Validation of a screening battery to predict driving safety in people with Parkinson s disease

Transcription

1 Manuscript title. Validation of a screening battery to predict driving safety in people with Parkinson s disease Names and institutional affiliations of the authors. H. Devos, PhD, 1,2,3 W. Vandenberghe, MD, PhD, 4,5 A. Nieuwboer, PhD, 1 M. Tant, PhD, 6 W. De Weerdt, PhD, 1 J. D. Dawson, Sc.D., 7 E. Y. Uc, MD 3,8 1 Department of Rehabilitation Sciences, KU Leuven, Heverlee, Belgium 2 National Advanced Driving Simulator, University of Iowa, Iowa City, IA, USA 3 Department of Neurology, University of Iowa, Iowa City, IA, USA 4 Department of Neurology, University Hospitals Leuven, Leuven, Belgium 5 Department of Neurosciences, KU Leuven, Leuven, Belgium 6 CARA Department, Belgian Road Safety Institute, Brussels, Belgium 7 Department of Biostatistics, University of Iowa, Iowa City, IA, USA 8 Neurology Service, Veterans Affairs Medical Center, Iowa City, IA, USA Name and address for corresponding author. Hannes Devos KU Leuven, Faculty of Kinesiology and Rehabilitation Sciences, Department of Rehabilitation Sciences, Tervuursevest 101, BE-3001 Heverlee Tel Fax hannes.devos@faber.kuleuven.be addresses co-authors. wim.vandenberghe@uzleuven.be; alice.nieuwboer@faber.kuleuven.be; mark.tant@bivv.be; willy.deweerdt@faber.kuleuven.be; jeffrey-dawson@uiowa.edu; ergun-uc@uiowa.edu. Word count. Number of characters in the title: 94; number of words in the abstract: 150; number of words in the body: 1,506; number of tables: 2 Running title. Fitness to drive in PD

2 Key words. Parkinson s disease; automobile driving; fitness to drive; driving simulator Financial disclosure/conflict of interest. The authors report no financial disclosures. The authors report no conflict of interest. Funding sources. This study was self-supported by the Department of Rehabilitation Sciences of KU Leuven and CARA, Belgian Road Safety Institute. Dr. Uc is supported by a Merit Review grant from Rehabilitation Research and Development Branch, the US Department of Veterans Affairs.

3 Devos et al 3 ABSTRACT We enrolled 60 drivers with Parkinson s disease (PD) to validate our original screening battery (consisting of contrast sensitivity, UPDRS-motor section score, Clinical Dementia Rating scale score, and disease duration) to predict driving fitness decisions (pass-fail) by a state agency where drivers underwent detailed visual, cognitive, and on-road testing. Twentyfour participants (40%) failed the driving evaluation. The screening battery correctly classified 46 (77%) participants (sensitivity and negative predictive value = 96%; specificity and positive predictive value = 64%). Adding other clinical predictors (e.g., age of onset, Hoehn-Yahr stage instead of UPDRS-motor score) failed to improve the specificity of the model when the sensitivity was kept constant at 96%. However, a driving simulator evaluation improved the specificity of the model to 94%. Our original clinical battery proved to be a valid screening tool that accurately identifies fit drivers with PD and select those who need more detailed testing at specialized centers.

4 Devos et al 4 INTRODUCTION Counseling patients with neurological disorders on their driving fitness is a complex process.(1) There is no concensus on which tests to use to determine driving safety.(2;3) Several clinical screening tools have been shown to predict driving fitness more accurately than the appraisal of physicians.(4-6) The predictive accuracies of screening tools to determine driving fitness in Parkinson s disease (PD) range between 72% and 90% compared with the ratings of driving evaluation experts.(7-9) Small sample sizes, clinical heterogeneity of PD symptoms, and differences between clinical screening tools may account for the wide range in predictive accuracies.(7-9) We previously developed a short clinical battery, consisting of contrast sensitivity (CS), Clinical Dementia Rating (CDR), Unified Parkinson s Disease Rating Scale-motor section (UPDRS III), and disease duration, which correctly classified 90% of PD drivers.(7) When the result of a driving simulation was added to the clinical model, the accuracy increased to 97.5%.(7) of the aforementioned screening tools have been cross-validated which limits their application in clinical practice.(7-9) The aim of this study was to validate our screening battery in a different sample of PD drivers. METHODS Patients From March 2008 to March 2012, 70 PD patients recruited through the Movement Disorders Clinic, University Hospitals Leuven, and the Center for Evaluation of Fitness to drive and Car Adaptations (CARA), were informed about the study. Nine persons declined participation and one dropped out before testing due to medical reasons. Sixty persons gave informed consent. The subgroup recruited from the University Hospitals (n=29) presented with no apparent driving problems, while those recruited from CARA (n=31)

5 Devos et al 5 were self-referred (n=4) or referred by physicians (n=22), insurance companies (n=4) or prosecution office (n=1) because of difficulties with driving. Inclusion criteria were valid driver s license; mileage >300 km/year; Hoehn-Yahr (H&Y) stage I III ( on ); and binocular visual acuity 20/40. Exclusion criteria were deep brain stimulators; unpredictable motor fluctuations; and any comorbidity hindering safe driving. Consistent with the guidelines on driving in dementia, persons with CDR scores >1 were also excluded.(10;11) All participants were assessed in the on phase. The study was approved by the University Hospitals Leuven ethics committee. In-clinic screening battery In addition to the general evaluation, a clinical screening battery was administered, encompassing CS, CDR, UPDRS III, and disease duration (Table 1).(7) The Pelli-Robson chart was used to measure CS.(12) The CDR assessed six domains of cognitive and functional performance.(13) The UPDRS III on was used to rate motor symptoms.(14) Disease duration was defined as the time interval in years between onset of motor symptoms and driving evaluation, as determined by the patient. Driving simulation The driving simulator was a full-sized Ford fiesta 1.8 and ran on STISIM software, version (STI Inc., Hawthorne, CA). Patients were exposed to several urban and suburban interactive driving scenarios projected onto a flat screen. Thirty items of driving skill were scored on a 4-point ordinal scale by the principal investigator (H.D) who was blind to the performance on the CARA evaluation (Table 1).(7) CARA assessment

6 Devos et al 6 The official driving evaluation comprised visual, neuropsychological, and on-road tests.(7;15) The 20-km on-road driving evaluation started in a suburban area with low to moderate traffic, progressed to a highway section, and ended in an urban area. The on-road checklist contained 13 items on a 4-point ordinal scale.(7;15) Following the CARA evaluation, patients were categorized as pass when they were fit to drive without restrictions. They were categorized as fail when they were either fit to drive with restrictions in time (e.g., not at night), distance (e.g., not in unfamiliar areas), or speed (e.g., not on highways), or unfit to drive. Statistical analysis Differences between groups were compared using unpaired t-tests, Wilcoxon Rank Sum tests, or Fisher s Exact tests. P-values< 0.05 were considered significant. Logistic regression analysis was used to investigate whether the inclusion of additional tests could improve predictive accuracy, when sensitivity was kept constant. Statistical procedures were performed with the SAS Enterprise 4.3 software. RESULTS Participants in the present study were older at time of disease onset (mean (standard deviation), (59.98 (9.94) vs (9.68); P =.01) and at time of testing (66.07 (9.61) vs (9.42); P =.02) than in our previous study.(7) They also had more years of driving experience (43.55 (10.46) vs (9.81); P =.03). They were involved in more accidents (median (Q1 Q3), 0 (0 1) vs 0 (0 0); P =.0006), and performed worse on CS (1.65 ( ) vs ( ); P =.0003) and UPDRS III (26.50 (21 37) vs (12

7 Devos et al ); P < ). H&Y stage, disease duration, and medication intake did not differ between the current and previous cohorts. From the current sample, 36 (60%) passed the CARA evaluation and 24 (40%) failed (13 fit to drive with restrictions and 11 unfit to drive ). Three items of the screening battery differentiated between the pass-fail groups (Table 1). To predict the pass-fail outcome, the previously developed discriminant equation was applied: (CS*16.95) (CDR*4.17) (UPDRS III*0.16) (disease duration*0.33) A negative outcome indicates that the patient is predicted to fail the CARA evaluation. The clinical model correctly classified 46 drivers in the respective pass-fail categories, corresponding to a predictive accuracy of 77%, sensitivity and negative predictive value (NPV) of 96%, and specificity and positive predictive value (PPV) of 64% (Table 2). No differences in predictive accuracies were found between persons referred through the University Hospitals Leuven (79%) and those referred to CARA from other places (74%). Univariate analysis showed that persons who failed were older at PD onset and at time of testing, were more disabled, but reported less traffic tickets as compared to those who passed (Table 1). Only age at onset remained significant in the multivariate regression model including the four variables of the clinical screening battery (odds ratio = 1.13 (95% confidence interval, , P =.03). The clinical screening battery with age of onset correctly identified 83% of patients (sensitivity = 83%; specificity = 86%; PPV = 80%; NPV = 89%). However, when we kept sensitivity constant at 96%, age of onset did not improve specificity compared to the original model. We explored H&Y stage instead of UPDRS III, but this model provided the same sensitivity and specificity as the original one.

8 Devos et al 8 Participants who passed the CARA evaluation scored better on the driving simulator evaluation (Table 1). When sensitivity was held constant at 96%, the driving simulator evaluation increased the specificity from 64% to 94%. The NPV of this enriched model was 97% and the PPV was 92%. DISCUSSION In this study, we cross-validated our previously established clinical screening battery to predict driving fitness in PD as determined by an official driving evaluation center that used detailed neuropsychological testing and a road drive. Compared to the previous study,(7) the sensitivity of our clinical battery improved from 91% to 96% and the specificity fell from 90% to 64%. These changes can be explained by differences between the current and previous cohorts. The clinical severity of PD and the fitness to drive failure was higher in the current (40%) compared to the past cohort (27.5%).(7) As the model has high sensitivity and NPV, one can confidently state that a PD driver who passes the screening is indeed fit to drive and does not need further evaluation. However, as the model has only modest specificity and PPV (64%), a substantial portion (36%) of drivers who fail the screening may still have the chance to pass a detailed, multidisciplinary driving evaluation. To avoid potentially unfair revocation of their driving privileges, PD drivers who fail the proposed screening battery should be referred for further evaluation. Our screening battery evaluated visual, cognitive and motor functions, and disease duration. Reduced CS is a common symptom in PD(16;17) and is associated with poor outcomes on driving tests.(7;8;18) The CDR, although found to be useful in determining driving fitness in dementia,(10;11) has not been widely used to predict driving fitness in PD. The literature on the predictive value of motor symptoms to driving outcomes is

9 Devos et al 9 mixed.(7;8;19-25) An evidence-based review found that UPDRS off motor scores were predictive of on-road driving performance (Level B evidence).(3) Motor dysfunction becomes an important predictor when speed of behavior is critical for responding to sudden hazards.(26) We added age of PD onset to the model as it was a significant multivariate predictor of the fitness to drive decision and as progression to disability does not solely depend on disease duration.(27) However, this alternative model did not improve the specificity when sensitivity was kept constant. Thus, we recommend using the original model that has been cross-validated in two different PD populations. The inclusion of the driving simulation increased the specificity of the clinical model, suggesting that driving simulation is strongly associated with actual on-road performance in PD.(28) Yet, the widespread implementation of simulators in driving fitness evaluation is hampered by the limited access to these tools and lack of standardized scenarios across centers. Our clinical battery alone represents a valid tool with strong applicability in daily practice to decide which drivers with PD require more detailed evaluation at specialized centers. ACKNOWLEDGMENT The authors thank the CARA staff (B. De Winde, T. Nijs, M. Strypstein, T. Vanleeuwe, K. Verkammen, and M. Van Weerst), to conduct the fitness to drive assessments and R. De Weze for organizing them. They are also grateful to V. Froyen, who assisted in data collection and statistical analysis as part of her master s dissertation. H. Devos is a Postdoctoral Fellow and W. Vandenberghe is a Senior Clinical Investigator of the Research Foundation Flanders. H. Devos received a fellowship of the Belgian American Education Foundation (B.A.E.F.).

10 Devos et al 10 AUTHOR S ROLES: Authors: Hannes Devos, Wim Vandenberghe, Alice Nieuwboer, Mark Tant, Willy De Weerdt, Jeffrey D. Dawson, Ergun Y. Uc 1. Research project: A. Conception, B. Organization, C. Execution 2. Statistical analysis: A. Design, B. Execution, C. Review and Critique 3. Manuscript preparation: A. Writing of the first draft, B. Review and Critique Hannes Devos: 1. A, B, C 2. A, B 3. A Wim Vandenberghe: 1. B, C 2. C 3. B Alice Nieuwboer 1. A, B 2. C 3. B Mark Tant 1. B, C 2. C 3. B Willy De Weerdt 1. A, B 2. C 3. B Jeffrey D. Dawson 2. A,C 3. B Ergun Y Uc 1. A 2. A, C 3. B FULL FINANCIAL DISCLOSURE OF PRECEDING 12 MONTHS Hannes Devos Stock ownership in medically-related fields Consultancies

11 Devos et al 11 Advisory boards Partnership Honoraria Grants Postdoctoral fellowship Research Foundation Flanders; Travel grant Research Foundation Flanders; Belgian American Educational Foundation Fellowship Wim Vandenberghe Stock ownership in medically-related fields Consultancies Advisory boards Partnership Honoraria Grants Alice Nieuwboer Stock ownership in medically-related fields Consultancies Advisory boards Partnership Honoraria Grants

12 Devos et al 12 Mark Tant Stock ownership in medically-related fields Consultancies Advisory boards Partnership Honoraria Grants Willy De Weerdt Stock ownership in medically-related fields Consultancies Advisory boards Partnership Honoraria Grants Jeffrey D Dawson Stock ownership in medically-related fields Consultancies Advisory boards Partnership Honoraria Reviews for Neurosurgery journal, NIH DSMB Participation Grants NIH-R01-HL082711, NIH-R01-AR054439, NIH-T15-HL097622, NIH-R01-HD062507,

13 Devos et al 13 NIH-R01-HL091917, ACS-RST CPPB, VA Merit Review B6261R, VA Merit Review RX Ergun Y Uc Stock ownership in medically-related fields Consultancies Advisory boards Partnership Honoraria Grants VA Merit Review B6261R, VA Merit Review RX REFERENCES (1) Bacon D, Fisher RS, Morris JC, Rizzo M, Spanaki MV. American Academy of Neurology position statement on physician reporting of medical conditions that may affect driving competence. Neurology 2007; 68: (2) Devos H, Akinwuntan AE, Gélinas I, George S, Nieuwboer A, Verheyden G. Shifting up a gear: considerations on assessment and rehabilitation of driving in people with neurological disorders; An extended editorial. Physiother Res Int 2012; 17: (3) Crizzle AM, Classen S, Uc EY. Parkinson s disease and driving: An evidence based review. Neurology 2012; 79: (4) Devos H, Nieuwboer A, Tant M, De Weerdt W, Vandenberghe W. Determinants of fitness to drive in Huntington disease. Neurology 2012; 79;

14 Devos et al 14 (5) Nouri FM, Lincoln NB. Predicting driving performance after stroke. BMJ 1993; 307: (6) Heikkila VM, Turkka J, Korpelainen J, Kallanranta T, Summala H. Decreased driving ability in people with Parkinson's disease. J Neurol Neurosurg Psychiatry 1998; 64: (7) Devos H, Vandenberghe W, Nieuwboer A, Tant M, Baten G, De Weerdt W. Predictors of fitness to drive in people with Parkinson disease. Neurology 2007; 69: (8) Worringham CJ, Wood JM, Kerr GK, Silburn PA. Predictors of driving assessment outcome in Parkinson's disease. Mov Disord 2006; 21: (9) Classen S, Witter DP, Lanford DN, et al. Usefulness of screening tools for predicting driving performance in people with Parkinson's disease. Am J Occup Ther 2011; 65: (10) Dubinsky RM, Stein AC, Lyons K. Practice parameter: risk of driving and Alzheimer's disease (an evidence-based review): report of the quality standards subcommittee of the American Academy of Neurology. Neurology. 2000; 54: (11) Iverson DJ, Gronseth GS, Reger MA, Classen S, Dubinsky RM, Rizzo M; Quality Standards Subcomittee of the American Academy of Neurology. Practice parameter update: evaluation and management of driving risk in dementia: report of the Quality Standards Subcommittee of the American Academy of Neurology. Neurology.2010; 74:

15 Devos et al 15 (12) Pelli DG, Robson JG, Wilkins AJ. The design of a new letter chart for measuring contrast sensitivity. Clin Vis Sci 1993; 34: (13) Morris JC. The Clinical Dementia Rating (CDR): current version and scoring rules. Neurology 2010; 43: (14) Movement Disorder Society Task Force on Rating Scales for Parkinson's Disease. The Unified Parkinson's Disease Rating Scale (UPDRS): status and recommendations. Mov Disord 2003; 18: (15) Akinwuntan AE, Devos H, Feys H, et al. Confirmation of the accuracy of a short battery to predict fitness to drive of stroke survivors without severe deficits. J Rehabil Med 2007; 39: (16) Archibald NK, Clarke MP, Mosimann UP, Burn DJ. The retina in Parkinson's disease. Brain 2009; 132: (17) Uc EY, Rizzo M, Anderson SW, Qian S, Rodnitzky RL, Dawson JD. Visual dysfunction in Parkinson disease without dementia. Neurology 2005; 65: (18) Uc EY, Rizzo M, Anderson SW, Sparks J, Rodnitzky RL, Dawson JD. Impaired visual search in drivers with Parkinson's disease. Annal Neurol 2006; 60: (19) Radford K, Lincoln N, Lennox G. The effects of cognitive abilities on driving in people with Parkinson's disease. Disabil Rehabil 2004; 26: (20) Zesiewicz TA, Cimino CR, et al. Driving safety in Parkinson's disease. Neurology 2002;59:

16 Devos et al 16 (21) Madeley P, Hulley JL, Wildgust H, Mindham RH. Parkinson's disease and driving ability. J Neurol Neurosurg Psychiatry 1999; 53: (22) Dubinsky RM, Gray C, Husted D, et al. Driving in Parkinson's disease. Neurology 1991; 41: (23) Uc EY, Rizzo M, Johnson AM, Dastrup E, Anderson SW, Dawson JD. Road safety in drivers with Parkinson disease. Neurology 2009; 73: (24) Lings S, Dupont E. Driving with Parkinson's disease. A controlled laboratory investigation. Acta Neurol Scand 1992; 86: (25) Wood JM, Worringham C, Kerr G, Mallon K, Silburn P. Quantitative assessment of driving performance in Parkinson's disease. J Neurol Neurosurg Psychiatry 2005; 76: (26) Uc EY, Rizzo M, Anderson SW, Dastrup E, Sparks JD, Dawson JD. Driving under low-contrast visibility conditions in Parkinson disease. Neurology 2009; 73: (27) Kempster PA, O Sullivan SS, Holton JH, Revesz T, Lees AJ. Relationships between age and late progression of Parkinson s disease: a clinico-pathological study. Brain 2010; 133: (28) Lee H, Falkmer T, Rosenwax L, Cordell R. Validity of driving simulators in assessing drivers with Parkinson s disease. Adv Transport Stud 2007; 4: S81-S90.

17 Devos et al 17 TABLE LEGENDS Table 1. Characteristics of pass and fail drivers Abbreviations: CDR = Clinical Dementia Rating; CS = contrast sensitivity; ESS = Epworth Sleepiness Scale; UPDRS II =Activities of the Daily Living section of the Unified Parkinson s Disease Rating Scale; UPDRS III = Motor section of the Unified Parkinson s Disease Rating Scale. Significant P values are indicated in bold. a = Wilcoxon Rank Sum test; b = unpaired t-test; c = Fisher s Exact test. Table 1. Characteristics of pass and fail drivers Variable Pass category N = 36 Fail category N = 24 P value Clinical screening battery CS, / ( ) 1.65 ( ).0002 a UPDRS III (on), / (19 29) 37 ( ).0008 a CDR, / 3 0 (0 0) 0 (0 0.5).001 a Disease duration, y 5 ( ) 6 (5 8.5).07 a Driving simulation Total simulator score, / ( ) 100 (84 106) <.0001 a Descriptive, medical and driving characteristics Age at onset, y ± ± b Age at moment of testing, y ± ± b Male gender, n (%) 32 (89) 17 (79).46 c Education, y ( ) 12 (12 16).87 a Hoehn and Yahr (on), / 5 2 (1.75 3) 3 (2 3) <.0001 a Anticholinergic medication, n (%) 3 (8) 1 (4).64 c MAOB-inhibitors; n (%) 15 (42) 7 (29).42 c Amantadine, n (%) 4 (11) 2 (8) 1.00 c LED, mg/day 385 ( ) 538 ( ).38 a UPDRS II (on), / 12 9 (4 12) ( ).05 a ESS, / 24 5 (4 6) 4.5 (2 8).67 a Driving experience, y ± ± a Mileage, 10³ km/year 10 (6 17.5) 8 (5 10).83 a Accidents (preceding 5 years), n 0 (0 0.5) 0 (0 1).42 a Traffic tickets (preceding 5 years), n 0 (0 1) 0 (0 0).005 a

18 Devos et al 18 Table 2. Predictive accuracy of the clinical screening battery Official fitness to drive decision Screening Fail Pass Total battery Fail 23 (a) 13 (b) 36 Pass 1 (c) 23 (d) 24 Total Predictive accuracy: ((a + d) / (a + b + c + (d)) * 100= (( ) / ( )) *100= 77% Sensitivity: ((a) / (a + c)) * 100 = (23 / 24) * 100 = 96% Specificity: ((d) / (b + d)) * 100= (23 / 36) * 100= 64% Positive predictive value: ((a) / (a + d)) * 100 = (23 / 36) * 100 = 64% Negative predictive value: ((d) / (c + d)) * 100 = (23 / 24) * 100 = 96%