UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Driving performance in mild dementia Tallman, Karen S. 1992

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1993_spring_phd_tallman_karen.pdf [ 9.18MB ]
Metadata
JSON: 831-1.0086358.json
JSON-LD: 831-1.0086358-ld.json
RDF/XML (Pretty): 831-1.0086358-rdf.xml
RDF/JSON: 831-1.0086358-rdf.json
Turtle: 831-1.0086358-turtle.txt
N-Triples: 831-1.0086358-rdf-ntriples.txt
Original Record: 831-1.0086358-source.json
Full Text
831-1.0086358-fulltext.txt
Citation
831-1.0086358.ris

Full Text

to the required standardDRIVING PERFORMANCE IN MILD DEMENTIAbyKAREN SHEPARD TALLMANB.Sc., The University of British Columbia, 1976M.A., The University of British Columbia, 1986A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIES(Department of Psychology)We accept this as conformingTHE UNIVERSITY OF BRITISH COLUMBIANovember 1992© Karen Shepard Tallman, 1992In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature) Department of `r.. C^yThe University of British ColumbiaVancouver, CanadaDate 1\hq ‘;‘) L^19(12DE-6 (2/88)iiABSTRACTThe automobile driving performance of 18 mildly demented subjects was compared to thatof 18 normal elderly and 18 mid-age controls. Driving tasks were grouped according to athree-level hierarchical model that characterizes driving as a problem solving skillinvolving (a) low-level vehicle control skills, (b) intermediate level manoeuvring skills inresponse to on-road events, and (c) higher level driving-related judgemental abilities.Driving simulator measures of brake reaction time and steering accuracy were selected torepresent the lowest level of hierarchy. A Motor Vehicle Branch (MVB) road test and atest of emergency braking distance indexed the intermediate level. Driving-relatedjudgment was assessed by evaluating the accuracy of subjects' appraisals of their owndriving skills, and by examining whether the demented subjects evidenced an increasedlevel of driving avoidance that might be commensurate with the extent of their drivingdeficits. Overall the demented subjects performed significantly less well than did controlson the driving behaviour measures. Particularly striking were their deficits at the highestand lowest levels of the hierarchical task analysis. They were markedly impaired on thedriving simulator tasks and they showed a clear tendency to over-estimate their drivingcompetence relative to their actual performance. However, from a practical standpoint itwas noted that despite significant group differences at the intermediate level, the overlapin performance scores between the demented and the normal elderly was considerable forthese in-car tasks. Also, although the mildly demented subjects had significantly moredemerit points on the MVB road test than did the elderly controls, nearly 70% were ableto pass the licensing exam. Mildly demented drivers might best be characterized as111having marginal driving abilities, a fact which may pose considerable challenges toclinicians and policy makers. A second component of the study involved evaluation ofthe correlations between the driving measures and several common psychometric tests ofattention, perception, and psychomotor speed. After group membership was accountedfor, the psychometric tests failed to add precision to the prediction of driving performance.ivTable of ContentsAbstract^  iiList of Tables  viiiAcknowledgment^Introduction 1Component one: Performance-based measurement^ 4Component two: Psychometric prediction of drivingperformance^ 4Funding and administration of the study^ 5Literature Review^ 7The Practical Context: Why Study the Driving Performance of Those withDementia?^ 7Demographic trends^ 7Needs of clinicians: Fulfilling ethical and legalresponsibilities^ 8Retrospective investigations of the accident rates of dementeddrivers^ 11Performance based research with dementing drivers^ 13The Conceptual Context: Approaches to UnderstandingDriving Behaviour^ 17Models of the Driving Task^ 18Human-factors and information processing approaches to the drivingtask^ 19Cognitive and hierarchical models of driverbehaviour^ 20Models of Driver Behaviour and their Implications for Evaluation of Normaland Demented Elderly Drivers^ 22Are the Normal Elderly Competent Drivers? : Comment on the Choice of aComparison Group for Dementing Drivers^ 29Studies of Brain Damaged Drivers: Examples of Psychometric Predictionof Driving Ability^ 33Method^ 41Subjects 41Diagnostic procedures: Overview^ 42Characteristics of the Demented Sample and the Control Groups^43Measures^ 48Descriptive Measures^ 49Driving Interview 51Motor Vehicle Branch Driving Record^ 53Visual acuity^ 54Driving Behaviour Measures^ 54Operational level measures 55Manoeuvring level measures^ 57viStrategical level measures^ 62Psychometric Measures 65Procedures^ 72Procedures for Older Participants^ 72Procedures for Mid-Aged Participants 74Results^ 76Descriptive Measures^ 76Driving history 78Driving exposure^ 79Motor Vehicle Branch Driving Records^ 89Visual acuity^ 89Driving Behaviour Measures^ 92Operational level measures: Group differences^ 93Manoeuvring level measures: Group differences 95Strategical level measures: Group differences^ 97Distinguishing between normal elderly anddementing drivers^ 104Sex differences on the driving behaviour measures^ 113Relation Between Psychometric Tests and Driving Measures 114Discussion^ 128Group Differences in Driving Behaviour^ 128Group Differences: Background Issues 129viiGroup Differences: Subject Characteristics and Descriptive Variables^ 134Group Differences: Driving Behaviour Measures^ 138Operational level^ 139Manoeuvring level 141Strategical level^ 143Psychometric Prediction of Driving Behaviour^ 152The Conceptual Context: How Fares the Hierarchical Model?^ 153The Practical Context: Where to From Here?^ 156References^ 164Appendix A: Subject Selection Procedures and Diagnostic Criteria^ 174Appendix B: Driving Interview Questions Phrased for Interviewof Collaterals^ 204Appendix C: Procedures for Scoring Driving Avoidance and DrivingProblem Questions from the Driving Interview^213Appendix D: Information on Selected Driving Behaviour Measures^228Appendix E: Copy of Consent Form^ 232viiiList of TablesTable 1: Demographic Information for Dementia and Control Groups^44Table 2: Health Problems of Dementia and Control Groups^46Table 3: Prescription Medications used by Dementia and Control Groups^47Table 4: Study Measures^ 49Table 5: Driving Exposure: Miles in Previous Year^ 80Table 6: Driving Exposure: Trips per Week 83Table 7: Driving Exposure: High Risk Situations^ 85Table 8: Driving Exposure: Avoidance of Driving Situations^88Table 9: Motor Vehicle Branch Record of Tickets andAccidents for Previous Five Years^ 90Table 10: Visual Acuity for Dementia and Control Groups^ 91Table 11: Driving Behaviour: Operational Level^ 94Table 12: Driving Behaviour: Manoeuvring Level 96Table 13: Driving Behaviour: Strategical Level - Means and StandardDeviations for Cone Avoidance Task^ 98Table 14: Driving Behaviour: Strategical Level - Difference Scores betweenPredicted Hits and Actual Hits on the Cone Avoidance Task^ 101Table 15: Driving Behaviour: Strategical Level - Participant and CollateralRatings of Driving Problems^ 103Table 16: Results of Discriminant Analysis with Driving Behaviour Measures^ 106ixTable 17: Proportion of Subjects with Scores Falling in the Extreme Regionsof the Score Distributions for the Normal Elderly^ 110Table 18: Mean Performance Scores on Psychometric Tests 115Table 19: Correlations Between Psychometric Tests and DrivingMeasures Within the Demented Group^ 119Table 20: Correlations Between Psychometric Tests and Driving Measuresin Combined Group of Older Subjects and Point BiserialCorrelations of Group Membership with Driving Measures^ 123Table 21: Part Correlations Between Psychometric Tests and Driving Measures^ 126AcknowledgementThe research reported in this thesis would not have been possible without thesupport and enormous goodwill of many individuals and several organizations. First,without Dr. B. Lynn Beattie this study would never have happened. Through Dr. Beattieand the Clinic for Alzheimer Disease and Related Disorders, the Driving and Aging Studywas funded, encouraged, and given a home base. Also, I am grateful to Dr. Eric Eich, mydepartmental supervisor, and to Dr. Holly Tuokko, Supervising Psychologist at theAlzheimer Clinic, for their guidance and support. Financial support for the Driving andAging Study was provided by Health and Welfare Canada (NHRDP), the AlzheimerSociety of British Columbia, the Insurance Corporation of British Columbia, and theMotor Vehicle Branch of British Columbia. The Alzheimer Society of Canada kindlygranted me a two year Doctoral Training Award to undertake this work.While the research report that follows tells one of the stories associated with theDriving and Aging Study, the other story is about the generosity and commitment of thesmall army of people who contributed to the study. I regret that I can do little more thanhint at the human side of this project. First, I wish to thank the subjects for their patienceand, in many cases, courage in the face of a daunting amount of testing. Also, Iacknowledge substantial assistance in the construction of the off-road driving track fromMr. Mike McNabb and members of the UBC Accident Research team headed by Dr.Frank Navin. Mr. Al Lund, director of Pacific Traffic Education Centre, is warmlythanked for his part in the design of the Cone Avoidance task and for his enthusiasticsupport when I most needed it. Also, I am grateful to Mr. Lund and to Sergeant WarneLynd who each gave up many of their Sundays to man the dual-controls for theemergency braking tests. Despite the hectic schedule, testing days at GF StrongRehabilitation Centre were a pleasure and for this I especially want to thank RalphCheesman, Lana Gudeit, Erin Harris, Spencer McDonald, Dr. Joann Miller, FredVandenboer, and David Wong. Psychological testing was carried out by a group of UBCgraduate students to whom I am truly grateful--thank you one and all! I would also liketo acknowledge the following friends of the Driving and Aging Study: Don Chaput, PeterCooper, Tim and Kathyrn Dunne, GF Strong Rehabilitation Society, Dr. Charles Laszlo,Bev McClean, Silent Witness, Scottish Line Painters, John Wong, and the staff at theAlzheimer Clinic. I note with sadness the passing of Mr. Des Vosper, the former ChiefExaminer for the Motor Vehicle Branch. Mr. Vosper designed the test course for theMVB road test and his support for the study lent credibility to our early negotiations forthe resources needed to carry out this project.1INTRODUCTIONIn North America the automobile has become an omnipresent fact. For enormousnumbers of citizens access to basic necessities requires the mobility afforded by theprivate car. Recent U.S. census information indicates that people over 60 rely on privatevehicles for more of their transportation needs than do those aged 16 to 60 (Rosenbloom,1988). Accident statistics show that although the aged have fewer accidents than doyounger age groups, they also drive significantly fewer miles. When accident rates arecalculated on a per-mile basis, the relation of age to accidents is described by a classic Ushaped curve: the young beginning driver and the aged driver have the greatest number ofaccidents per mile driven (McFarland, Tune, & Welford 1964; Waller,1988).Many factors underlying these statistics need to be studied. It has been repeatedlydemonstrated that age alone is an unreliable predictor of driving ability (Waller, 1988).Traffic safety experts believe that debilitating medical conditions associated with agingmay be an important factor in the increased accident risk of the elderly. Thus, one usefulstrategy for achieving a more specific understanding of the driving problems of the agedis to identify and study potentially high-risk groups of elderly drivers.Dementia is a condition with a strong age-related incidence. About 5% of thoseover age 65 are believed to suffer from severe dementia and estimates of the prevalenceof mild dementia in those over age 65 range from 3 to 15% (Katzman, 1985; Mortimer,Schuman,& French, 1981). In the later stages of a dementing disorder there is devastatingfunctional deterioration at all levels of behaviour. There is every reason to expect that2those with significant global cognitive impairment would be a great danger to themselvesand others if they were to drive. However, most moderately to severely dementedindividuals are too incapacitated to even attempt to operate a car.In the early stages of a dementing disorder the situation is not so clear cut.Alzheimer Disease, which accounts for 50 to 70% of dementing disorders, has aninsidious onset that may not be apparent to patients or their family members for aconsiderable period of time. Albert (1981, pp. 844-845) makes an important point whenshe states:Patients being evaluated for Alzheimer's Disease are mostoften in the early stage of the illness....The most strikingthing about these individuals is that to the casual observerthey appear normal. Their social skills are maintained andtheir behaviour is generally appropriate.Experience at the University Hospital Clinic for Alzheimer Disease and RelatedDisorders and other research centres suggests that 50% or more of individuals who wereregular drivers prior to the onset of their illness will still be driving 2 to 3 years after theappearance of their first symptoms (Carr, et al. 1991; Friedland et al. 1988; Gilley et al.1991; Lucas-Blaustein, Filipp, Dungan, & Tune, 1988)The problem of assessing the driving competence of an older and potentiallydementing driver is particularly relevant to two different groups: licensing authorities whoare under increasing pressure from the courts to defend their licensing practices (Waller,1988), and clinicians who have ethical and legal responsibilities to protect and report unfitdrivers. The information most relevant to each of these groups derives from compatiblebut somewhat different approaches to the investigation of cognitively impaired older3drivers: the driving performance based approach, and the psychometric screeningapproach.The former approach emphasizes the testing of actual driving ability. Research isneeded to determine the particular nature of the driving difficulties experienced bydementing drivers, and to identify on-road or simulator measures that are helpful indetecting individuals who are at risk when driving. One practical goal of theperformance-based approach is the development of specialized driving tests that are bothsensitive and specific to the driving problems of cognitively impaired older drivers. Werethere a desire on the part of licensing administrators to implement specialized testingprocedures for high-risk older drivers, this approach would generate the neededperformance-based measures of driving skill.The psychometric-screening approach involves the estimation of potential drivingrisk from an individual's performance on tests that have a demonstrated association withdriving disability. The screening approach is particularly attractive to clinicians who mustmake recommendations to licensing authorities about their patients' fitness to drive. Werescreening measures available that could be readily administered in the office, clinicianswould have an objective means of determining whether their patient should be referred tothe Motor Vehicle Branch (MVB) for further assessment. Commonly used psychometrictests of perception, attention, and psychomotor speed are plausible candidates as predictorsof driving performance.The development of technologies for fair and valid assessment of older drivercompetence is a long range, and still distant goal. Work on this problem has only just4begun. Data characterizing demented drivers are scant and so far published studies havebeen limited to retrospective examinations of accident rates and driving patterns.However, a handful of recent conference presentations herald the initiation of researchprograms aimed at prospective measurement of driving behaviour and its putativecorrelates. The most rudimentary foundations for future research are still requiredincluding the empirical demonstration of measurable differences between the drivingperformance of mildly demented individuals and normal elderly persons. At this point intime the contribution of any individual study can at best be a step in the direction ofclarifying the conceptual, methodological, and ethical issues that must be confronted asour understanding progresses. It is in this context that the present study was undertaken.The project has two components corresponding to the above discussion ofperformance-based measurement and psychometric prediction of driving performance.Component one: Performance-based measurement.  The central aim of this projectwas to determine whether there were measurable deficits in the driving performance ofmildly demented drivers relative to normal elderly controls. Experimental demonstrationof impaired driving competence is considered to be a necessary first step in formulatingresponsible policy about cognitively impaired older drivers (Drachman, 1988).In accordance with the multi-dimensional nature of driving, five measures ofdifferent facets of driving behaviour were included. Also, a mid-age control group wasincluded in the study to assist in detecting tests with minimal age-effects.Component two: Psychometric prediction of driving performance.  The secondcomponent of the project was directed toward exploration of correlations between5psychometric measures that are sensitive to dementia and the driving performancemeasures. Psychometric tests of attention, perception, everyday functional ability andjudgement, and psychomotor speed were selected on the basis of both previous researchon psychometric prediction of driving performance in brain damaged individuals, and onthe suggestions of researchers concerned with developing strategies for assessing impairedolder drivers (Hopewell & van Zomeren, 1990; Kaszniak, Keyl, & Albert, 1991).Participants in the study consisted of 18 older drivers who suffer from milddementia, 18 age-matched cognitively intact subjects, and 18 mid-aged drivers.Participants were tested on a driving simulator, a standard Motor Vehicle Branch roadtest, and at an off-road driving circuit where emergency braking skills and vehiclemanoeuvring skills were tested. All subjects were assessed on the battery ofneuropsychological tests and functional rating scales used for evaluation of dementia atAlzheimer Clinic, and on a battery of putatively "driving-related" psychometric measuresof attention, perceptual functioning, psychomotor speed, and functional status. In addition,each subject and a collateral were interviewed about the participants' driving habits andproblems.Funding and administration of the study. Data for this dissertation were collectedin conjunction with the Driving and Aging Study--a three year project administeredthrough the Clinic for Alzheimer Disease and Related Disorders at University Hospital,UBC Site, Vancouver, British Columbia, and funded by the National Health Research andDevelopment Program (NHRDP). I am a Research Associate at the Clinic, and in thatcapacity I designed the Driving and Aging Study. In collaboration with Dr. B. L. Beattie6I wrote the application for funding and have been responsible for overseeing allcomponents of the project. In August 1991, the Alzheimer Society of British Columbiaawarded me the funds necessary to add the mid-age control group.Data collection for this study was extensive involving neuropsychological andpsychometric assessment, road testing, driving simulator assessment, and evaluation ofemergency stopping skills at the off-road track. Several PhD level psychology graduatestudents supervised by Dr. H. Tuokko and myself were employed to undertake thepsychometric assessment. Examiners from the B.C. Motor Vehicle Branch and the B.C.Safety Council were hired to conduct the MVB Road Tests. Assessments on the drivingsimulator were conducted by students from the School of Rehabilitation Medicine. Thesenior driving instructor for the RCMP driver training program and the Coordinator ofdriver training programs for the Justice Institute were hired to oversee the dual-controls inthe test vehicle during the emergency braking tasks. Members of the UBC AccidentResearch Team headed by Dr Frank Navin were instrumental in the design andconstruction of the test equipment used at the off-road track.7LITERATURE REVIEWThe literature review for this project comprises three sections: The Practical Context;the Conceptual Context; and Studies of Brain Damaged Drivers.The Practical Context: Whv Study the Driving Performanceof Those with Dementia? Demographic trends. It is common knowledge that the population of NorthAmerica is aging. In 1941, about 7% of Canadians were over age 65. By 2025, theproportion will be close to 19% (Gutman, Gee, Bojanowski, & Mottet, 1986). Also thereis aging within the over 65 population itself. Gutman et al. (1986) project that in BritishColumbia by 2001, there will a 70% increase in the population aged 65 and over, and a128% increase in the group aged 80 and over. Dementia has a strongly age-relatedincidence. Given the proportionally greater increase in the growth of the very-oldpopulation, there is an expectation that the prevalence of dementia in Canada will increasefrom about 5.6% of those over age 65 (or 132,000 cases) in 1981 to 7.4% (or 324,000cases) in 2006 (McEwan, Donnelly, Robertson, & Hertzman, 1991).There is a clear trend towards greater suburbanization of older citizens with anassociated increase in reliance on the private automobile for access to needed services.Rosenbloom (1988, p. 24) made the following statement:The aging of society raises significant questions aboutmobility and transportation of the elderly. Strikingly, almost8all of the "new" old will drive cars; the majority will live insuburban or relatively low-density urban settings. Thesepeople will have made a variety of decisions, and structuredtheir social and economic lives, in response to their lifelongaccess to the private carMost experts agree that over the next few decades there will be significantly largernumbers of drivers in the older age groups. In this older population, the high prevalenceof debilitating age-related disorder such as dementia may result in a growing number ofimpaired older drivers. However, policy makers caution that the introduction of arbitrarilyrestrictive licensing procedures for older drivers would result in reductions in the mobilityand self-sufficiency of older citizens which would be extraordinarily costly for society.Rosenbloom (1988) argues that the goal of responsible public policy must be thepromotion of maximum mobility for older persons, with concomitant concern for thesafety of the older drivers themselves and the rest of the public. The development of teststhat are valid indicators of driving risk in the elderly would be a step in the rightdirection.Needs of clinicians: Fulfilling ethical and legal responsibilities. In geriatricassessment and treatment settings, the question of whether an elderly person is sufficientlycognitively intact to continue to drive is frequently encountered and often presentsdifficult ethical and legal problems for practitioners. This issue has recently become thesubject of several journal articles and conference presentations by geriatricians,psychologists, occupational therapists, and lawyers (Carr et al., 1991; Coopersmith,Korner-Bitensky, & Mayo, 1989; Drachman 1988; Kaszniak, Keyl, & Albert, 1991).The ethical codes for both physicians and registered psychologists include the duty9to warn patients of foreseeable danger to themselves or others should they continue todrive in the presence of a condition known to have the potential to impair driving ability.In addition to ethical guidelines, many jurisdictions specify legal responsibilities forpractitioners. In British Columbia, the legal responsibilities of physicians, registeredpsychologists, and optometrists are outlined in the Motor Vehicle Act. They must beaware of conditions that may interfere with their patients' abilities to safely operate amotor vehicle, counsel those at risk not to drive, and report those who disregard thewarning to the Motor Vehicle Branch (MVB). Also, in B.C. (and some other Canadianprovinces) physicians play a central role in the relicensing of elderly drivers who require amedical certificate to renew their drivers licence when they reach age 75, 80, and everytwo years thereafter.Although the MVB has final responsibility in licensing decisions, a practitionerfaced with a potentially driving impaired patient must make an immediate decision aboutwhether to direct their patient to stop driving because of potential for harm. In the caseof routine medical assessments for fitness to drive, physicians must make a judgementabout whether to recommend licence renewal. The legal consequences of failure toexercise care in these decisions can be significant. Recent court decisions in the U.S.have found physicians liable for failing to warn patients suffering from disorders such ascongestive heart failure, history of stroke, and epilepsy of foreseeable danger whileoperating a motor vehicle (Coopersmith et al., 1989).The decision to recommend the removal of driving privileges from a moderately orseverely demented individual is straight forward; indeed, it is rare for these individuals to10still be able to drive. The more difficult scenario involves an individual with mildsymptoms of cognitive decline, often only obvious in the area of new learning ability,who insists that their driving skills are still intact. At present, clinicians have very littleinformation to assist them in assessing the extent to which these patients are more at riskthan the average elderly driver. For an issue as sensitive as recommendations affectingdriving privileges it would, at the very least, be helpful to have research documenting theoccurrence of specific driving problems in this group. Even more useful would bevalidated screening measures that would provide clinicians with objective grounds forreferring older drivers to the MVB for further consideration. Coopersmith et al. (1989, p.378) make the following recommendations:There is clear need for the development of objective,rigorous and scientific measures of driving ability that permitthe screening of patients whose driving abilities are difficultto ascertain, such as those with functional, cognitive, andperceptual dysfunctions commonly associated withAlzheimer's Disease, stroke, and head injury. Such measureswould help physicians judge the severity of impairment andwould diminish the subjective component of determiningmedical fitness to drive.To date the bulk of publications related to dementing drivers are position papers--outlining legal and ethical issues, the need for future research, and often providingrecommendations for clinicians (for e.g., Carr et a. 1991; Coopersmith et al., 1989;Drachman, 1988; Donnelly & Karlinsky 1990; Graca, 1986; Hopewell and van Zomeren,1990; Kaszniak, Keyl, & Albert, 1991; Parasuraman & Nestor, 1991; Reuben, 1991). Inaddition, over the past two years an assortment of empirical studies of dementing drivers(mostly pilot projects) have been presented at conferences (Coyne, Feins, Powell, &11Joslin, 1990; Hunt, Edwards, Morris, & Mui, 1990; Odenheimer, Beaudet, Grande, Albert,Jette, & Minaker, 1990; Keyl, Rebok, Bylsma, & Rodman, 1990; Wild, Kaye, &Campbell, 1991), or have been submitted for publication (Cooper, Tallman, Tuokko, &Beattie, in press; Logsdon, Teri, & Larson, submitted; Tallman, Cooper, & Beattie, inpress). Although there soon will be a larger and more diverse literature on dementingdrivers, the published studies that are currently available involve retrospective analysis ofdata obtained from interviews with caregivers.Retrospective investigations of the accident rates of demented drivers. Of the fouror so available articles that present substantive data about dementing drivers (Friedland etal., 1988; Gilley et al. 1991; Lucas-Blaustein, Filipp, Dungan, & Tune, 1988; and Waller,1967 cited in Kaszniak et al. 1991), far and away the most influential and controversialhas been that of Friedland et al. (1988).Friedland and colleagues investigated a group of 30 patients with probableAlzheimer Disease (AD) of about 5.5 years duration that were part of a National Instituteof Health longitudinal study. A control group comprised twenty carefully screenedindividuals who were healthy with no evidence of medical or psychiatric disease onhistory, physical exam, or laboratory results. Although a variety of parameters wereinvestigated, the central focus of the study was to compare the demented and healthyelderly groups with respect to the frequency of traffic accidents for the preceding fiveyears. Information about accidents was based on caregiver report. They found thatwhereas 47% (n=14) of the 30 drivers with AD had been in an accident in the precedingfive years, accidents had occurred for only 10% (n=2) of the control subjects in the same12time period. Friedland et al. (1988) concluded:We recommend that patients with the diagnosis of [ dementiaof the Alzheimer type ] not drive a motor vehicle. (p. 785)Another published estimate of the accident risk for drivers with dementia isavailable from Lucas-Blaustein et al. (1988). They retrospectively evaluated 53 dementiacases and found that 31% had been in an accident in the preceding five-year period; theydid not have a control sample of normal elderly. They also concluded their article withthe categorical recommendation that individuals diagnosed with dementia should not drive.Reactions to the recommendations of Friedland et al. (1988) have been mixed.The study has been criticized for overgeneralizing in the face of small samples and acontrol group of high functioning individuals who are unlikely to represent average elderlydrivers (Carr, et al., 1991; Kaszniak, et al. 1991). The most vehement reaction to theFriedland et al. paper comes from Drachman (1988). He asserts that it is a mistake toutilize diagnostic status as the criteria for categorically prohibiting AD patients-fromdriving. He argues that the diagnostic criteria for Alzheimer Disease are still evolvingand asks:how subtle a decline in memory or cognitive ability will besufficient to diagnose "possible AD"? Will individuals whoseWechsler Adult Intelligence Scale scores have fallen from140 to 120 lose the right to drive? If peripheral markers cansuccessfully identify the accumulation of increased numbersof neuritic plaques and neurofibrillary tangles, Alz 50antibodies, or amyloid deposits, will patients with thesemarkers be deprived of their licenses? (p. 787)Drachman then goes on to state:In the case of medically impaired drivers, I believe that13limitation of the privilege to drive should be based ondemonstration of impaired driving competence, rather than astigmatizing label such as Alzheimer Disease. The real issueis whether patients with dementia and driving incompetencecan accurately be detected and prevented from driving.Failing the ability to distinguish between competent andincompetent drivers, other issues will certainly arise...Here,the need is for specialized testing in an appropriateenvironment-preferably a driving test, or a simulator that canpresent numerous situations that make demands on themechanical operation of the vehicle, emergency responses,night driving, the ability to interpret signs and signals, etc.(p. 787)Recent papers continue to echo the concerns expressed by Drachman (Carr et al.,1991; Kaszniak et al., 1991; Owsley, Ball, Sloane, Roenker, & Bruni, 1991). Whileundoubtedly the examination of accident rates is a critical first step in characterizing thedriving problems of dementing patients, unlike Friedland et al. (1988) and Lucas-Blausteinet al. (1988), most investigators do not feel that recommending the categorical removal ofthe driving privileges of dementing individuals on the basis of the currently availableinformation about their accident rates does justice to the problem.Performance based research with dementing drivers. In the past few years anumber of research groups across North America have initiated performance-basedinvestigations of dementing drivers in order to augment the information available from theretrospective studies. To date four preliminary reports of projects that have utilizeddriving performance measures have been presented at conferences and summariespublished as abstracts (Fitten et al., 1991; Hunt et al., 1990; Keyl et al., 1990;Odenheimer, et al. 1991). All principal authors were contacted by mail and a writtenversion of their presentation was requested. One group provided the text of the talk from14which the published abstract was derived.A copy of the 1990 presentation by Keyl et al. was made available by the authors.The investigators utilized a driving simulator that monitored subjects' responses inbraking, steering, acceleration, speed, signalling, reaction time, and cue recognition asparticipants responded to two filmed roadway sequences.Error rates for the various performance categories were compared betweengroupings within a demented sample (n=23) and groupings within a normal elderly sample(n=37). The demented sample was dichotomized into those who had experienced 2 ormore near misses (i.e., near accidents) in the previous two years versus those with 1 or 0near misses as reported by an informant. The 5 demented subjects who had experienced 2or more near misses did significantly less well than the 18 demented subjects with 1 or 0near misses with respect to the number of braking errors. The two groups of dementedwere also described as differing (2 = .066) for total score on a city-driving film sequence.The investigators also dichotomized the demented sample with respect to caseswith one or more accidents versus those with no accidents in the preceding two years.They report that there was a significant difference (2 = .063) in the cue recognition scoresof the two demented groups such that the 4 demented subjects who had been in 1 or moreaccidents did less well than the 19 demented subjects with no accidents. The authors didnot present comparisons of the normal controls and the demented subjects for either thesimulator tasks or for accident rates or near misses although cursory inspection of the datasuggests that the groups may have differed with respect to proportion of subjects with 2 ormore near misses but not for accident rates.15While it would be very useful to have tests that are effective at distinguishingdemented individuals who are likely to be involved in accidents or near misses fromdemented cases who are not, the Key! et al. study has set a difficult goal for itself.Despite one statistically significant difference and two near significant results within thedemented group, one wonders about the reliability of the findings. Problems arise whenaccident rates are utilized on an individual basis as the criterion for driving competence(unless an individual has had several accidents, which is the exception even for quiteimpaired drivers). Also, the use of accidents is problematic when, as was the case for theKey! et al. study, the demented and the normal elderly do not differ with respect toaccident rate. Accidents are multi-determined, semi-random events and difficult to predicton an individual basis (see McKenna, Duncan, & Brown, 1986 for a clear statement of theissues involved in attempting to find associations between measures of behaviour andaccident rates). Near misses may be an improvement over accidents as a measure ofdriving competence, however one can imagine myriad problems with defining a near missand with the use of caregiver estimates or with self-report of these events.Hunt et al. (1990) present a comparison of three groups of drivers; mildlydemented subjects, very mildly impaired subjects diagnosed as questionable seniledementia of the Alzheimer type, and normal elderly controls. They found that whereas allthe normal controls (n=18) and all the subjects in the questionable dementia group (n=14)were judged capable of driving on the basis of a global rating on a one hour road test,64% (n=9) of the mildly demented group were not considered competent to remain on theroad. They concluded that a minority of senile dementia of the Alzheimer type patients16retain adequate driving skills.Fitten et al. (1991) reported preliminary data from a study that included a group ofdemented patients (n=8), a group of diabetic patients (n=5), and a group of healthy elderlycontrols (n=21). Subjects were tested on a specially designed road test. The investigatorspresent the average driving test scores for the three groups which show that the dementedhad a lower average (M = 20.8) than did the two comparison groups VI = 32.6 for thenormal controls and M = 33.8 for the diabetic group), however they do not state whetherthe differences between groups were statistically significant. They do report a statisticallysignificant difference in driving performance between the 3 least impaired dementedsubjects and 3 of the diabetic subjects matched to the demented group for Mini MentalState Exam (MMSE) scores. They conclude that significant driving impairment may bepresent in early Alzheimer Disease and that MMSE scores may not be helpful in detectingthose with driving problems.These investigators utilized an interesting methodology by matching mildlydemented subjects with diabetic subjects on MMSE scores. The sample sizes are toosmall to warrant strong conclusions but the method holds promise. It would be extremelyuseful to determine whether mild deficits as measured by mental status screeninginstruments are indicative of driving risk in non-demented populations of older drivers.Odenheimer et al. (1991) investigated a group of subjects described as having "arange of cognitive abilities" (p. 84) though they do not specify diagnoses nor the ranges ofscores on measures of cognitive functioning. They examined the bivariate relationshipsbetween cognitive tests and a road test and found several significant correlations. The17correlations between a road test summary score and the following psychometric measureswere: Mini Mental State Exam, r = .72; Boston Naming Test, r = .52; Trail Making testPart A, r = .52; and Trail Making test Part B, r = .45. Unfortunately it is difficult toevaluate the meaning of these correlations in the absence of information about thecomposition of the group from which they were derived. We do not know if the overallgroup was composed of two samples, one normal and one demented, or whether theinvestigators were able to obtain a representative sample of the elderly population. Alsowe do not know if the subjects with cognitive impairment were selected for participationin the study on the basis of the cognitive tests.In summary, there are now a small number of preliminary reports of investigationsthat have involved performance-based measurement of cognitively impaired drivers. Theproject described in this thesis has goals that overlap with a number of these studies.Unfortunately, until these investigations are published as articles with details of the studymethods, it will be difficult to evaluate the actual similarities between the projectdescribed in this document and those of other research groups.The Conceptual Context: Approaches to Understanding Driving Behaviour Driving an automobile occurs in the complex context of the driver-vehicle-roadwaysystem. Literature in the field of traffic safety is currently divided into methodologicallydistinct areas of driver behaviour, vehicle dynamics, and roadway engineering.18This discussion is concerned with issues related to the study of driver behaviour.As will be seen in the following review, those who study driver behaviour are currentlygrappling with difficult and fundamental conceptual problems related to specifying whatshould be studied and how to study it. This creates a challenging situation for researchersconcerned with the driving problems of high-risk groups such as demented or head injureddrivers. There are a variety of opinions about what constitutes an appropriate (orinappropriate) approach or model for investigating the driving skills of these drivers.Notwithstanding the assertions of proponents of a particular viewpoint, available data areinsufficient to demonstrate the superiority of one approach over another. For the presentstudy, two approaches were particularly relevant: the perceptual-motor skill model basedin the human factors and information processing traditions, and the hierarchical model thatemphasizes higher order problem-solving behaviour. Fortunately, there is no imperative tochoose one approach to the exclusion of the other. Indeed, the variables of interest in theskill model, such as perception, attention, and psychomotor skills, are embedded withinthe hierarchical approach.Models of the Driving TaskAnalysis of the driving task, or what is more recently referred to as modelling ofdriving behaviour, has presented many conceptual challenges, and remains a topic oflively discussion (Fuller,1984; Kramer & Rohr,1982; Michon, 1988). At issue is thedevelopment of a comprehensive description of the task demands and driver behavioursthat constitute competent operation of a vehicle.For purposes of this review, it is noted that much of the literature on the driving19task can be roughly divided between two paradigms: (a) the human factors andinformation processing approaches; and (b) the more recent psychologically basedmotivational, cognitive, and hierarchical models. These paradigms differ in terminology,focus, and application, as will be apparent from the quotes which follow.Human-factors and information processing approaches to the driving task. Themajority of research involving on-road driving performance is in a human-factorstradition. This research uses what Summala (1985) terms the "skill model of driverbehaviour" which presumes that driving is primarily a perceptual-motor skill and thataccidents are largely a result of failures in this skill. In the 1970's researchers in the areaof human factors began to use information processing terminology in discussing thedriving task. Shinar's (1978) text Psychology on the Road summarizes much of theAmerican human-factors work and provides the following descriptions of the driver andthe driving task:In summary, the driver can best be thought of as aninformation processing channel in which the central decision-making component is rate limited, and the peripheralattentional and perceptual mechanisms function to select themost important cues to be processed by the central decisionmechanism. (p. 71)andMost of overt driving behaviour can be categorized asbelonging to one of two categories: lateral control throughsteering and longitudinal control through acceleration,deceleration, and braking. In addition to the hand movementsinvolved in steering and the foot movements involved inbraking and acceleration, there is also the movementinvolved in shifting the right foot from the accelerator to thebrake pedal or from the brake pedal to the accelerator pedal.(p. 92)20This view of the driver and the driving task is embedded in the selection ofresearch topics and methods in the human-factors approach. Considerable information hasbeen collected about drivers' competencies in visual search tasks, perception of roadwaygeometry, judgement of distance and speed, reaction time under various roadwayconditions, and motor-response capability for lateral and longitudinal control (Shinar,1978). This work is often carried out with single subjects or small groups and may findapplication by traffic engineers in road-way and traffic sign design.Cognitive and hierarchical models of driver behaviour. There has been increasingdissatisfaction among some traffic-safety researchers with the perceptual-motor skill modelof the driving task. Naatanen and Summala (1976, p. 36) state:A profound misunderstanding of the basic nature of thedriver's task by many workers in the field has led research infruitless directions: little attention has been paid to thedriver's ability to compensate for changes in the degree ofdifficulty of traffic situations by modifying his efforts(attention, vigilance), or even to the driver's ability todetermine the nature and degree of the difficulty of thesevarious situations.... Driving indeed should not be understoodas involving a forced-pace task in which the driverprincipally has only a responsive role in his interaction withthe traffic situation; instead his active role and initiative onthe road should be given sufficient notice. Such a bias hasled to considerable over-emphasis on the perceptual-motoraspects of the driver's performance, both in research and inattempts to prevent accidents, at the expense of the muchmore important motivational and cognitive aspects of driving.In response to the unidimensionality of the perceptual-motor skill models, anumber of risk perception models and hierarchical task analyses were introduced that haveattempted to describe the driving task in terms of a larger context that includes drivers'goals and problem solving tactics (Groeger, 1987; Michon, 1985).21A useful general description of the drivers' task that incorporates the notion of thedriver as an active problem solver was introduced in 1971 by Michon and is referred to asthe hierarchical model. The aspects of Michon's task description that are relevant to thepresent research concern what Michon (1985) terms "the hierarchical structure of the roaduser task". He claims that driving is best conceptualized as a problem solving task thatcan be divided into three hierarchically structured levels of behaviour: strategical, tactical,and operational. The strategical is the highest level of Michon's hierarchical task description and isthought of as the level where goal formulation and planning take place, usually prior toactual driving. Strategical level activities include the determination of trip goals, selectionof route, consideration of traffic conditions and weather, and evaluation of costs and risksassociated with particular driving goals.The manoeuvring level occupies an intermediate position in Michon's hierarchy,and involves exercising manoeuvring control to allow negotiation of the vehicle within theprevailing circumstances on the road. The driving task at the manoeuvring level involvesdecisions and behaviours related to interacting with other vehicles and the trafficenvironment such as obstacle avoidance, speed adaption, turning, lane selection, andovertaking.The lowest level of the hierarchy, the operational or control level, involvesperceptual processing of the driving environment and the various behaviours required tocontrol the movement of the vehicle such as steering, braking, and accelerating. Theoperational level in Michon's hierarchical model encompasses the behaviours which have22been the focus of the perceptual-motor skills models of the driving task.The model is hierarchical in that the decisions made at a higher level affect thetask demands at the lower levels. For instance, an action at the tactical level, such as adecision to pass another car, requires a series of operational level actions. Similarly, adecision at the strategical level to avoid fast-moving heavy traffic obviates the need toengage in potentially high-risk tactical and operational level actions. Michon's modelprovides a characterization of the driving task that incorporates the active role drivers playin selecting aspects of their driving environment, and as we shall see, the model alsopoints the researcher toward the investigation of variables that were previously neglectedby research focused primarily on perceptual-motor skills.Models of Driver Behaviour and their Implications for Evaluation of Normal andDemented Elderly Drivers. There are non-trivial differences between characterizations of the older driverdepending on whether an investigator is inclined towards the perceptual-motor skill modelof driving, or the hierarchical task description. Within the traditional human-factors skillsmodels, the key variables that are believed to predict driving problems are deficits in theareas of perception, attention, and reaction time (Shinar, 1978). Expectations about theperformance of older drivers originating in this tradition are predicated on the wellestablished findings that older people experience declines in perceptual-motor functioning.Because of age-related deterioration in these skills, it would be predicted that elderlydrivers would experience marked declines in their driving abilities and would thereby beat greater risk than their middle-aged counterparts for experiencing and causing mishap on23the road. There is a large body of literature showing that there are indeed measurableage-related declines in the areas of perceptual-motor and attentional functioning (seeKausler, 1982, and Salthouse, 1985 for reviews). However, there is an increasingreluctance to assume a direct relationship between these operational-level deficits and theability to be a safe participant in the driving environment.One reason for this reluctance is the lack of data linking deficits in psychomotorskills with traffic accidents. Hopewell and van Zomeren (1990, p. 311) make thefollowing comment:One reasonable and yet erroneous idea is that numerouspsychomotor abilities, such as visual scanning, attention, andreaction time, interact in forming a hypothetical construct ofdriving skill, and that the more 'driving skill' one has thebetter a driver one will be...This popular notion, combinedwith the idea that knowledge of traffic laws is important, hasbeen the basis for all state licensing examinations...Asubstantial body of research in this area has demonstratedthat knowledge of driving regulations as well as basicpsychomotor abilities, such as coordination and reaction time,are among the least predictive factors of accident risk.There is one available study by Barrett et al. (1977) that links complex reactiontime and perceptual skill to accident rates in older commercial drivers. For a group ofolder drivers (aged 43 to 64) these investigators found a multiple correlation of .65between accidents and two predictor tests, the rod and frame test and a reaction time taskin which subjects were to respond to photographs of road scenes with embedded trafficsignals. Otherwise, as McKnight points out in his 1988 review, there are no compellingdata linking accident rates with measures of reaction time, motor skills, or motordisabilities in aged populations. McKenna, Duncan, and Brown (1986) discuss the24findings of Barrett and colleagues which also included an examination of the performanceof younger subjects for whom there was no correlation between accident rate and testperformance. McKenna et al. question whether the correlation between the tests ofinformation processing and accident rates found in the older group is indicative of acausal relationship. Given the combination of age-related declines in test performance andthe significant correlations between age and accident rates within both the younger andolder subject groups evaluated by Barrett et al. (1977) and Mihal and Barrett (1976),McKenna et al. suggest that the multiple correlation reported by Barrett et al. may bemediated by an age effect within the group of older subjects.Another source of resistance to equating psychomotor skills with drivingcompetence comes from those investigators who favour the hierarchical approach to thedriving task. When the driving task is viewed in the larger context afforded by Michon'shierarchical task description, the linkage between perceptual-motor skills and drivingcompetence is less direct than in the skills-model task descriptions. It has beenhypothesized by several investigators that the elderly and also the head-injured driverssuffer from impairments at the operational level of driving but may nonetheless continueto be safe drivers if they are able to compensate at the higher strategical and manoeuvringlevels (Brouwer, Rothengatter, & van Wolffelar, 1988; Kaszniak, Keyl, & Albert, 1991;Rothengatter & de Bruin, 1987; van Zomeren, Brouwer, & Minderhoud, 1987; vanZomeren, et al., 1988; van Wolffelar, Rothengatter, & Brouwer, 1987) .Rothengatter (1988) states that evidence suggestive of a marked general decline inthe mobility skills of the elderly is less abundant than might be expected based on the25results of laboratory research by experimental psychologists. He points out that, like theresearch on head-injury and driving, studies of elderly drivers using both laboratorypsychomotor tests and on-road driving measures sometimes fail to show a correlationbetween the two types of measurement methods (see Korteling, 1990, for an example ofthis in elderly drivers). His interpretation of these results is that the actual drivingbehaviour of normal elderly drivers is not directly related to the declines in theirpsychomotor and attentional skills. Rothengatter (1988, p. 3) states:Obviously, most of the elderly are able to maintain asufficient  level of driver performance by adaption of theirbehaviour on the strategic level through selective exposureand on a tactical level by selecting behaviours that are, as itwere, more error resistant. In order to do so, an awareness ofthe problems that may be encountered on an operational levelis a prerequisite.Similarly, van Zomeren et al. (1988) hypothesize that drivers impaired by age or head-injury may remain safe drivers if they are able to adopt strategies aimed at compensatingfor slowing in perception and response time. They suggest that psychometric predictionof driving skills in brain damaged drivers has relied too heavily on the measurement ofmental and motor slowness and visual search proficiency to the exclusion of morecomplex tests that might assess "the global cognitive skills necessary for psychologiccompensation of operational level impairments" (p. 95). Van Zomeren et al. (1988, pp.95-96) conclude with the following statements:There is ample evidence that people who have sustained headinjuries exhibit a slowing down in cognitive processes...Further, a comparable slowing down of informationprocessing has been demonstrated in elderly people.Nevertheless, aged drivers are not necessarily dangerousdrivers, and some of the patients in the present study were26classified by the external expert as demonstrating satisfactorydriving skill. It seems, then, that both aged and brain-damaged subjects can compensate, in principle, for theirshortcomings on the operational level. This compensation canbe explained from the hierarchic nature of the driving model.By making the right decisions on the strategic and tacticallevels, subjects can to a large extent avoid time pressure...This line of reasoning leads to the final hypothesis that theinstrumental shortcomings of head-injured drivers do notresult in dangerous driving so long as the subject is able tocompensate for them by adapting his behaviour on higherlevels of task performance. In particular, the subject must beaware of his own deficits and their consequences for trafficparticipation. Hence the conjecture that insight and self-criticism may be more important for a patient's fitness todrive than the degree of his cognitive deficits.No experimental evidence is yet available to substantiate their hypothesisconcerning the role of insight and self-criticism as a mediator of risk for cognitivelyimpaired drivers. Van Zomeren et al. (1987) seem to derive their assumptions about therole of insight, or awareness of deficits, from clinical reports that individuals with righthemisphere lesions or frontal lobe disorder often appear to lack awareness of their deficitsor seem unable to engage in self-critical thinking. It appears that patients with these typesof brain-damage are more likely to engage in dangerous impulsive driving behaviours.Also the authors give examples of compensation behaviour in head-injured drivers withslow psychomotor responses but good strategical- and manoeuvring- level planning skills.These drivers were considered to be "safe" and were observed to be good at "anticipatorydriving" suggesting that they planned ahead so as to circumvent their reaction timedifficulties.Van Zomeren et al. (1987) have suggested that the critical factors underlying27dangerous driving in head-injured people may not be the simple presence of deficits inoperational level skills (such as perception and reaction time), but rather a lack ofjudgement and a failure of insight (or "self-criticism") that then precludes the possibilityof adequate strategical level behaviour. If there is any substance to these suggestions,operational level deficits which may be present in both the normal elderly and the mildlydemented could be less important than deficits in strategical level skills such as judgementand planning. From this perspective the hierarchical model may be particularly helpful inguiding decisions about what types of driving behaviour measures would be useful whenattempting to characterize the particular deficits of dementing drivers.Recently McGlynn and colleagues have published an extensive review of theliterature on deficits in awareness in neuropsychological syndromes (McGlynn & Schacter,1989) and have carried out experimental evaluations of this phenomena in patients withHuntington's disease (McGlynn & Kaszniak, 1991a) and with Alzheimer's disease(McGlynn & Kaszniak, 1991b).The recent experimental work by McGlynn and Kaszniak (1991b) is particularlyrelevant when considering the measurement of strategical level behaviours in the presentstudy. These investigators developed two procedures to objectively evaluate whether ADpatients were aware of the extent of their memory problems. One procedure focused onevaluating patients' awareness of their memory problems in everyday situations. Usingrelatives ratings as the standard, patients were found to significantly underestimate theextent of their memory problems in everyday life. To evaluate whether the inaccurateself-ratings of the AD patients were due to a general deficit in the ability to make such28judgements, the AD patients were also asked to rate their relatives' memory functioning.There was a good match between the patients' ratings of their relatives' memory abilitiesand the self-reports of these relatives. This suggests that although the AD patients over-rate their own memory ability, they are able to accurately assess others' memoryfunctioning. The second procedure involved evaluating the accuracy of patients'expectations when asked to predict their scores on a series of standard memory tests.Accuracy ratios were computed by dividing predicted performance by actual performance.On all 12 memory tests administered, the AD patients predicted that they would obtainbetter test scores than they actually did. Accuracy ratios were also calculated betweenpredictions made by the relatives and the test performances of the AD patients.Comparing the two sets of ratios it was clear that the AD patients viewed their ownabilities more favourably than did their relatives. Finally, it was shown that although thedemented were not very accurate in predicting their own test scores they were able to do amuch better job when predicting scores for their relatives. Thus, both proceduresindicated that the AD patients were able to make accurate estimates of others' abilities butnot of their own.The possibility that the mildly demented subjects may have difficulty evaluatingtheir own driving skill will be explored in the present study. Of particular interest will beto determine if the over-confidence of demented individuals noted by McGlynn andKaszniak (1991b) with respect to cognitive abilities, also occurs when they are asked toevaluate their driving abilities.29Are the Normal Elderly Competent Drivers?: Comment on the Choice of a Comparison Group for Dementing Drivers.In this project, the identification of deficits in the driving abilities of dementing subjectswas based on performance discrepancies relative to the attainment of the normal elderlycontrols. One might question whether normal aged controls are an appropriate comparisonsample and whether there are valid reasons to assume that the normal elderly arecompetent drivers. Unfortunately, there is no "gold standard" criterion for drivingcompetence. If there were a well-established definition of driving competency and anaccompanying assessment measure, a project such as this would be quite straight-forward.Given a validated task analysis and an accompanying competency test, one would simplyuse this instrument to determine whether elderly cognitively impaired drivers were able toperform to criterion. However, in the absence of these established tools, one mustapproach the issue in a far more relativistic frame, comparing one group to another,comparing one test to another, trying to bootstrap one's way to a well-reasoned estimateof the driving risks cognitively impaired drivers might pose to themselves and others.Despite the current absence of evidence demonstrating that the majority of normalelderly (or any other group for that matter) are "competent" drivers, there are nonethelessnon-trivial reasons for selecting this group as the comparison standard for dementingdrivers.At this juncture, it is useful to make explicit a supposition that runs throughout thisproject and indeed the majority of the literature discussing the aged driver. I assume thatthe average medically fit elderly driver should be permitted to drive notwithstanding the30possibility of some measurable declines in driving skills. Recent discussions of "theelderly driver problem" (e.g., the report of the Committee for the Study on ImprovingMobility and Safety for Older Persons,Transportation Research Board, Volume 1, 1988)make it clear that neither researchers nor policy makers have as part of their agenda anydesire to see the automobile driving of seniors curtailed. Indeed, recommendations frommost experts advocate measures that will support continued driving by elderly individualsfor as long as possible.Economics is one factor underlying the desire to minimize interference andmaximize private vehicle access for the aged. Rosenbloom (1988) formulated a model topredict costs at the turn of century if the transportation costs of older citizens who hadlost the ability to drive were borne by the public sector in the U.S. Her model predictedthat if 1 in 10 of the trips needed by those with severe physical limitations were publiclyfunded, the cost would amount to several billion dollars per year. Clearly, it would bevery costly for society to discourage or impede access of healthy older persons to theindependent mobility allowed by private automobiles. Indeed, if Rosenbloom's costprojections are credible it appears that creative planning is needed to encourage andmaintain personally funded automobile use by the majority of the increasingly largepopulation of older citizens.Another factor that favours older drivers is the fact that as a group they have feweraccidents in absolute terms than do any other age group (Barakat & Mulinazzi, 1987;Evans, 1988; Rosenbloom, 1988; Transportation Research Board, 1988; van Wolffelar al.,1987; Yanik, 1985). Yanik (1985) presented data on the age distribution of drivers31involved in accidents in 1982 for nearly 10 million drivers. Whereas 29% of those aged25 to 34, and 15% of those aged 35 to 44 had been in an accident, only 7% of those over65 were in traffic accidents. Yanik states (1985, p. 5):It should be clear at this point that the elderly form arelatively safe group of drivers. They are involved in feweraccidents than drivers of other age groups, and areconsistently underrepresented among most accident scenariosconsidered. At the same time however, there is the distinctpossibility that elderly drivers are involved in accidents wellout of proportion to the time they spend on the road whencompared to the driving exposure of other driversYanik's statement appears to contain contradictory information: while maintaining that theelderly as a group are safe drivers, he also points out that they experience a higher per-mile accident rate than do other drivers. However, these apparently inconsistentstatements can be reconciled by taring into account the driving exposure of older drivers.Evans (1988) documents the fact that older drivers voluntarily curtail the amount ofdriving they do. According to Evans (1988, p. 193):Greater than any increase in driver risk with increasing age isdeclining distance of driving. For example, male drivers aged70 and older travel, on average, 9,300 km/year, compared to31,000 km/year for 35-39-year-old drivers; the correspondingvalues for female drivers are 4,300 km/year and 12,600km/year respectively. The problem of aging may thus bemore one of reduced mobility than of reduced safety. Asmental and sensory abilities decline, the response is lessdriving, especially under conditions of elevated risk, ratherthan a net increase in risk from driving, and even less so anincrease of risk imposed on other road users.Carp (1988) points out that the milage reduction of older drivers is not simply dueto their no longer needing to drive to work. Travel totals are not equalized by subtractingwork trips for younger groups; even when employment related driving is factored out,32younger drivers still log more miles than do older ones.The hierarchical model provides for an interesting perspective on the decreasedmilage of at least some older drivers. In the hierarchical scheme, the decision to decreaserisk by voluntarily decreasing driving exposure is a strategical level behaviour that can beseen as part of an individual's repertoire of driving behaviors. Considering the lowaccident rates for elderly drivers as a group, it would appear that the decision not to drivevery often or very far is an adaptive strategy that helps decrease accident risk of seniorsbelow the level experienced by other age groups.In summary, the use of normal elderly drivers as the comparison standard fordementing drivers is appropriate for the following reasons. Because there are no validateddriving competency indexes, definition of driving competency (or incompetency) must bebased on assessing the similarities or differences in performance of a target group to agroup selected as the comparison standard. Society is reconciled to the ongoing trafficparticipation of medically fit older drivers, notwithstanding some declines in their level ofdriving skill. Since we tolerate normal aging-related declines in driving skill, it is notcrucial to determine whether the mildly demented have experienced declines in drivingskill relative to their skills in middle age, or whether they are less able than youngerdrivers. Rather, the important question is whether mildly demented drivers experienceimpairments in driving related skills in excess of deficits that might be expected on thebasis of advanced age alone.33Studies of Brain Damaged Drivers: Examples of Psychometric Prediction of Driving Ability In the past decade there have been a small number of empirical investigations ofthe driving problems of brain damaged drivers. Some are interesting examples ofpractical applications of the skill-model of driving in clinical populations. Like thepresent project, studies of brain damaged drivers are typically designed to contributeinformation that will be helpful to clinicians who must make recommendations aboutfitness to drive. The goal of many of these studies was to examine the possibility ofselecting or developing screening measures that would provide clinicians with an objectivemethod for estimating driving risk. The human-factors approach to driving underlies thetest selection strategy for most of these investigations, with emphasis placed on measuresof perceptual ability, attention, and psychomotor speed.Unfortunately, despite recent conference presentations no articles are yet availablethat present details of performance-based studies of dementing drivers. However,investigations of brain damaged drivers grapple with many of the same issues. Researchon brain damaged drivers has been instructive in formulating the approach taken in thisproject with respect to both conceptual issues and the selection of psychometric tests. Afew studies warrant special attention because specific tests from these investigations havebeen adopted for the present project or because the investigators raise useful theoreticalissues. Six studies will be discussed in some detail to provide an indication ofmeasurement approaches, the types of psychometric tests used, and the magnitude and34consistency of correlations between psychometric tests and various measures of drivingperformance.(1) Sivak, Olson, Kewman, Won, and Henson (1981) assessed 23 brain damagedsubjects and 18 non-brain damaged controls on a set of off-road driving tasks, an in-trafficdriving test, and a battery of psychometric measures of perceptual and cognitive abilities.The brain damaged group performed significantly less well than controls on 3 of 5 off-road driving tasks, on 5 of 9 in-traffic tasks, and on 11 of the 12 psychometric measures.Different patterns of correlations between a composite index of the driving performancemeasures and the psychometric tests were found for the brain damaged versus non-braindamaged groups. For the brain damaged subjects, the composite index of the driving testscorrelated significantly with the WAIS-R Picture Completion (i= .72) and PictureArrangement (_r= .46) subtests, and a measure of visual stereoscopic depth of field (_r=.52). For the non-brain damaged, significant correlations (ranging from .52 to .77) werefound between the composite driving index and Porteus Maze, Rod-and-Frame, AbstractReasoning, and Ayres Space tests. There were no significant correlations between drivingmeasures and the Digit Symbol, Arithmetic, Digit Span, and Vocabulary subtests of theWAIS-R, the Motor-Free Visual Perception Test, or the Southern California Figure-Ground Visual Perception Test for either group. Sivak et al. recommend replicationstudies and conclude that their results provide ground for optimism about the prospects ofpsychometric screening for potentially serious driving-related problems. However, theyemphasize that different sets of tests may be required for predicting the drivingperformance of brain damaged versus non-brain damaged drivers.35(2) Stokx and Gaillard (1986) investigated 13 patients who had recovered fromsevere concussion and found that a measure of choice reaction time correlated with timetaken to perform a slalom driving task (=.69), but not with measures of braking and gearshifting speed. There were no significant correlations between driving tests and measuresindexing perception of degraded stimuli or memory set size.(3) Engum, Cron, Hulse, Pendergrass, and Lambert (1988) administered a largebattery of visual attention and scanning tasks from a computerized cognitive rehabilitationprogram including Trail Making, Picture Completion, and the Digit Symbol subtests of theWAIS-R. The majority of the attentional, visual scanning, and psychometric tests werepredictive of pass-fail driving classifications given subjects by a psychologist and anindependent driving tester. The only item in the test battery that failed to distinguishbetween the pass-fail groups was a measure of brake-pedal reaction time in a laboratorysimulation.(4) In a study involving 20 brain damaged subjects, Van Wolffelaar, van Zomeren,Brouwer, and Rothengatter (1988) found that a choice reaction time measure, the TrailMaking Test, a driving simulator compensatory tracking task (steering a straight courseunder changing "cross-wind" forces), and a test of eye/hand coordination predicted abilityto maintain a straight course while driving on the open highway. A reaction time test anda problem-solving task modeled after the Tower of Hanoi predicted the speed of trafficmerging decisions. The Picture Completion subtest of the WAIS-R and an embeddedfigures task did not correlate with any of the driving measures. None of the psychometrictests predicted performance on the Test for Advanced Drivers (TAD) devised by the36Dutch Automobile Association to assess traffic insight and risky driving habits.(5) Hopewell and van Zomeren (1990) reviewed the literature on the prediction ofdriving in neuropsychologically impaired populations. They underscore the importance,and difficulty, of developing adequate conceptual and methodological underpinnings fordetermining impaired individuals' fitness to drive. They point out that historicallydecisions about the driving fitness of neurologically impaired individuals have been madeby physicians who "diagnose" driving impairment based an individuals' medical status.Hopewell and van Zomeren argue that a medical approach will not yield a usefulunderstanding of impaired driving behaviour nor lead to the development of systematicmethods for assessment of driving-related impairments. They promote an approach to theunderstanding of driving impairment that is based on a psychological task analysis. Theystate:The operation of a motor vehicle is, after all, a functionalbehaviour. As such, the inability to drive a car is not amedical disease to be diagnosed or treated but rather a set ofgreatly impaired functional behaviours that may be evaluated.Although medical conditions undoubtedly affect drivingabilities, and physicians should be a part of the overalldecision-making process, the evaluation and prediction ofcomplex behaviour must remain within the realm of thepsychological, rather than the medical, model. (p. 311)Further, they recommend that attempts at prediction of driving performance emphasizecomplex, multifunctional neuropsychological tests. It is their contention that previouspsychological investigations of driving impairment in brain damaged patients have beenmisguided in over emphasizing perceptual and psychomotor abilities. The authors proposethat tests like Trail Making and Digit Symbol, which require mental flexibility, problem37solving, and complex visuomotor responding, are likely to be better predictors of drivingcompetency than are the more discrete measures of visual scanning, perceptual ability, andreaction time. Evidence consistent with this prediction has been reported by Gouvier etal. (1989) and Hale et al. (1987).(6) Finally, a report by van Zomeren, Brouwer, Rothengatter, and Snoek (1988)has proved to be one of the more influential papers and is regularly cited regarding theauthors' critique of the psychomotor skills approach to assessing driving ability and theirclaim that measures of lower level information processing skills are poor predictors of on-road driving performance for brain damaged individuals. Points raised by van Zomeren etal. (1988) have provided the impetus for aspects of the design of this project. However,although the paper introduces some interesting hypotheses, there are prominent flaws intheir study design and data interpretation that are difficult to ignore.The paper reports on an investigation of group differences and within-groupcorrelations for two on-road measures and 12 psychological tests. The study subjectswere a small and heterogeneous group of 9 head injured drivers, and 9 normal controls.The battery of psychological tests consisted of the Trail Making Test, Stoop Color WordTest, Benton Visual Retention Test, a 15 word verbal memory test, the PictureCompletion, Picture Arrangement, and Digit Symbol subtests from the WAIS-R, two testsof motor speed, four-choice reaction time, and two tests of visual scanning. Drivingmeasures were the same tasks used by van Wolffelaar et al. (1988; described above)comprising a measure of lateral position control on the open highway (an index of theability to steer a straight course), and the Dutch Automobile Association Test for38Advanced Drivers.The head-injured were impaired relative to controls on 9 of the 12 psychometricmeasures. Also, they performed significantly less well on the lateral position control(LPC) task. They engaged in more swaying and deviation from the centre line of the roadthan did controls. However, the actual amount of deviation was not considered to be ofpractical significance since only one of the drivers experienced difficulty keeping the carfrom crossing the boundary of the travel lane. Lateral position control scores correlatedsignificantly with the Benton Visual Retention Test t=-.76), the Trail Making Test(r=.83), and a test of motor speed and dexterity (r=.96).On the more demanding in-traffic measure, the Test for Advanced Drivers (TAD),the subject groups did not differ in the number of driving errors made during an hour longtest drive. However, on a qualitative basis, the errors made by head-injured drivers wereconsidered more serious. Five of the nine head-injured subjects were classified asinsufficient drivers using standard rating procedures whereas none of the controls were.There were no significant correlations between psychological tests and error scores onTAD in the head-injured group.Van Zomeren and co-workers engaged in an extended discussion of reasons why"the test battery had no predictive value at all" (p. 95). They suggest that the predictivefailures of the psychometric tests are likely to be mediated by hidden variables thatenhance driving competence despite measurable deficits in basic psychologicalfunctioning. Two likely candidates were discussed. One possibility is that the perceptual,decisional, and motor skills used in the driving task may become automatized and so fail39to show the injury-related declines that are manifested on the unfamiliar psychologicaltests of these abilities. A second possibility is that even if the head-injured drivers do infact experience deficits in basic instrumental driving skills, they may engage incompensatory adjustments so that their overall level of driving performance is unimpaired.The authors hypothesize that the lack of correlations between the Test forAdvanced Drivers and the psychological battery may have occurred because the head-injured drivers, despite real neuropsychological deficits, were able to perform competentlyin the actual driving situation by strategically adapting their driving speed and manoeuvresto compensate for their instrumental shortcomings. Unfortunately the discussion does notadequately integrate the full range of their findings.They do not discuss the implications of the apparent predictive successes of thepsychometric tests with respect to the LPC task. It would appear that the investigatorswere not convinced that the LPC measure provided a clinically useful index of drivingproblems and hence were not inclined to place much emphasis on the rather largecorrelation coefficients between this test and the psychometric measures. However, theseresults do suggest that psychometric tests can be sensitive to deficits in vehicle handlingskills. Also, in suggesting that the non-significant correlations between the Test forAdvanced Drivers error scores and the psychological test battery may have been the resultof compensation (i.e., competent driving despite psychological deficits), they seem to havelost sight of the fact that 5 of the 9 head-injured drivers were rated as insufficient by thedriving examiner suggesting that the majority of the drivers may not have been drivingcompetently in the first place.40Finally, the authors failed to consider some of the more mundane methodologicalreasons why the psychological test battery might have failed to predict error scores on theAdvanced Driving Test. Specifically, it is likely the outcome of regression analyses wereseriously compromised by restricted test score ranges, low sensitivity of the TAD errorscore to brain injury, and a very small sample size.Despite the weaknesses of the van Zomeren et al. (1988) study, the issues theyraise in their discussion may be valuable. The interpretation of the study data appears tobe strongly influenced by their interest in the concept of compensation and by argumentsdeveloped in their 1987 review of literature on acquired brain damage and driving inwhich they introduce Michon's hierarchical model into the clinical literature (vanZomeren, Brouwer, & Minderhoud, 1987). This model is llidey to be helpful in futureresearch on elderly and cognitively impaired drivers.41METHODSubjectsSubjects in this study were 18 demented, 18 normal elderly, and 18 mid-agedrivers who participated in the Driving and Aging Study between February 1990 andOctober 1992. Subjects were recruited in a number of ways. The majority of thedementia cases were solicited through the Clinic for Alzheimer Disease and RelatedDisorders (hereafter referred to as the Alzheimer Clinic) with a small number of casescoming by referral from other geriatric assessment centres or private practitioners.Several different recruiting strategies were used in an attempt to obtain larger numbers ofdementing drivers. Announcements about the study were placed in newspapers andnewsletters for medical practitioners and families of dementing individuals. Radio andtelevision coverage of the study included requests for cognitively impaired drivers.Medical practitioners were informed about the study through mail-outs and several talksgiven at medical rounds and various meetings. Normal elderly controls were recruitedthrough the same media coverage methods that were used to recruit the demented and alsowith separate adds that requested elderly drivers with 12 or fewer years of eduction.Healthy spouses of demented drivers were encouraged to take part as participants in thenormal control group. The mid-age controls were recruited through radio and newspaperpublic-service announcements and through the hourly workers office of EmploymentCanada.Selection criteria for the dementia sample included the following: (a) evidence of42cognitive impairment and decline in everyday functioning sufficient to satisfy criteria fordementia used at the Alzheimer Clinic, (b) mild cognitive impairment defined as a MiniMental State Exam (MMSE) score between 27 and 20, (c) driving within one month ofrecruitment for the study, and (d) absence of exclusionary medical criteria such as majorpsychiatric disorder, Parkinson's Disease, severe arthritis or other disorders interferingwith physical mobility, marked loss of vision, mental retardation, delirium, or neurologicaldisorders resulting from stroke, tumour, or head injury.Individuals were considered suitable to be included in the normal elderly controlgroup if they demonstrated (a) an absence of significant cognitive impairment, (b) anMMSE score of 29 or 30, (c) an absence of exclusionary medical criteria, (d) and were 60years of age or older and driving within one month of recruitment to the study.Individuals suitable for the mid-age control group met the same criteria as wereapplied to the normal elderly sample except that the age range for the mid-age controlswas between 35 and 45 years.Diagnostic procedures: Overview.The process of establishing whether potential study subjects met selection criteriaproceeded by the following steps: (a) all subjects were assessed on the battery ofneuropsychological measures and symptom severity rating scales used for diagnosticassessment at the Alzheimer Clinic; (b) collateral information about changes in cognitiveand everyday functioning was obtained from a significant other and/or a referringphysician using the Present Functioning Questionnaire; (c) a medical questionnaire,providing information related to exclusionary medical conditions, was filled out by43participants or by participants with the help of a collateral if cognitive impairment wassuspected; (d) test scores on the neuropsychological battery were compared to ageappropriate norms and rated with respect to level of impairment; (e) the test score profileand the collateral information were used by an experienced neuropsychologist to assignimpairment ratings on the Functional Rating Scale (1-41(S) which then formed the basis fordiagnoses of dementia using the standard procedures established at the Alzheimer Clinic.Appendix A provides further details of diagnostic procedures, scores for the three subjectgroups on the neuropsychological tests and Clinic rating scales and copies of theneuropsychological test profile form, the Functional Rating Scale, the Present FunctioningQuestionnaire, and the medical screening questionnaire.Characteristics of the Demented Sample and the Control Groups Of 29 cognitively impaired elderly who participated in the Driving and AgingStudy, seven were very mildly affected and did not meet all the Functional Rating Scale(FRS) criteria for dementia, while four others had MMSE below 20 making them tooseverely impaired to be included in the group of mildly demented. Thus there were 18individuals (12 males and 6 females) who satisfied FRS criteria for dementia and hadMMSE scores between 27 and 20. These subjects ranged in education from 6 to 16 years= 11.6; SD = 2.6), and had an average age of 73.6 years (SD = 6.9). Table 1 presentsbasic demographic information for the three groups of subjects.Groups of 18 normal elderly and 18 mid-age subjects were selected in order tomatch the control groups with the dementia group as closely as possible for the proportionof males and females, for education level, and for age in the elderly groups. The average44Table 1Demographic Information for Dementia and Control GroupsMeasureGroupsDemented(n=18)NormalElderly(n=18)Mid-ageControls(n=18)ANOVAnAgeM 73.61 * 72.61 a 40.83 b < .001SD 6.99 6.08 3.29Range 58-83 67-87 35-46Years of EducationM 11.61 12.11 12.72 , .214SD 2.57 2.76 2.67Range 6-16 6-16 7-16SexMale 12 12 12Female 6 6 6Note. Superscripts appearing in the body of the table are associated with post-hoc pairwisecomparisons between group means. Groups with different superscripts are significantlydifferent at the .05 level.45age of the 12 males and 6 females in the normal elderly control group was 72.6 years CSD= 6.1) with education ranging from 6 to 16 years LM = 12.1; SD = 2.8). The mid-agecontrol group also comprised 12 males and 6 females, ranged in education from 7 to 16years^= 12.7, SD = 2.7), and had an average age of 40.8 years (range 35 to 46 years,SD = 3.3).Individuals included in the control groups were allowed a maximum of one erroron the MMSE, thus all controls had MMSE scores of 29 or 30. The mean MMSE scorefor the normal elderly was 29.66 02 = .48) and for the mid-age group the mean was29.61 (S12 = .50). The demented group had a mean MMSE score of 23.56 (a3 = 2.23).Following a significant ANOVA for MMSE score, F(2, 51) = 122.1, 2. <.001, pairwisecomparisons indicated that both control groups differed significantly from the dementedbut not from each other.All subjects in the study had valid drivers licences and drove at least once a week.Each of the three groups included two subjects who had at some point driven as part oftheir livelihood (e.g., cab driver, delivery van operator, travelling salesman). Other detailsof driving history are presented later in the Results section.Table 2 shows the number and type of health problems reported by subjects in thestudy. In both the demented and normal elderly groups 10 individuals suffered from achronic medical disorder while only two subjects in the mid-age group had healthproblems. Table 3 shows the number of prescription medications used by study subjects.The older groups were essentially equivalent with respect to medication use whilesignificantly fewer medications were used by the mid-age group.46Table 2Health Problems of Dementia and Control GroupsGroupsCondition^Demented(n=18)NormalElderly(n=18)Mid-ageControls(n=18)N % N % N %Hypertension^5 27.0 6 33.0 1 5.5Arthritis 1 5.5 2 11.1 -Diabetes^1 5.5 - -Cancer 2 11.1 2 11.1 1 5.5Heart Disease^1 5.5 2 11.1Other^6 33.0 3 16.6 1 5.5Total number of complaints16 15 3Number of individuals with no complaints10 10 16Note. This table summarizes participant's medical histories with respect to more seriousmedical conditions. The data were obtained from a questionnaire filled out by the normalelderly and mid-age participants and by demented participants with assistance from acollateral.47Table 3Prescription Medications used by Dementia and Control GroupsGroupsNormalMedication^Demented^Elderly(n=18) (n=18)Mid-ageControls(n=18)N^%^N % N %Hormones*^2^11.1^1 5.6 1 5.6Antihypertensive^1^5.6^- -Analgesic/^2^11.1^2 11.1 2 11.1Anti-inflammatoryCardiovascular^4^22.2^3 16.6Gastrointestinal - - 1 5.6Corticosteroid^2^11.1^1 5.6 1 5.6Other^2^11.1^2 11.1 1 5.6Total number of medications13^12 6Number of individuals with no prescription medications12^12 14Note. The data in this table were obtained from a questionnaire filled out by the normalelderly and mid-age participants and by demented participants with assistance from acollateral.' Medications in this category include antidiabetic agents, thyroid replacement, and oralcontraceptives48Because the two older groups were so similar with respect to health and medicationstatus, it is unlikely that these factors would confound differences in driving orpsychometric test performance between the older subject groups. There were, however,health and medication status differences between the younger and older groups. While intheory this could be problematic, the results of this study are unlikely to have beenaffected by these differences. Neither the scores on the neuropsychological measures(presented in Appendix A), nor the driving behaviour measures, show any particularadvantage for the mid-age controls over the normal elderly.MeasuresMeasures will be discussed in three sections: (a) descriptive measures, (b) drivingbehaviour measures, and (c) psychometric tests. Table 4 provides an overview of thestudy measures.Descriptive Measures To describe relevant driving-related subject characteristics, several aspects of theparticipants' driving habits and history were documented. The variables included in thisgroup are important to document because they have the potential to confound the groupdifferences that are the focus of this study. Of particular concern were factors such asyears of driving experience, frequency of car use, driving exposure in higher-risk drivingsituations, and avoidance of driving situations. Because duration of driving experienceand amount of exposure are accident risk factors in their own right, it was important to49Table 4Study MeasuresDescriptive Measures Driving History:Driving interview question related to years of driving.Driving Exposure:Driving interview questions related to annual milage,frequency of night and rush hour driving, number of tripsper week, and extent of driving avoidance.Motor Vehicle Branch Driving RecordMotor Vehicle Branch Screening of Visual AcuityDriving Behaviour Measures Operational Level Measures:Brake reaction time on driving simulator.Steering accuracy on driving simulator.Manoeuvring Level Measures:Motor Vehicle Branch Road Test.Stopping distance for hazard avoidance at off-road track.(table continued)50Table 4 (continued)Study MeasuresStrategical Level Measures:1) Assessing the accuracy of self-appraisal:Cone Avoidance task and comparison of subjects'self-ratings of driving problems to the ratings ofcollaterals.2) Assessing compensatory avoidance:Examination of the relation between measures of drivingexposure and measures of driving behaviour and problems.Psychometric Tests Letter cancellation timeStroop color-word interference timeChoice reaction timeTrail Making Test - Part BPicture Completion subtest of the WAIS-RComprehension subtest of the WAIS-RDirect Assessment of Functional Status51consider the extent to which the groups in the present study might differ on thesebackground variables.Driving Interview. The bulk of information about subjects' driving histories, habits,and everyday driving problems was derived from a structured interview administered tosubjects and, when possible, to a collateral familiar with the subject's driving habits andproblems. Two versions of the Driving Interview were constructed, one phrased to obtaininformation from subjects about their own driving, the other for eliciting informationabout the subjects from their collaterals. Appendix B. contains copies of the collateralversion of Driving Interview questions that were administered for this project.The interview questions used in this study were a subset of questions from the OlderDriver Interview Schedule designed by the Gerontology Research Centre at Simon FraserUniversity for the Older Driver Study carried out by researchers at the InsuranceCorporation of British Columbia. Rothe (1990) provides a detailed discussion of thedevelopment of the Older Driver Interview Schedule and the results from a survey of 900older drivers using this instrument.Subjects in the present study were interviewed at the end of the psychometric testingsession by a graduate student research assistant. Collaterals were interviewed separatelyby the study coordinator, in person if they accompanied the subject, or by telephone ifthey were not available in person. Except for items related to the number of times eventsoccurred, the majority of questions were presented with a forced choice format. Responseoptions varied depending on the question. For example, some questions were in simpleyes, no format, while for others the respondent might be asked to indicate whether52the subject frequently, sometimes, seldom, or never had difficulty with a specified actionor situation. A don't know / no response category was among the scoring options for allitems.In this discussion of the Driving Interview a distinction will be made betweenquestions and items. Several of the topics covered in the Interview were presented asmulti-item questions. For these questions the interviewer provided the respondent with abrief outline of the topic of the question and explained the set of response options. Thenthe list of situations (items) were presented, each of which was coded in terms of thesame set of response categories. Written versions of the interviews were available inlarge type for respondents who wished to read along or found it helpful to be able to seethe response options. Typically the interview took 30 to 40 minutes to complete.The questions eliciting information about the number of years of driving andnumber of miles driven per year were presented with forced choice formats requiring therespondent to choose the appropriate year or milage range from six possible options.The questions about driving at night and in rush hour simply required the respondentto state the number of times per week the subject drove in each situation. Those whodrove less than once a week in rush hour or at night were assigned a score of zero for thatsituation.An estimate of the of trips per week was obtained by asking respondents to specifyhow many times per week the subject used their car to carry out ten activities such asshopping, visiting family or friends, going to the doctor etc. A score of zero was assignedto any activity for which there was less than one trip per week. The overall trips per53week score was the sum of the trips for the ten activities.The question related to driving avoidance comprised eight items, each consisting of adriving situation such as driving on freeways, in heavy traffic, in rain, in fog etc. Foreach item the respondent was presented with three response options and was asked to pickthe one that best described the subjects' avoidance behaviour. The response options were:(a) don't drive in situation, (b) try to avoid situation, and (c) don't avoid situation. Ratherthan treat each item separately, a summary score was computed to provide an index of theextent of avoidance across the eight items. The method used to obtain the index ofdriving avoidance is detailed in Appendix C along with a discussion of the scoring of theother multi-item questions from the Driving Interview.Motor Vehicle Branch Driving Record. Subjects signed a consent form allowingtheir driving record to be searched. MVB driving records contain a five year summary oftraffic convictions and police reported accidents. In addition to providing this listing ofthe traffic tickets and accidents for each subject, the MVB made the information containedin the police accident reports available. In principle, police accident reports shouldprovide information about probable fault, causal factors, and an estimate of the damagecosts for each recorded accident. Unfortunately, although accidents listed on the MVBrecord are technically police reported, in fact the majority are not police attended.Accident reports are usually filed by drivers themselves in the days following theaccident'. Often these driver-filed reports contain little more than the date and location'Accidents documented in this way are nonetheless termed "police reported accidents"because the driver files the report with the police who in turn submit this information to theMVB. Thus, from the perspective of the MVB these are police reported accidents.54of the accident. Accidents that do not involve injury or costly repairs are not usuallyreported to the police and hence are not included on the MVB driving record. Thedifficulties with accident rate data are widely acknowledged by researchers (see Ball &Owsley, 1991, for a recent discussion). There is no simple response to the problem. It isdifficult to categorically ignore accident rate data, but clearly this information must beinterpreted with caution. Notwithstanding the problems with the MVB accident records,most researchers seriously question the validity of self-report measures of accidentfrequency. Owsley et al. (1991) report on an unpublished study conducted by theirresearch group in which they found that older drivers with the largest number of state-recorded accidents were also those individuals who tended to under-report accidents on aself-report questionnaire.Vim/al acuity. Assessment of vision was based on the pass / fail visual screeningprocedures used by the MVB as part of the standard licensing procedures. The purpose ofthe visual screening was to ensure that there were no gross discrepancies between groupsin visual acuity and to exclude any drivers who were seriously visually impaired. Allsubjects were assessed by an experienced examiner on a Keystone vision tester (Schieber,1988). The MVB requires that licensed drivers have at least 20/40 vision as assessed onthe Keystone vision tester.Driving Behaviour Measures The organizing principle for selection and/or development of driving behaviourmeasures was the hierarchical model of driving proposed by Michon (1985). This modelwas not considered to be a formal theory of driving behaviour. Rather, it was chosen for55its heuristic value. The hierarchical model provides a multidimensional conceptualizationof driving behaviour that incorporates basic psychomotor skills, driver's on-roadinteractional behaviour, and also highlights cognitive skills associated with trip planningand risk avoidance.Operational level measures. To obtain indices of operational level skills, simulatorbrake reaction time and simulator steering accuracy were assessed on the ComputerizedDriving Assessment Module (CDAM) located at GF Strong Rehabilitation Centre (seeAppendix D for a copy of CDAM output)Testing procedures were highly standardized and followed the semi-automatedassessment protocol routinely used for driver evaluation at the Centre. Prior to datacollection subjects were provided with a thorough introduction to the controls onsimulator, five practice trials on the braking task, and ten practice stimuli on the steeringtask. None of the control subjects required further practice but some of the dementedsubjects were given 10 to 15 minutes of additional time to become familiar with theequipment. CDAM was operated by one of two senior students from the School ofRehabilitation Medicine at the University of British Columbia.CDAM consists of an automobile seat, dashboard with speedometer, brake and gaspedals, steering wheel, computer monitor for display of instructions, and a double arc oflight-emitting diodes (LEDs) set at eye level subtending 190 degrees of visual field whichgenerate stimuli for the steering tasks. CDAM has an auditory channel that simulates acar engine accelerating and decelerating, thereby providing auditory feedback about"speed." Tasks were presented in a fixed order, and were generated and scored through a56software program developed by the Departments of Electrical and Clinical Engineering atUBC.The brake reaction time measure comprised the average of three braking trails. Thesubject was instructed to maintain a speed of 50 kph while monitoring the screen for theappearance of a large stop sign. Brake reaction time was recorded in milliseconds andcorresponded to the interval between the presentation of the word 'STOP' and the momentwhen the subject fully depressed the brake pedal.For the steering task, subjects were instructed to use the steering wheel to track aseries of lights that appeared in different positions in the upper display band of the LEDarc. The lights changed position at intervals ranging from 1 to 10 seconds. The steeringwheel controlled the LEDs in the lower band of the arc. The object of the steering taskwas for subjects to match, as quickly and accurately as possible, the steering wheelcontrolled LEDs to the position of the computer generated LEDs for 14 changes in lightposition. Steering accuracy was computed by summing the areas of deviation between thecurve describing the position of the computer generated lights and the curve generated bythe steering actions of the driver. Steering deviations occurred when subjects reactedslowly and / or from over- and under-steering. At the same time that subjects werecarrying out the steering task they were instructed to "maintain 50 kph." A speedometerwas available and subjects were also able to gauge changes in simulator speed from theauditory feedback that accompanied acceleration and deceleration on CDAM. Thus, thesteering deviation task functioned as a measure of performance in a dual task situation.57Manoeuvring level measures.  Manoeuvring level skills were assessed on the MVBRoad Test and on a measure of stopping distance in response to a moving hazard.The MVB Road Test is a performance-based examination that systematicallysamples drivers' knowledge of the rules and conventions of the road and their abilities tosafely interact with traffic in everyday driving situations. All subjects were examined ona standard MVB-format road test in a dual control car. The Road Test was laid out bythe retired Senior Regional Coordinator and Chief Examiner for the British ColumbiaMVB. Testing was done by one of three trained licensing examiners and scored using thestandard MVB Road Test Form (see Appendix D for a copy of the scoring sheet). Testscores on the Road Test were the sum of demerit points earned during the test drive.Following procedures used for Class 5 licensing tests, drivers were considered to havefailed the road test if they obtained 40 or more demerit points or engaged in a dangerousdriving action. Typically the examiner discontinued road testing once a subject met thefailure criteria. Subjects who failed the road test due to engaging in a dangerous actionwere assigned 40 demerit points.Participants were informed that all test results were confidential and that their driverslicence was not threatened by taking part in the study. They were provided with copies oftheir MVB road test and visual screening forms and with written feedback from theexaminer about any driving actions that had resulted in demerit points. Subjects whoreceived 40 demerit points were informed verbally and in writing that if the test had beenan official licensing exam they would have failed, and they were given a verbal andwritten recommendation to see their family doctor about whether they should continue to58drive. Also, subjects who did poorly on the MVB test were given the opportunity to signa release form permitting study staff to forward the test results to a family member or thesubject's family doctor.On the MVB road test vehicle control skills were tested in the followingcategories: stopping and starting the car, steering, signalling, backing up, starting the car,right turns, and left turns. Knowledge of traffic regulations was tested by observing driverbehaviour in the following situations: at stop signs, traffic lights, intersections, taking andyielding right of way, regulating speed appropriate to speed limits and traffic flow,selection of position on the road or traffic lane, actions at intersections, and travel throughrestricted speed zones around schools and playgrounds.The second test of manoeuvring-level skill was a measure of stopping distance forhazard avoidance. This measure of accident avoidance skill helped to round out theassessment of manoeuvring level skills provided by the MVB Road Test. Whereas theMVB test is primarily an index of overlearned driving actions carried out under low timepressure, the hazard avoidance task provided a measure of subjects' capacities to respondquickly and appropriately to an unexpected event.The Hazard Avoidance task was done in the same 1985 two door Pontiac 6000outfitted with dual controls as was used for the MVB road test. An off-road drivingcircuit was specially constructed for this project at the Pacific Traffic Education Centre(PTEC) located on an abandoned airport runway. The driving circuit consisted of a tenfoot wide lane painted on the pavement in a shape approximating a capital letter B. Thecurved portions of the circuit were constant radius turns with an 80 foot radius for each of59the two loops. The distance for a single trip around the circuit was 404 meters (measuredfrom centre line), therefore slightly less than 1/2 km. Appendix D contains a scalediagram of the circuit indicating the location of tasks.The hazard avoidance manoeuvre was one of a series of braking and steering tasksthat occurred during the course of a ten minute testing period on the circuit. The otheremergency braking test conducted at the driving track was not suitable for inclusion inthis study due to frequent equipment failure in the automatic triggering device. Thehazard avoidance task was selected for this project because it was mechanically reliableand presented a very compelling emergency braking situation in that failure to stop thetest vehicle in a fairly efficient manner would result in actual collision with a life sizestyrofoam mock vehicle.Before starting the circuit testing, drivers were presented with both written andverbal instructions informing them that at some point an obstacle would move into theroad in front of the car and they would be required to stop as quickly as possible. Indevising the standard instructions for the hazard avoidance measure, consideration wasgiven to issues concerning subjects' affective reactions to the task particularly should theycollide with the styrofoam vehicle. Although impaired driving performance was in nosense treated lightly, the approach taken to introducing subjects to the task was intendedto put them at ease and indicate that the testing process was somewhat playful.The circuit testing was done at 20 kilometres per hour (kph) +/- 4 km. Subjectshad two practice runs around the track to become familiar with the lay out and withtravelling at the designated speed. In a few cases, extra practice runs were given for60subjects who had difficulty maintaining speed. Vehicle speed was monitored by an on-board computer and the subjects were reminded verbally and by road signs to alwaysmaintain 20 kph. Testing was conducted with the experimenter in the back seat of the carand one of two experienced RCMP driving instructors covering the over-ride controls. Ifnecessary the driving instructor corrected the vehicle speed using the dual controls.The "hazard" in the hazard avoidance task was a life-size styrofoam facade of acar attached to a light-weight aluminum frame that moved on hidden bicycle wheels. Themovement of the mock car was automatically triggered when the test car passed over aswitch activating a mechanism that pulled the styrofoam facade across the road in front ofthe approaching vehicle. The distance between the trigger switch and the styrofoam carwas 22.5 meters. This distance can be conceptualized as the distance-to-collision. In theabsence of over-ride braking by the driving instructor, subjects who were unable to stopthe test vehicle within the 22.5 meter distance would collide with the styrofoam car. Thetest vehicle was travelling at approximately 5.6 meters per second allowing just over 4seconds for the driver to come to a stop. The dependent measure for this task was thevehicle stopping distance which was measured from the trigger switch to the front wheelof the test car.In order to create a situation simulating some of the demands of on-road driving,the hazard avoidance braking manoeuvre occurred on a "busy" 40 meter portion of thecircuit that also included a stop light and a directional arrow. The sequence of events justprior to the braking manoeuvre started when subjects came around a blind curve at the farend of the circuit onto the 110 meter straight section. This marked the beginning of61subjects' 5th trip around the circuit. Approximately half way along the straight-away atraffic signal was automatically triggered. On the 5th trial the light was green allowing thesubject to continue. One and a half seconds later, a directional arrow located 20 metersbeyond the traffic light was triggered. Subjects had to monitor the arrow on each triparound the track in order to determine whether they were to go straight ahead or off to theright. On this pass the arrow directed them straight ahead. Approximately 1 second afterthe directional arrow lit-up, the mock vehicle began to move across the travel-lane at arate of about one meter per second. Thus, within a 2.5 second period subjects wererequired to monitor a traffic signal, a directional arrow, and detect the beginning of themovement of the styrofoam vehicle.The resting position of the styrofoam car was about 5 meters beyond, and acrossthe travel-lane from the large box housing the lights for the directional arrow. The boxwas intended to act as a barrier so that subjects could not swerve to the right to avoid themoving object coming across the road from their left. Swerving to left was not attemptedby any subjects and would have resulted in a collision with the rear of the mock vehicleor in landing in a ditch beyond the edge of the pavement. Thus, a braking action was theonly viable response to avoid collision with the styrofoam car.There were four occasions when subjects failed to react quickly enough to themoving styrofoam car and it was necessary for the driving tester to use the dual brake inan attempt to avoid collision. The over-ride braking was successful on two occasionswhile in the other two cases there were low impact collisions with the styrofoam vehicle.When over-ride braking or collision occurred, the full distance of 22.5 meters was62assigned as the subject's stopping distance.Strategical level measures. Of special interest in this study was the possibility thatthere might be group differences in the quality of strategical level behaviour. However,representing the strategical level presented interesting challenges. Of the three levelscomprising Michon's hierarchial model, the strategical level is the most abstract anddifficult to operationalize. In a general sense, strategical level behaviour comprises thejudgements and problem solving cognitions that drivers engage in as they plan andexecute their driving actions. Clearly there are myriad behaviours and cognitions thatcould be investigated in the context of strategical behaviour.For this project emphasis was placed on a subset of strategical level behavioursrelated to self-appraisal and compensatory risk avoidance. Two approaches to evaluatingthe accuracy of subjects self-appraisals' were examined: the Cone Avoidance Task and acomparison between self-ratings and collateral ratings of driving problems. Compensatoryrisk avoidance was explored by examining the relationships between the Driving Interviewdata related to driving exposure and the various measures of driving problems.The Cone Avoidance Task was comprised of three trials during which the subjectwas required to repeatedly manoeuvre the test car through a course of traffic cones hittingas few as possible. The cones were configured in the shape of a lower case letter h withright angle corners. Lane width was three meters, the long side of the h was 15 meters,with cones at 1 meter intervals. The task began with the driver facing the bottom-rightleg of the h configuration. The driver was asked to enter the open lane on the short leg ofthe cone-configuration, make a sharp left at the first corner, another sharp left at the63second corner, back the car up the length of the 15 meter long side, and come out throughthe same route they had entered. One examiner stayed in the car with the driver throughout the task while another examiner walked through the course before the first trial toinsure that the driver knew precisely what the route was. Drivers were informed that theycould do the task at what ever speed they wished and that they could reverse andreposition the car as needed in order to make the tight turns while avoiding the cones.The vehicle handling portion of the task was similar to what one might encounter whenpulling in and out of a small space in a tightly packed parkade.The task was done three times. On each trial the participant was asked to estimatethe number of cones they expected to hit. After each trial, just prior to making predictionsfor the next trial, the driver was told the number and shown the position of cones that hadbeen hit. The dependent measures were the absolute value and the signed value of thedifference between the sum of predicted and the sum of actual hits on trials two and threeof the task. The first trial provided experience at both the prediction and manoeuvringcomponents of the task. The difference between predicted and actual hits for the secondand third trials provided an index of the subject's ability to estimate the difficulty of thetask while taking account of their performance on the previous trial(s).The average of the absolute values of the difference scores provided an index ofthe overall accuracy of each groups predictions where larger means would be indicative ofless accurate prediction. The signed values of the difference scores were also examined toevaluate the direction of mismatches between predicted and actual hits. A negative score(i.e., actual hits greater than predicted hits) suggests that the driver is overconfident and64does not adequately assess the difficulty of the task in relation to their own vehiclemanoeuvring skills.The other approach to examining the accuracy of subjects' self-appraisals involveda comparison of the self-ratings of subjects to the ratings of their collaterals on theDriving Interview questions related to driving problems. The Driving Interview includedfive multi-item questions (comprising 56 items in total) eliciting information about drivingproblems the subject might be experiencing. The questions were concerned with drivingfaults, changes in driving ability, concerns about deficits in driving skill and drivingmishaps, difficulties with driving manoeuvres, and the extent to which various factorsinterfered with driving. Not surprisingly perhaps, given the relatively homogeneouscontent of the questions, the relationships of self-ratings to other-ratings were quite similaracross the five different questions. Consequently, rather than treating each questionseparately, it was decided that it would be advantageous to derive a single summary scorefor subject-rated driving problems and similarly a single summary score for collateralrated problems.Derivation of a summary score to represent the combined ratings on the fivedriving problem questions involved a number of considerations related to selecting themost appropriate statistical model for data reduction. After considering several options,principal component analysis was selected as the most appropriate technique. Thestandardized weights derived from the first principal component were applied to the2 The questions under discussion are numbers 27 (faults), 31 (changes), 33 (concerns),35 (difficulties), and 37 (interferences) on the collateral version of the Driving Interview. Thesame questions were asked of participants. See Appendix B for copies of the questions.65driving problem question scores to allow for computation of a summary score that was anoptimal linear combination of the five questions. Weights derived from the first principalcomponent have the virtue of generating a linear combination of variables that has thehighest internal consistency and explains the most variance in the data set from which theweights were generated (Tabachnick & Fidel!, 1983). Appendix C presents the scoringmethods for the individual driving problem questions and the specifics of the principalcomponent analysis.In addition to investigating the accuracy of self-appraisal, compensatory riskavoidance was also explored. This construct was addressed by examining the relationshipbetween the measures of driving exposure (as outlined in the section on descriptivemeasures) and the various indices of driving performance and driving problems.Psychometric Measures The second component of this project was the exploration of the relations between agroup of psychometric tests and the measures of driving performance. The purpose of thiscomponent was to determine whether commonly used measures of cognitive or functionalability might be highly correlated with driving ability. Psychometric predictors of drivingperformance would be helpful to clinicians by providing an office-based estimate of thedriving risk of demented patients. The following paragraphs provide a brief review of theliterature associated with the seven tests selected for this study.1. Letter cancellation. Cancellation tests are thought to require visual scanning,attentional focusing, and the ability to sustain attention (Lezak, 1982). There are a widevariety of cancellation tasks available in the literature. The version used for this study66was a form developed by Diller et al. (1974) found to be sensitive to deficits in olderstroke patients. Huff et al. (1987) used the Diller cancellation task and found that itdistinguished between normal elderly and dementia patients. A similar cancellation taskdistinguished mildly demented from normal elderly controls and was sensitive to theseverity of overall impairment in a group of mildly to severely impaired dementia patients(Botwinick, Storandt, & Berg, 1986; Botwinick, Storandt, Berg, & Boland, 1988).However, Allender and Kaszniak (1987), and Vitaliano, Russo, Breen, Vitiello, and Prinz,(1986) suggest that performance on cancellation tasks is relatively preserved during theearly stage of a dementing disorder and then becomes a very sensitive measure ofdeterioration in the moderate to severe stages. Notwithstanding the inconsistencies in theliterature, the letter cancellation test is believed to be useful as a measure of aspects ofattention . such as vigilance, concentration, and selectivity.The letter cancellation test form contained six lines of upper case letters with 52characters per line. For this project, the test form was enlarged so that older individualscould see the letters clearly. The subject was instructed to cancel all the C's and E's inthe six rows as quickly as possible. There were 18 target characters randomlyinterspersed in each row. The dependent variable was the error score which includedomissions as well as inappropriate cancellation of non-targets.2. Stoop Color-Word Interference task. The S troop task is usually characterizedas a measure of focused attention, assessing an individual's ability to ignore onedimension of a stimulus while selectively attending to another. Some describe it as ameasure of freedom from distractibility (Lezak, 1983). Grady et al. (1989) report that AD67patients have been shown to have difficulty on this task as well as on others that involvecompeting processes. Koss, Ober, Delis, and Friedland (1984) found sizable Stroopinterference effects for mildly demented patients.A shortened version of the standard Stroop test was used in this project. Incontrast to the original Stoop task which comprises 3 cards with 50 stimuli per card(Lezak, 1982), the test materials for the modified Stroop were two cards: one with 24coloured dots for the color naming trial, the other with 24 color-word stimuli for theinterference trial. The dependent variable of interest was the time taken to complete theStroop color-word (or interference) portion of the task. In this condition, subjects arepresented with a page of color-names that are printed in non-matching coloured inks. Forexample, the word red might be printed in green ink. The subject is instructed to namethe color of the ink while ignoring the word itself and to do so as quickly and accuratelyas possible.3. Choice Reaction Time. Several investigators have found that measures ofchoice-reaction time are correlated with driving performance. Also both simple andchoice reaction time (RT) measures have been shown to differentiate normal elderly fromdementing individuals although choice RT appears more sensitive. Pirozzolo, Christensen,Ogle, Hansch, and Thompson (1981) investigated the relative sensitivities of unwarnedsimple RT and 4-choice RT. The choice RT task was both sensitive and specific: 92% ofdemented cases were correctly classified with no errors for the normals. A study byFlicker, Bartus, Crook, and Ferris (1984) explored the correlations between choice RT anda variety of measures of cognitive and functional status in demented patients. Choice RT68was significantly correlated with scores on a mental status questionnaire, measures ofinstrumental activities and activities of daily living, measures of recent verbal memory,and timed tests requiring visual-motor integration. In contrast, choice RT was poorlycorrelated with predominantly motor tasks, such as finger tapping speed.The choice RT measure for this project was the average of 30 trials of warnedthree-choice RT to randomly ordered red, green, and blue coloured lights. Subjects wereinstructed to place the index finger of their dominant hand on a resting spot between trialsand to respond by pressing the appropriately coloured telegraph key as soon as a colouredlight appeared. Stimuli were presented and reaction times measured using a LafayetteInstrument Model 6302 BX Multi Choice Reaction Timer.4. Picture Completion. The Picture Completion subtest of the WAIS-R certainlytaps attention to visual detail. Interestingly though, Picture Completion has the highestweighting of any Performance Scale subtest on the WAIS general ability factor, and onlymodest weighting on the visuospatial factor of the WAIS. Also, despite its visual content,Picture Completion correlates more with Verbal Scale subtests than with PerformanceScale subtests. Lezak (1983, p. 275) says: "There are also reasoning components to thistest involving judgements about both practical and conceptual relevancies... J. Cohenconsiders this test to be a nonverbal analogue of Comprehension." As discussed earlier,poor performance on Picture Completion was significantly associated with poorer drivercompetence for brain damaged subjects in studies done by Sivak et al. (1981) and Engumet al. (1988).The standard WAIS-R test stimuli consist of booklet of 20 pictures of human69figures and familiar objects or situations in which some important feature is missing. Thesubject is permitted to either name the missing part or simply point to where it should be.There is a 20 second time limit for each picture.5. Trail Making Test - Part B (TM-B). TM-B is a commonly usedneuropsychological measure and is very sensitive to brain damage. Lezak (1983) suggeststhat performance on TM-B involves complex tracking, attentional switching, flexibility,and executive functions. Of the two forms of the Trail Making Test the Part A form (TM-A) is most frequently referred to in the dementia literature. TM-A is part of a battery oftests developed by Berg and colleagues (1984) that has been found to be sensitive to verymild dementia and predictive of decline (Botwinick, Storandt, Berg, & Boland, 1988;Morris & Fulling, 1988; Storandt & Hill, 1989; Tierney, Snow, Reid, Zorzitto, & Fisher,1987). It is likely that the TM-B is also quite sensitive to dementia, but it has been lesspopular because it takes longer and is somewhat less reliable. However, reliability isconsidered adequate; in one study Klonoff, Fibiger, and Hutton (cited in Boll, 1981, p.600) found that TM-B had the highest retest reliability (r= .87) of any measure in theHalstead-Reitan Neuropsychological Battery. As mentioned previously, TM-B is favouredby some investigators concerned with psychometric prediction of driving performance inbrain damaged individuals. Engum et al. (1988) and van Wolffelaar et al. (1988) foundthat the Trail Making Test (they do not specify whether they used Part A, Part B, or both)was correlated with indices of driving ability. Based on a review of the literature on theneuropsychological prediction of driving, Hopewell and van Zomeren (1990) suggest thatTM-B is a particularly good candidate as a predictor test because it is a complex multi-70functional information processing task as is driving.TM-B test comprises an 8 by 11 inch sheet on which 13 numbers and 13 lettersenclosed in circles are distributed across the page. The subject is instructed to draw a linethat links the numbers and letters in ascending sequence while alternating betweennumbers and letters. The subject starts with the letter A, goes to the number 1, then to theletter B, then to the number 2, and so on. The time to correctly compete the task isscored. If a mistake is made, the examiner alerts the subject who is then asked to correctthe error before going on. The test is discontinued if not completed in five minutes.6. Comprehension. The Comprehension subtest of the WAIS-R taps practicalreasoning and common-sense judgement (Lezak, 1983). Berg et al. (1984) have used theComprehension subtest in several studies of dementing subjects. They found that the testwas able to discriminate mildly impaired individuals from normals and also was sensitiveto longitudinal decline in dementing patients. The Comprehension subtest does not appearto have been used in studies of brain damaged drivers. It is of interest in this projectbecause it taps practical reasoning and judgement which may be associated withstrategical level skills.The Comprehension subtest comprises 11 open-ended questions assessing the subject'sunderstanding of societal conventions, appropriate behaviour, and includes three proverbsthat the subject must interpret. The test was administered and scored according to theWAIS-R instruction manual.7. Direct Assessment of Functional Status. In order to document deficits infunctional status, a recently developed behavioral measure of everyday task performance71was used. The scale, known as the Direct Assessment of Functional Status (DAFS), is awelcome addition to the field: the instrument appears to have high interrater and test-retestreliabilities; convergent validity with indexes of symptom severity, and discriminantvalidity with respect to correctly classifying normal elderly versus those suffering fromAD (Loewenstein, Amigo, Duara, et al., 1989). Prior to the introduction of the DAFS,most instruments for assessing the functional status of dementing patients were based onself- or caregiver-report. These instruments may be invaluable for some less observablebehaviours, but they suffer from various reporter biases. Also, functional assessmentinstruments have typically focused on rudimentary self-care functions and neglectedassessment of higher-order functional abilities.The overall score summed across the DAFS subscales was used in this project. Thefive functional domains assessed by the DAFs are: Time orientation (clock setting andreading), Communication (use of a telephone, addressing of a letter), Transportation(interpreting road signs), Financial (counting money, making change, balancing a chequebook), and Shopping (picking out objects from a written shopping list and memory for averbal shopping list). The DAFs was included in order to investigate whether competencein everyday tasks was predictive of driving ability. Hopewell and van Zomeren (1990)have suggested that the search for predictors of driving behaviour may be better served bymultifunctional tests than by measures of discrete cognitive functions.72ProceduresProcedures for Older Participants The two groups of older participants were tested on three separate occasions in themanner detailed below.Day One procedures for older participants:  The first day of testing took place at theAlzheimer Clinic and was primarily devoted to diagnostic and psychometric assessment.When subjects arrived at the Clinic on the first day, approximately 45 minutes weredevoted to a detailed explanation of study procedures during which time the full consentform was read aloud to the subject (see Appendix E for a copy of the consent form).Participants were informed that they could withdraw from the study at any time for anyreason, and that upon completion of the project they would receive $50.On Day One participants were tested on the battery of diagnostic tests (requiringapproximately two hours) and on the 'driving-related' psychometric test battery (alsoapproximately two hours). The participant and their collateral were interviewed separatelyon the structured Driving Interview requiring approximately 45 minutes.Typically, two older participants were seen at the Clinic on a given testing day.Approximately half the older subjects were given the diagnostic battery first and thedriving related psychometric test battery second.Day Two procedures for older participants: Subjects attended GF StrongRehabilitation Centre on a Saturday and took part in three blocks of testing involving theDirect Assessment of Functional Status, the Computerized Driving Assessment Module,73and the MVB road test and visual screening. Typically four subjects were seen on agiven testing day. Participants were scheduled to arrive at one hour intervals and toprogress through the three test blocks in the same order. Approximately three hours wereneeded to complete the testing at this site.Day Three procedures for older participants: Subjects were transported, or came ontheir own, to the off-road test track at the Boundary Bay Airport. Six to eight subjectswere tested on a given day which were always booked on Sundays. The tests of brakingskills and the Cone Avoidance Task took approximately 30 minutes per subject. During allcircuit testing an RCMP driving instructor handled the dual-controls on the instrumentedtest vehicle. The experimenter rode in the back seat to give test instructions and tooperate the on-board computer that monitored vehicle velocity. Testing was done in thesame order for each participant. Two technicians were needed outside the car tocoordinate the placement of equipment for the successive tasks and to measure thestopping distances.Study completion time for older subjects.  The time interval between Day One andDay Three testing varied from as little as three days for some subjects to as much as 5months for two participants. The average number of days between Day One and DayThree was 40.8 (SD = 42.4) for the demented subjects and 41.28 (51.3 = 35.3) for thenormal elderly controls. In both the older groups a little over half the subjects were ableto complete the study within a month and approximately 80% of both groups hadcompleted within two months. For these subjects the interval between the first and thirdtest days was primarily a function of the frequency of Day Three testing. Due to the74considerable expense associated with rental and staffing at the driving track, test dayswere not scheduled until there were six to eight participants who had completed the firsttwo days of assessment. Delays arose when subject recruitment was slow or from theoccasional necessity to cancel a scheduled Day Three due to winter weather. For sevensubjects, four from the demented group and three from the normal elderly group, it tooklonger than 60 days for the three testing days to be completed. These delays were due totravel absences on the part of four subjects and repeated scheduling difficulties for theother three.Procedures for Mid-Aged ParticipantsFor the mid-aged control group, Day One and Day Two were combined and tookplace on a Saturday at GF Strong Rehabilitation Centre. Four participants were seen on agiven testing day. All arrived at the same time and rotated through four test blocks: (a)the neuropsychological diagnostic battery, (b) the driving-related battery of psychometrictests and the Driving Interview, (c) the Direct Assessment of Functional Status (DAFS)and, (d) CDAM and the MVB Road Test. The order of tests within each block wereidentical to that received by the older subjects. Test procedures at the driving track werethe same as for the older participants. Mid-age controls were paid $80 upon completionof the study. Funding for the mid-age control group was obtained separately from, andlater than, the grant for the older subjects. Experience in recruiting the older subjectssuggested that people found the study time consuming and it was decided that it would bewise to budget for extra subject money when applying for funding for the mid-agecontrols. Unfortunately it was not possible to increase the honorarium paid to the older75subjects.Study completion time for mid-age subjects. The average time interval to completethe two days of testing required for the mid-age subjects was 7 days (S2 = 5.4). Sevenof the group completed the study in a single weekend, another seven completed withineight days, and the remaining two completed the study within three weeks. The markeddifferences in testing interval for the mid-age controls compared to the older groups was afunction of the ease with which mid-age individuals could be recruited and the groups ofsix to eight subjects assembled for a test day at the track.76RESULTSThe results from this study will be described in three sections: (a) descriptivemeasures, (b) driving behaviour measures, and (c) relations between psychometric testsand driving performance.Descriptive Measures The results presented in the following subsections devoted to driving history andexposure are based on data obtained from the Driving Interview. Ratings were obtainedfrom subjects and, whenever possible, also from a collateral familiar with the subject'sdriving habits. Collateral ratings for the descriptive questions were obtained because therewas concern that demented subjects would have difficulty recalling factual informationrelated tO their own driving history and habits. Also, as will be discussed later, collateralratings for all groups were desirable for the less "neutral" questions on the DrivingInterview that requested information about the occurrence of driving problems. Forconsistency, tables will include both participant and collateral ratings (where available) foreach of the three groups.Collaterals were available for all 18 demented subjects. Of these, 11 were spousesof the demented subjects, 3 were offspring, 3 were friends, and 1 was a sibling. For thenormal elderly, 10 of the collaterals were spouses, 3 were offspring, 3 were friends, and 2normal elderly subjects (both spouses of demented individuals) had no suitable collateralfor the Driving Interview. For the mid-age controls, 8 collaterals were spouses, 1 was anoffspring, 4 were friends, and 5 mid-age controls were unable to provide the name of an77individual familiar with their driving.In the remainder of this section the following convention will be used whendiscussing responses to the forced choice questions on the Driving Interview: a validresponse will refer to a response that was one of the descriptor options presented by themultiple-choice questions. An item might be without a valid response for one of threereasons: (a) the respondent might have refused to select one of the available options, (b)the respondent might have said they did not know the answer to the question, or (3) theremay not have been the opportunity to ask the question, as would be the situation if nocollateral were available. All items without valid responses were classified as missingdata.For the normal elderly and the mid-age groups there were considerable amounts ofmissing data for collateral ratings on some of the Driving Interview questions. Themajority of missing data for the normal elderly were attributable to don't know answerson the part of collaterals and to the fact that collaterals were not available for two of thenormal elderly subjects. For the mid-age controls, missing data were due to both theabsence of collaterals for five subjects and also to don't know answers from some of theavailable collaterals.While unfortunate, the missing data for control group collaterals was notconsidered to be a serious problem for the questions considered in this section. Althoughit would clearly have been better to have a full set of collateral data, it is reasonable toaccept the self-ratings of the subjects in the normal control groups with respect to drivinghistory and exposure. The control subjects were carefully screened to rule out cognitive78impairment and they would be expected to be at least as well, if not better, informedabout their own driving history as a collateral would be.Driving history. The question related to duration of driving experience waspresented within a multiple-choice format for which respondents were requested toindicate the number of years the subject had been driving by selecting the appropriaterange (in spans of 10 years) from six options. The following is based on collapsing theoriginal six ranges into three 20-year-spans.Demented subjects and their collaterals both reported that 16 (88%) of thedemented had been driving for 40 or more years while 2 (11%) had been driving for 20 to39 years. Based on the self-reports of the normal elderly controls, 12 (67%) had beendriving for 40 or more years and 6 (33%) had been driving for 20 to 39 years. The self-report of the mid-aged indicted that 14 subjects had been driving for between 20 and 39years and 4 subjects had driven for 5 to 19 years.The primary purpose of this question was to assess whether there might betroublesome differences between the demented and normal elderly group in the amount ofprevious driving experience. Of particular concern would be a situation in which thedemented were found to have markedly less driving experience than the normal elderly,thereby confounding deficits that might be found in the driving skills of those in thedemented group. This was clearly not the case for the subjects in this study. Comparingthe data for the demented with the self-reports of the normal elderly, it appears that anydifferences between the two older groups tend in the direction of the demented havingmore years of driving experience than do the elderly controls.79Driving exposure. Driving exposure was investigated by examining five differentquestions from the Driving Interview. Table 5 presents the ratings of participants andtheir collaterals for the number of miles driven over the preceding 12 months. Both theself-reports of the demented and the ratings of their collaterals indicated that the dementedsubjects had less milage in the past year than did the normal elderly. However, inspectionof the data also shows that the self-ratings of the demented subjects and the ratings oftheir collaterals were significantly different with respect to the proportion of subjects inthe lowest milage category (less than 5,000 miles) versus those in higher milagecategories, X' (1, N = 33) = 4.28, 2. <.05. Fifteen (83%) of the collaterals for thedemented reported that the subject drove less than 5,000 miles in the previous yearwhereas 8 (50%) of the demented placed themselves in this low milage category, and 5(28%) of the demented subjects claimed to have driven 5,000 to 7,900 miles in theprevious year.It is clear from Table 5 that considerably more of the mid-age controls (i.e., 44%)fell in the highest milage category (12,000 or more miles / year) compared to the smallnumbers of elderly who drove that much ( 5.6% of the demented, and 5.6% or 11.1% ofthe normal elderly). Thus, it appears that the mid-age subjects had a greater annualmilage than did the older subjects and it seems likely that the demented drove fewer milesin the 12 months preceding participation in the study than did the normal elderly.The remaining four questions about driving exposure (trips per week, frequency ofdriving after dark, frequency of driving in rush hour, and extent of driving avoidance) lentthemselves to parametric analyses to assess the occurrence of both group and raterTable 5Driving Exposure: Miles in Previous YearGroupsNormal^Mid-ageDemented^Elderly ControlsMiles Driven per Year:Less than 5,000 miles per yearP. Report' 50.0% 33.3% 11.1%C. Report 83.3% 33.3% 16.7%5,000 to 7,900 miles per yearP. Report 27.8% 27.8% 22.2%C. Report 5.6% 11.1% 22.2%8,000-11,999 miles per yearP. Report^5.6%^27.8%^22.2%C. Report 5.6%^16.7%12,000 or more miles per yearP. Report 5.6% 11.1% 44.0%C. Report 5.6% 5.6% 5.6%(table continued)80Table 5 (continued)Driving Exposure: Miles in Previous YearGroupsNormal^Mid-ageDemented^Elderly ControlsMiles Driven per Year (continued):DK/missingbP. Report^11.1%C. Report^5.6%^44.5%^38.9%a P. Reports are the participants' self-reports to interview items. C. Reports are thereports of the collaterals.b In some cases the respondents did not know (DK) the answer to the item; or thequestion was not asked or the answer could not be coded in the available responsecategories (missing).8182(i.e., self-ratings vs. collateral ratings) differences. These questions were analyzed with 3X 2 (Group X Rater) mixed analyses of variance with rater as the repeated measure.Significant main effects were followed up with post-hoc pairwise multiple comparisonsusing the Spjotvoll-Stoline modifications of Tukey's HSD test which is appropriate whenthere are marginal differences between groups in sample size (Kirk, 1982).The average number of trips per week as reported by both subjects and collateralsare presented in Table 6. The main effect for group was significant, F(2, 30) = 3.30, 2<.05. However, none of the follow-up pairwise comparisons between groups weresignificant. Reduced sample sizes and high within-group variability are likely causes forthe absence of significant pairwise differences. The main effect for rater was notsignificant nor was the interaction of Group by Rater.As can be seen in Table 6, there are a considerable amount of missing data for thisquestion. The values presented in the table include only data for subjects for whom therewere valid ratings from both the subject and his or her collateral. Fifteen of the 18subjects in the demented group had valid ratings from both raters (self and collateral); themissing data for this group was mainly attributable to subjects who were unable to recallhow many trips they took for the various purposes discussed in the question.Unfortunately, there were valid data for only 7 pairs of ratings for the normal elderly.Missing data for 3 collaterals of normal elderly was due to an error in Driving Interviewadminstration on this particular question. For another 5 normal elderly subjects,collaterals' ratings were unavailable due to don't know answers, and 2 normal elderlysubjects had no collateral. For the mid-age subjects the majority of missing data was aTable 6Driving Exposure: Trips per Week GroupsNormal^Mid-ageMeasure^Demented^Elderly ControlsNumber of car trips per week: Participant reportM 8.37 8.16 12.10SD 6.21 4.18 7.06n 15 7 11Number of car trips per week: Collateral reportM 6.51 6.82 11.09SD 4.17 4.19 4.04n 15 7 118384function of the fact that 5 subjects were unable to identify a collateral familiar with theirdriving.The means and standard deviations for the two questions related to driving inhigher risk situations are presented in Table 7. For the question concerned with thefrequency of driving after dark, significant main effects were found for both Group, F(2,42) = 10.5, n <.001, and Rater F(1,42) = 5.97, E <.02. Post-hoc pairwise comparisons ofgroup means showed that while the demented LM = .82) and normal elderly 1Vl = 1.2) didnot differ from each other with respect to the number of times per week they drove afterdark, both of the older groups drove less at night than did the mid-age controls^= 3.2).The main effect for Rater was significant with the overall average of self-ratings =1.9) being greater than the average of collateral ratings 1VI = 1.3). The interaction ofGroup by Rater was not significant.For the last two exposure questions--those related to driving in rush hour and toextent of driving avoidance--groups were of unequal sizes and there were failures to meetthe assumptions of multivariate homogeneity of variance-covariance matrices and/orunivariate homogeneity of variance that are associated with repeated measures analysis ofvariance.3 For these analyses the correlations between the sample sizes and the samplevariances of the three groups were positive in sign resulting in F tests that would be3 Because there were only two raters there was no need to test assumptions related tohomogeneity of covariance that are associated with repeated measures analysis of variance(Winer, 1971, p. 537). Using SPSS-X MANOVA, tests of univariate homogeneity of variance(the Bartlett-Box procedure) and the multivariate test for homogeneity of variance-covariancematrices (Box's M test) were carried out prior to each repeated measures analysis of variance.Table 7Driving Exposure: High Risk Situations GroupsNormal^Mid-ageMeasure^Demented^Elderly ControlsTimes per week drive after dark: Participant reportM 1.35 1.46 3.15SD 1.49 1.35 1.77n 17 15 13Times per week drive after dark: Collateral reportM 0.41 0.87 3.15SD 1.27 1.88 1.99n 17 15 13Times per week drive in rush hour: Participant reportM 1.18 0.71 3.92SD 2.08 0.91 1.78n 17 14 1285(table continued)Table 7 (continued)Driving Exposure: High Risk Situations GroupsNormal^Mid-ageMeasure^Demented^Elderly ControlsTimes per week drive in rush hour: Collateral reportM 0.84 0.71 4.17SD 1.70 1.49 1.74n 17 14 128687conservative with respect to Type I error (Glass, Peckman & Sanders,1972; Tabachnick& Fidell, 1983, p. 233). Given that the larger variances and covariances were associatedwith the larger samples (hence the conservative F tests) and because the probability levelsassociated with the tests were mainly quite small (ranging from .02 to .001), the outcomesof the analyses were accepted as reasonable estimates of the likelihood of significanteffects for the variables in question.The groups differed significantly with respect to the number of times per weeksubjects drove during rush hour, F(2, 40) = 22.57, 2 < .001. Follow-up multiplecomparisons revealed that the mid-age controls drove more frequently during rush hour( = 4.0 times per week) than did the demented (M = .97) or the normal elderly (M =.71). The older groups did not differ from each other. Neither the main effect for Raternor the interaction of Group by Rater were significant.The results for the driving avoidance question were very similar to those obtainedfor the question about rush hour driving. The group averages for subjects' and collaterals'ratings are presented in Table 8. The main effect for group was significant, F (2,41) =7.37, 2 < .002. The demented LM, = 52.9) and the normal elderly Nl = 52.6) did notdiffer from each other, while both older groups evidenced more avoidance behaviour thandid the mid-age controls 1Vi = 43.4). There was no significant difference between ratersnor was the interaction of Group by Rater significant.In overview, on all questions except that related to trips per week, the older groupsdrove less than the mid-age group. The demented and normal elderly did not differ fromeach other in exposure to night time and rush hour driving nor in the extent of avoidanceTable 8Driving Exposure: Avoidance of Driving SituationsGroupsMeasure DementedNormalElderlyMid-ageControlsAvoidance of driving situations: Participant reportMa 52.88 52.64 43.42SD 11.21 7.87 8.23n 15 16 13Avoidance of driving situations: Collateral reportM 55.93 49.61 43.64SD 12.02 8.09 4.47n 15 16 13'Scores for both participant and collateral reports were standardized and scaled tohave a mean of 50 and standard deviation of 10 across the three groups. Higherscores indicate more avoidance.8889of demanding driving situations. However, the demented appear to have driven fewermiles in the previous 12 months than did the normal elderly. Issues associated withdifferences between the raters for the demented group and inconsistencies between thedifferent exposure estimates will be addressed in the discussion section.Motor Vehicle Branch Driving Records.  Table 9 presents the data obtained fromMVB records of accidents and traffic violations (tickets) for the five year period prior tostudy participation. Included in the table are the numbers of accidents and tickets for thethree groups and the proportion of subjects in each group with one or more accidents orone or more tickets. Because accident rates in small samples are not likely to be reliableindicators of the population rates, the data have not been subjected to statistical analysesand are included merely for descriptive purposes.Visual acuity. Table 10 presents the results of visual screening using the Keystonevision tester. The MVB criteria for passing the test requires that individuals have abinocular visual acuity of 20/40 or better, with corrective lenses if needed. Twoindividuals, one in the normal elderly group and one in the demented group did not meetthis criterion. The visual acuity of the individuals who failed the screen was between20/40 and 20/50. Although they would have been denied licence renewal on an officialMVB test, these individuals were regular drivers and were not considered to besufficiently visually handicapped as to be excluded from the study. One demented subjectcould not be tested because he was unable to name the letters presented as the stimuli forthe acuity test. From this individual's performance on psychometric tests, and based onthe report of his spouse, there was no reason to suspect that he suffered from significantTable 9Motor Vehicle Branch Record of Tickets and Accidents for Previous Five YearsGroupsNormal^Mid-ageMeasure^Demented^Elderly Controls(n=18) (n=18)^(n=18)Subjects with one or more accidents in previous 5 years^8^5^6Total number of accident for group in previous 5 years10^7^8Subjects with one or more tickets in previous 5 years4^7^7Total number of tickets for group in previous 5 years10^12^3090One subject in the mid-age group had 20 tickets in the 5 year periodTable 10Visual Acuity for Dementia and Control GroupsGroupsNormal^Mid-ageMeasure^Demented^Elderly Controls(n=18) (n=18)^(n=18)N^%^N^%^N^%Keystone Assessment of Visual AcuityPass'FailUntestable161188.85.65.617194.45.618 100a A pass on this test corresponds to a visual acuity of 20/40 or better.9192visual impairment. Rather, the deficit exhibited on the visual acuity test was believed tobe part of a language disorder involving deficits in letter recognition and naming. Apartfrom the these three individuals, all other subjects passed the screening test.Driving Behaviour Measures The results of the analyses of the driving behaviour measures will be discussed infive subsections. The basic finding related to group difference on the driving behaviourmeasures will be discussed in three sections corresponding to the three levels of thehierarchical model. A fourth subsection will be devoted to examining the extent to whichperformance on the driving measures is helpful in discriminating between normal elderlyand dementing elderly drivers. Finally, sex differences on the driving measures will bebriefly discussed.Unless otherwise specified, one-way ANOVAs were conducted with groups ofequal sizes and were followed by post-hoc pairwise comparisons of group means usingTukey's Honestly Significant Difference (HSD) test (Kirk, 1982, pp. 116-118). However,there were sizable differences in group variances for the simulator brake reaction time andsteering tasks, the demerit points on the road test, and for the actual-cone-hit scores on thesecond and third trials of the Cone Avoidance Task. Although these differences constitutea violation of the assumption of homogeneity of variance, Box (described in Howell,1982, p.297) has shown that even when variances are unequal a valid and conservativetest of the significance of group differences can be carried out if the degrees of freedomfor Fcrilical are altered from df=(k-1, k(n-1)) to df=(1, n-1). For the analyses of groupdifferences for the five variables named above, the Fcritical values with degrees of freedom93adjusted for heterogeneity of variance (1 df for the numerator, and 17 df for thedenominator) were: 4.45 at 2 <.05; 8.40 at 2. <.01; and 15.72 at 2 <.001. Follow-upmultiple comparisons for these variables were conducted using Scheffe's procedure whichis conservative and robust with respect to heterogeneity of variance (Kirk, 1982, p.121).Operational level measures: Group differences.  The means and standard deviationsfor two driving simulator measures are presented in Table 11.For simulator brake reaction time there was a significant effect for group T =15.7, 2 < .01 with df = (1, 17) as per the Box procedure for adjusting Fes,, whenvariances are heterogeneous). The demented group had a markedly longer mean reactiontime than either of the control groups based on follow-up multiple comparisons usingScheffe's method. The normal elderly and mid-age controls did not differ significantlyfrom each other for simulator braking time.The brake reaction time scores were positively skewed for the older groups,particularly for the demented group. The medians may be a better indicator of centraltendency for the brake reaction time scores and are presented along with the means inTable 11. Although the differences between the medians for demented group and the twocontrol groups are somewhat attenuated compared to differences between the means, thereis still a very significant group effect. The outcome of a Kruskal-Wallis one-way analysisof variance was highly significant, X' (2, N = 54) =17.7, 2 . < .001. The mean ranks forthe demented, normal elderly, and mid-age groups were 40.2, 22.3, and 19.8 respectively.On the steering deviation task the group effect was highly significant T. = 89.5, 2< .001 with the degrees of freedom for the Ferrue., adjusted to [1,16]). The demented94Table 11Driving Behaviour: Operational LevelGroupsNormalMeasure^Demented^Elderly(n=18) (n=18)Mid-ageControls(n=18)ANOVA2.Simulator Brake Reaction Time (seconds)M^ 2.34 a^0.78 b 0.63 b <.01SD 1.71^0.40 0.12Range^.45-6.0^.40-1.8 .41-.93Median Simulator Brake Reaction Time (seconds)1.89^0.66 0.60Steering Deviation on Simulator (area under curve mm2)M^ 705.6 a^380.4 b 252.3 c <.01SD 121.2^122.1 50.5Range^477-952^227-686 166-343Note. Given marked heterogeneity of variance, the significance of the F tests wereevaluated by Box's procedure for adjusting the degrees of freedom for F ^(described inHowell, 1982, p.297). Pair-wise multiple comparisons were conducted by Scheffe'smethod. Groups with different superscripts are significantly different at the .05 level.Superscript' is associated with the lowest performance level.95performed significantly less well IVI = 705.6 mm2) than did either of the control groups.Also, the mean steering deviation score for the normal elderly VI = 380.4 mm2) wassignificantly larger than for the mid-age group IVI = 252.2. mm2). There was onedemented subject for whom a steering score was not available due to equipment failure.Manoeuvring level measures: Group differences. Group means and standarddeviations for the MVB road test and the measure of stopping distance in response to themoving hazard are presented in Table 12.A one-way ANOVA with MVB demerit points as the dependent variable wassignificant (..E = 14.8, p < .01 with (1,17) as the adjusted df for Fcriial). Follow-uppairwise comparisons using Scheffe's procedure indicated that the demented hadsignificantly more demerit points than did either control group. Also the normal elderlyand mid-tage groups differed significantly from each other with the normal elderlyperforming less well on the road test than did the mid-age controls.The maximum attainable score on the MVB road test is 40 demerit points which,once accumulated, results in discontinuation of the test and constitutes a failure. Fivesubjects (28%) in the demented group failed the road test while no subjects in eithercontrol groups failed. The chi-square test of independence was significant (X2 (2, N =54) = 11.1, p. < .004) indicating that the difference in failure rate between the dementedand the controls was greater than would be expected by chance alone.For the second manoeuvring level task there was also a significant effect for group.The demented group had a longer emergency stopping distance in response to the movingstyrofoam vehicle than did either of the control groups, F(2,51) = 5.2, p < .01. The96Table 12Driving Behaviour: Manoeuvring LevelGroupsNormal^Mid-age^ANOVAMeasure^Demented^Elderly Controls 2.(n=18) (n=18) (n=18)Total Demerit Points on Road TestM^ 32.11 a^21.38 b 13.00 c <.01*SD 13.47^10.15 6.46Range^10-40^0-37 2-22Stopping Distance in Response to a Moving Hazard (meters)M^ 18.10 '^15.85 b 15.32 b , <.008SD 2.59^2.34 3.15Range^15-22.5^11-19.5 9-19.4Note. Superscripts appearing in the body of the table are associated with post hoc pairwisecomparisons between group means. Groups with different superscripts are significantlydifferent at the .05 level. Superscript' is associated with the lowest performance level.*Given marked heterogeneity of variance, the significance of the F test was evaluated byBox's procedure for adjusting the degrees of freedom for Fcritica (described in Howell,1982, p.297). Pair-wise multiple comparisons conducted by Scheffe's method.97normal elderly and the mid-age groups did not differ in their stopping distances.Collisions or near-collision (i.e., a situation in which impact was averted only because ofoverride braking in the dual control car) occurred for four of the demented group (22%)and for none of the subjects in the control groups.Strategical level measures: Group differences.  As mentioned previously, twocomponents of strategical level behaviour were selected for examination: the accuracy ofsubjects' self-appraisals and the concept of compensatory avoidance.The discussion of compensatory avoidance will involve an inquiry about therelations between exposure measures and indices of driving problems. These measureshave already been discussed in various subsections of this chapter. No additionalstatistical analyses will be carried out and it seems most appropriate to address the topiclater on in the discussion section.The accuracy of subjects' self-appraisals was assessed in two different contexts: onthe Cone Avoidance task and by comparing subjects' self-ratings to their collaterals'ratings on the Driving Interview questions related to driving problems.The results of the Cone Avoidance task are presented in Tables 13 and 14. Inorder to provide an overview of the results for this task, Table 13 presents means andstandard deviations for predicted cone hits and actual cone hits on a trial by trial basis.On each of the three trials the demented group hit significantly more cones than didmembers of either control group. There were no significant differences between groupswith respect to predicted hits.In focusing on the accuracy of self-appraisal, the relation between predicted hits98Table 13Driving Behaviour: Strategical Level - Means and Standard Deviations for Cone Avoidance Task GroupsNormal^Mid-age^ANOVAMeasure^Demented^Elderly Controls n(n=18) (n=18) (n=18)Trial One: Mean predicted and actual hitsM Predicted^0.55 1.44 2.11 .060SD^0.92 1.85 2.61M Actual^3.67 a 1.88 b 0.94 b < .002SD 2.70 2.39 1.35Trial Two: Mean predicted and actual hitsM Predicted^1.00 1.28 0.67 .587SD^1.68 2.08 1.49M Actual^2.72 ' 0.44 b 0.55 b <.01*SD 2.42 0.61 1.15Trial Three: Mean predicted and actual hitsM Predicted^0.89 0.17 0.50 .185SD^1.64 0.38 1.09(table continued)99Table 13 (continued)Driving Behaviour: Strategical Level - Means and Standard Deviations for Cone Avoidance Task GroupsNormal^Mid-age^ANOVAMeasure^Demented^Elderly Controls P.(n=18) (n=18)^(n=18)Trial Three: continuedM Actual^2.39 a^0.28 b^0.22 b^<.01*SD 2.42^0.61^1.15*Given marked heterogeneity of variance, the significance of the F tests were evaluated byBox's procedure for adjusting the degrees of freedom for Fa  (described in Howell,1982, p.29'7). Pair-wise multiple comparisons were conducted by Scheffe's method.100and actual hits was considered to be more informative than the results for either of thesevariables taken separately. The means presented in Table 14 are difference scorescomputed by subtracting the sum of actual hits for Trials 2 and 3 from the sum ofpredicted hits on the same two trials. The first trial of the task served the function ofproviding subjects with experience in the vehicle-manoeuvring and prediction componentsof the task. After the first trial (and on each subsequent trial), subjects were givenfeedback about the number and position of displaced cones. Thus, predictions made forthe second and third trials were "educated" estimates based on experience with the task.Two approaches to quantifying the relations between predicted and actual hits arepresented in Table 14. The top of the table presents the group averages and standarddeviations for the absolute value of the difference scores derived from subtracting actualhits froni predicted hits. A significant group effect was found with the absolute value ofdifference scores as the dependent measure ( (2,52) = 9.0, 2 < .001). Follow-up multiplecomparisons indicated that the difference between predicted hits and actual hits wassignificantly larger for the demented group than for either of the control groups. Thus,irrespective of the direction of the mismatch between predicted and actual hits, thedemented were less accurate in their predictions than were the controls.The lower portion of Table 14 shows the average of the signed difference scores.A one-way ANOVA on the signed difference scores was significant LF (2,52) = 9.6,<.001). The demented group differed from both control groups on post-hoc comparisons.The advantage of the signed difference scores is that the direction of the mismatchbetween predicted and actual hits is preserved. For instance, the fact that the mean for the101Table 14Driving Behaviour: Strategical Level - Difference Scores between Predicted Hits and Actual Hits on the Cone Avoidance TaskGroupsNormal^Mid-age^ANOVAMeasure^Demented^Elderly Controls n(n=18) (n=18)^(n=18)Cone Avoidance Task: Mean of absolute difference scores between predicted and actualhits (summed across trials 2 and 3).MSD .4.11^a3.160-131.39 b1.750-50.94 b2.130-9<.001RangeCone Avoidance Task: Mean of signed difference scores between predicted minus actualhits (summed across trials 2 and 3).M -3.22 ' 0.77 b^0.39 b <.001SD 4.11 2.14 2.30Range -13 to 6 -3 to 5 -2 to 9Note. Superscripts appearing in the body of the table are associated with post-hoc pairwisecomparisons between group means. Groups with different superscripts are significantlydifferent at the .05 level. Superscript' is associated with the lowest performance level.102demented is a negative score indicates that subjects in this group were likely to under-predict the number of cones they would hit relative to their actual hits. Means derived bysumming across signed scores could, in some circumstances, be difficult to interpret.However, in this study there was a high degree of consistency within groups with respectto the direction of the mismatch between predicted and actual hits. On a case by casebasis, 14 of the 18 demented subjects under-predicted the number of cones they would hitacross Trials 2 and 3, while only 4 normal elderly and 4 mid-age controls had negativedifference scores across those same trials.In addition to the Cone Avoidance task, a second set of measures was examined toevaluate the accuracy of subjects' self-appraisals. A comparison was made betweensubjects' self-ratings of driving problems and the ratings given by their collaterals on thesame set of items. The derivation of the driving problem scores (which are based onresponses to 56 items) is discussed briefly in the Measures section and is detailed inAppendix C. Means and standard deviations for participant and collateral scores fordriving problems are presented in Table 15. For clarity of presentation the summaryscores were standardized and scaled in T-score metric such that they have a mean of 50and a standard deviation of 10.Unfortunately, the number of subjects included in this analysis is considerablyreduced compared to the sample sizes available for the driving behaviour measures.Because the key focus in this analysis involved the comparison of subjects' self-ratings tothe ratings of their collaterals, it was necessary to have valid responses from both raterson the majority of the large number of driving problem items covered in the Driving103Table 15Driving Behaviour: Strategical Level - Participant and Collateral Ratings of Driving Problems GroupsNormal^Mid-ageMeasure^Demented^Elderly Controls(n=10) (n=11)^(n=10)Driving Problems: Participant report'M 51.57 49.66 49.38SD 9.64 7.36 7.70Range 37-71 37-61 41-65Driving Problems: Collateral reportM 61.23 45.44 43.19SD 13.46 6.65 3.79Range 42-92 37-59 38-51' The derivation of the Driving Problems scores is described in Appendix C. Scores forboth participant and collateral reports are a linear combination of results from five multi-item Driving Interview questions. The scores were standardized and scaled to have a meanof 50 and standard deviation of 10 across the three groups. Higher scores indicate moreproblems104Interview. Although there was provision for some prorating of missing data, there wasstill a need to drop between 7 and 8 subjects per group due primarily to missing collateraldata. Appendix C presents details of the procedures for handling missing data on theDriving Interview questions.The driving problem data were analyzed with a 3 X 2 mixed analysis of variancewith rater serving as the repeated measure. Although the main effect of group wassignificant, more important was the fact of a significant interaction of Group by Rater T(2,27) = 6.03, 2 <.02). Tests of simple effects of the interaction showed that thedemented subjects^= 51.5) reported significantly lower levels of driving problems thandid their collaterals 1y = 61.8). There were no significant differences between self-reports and the collateral reports for the control groups.The results from the driving problem ratings, although compromised by the smallsample sizes, echo the outcome on the Cone Avoidance task. In both cases the dementedsubjects presented themselves in a manner that was indistinguishable from the way normalcontrols presented themselves. However, the information from external sources (ie., actualhits for the Cone Avoidance task and the ratings of collaterals for the driving problemquestions) indicated that the demented subjects were performing significantly less wellthan were the controls.Distinguishing between the normal elderly and the dementing drivers. As notedearlier, there were significant group differences between the demented group and thecontrol groups on all measures of driving performance. For practical purposes it would beuseful to go beyond the initial demonstration of statistically significant group differences105to examine the extent to which the driving behaviour tests are helpful in distinguishingbetween individuals within the two groups of older subjects. Two approaches to exploringthis issue are reported below.A discriminant function analysis was performed on the five driving behaviourmeasures in order to identify the combination of measures that provided optimaldiscrimination between the two older groups. Also, a count was made of the number ofsubjects in each group with test scores falling at least one standard deviation (SD) abovethe mean (in the direction of greater impairment) of the normal elderly control group.Because the differences between the normal elderly and the demented are ofspecific interest from a practical point of view, the following analyses focused on justthese two groups and did not include the mid-age group. A two group discriminantanalysis 'was carried out using a stepwise method with minimization of Wilk's lambda asthe variable selection criteria. Three of the five driving measures met criteria for entryinto the equation after which the remaining variables did not result in significant gains inthe accuracy of subject classification. The results of the discriminant analysis are seen inTable 16. The weighted composite of scores from the driving simulator steering task, theCone Avoidance task, and the MVB Road Test defined a highly significant discriminantfunction (X2 (3, N = 35) = 41.39, 2 < .0001).Subject classification was carried out through the SPSSX Discriminant programwhich assigns subjects to groups based on Bayes' theorem (see Norusis, 1988, pp. 82-83).With two groups of subjects the prior probability of group membership was 50% for eachgroup. The overall hit rate was 91.4% for successful discrimination of subjects in theTable 16Results of Discriminant Analysis with Driving Behaviour Measures Measure'Raw-Score DiscriminantFunction CoefficientsStandardized-NormalizedDisc. Fn. CoefficientsSteering Dev. .00746 .85115Cone Task - .15004 -.45941MVB Test .02549 .23886(Constant) -4.83049X2 (3) = 41.39, 2 < .001Actual GroupMembershipPredicted Group MembershipDemented Normal ElderlyDemented 94.1% 5.9%(n=17) (16) (1)Normal Elderly 11.1% 88.9%(n=18) (2)^ (16)106(table continued)107Table 16 (continued)Results of Discriminant Analysis with Driving Behaviour Measures Classification Efficiency'Overall Hit Rate^91.43%False Postives^5.71% (relative to demented group)False Negatives^2.86% ( "^)a Steering Dev. = steering deviation on the driving simulator, MVB Test = demerits pointson the MVB Road Test; Cone Task = number of predicted hits minus actual hits on theCone Avoidance task.Case classification based on Bayes' Rule (see Norusis,1988, p.82) with prior probabilityof .5 for each group.108demented group from those in the normal elderly control group. Of the 35 cases includedin the analysis, 5.7% were misclassified as false positive (classified as demented whenthey were actually normal elderly) and 2.9% of cases were false negatives (classified asnormal when in fact the subject was suffering from dementia).The standardized-normalized discriminant function coefficients presented in Table16 indicate the metric-free contribution of each measure to the discrimination (Hakstian &McLean, 1989). The simulator steering task played a prominent role in differentiating thedemented from the normal elderly subjects. It is interesting to note that optimaldiscrimination between the groups was obtained by a weighted composite of three tests--each representing a different level of the hierarchical driving model.Another approach to gauging the practical significance of group differences on thedriving measures is to evaluate the extent of overlap between the score distributions of thenormal elderly and the demented. In clinical settings it is not uncommon to designate as"impaired" scores that fall beyond some specified point in the (hypothetically normal)distribution of test scores associated with a normal population. Thus, impairedperformance is defined normatively in terms of a score that would occur in only arelatively small proportion of the normal population; for example an "abnormal" scoremight be one that fell at least one, or perhaps two, standard deviations beyond the meanof a sample of normal individuals.In an exploratory context, it is informative to examine the overlap between thescore distributions for the demented and for the normal elderly subjects who participatedin this study. However, given the small size of the "normative" sample, it must be109recognized that the generality of these results is not assured.All measures except the Cone Avoidance task were scaled such that lower scoreswere associated with less impairment and hence scores in the regions above the mean plus1 and 2 SDs are associated with poorer performance. As shown in Table 17 the reverseapplies to the Cone Avoidance task for which lower scores are indicative of poorerperformance. Scores on the Cone Avoidance task are difference scores and the scaling isarbitrary: results would have been statistically identical had predicted cone hits beensubtracted from actual cone hits instead of the other way around. For economy ofexpression the following discussion will be phrased in terms consistent with the scaling ofthe four measures for which higher scores indicate less adequate performance. In the caseof the Cone Avoidance task, the reader is asked to translate the terms "above the meanplus 1 or 2 SDs" to "beyond the mean plus 1 or 2 SDs" in the direction of less adequateperformance.Table 17 presents the percentage of subjects in each group with scores that were(a) less than 1 SD above the mean of the distribution for the normal elderly (i.e., scoresthat were in the "average" and better-than-average range of performance), (b) above themean plus 1 SD placing them beyond the 84th percentile of scores for the elderlycontrols (assuming a normal distribution), and (c) above the mean plus 2 SDs which is theregion of a normal distribution that contains only 2% of values.Given our knowledge of the results of the discriminant analysis, it is not surprisingthat the simulator steering deviation task was the test that yielded the least overlapbetween groups. The difference between the demented and normal elderly was quiteTable 17Proportion of Subjects with Scores Falling in the Extreme Regions of the Score Distributions for the Normal ElderlyGroupsRegion of^ Normal^Mid-ageDistribution Demented^Elderly^Controlsof Normal (n=18)^(n=18) (n=18)Elderly Scores'Simulator Brake Reaction TimeBelow M +1 SD^27%^83%Above M +1 SD^77%^17%Above M +2 SD^61%^5%Steering Deviation on Simulator'Below M +1 SD^6%^78%Above M +1 SD^94%^22%Above M +2 SD^70%^5%Total Demerit Points on Road Test100%100%Below M +1 SD^56%^83%^100%Above M +1 SD^44%^17%Above M +1.9d^28%(table continued)110111Table 17 (continued)Proportion of Subjects with Scores Falling in the Extreme Regions of the Score Distributions for the Normal ElderlyGroupsRegion of^ Normal^Mid-ageDistribution Demented Elderly^Controlsof Normal (n=18)^(n=18) (n=18)Elderly Scores'Stopping Distance in Response to a Moving HazardBelow M +1 SD 67% 83% 83%Above M +1 SD 33% 17% 17%Above M +2 SD 22%Cone Avoidance Task: Predicted hits minus actual hits'Above M +1 SD 28% 95% 95%Below M +1 SD 72% 5% 5%Below M +2 SD 44%Note. The percentage values presented in the table have been rounded to the nearestinteger. The values reported for the proportion of subjects with scores above M + 1 SDalso includes individuals whose scores are in the region above M + 2 SD' The majority of the driving measures were scaled such that a larger score was associatedwith greater impairment. The scores falling below M + 1 SD are those in the "lower" 84%of the distribution of scores for the normal elderly (assuming a normal distribution) andrepresent better performance than the scores falling above this region. The ConeAvoidance task was scaled such that lower scores were associated with poorerperformance.b Due to equipment malfunction one demented subject did not have a score for thesteering deviation task; percentages for the demented group on this measures are based onan n of 17.112dramatic on this measure. Whereas nearly 80% of the normal elderly had scores that fellbelow the mean plus 1 SD, only 1 demented subject had a score in this lower region ofthe distribution of normal elderly scores. For the demented sample, 94% of subjects hadscores that lay at least 1 SD beyond the mean of the normal elderly group, and further,70% of the demented group had scores that were more than 2 SDs above the mean of thenormal elderly.The next most discriminating measure was simulator brake reaction time on which77% of the demented scored at least beyond the mean plus 1 SD and 61% of the grouphad scores that were 2 SDs beyond the mean of normal elderly groups. Although thismay seem somewhat at odds with the results of the discriminant analysis, because the twosimulator measures are intercorrelated, the brake reaction time task did not meet entrycriteria for the discriminant analysis once the steering deviation task had been entered.Pooling the number of "extreme" scores (in the direction of poor performance) persubject across the five driving measures yielded the following results:1. In the normal elderly group, 6 subjects (33%) had one score (out of the fivedriving measures) that fell between the mean plus 1 SD and the mean plus 2 SDs of thedistribution of scores associated with the particular measure. Two subjects had one scorethat was beyond the mean plus 2 SDs of the distribution for the normal elderly group.Only 1 subject in the normal elderly group had scores above the mean plus 1 SD onthree of the five driving tests.2. All 18 subjects in the demented group had scores on one or more tests that wereat least 1 SD above the mean of the normal elderly group. Sixteen subjects (89%) had a113score on at least one test that fell in the region beyond the mean plus 2 SDs of the rangefor normal elderly scores. Fourteen demented subjects (78%) had scores on three or moreof the driving measures that were greater than the mean plus 1 SD of the scoredistributions for the elderly controls.3. In the mid-age group, 4 subjects (22%) had a score on one of the five drivingmeasures that was greater than the mean plus 1 SD of the score range for the normalelderly control group. None of the mid-age subjects had extreme scores on more than onetest and none scored above mean plus 2 SD on any of the five tests.4. For the samples investigated in this study, the following decision rule wouldyield an overall hit rate of 88.2% for dichotomous classification of the normal elderly anddemented subjects. Subjects would be predicted to be in the dementia group if they had aminimum of two extreme test scores (in the impaired direction), at least one of which wasmore than 2 SDs above the mean of the normal elderly control group. This rule wouldresult in misclassification of 1 normal elderly subject (false positive rate = 2.9%) and 3demented subjects (false negative rate = 8.6%). Of incidental interest, note that none ofthe mid-age subjects would be misclassified as demented using the above decision rule.Sex differences on the driving behaviour measures.  Sex effects were examined byconducting a 3 X 2 (Group by Sex) analysis of variance for each of the driving behaviourmeasures. For all five analyses the Group effect was highly significant as would beexpected from the results of the one-way analyses presented earlier. Significant sexeffects were encountered on two of the measures. For the driving simulator steering taskthe main effect of sex was significant, F (1, 46) = 10.4, 2 < .01, with males having greater114steering accuracy 1VI = 423.19 mm 2) than females^= 493.73 mm2). Also the effect ofsex was significant for the MVB Road Test, F (1, 46) = 5.05, p > .05, with males havingfewer demerit points 1VI = 19.23) than females 1VI = 24.35). The three study groups wereeach composed of 12 males and 6 females. There were no significant Group by Sexinteractions on any of the driving measures.Relation Between Psychometric Tests and Driving MeasuresSeven psychometric tests measuring a range of functions including attention,perceptual skill, everyday functional ability and judgement, and psychomotor speed wereadministered to all subjects. Table 18 presents the group means and standard deviationsfor these measures.Group differences were examined using one-way ANOVAs with groups of equalsizes. The Trail Making Test and the Picture Completion and Comprehension subtests ofthe WAIS-R had reasonably homogeneous variances across the three groups. For thesethree measures, Fcritical values were obtained using the usual method for determining thedegrees of freedom for the tabled value of F (i.e., k-1 df for numerator, and k(n-1) df forthe denominator, where k = number of groups). Also for these three measures post-hoccomparisons were done using Tukey's HSD test.For the other four psychometric tests F„iticai values were based on an adjustednumber of degrees of freedom (ie., df for Faitica, = (1,n-1); for these analyses was the df=1, 17) as recommended by Box in situations where there are large difference betweengroup variances. Multiple comparisons between group means for the Letter Cancellationerrors scores, the Stoop task, choice reaction time, and the Direct Assessment115Table 18Mean Performance Scores on Psychometric Tests GroupsMeasure^Demented(n=18)NormalElderly(n=18)Mid-ageControls(n=18)ANOVA2Letter Cancellation ErrorsM^ 9.44 a 3.70 b 4.27 b <^.05 *SD 7.21 3.71 1.00Range^0-22 0-12 0-18S troop Color-Word Naming Time (seconds)M^ 92.55 ' 35.08 b 25.78 b <.001'SD 45.85 13.63 5.47Range^41-199 22-68 16-39Mean Choice Reaction Time (milliseconds)M^ 103.79 ' 72.27 b 62.58 b <001*SD 39.06 13.41 7.78Range^73-245 56-107 50-77Trail Making Test - Part B Time (seconds)M^ 278.62 ' 91.75 b 73.85 b < .001SD 39.92 32.00 34.30Range^170-300 42-155 36-190(table continued)116Table 18 (continued)Mean Performance Scores on Psychometric Tests GroupsMeasure^Demented(n=18)NormalElderly(n=18)Mid-ageControls(n=18)ANOVA2WAIS-R Picture Completion SubtestM^ 8.11 a 12.89 b 11.38^b < .001SD 3.69 2.32 2.76Range^2-17 9-17 5-16WAIS-R Comprehension SubtestM^ 8.27 a 14.22 C 11.94 b < .001SD 3.67 1.70 1.86Range^3-17 11-18 9-16Direct Assessment of Functional Status - Total ScoreM^ 53.66 ' 77.72 b 78.11^b <.001*SD 10.71 1.99 1.81Range^34-68 74-80 75-80Note. Superscripts appearing in the body of the table are associated with post-hoc pairwisecomparisons between group means. Groups with different superscripts are significantlydifferent at the .05 level. Superscript' is associated with the lowest performance level.*Given marked heterogeneity of variance, the significance of the F tests were evaluated byBox's procedure for adjusting the degrees of freedom for F crit,cal (described in Howell,1982, p.297). Pair-wise multiple comparisons were conducted by Scheffe's method.117of Functional Status total scores, were computed using Scheffe's procedure.All F-tests were significant, most with probability levels less than .001. Post-hocmultiple comparisons revealed that for all tests the demented performed significantly lesswell than did the control groups. On one measure only, the Comprehension subtest of theWAIS-R, there was a difference between control groups with the normal elderly groupattaining a significantly higher average score than did the mid-age controls. Thus, theonly age effect detected on the psychometric tests was in the direction of the normalelderly subjects performing better than the mid-aged group.The purpose of administering the psychometric measures was to explore thecorrelations between "office-based" tests and the measures of driving performance. Atthis point it is useful to ask: what might be the practical application of specialized driving-related psychometric tests? For clinicians, the most useful tests would be those that couldhelp predict which cognitively impaired individuals would be likely to experience drivingdifficulties. The most common situation in which these tests would be used is withmildly impaired individuals who have received a diagnosis of dementia, but are not sodeteriorated that driving is considered out of the question. Thus, what is being called forare tests that are correlated with driving abilities within a mildly impaired group ofindividuals.Two different approaches to exploring these relations were employed. The firstwas to examine the matrix of bivariate correlations between driving measures andpsychometric tests within the group of demented subjects. The second approachcomprised a series of covariate analyses that provide specific information about the118contribution of psychometric test scores to the prediction of driving performance aftergroup membership has been accounted for.Prudence is called for when interpreting these correlational analyses given thesmall number of subjects available in this project. Also, because of the small samplesizes and numerous tests, the use of multiple regression could not be justified. Providingone is mindful of the limitations of the available data, examination of the bivariaterelations raises some interesting issues that may be instructive when planning futureresearch.The matrix of Pearson product-moment correlations between the psychometric testsand the driving measures for the sample of 18 demented drivers is shown in Table 19.None of the 35 correlations were significant using a liberal one-tailed test with alphaequal to '.05. Lack of statistical power undoubtedly plays a role in the failure of any ofthe correlations to attain significance. As shown by Cohen (1988; p. 86) given a samplesize of 18 subjects, if the actual population correlation was .40 the probability of rejectingthe null hypothesis with a one-tailed test at alpha .05 would only be 0.52.Notwithstanding the meagre statistical power afforded by the available sample size, itmust be recognized that 80% (28 / 35) of the correlation coefficients in Table 19 are lessthan r = .30. Although obtaining a larger sample would increase the power of thesignificance tests, there is no guarantee that these statistically significant correlationswould be of any practical import in accounting for variance in the driving behaviourmeasures.The coefficients in the correlation matrix presented in Table 19 speak most directly119Table 19Correlations Between Psychometric Tests and Driving Measures Within the Demented Group Psychometric TestaDriving Measureb LC ST RT TM WP WC DFLog Brake RT -.02 -.07 .17 -.07 -.40 -.01 -.10Steering Dev. .21 .20 -.14 .18 -.42 -.16 -.28MVB Test -.03 -.34 .09 -.08 .19 .16 .08Stop Distance -.09 -^.38 .29 -.18 .18 .14 -.14Cone Task -.11 .27 -.35 -.05 .32 .35 .26Note. n=18 except for a missing score for one subject on the steering deviation task. Noneof the above Pearson correlation coefficients were significant at alpha .05.a LC = Letter Cancelation; ST = Stroop; RT = Choice Reaction Time; TM = Trail MakingTest - Part B; WP = WAIS-R Picture Completion subtest; WC = WAIS-R Comprehensionsubtest; DF = Direct Assessment of Functional Status.b Log Brake RT = Log of simulator brake reaction time; Steering Dev. = simulatorsteering deviation; MVB Test = demerit points on MVB Road Test; Stop Distance =vehicle stopping distance on Hazard Avoidance task; Cone Task = predicted minus actualcones hit on the Cone Avoidance task.120to the contribution the individual psychometric tests might be expected to make to anoffice-based prediction of driving behaviour. However, in simply looking at the relationswithin the demented group we are not informed about the larger context that involves theassociations between group membership, driving performance, and scores on psychometrictests.The set of analyses to be described below do not alter the conclusions that wouldbe derived from examining Table 19 but they place these results in a somewhat differentperspective. Rather than simply examining the correlation coefficients for the dementedgroup on its own, another way to evaluate the relation between the driving measures andthe psychometric tests is to ask whether the psychometric tests significantly add to ourability to predict driving performance beyond what we might predict on the basis of ourknowledge of group membership.The statistical procedure selected for addressing this question was conceptuallyidentical to an analysis of covariance with the group factor entered first. However, ratherthan run the analysis as an ANCOVA it was conducted as a regression analysis withdummy coding for the group and interaction effects (Tabachnick & Fidell, 1983, p. 182;Younger, 1985, chap. 14). In these analyses group membership was dummy coded andentered as the first variable followed by the psychometric test scores (the "covariate") andfinally by the interaction term for covariate by group. This sequence of statisticalThe correlation matrices presented in Tables 24 and 25 are based on the 36 subjects inthe demented and normal elderly groups. Although these analyses might have been done with themid-age controls included, the practical concerns addressed here revolve around the differencesbetween the two groups of older drivers. For that reason, the mid-age group was dropped fromthe regression analyses as well as from the discriminant analysis discussed earlier.121operations reverses the familiar ANCOVA objective of evaluating group difference afteradjusting for the variance associated with the covariate. In the analyses reported here theobjective was to determine the significance of the relationship between the dependentmeasures (i.e., the driving tests) and the covariate (i.e., the psychometric tests) afteradjusting for group membership.There were a number of advantages associated with running the analyses with aregression model. In addition to generating all the F-ratios, t-values, and interaction termsthat are the core of an ANCOVA, the regression approach also provided a correlationcoefficient for the relation between the group factor and the dependent variables (i.e., thepoint biserial correlations between the dummy coded group variable and the drivingmeasures), and a part-correlation expressing the association between the covariate and thedependent variables after adjusting for group membership. Another advantage was theflexibility of the SPSS-X Regression program for examination of outliers and of residualsin order to detect serious violations of the assumptions of normality and homoscedasticity.Separate regression analyses were conducted for each driving measure paired witheach psychometric test. For all analyses the standardized residuals were plotted againstpredicted values to examine whether there might be systematic patterns in the residualssuggestive of greater variability in portions of the data (i.e., heteroscedasticity). Thescatter of residuals were reasonably random for all sets of analyses except those thatinvolved simulator brake reaction time as the dependent variable. Also skewness andkurtosis of residuals (and the associated standard errors) were computed. For all analyses,except the sets associated with simulator brake reaction time and the Cone Avoidance122task, the Z scores associated with indices of skewness and kurtosis were less than + 2.58.Tabachnick and Fide11 (1983, p. 79) state that Z values in excess of + 2.58 would lead torejection of the assumption of normality of the distribution at p < .01.The simulator brake reaction time measure was significantly positively skewed.Also, for this dependent measure there was evidence of an increased spread of residuals atlarger values of the predicted variable. Following a log- transformation of the brakereaction time data, skewness was markedly reduced and the distributions of plottedresiduals were improved. In the set of analyses in which the Cone Avoidance task servedas the dependent variable the residuals were not normally distributed. However, despitesignificant kurtosis (Z scores for kurtosis ranging from 4.9 to 6.5) the Z scores forskewness never exceeded 2.58 and the residual plots were satisfactory with no particulartrend suggestive of heteroscedasticity. Accordingly, the scores for the Cone Avoidancetask were left untransformed.Prior to examining the correlation coefficients, it was established that none of theinteractions between group membership and the psychometric tests yielded significant F-tests. The significance test associated with the interaction term tests the assumption ofequality of regression slopes between groups. Should this test yield a statisticallysignificant result, it would suggest that regression lines for the groups should be fittedseparately.Taken together, the three matrices of correlations shown in Tables 20 and 21provide an overview of the different facets of the relations between driving performance,group membership, and psychometric test scores.Table 20Correlations Between Psychometric Tests and Driving Measures in Combined Group of Older Subjects and Point Biserial Correlations of Group Membership with Driving Measures Psychometric TestaDriving Measureb LC ST RT TM WP WC DFLog Brake RT .31 .44 .46 .60 -.57 -.53 -.62Steering Dev. .41 .64 .37 .81 -.59 -.67 -.74MVB Test -.08 -.07 .16 .28 -.16 -.18 -.28Stop DiStance .11 .12 .38 .37 -.26 -.27 -.42Cone Task -.24 -.20 -.48 -.50 .53 .53 .54Point Biserial Correlation of Driving Measures with Group MembershipLog Brake RT r (36) = .61, p < .0001Steering Dev. r (35) = .81, 2 < .0001MVB Test r (36) = .34, p < .044Stop Distance r (36) = .43, D. < .009Cone Task r^(36) = .53, 2 < .001(table continued)123124Table 20 (continued)Correlations Between Psychometric Tests and Driving Measures in Combined Group of Older Subjects and Point Biserial Correlations of Group Membership with Driving Measures Note. N=36 except for the steering deviation measure where N=35.a LC = Letter Cancelation; ST = Stoop; RT = Choice Reaction Time; TM = Trail MakingTest - Part B; WP = WAIS-R Picture Completion subtest; WC = WAIS-R Comprehensionsubtest; DF = Direct Assessment of Functional Status.b Log Brake RT = Log of simulator brake reaction time; Steering Dev. = simulatorsteering deviation; MVB Test = demerit points on MVB Road Test; Stop Distance =vehicle stopping distance on Hazard Avoidance task; Cone Task = predicted minus actualcones hit on the Cone Avoidance task.125The entries in the matrix at the top of Table 20 are the correlations between thedriving measures and the psychometric tests before group membership is accounted for.The size of these correlations are largely a function of the fact that they are based on thecombined groups of demented and normal elderly (N=36) who differ significantly on eachof the variables that make up the bivariate coefficients. The matrix is presented in orderto contrast these "uncorrected" correlations with the matrix of part-correlations that arisewhen group membership is accounted for.The lower half of Table 20 shows the point biserial correlations between thedriving behaviour measures and the dummy coded group membership variable that wereobtained in the first step of the regression analysis. These correlations reflect the nowfamiliar group differences on the driving behaviour measures that have been reported inthe context of the one-way analyses of variance, the discriminant analysis, and thediscussion of overlaps between the score distributions of the normal elderly and thedemented. It is interesting to revisit these groups difference in driving performance in acorrelational context. Contrasting the point biserial coefficients with the correlationsbetween the psychometric tests and driving behaviour (presented in the matrix at the topof Table 20) illustrates the fact that with few exceptions, group membership is morestrongly associated with the driving performance measures than are scores on thepsychometric tests.The part-correlation coefficients displayed in Table 21 indicate the correlationsbetween the psychometric tests and the driving measures after adjusting for groupmembership. Probability levels associated with the F Change statistic for these126Table 21Part Correlations Between Psychometric Tests and Driving MeasuresPsychometric TestaDriving Measureb LC ST RT TM WP WC DFLog Brake RT .31 .44 .46 .60 -.57 -.53 -.62Log Brake RT .05 .06 .19 .12 -.26 -.14 -.20Steering Dev. .06 .12 .03 .16 -.12 -.12 -.11MVB Test -.01 -.20 -.01 -.17 .05 .09 -.01Stop Distance -.06 -.22 .20 -.06 .01 .05 -.12Cone Task -.01 .19 -.26 -.04 .25 .21 .17Note. N=36 except for the steering deviation measure where N=35.a LC = Letter Cancelation; ST = Stroop; RT = Choice Reaction Time; TM = Trail MakingTest - Part B; WP = WAIS-R Picture Completion subtest; WC = WAIS-R Comprehensionsubtest; DF = Direct Assessment of Functional Status.b Log Brake RT = Log of simulator brake reaction time; Steering Dev. = simulatorsteering deviation; MVB Test = demerit points on MVB Road Test; Stop Distance =vehicle stopping distance on Hazard Avoidance task; Cone Task = predicted minus actualcones hit on the Cone Avoidance task.127coefficients range from .056 to .98, thus none of the part-correlations were significant atalpha .05. Comparing the matrix of part-correlation coefficients in Table 21 with thematrix of "uncorrected" correlations in Table 20, again reinforces the fact that once groupmembership is taken into consideration, the psychometric scores do not provide forincreased precision in predicting driving performance.128DISCUSSIONIn the introductory sections of this thesis the structure of the study was outlined interms of two interwoven contrasts. The first involved characterizing the project as havingtwo principal goals or components: a primary aim of evaluating group differences indriving performance, and a secondary aim of exploring the psychometric prediction ofdriving behaviour. The second contrast comprised a distinction between the practicalcontext of the study framed with respect to the information needs of clinicians or licensingauthorities, and the conceptual context that subsumed issues related to the measurementand characterization of driving behaviour itself.The following pages will largely be organized around the discussion of the groupdifferences on the driving behaviour measures. Issues related to the practical andconceptual contexts of the study will be addressed as they arise in the course ofexamining these findings. A separate section will be devoted to discussion of the resultsconcerned with the psychometric prediction component of the study. Finally, twosummary sections will present conclusions and recommendations that arise fromconsideration of the conceptual and practical issues that were the focus of this study.Group Differences in Driving BehaviourFor clarity, the discussion of group differences will be organized into subsectionsaddressing background issues; subject characteristics and descriptive variables; and group129differences on the driving measures at each of the three levels of the hierarchical model.As a starting point it is useful to inquire about the contribution of this study toanswering the question: Are there measurable deficits in the driving performance of mildlydemented elderly?Statistically significant differences were found between the demented sample andboth of the two control groups on all five driving behaviour measures. Figure 1 providesa graphic representation of these differences. While there are some options to beconsidered when evaluating the practical implications of these results, the fact that thesedifferences occur across the range of the driving-related measures is the central finding ofthis study. As was pointed out in the introduction, the current status of research in thisarea is such that it is necessary to establish rudimentary facts, the most obvious being theneed to empirically demonstrate that there are detectable differences in drivingperformance between normal elderly and mildly demented individuals.Group Differences: Background Issues.At this juncture it may be helpful to review information about the current practicesof clinicians with respect to decisions about dementing drivers. The developing body ofresearch in this field, including this study, is largely pragmatic and has been shaped by thepresent circumstances and concerns of practitioners. Clearly, the problem of dementingdrivers did not materialize overnight. However, several demographic and socio-culturaldevelopments have collectively come to bear over the past decade raising the profile ofthis issue to the status of "an important public health problem" (Friedland, 1988, p. 785)18O03• 160)14C0.0. 120rn1Demented^Normal Elderly Mid-Age ControlGroupsNormal Elderly Mid-Age ControlGroupsDementedOOa)0Ca)0Operational Level130 Simulator BrakeReaction Time (sec) Simulator SteeringDeviation (area in mm2)32.5E 2g 1.5U(u•^1CC0.50 Demented^Normal Elderly MKI-Age ControlGroupsDemented^Normal Elderly Mid-Age ControlGroupsManoeuvring LevelDemerit Points onMVB Road TestDemented^Normal Elderly Mid-Age ControlGroupsStopping Distance forHazard Avoidance (meters)Strategical LevelCone Avoidance TaskAccuracy ScoresFigure 1. Performance on measures of driving behavior131and emphasising gaps in the established mechanisms for regulating the licensing of thesedrivers. In the past five years there have been a number of position papers by cliniciansexamining the nature of their responsibilities and contributions to the evaluation of drivingrisk. These discussions frequently conclude by indicating the need for empiricalconfirmation that the driving skills of demented individuals are actually impaired.It must be acknowledged, however, that to some the demonstration of differencesin driving ability between demented subjects and normal elderly controls would appear analmost trivial exercise. Those of us who conduct research in this area are familiar withexpressions of surprise and incredulity when we describe what we do--we are asked "youmean you would actually consider allowing these people to drive?" Also, given that mostdementing disorders are chronic deteriorative diseases that will inevitably cause theeventual' cessation of driving, some may be puzzled by the recent appeals of clinicians forinvestigation of the relation between dementia and driving safety.To make sense of the concerns of clinicians and of the goals of researchers, certainfeatures of the prevailing situation must be appreciated:1. First is the simple fact that a sizable proportion of mildly demented individualsare active drivers for months and often years after the onset of their symptoms. Thus, thebackdrop to research on this problem is the status quo which finds many mildly dementedindividuals on the road. For example, Gilley et al. (1991) surveyed approximately 400drivers evaluated at a dementia clinic. Eighty-two percent of the demented patients (n =330) continued to drive after the onset of symptoms. At the time of the survey 240patients had stopped driving. The group was divided into two sub-samples allowing a132comparison between cases with non-Alzheimer Disease dementias (mostly of vascularetiology) and those with probable or possible Alzheimer Disease (AD). They found that50% of the non-AD cases had stopped driving at 24 months post symptom onset whereasthe 50% mark for driving cessation by the AD patients was not reached until 34 monthsafter onset.2. Gilley et al. (1991) did not indicate why individuals stopped driving, however, iftheir clinic is like many, the majority of those who stopped would have done so becauseof a personal decision (or incapacity) or due to family pressure. At present it appears thatinterventions by clinicians or licensing authorities are the exception not the rule. Coyne etal. (1990) found that 70% of subjects in their retrospective study had stopped drivingwithout the intervention of a clinician. Similarly, Kaszniak et al. (1991) reported thatonly 6 of 39 older patients who suffered from either major depression (n=18) or milddementia (n=21) had been advised to stop driving by a medical professional.In addition, clinicians have noted that if intervention is attempted there is oftenresistance from the patient (Kaszniak et al., 1991, p. 533,) or from family members (Carret al., 1991, Hopewell and van Zomeren, 1990). Lucas-Blaustein et al. (1988) weresurprised to find that 60% of the relatives of demented subjects who were current driversrated the patient as a "safe driver." Presumably a proportion of these families would notbe particulary sympathetic to a practitioner's attempt to revoke the driving privileges ofthe patient.Reading between the lines of recent discussions about older drivers, it appears thatat present it is usually up to dementing drivers and their families to decide when driving133should cease. Despite legal guidelines, the actual role of clinicians in the decision makingprocess is still being defined (Reuben, 1991).3. A third factor that places practitioners in a difficult position, fuelling the callfor empirical research, is the contradictory nature of available recommendations aboutdementing drivers. On the one hand there is the advice of Friedland et al. (1988) whobelieve that driving should be discontinued once a diagnosis of dementia is made and whostate " ... our data show that there is no initial period of disease during which driving issafe" (p. 785). On the other hand, researcher-clinicians such as Hopewell and vanZomeren (1990) state "In most cases of early dementia the patient suffers primarily fromforgetfulness and is usually concerned and perhaps depressed over the beginning illness.In this stage--that is, when insight and self-criticism remain preserved--driving should notas a rule' be prohibited" (p. 322).4. Finally, practitioners are aware that removal of driving privileges may have adevastating impact on the independence and psychological well-being of their patients.Coordinated services to support the independence of elderly who have lost independentmobility are rare. Thus, the decision to recommend withdrawal of a driver's licence mayhasten the dependence of citizens who are otherwise willing and able to care forthemselves. This runs counter to the usual role of clinicians who typically see themselvesas advocates for the optimization of their patients' independent functioning.Considering these four factors, the appeals by clinicians for empiricaldemonstration of the existence of driving problems in the mildly demented do not seemsurprising. Although few would be surprised by results showing that mildly demented134individuals were having more difficulty on the road than were the normal elderly, itnonetheless seems reasonable to solicit proof of these deficits. If changes to the statusquo are to be in the direction of instituting more restrictive policies, those who must makethese difficult decisions will be aided by concrete documentation of the specific nature ofthe performance problems.Group Differences: Subject Characteristics and Descriptive Variables.Returning to the findings of the present study, the group differences on the drivingperformance measures occurred in the context of well-matched subject samples. Inselecting the group of normal elderly controls particular attention was given to assuringthat age, education levels, and gender proportions were comparable between the twogroups of older subjects.Criteria for selection of the demented sample were set to ensure that the subjects inthis group were characteristic of the type of mildly impaired individuals about whomclinicians are especially concerned. The requirement that the cognitively impairedsubjects meet research diagnostic criteria for dementia provides a certain homogeneity inthe sample and communicates useful information about the participants to clinicians andother researchers. The additional use of severity criteria (comprising a range of MiniMental State Exam scores from 20 to 27 points for the demented group) was intended toexclude more severely impaired individuals whose driving deficits could easily beanticipated without elaborate testing. Inclusion of more severely impaired cases wouldhave contributed to more dramatic group differences but would have made it difficult todetermine the extent of impairment in the mildly demented drivers.135Descriptive information derived from the Driving Interview was collected in orderto obtain an overview of the driving patterns of subjects in the study. Driving exposure isan important factor from a methodological standpoint because it is a moderator variable inaccident risk and, as a correlate of driving experience, it can interact with performance-based measures of driving ability (Retchin, Cox, Fox, & Irwin, 1988).Also, exposure data is of particular interest because the presence (or absence) ofvoluntary reductions in driving exposure is of concern from a conceptual standpoint whenattempting to characterize strategical level driving behaviour. In particular, it would beuseful to know whether demented drivers typically have less driving exposure than do thenormal elderly. It seems clear that for exposure reductions to result in meaningful riskcompensation, the reduction must be in some way commensurate with the extent ofdriving problems.The significance of reported reductions in exposure for an individual dementeddriver may need to be placed in a broader normative context. Given that a drop in drivingexposure (relative to mid-age levels) appears typical of normal elderly, it would be usefulto know whether the exposure- declines for demented individuals are of the samemagnitude as the characteristic age-associated reductions, or whether there is a tendencyfor even greater curtailment of driving in this group.The first issue when examining the exposure data available in this study was toestablish whether the results were consistent with past research showing that older peopledrive less than mid-age individuals. Fortunately, a stable pattern of results emerged acrossthe multiple exposure measures and the two raters (i.e., subjects and their collaterals) with136respect to comparing the exposure rates of the older groups to those of the mid-agecontrols. Both of the elderly groups evidenced the characteristic tendencies to drive fewermiles and to engage in more avoidance of demanding driving situations than do the mid-age persons. The exception to this generalization were the results from the question aboutthe number of trips per week for which no significant differences were found betweengroups. But, as discussed in the results section, the data from this measure were flawedin several ways.In addition to differences between the older groups and the mid-age group, theexposure data were also examined for consistent differences between the two groups ofolder subjects. The evidence regarding exposure differences between the normal elderlyand the demented groups was somewhat contradictory. Results show that, on the onehand, the demented drove fewer miles in the previous 12 months, but at the same timethey did not differ with respect to the frequency of driving after dark or in rush hour, norin their tendency to avoid demanding driving situations. In overview it appears that thedemented subjects experience about the same rate of exposure to higher-risk drivingsituation as do the normal elderly, although confidence in this generalization is somewhatcompromised by discrepancies between the ratings of the demented subjects and theratings of their collaterals. There was a significant difference between the dementedsubjects and their collaterals in estimates of annual milage, and a similar tendency on theother exposure measures for the collaterals to indicate less exposure than was claimed bythe demented.Although it is not possible to determine which ratings are more veridical, the most137obvious alternative explanations for the mismatch between the demented subjects and theircollateral have interesting implications. First, if the demented subjects' self-ratings areinaccurate, it appears that they err in an interesting way: the self-reports of the dementedsubjects are surprisingly similar to the self-reports of the normal elderly. This raises thepossibility that despite objective changes in behaviour, the demented view themselves asfunctioning as they had before the onset of their symptoms.On the other hand, if the collaterals' ratings were found to be inaccurate, it wouldsuggest that over-reliance by clinicians on caregiver reports of exposure might be ill-advised. Given the current clinical practice of obtaining exposure information (that mayhave a significant bearing on licensing recommendations) from collaterals, this is not atrivial issue.The results from the measures of driving exposure suggest some areas of practicalimport for clinicians. If estimates of exposure play an important role in recommendationsabout driving privileges the following factors should be considered. Because patient andcollateral reports of driving exposure may differ it would be useful to press for objectivemilage estimates such as might be available from auto-servicing records or from acaregiver who could unobtrusively record the amount and type of exposure over adesignated period of time. When collaterals' reports are solicited it would be useful todocument the nature of their familiarity with the subject's driving behaviour. If thesubject drives alone most of the time the collateral may not be able to provide accurateinformation. Also, factors other than familiarity that might contribute to biased reportingshould be evaluated. It would be wise to consider the possibility that the patient and also138the collateral may have a tacit agenda with respect to whether the patient should continueto drive. Collaterals may wish to protect the patients' driving privileges or alternativelythey may hope someone will take immediate action and revoke the patient's licence.These agendas may influence the accuracy of reported exposure. Finally, even if thepatient is reported to have reduced his or her driving exposure, it is important toremember that most normal elderly also reduce their exposure. In weighing informationabout reductions in exposure, emphasis should be on determining whether the reductionsare commensurate with the level of the individual's impairments.Group Differences: Driving Behaviour Measures To date the hierarchical model has principally been a conceptual tool, not ameasurement paradigm. The present study represents an exploration of thismultidimensional approach in an applied context. Although unwieldy, the simultaneousmeasurement of several aspects of driving behaviour across the same set of subjects ishelpful both in characterizing subjects' driving abilities and in increasing ourunderstanding of the driving task itself.Because of restricted sample sizes, it was not feasible to interpret theintercorrelations among the driving measures although that is an issue of obvious interest.It is hoped that the larger sample sizes that will be available when the Driving and AgingStudy has completed data collection will make these analyses possible. Although thecorrelational avenue was not available in the present study, by examining specific taskdemands and the group differences on measures at each of the three levels, it is possibleto explore some of the strengths and weaknesses of the present instantiation of the139hierarchical model.In offering the following critique of the driving performance measures and thehierarchical model I am aware that more weaknesses than strengths are discussed.Although shortcomings are highlighted, this does not imply dismissal of the driving tasksnor is it an indication that I consider the study to have been a failure. The drivingmeasures out-performed my expectations and the consistency within the data set is asmuch as I could have hoped for. As pointed out previously, the primary aim of this studywas to determine if there were detectable group differences in driving behaviour betweenmildly demented and normal elderly subjects. Not only was this primary objectiverealized but several auxiliary concerns were also addressed. Given the paucity of researchin this area, the present study was by defmition exploratory and was not expected to resultin definitive solutions to the many problems that must be grappled with. What follows isoffered as a contribution to a fledgling body of knowledge in which fundamental issuesrelated to construct defmition and measurement methodology are still at the forefront andremain largely unresolved.Operational level. The Computerized Driving Assessment Module (CDAM)indices of brake reaction time and steering accuracy were included on the assumption thatsimulator technology would be useful for standardized assessment of basic vehicle-controlskills. Compared to tasks at other levels of the hierarchical model, the operational levelmeasures provided the greatest discrimination between the demented and the normalelderly. Also, there was a significant age effect on the steering accuracy task. Althoughthese results appear to be consistent with predictions that would follow from accounts of140the hierarchical model in head-injured and elderly drivers (eg., Brouwer et al., 1988;Hopewell & van Zomeren, 1990; van Zomeren et al., 1987) two issues--one of practicaland the other of conceptual import--deserve mentioning.From a practical standpoint, a comparison of the score distributions for theoperational level measures with those of the more realistic manoeuvring level taskssuggest that attempts to generalize from CDAM performance to on-road behaviour may beill fated. Driving simulators undoubtedly have clear advantages over on-road tests withrespect to task standardization and precision of measurement. However, in terms of theeventual development of measurement technology for screening older drivers, there is anobvious concern about whether simulators will function as valid indices of the actualdriving abilities of demented subjects. In practical settings the high sensitivity of thesimulator may be to its detriment. If simulators are to be used in a practical contextassociated with licensing decisions, it must be demonstrated (not assumed) that they are,at the very least, predictors of on-road behaviour and better that they are associated withan individual's ability to drive safely.From a conceptual standpoint it is important to probe the task demands of thissimulation and ask, with hindsight, what is CDAM actually measuring in these subjectgroups? One possibility is that for the demented subjects simulator performance is asmuch a function of their deficits in language comprehension and new learning ability as itis an indicator of decrements in basic vehicle-control skills.Our observations of the behaviour of the demented subjects during CDAM testingsuggest that contrary to expectations, the driving simulator provided anything but an141assessment of basic overlearned vehicle-handling skills. Several of the demented subjectshad problems remembering and comprehending the task instructions, and despite thepresence of automobile controls, the cognitively impaired subjects often did not respondas one would in a vehicle. Examiners reported that demented subjects would often lookdown at the brake pedal before responding, or would respond verbally by reading the stopsign rather than depressing the brake pedal. In addition to the novelty of the simulatortasks, both simulator measures (but especially the steering accuracy task) were multi-trialdual-tasks that involved regulating the accelerator pedal to maintain a constant reading onthe speedometer while tracking a series of lights or monitoring the video screen for a stopsign.Dual-task decrements are ubiquitous in the normal-aging literature. Indeed, giventhe respcinse demands of the steering accuracy task it would have been surprising if thenormal elderly had not shown a deficit relative to the mid-age subjects. It is also easy toimagine how difficulties with dual-tasks could be intensified in dementia. It seems likelythat despite the attempt to reproduce an automobile-like-environment, for the dementedsubjects the simulator did not feel like a car nor did the tasks feel like actual driving.Thus, the simulator stimuli were not likely to be eliciting the automatized perceptual-motor driving skills of the demented subjects, but rather functioned like novel laboratorytests of complex information processing skills. This may be less true for the normalcontrol groups although this would require a more through exploration than is possiblehere.Manoeuvring level. The manoeuvring level tasks are the closest to "real" driving142behaviour of all the performance measures included in the study. Also, interestingly,these measures were the tests on which there was the greatest overlap between the scoresof the demented and those of the normal elderly. This finding has important practicalimplications. Put simply, although the manoeuvring level performance of the dementingsubjects was indicative of worrisome difficulties while driving, it was also the case that,from a normative perspective, the majority of the demented subjects were not dramaticallyimpaired. For example, on the MVB Road Test the scores for nearly half the mildlydemented lie beyond the 80th percentile (31 demerit points) of the distribution of scoresfor the normal elderly, however 13 of the 18 demented subjects passed the road test giventhe MVB cutoff score of 40 demerit points. Given that the MVB test is the currentstandard by which licensing decisions are made, by that definition most of the dementedsubjects were able to demonstrate what is deemed by society to be an adequate level ofdriving skill. Similarly, although there was a significant difference between the normalelderly and the demented group for emergency stopping distance, nearly 70% of thedemented subjects had stopping distances that fell within the average range (ie., scoresless than 1 SD above the mean for the normal elderly).For a behaviour like driving skill, which is characterized by significant individualvariation in normal populations, the range of abilities that can be normatively designatedas significantly impaired  must be limited to the more extreme performances that are trulyuncharacteristic of the normal population. Failure to do so would result in sizableproportions of a normal population being arbitrarily designated as "impaired", which isobviously at odds with a normative approach, and is also contrary to societal priorities143that include the maximization of independent mobility for older persons. If a normativecriteria for impairment on the manoeuvring level tasks is set such that scores must liebeyond 2 SDs from the mean for the normal elderly, then approximately 30% of thedemented sample qualify for a label of significant impairment. Some of the implicationsthese results might have for clinicians will be taken up in my concluding remarks.Strategical level. One of the strengths of the hierarchical model is that itspecifically addresses the higher cognitive dimensions of driving that have not found theirway into the working language of the psychomotor skills approaches. From the standpointof furthering our understanding of the phenomenology of driving behaviour in populationswith deficits due to injury, disease, or age, it is especially advantageous to have aframework that helps place different sorts of deficits in perspective. For instance, fromthe hierarchical model we can immediately see why a conscientious elderly driver withsensory losses and slowed reaction times would be a better driving risk than a physicallyfit 30 year old who, due to a previous head injury, is impulsive but still performs in theaverage range on psychometric tests. While common sense suggests this without the needof a formal model, we can become prisoners to our assessment tools. If these twohypothetical drivers were to be assessed from a psychomotor information processingapproach, the elderly driver would appear more disabled. Van Zomeren and colleagues(Brouwer, et al., 1988; Hopewell & van Zomeren, 1990; van Wolffelaar et al., 1988; vanZomeren et al., 1987, 1988,) made a clear contribution when they advanced the concept ofstrategical level behaviour in the context of head-injured drivers. In so doing theyprovided a handle for conceptualizing clinically meaningful aspects of driving-related144judgement skills that previously had been too abstract to be operationalized.Rather than having the concept of "judgement" stand as a global intuitiveconstruct, by using the notion that drivers may differ in the quality of their strategicallevel problem-solving skills, we are able to be more concrete about what we might meanby the terms "good" or "poor" judgement as applied to driving behaviour. Some examplesof partially distinguishable aspects of driving-related judgemental abilities would include:the ability to consider risk factors associated with specific road-way or vehiclecharacteristics, insight into personal deficiencies that might impact in a permanent, or atransitory way, on driving skill, and the ability and motivation to make and follow throughon decisions to curtail exposure, even when "inconvenient." Obviously the need to makejudgements of this sort is not just the purview of the compromised driver. All of usconfront' situations that require the exercise of driving-related problem solving. Driversmay face a situation where they must decide whether or not to drive when they suspecttheir vehicle may not be in top running condition. Individuals on long journeys mustevaluate whether they are alert enough to continue driving hour after hour. Someotherwise responsible people may have difficulty resisting the impulse to drive the "fewmiles" to get home even when they have had too much to drink. Thus, we must all makestrategical decisions; however, these decisions take on special significance in those whosesensory, motor, or cognitive functioning is marginal.The aspects of strategical level behaviours that were explored in this studyinvolved: (a) evaluating the accuracy of self-appraisals with respect to driving abilities,and (b) attempting to determine whether, in light of increased levels of driving problems145relative to the normal elderly, the demented group also engaged in "extra" exposurereduction.When we examine the accuracy of the demented subjects' self-appraisals on theCone Avoidance task the results are quite striking. First, in an overall sense the dementedwere less accurate than control subjects when predicting how many cones they would hitwhile manoeuvring through the relatively narrow lane of traffic pylons. While this isinteresting in its own right, the more dramatic finding is that the mismatches betweenpredicted hits and actual hits were almost always in the direction of under-predicting theactual number of hits. Across trials two and three, 14 of the 18 demented subjects under-predicted their cone hits compared to only 4 subjects with under-predictions in each of thecontrol groups.In describing the relation between predicted cone hits and actual cone hits, I havestated that the demented subjects were "over-confident." This is accurate in as much as itdescribes the fact that this group consistently predicted fewer cone hits than actuallyoccurred. However, these over-confident ratings do not necessarily mean that thesubjective experience of the demented subjects was one of over-confidence. However, ifother reasonable alternatives could be ruled out, the implication of these findings wouldbe a matter for considerable concern. That is, we would be forced to conclude that thesedrivers were over-confident to the extent that even when faced with clear objectiveevidence that things were not as they predicted, they nonetheless were unable to adjusttheir erroneous self-appraisals. Although, the data from the present study do not allow usto rule out all the various alternatives to an over-confidence interpretation, it is worth146examining some of the possible explanations for the data.The demented subjects are by definition memory-impaired, and this could, inprinciple, compromise their ability to take past performance into consideration. In fact,memory problems are an unlikely candidate as the cause of the mismatch betweenpredicted hits and actual hits. The subjects in this study were not so profoundly amnesicthat they were unable to maintain an instructional set from minute to minute; moreover,the memory load for the prediction component of the Cone Avoidance task was low. Ontrials two and three, subjects' ratings were elicited as they watched the experimenter tapeach cone that they had hit on the previous trial. At the same time the driving tester wasin the vehicle with the subject informing them of the number of cones they had hit on theprevious trial and then asking them to make their predictions for the next trial. Also,requests 'for reminders of previous predictions or actual hits were always honoured. Whatis more, if memory deficits were the primary cause of the mismatch between predictedand actual hits, one would not expect such a consistent pattern of under-prediction by thedemented subjects.More difficult to disentangle are the kinds of response biases, other than one of an"over-confidence bias," that may have affected the subjects' ratings. The most seriouschallenge to an over-confidence interpretation of the present data is the possibility that theprediction component of the task was simply too abstract or difficult for the dementedsubjects. If this were the case then the consistency in demented subjects' predictionsmight be explained in terms of something we could call a "simplicity bias." Conceivably,the demented subjects might have found it easier to make simple predictions of 1 or 0147cone hits rather than try to weigh the information about past trials.Another possibility competing with a straightforward over-confidence interpretationwould be if the demented subjects were responding in a way that they believed madethem appear confident and was consistent with socially desirable behaviour. They mighthave felt that denying problems was a better tactic than admitting them.In trying to choose among the different interpretations of these results, some of theexperimental procedures described in the McGlynn and Kaszniak (1991a, 1991b) paperswould be welcome. In particular, it would have been useful to attempt to determinewhether the effects observed on the Cone Avoidance task were simply the result of ageneral deficit in the ability to make predictions or ratings. McGlynn & Kaszniak(1991b) ruled out what they termed a "general estimation deficit" (p. 187) in their subjectsby showing that although the AD patients were inaccurate when rating their own abilities,they were able to generate rating of their relatives that closely matched the self-reports ofthese relatives.For the Cone Avoidance task, introducing a check such as that used by McGlynn& Kaszniak would seem to require asking subjects to make predictions of cone hits whilethey observed the performance of a significant-other doing the Cone Avoidance task.Although such a task would be fascinating, neither the results obtained from it nor theresults of McGlynn and Kaszniak actually entirely rule out either a simplicity bias or asocial-desirability bias as long as the demented subjects are rating normal relatives whotypically function well. In order to positively attribute the prediction inaccuracies ofdemented subjects to a failure of awareness or to over-confidence, one would have to148determine that demented subjects were able--and willing--to perceive deficits in others butnot themselves.Thus, although the fmdings for the Cone Avoidance task are consistent in showingthat the demented persistently under-estimated the difficulty of the task, it is not possibleto be certain that these inaccurate predictions are not the result of either choosing thesimplest prediction options or trying to make a good impression on the experimenters.However, the two additional indices of strategical level behaviour--derived from theInterview ratings of driving problems and from the examination of compensatoryavoidance--lend weight to the impression that the over-confidence seen on the ConeAvoidance task is not simply an artifact of the task's response requirements.The outcome of the analyses of the driving problem questions from the DrivingInterview are consistent with the view that the demented exhibit a characteristic tendencyto underestimate their driving disability. On the Interview questions the dementedsubjects rated themselves as having significantly fewer problems than the ratings of theircollaterals would suggest.The second aspect of strategical behaviour that was of interest is associated withcompensatory risk reduction. When the information about driving exposure is examinedtogether with the data for driving performance and rating of driving problems, one is leadto question whether the demented subjects respond to the fact of their deteriorated drivingskills in a "normal" or insightful manner. The question is whether, given declines indriving ability, the demented typically "compensate" by reducing their driving exposure byan amount commensurate with their driving deficits. An auxiliary question is whether149there is any evidence that the normal elderly group engage in behaviour suggestive ofcompensatory risk reduction.First, on the measures of driving ability, there is clear evidence that on all tasksthe demented were having more driving difficulties than were their normal elderlycounterparts. Also, there were indications that the normal elderly had mild deficits indriving skill. Relative to the mid-age group, the normal elderly did significantly less wellon the driving simulator measure of steering accuracy and on the MVB road test. Whilethere is no direct proof that these deficits are the result of age-related declines in drivingability, it would run counter to much of the literature were there to be no age effects onany of the driving tasks.The next group of measures that play a role in assessing risk compensation are thedriving exposure indices. Comparing the two older groups to the mid-age group, therewas clear evidence of reduced exposure for both groups of older subjects. As discussedpreviously, when the exposure ratings for the demented and normal elderly groups werecompared there were some inconsistencies between the different indices. On the one hand,the demented and normal elderly did not differ with respect to the frequency of night timeor rush hour driving nor in the extent of their avoidance of demanding driving. But, itappears that the demented group may have driven fewer miles than did the normal elderlyin the 12 months preceding the study. Although the ambiguity cannot be erased, theremay be meaningful differences in the implications of exposure as indexed by a "simple"milage estimate, versus an index that conveys information about the propensity of subjectsto drive in demanding situations. When considering compensatory risk reduction the150emphasis is on drivers' selective exposure to demanding (i.e., higher risk) drivingsituations, not on their total milage. So, although it would be incorrect to statecategorically that the demented group did not differ from the normal elderly group withrespect to driving exposure, it does seem that the older groups do not differ significantlyin their exposure to (or avoidance of) the types of demanding driving situations that aremost important from the perspective of risk compensation.To return to the questions posed a few paragraphs ago, we are interested inexploring whether there is any evidence for risk compensation by either group of oldersubjects. It is perhaps easiest to start with the normal elderly group. We saw that therewere significant differences between the mid-age and the normal elderly on two drivingtasks, providing evidence of mild deficits in driving ability for the older controls. Wealso saw that compared to the mid-age group, the normal elderly drove less frequently atnight and in rush hour, engaged in more avoidance of demanding driving situations, anddrove fewer miles in the 12 months prior to taking part in the study. Certainly all the keycomponents of risk compensation are present here.Having established at least tentative evidence for risk compensation in the normalelderly, attention can be turned to the dementing drivers. Relative to the normal elderly,the demented have more driving problems and similar amounts of exposure to demandingdriving situations. Thus, while there is evidence that the demented have less exposurethan do younger drivers, the magnitude of reductions in driving exposure relative to theextent of driving deficits is not proportional to that seen in the normal elderly. Thus, theresults suggest that while the demented do avoid more than younger drivers, the decreased151exposure is not commensurate with the increased driving problems they experience.Given the limitations of the information available in this study it is not possible tomeasure the real-world impact of the exposure reductions that occur in the older groups.Without external evidence, such as documentation of a decreased accident rate for thosewho engage in higher levels of selective exposure reduction, we cannot say with certaintythat the driving reductions of normal elderly are an effective risk compensation strategy.In the absence of an empirically derived ratio that specifies the optimal relation betweenproblem level and amount of avoidance, it is difficult to say how much avoidance is"enough." Nonetheless, there is mounting evidence for what appears to be anapproximately two-fold (Cooper, et al., in press) increase in accident occurrence fordementing drivers relative to the overall elderly population. Thus, we have reason tobelieve that even if the demented do curtail their driving exposure, the reduction is notsufficient to offset their increased disability.Considering all the results of the strategical level measures together, we need to beconcerned about the abilities of mildly demented drivers to make appropriate driving-related judgements. Subjects in the demented group were poor judges of their owndriving ability on the Cone Avoidance task, they under-rated their driving problemsrelative to ratings by their collaterals, and they did not appear to significantly curtail theirexposure beyond what one would expect for normal elderly drivers.These results are interesting in the context of comments by Hopewell and vanZomeren (1990) presented earlier in this chapter. They suggested that insight and self-criticism usually remained preserved in early dementia, and thus for very mildly impaired152people there were no clear counter-indications to driving. The findings of this study showthat subjects with symptomatology that would be consensually agreed upon as mildalready appear to have deficits in insight and self-criticism. Clinicians are aware of thesephenomena in more advanced dementia, but it is interesting to have an empiricaldemonstration of this problem in the context of a behaviour like driving where theconsequences of impaired judgement could be very damaging to both the patient and toinnocent bystanders.Psychometric Prediction of Driving BehaviourThe results from this component of the study have an interesting twist to them. Aswas shown in some detail, once group differences are accounted for, the psychometrictests do not add any extra precision to our ability to predict the driving performance ofdemented subjects. The twist is that the psychometric tests used in this study are close"cousins" to the measures in the diagnostic test battery. Most of the tests in thepsychometric battery were selected because they are familiar assessment instruments inclinical settings.Had the "driving-related" psychometric test battery been used for the purpose ofclassifying the older subjects into cognitively-impaired versus cognitively-intact groups(based on norms in the neuropsychological literature), the new groups would probablyresemble those defined by the Alzheimer Clinic diagnostic battery. Also, if these newgroups were compared with respect to their driving performance, it is likely that the153between group differences would be similar to the differences already observed betweenthe demented and normal elderly groups.Thus, in summary, it is not the case that the psychometric tests of perception,attention, judgement, and psychomotor skills are unrelated to the driving measures.Rather, because they tap many of the same deficits that are assessed by tests in thediagnostic battery, they do not add anything extra. The message of the correlationalanalyses is that if we wish to add precision to our prediction of the driving skill of thosewith documented cognitive impairment, we would be wise to pursue measures that bearlittle resemblance to the tests in the neuropsychological assessment battery. This will bea considerable challenge given the wide range of functions that are considered whenmaking a diagnosis of dementia.The Conceptual Context: How Fares the Hierarchical Model? Although the hierarchical model is a loose conceptual framework that was notoriginally proposed as a measurement model, it has served this study well especially inclarifying the different dimensions of driving behaviour that might be affected by age anddementia. What the hierarchical model lacks in specificity, it makes up for in breadth ofperspective. From the lead given by Michon and clinical researchers working with brain-damaged drivers, one is able to generate a set of measures that yield a broad profile ofdriving behaviour. The stepwise discriminant analysis showed that the optimalclassification of subjects was obtained with a weighted composite of three measures--one154from each of the levels of the model.However, the hierarchical model provides less help in analyzing specificcomponents within the complex and integrated sequences of actions that are driving. Thetask descriptions that arise from a psychomotor or information processing approach todriving are potentially very precise. In comparison, the analysis that could currently begenerated from the hierarchical model might seem superficial. Although the hierarchicalapproach and the psychomotor information processing approach need not be seen ascompeting models, the differences in emphasis and focus do lead to quite differentconclusions. This has practical implications for clinicians who must weigh various kindsof information when evaluating an individual's fitness to drive. The "hierarchically-minded" assessor might weigh the data from simulator assessments or psychometricassessments differently than would someone who viewed driving as primarily a perceptualand psychomotor skill.Unfortunately, the two areas that are major strengths of the hierarchical model arealso those in which it has the least to offer--when our goal is to assess driving ability.The model re-directs our attention from perceptual-motor skills to judgement abilities andto manoeuvring level behaviour in an on-road environment. However, once re-directedwe are on our own when it comes to task selection and developing criteria for adequateperformance. Because it is usually for these things that we turn to conceptual models inthe first place, it is clear that for now the hierarchical model falls short as a completedescription of driving behaviour that might function as the underpinning for asophisticated approach to measurement and analysis of such behaviour.155Although the tasks used in this study were loosely slotted into the hierarchicalframework, the goals of the study did not include conducting a rigorous hierarchical taskanalysis of driving behaviour. Instead, the selection of individual measures was based onthe study's practical goals--to contribute information that would be helpful to those whomust make decisions about the driving privileges of dementing individuals. Also, becausethe model does not provide criterion by which competency can be defined, this studyrelied on a normative approach involving the comparison of demented drivers to theircognitively intact age-mates. Future driving research would benefit from a more formaltask analysis of the hierarchical levels.As a starting point for a formal theory of driving actions the hierarchical model isvaluable for its wide scope, attention to driving-related problem solving skills, andbecause, under the rubric of the manoeuvring level, it incorporates in-traffic interactionalbehaviour. Because the manoeuvring level is where actual driving takes place, it can beconceptualized as the final common pathway through which drivers' operational levelskills and their strategical deliberations become manifest. Given that problems inoperational level skills may have different ramifications than would deficits in strategicallevel judgements, it would be useful to focus on disentangling strategical and operationalbehaviour as they occur in actual traffic situations. It should be possible to identifyeveryday driving situations that call on definable aspects of strategical or operational levelabilities. One could then work out a scoring system for characterizing individuals'operational and strategic behaviours in everyday driving situations. This might move uscloser to a conceptually-based measure of driving competency and also would provide a156scoring system that is more than a simple checklist of discrete driving actions divorcedfrom their context and their implications.For the group of mildly demented drivers in the present study, the outcome ofperformance-based measurement of driving behaviour yielded indeterminate results. Therange of scores on the manoeuvring level tests indicated that most of these drivers wereneither "normal" nor were they grossly impaired when compared to the elderly controlgroup. We clearly need to be concerned about marginal drivers even though they "pass"road tests. This poses difficulties for the development of an assessment strategy intendedto entirely replace clinical judgement with objective performance-based measures ofdriving skill. However, even if clinical judgement remains an important component in theassessment of driving fitness, there is still a need for the development of betterperformance-based evaluations of driving ability. We need only look at the automobiledependence of the large and aging baby-boom generation to assure ourselves that theproblem of evaluating the driving skills of older drivers is not going to go away. Themethodological issues involved in the design and validation of better measures of drivingskill are many and difficult. Nonetheless, there is no doubt that this work should be done.The hierarchical model is not sufficient on its own to support such a program, but it doeshave a contribution to make.The Practical Context: Where to From Here? We have seen that the demented do less well than the normal elderly on all the157driving measures. However, on the most "realistic" measures, the differences betweengroups are less pronounced than on the highly sensitive driving simulator tests. Withrespect to in-office prediction of driving ability, the results of the present study suggestthat once cognitive impairment has been reliably documented (i.e., by neuropsychologicalassessment) further psychometric testing is not likely to add precision to predictions aboutdriving ability.In making my concluding remarks I wish to return to the MVB Road Test results,which I believe illustrate an important point. As noted earlier, we found that while almost70% of the mildly demented drivers were able to pass the MVB test, many madesignificantly more errors than did subjects in the other groups. I believe that this is anaccurate indication of what the driving abilities of mildly demented drivers are. Most aremarginal drivers.The demented drivers in this study were primarily individuals contacted during theprocess of diagnostic assessment at the Alzheimer Clinic. They are likely to be typical ofindividuals seen at other multidisciplinary geriatric assessment centres. As a group, thesemildly demented individuals can be expected to differ from normal elderly in almost anyaspect of driving skill one might measure. Also, they can be expected to deteriorate overtime. But, when diagnostic teams encounter these drivers at their first presentation at aspecialty clinic, I would predict that well over half will be "getting by" in everydaydriving situations and would be able to pass an MVB licensing road test.If this prediction is correct, clinicians and policy makers must grapple with acomplex situation. It appears that there are still-to-be-resolved discrepancies between the158ethical /legal, clinical, and pragmatic facets of the dementing driver issue. While it isunlikely that major changes in licensing policy can, or even should, be instituted in thenext few years, recent position papers on the topic suggest that there are some who wouldlike to see significant and immediate changes to the status quo (e.g., Reuben, 1991).Given the complexity of the issues it seems possible that hasty changes, made in advanceof larger scale studies, may result in even less effective decision making procedures formaking decisions about demented drivers than are currently in place.Recently two articles (Carr et al. 1991, and Reuben, 1991) representing verydifferent perspectives appeared side by side in the same issue of the Journal of the American Geriatric Society (Vol. 39, November, 1991). In light of the findings of thisstudy I am moved to make an editorial comment on the implications of the twoapproaches.Reuben (1991) argues that all jurisdictions should institute a mandatory reportinglaw such as the one in California that requires physicians to report all individualsdiagnosed with dementia to the local health department who then notify the state MotorVehicle Department. With this system, decisions about driving privileges are solely in thehands of the licensing authorities. In Reuben's view all parties are served better when thephysician's role in assessing fitness to drive is confined to that of a diagnostician legallyrequired to report those diagnosed with dementia. He states (p. 1138):Given the probable link between dementia and motor vehiclecrashes and the tenuous link between current clinicalassessment of dementia and crashes, we may best serve bothour patients and society by establishing a diagnosis ofdementia in patients who meet accepted clinical criteria. Inso doing, we have performed a screening function that159identifies patients at higher risk for having crashes. Once thisinformation is gathered, it is then the state's responsibility todetermine whether the risk is high enough to warrantrevoking the driving privilege.Although there are points in favour of Reuben's approach there are seriousshortcomings that he has not addressed. First are the problems inherent in a processwhereby the clinical diagnosis of dementia acts as a sort of cutting score that thenprecipitates a mandatory report. It is important to note that a mandatory reporting lawwhich specifies the reporting of a diagnostic entity like dementia is very different fromone that requires reporting those suspected to be unfit to drive. Reuben does not indicatewhether he would advocate mandatory reporting of all conditions that might negativelyimpact on driving. But clearly the reality of the relation between driving impairment,medical disabilities, aging, and cognitive functioning are not so black-and-white that wecan ignore the possibility of driving problems in individuals with other medical conditionsor in those who fail to meet diagnostic criteria for dementia.Also, it is important to note that unlike Friedland et al. (1988), Reuben (1991) doesnot appear to be suggesting that individuals diagnosed with dementia should categoricallyhave their licences revoked. He states: "The two critical components--mandatoryreporting and a fair process of evaluation--must both be present if we are to provide thebest possible outcome for our demented patients and society " (p. 1138). It would appearthat Reuben believes that the procedures of licensing authorities will somehow solve theproblem of dementing elderly drivers. I would argue, in light of the present study results,that this is not the case. We would expect a sizable proportion, even a majority of mildlydemented individuals, to pass a licensing test but continue to be marginal drivers in need160of regular evaluation. Unfortunately, Reuben's recommendations pass the buck tolicensing authorities who are less well-equipped to manage marginally impaired driversthan are clinicians.We are still a long way from the development of a "gold-standard" index ofindividuals' abilities to drive safely. In fact, I doubt we will ever have such a thing.Therefore, I believe we must realize that the solution to the problem of mildly dementeddrivers cannot rest simply on the use of road tests. Although they are a reasonable part ofthe assessment process, the fact that many mildly demented drivers do not fail is noassurance that their driving problems can be ignored. Current road tests function asscreening measures for gross incompetence--not as measures of the ability to drive safely.Having said this, it would seem we are led to choose between two very difficultoptions. The first is to take a categorical stand and simply revoke driving privileges whena diagnosis is made (e.g., Friedland et al., 1988). Future research may show that this isthe most reasonable approach despite its many limitations. Longitudinal studies are verymuch needed to chart the course of deterioration in the driving skills of dementingindividuals. If it turns out that their risks of mishap spiral sharply and quickly upwards, itwould be misguided to attempt to maximize the period of time they are allowed to stay onthe road. Until such research is available I would opt for the "management" approach thatis thoroughly described in the article by Carr et al. (1991).Carr et al. (1991) describe a model for a multidisciplinary approach to theevaluation and management of dementing drivers that is well balanced, respecting theneeds of the patient while also attempting to fulfil the responsibilities of clinicians. They161make comprehensive recommendations about the kinds of information that should begathered, various professionals that might be called in, and the , sequencing of assessmentprocedures and interventions. They advocate regular systematic assessment andindividualized recommendations that incorporate a wide range of information includingexposure characteristics, level of cognitive impairment, assessment of the impact ofmedications and concurrent medical disorders, caregiver reports and other factors. Theirapproach covers nearly all the issues one could come up with in terms of assessing andmonitoring marginally impaired drivers. The program resembles those of rehabilitationcentres which have developed considerable expertise in the management of driving forbrain injured individuals.However, two problems with the Carr et al. (1991) approach should be mentioned.First, it appears to be expensive and time-consuming. The approach is premised on accessto the resources of a multidisciplinary team and would not be feasible for busy physiciansin private practice. Second, and more serious, it is not clear how this team would knowwhat to monitor nor precisely how to weigh the different types of information available tothem. While this shortcoming is not unique to their process, for physicians like Reuben itmakes the endeavour suspect from ethical and legal perspectives.In closing I would like to add an observation and a recommendation for a ratherlow-tech addition to the procedures discussed by Carr et al. Based on the tone of thediscussions presented in the position papers, and on my own experience with a numberdementing drivers, I have the distinct impression that it is rare for anyone, even a familymember, to discuss the issue of driving in much detail with the afflicted person. Informal162discussions with physicians have reinforced my belief that many find this to be anuncomfortable topic. Without wishing to appear flip, I would venture that talk ofcurtailing driving privileges ranks along with sex and death as a topic that many cliniciansdread having to bring up.One of the goals of the diagnostic process should be to develop a plan formonitoring and managing driving problems. I think this should be done, if at all possible,in the form of a set of negotiations with the patient and their family members. Ideally,the patient "negotiates" to stop driving on their own before the clinician or licensingauthorities are forced to remove their driving privileges. I believe the likelihood of apersonal decision to stop driving would be greatly increased in marginal drivers if thetopic of driving privileges and problems were more openly and frequently discussed byclinicianS. Without doubt there will be a proportion of people who will not voluntarilystop driving. The present study demonstrated that many demented drivers underestimatetheir driving impairment. Nonetheless, I am convinced that many drivers would give updriving sooner and with less upset if there were a programmatic way to keep the questionof driving competence in the foreground. In a sense the process is one of educating thedementing drivers and their family members about the possibility of driving problems.The process is also one of overcoming complacency and denial-- many people don't wantto believe that things are not as they once were and that they are not able to do what oncecame easily. Although the fact is that loss of driving privileges may be very painful forpatients, even if they have been involved in a negotiation process, current wisdom aboutother kinds of losses suggests that people cope better if they are prepared and have been163able to "work through" some of their feelings before the event occurs.Regular monitoring of dementing drivers by multidisciplinary teams may be areasonable approach to managing the problems associated with marginal drivers who areable to pass standard licensing exams. Compared to denying driving privileges to all whoqualify for a diagnosis of dementia, a monitoring process certainly has the advantage ofproviding a mechanism to allow these individuals to retain their driving privileges for aslong as possible. Rigorous evaluations of the safety records of drivers who have beenmanaged by teams such as the one described by Carr et al. (1991) are essential. If themanagement approach is successful, the mishap rates for these patients should, at the veryleast, not exceed those of other groups of medically impaired older drivers. Largesamples and longitudinal research will be needed to resolve this issue.164REFERENCESAlbert, M.S. (1981). Geriatric neuropsychology. Journal of Consulting and ClinicalPsychology, 49 835-850.Anthony, J.C., Le Resche, L., Unaiza, N., Von Koroff, M.R. & Folstein, M.R. (1982).Limits of the "Mini-Mental State" as a screening test for dementia and deliriumamong hospital patients. Psychological Medicine, 12 397-408.Ball, K. & Owsley, C. (1991). Identifying correlates of accident involvement for theolder driver. Human Factors, 33, 583-595.Barakat, S.J. & Mulinazzi, T.E. (1987). Elderly Drivers: Problems and needs forresearch. Transportation Quarterly,  41 198-206.Barrett, G.V., Mihal, W.L., Panek, P.E, Sterns, H.L. & Alexander, R.A. (1977).Information processing skills predictive of accident involvement for younger andolder commercial drivers. Industrial Gerontology, 4 173-182.Bayles, K.A. & Kaszniak, A.W. (1987). Communication and cognition in normal agingand dementia. Boston: College Hill/Little Brown.Berg, L., Danziger, W.L., Storandt, M., et al (1984). Predictive features in mild seniledementia of the Alzheimer type. Neurology, 34, 563-569.Blaaw, G.J. (1982). Driving experience and task demands in simulator and instrumentalcar: A validation study. Human Factors, 24 473-486.Boll, T.J. (1981). The Halstead-Rieten neuropsychological battery. In S. Filskov & T.J.Boll (Eds.), Handbook of clinical neuropsychology (pp. 577-607). New York:John Wiley and Sons.Botwinick, J., Storandt, M. & Berg, L. (1986). A longitudinal behavioral study of seniledementia of the Alzheimer type. Archives of Neurology, 43, 1124-1127.Botwinick, J., Storandt, M., Berg, L. & Borland, S. (1988). Senile dementia of theAlzheimer type: Subject attrition and testability in research. Archives ofNeurology, 45 493-496.165Branconnier, R.J. (1986). A computerized battery for behavioral assessment inAlzheimer's Disease. In L.W. Poon (Ed.), Handbook for clinical memoryassessment of older adults (pp. 189-196). Washington, DC: AmericanPsychological Association.Brouwer, W.H., Rothengatter, J.A. & van Wolffelaar, P.C. (1988). Compensatorypotential in elderly drivers. In J.A. Rothengatter & R. de Bruin (Eds.), Road userbehaviour: Theory and research  (pp.296-301). Assen: Van Gorcum.Buschke, H. (1984). Cued recall in amnesia. Journal of Clinical and ExperimentalNeuropsychologv, 6 433-440.Carp, F.M. (1988). Significance of mobility for the well-being of the elderly. InTransportation in an aging society: Improving mobility and safety of older persons(vol. 2, pp. 1-20). Washington,DC: National Research Council, TransportationResearch Board.Carr, D., Jackson, T.W., Madden, D.J., & Cohen, H.J. (1992). The effect of age ondriving skills. Journal of the American Geriatric Society, 40 567-573.Carr, D., Schmader, K., Bergman, C., et al (1991). A multidisciplinary approach in theevaluation of demented drivers referred to geriatric assessment centers. Journal ofthe American Geriatric Society, 39, 1132-1136.Cohen, J. (1988). Statistical power analysis for the behavioral sciences.  (2nd ed.). NewJersey: Lawrence Erlbaum Associates.Cooper, P.J., Tallman, K., Tuokko, H. & Beattie, B.L. (1992). Vehicle crashinvolvement and cognitive deficit in older drivers. Journal of Safety Research,  (inpress)Coopersmith, H.G., Korner-Bitensky, N.A. & Mayo, N.E. (1989). Determining medicalfitness to drive: Physicians' responsibilies in Canada. Canadian Medical Association Journal, 140, 375-378.Coval, M., Crockett, D., Holliday, S. & Koch, W. (1985). A multi-focus assessmentscale for use in frail elderly populations. Canadian Journal on Aging, 4, 101-109.Coyne, A.C., Feins, L.C., Powell, A.L. & Joslin, B.L. (1990).Driving patterns of dementia diagnostic clinic outpatients [Abstract].Proceedings of the 43rd Annual Scientific Meeting of the GerontologicalSociety of America, 30, 194A.166Crockett, D., Tuokko, H., Koch, W. & Park, R. (1989). The assessment of everydayfunctioning using the Present Functioning Questionnaire and the Functional RatingScale in elderly samples. Clinical Gerontologist, 8 3-25.Diller, L., Ben-Yishay, Y., Gerstman, L.J., Goodkin, R., Gordon, W. & Weinberg, J.(1974). Studies in cognition and rehabilitation in hemiplegia (RehabilitationMonograph No. 50). New York: New York University Medical Center Institute ofRehabilitation Medicine.Donnely, R. & Karlinsky, H. (1990). The impact of Alzheimer's disease on drivingability: A review. Journal of Geriatric Psychiatry and Neurology, 3 71-76.Drachman, D.A. (1988). Who may drive? Who may not? Who shall decide?. Annals ofNeurology, 24 787-788.Engum, E., Cron, L., Hulse, C., Pendergrass, T. & Lambert, C. (1988). Cognitivebehavioral drivers inventory. Cognitive Rehabilitation, 6 34-50.Flicker, C., Bartus, R.T., Crook, T.H. & Ferris, S.H. (1984). Effects of aging anddementia upon recent visuospatial memory. Neurobiology of Aging,  5 275-283.Freund, K. (1991). The politics of older driver legislation. [Abstract]. Proceedings of the43rd Annual Scientific Meeting of the Gerontological Society of America,  31 156.Friedland, R., Koss, E., Kumar, A., Gaine, S., Metzler, D., Haxby, J.V., & Moore, A.(1988). Motor vehicle crashes in dementia of the Alzheimer type. Annals ofNeurology, 24 782-786.Fuller, R. (1984). A conceptualization of driving behavior as threat avoidance.Ergonomics, 27 1139-1155.Gilley, D.W., Wilson, R.S., Bennett, D.A., Stebbins, G.T., Bernard, B.A., Whalen, M.E.,& Fox, J.H. (1991). Cessation of driving and unsafe motor vehicle operation bydementia patients. Archives of Internal Medicine, 151 941-946.Glass, G.V., Peckham, P.D. & Sanders, J.R. (1972). Consequences of failure to meetassumptions underlying the analysiss of variance and covariance. Review ofEducational Research, 42, 237-288.Gouvier, W.D., Maxfield, M.W., Schweitzer, J.R., et al (1989). Psychometric predictionof driving performance among the disabled. Archives of Physical Medicine andRehabilitation, 70, 745-750.Graca, J.L. (1986). Driving and aging. Clinics in Geriatric Medicine, 2, 577-599.167Grady, C.L., Grimes, A.M., Patronas, N., Sunderland, T., Foster, N.L. & Rapoport, S.I.(1989). Divided attention, as measured by dichotic speech performance, indementia of the Alzheimer type. Archives of Neurology, 46 317-320.Graf, P., Tuokko, H. & Gallie, K. (1990). Attentional deficits in Alzheimer's Diseaseand related dementias. In J.T. Enns (Ed.), The development of attention: Researchand theory (pp. 527-544). New York: Elsevier Science Publishers.Groeger, J.A. (1988). Underlying structures: Driver models and model drivers. In J.A.Rothengatter & R. de Bruin (Eds.), Road user behavior: Theory and research  (pp.518-526). Assen: Van Gorcum.Gutman, G.M., Gee, E.M., Bojanowski, B.C. & Mottet, D. (1986). Fact book on aging inBritish Columbia. Burnaby, B.C.: Gerontology Research Centre.Hakstian, A.R. & McLean, P.D. (1989, April). Unipolar depression versus normalpopulations: Characteristics and four-item scale for efficient screening ofdepression. An extended report.  (Available from R. Hakstian, Department ofPsychology, University of British Columbia, V6T 1Y7).Hale, P.N, Schweitzer, J.R., Slipp, M. & Gouvier, W.D. (1987). A small-scale vehicle ofassessing and training driving skills among the disabled. Archives of Physical Medicine and Rehabilitation, 68 741-742.Hopewell, C.A. & van Zomeren, A.H. (1990). Neuropsychological aspects of motorvehicle operation. In D. Tupper & E. Cicerone (Eds.), The neuropsychology ofeveryday life (pp. 307-334). Boston: Kluwer Academic Publishers.Howell, D. C. (1982). Statistical methods for psychology(2nd  ed.). Boston: PWSHuff, J.F., Becker, J.T., Belle, S.H., Nebes, R.D., Holland, A.L. & Boller, F. (1987).Cognitive deficits and clinical diagnosis of Alzheimer's disease. Neurology, 37,1119-1124.Hughes, C.P., Berg, L., Danziger, W.L., Cohen, L.A. & Martin, R.L. (1982). A newclinical scale for staging dementia. British Journal of Psychiatry, 140 566-572.Hunt, L.A., Edwards, D., Morris, J.C. & Mui, A. (1990). Driving performance in MildSenile Dementia of the Alzheimer Type [Abstract]. Proceedings of the 43rd AnnualScientific Meeting of the Gerontological Society of America,  30 194A.Kaszniak, A.W., Keyl, P.M. & Albert, M.S. (1991). Dementia and the older driver.Human Factors, 33, 527-537.168Katzman, R. (1985). Aging and age-dependent disease: Cognition and dementia. InCommittee on an Aging Society.Institute of Medicine and National ResearchCouncil (Ed.), America's aging, health in an older society (pp. 129-152).Washington, DC: National Academy Press.Kausler, D.H. (1982). Experimental psychology and human aging. New York: JohnWiley & Sons.Keyl, P.M., Rebok, G., Bylsma, F. & Rodman, A. (1991). Simulated driving perfomancein Alzheimers patients in relation to accident involvement and near misses.[Abstract]. Proceedings of the 44th Scientific Meeting of the Gerontological Society of America, 31 155.Kirk, R.E. (1982). Experimental design: Procedures for the behavioral sciences  (2nded.). Belmont, CA: Wadsworth.Korteling, E. (1990). Perception-response speed and driving capabilities ofbrain-damaged older drivers. Human Factors, 32, 95-108.Koss, E., Ober, B., Delis, D. & Friedland, R. (1984). The Stroop Color-Word Test:Indicator of dementia severity. International Journal of Neuroscience,  24 53-61.Kramer,'U. & Rohr, G. (1982). A model of driver behavior. Ergonomics, 25 891-907.Lezak, M.D. (1983). Neuropsychological assessment (2nd ed.). New York: OxfordUniversity Press.Loewenstein, D.A., Amigo, E., Duara, R., et al (1989). A new scale for the assessmentof functional status in Alzheimer's disease and related disorders. Journals ofGerontology: Psychological Sciences,  44 114-121.Logsdon, R. & Teri, L. (1990). Driving and Alzheimer's Disease. Proceedings of the 43rd Annual Scientific Meeting of the Gerontological Society of America,  30 53A.Loutzenheiser, D.W. & King, C.C. (1976). Geometric design. In J.E. Baerwald (Ed.),Transportation and traffic engineering handbook (pp. 594-674). Toronto: PrenticeHall.Lucas-Blaustein, M.J., Filipp, L., Dungan, C. & Tune, L. (1988). Driving in patient withdementia. Journal of American Geriatric Society,  36, 1087-1091.McEwan, K., Donnelly, M., Robertson, D. & Hertzman, C. (1991). Mental health problems among Canada's seniors. Ottawa: Department of National Health andWelfare.169McFarland, R.A., Tune, G.S. & Welford, A.T. (1964). On the driving of automobiles byolder people. Journal of Gerontology, 19 190-197.McFarland, R.A., McGee, H.W., Moore, W., Knapp, B.G., & Saunders, J.H. (1978).Decision sight distance for highway design and traffic control requirements.Washington, DC: Federal Highway Administration. (FHWA-RD report number78-78, available from NTIS).McGlynn, S.M. & Kaszniak, A.W. (1991a). Unawareness of deficits in dementia andschizophrenia. In G.P. Prigatano & D.L. Schacter (Eds.), Awareness of deficits after brain injury: Clinical and theoretical issues  (pp. 84-110). New York:Oxford University Press.McGlynn, S.M. & Kaszniak, A.W. (1991b). When metacognition fails: Impairedawareness of deficit in Alzheimer's disease. Journal of Cognitive Neuroscience, 3,183-189.McGlynn, S.M. & Schacter, D.L. (1989). Unawareness of deficits in neuropychologicalsyndromes. Journal of Clinical and Experimental Neuropsychology,  11 143-205.McKenna, F.P., Duncan, J. & Brown, I.D. (1986). Cognitive abilities and safety on theroad: A re-examination of individual differences in dichotic listening and searchfor embedded figures. Ergonomics, 29 649-663.McKnight, A.J. (1988). Driver and pedestrian training. In Transportation in an agingsociety: Improving mobility and safety of older persons (vo1.2, pp. 101-133).Washington, DC: National Research Council, Transportation Research Board.Michon, J.A. (1985). A critical view of driver behavior models: What do we know; whatshould we do?. In L. Evans & R.C. Schwing (Eds.), Human behavoir and traffic safety (pp. 485-520). New York: Plenum Press.Michon, J.A. (1988). Driver models: How they move. Proceedings of the SWOV Conference Traffic safety theory and research methods (pp. 1-19). Amsterdam:North Holland.Morris, J.C. & Fulling, K. (1988). Early Alzheimer's Disease. Archives of Neurology,4S 345-349.Mortimer, J.A., Shuman, L.M. & French, L.R. (1981). Epidemiology of dementingillness. In J.A. Mortimer & L.M. Schumen (Eds.), The epidemiology of dementia(pp. 3-23). New York: Oxford University Press.170Naatanen, R. & Summala, H. (1976). Road user behavior and traffic accidents.Amsterdam: North Holland.National Research Council (1988). Transportation in an aging society: Improvingmobility and safety of older persons (Vol 1). Washington, DC: Author.Norusis, M.S. & SPSS Inc. (1988). SPSS-X advanced statistics guide (2nd ed.).Chicago: SPSS Inc.Odenheimer, G., Beaudet, M., Albert, M., Grande, L., Jette, A. & Minaker, K. (1991).Cognitive performance correlates with driving performance in the elderly[Abstract]. Proceedings of the 44th Annual Scientific Meeting of the Gerontological Society of America, 31 160.Owsley, C., Ball, K., Sloane, M.E., Roenker, D.L. & Bruni, J.R. (1991). Visual /cognitive correlates of vehicle accidents in older drivers. Psychology and Aging,6 403-415.Parasuraman, R. & Nestor, P.G. (1991). Attention and driving skills in aging andAlzheimer's disease. Human Factors, 33 539-557.Pirozzolo, F.J., Christensen, K.J., Ogle, K.M., Hansch, E.C. & Thompson, W.G. (1981).Simple and choice reaction time in dementia: Clinical Implications. Neurobiologyof Aging, 2, 113-117.Reisberg, B., Ferris, S.H., Anand, R., Buttinger, C., Borenstein, J.1, Sinaiko, E. & deLeon, M.J. (1984). Clinical assessment of cognition in the aged. In C.A.Shamoian (Ed.), Biology and treatment of dementia in the elderly (pp. 15-37). Washington, D.C.: AmericanPsychiatric Press.Reisberg, B., Ferris, S.H., Anand, R., de Leon, M.J., Schneck, M.K. & Crook, T. (1985).Clinical assessment of cognitive decline in normal aging and primary degenerativedementia: Concordant ordinal measures. In P. Pinchot, P. Berner, R. Wolf & K.Thau (Eds.), Psychiatry (Vol. 5, pp. 333-338). New York: Plenum Press.Reisberg, B., Ferris, S.H., Borenstein, J., Sinaiko, E., de Leon, M.J. & Buttinger, C.(1986). Assessment of presenting symptoms. In L.W. Poon (Ed.), Handbook forclinical memory assessment of older adults (pp. 108-128). Washington, DC:American Psychological Association.Reisberg, B., Schneck, M.K., Ferris, S.H., Schwartz, G.E. & de Leon, M.J. (1983). BriefCognitive Rating Scale (BCRS): Findings in primary degenerative dementia.Psychopharmacology Bulletin, 19, 47-50.171Retchin, S.M., Cox, J., Fox, M., & Irwin, C. (1988). Performance-based measurementsamong elderly drivers and non-drivers. Journal of the American Geriatric Society,36, 813-819.Reuben, D.B. (1991). Dementia and driving. Journal of the American Geriatric Society,39 1137-1138.Rosenbloom, S. (1988). The mobility needs of the elderly. In Transportation in an agingsociety: Improving mobility and safety of older persons (Vol.2, pp. 21-71).Washington, DC: National Research Board, Transportation Research Board.Rothe, J.P. (1990). The safety of elderly drivers: Yesterdays young in today's traffic.New Brunswick, NJ: Transaction Publishers.Rothengatter, J.A. (1987). Introduction: Current issues in road user research. In J.A.Rothengatter & R. de Bruin (Eds.), Road users and traffic safety (pp. 1-10).Assen: Van Gorcum.Rothengatter, J.A. & de Bruin, R. (1987). Road users and traffic safety.  Assen: VanGorcum.Salthouse, T.A. (1985). A theory of cognitive aging. Amsterdam: North Holland.Schieber, F. (1988). Vision assessment technology and screening older drivers: Pastpractices and emerging techniques. In Transportation in an aging society: Improving mobility and safety of older persons  (Vol.2, pp. 325-378).Washington, DC: National Research Council, Transportation Reseach Board.Shinar, D. (1978). Psychology on the road: The human factor in traffic safety. NewYork: John Wiley & Sons.Sivak, M., Olson, P., Kewman, D., Won, H. & Henson, D. (1981). Driving andperceptual/cognitive skills: Behavioral consequences of brain damage. Archives ofPhysical Medicine and Rehabilitation,  62, 476-483.Stokx, L.C. & Galliard, A.W.K. (1986). Task and driving performance of patients with asevere concussion of the brain. Journal of Clinical and ExperimentalNeuropsychology, 8 421-436.Storandt, M. & Hill, R.D. (1989). Very mild senile dementia of the Alzheimer type.Archives of Neurology, 46, 383-386.Summala, H. (1985). A pessimistic prediction?. In L. Evans & R.C. Schwing (Eds.),Human behavior and traffic safety (pp. 260-278). New York: Plenum Press.172Tabachnick, B. & Fidel, L.S. (1983). Using multivariate statistics. New York: Harper& Row.Tallman, K.S., Beattie, B.L. & Tuokko, H. (1991). Investigating the driving problems ofpersons with Alzheimer Disease: Research strategies. Proceedings of the 16th Annual Conference of the Institute of Transportation Engineers, 1 105-122.Tallman, K.S., Cooper, P.J. & Beattie, B.L. (1992). Risk compensation in older drivers.Recovery, (in press)Teng, E.L., Chui, H.C., Schneider, L.S. & Metzger, L.E. (1987). Alzheimer's dementia:Performance on the Mini-Mental State Exam. Journal of Consulting and ClinicalPsychology, 55 96-100.Tierney, M.C., Snow, G., Reid, D., Zorzitto, J. & Fisher, R.H. (1987). Psychometricdifferentiation of dementia: Replication and extension of the findings of Storandtand coworkers. Archives of Neurology, 44 720-722.Tuokko, H., Crockett, D., Beattie, B.L., Horton, A. & Wong, M. (1986). The use ofrating scales to assess psycho-socialfunctioning in demented patients [Abstract]. Journal ofClinical and Experimental Neuropsychology, 7 651.Tuokko, H. & Crockett, D. (1987). Central cholinergic deficiency WAIS profiles in anondemented aged sample. Journal of Clinical and Experiemental Neuropsychology, 9 225-227.Tuokko, H. & Crockett, D. (1990). Assessment of everyday functioning in normal andmalignant memory disordered elderly. In D.E. Tupper & K.D. Cicerone (Eds.), Theneuropsychology of everyday life: Issues in development and rehabilitation (pp.135-181). Boston: Kluwer Academic Publishers.van Wolffelaar, P.C., Rothengatter, J.A. & Brouwer, W.H. (1987). Compensationstrategies in elderly car drivers. In E.D. Megaw (Ed.), Contemporary Ergonomics: Proceedings of the Ergonomics Society's 1987 Annual Conference (pp. 95-100).London: Taylor & Francis.van Wolffelaar, P.C., van Zomeren, A.H., Brouwer, W.H. & Rothengatter, J.A. (1988).Assessment of fitness to drive of brain damaged persons. In J.A. Rothengatter &R. de Bruin (Eds.), Road user behavior: Theory and research (pp. 302-309).Assen: Van Gorcum.173van Zomeren, A.H., Brouwer, W.H. & Minderhoud, J.M. (1987). Acquired brain damageand driving: A review. Archives of Physical Medicine and Rehabilitation, 68697-705.van Zomeren, A.H., Brouwer, W.H., Rothengatter, J.A. & Snoek, J. (1988). Fitness todrive a car after recovery from severe head injury. Archives of Physical Medicine and Rehabilitation, 69 90-96.Vitaliano, P.P., Russo, J., Breen, A.R., Vitiello, M.V. & Prinz, P.N. (1986). Functionaldecline in the early stages of Alzheimer disease. Psychology and Aging, 1, 41-46.Waller, P.F. (1988). Renewal licensing of older drivers. In Transportation in an agingsociety: Improving mobility and safety of older persons  (Vol.2, pp. 72-100).Washington, DC: Transportation Research Board.Wild, K.V., Kaye, J.A. & Campbell, G. (1991). Predictors of driving status in dementedelderly [Abstract]. Proceedings of the 116th Annual Meeting of the American Neurological Association (pp.61). Seattle, WA.Winer, B.J. (1971). Statistical principles in experimental design. New York:McGraw-Hill.Yanik, A. (1985). What accident data reveal about elderly drivers  (Research Rep. No.851688). Warrendale, PA: Society of Automotive Engineers.Younger, M.S. (1985). A first course in linear regression  (2nd ed.). Boston: DuxburyPress.174Appendix ASubject Selection Procedures and Diagnostic CriteriaTable of ContentsI. Copies of Test Forms used for Diagnostic Assessment 1.Alzheimer Clinic Neuropsychological Test Profile^ 1762. Mini Mental State Exam^ 1773. Functional Rating Scale - Diagnostic Criteria^ 1784. Present Functioning Questionnaire^ 1805. Medical Questionnaire^ 182II. Description of Subject Selection Procedures and Diagnostic AssessmentSubject Selection Criteria^ 186Selection criteria for dementia sample^ 186Selection criteria for mid-age controls 186Exclusionary medical criteria^ 186Severity criteria^ 187Diagnostic Procedures and Assessment Instruments^ 189Diagnostic procedures: Overview^ 189Assessment instruments^ 189Diagnosis^ 192175Group Characteristics on Neuropsychological Tests and Clinic RatingScales^ 193Alzheimer Clinic Psychological Assessment Battery^ 193Present Functioning Questionnaire^ 197Functional Rating Scale^ 197Summary of results on diagnostic tests and rating scales^ 198List of TablesTable A-1 Alzheimer Clinic Neuropsychological Test Battery^ 190Table A-2 Test scores on Alzheimer Clinic Battery^ 194Table A-3 Present Functioning Questionnaire: Collateral report of problems ineveryday functioning^ 199Table A-4 Impairment levels on the Functional Rating Scale^ 201176University Hospital - UBC SiteAlzheimer ClinicPSYCHOLOGICAL TEST BATTERY PROFILE(compared to similarly aged functioningpersons)Date Assessed:  ^Native Language: ^Assessed by:  ^Handedness: ^SCORESTEST NAME VeryImpairedBelowAverage Average GoodMAS :^Social BehaviorAuditory Receptive Lang.Visual Receptive Lang.Mental StatusOrientationMoodExpressive Lang.AccessibilityWAIS-R:^InformationDigit SpanSimilaritiesDigit SymbolBlock DesignLURIA-NEBRASKA:^DesignsImmediate RecallReproductionDelayed RecallWORD FLUENCY:BUSCHKE CUED RECALL:Immediate Recall (FRI)Retrievel (Total FR)Acquistion (Total TR)Retention (TR7) ---CLOCKS:^DrawReadSetTAPPING:^Right HandLeft HandGRIP:^Right HandLeft HandPRESENT FUNCTIONINGQUESTIONNAIRE:PersonalityEveryday ‘LanguageMemorySelf-CareACTUAL TIME:EXAM DATE:177MINI—MENTAL STATEScreening Visit( mo I day I yr )^(24—Hour)SCOREORIENTATION1. , What is the (year) (season) (date) (day) (month)?^,.--(maximum score: 5);..^ ..,2. Where are we (province) (country) (town) (hospital) (floor)?(maximum score: 5)REGISTRATIONName 3 objects (ball, flag, tree) One second to say each. Then, askthe patient all 3 after you have said them. Give one point for each1ItIrWak^maweet-maxir^ ,repeat'1all 3...• Count number of trials and record.^(No. of Trials)ATTENTION AND CALCULATION— Serial Ts— Spell "world" backwards. One point for each correct answer.(maximum score: 5)RECALLAsk for the 3 objects repeated above. Give 1 point for each correctanswer. (maximum score: 3)LANGUAGE1. Name a pencil and watch. (2 points)2. Repeat the following, "No ifs, ands or buts." (1 point)3. Follow a three—stage command: "Take a paper in your right hand,fold it in half, and put it on the floor." (3 points)4. Read and obey the following: Close your eyes (1 point)Write a sentence ( 1 point)Copy design^(1 point)(maximum score: 9)Fourth Assessment:Fifth Assessment:explanatory statements)the following abilities (see3 3 42ONSET:  ^DATE OF ONSET: ^  AGE AT ONSET:(see explanatory definition)DURATION:^(Number of months since first noticed).First Assessment:Second Assessment:Third Assessment:Impairment of memory and three or more o2^i^3^4^51ABILITIES1Mlemorl,Social/OccupationalHomo andHobbiesPersonal CarsLanguageProblem solving/ReasoningAffectOrientation4 S i 2 2 3 4 342 23 4[5 1 5 1Functional Rating ScaleUNIVERSITY HOSPITAL - UBC SITEALZHEIMER CLINICDIAGNOSTIC CRITERIADIAGNOSIS OF ALZHEIMER DISEASEFirst Assessments Date: ^Comments: Second Assessments Date: ^Comments: Third Assessments -Date:  ^Dx:  ^Modifiers: ^Comments: ^Fourth Assessments Date:  ^Dx:  ^Modifiers: ^Comments: ^Fifth A^nt: Date:  ^Dx:  ^Modifiers: ^Comments: ^178Dx:^ Modifiers: ^Dx:^ Modifiers: ^HtoUHI ZHElHC0440 gN?•IEelHealthy(1) Questionable(1) Mild(3) Moderate(4)Severe(5)MauryNo deficit orInconsistentforgetfulness evidentonly on clinicalinterviewVariable symptomsreported by patient orrelative, seeminglyunrelated to level offunctioningMemory losses whichinterfere with dailyliving, more apparentfor recent eventsKModerate memory loss,only highly learnedmaterial retained, newmaterial rapidly lostSevere memory loss,unable to recallrelevant aspects ofcurrent life, verysketchy recall of pastlifeSocial/CommunityandOccupationalaNeither patient norrelative aware of anydeficitVariable levels offunctioning reported bypatient or relatives,no objective evidenceof deficits inemployment or socialsituationsPatient or relativeaware of decreasedperformance in demandingemploy:pant or socialsettings, appears normalto casual inspectionPatient or relativeaware of ongoinigdeterioration, does notappear normal toobjective observer,unable to perform job,little independentfunctioning outside homeMarked impairment ofsocial functioning, noindependent functioningoutside homeHome andHobbiesNo changes noted bypatient or relative Slightly decreasedinvolvement inhousehold tasks andhobbiesEngages in socialactivities in the homebut definite impairmenton some household tasks,some complicated hobbiesand interests abandonedOnly simplechores/hobblespreserved, mostcomplicatedhobbies/interestsabandonedNo independentinvolvement in home orhobbiesPersonalCareFully capable of self-care Occasional problemswith self-care reportedby patient/relatives orobservedNeeds prompting tocomplete tasksadequately (i.e.dressing, feeding,)Igiene)Requires supervision indressing, feeding,hygiene, and keepingtrack of personaleffectsNeeds constantsupervision andassistance with feeding,dressing, or hygieneetc.LanguageSkillsNo disturbance oflanguage reported bypatient or relativesubjective complaintof, or relativereports, languagedeficits, usuallylimited to word findingor namingPatient or relativereports variabledisturbances in suchskills as articulationor naming, occasionallanguage impairmentevident duringexaminationPatient or relativereports consistent,language disturbance,language disturbanceevident on examinationSevere impairment ofreceptive and/orexpressive language,production ofunintelligible speechProblem SolvingandReasoningSolves everyday problemsadequately Variable impairment ofproblem solving,similarities,differencesDifficulty in handlingcomplex problems Marked impairment oncomplex problem solvingtasksUnable to solve problemsat any level, trial anderror behavior oftenobservedAffectNo change in affectreported by patient orrelativeAppropriate concernwith respect tosymptomatologyInfrequent changes inaffect (e.g.irritability) reportedby patient or relative,would appear normal toobjective observerFrequent changes inaffect reported bypatient or relative,noticeable to objectiveobserverSustained alterations ofaffect, impaired contactwith reality observed orreportedOrientation Fully oriented Occasional difficultieswith time relationships Marked difficulty withtime relationships Usually disoriented totime and often to place Oriented only to personor not at allDifferentiation of Onset and ProgressionDescriptionC da 1^Neither patient nor relative is aware of any deficits or changes in behavior.2 Variable/fluctuating symptoms reported by patient or relative, seemingly unrelated to level of functioning.3^Gradual onset, presence of sustaind progressive deterioration.4 Abrupt onset, presence of sustained deterioration and stepwise progression.5^Abrupt onset, presence of sustained deterioration and unremitting progression.6 Abrupt onset, stable course.7^Abrupt onset, self-limited.8 Gradual onset, stable course.9^Gradual onset, fluctuating course.10 Abrubt onset, fluctuating course.11^Other.12 Gradual onset, presence of sustained deterioration and stepwise progression.PRESENT FUNCTIONING OUESTIONNAIRE^ 180Department of PsychologyUNIVERSITY HOSPITAL, UBC SITEInformant:Relationship to Patient:Comments:Personality:Being IRRITABLE or ANGRYBeing MISERABLE or DEPRESSEDBeing TENSE or PANICKYBeing APATHETICBeing AGITATED or HYPERACTIVEBeing ANXIOUS or AFRAIDStating that "THINGS AREN'T REAL"Complaining of "UPSETTING THOUGHTS"Being AGGRESSIVEBeing SUSPICIOUSBeing INSENSITIVE TO OTHERS' FEELINGSShowing INAPPROPRIATE SMILES OR LAUGHTERShowing DECREASED HOBBY INVOLVEMENTTALKING TO IMAGINARY OTHERSExhibiting INAPPROPRIATE SEXUAL ACTIVITIESEveryday Tasks:'^Problems PERFORMING HOUSEHOLD TASKSProblems HANDLING MONEYProblems SHOPPINGProblems•FINDING THEIR WAY INSIDE A HOUSE OR BUILDING,Problems FINDING THEIR WAY AROUND FAMILIAR STREETSProblems RECOGNIZING SURROUNDINGSProblems RECOGNIZING THE DATE OR DAY OF WEEKProblems RECOGNIZING THE TIME OF.DAYAWAKING AT NIGHT AND THINKING IT IS DAYProblems READING (except as caused by poor vision)Problems PERFORMING JOBProblems DRIVING A CAR (if could before)Language skills:Problems FINDING WORDS TO EXPRESS HIM/HERSELFLOSING HIS/HER VOCABULARYProblems PRONOUNCING WORDSProblems UNDERSTANDING OTHERSMore frequently SLURRING WORDSProblems CLIPPING ENDS OFF WORDS OR SENTENCESMore frequently STUTTERINGProblems FINDING NAMES FOR COMMON OBJECTSProblems FORMING ANY INTELLIGIBLE SPEECH181PRESENT FUNCTIONING QUESTIONNAIREDepartment of PsychologyUNIVERSITY HOSPITAL, UBC SITEMemory functions:Problems REMEMBERING PREVIOUS ACTIONSON THE SAME DAYProblems REMEMBERING PAST LIFE EVENTSASKING QUESTIONS REPEATEDLY (despite answers)Problems RECOGNIZING FACES OF OLD FRIENDS OR FAMILYProblems RECOGNIZING NAMES OF OLD FRIENDS OR FAMILYProblems REMEMBERING NEWLY—INTRODUCED PERSONSProblems LEAVING STOVE BURNERS, WATER TAPS,AND LIGHT SWITCHES TURNED ONProblems MAINTAINING A TRAIN OF THOUGHTProblems CONCENTRATING-Problems REMEMBERING WHERE HE/SHE PLACED OBJECTSINCREASINGLY FRUSTRATED over problemsremembering or thinkingINCREASINGLY DEFENSIVE about problems rememberingProblems REMEMBERING IMPORTANT PERSONAL DATESSEEMINGLY UNAWARE OF IMPORTANT CURRENT EVENTSDOES NOT KNOW OWN NAMESelf-care functions;EATING MESSILY WITH SPOON ONLYEATING MESSILY WITH SPOON AND WITH SOLID FOODSHAS TO BE FED BY SOMEONE ELSE.Problems DRESSING, e.g., OCCASIONAL MISPLACED BUTTONS'Problems DRESSING, e.g., PUTS ON CLOTHESIN WRONG SEQUENCEProblems. DRESSING, e.g., UNABLE TO DRESS SELFProblems with OCCASIONALLY WETTING BEDProblems with FREQUENTLY WETTING BEDDOUBLY INCONTINENTDOES NOT WASH HIM/HERSELF ENOUGHMUST BE BATHED BY SOMEONE ELSEGROOMING (combing of hair, etc.) INADEQUATEMUST BE GROOMED BY SOMEONE ELSENEEDS CONSTANT SUPERVISION IN CARING FOR SELFtAjt UNIVERSITY'Lim- HOSPITALInvestigators:B. L. Beattie, MD, FRCPCH. Tuokko, Ph.D.University Hospital, UBC SiteHEALTH QUESTIONNAIRE182The purpose of this project is to investigate the driving behaviourof senior citizens with and without memory complaints. At the end ofthe study we hope to have more objective criteria for assessingdriving competence.Each person in the study will be asked to do a number of differenttasks involving language, memory, motor skills, and driving ability.As part of this study, we would like you to fill out this simplehealth questionnaire which will take approximately ten to fifteenminutes of your time. If you complete this questionnaire, it will beassumed that you consent to providing us with this information.If, for any reason, you wish to withdraw from the study, you are freeto do so at any time without further comment. In order to ensureconfidentiality, names will be replaced by research numbers on allinformation collected.DRIVING AND AGINGSTUDY(604) 228-7926UBC SITE2211 Wesbrook MallVancouverBritish ColumbiaCanada V6T 2B5(604) 228-7121University Hospitalalso includes theShaughnessy Site and theGeorge Derby CentreMedical QuestionnaireName^Age ^AddressPhoneWithinQUESTIONNAIRE - PART 1the past year, have you suffered from: Please circleYES^or NO1. Cough ^ YES^ NO2. Sputum production ^ YES^ NO3. Chest pain^ YES^ NO4. Shortness of breath ^ YES^ NO5. Lightheadedness or faintness^ YES^ NO6 -: Loss of appetite^ YES^ NO7. Change in weight YES^ NO8. Nausea or vomiting^ YES^ NO9. Change in bladder function^ YES^ NO10. Change in skin or hair YES^ NO11. Fever, sweats, or chills^ YES^ NO12. Loss of energy^ YES^ NO13. Easy bruising YES^ NO14. Have you ever had high blood pressure' YES^ NO15. Do you have a pacemaker' YES^ NO16. Have you had any artificial joint implants' YES^ NO17. Do you have diabetes' YES^ NO183Medical Questionnaire18. Are you on any medications? (if yes, please list)YES^ NOA. PrescriptionB. Over-the-counter19. What is your usual alcohol intake?A. Not at allB. Less than 5 drinks/weekC. 5 to 10 drinks/weekD. More than 10 drinks/week20. Please list other medical problems that you have had^(ifany).QUESTIONNAIRE - PART 2A.^Within the past year, have you suffered from:^Please circleYES^or^NOor^NO1. Headaches ^ YES^ NO2. Seizures YES^ NO3. Blackouts ^ YES^ NO4. Memory impairment^ YES^ NO5. Confusion^ YES^ NO6. Trouble with vision (not improved by glasses)..YES^ NO7. Numbness, pins and needles or loss ofsensation^ YES^ NO8. Weakness, slowness of movfement, orincoordination ^ YES^ NO9. Difficulty with speech^ YES^ NO10. Difficulty with swallowing YES^ NO11. Difficulty walking^ YES^ NO184Medical Questionnaire12. Difficulty with control of bladder orbowels^ YES^ NOB.^1.^Have you ever suffered a major headinjury') ^YES^ NO2. Have you ever had surgery on the head' ^YES^ NO3. Have you ever had a stroke' ^YES^ NO4. Have you ever been depressed' ^YES^ NO5. Have you ever had any other psychiatricdiagnosis or treatment? Comment if youwish^ YES^ NO185€186Subject Selection Criteria Selection criteria for dementia sample: (a) evidence of cognitive impairment anddecline in everyday functioning sufficient to satisfy the Functional Rating Scale criteria fordementia used at the Alzheimer Clinic, (b) mild cognitive impairment defined as a MiniMental State Exam (MMSE) score between 27 and 20, (c) driving within one month ofrecruitment for the study, and (d) an absence of exclusionary medical criteria.Selection criteria for normal elderly controls:  (a) an absence of significant cognitiveimpairment as indexed on the Functional Rating Scale, (b) an MMSE score of 29 or 30, (c)an absence of exclusionary medical criteria, and (d) 60 years of age or older and drivingwithin one month of recruitment to the study.Selection criteria for mid-age controls: (a) an absence of significant cognitiveimpairment as indexed on the Functional Rating Scale, (b) an MMSE score of 29 or 30, (c)an absence of exclusionary medical criteria, (d) between age 35 and 45, (e) education levelsmatched to the elderly group.Exclusionary medical criteria. The Medical Questionnaire provided the informationneeded to determine whether potential subjects suffered from any of the followingexclusionary medical conditions: major psychiatric disorder, Parkinson's Disease, severearthritis or other disorders interfering with physical mobility, marked loss of vision, mentalretardation, delirium, or neurological disorders resulting from stroke, tumour, or head injury.The medical questionnaire also provided data concerning the number of other diseases andthe number of prescriptions medications used by members of each group.187Severity criteria. A score range on the Mini Mental State Exam from 20 to 27comprised the impairment range for the dementia cases included in this study. This rangeis considered indicative of mild dementia--the disorder of chief concern in this project. Asdiscussed in the introduction to the thesis, moderately to severely impaired individuals canbe assumed a priori to have markedly impaired driving skills. In this study the concern waswith individuals who satisfy diagnostic criteria for dementia but have mild symptoms.The Folstein Mini Mental State Exam (MMSE) is a cognitive status screening measuredesigned to assess pathological impairment in the areas of orientation, concentration,immediate and delayed recall, expressive and receptive language, and visuospatialconstructional ability. The test is considered by many to be a useful instrument for gradingthe severity of dementia (Anthony, LeResche, Unaiza, VonKoroff, & Folstein,1982;Branconhier,1986; and Reisberg, Ferris, Borenstein, Sinaiko, deLeon, & Buttinger, 1986).For this study MMSE scores were not considered during diagnosis thereby providing anindependent source of information on symptom severity.A useful discussion of the characteristics of individuals with scores ranging from 20to 27 points on the MMSE has been provided by Reisberg et al. (1986). Extensive work byReisberg and colleagues (Reisberg et al., 1983, 1984, 1985, 1986) resulted in the developmentand validation the Global Deterioration Scale as an instrument for staging deterioration incognition and behaviour in Alzheimer Disease. A stage involving mild cognitive declineconsistent with 'incipient' Alzheimer Disease is described as follows:The early confusional phase (GDS=3), in which there occur theearliest clear-cut, clinically evident deficits in occupational andsocial functioning, appears to represent a borderline conditionbetween normal aging and AD. (Reisberg et al., 1986, p. 111)188Reisberg et al. (1986, pp. 120-121) describe the relationshipbetween the early confusional stage and MMSE scores asfollows:The borderline between early AD and benign senescentforgetfulness occurs in the MMSE range of 20 to 27,corresponding to the early confusional phase (GDS=3)... It isinteresting to note that these results with respect to the MMSEborders between normal aging and AD, based on prognosticstudies of Reisberg, Ferris, Shulman, et al. (in press), are invery close agreement with earlier studies of Folstein et al.(1975) and others with respect to the borders between normaland pathological aging.Two studies, one of healthy elderly and the other of demented elderly, indicate that acutting score of 27 points on the MMSE should result in little overlap between normal elderlyand dementing individuals. Huff et al. (1987) reported on 85 medically screened normalelderly. ' MMSE scores ranged from 26 to 30; only 3 individuals had scores 28 or below.Teng, Chui, Schneider, and Metzger (1987) investigated 141 carefully diagnosed patientsmeeting DSM III criteria for Primary Degenerative Dementia (and presumed to suffer fromAlzheimer Disease) and found that MMSE scores for these individuals ranged from 0 to 27.No individual with diagnosable dementia was found to have a MMSE score above 27.Thus, several lines of convergent evidence suggest that use of a range of MMSEscores between 20 and 27 is more than mere convenience. The work of Reisberg andcolleagues and the research of others cited in their review (Reisberg et al.,1986) suggest thatthis score range identifies a clinically meaningful group of individuals with mildsymptomatology. And, although it is certainly possible for non-demented individuals to scorein the 20 to 27 range, investigators have found that the upper-end cutting score of 271 89(distinguishing normal functioning from impairment) results in a very little misclassificationof intact versus dementing individuals.Diagnostic Procedures and Assessment Instruments Diagnostic procedures: Overview.  As described in the Method section the process ofestablishing the presence or absence of dementia proceeded in the following manner: (a) allsubjects were assessed on the battery of neuropsychological measures and symptom severityrating scales listed in Table A-1 which are used for diagnostic assessment at the Clinic forAlzheimer Disease and Related Disorders, University Hospital, UBC Site, Vancouver, B.C.(hereafter referred to as the Alzheimer Clinic); (b) collateral information about changes incognitive and everyday functioning was obtained from a significant other and/or a referringphysician using the Present Functioning Questionnaire; (c) the medical questionnaire,providing information related to exclusionary medical conditions, was filled out byparticipants or by participants with the help of a collateral if cognitive impairment wassuspected; (d) test scores on the neuropsychological battery were compared to age appropriatenorms and rated with respect to level of impairment; (e) the test score profile (see copy ofAlzheimer Clinic Psychological Test Profile in this appendix) and the collateral informationwere used by an experienced neuropsychologist to assign impairment ratings on theFunctional Rating Scale (FRS) which then formed the basis for diagnoses of dementia usingthe standard procedures established at the Alzheimer Clinic.Assessment instruments.  The measures included in the neuropsychological test batteryare presented in Table A-1. These measures have been in use at the Alzheimer Clinic for thepast seven years and have established convergent and predictive validity for the assessmentTable A-1Clinic for Alzheimer Disease and Related Disorders Neuropsychological Test Battery and Rating Scales. Subtests of Wechsler Adult Intelligence Scale - Revised:Information, Digit Span, Similarities, Digit Symbol, BlockDesign.Multi-focus Assessment Scale-Revised (MAS-R).Luria Nebraska Battery Item 227 - Visual MemoryWord Fluency (F,A,S).Cued Recall Procedure for Memory AssessmentFinger Tapping Speed (right and left hand)Geriatric Depression Scale (GDS).Personal Functioning Questionnaire (PFQ)190191of pathological cognitive deterioration (see Coval, Crockett, Holliday, & Koch, 1985; Tuokko,Crockett, Beattie, Horton, & Wong, 1986; Tuokko & Crockett, 1987; Tuokko & Crockett,1990).Normative data for the neuropsychological test battery have been obtained locallyfrom a sample of 78 elderly volunteers with demographic characteristics similar to those ofpatients seen at the Alzheimer Clinic (Tuokko, 1986, unpublished data). The normativesample allows identification of deficits that are in excess of what would be expected on thebasis of age alone. Cutoff scores have been established for each measure. Test performanceis rated as either 'very-impaired' (greater than 2 SD below normal sample mean), 'below-average' (greater than 1 SD below normal sample mean), 'average', or 'above-average'relative to the age-norms. Using these test norms, a profile of test scores was constructedwhich, along with the collateral information, formed the basis of diagnostic ratings on theFunctional Rating Scale (described below).The Present Functioning Questionnaire (PFQ) (Crockett, Tuokko, Koch, & Parks, 1989)is an interview instrument used to obtain collateral information useful in diagnosing dementia.The PFQ comprises 65 items divided into 5 subscales soliciting information from collateralsabout deterioration in personality and affect, everyday tasks, language skills, memoryfunctions and, self-care functions. Three demented individuals did not have collateralsavailable for interview on the PFQ. For these individuals information necessary for diagnosis(related to decline in functioning) was obtained from physician records. One normal elderlycontrol whose spouse was a demented participant in the study did not have an alternativecollateral for the PFQ and so interview information was missing for this individual. Half the192mid-age controls were not able to provide the name of a suitable collateral to be interviewed.Diagnosis. Diagnoses of dementia were made by an expert clinician (Holly Tuokko,PhD, Registered Psychologist, Supervising Neuropsychologist Clinic for Alzheimer Diseaseand Related Disorders) using the Functional Rating Scale (FRS) (Tuokko & Crockett, 1990).The FRS is a multidimensional rating scale based on the widely used Clinical DementiaRating (Hughes, Berg, Danzinger, et al., 1982) and is used routinely at the Alzheimer Clinicfor diagnosis of all referred patients (see this appendix for a copy).The FRS comprises eight dimensions: Memory, Social/Community and OccupationalFunctioning, Home and Hobbies, Personal Care, Language Skills, Problem Solving andReasoning, Affect, and Orientation. Subjects are rated on each dimension using a 5-pointscale with dimension-related descriptors for: (1) healthy, (2) questionable, (3) mild, (4)moderate and,(5) severe levels of impairment.From the profile of neuropsychological test scores and the information obtained fromthe collateral, ratings on the FRS were made for each subject. The Cued Recall Procedurefor Memory Assessment (Buschke, 1984), the visual memory subtest from the Luria Battery(Item 227), the Early Memory Scale from the MAS-R, the Information subtest of the WAIS-R, and the Memory Problems subscale of the PFQ provided the input for assessing thepresence and severity of memory impairment. Language skills were assessed from the MAS-R Communication scales (evaluating auditory and visual language comprehension), the MAS-R scale for Expressive Language Skill, the Word Fluency test, and the Language Problemssubscale of the PFQ. Ratings on the Problem Solving and Reasoning dimension of the FRSwere based on the Similarities and Block Design subtests of the WAIS-R. Orientation was193assessed on the 11 item Present Orientation Scale of the MAS-R and the Everyday Taskssubscale of the PFQ. Information for ratings on the FRS dimensions Social/Community andOccupational, Home and Hobbies, and Personal Care was obtained from the PFQ. FRSAffect rating were based on the Geriatric Depression Scale and the Personality and Affectsubscale of the PFQ.Alzheimer Clinic practice for diagnosing dementia requires a minimum rating of 3(mild impairment) for memory, ratings of 3 or more on at least three other dimensions of theFRS, and evidence that the deficits constitute a decline from previous levels of functioning.A rating of 3 for memory corresponds to the following descriptor: "Memory losses whichinterfere with daily living; more apparent for recent events" (Tuokko & Crockett, 1990, p.141). The absence of pathological deterioration was assumed for individuals who scored nomore than 2 on two or fewer dimensions of the FRS.Group Characteristics on Neuropsychological Tests and Clinic Rating Scales Alzheimer Clinic Psychological Assessment Battery. Table A-2 presents groupmeans, standard deviations, and probability levels associated with one-way ANOVAs for theneuropsychological tests and the Geriatric Depression Scale. For all measures except theMulti-focus Assessment Scale-Revised (MAS-R) and the Cued Recall task, F tests wereconducted in the usual manner and follow up comparisons were carried out using Tukey'sHSD test. However, for the MAS-R and the Cued Recall task there were marked differencesin group variances, particularly between the demented group and the mid-age controls.Although these differences constitute a violation of the assumption of homogeneity ofvariance, Box (cited in Howell, 1982, p.297) has shown that even when variances are194Table A-2Test Scores on Alzheimer Clinic Assessment BatteryGroupsMeasure^Demented(n=18)NormalElderly(n=18)Mid-ageControls(n=18)ANOVAPWAIS-R Information SubtestM^ 7.29 a 13.22 C 11.05 b < .001SD 2.59 2.82 2.57Range^3-12 7-18 6-15WAIS-R Digit Span SubtestM^ 9.68 a 11.88 b 10.28 < .030SD 2.91 2.32 1.99Range^5-15 9-19 7-16WAIS-R Similarities SubtestM^ 7.70 a 12.89 b 11.78^b <^.001SD 3.27 2.47 2.02Range^1-14 6-16 8-15WAIS-R Block Design SubtestM^ 7.66 a 13.39 C 10.78^b <^.001SD 2.57 2.75 3.02Range^4-11 7-18 3-15^ (table continued)195Table A-2 (continued)Test Scores on Alzheimer Clinic Assessment BatteryGroupsMeasureNormal^Mid-age^ANOVADemented^Elderly Controls 2.(n=18) (n=18) (n=18)Multi-focus Assessment Scale - Revised (MAS-R)M^54.05 a^58.94SD 4.35^2.68b 59.440.5159-60b <.01*Range^47-59^49-60Immediate Reproduction of Five Geometric FiguresM^, 2.61 a^4.44 b 4.72 b < .001SD 0.77^0.85 0.57Range^1-4^2-5 3-5FAS Word Fluency TestM^ 25.83 a^41.50 b 43.76 b < .001SD 11.76^11.68 10.79Range^9-54^20-60 26-72Cued Recall Procedure for Memory Assessment: Sum of Trials 1-3M^ 9.22 '^28.55 b 30.27 b < .01*SD 6.12^4.17 2.72Range^0-20^18-34 26-35^ (table continued)196Table A-2 (continued)Test Scores on Alzheimer Clinic Assessment Batters,GroupsNormalMeasure^Demented^Elderly(n=18) (n=18)Mid-ageControls(n=18)ANOVA2Finger Tapping Speed (mean taps per 10 sec. with right hand)M^41.68 a^44.12 47.60 b < .016SD 6.33^6.32 4.99Range^31-53^34-59 38-55Finger Tapping Speed (mean taps per 10 sec. with left hand)M^37.01 a^39.34 43.80 b < .005SD 5.95^7.25 4.36Range^27-48^26-56 35-51Geriatric Depression ScaleM^ 5.11^3.27 4.33 .309SD 3.64^2.82 4.48Range^1-15^0-9 0-13Note. Superscripts appearing in the body of the table are associated with post-hoc pairwisecomparisons between group means. Groups with different superscripts are significantlydifferent at the .05 level. Superscript' is associated with the lowest performance level.* Given marked heterogeneity of variance, the significance of the F test was evaluated byBox's procedure for adjusting the degrees of freedom for F crilical (described in Howell, 1982,p.297). Pair-wise multiple comparisons conducted by Scheffe's method.197unequal, a valid and conservative significance test can be carried out if the degrees offreedom for Fcritical are altered from F,( k-1, k(n-1) ) to F, (1, n-1). For the MAS-R and CuedRecall, the adjusted Fcritica with (1, 17) degrees of freedom is 15.72 at p . < .001. The Fobtainecivalues were 17.8 for the MAS-R and 118.5 for the Cued Recall Procedure indicating thatthere were significant main effects for group on both measures. Follow-up multiplecomparisons were carried out using Scheffe's procedure.Present Functioning Questionnaire.  Table A-3 presents group means and standarddeviations on the five subscales of the PFQ. The collaterals for the mid-age controls did notendorse any of the 65 problem items for subjects in this group and the average problemratings for the normal elderly controls were also very low. The data derived from thismeasure were not formally analyzed because of unequal group sizes, the absence of varianceon all subscales for the mid-age group, and the discrepancies between the variances of thenormal elderly and the demented groups. Nonetheless, with the exception of the subscaleconcerned with self-care, it is quite obvious that the collaterals for the demented participantsobserved more problems in these subjects than did the collaterals for the control subjects.Functional Rating Scale. Table A-4 presents data derived from the FRS ratings thatare the basis for diagnostic classifications. As with the PFQ data, the FRS ratings do notlend themselves to parametric analysis due to the absence of variance in the control groupsfor several of the dimensions. However, on all dimensions except Personal Care, there areclear differences between the demented group and the two control groups. Note that thevalue 1 designates healthy while a value of 3 indicates a mild though pathological deficit.Summary of results on diagnostic tests and rating scales.198The overall pattern of scores on the diagnostic assessment battery is, not surprisingly,characterized by the demented group doing significantly less well than both of the twocontrols groups. Given the selection criteria it could hardly be otherwise. However, thereare a few exceptions to this general trend that are worthy of mention.On the WAIS-R Digit Span subtest although the demented group and the normalelderly group had significantly different scores, the mid-age subjects had an average DigitSpan score that was intermediate and did not differ significantly from either of the two oldergroups. On the Finger Tapping speed measures the demented and mid-age groups differedfor both right and left handed tapping while the normal elderly group's tapping speed wasintermediate and did not differ significantly from either the demented or mid-age groups.None of the groups differed significantly on the Geriatric Depression Scale. For the relatedPFQ and FRS subscales of Self-Care and Personal Care there was no indication that thedemented were impaired relative to the control groups.The overall pattern of group differences between the demented and the control groups,combined with the handful of measures on which groups were similar, is a welcome outcome.This pattern of results indicates that although the demented group was measurably impaired,there was nonetheless relative preservation of self-care ability, motor skill and persistence(indexed by finger tapping speed) and, aspects of attention (indexed by Digit Span).Essentially normal functioning in these areas concomitant with cognitive impairment isconsistent with mild or early dementia and encourages confidence that the study selectioncriteria were appropriate for identification of the type of marginally impaired individuals thisproject was designed to investigate.Table A-3Present Functioning Questionnaire: Collateral Report of Problems in Everyday Functioning GroupsMeasureCollateralsforDemented(n=15)CollateralsforNormalElderly(n=17)CollateralsforMid-ageControls(n=9)Personality and Affect ChangesM 3.93 0.29 0.00SD 2.37 0.77 0.00Range 0-8 0-3 0Problems with Everyday TasksM 3.73 0.06 0.00SD 3.41 0.24 0.00Range 0-9 0-1 0Problems with Language SkillsM 2.33 0.00 0.00SD 2.32 0.00 0.00Range 0-8 0 0199(table continued)Table A-3 (continued)Present Functioning Questionnaire: Collateral Report of Problems in Everyday Functioning GroupsCollaterals^Collaterals^Collateralsfor^for^forDemented^Normal^Mid-ageMeasure^ Elderly Controls(n=15) (n=17) (n=9)Problems with Memory FunctionsM 6.66 0.41 0.00SD 3.11 1.06 0Range 0-13 0-4 0Problems with Self-Care FunctionsM 0.60 0.00 0.00SD 1.68 0.00 0.00Range 0-6 0 0Total Problems for all PFQ ScalesM 17.26 0.76 0.00SD 9.49 1.35 0Range 0-34 0-4 0200Table A-4Impairment Levels on the Functional Rating Scale (FRS) GroupsMeasure^Demented(n=18)NormalElderly(n=18)Mid-ageControls(n=18)Memory ImpairmentM 3.61 1.17 1.00SD 0.50 0.38 0.00Range 3-4 1-2Social/Community and Occupational ImpairmentM 2.77 1.00 1.00SD 0.46 0.00 0.00Range 2-3Home and Hobby ImpairmentM 2.89 1.05 1.00SD 0.58 0.23 0.00Range 2-4 1-2(table continued)201Table A-4 (continued)Impairment Levels on the Functional Rating Scale (FRS) GroupsMeasure^Demented(n=18)NormalElderly(n=18)Mid-ageControls(n=18)Personal Care ImpairmentM 1.28 1.00 1.00SD 0.83 0.00 0.00Range 1-4Language ImpairmentM 2.93 1.00 1.00SD 0.77 0.00 0.00Range 1-3Problem Solving and Reasoning ImpairmentM 2.89 1.05 1.00SD 0.83 0.23 0.00Range 1-4 1-2202Table A-4 (continued)Impairment Levels on the Functional Rating Scale (FRS) 203MeasureGroupsDemented(n=18)NormalElderly(n=18)Mid-ageControls(n=18)Affect DisturbanceM 3.00 1.11 1.06SD 0.59 0.47 0.24Range 1-4 1-3 1-2Orientation ImpairmentM 2.65 1.00 1.00SD 0.93 0.00 0.00Range 1-4204Appendix B.Copies of Driving Interview Questions Phrased for Interview of Collaterals Table of ContentsQuestion 8. Number of times per week subject drives in morning or eveningrush hour^ 205Question 9. Number of times per week subject drives after dark^ 205Question 12. Annual milage^ 206Question 15. Number of trips per week^ 206Question27. Driving faults^ 207Question 31. Changes in driving ability over time^ 208Question 33. Concerns about driving^ 209Question 35. Difficult driving manoeuvres 210Question 36. Avoidance of driving situations^ 211Question 37. Factors that interfere with driving ability^ 212DRIVING PATTERNSNow I'd like to ask you about ^ driving patterns.8. In an average week, how many days does ^ usuallydrive during the morning or evening rush hour(if less than once a week, codeas 8; if never, code as zero)[13.K./ho response] ^ 99. In an average week, on how many days does ^usuallydrive after dark?(if less than once a week code as 8;if never, code as zero)[D.K./no response] ^  9205shopping ^driving for pleasure visiting family/friends ..going to church ^attend social/culturalevents or entertainment ..engage in sports (e.g. golf)or attend a sporting event(e.g., football game) ....health care service ^volunteer activityattend meetings ^12. Approximately how many miles does .... drive a year?Would you say:less than 5,000 miles ^ 15,000 - 7,999 miles ^ 28,000 - 11,999 miles ^ 312,000 or more miles, or ^ 4do not drive at present time ^ 5[D.K./no response] ^ 9*15. People use their cars for a number of purposes.Does ^ use the car for:206Y NDK/^NO OF TIMESNR^PER WEEK1 2 91 2 91 2 91 2 91 2 91 2 91 2 91 2 91 2 9getting to/from work ....^1 2^9vacation travel  ^1 2^927. (a) I'm going to read you a list of common driving faults peopleoften make. Please tell me which ones^ makes:207drives too fast for conditionsspeeds ^fails to yield right of way ^failure to come to complete stopat stop sign ^runs a red light runs a yellow light ^drives left of centre ^makes an improper pass ^makes an improper turn ^follows too closely moves too slow for traffic flow ..fails to signal  direction (left/right confusion).loses way on familiar routes ...problems with parking ^problems with backing up ^becomes confused or disoriented.[D.K./no response] ^YES NO DK/Na1 2 91 2 91 2 91 2 91 2 91 2 91 2 91 2 91 2 91 2 91 2 91 2 91 2 91 2 91 2 91 2 91 2 91 2 9CHANGES OVER TIME31. Under certain conditions, some older people feel they drivebetter than they used to, some feel they drive the same asthey used to while others feel they don't drive as well.I'm going to read you a list of 12 different driving conditions.For each condition I read, please tell me whether you feel  drives better, the same or worse than at age 35-45 (or specify age when started to drive if later than 45)Not asBetter Same Well^DK/NRnight driving ^ 1coping with headlightglare ^ 1winter driving^1driving in rain & fog 1driving in snow, sleetor slush ^ 1freeway driving ^ 1driving in citystreets ^ 1driving during rushhour ^ 1driving when tired orupset ^ 1driving after drinking 1driving aftermedication ^ 1holiday/vacationdriving ^ 1Comments:2082 3 92 3 92 3 92 3 92 3 92 3 92 3 92 3 92 3 92 3 92 3 92 3 920933* What are your greatest concerns about ^ driving?Y N DK/NR1 2 91 2 91 2 91 2 91 2 91 2 91 2 9loss of attention orconcentration ^losing licence or having toquit driving ^getting hit ^injuring a pedestrian ^losing driving abilitieshitting another car ^other (specify) ^210DIFFICULT DRIVING MANOEUVRES35. There are various types of driving manoeuvres that some peoplefind difficult. I'm going to read you a list of these and foreach one I mention, please tell me how frequently ^finds troublesome.(a) Does ^ frequently, sometimes, seldom or never find ittroublesome to:(b) Have you noticed more difficulty in the past 5 years?(Examiner:^lay down card with responses down for subject)Freq Some Seld.Never NR^Y^N^DK/NR(a)change lanes ^ 1 2 3 4 9 1 2 9(b)steer your car ^ 1 2 3 4 9 1 2 9(c) read traffic signs . 1 2 3 4 9 1 2 9(d)make R. turn  ^1 2 3 4 9 1 2 9(e) make L. turn at astop light or stopsign  ^1 2 3 4 9 1 2 9'(f) make L. turn atuncontrolled inter-section  ^1 2 3 4 9 1 2 9(g)keep up with traffic 1 2 3 4 9 1 2(h)park your car ^ 1 2 3 4 9 1 2 9(i)pass other cars^1 2 3 4 9 1 2 9(j)back your car  ^1 2 3 4 9 1 2 9(k) enter stream of citytraffic  ^1 2 3 4 9 1 2 9(1) enter freeway  ^1 2 3 4 9 1 2 9(m) keep car in its lane 1 2 3 4 9 1 2 9(n) judge distances^1 2 3 4 9 1 2 9(o) keep an appropriatedistance  ^1 2 3 4 9 1 2 936. I'm going to read you a list of conditions underwhich some people don't drive or try to avoiddriving.Please tell me whether this is a condition whicheither tries to avoid driving in or does not drive in:DoesTries not Doesto^drive not DK/avoid in^avoid NR2112 3 92 3 92 3 92 3 92 3 92 3 92 3 92 3 92 3 9at night ^ 1in rain  ^1in fog ^ 1in snow, sleet orslush ^ 1on freeways ^ 1in rush hour ^ 1in heavy traffic areas(bridges, tunnels,congested) ^ 1in heavy holidaytraffic  ^1other (specify)121237* Do any of these factors significantly interfere with ^driving ability?FREQUENTLY SOMETIMES SELDOM NEVER DKpoor hearing ^ 1 2 3 4 9poor vision ^ 1 2 3 4 9poor night vision ... 1 2 3 4 9rain ^ 1 2 3 4 9glare 1 2 3 4 9can't turn head ^ 1 2 3 4 9fatigue ^ 1 2 3 4 9concentration ^ 1 2 3 4 9memory ^ 1 2 3 4 9weakness, pain orstiffness ^ 1 2 3 4 9213Appendix C.Driving Avoidance and Driving Problems:Procedures for Scoring and Tables of Summary ScoresTable of ContentsI. Scoring Procedures for Multi-item Driving Interview Questions Scoring method for driving avoidance question^ 214Scoring methods for driving problem questions 216II. Data Reduction Method for Driving Problem Questions^ 219III. Tables of Summary Scores for Driving Problem Questions Table C-1:^Driving Problem Questions: Scores for the Five Questions andRepeated Measures ANOVA's for each Question^222Table C-2:^Summary Scores for Linear Combination of DrivingProblem Questions^ 227214Scoring Procedures for Multi-item Driving Interview Questions Scoring procedures for driving avoidance question. The Driving Interviewquestion related to driving avoidance was number 36 on the Collateral Interview Scheduleand number 42 on the Participant Interview Schedule (see question 36 Appendix B). Thefollowing description of the scoring method and procedures for handling missing data forthe driving avoidance question will also serve as an example of the methods applied toscoring the multi-item driving problem questions from the Driving Interview (i.e.,questions 27, 31, 33, 35, and 37 in Appendix B). It would be redundant to present adetailed computational description of the scoring of each multi-item question. Therefore,although the following discussion focuses on the scoring of the avoidance question, it isalso intended to communicate the general strategy for scoring the other multi-itemquestionS discussed below.Many of the Driving Interview questions, including the question on drivingavoidance, were composed of several items each pertaining to a different driving situation.For the question on driving avoidance the respondent was informed that the interviewerwas going to read a list of eight driving conditions under which some people don't driveor try to avoid driving. The list of conditions presented were: driving at night; driving inrain; driving in fog; driving in snow,sleet or slush; driving on freeways; driving in rushhour; driving in congested traffic area such as bridges and tunnels; and driving in heavyholiday traffic. The respondent was given three response options for each drivingcondition:(a) don't drive in condition, (b) try to avoid condition, and (c) don't avoidcondition. Also the interviewer had the option of scoring the item as don't know / no215response if the respondent was unable to answer.Rather than carry out analyses for group differences on each item separately, asummary score that provided an overall index of driving avoidance was desired. In orderto combine the ratings for the eight situation, numeric values were assigned to each of theresponse options such that the more avoidance a subject engaged in, the higher their score.Don't avoid was assigned a value of 0 points, try to avoid a value of 1 point, and don'tdrive a value of 2 points.After calculating the sum of points for the eight driving situations, this value wastransposed into a proportionalized score. The index of driving avoidance computed foreach individual was expressed as a percentage of the maximum possible avoidance score.The maximum score possible for the avoidance question was 16 points corresponding tothe sum of 2 points (2 = don't drive) for each of the eight items. For each subject thesum of their responses on the eight avoidance items was divided by the maximum score(i.e., 16) and multiplied by 100 to obtain an index of avoidance expressed as a percentageof the maximum score for the question. For example, an individual who did not drive inthree situations (corresponding to 6 points), who tried to avoid driving in three situation(corresponding to 3 points), and who did not avoid at all in two situations (correspondingto 0 points) would have a sum of 9 points yielding an avoidance index score of 56.2%(because (9/16)100 = 56.2).The procedures outlined above describes how the avoidance index was computedwhen there were no missing data for any of the eight items. This method was applied tothe self-ratings of all subjects except one. However, in some instances the collaterals (and216in one instance, a subject) chose the don't know response option for items they did notfeel knowledgable about, resulting in missing data. A threshold for missing data in multi-item questions was set such that a respondents data would be included in the analysesonly if a minimum of 75% of items had valid answers (i.e., not scored as don't know orno response). Providing this criteria was satisfied, a percentage score could be computedbased on an adjusted maximum. For example, the tolerance for missing values for theavoidance question was two items. If a respondent did not answer two items, the sum oftheir ratings on the remaining six items would be divided by 12 (the maximum possiblescore for six items) to yield a percent-of-maximum score that was comparable to thescores for subjects without missing scores. This method of handling missing valuesresulted in scores that were essentially identical to those which would have resulted fromprorating missing scores by mean substitution.A further refinement of the method for handling missing data was used in caseswhere the responses of subjects and collaterals were to be directly compared. In thatsituation it was important that the summary scores for subjects and collaterals were basedon the same set of items. To accomplish this, analyses were based only on items forwhich valid responses were available for both subjects and their collaterals. Data forsubject-collateral pairs were included in analyses only if at least 75% of items had validanswers.Scoring procedures for driving problem questions.  The five Driving Interviewquestions related to driving problems were: (a) driving faults -- number 27 on theCollateral Interview Schedule and number 33 on the Participant Interview Schedule, (b)217changes in driving ability -- number 31 on the Collateral Interview Schedule and number37 on the Participant Interview Schedule, (c) concerns about deficits in driving skill anddriving mishaps -- number 33 on the Collateral Interview Schedule and number 39 on theParticipant Interview Schedule, (d) difficulties with driving manoeuvres -- number 35 onthe Collateral Interview Schedule and number 41 on the Participant Interview Scheduleand, (e) the extent to which various factors interfered with driving -- number 37 on theCollateral Interview Schedule and number 43 on the Participant Interview Schedule (seeAppendix B for copies of the questions).A single score for each of the five questions (across all items addressed in theparticular question) was derived for each respondent and was expressed as a percent of themaximum possible score in the same manner as described above for the question onavoidanc'e.Missing data were an important concern in scoring the driving problem questions.Because many of the driving problem items required considerable familiarity withsubjects' driving behaviour, there were occasions in which collaterals felt unable to selecta scoreable response (or "valid" response) and instead chose the don't know option. Inderiving an overall score for any of the driving problem questions an item was consideredto have valid answer if: (a) the response option selected by the respondent was one of thescoreable responses (i.e., not don't know / no response), and (b) both the subject and thecollateral had provided a valid answer to the item. As with the driving avoidancequestion, a tolerance limit for missing data was set such that a summary score for thequestion would be computed only if a minimum of 75% of items had a valid answer.218For the driving faults question the respondent was presented with 17 drivingactions (such as speeding, driving too slowly, running red lights, failing to signal turnsetc.) that were described as "common driving faults that people often make".Respondents were asked to indicate in^/ no format whether the subject made any ofthe driving errors listed. The summary score for the faults question was obtained byassigning a score of 1 to each yes (fault) answer, adding the number of faults, andexpressing the total score as a proportion of the maximum possible for the question.The question related to changes in driving skills was phrased differently for oldersubjects than for the mid-age controls. Older subjects and their collaterals were asked toindicate, for nine different driving situations, whether the subject drove better, the same ,or not as well as when they were middle-aged (between ages 35 and 45). The mid-agecontrols and their collaterals were asked to rate present driving ability relative to the firstsix years the subject had their licence. The issue of interest was whether there had beenan increase in driving problems for the older drivers. In accordance with this focus, ascore of 1 point was assigned to the not as well response while better and the same wereassigned a score of 0. As with the other multi-item questions the final score for thequestion was expressed a proportion of the maximum possible score.For the question related to concerns about driving, the respondents were asked toindicate in yes / no format whether they were concerned about the subjects' driving inrelation to six problem behaviour and situations. The content of the items was quitediverse, ranging from concerns about the loss of attention and concentration ability toconcerns about the subject injuring a pedestrian or hitting another car. A response of yes219was scored as 1 point whereas a response of no, indicating no concern about the situation,was scored as 0.Subjects and collaterals were asked to indicate whether the subject frequently ,sometimes, seldom, or never experienced difficultly carrying out 15 different drivingmanoeuvres such as changing lanes, parking, entering a freeway etc. The responses werescored such that those with more frequent difficulties would have higher total scores. Thefrequently response was assigned a value of 3, sometimes a value of 2, seldom a value of1, and never was assigned a value of 0.The question on interfering factors was also rated using the response categories offrequently , sometimes, seldom, or never. For this question respondents were presentedwith ten items involving a range of subject characteristics such as poor vision , memory,weaknesS or stiffness, and a couple of driving conditions such as rain and glare. For eachitem the respondent was asked to indicate the frequency with which the characteristic orcondition significantly interfered with the subject's driving ability. The values assigned tothe response options were the same as for the driving manoeuvre difficulties discussedabove.II. Data Reduction Method for Driving Problem Questions.The pattern of relationships between subject responses and collateral responses tothe driving problem questions were quite similar across the five questions. Table C-1 ofthis appendix presents subject and collateral means for the individual driving problemquestions. The relatively consistent pattern across the different questions suggested that itwas reasonable to reduce the data in order to derive an overall index of driving problems220that combined the five questions. An overall score for driving problems was desirable inthe context of simplifying the exploration of strategical level behaviour.Several methods of data reduction were considered ranging from simple addition ofthe five driving problem scores to factor analysis. Principal component analysis (PCA)was selected as the most appropriate procedure for generating weights that would allowfor computation of an optimal linear combination of the five questions. The followingfive paragraphs describe the specifics of the data reduction procedures and their rationale.1. The collaterals ratings on the five driving problem questions constituted theinput data to the PCA. These scores were selected because collateral ratings were likelyto be the most reliable data. It was decided that the inclusion of demented subjectsratings could degrade the analysis and result in weights that were less optimal than thosederived from the best available data.2. The scores entered into the PCA were mean deviated using the appropriatemean, i.e., the mean associated with the group from which the respondent was drawn.For example, the driving fault scores for respondents who were collaterals for the normalelderly were mean deviated using the mean for that groups ratings on the driving faultsquestion. Similarly the scores for respondents who were collaterals for the demented weremean deviated using the mean from that group.3. The analysis was run under SPSS-X Factor with PCA as the designatedextraction technique and a Criteria command specifying that only one component beextracted.4. The factor 1 matrix yielded the following coefficients for the five driving221problem questions: Faults .809, Changes .750, Concerns .385, Difficulties .832, andInterferences .846. The first principal component accounted for 55.5% of the variance inthe data set.5. The factor 1 coefficients were divided by the average standard deviation for thevariable and applied as weights to respondents scores on the five driving questions. Theresulting value was a summary score that was the optimal linear combination of thedriving problem questions. The group means for the summary index of driving problemsare presented in Table C-2.222Appendix C: Table C-1Driving Problem Questions: Scores and Repeated Measures ANOVA'sGroupsNormal^Mid-ageDemented^Elderly^ControlsDriving Faults: Participant ReportMa^ 8.63^13.23^15.38SD 9.13^8.72^8.50n^ 15 16 13Driving Faults: Collateral ReportM^ 18.06^4.78^13.12SD 17.55^5.77^9.35n^ 15 16 13Repeated Measures Analysis of VarianceSource^df^MS^F^Prob.Group^2^236.4^1.99^.154Rater 1 3.9^0.04^.840Group by Rater^2^633.5^6.35^.004aScores are expressed as percentages of the maximum possiblefor the question. Larger scores indicate higher problemratings.(table continued)223Appendix C. Table C-1 (continued)Driving Problem Questions: Scores and Repeated Measures ANOVA'sGroupsNormal^Mid-ageDemented^Elderly^ControlsDeterioration in Driving Ability: Participant ReportMa^25.12^13.19^20.00SD 21.81^21.74^32.62n^ 12 12 10Deterioration in Driving Ability: Collateral ReportM^ 40.61^20.60^10.00SD 32.93^22.17^14.29n^ 12 12 10Repeated Measures Analysis of VarianceSource^df^MS^F^Prob.Group^2^2218.2^3.25^.053Rater 1 312.0^0.54^.469Group by Rater^2^910.0^1.57^.225aScores are expressed as percentages of the maximum possiblefor the question. Larger scores indicate higher problemratings.^ (table continued)224Appendix C: Table C-1 (continued)Driving Problem Questions: Scores and Repeated Measures ANOVA'sGroupsNormal^Mid-ageDemented^Elderly^ControlsConcerns about Driving Ability: Participant ReportMa^30.62^37.71^35.89SD 23.86^29.98^21.35n^ 16 16 13Concerns about Driving Ability: Collateral ReportM^41.87^28.12^14.10SD 24.40^33.18^24.39n^ 16 16 13Repeated Measures Analysis of VarianceSource^df^MS^F^Prob.Group^2^937.9^1.12^.335Rater 1 1003.2^1.69^.201Group by Rater^2^2051.4^3.46^.041aScores are expressed as percentages of the maximum possiblefor the question. Larger scores indicate higher problemratings.(table continued)225Appendix C: Table C-1 (continued)Driving Problem Questions: Scores and Repeated Measures ANOVA'sGroupsNormal^Mid-ageDemented^Elderly ControlsDifficulty with Driving Manoeuvres: Participant ReportMa^ 11.03^13.19^10.42SD 6.05^9.79^13.25n^ 15 15 13Difficulty with Driving Manoeuvres: Collateral ReportM^17.98^8.92^5.30SD '^16.05^14.24^5.63n 15 15 13Repeated Measures Analysis of VarianceSource^df^MS^F^Prob.Group^2^308.7^.161^.161Rater 1 13.5^0.12^.726Group by Rater^2^529.5^3.04^.0594Scores are expressed as percentages of the maximum possiblefor the question. Larger scores indicate higher problemratings.(table continued)226Appendix C: Table C-1 (continued)Driving Problem Questions: Scores and Repeated Measures ANOVA'sGroupsNormal^Mid-ageDemented^Elderly ControlsInterferences with Driving Ability: Participant ReportMa 18.69 17.38 10.05SD 18.06 10.87 10.07n 14 16 13Interferences with Driving Ability: Collateral Report20.32 6.94 4.90SD 13.85 8.06 6.61n 14 16 13Repeated Measures Analysis of VarianceSource df MS F Prob.Group 2 999.48 7.95 .001Rater 1 462.32 2.87 .098Group by Rater 2 272.38 1.69 .198aScores are expressed as percentages of the maximum possiblefor the question. Larger scores indicate higher problemratings.227Appendix C: Table C-2Summary Scores for Linear Combination of Driving Problem Questions GroupsNormal^Mid-ageDemented^Elderly ControlsSummary Score for Driving Problems: Participant ReportMa^ 51.57^49.66^49.38SD 9.64^7.36^7.69n^ 10 11 10Summary Score for Driving Problems: Collateral ReportM^ 61.23^45.43^43.19SD 13.46^6.65^3.79n^ 10 11 10Repeated Measures Analysis of VarianceSource^df^MS^F^Prob.Group^2^613.9^8.78^.001Rater 1 1.5^0.01^.901Group by Rater^2^376.6^4.89^.015aScores are expressed in T score metric (mean of 50, SD of 10across the three groups). Higher scores indicate higherproblem ratings.228Appendix DInformation Associated with Selected Driving Behaviour Measures Table of ContentsMotor Vehicle Branch Road Test Form^ 229Figure D-1: Computerized Driving Assessment Module (CDAM)Output for Steering Task^ 230Figure D-2: Diagram of Boundary Bay Driving Track 231REGULATION KNOWLEDGEJ.^1. Approach too Fast 5STOP SIGNS^2. Stops too Far Ahead/BackR.R. XING 53. Blocks Crosswalk 54. Fails to Come to Complete Stop 105. Fails to Observe Before Starting 106. R.R. Xing Stop Procedure 57. Violation FK.^1. Fails to Anticipate 5TRAFFIC^2. Speeds Up to Make GreenUGHTS103. Speeds Up to Make Amber 104. Rift 10 Stop on Amber 105. No Caution - Flashing Amber/Green 106. Stops on Green - Obstructing 57. Starts Before Green 108. Falls to come to a Complete Stop 109. Violation - Red/Flashing Red FL^1. Uncertain - Take/Yield 5RIGHT^2. AssumesOF WAY103. Stops Incorrectly - To Grant 54. Fails to Yield to Veri/Ped 105. Violation/Dangerous Action FM.^t Slow - Not Obstructing 2SPEED^2. Tendency to Hurry 53. Slow - Obstructing 54. Too Fast - Ability/Conditions '55. Uneven Speed Control 56. Violation/Dangerous Action^• FN.^1. Too Far Left/Right 5 -ROAD OR^2. Late ChoosingLANE USE 53. Wrong Lane 104. Straddles 105. Fails to Observe Blind Spot 106. Change/Unsafe Location 107. Dangerous Action F0.^1. Stops Unnecessarily 5INTER-^2. Fails to Observe/LateSECTIONS103. Depends on Others for Safety 104. Dangerous Action FP.^1. Uncertain l5/-PASSING^2. Fails to Sou^orn 23. Fails^eck Rear 104. s to Yield 10. Cuts in too Ouicy.ia,,ci / F-^6. Violations/D^erous Action FO. 1. Too Close 5FOLLOW-ING^2. Dangerous Action FR. .^1.^ruck/Tractor^us 5AIR-BRAI Tra i ler IMPROPER^"//,../ 5USE^3. Parki 5TOTAL DEMERITSTest Chsoueltheahon Class 2. 35 4 mole Man 30 Dements. or F) Failure onany Maneuver Class 5 more than 40 Dements. or (F) Failure on any Manoeuver.(:)11.0UnoAJInrY,$dl ,On Nip^55 , .:, in or r•arn.,e ,105510555F55F22510555F25555F25555F229VEHICLE OSLO FOR TEST. YEAR^STYLE^LIG G V WProvince of^Ministry ofBritish Columbia Transportationand HighwaysMOTOR VEHICLE DEPARTMENTROAD TEST RESULTS-re^ First NamesLic No I I I 1 I I I 1 Road TeM^Signature ofExaminee^QUALIFICATIONS: Pre-Trip - YES ^ NO ^ Air Test - YES ^ NO ^ Road Test - YES ^ NO ^AixidwanYing(NNWIF BUS. - ADULT SEATING CAPACITYAnalysis: Road Test Failure^ I. ACCIDENT^^ 2. ACCUMULATED DEMERITS^ 3. OtSCONTINUE0 TEST^^ 4. ADMITS CANNOT PERFORMi.".1^1 I I I 1 IC.SHIFTINGA.STARTINGSTOPPINGVEHICLE CONTROL1. Lacks Knowledge/Equipment2. Rides Clutch/Brake 3. Fails to Cover Brake (Auto) 4. Emergency Brake Set/Stalls.5. Races Engine 6. Unnecessarily Fast Start 7. Fails to Check Conditions/Blind Spot ^108. Dangerous Action 1. Uncertain - Forward/Backward ^52. Improper Posture - Hands/Arms ^23. Unnecessary One Arm bt6braig r '4.Wanders ^105. Dangerous Action1. Uncertain2. Shifts^'3.4. Shifts Too Soon/Late5. Improper Use6. I/78. Coasts - NeutraWClutch in9. Cannot Shift Gears1. Too Soon/Late 2. Improper/Poor3. Wrong Signal Given 4. No Signal Given 1. Uses Mirror Only (Classes 4 & 5) 2. Poor Observation - Before/While 3. Speed Excessive/Inconsistent 4. Faits to Sound Horn 5. Lacks Skill - Steering/Eiraking 6. Dangerous Action 1. Incorrect Use - Clutch/Accelerator2. Rolls Back 3. Loses Control 1. Improper Position • Vehicte/VVheels2. Bumps Curb 3. Fails to Set Brake/Gear 4. Poor Observation - Before/While 5. Unable to Park - Each Attempt6. Strikes Front or Rear Car 7. Climbs Curb 8. Dangerous Action1. Brakes in Turn2. Tends to Turn Wide 3. Ends in Wrong Lane 4. Cuts Corner 5. Fails to Observe - Before Turn6. Dangerous Action 1. Brakes in Turn2. Tends to Turn Wide3. Ends in Wrong Lane 4. Cuts Corner 5. Fails to Observe - Before Turn •6. Dangerous ActionH.RIGHTTURNS4SAWNGE.BACKINGF.STARTINGON HILLG.PARKINGB.STEERING25522555F255 5F25mV 2013 loom) W-529a) 0^20^40^60^80^100^120230LRP L 09/29/90b) -L=,111011•■■ -1 iP L 09/29/90Total area under the steering wheel deviation curve is 619.2000P L^09/29/90Figure D-1: Computerized Driving Assessment Module (CDAM) Output for SteeringTask. Panel (a) shows the actual steering wheel position during the task. Panel (b) showsthe ideal steering wheel position for perfect tracking of the changing LED stimuli. Panel(d) shows the deviation from the ideal steering wheel position.2C grid.Sccle (IL)01111111-11_50^1000231Figure D-2: Diagram of off-road test circuit at Boundary Bay Driving Track. (1: visualbarrier; 2: Traffic signal; 3: Directional arrow; 4: styrofoam vehicle for hazardavoidance task).Appendix EConsent Form232CONSENT FORM AND STUDY DESCRIPTIONDRIVING AND AGING STUDYInvestigators:B. L. Beattie, M.D, FRCPCH. Tuokko, Ph.D.University Hospital - UBC SiteResearch Associate:Karen Tallman, Ph.D. CandidateUniversity Hospital, UBC SiteOVERVIEW AND INTRODUCTIONThis Consent Form is intended to provide a detailed description of theDriving and Aging Study. This form will be reviewed with eachparticipant by a member of the research team before it is signed.Please feel free to ask for clarification if any portion of thedescription is unclear. Each participant will be given a copy of thisform for their own reference.We ,are very grateful that you are considering taking part in ourproject. We recognize that the testing we wish to do is quiteextensive and time consuming. We will make every attempt to make yourparticipation in the Driving and Aging Study pleasant and interesting.We believe this project has the potential to provide valuableinformation which will benefit older drivers in the future. Atpresent in British Columbia, drivers are required to have a medicalexamination to determine their ability to continue to drive when theyreach age 75, again at age 80, age 82 and so on. Recently, doctors inour province and across Canada and the U.S. have emphasized the needfor research on the aging driver in order to provide the grounds forfair and objective decision making. We believe the Driving and AgingStudy will make a valuable contribution to scientific knowledge aboutthe driving behaviour of older individuals. It will assist in theongoing process of developing a sound, scientifically established,basis for public policy in the decades to come.The specific goal of this research project is to address importantissues related to objective measurement of the driving performance ofolder individuals with, and without memory complaints. This is thefirst study of its kind. Because so little is known about the drivinghabits of older individuals, the project has been designed to collecta great deal of information about each participant. This study hasbeen funded by Health and Welfare, Canada, and has the active supportof the Motor Vehicle Branch of British Columbia, the InsuranceCorporation of British Columbia, the Justice Institute and PacificTraffic Education Centre, G. F. Strong Rehabilitation Centre, and theUBC Accident Research Team.233-2--From the outset, we wish to make it very clear that all results ofthis study will be entirely confidential. YOUR DRIVER'S LICENSE WILLNOT BE THREATENED BY THE OUTCOME OF TESTING. All the test sheets forthis study will be marked with a number code insuring theconfidentiality of the findings. Files will be kept in lockedcabinets and will be seen only by the study staff. When theresearch report is written, there will be no identification ofindividuals and your anonymity will be strictly protected.All test procedures are safe and your comfort is very important to us.We wish to collect measures of performance on a wide variety ofvisual-perceptual tests, measures of memory and problem solvingskills, questionnaires requesting information from you and a friend orrelative about your driving habits and attitudes, and a number ofmeasures of performance on simulated and real driving tasks. As willbe described below, you will take part in some road testing. Thistesting will be done in a car with two sets of controls (two steeringwheels, and two sets of gas and brake pedals). This is to ensure yoursafety during the actual driver testing portions of the study. Wehave hired professional road testers and driver trainers with specialinterests in the older driver to ride with you and to handle the dualcontrols in the testing car. You will be given time to becomeaccustomed to our testing vehicle and every effort will be made tomake you feel comfortable and relaxed during all testing proceduret.We realize how genuinely important the entitlement to drive is formost people in our society. Because of this importance, we recognizethat a study such as ours has the potential to make participants feelself-conscious and apprehensive about their performance on our tests.You may worry abdut being observed as you drive despite the fact thatyou have been reassured that your licence is not in jeopardy. We arestriving in the Driving and Aging Study to create an atmosphere freeof judgement. We wish to make it clear to those who take part in ourstudy that we are extremely grateful for your generosity in takingpart with us.At the end of the testing, you will receive $50.00. After completingthe Road Test and Visual Screening, you will receive a copy of yourtest results. As stated above, there are no repercussions to you oryour driving licence as a result of taking part in this study. Ifyour performance on the road test indicates that you are havingdifficulty with driving, we will recommend that you seek your familydoctor's advice about continuing to drive.234-3-TEST PROCEDURES AND TESTING SCHEDULEThe collection of data for this study will require three separatetesting sessions on three different days. Please note that Day 2 andDay 3 are somewhat dependent on the weather and your particulartesting schedule may be affected by poor weather conditions. We willkeep in close contact with you over the time you are involved in thestudy in order to keep you informed of any scheduling changes thatmight occur.PLEASE NOTE that some tests we use repeat the same questions. Weapologize for the repetition of items. We have chosen to retain theoriginal form of already established tests in order to allowcomparison of our findings with the findings of other researchers inthe field.Details of each of the testing days are outlined in the followingpages.Overview of Study Procedures: 235DAY ONEDAY TWODAY THREEUNIVERSITY HOSPITAL, UBC SITE. Date: ^Sign Consent Form. Complete 'Perception MeasuresTest Battery' and 'Memory and Language Test Battery'.Complete the Driving Information Interview. Assessdentof visual acuity and peripheral visual fields.Interview with relative or close friend about drivinghabits and daily functioning of participant.G.F. STRONG REHABILITATION CENTRE. Date:Road test in dual-control car.Visual Perceptual Test.Assessment of everyday functioning.Assessment on driving simulator.PACIFIC TRAFFIC EDUCATION CENTRE. Date:Road testing in dual-control car on driving circuit.Questionnaires on risk perception and assessmentof personal driving skills.OUR TELEPHONE NUMBER IS 228-7926e OUTSIDE OFFICE HOURS CALL 874-8498-4-DAY ONE: UNIVERSITY HOSPITAL UBC SITEThe first day of your participation in the study will take place atUniversity Hospital, UBC Site. Testing will take place in Roam Gof the Purdy Pavilion. You are asked to come with a relative or closefriend who is willing to provide us with some 'collateral' informationAbout your driving habits and everyday functioning. Your first daybegins at 9:00 a.m. and you will be finished at about 3:45 p.m. Therewill be a 45 minute luncheon break and a 15 minute midmorningrelaxation period. Additional breaks can be taken as desired by theparticipant.The measures contained in the 'Memory and Language Test Battery' andthe 'Perceptual Measures Test Battery' are standard psychologicaltests which assess a wide variety of functions which may relate todifferent aspects of the driving task. When we have collected thesemeasures from a large group of participants (approximately 150), wehope to be able to investigate the relationships between psychologicalmeasures of various skills and actual on-road performance of thedriving task.Approximate schedule for Day One (Actual Date^)9:00^9:30 Consent Form & Medical Information Form9:30 - 11:15 Either the 'Perceptual Measures Test Battery'or the 'Memory and Language Test Battery'11:15 -- 11:30 Rest Break11:30 - 12:30 Interview about driving habits & attitudesor Visual Field testing12:30 - 1:15 Luncheon Break1:15 - 2:45 Either the 'Perceptual Measures Test Battery'or the 'Memory and Language Test Battery'(which ever was not done in the am session)2:45 - 3:45 Interview about driving habits & attitudesor Visual Field testingNote: We wish to do an interview with a relative or closefriend about the participant's driving habits andattitudes, and selected aspects of daily functioning.This interview is expected to take approximately 30 to45 minutes, and will be scheduled to be as convenient4 as possible for all parties.236-5-^ 237DAY TWO: G. F. STRONG REHABILITATION CENTREThe second day of your participation in the study will usually be on aSaturday. This portion of the study will take place at G.F. StrongRehabilitation Centre where a computerized driving simulator ishoused. You will be given a map and explicit instructions about howto get to G. F. Strong. You will be involved in approximately 2 to 21/2 hours of testing on the day you come to G.F. Strong.The Road Test will consist of a driving assessment done in the samefashion as the testing done by the Motor Vehicle Branch. The testingwill be done in the area around G.F. Strong Centre and will beconducted in a dual control car. You will be given a copy of theresults of the road test. You will also be able to authorize Mr.Vosper to provide a second copy of these results for your relative orsignificant other. The Visual Perceptual Test is very similar to theVisual Field Testing that takes place on Day One, except that theequipment at G.F. Strong assesses reaction time rather than visualsensitivity. Following the Visual Perceptual Test, a measure offunctioning on several common everyday tasks will be done. The tasksinclude things such as operating a telephone, identifying money andcalculating the cost of various purchases, and a variety of otheractivities of daily liVing. The Computerized Driving AssessmentModule is a driving simulator constructed from parts of a car withcomputer controlled instructions and data collection. The taskspresented on the simulator are simple measures of 'steering','maintaining speed' and 'stopping'. There will be scheduled 15minute breaks between the tests and additional breaks may be taken asdesired.Approximate schedule for Day Two (Actual Date:TimeArrive Front Entrance of G. F. Strong Rehab. CentreRoad Test (45 minutes allotted)In dual control car with Mr. D. Vosper.Rest BreakVisual Perceptual Test & Functional Rating Scale.(60 minutes allotted)Rest BreakComputerized Driving Assessment Module(45 Minutes alloted)If you need to contact us on Friday P.M. or Saturday A.M., please trythe following numbers: 1) 228-7926 (leave message on answeringmachine) or 2) 874-8498 (home telephone of project staff member).-6-DAY THREE: PACIFIC TRAFFIC EDUCATION CENTRE, BOUNDARY BAY.Actual Date:Day Three will usually take place on a Sunday. The testing that takesplace on Day Three is the most likely of all our test procedures to bedisrupted by bad weather, particularly by icy conditions.The testing on Day Three will take place at the Pacific TrafficEducation Centre (PTEC) which is located on a closed airport runway atthe Boundary Bay Airport. PTEC provides training courses in drivingskills for Police, Ambulance, and Fire Department personnel. Mr. AlLund is the Programme Director for PTEC and will be assisting in ourDay Three testing as the 'other' driver in the dual control car.Also, Mr. Lund will be able to transport up to 6 people in a PTECmini-bus from the Vancouver area to the Boundary Bay Site. It is ourintention to arrange for approximately 6 participants to go to thePTEC track for testing on Sundays. Each participant will spendapproximately 20 minutes being tested on the driving circuit and willhave some questionnaires to fill out while other participants arebeing tested. Luncheon will be provided. For those participants whoare unable to go on Sundays, occasional testing periods during thework week may be arranged.The testing that will be carried out at PTEC is probably the mostexciting and innovative of our testing procedures. We have designed adriving circuit on the abandoned runway which is intended to providemeasures of steering and breaking control in response to 'surprise,'stop signs and obstacles. Testing will be done in the samedual-control vehicle that was used for the Road Test conducted by Mr.Vosper on Day Two.For the testing at PTEC, the car will be outfitted with a video camerato record the car's position on the track and with a computer whichwill record braking and accelerating times. You will not be asked todo anything that is dangerous to you or the car. You will generallybe travelling at very low speeds (approximately 20 km or 15 miles perhour). Because there is no danger of harm coming to you or others onthis abandoned runway, it is possible to test vehicle handling skillsunder 'surprise' conditions. The 'surprise' tasks will be unexpectedonly in the sense that the participant will not know when the stopsignor obstacle will appear. Participants will also be asked to makejudgements about distances and about their ability to perform variousdriving maneuvers. It is expected that the testing at PTEC will bechallenging but also fun. The track might be compared to a Go-Carttrack and the testing is intended to be somewhat playful. As with alltesting in this study, the participant is always free to withdraw atanytime without any personal consequence.If you need to contact us outside office hours about your testingsession at Boundary Bay, please call Karen at 874-8498.238239-7-RISKS, INCONVENIENCES AND DISCOMFORTS ASSOCIATED WITH PARTICIPATION INTHE STUDY:^Participation in the study will be time-consuming for you, but every effort will be made toarrange appointments at a convenient time.There are no discomforts anticipated; in factwe expect that most people will find participationquite interesting. In order to ensure your safety,during road testing and driving at the Boundary Baydriving track, all actual driving will be done in adual-control automobile with a professional drivingtrainer or examiner.You will receive $50.00 for participating in this study and apersonalized confidential evaluation of your driving skills.If for any reason you wish to withdraw from the study, you are free dodo so at any time without further comment. Also, we are happy at anytime to explain any aspect of our procedures.NAME:DOB:I agree to participate in this clinical research study on drivingperformance to be done under the auspices of the University Hospital,UBC Site under the direction of Dr. B. L. Beattie and Dr. HollyTuokko.I have been informed of the types of tasks that I will be expected to'do and of the time needed to take part in the study.I have been assured that all the information about me will be keptconfidential and that my identity as a participant in this study willnot be revealed. Documentation regarding me will be identified by anassigned study subject number for the purposes of data analysis.I understand that I may decline to enter the study and that I canwithdraw from the study at any time without any consequences to me.I understand that any enquiries I have about the study will beanswered.The purpose of this study and the inconveniences of the study havebeen presented to me in both verbal and written forms and I understandthese. explanations. I acknowledge that I have received a copy of theconsent form and the explanation of the study.Participant Signature^DateSignature of Relative/Significant Other^DateInvestigator

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0086358/manifest

Comment

Related Items