Section 5: Cerebrovascular Diseases

5.2. Cerebrovascular Accidents Prevalence

Chronic Impairment

There are a number of ways of determining fitness-to-drive: a) determination of relative risk through the use of crash statistics¹ , or b) the prediction of fitness-to-drive through the use of medical evaluations, neuropsychological testing, functional assessments (typically carried out by rehabilitation specialists), driving simulators, or actual on-road tests (specialized or regular). Each of the methodologies has their strengths and limitations. For a review, please see Ball and Owsley (1991) and Hakamies-Blomqvist (1998).

Despite the fact that cerebrovascular accidents (CVAs) are a leading cause of disability, there is a paucity of research on the relationship between the chronic effects of CVAs and motor vehicle crashes. That which is available comes from a study in Utah (Diller et al., 1998) (see Section 2.1 a., for details of the study) and from the State of Washington (Haselkorn, Mueller, and Rivara, 1998). Relevant to this discussion are results of the data available from the Diller et al. study for the neurological conditions category. Individuals with neurological conditions such as stroke, head injuries, cerebral palsy, multiple sclerosis, Parkinson's Disease, and progressive conditions (e.g., muscular atrophies, dystrophy, myasthenia gravis) and other brain and spinal cord diseases were included. Epilepsy was considered in a separate category. The sample consisted of 3,007 unrestricted drivers and 771 restricted drivers. Results revealed that drivers with neurologic impairments had a significantly increased risk of motor vehicle crash. For unrestricted drivers, the relative risk was 4.21 (CI = 3.86 - 4.60); for restricted drivers the relative risk for all crashes was 2.18 (CI = 1.72 - 2.78). Unfortunately, data on the relative risk of drivers with specific impairments within the neurological category are not available. Therefore, the relative risk of drivers with CVAs alone is not available.

Haselkorn et al. (1998) conducted a retrospective study comparing the driving records of four cohorts hospitalized with CVA, traumatic brain injury, fractured extremities, and appendicitis with driving records of age-, gender-, and ZIP code-matched non-hospitalized controls. Hospitalized records for 1992 for the cohorts were linked with Department of Licensing records for 1991 to 1993. Measures included the occurrence of crashes or citations for moving violations. Linkages identified 1,917 patients with CVA, representing 39 percent of patients hospitalized with CVA. Data from individuals with more than one disease classification were excluded from the study, resulting in a sample of 1,910 patients with CVA, with the majority (73 percent) of the CVA sample 60 years of age and older. Results of estimates of relative risk (RR) for crash revealed that the CVA sample did not have an elevated risk relative to other comparison group (RR = 0.8, CI = 0.6-1.4). The results suggest, therefore, that individuals post-CVA do not have an increased risk of motor vehicle crashes during the 12 months following hospitalization for the event.

However, a number of methodological limitations are noted. First, driving exposure was not taken into consideration. It is likely that individuals experiencing a stroke will drive less frequently (particularly in the first several months following their stroke) compared to the non-hospitalized comparison group. If driving exposure is indeed less, then the results underestimate the risk of crash. Second, based on available data, the authors were unable to determine the severity of brain damage. Given that the CVA sample was representative of only 39 percent of patients hospitalized with CVAs, it is unknown if only those with mild impairments were included. Such an occurrence also would underestimate the risk of crashes. Finally, the data included only state-recorded crashes, data susceptible to errors of omission.

The majority of studies reporting on assessments of fitness-to-drive following a CVA have employed neuropsychological testing and some type of on-road assessment. A summary of the studies is presented in Table 11.

In general, results of fitness-to-drive assessments of individuals who have suffered a cerebrovascular reveal that, of those presenting for assessment, approximately 50 percent or more fail the assessment. Methodological differences, however, make comparisons across studies difficult. For example, some studies include subjects from different diagnostic categories; when similar diagnostic categories are used (e.g., left and right CVA), the exact location and extent of damage often is unknown, and testing time post-stroke frequently differs or is unknown. In addition, there often is considerable variability in neuropsychological assessment batteries and criteria used for subject classification based on neuropsychological performance. For example, often batteries are chosen to include tests that are sensitive to areas of impairment: language-oriented functions for individuals with left-hemisphere damage and visuospatial functions in individuals with right-hemisphere damage. Often neglected, however, is attention to the generalized cognitive deficits that may result as a consequence of a stroke. Finally, on-road assessments differ dramatically in terms of assessment procedures and criteria used for scoring. In most cases, criteria are not well defined or are lacking. As noted by Springle, Morris, Nowachek, and Karg (1995), criteria used for evaluation of results are based on subjective criteria or no criteria. What is needed is a standardized driving evaluation procedure that has been shown, through research, to be valid and reliable.

As noted earlier, attention to generalized cognitive functioning often is neglected in assessment of individuals post-CVA. A number of studies have investigated the effect of a CVA on generalized cognitive functioning. Horn and Reitan (1990) compared the performance, based on an extensive battery of neuropsychological tests, of 60 patients with lateralized or diffuse cerebrovascular lesions to 20 controls. Results indicated that the group with cerebrovascular lesions performed significantly worse than controls on measures of cognitive functioning. Important to this discussion is the finding that significant neuropsychological impairments were noted in the group with cerebrovascular lesions that extended beyond the expected lateralized dysfunctions or selected impairments associated with the damaged hemisphere. Tatemichi, Desmond, Stern, Paik, and Bagiella (1994) examined cognitive function in 227 patients three months following admission for ischemic stroke. Results were compared to 240 stroke-free controls. Like Horn and Reitan, the authors were interested in focusing on general cognitive impairments rather than on describing the circumscribed deficits associated with focal impairments. Compared to controls, impairments were noted on all 17 neuropsychological items. Cognitive impairment, defined as failure on any four or more items, occurred in 35 percent of stroke patients and four percent of controls. Cognitive domains most likely to be impaired were memory, orientation, language, and attention. Finally, results from the Copenhagen Stroke Study (Pederson, Jorgensen, Nakayama, Raaschou, and Olsen, 1996) indicated that, of those stroke patients completing the Mini Mental Status Examination (MMSE) one week after stroke onset, 42 percent scored below the cut-off level of 24. Further, MMSE scores one week post-stroke were significantly correlated with functional assessment at discharge. Results of these studies underscore the importance of assessing for generalized cognitive functioning in individuals post-stroke.

A number of investigators have recognized the need for a standardized procedure for assessing individuals following a cerebrovascular accident (Engum, Pendergrass, Cron, Lambert, and Hulse 1988a; Galski, Bruno, and Ehle, 1992; Korner-Bitensky, Sofer, Kaizer, Gelinas, and Talbot, 1994; Nouri, Tinson, and Lincoln, 1987). Results of studies aimed at enhancing the precision and rigor of fitness-to-drive assessments in stroke patients are summarized in Table 12. The approaches taken by Engum et al. (1988a, 1988b, 1989, 1990) and Galski et al. (1990, 1992, 1993, 1996) are the most extensive and programmatic available. The approaches are ones that deserve special attention because they are likely to be instructive for future investigations in this area as well as in other areas.

Engum et al. (1988a) developed the Cognitive Behavioral Driver's Inventory (CBDI) specifically to provide rehabilitation professionals with a standardized battery for determining fitness-to-drive in individuals with a brain-injury (see Engum et al., 1988a for a description of the battery). Briefly, the battery consists of 10 tests measuring attention, concentration, rapid decision making, stimulus discrimination/response differentiation, visual scanning and acuity, and attention shifting. The criterion measure is on-road performance. The initial investigation provided data regarding internal consistency of the items in the battery (Cronbach's alpha = 0.949) and preliminary estimates of validity based on on-road driving performance. The battery was validated in 1989 on a sample of 175 brain-injury patients (Engum, Lambert, Scott, Pendergrass, and Womac, 1989). The correlation between outcomes on the CBDI (pass/fail) and road test outcome (pass/fail) was significant (r = 0.81,
p < .0001). Overall, of the 42 patients receiving a favorable pass on the CBDI results, 40 passed the road test. However, 7 of the 39 patients receiving a fail on the CBDI results passed the road test. The sensitivity of the CBDI was further evaluated in 1990 (Engum, Lambert, and Scott, 1990). Results of that investigation show the CBDI to be highly sensitive in discriminating between healthy controls, brain-injured patients passing the road test, and brain-injured patients failing the road test (see Table 12 for amore detailed description).

The approach taken by Galski et al. (1990, 1992, 1993, 1997) was to first critically assess evaluations developed at their facility to determine fitness-to-drive (Galski, Ehle, and Bruno, 1990). Results of that investigation revealed a lack of internal and predictive validity of the driver evaluation. A model was then developed for evaluating fitness-to-drive (Galski, Bruno, and Ehle, 1992). The utility of the model was determined in terms of amount of variance of the criterion (on-road evaluation) explained by the predictor variables (neuropsychological tests, driving simulator, and parking lot driving scores). The results of that investigation revealed that 93 percent of the on-road assessment (in traffic) was explained by three predictor variables. In a follow-up investigation Galski, Bruno, and Ehle (1993), using a larger sample size, determined the effectiveness of the evaluation methods developed in the 1992 investigation by discriminant function analysis and measurements of sensitivity. The methods were found to be highly sensitive in predicting outcomes. In their 1997 study, Galski, Ehle, and Williams explored the dimensions underlying their pre-driver assessment (neuropsychological battery) and measures from a driving simulator. Results of factor analysis indicated the presence of five underlying factors (higher order visuospatial abilities, basic visual recognition and responding, anticipatory braking, defensive steering, and attentional measures), accounting for 66 percent of the variance. As noted by the authors, the factors can be used as a basis for understanding what is measured in off-road evaluations and in determining fitness-to-drive following cerebral injury. Details of the approach are provided in Table 12.

¹ As noted above, disability laws may limit the use of risk analyses by State licensing authorities as a basis for making fitness-to-drive decisions with regard to categories of applicants with specific medical conditions.