participants completed visual ability and visual attention assessments, and participated in an on-road assessment, where seven types of driving errors and pass/fail outcomes were determined. Adjustment-to-stimuli (No.) and gap acceptance errors (commit/did not commit) significantly predicted passing/failing the on-road assessment, with an area under the curve of 91.6% ( p < .0001). With no gap acceptance errors committed, five adjustment-to-stimuli errors optimally determined pass/fail outcomes in PwMS. Furthermore, with no adjustment to stimuli errors committed, committing any gap acceptance errors also optimally determined pass/fail outcomes in PwMS. Further research may focus on visual, cognitive, and/or motor impairments underlying adjustment-to-stimuli and gap acceptance errors for eventual development of rehabilitation strategies for PwMS.

Development and Validity of Western University's On-Road Assessment.

Although used across North America, many on-road studies do not explicitly document the content and metrics of on-road courses and accompanying assessments. This article discusses the development of the University of Western Ontario's on-road course, and elucidates the validity of its accompanying on-road assessment. We identified main components for developing an on-road course and used measurement theory to establish face, content, and initial construct validity. Five adult volunteer drivers and 30 drivers with multiple sclerosis participated in the study. The road course had face and content validity, representing 100% of roadway components determined through a content validity matrix and index. The known-groups method showed that debilitated drivers (vs. not debilitated), made more driving errors ( W = 463.50, p = .03), and failed the on-road course, indicating preliminary construct validity of the on-road assessment. This research guides and empirically supports a process for developing a road course and its assessment.

Interrater reliability of Western University's on-road assessment.

BACKGROUND: Little empirical support exists for interrater reliability between evaluators from different backgrounds when assessing on-road outcomes of drivers. PURPOSE: We quantified interrater reliability of on-road outcomes between a certified driving school instructor (DI) and an occupational therapist and certified driver rehabilitation specialist (CDRS). METHOD: Both raters used the Global Rating Score (GRS) with two levels (pass, fail), the GRS with four levels (pass, pass with recommendations, fail remediable, fail), and the priority error rating score (PERS; most frequently occurring on-road errors in priority order) to assess 35 drivers (age, M = 48.31 years, SD = 9.76 years; 40% male; 86% with multiple sclerosis). FINDINGS: The DI and occupational therapist CDRS had excellent agreement on the GRS with two levels (κ = .892, p < .0001), GRS with four levels (κ = .952, p < .0001), and the PERS (κ = .847-.902, p < .0001), indicating interrater reliability. IMPLICATIONS: This research contributes to empirical support for the on-road assessment.