| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
a The Glennan Center for Geriatrics and Gerontology, Eastern Virginia Medical School, Norfolk
Correspondence: Barbara Freund, PhD, The Glennan Center for Geriatrics and Gerontology, Eastern Virginia Medical School, 825 Fairfax Ave #201, Norfolk, VA 23507. E-mail: freundbm{at}evms.edu.
Decision Editor: Laurence G. Branch, PhD
 |
Abstract |
|---|
 TOP Abstract Literature ReviewHypothesesMethodsResultsDiscussionReferences |
|---|
Key Words: Cognition • Aging • ADL and IADL impairments
The number and proportion of older people who drive increased dramatically during the past decades (Lundberg, Hakamies-Blomqvist, Almkvist, and Johansson 1998
), and this trend is expected to continue. Indeed, the proportion of persons aged 65 who will be driving is estimated to double between now and 2020 (U.S. Department of Transportation 1997). However, the oldest drivers, those 75 years old and older, crash at a rate second only to the youngest drivers, those up to 24 years old (O'Neill et al. 1992; Williams and Carsten 1989). This suggests that some older persons continue to drive even though their driving poses a serious risk to themselves and others on the road. Thus, it is essential to establish under what conditions older individuals continue to drive. Of particular concern is continued driving among the cognitively impaired, who may have difficulty interpreting environmental cues and responding appropriately. Especially persons in the early stages of cognitive decline may not recognize or may ignore the effect of their impairment on safe driving. Previous evidence relating cognitive functioning to driving behavior has yielded somewhat inconsistent results, most likely because these studies relied on diverse sample populations and failed to control for some important driver characteristics (Fox, Bowden, Bashford, and Smith 1997; Hunt, Morris, Edwards, and Wilson 1993; Lundberg et al. 1998). Our aim is to assess the impact of cognitive impairment on driving restriction and cessation, using nationally representative data from the Asset and Health Dynamics Among the Oldest Old (AHEAD) survey and controlling for other characteristics that have been associated with driving behaviors. Furthermore, we explore selected conditions (gender, alternative transportation) that may enhance or decrease cognitively impaired persons' inclination to restrict or cease driving.
|
Literature Review |
|---|
|
TOPAbstractLiterature ReviewHypothesesMethodsResultsDiscussionReferences |
|---|
Changes in driving skills and behaviors have been linked to impairment in cognition, as well as to other visuospatial and functional abilities (e.g., strength and motor ability, reaction time, certain medical conditions). Each of these, alone or in combination, may contribute to a decline in driving competency and involvement. However, some older people with these types of impairments continue driving and may be at risk for injury to themselves or others.
Cognitive Impairment and Driving Involvement
Several studies have documented effects of cognitive impairment on driving involvement. For example, miles driven are significantly reduced in individuals with cognitive impairment and Alzheimer's disease, although it is not clear at what level of cognitive impairment individuals start to restrict or cease driving. In a large North Carolina study of older drivers applying for driver's license renewal, individuals scoring in the lowest quartile on one of two cognitive tests (Trailmaking Test and the Short Blessed Cognitive Screen) were more likely to report driving fewer than 3,000 miles per year than were those scoring in the highest quartile (Stutts, Stewart, and Martell 1998). There is also evidence for driving cessation among the cognitively impaired. In a study of patients diagnosed with Alzheimer's disease (Trobe, Waller, Cook-Flannagan, Teshima, and Bieliauskas 1996), former drivers had poorer scores on measures of cognition (Mini-Mental State Examination (MMSE), Weschler Memory Scale, Blessed Dementia Scale) than did individuals still driving.
In addition to severity, duration of the impairment also affects driving involvement. Current drivers with dementia are more likely to have shorter disease duration and less overall impairment on cognitive tests. For instance, among individuals attending dementia clinics, those not driving had significantly lower scores on selected cognitive tests (MMSE, Boston Naming Test, Category Naming Test) and an illness duration more than twice that of individuals still driving (Gilley et al. 1991; Lucas-Blaustein, Filipp, Dungan, and Tune 1988).
Other studies exploring the relationship between cognitive impairment and driving involvement are gender specific, that is, they focused exclusively on one gender. In a study of older men with incident dementia, Foley and colleagues (Foley, Masaki, Ross, and White 2000) found that driving cessation increased with cognitive impairment, and they estimated that nationally, nearly 4% of male drivers 75 and older (approximately 175,000) have dementia. Diagnosis of dementia was determined by the presence of impairment in at least three of five domains including memory, language, visuospatial ability, and mood/personality. Severity was based on the Clinical Dementia Rating Scale. Among men with normal cognitive functioning, 78% were still driving compared with 62% of those who were not demented but had poor cognitive functioning. The association between cognition and driving has also been confirmed for women. Forrest, Bunker, Songer, Coben, and Cauley 1997 reported, for example, that former drivers experienced more memory problems and scored lower on MMSE than current drivers did.
These studies demonstrate that even though poor cognition promotes driving cessation, some cognitively impaired individuals continue to drive. These drivers are at greater risk for crashes as they no longer possess the cognitive ability to safely interpret and respond to the ever-changing driving environment (Ball and Owsley 1991; Ball, Owsley, and Sloane 1991; Ball, Owsley, Sloane, Roenker, and Bruni 1993; Johansson et al. 1996; Lundberg et al. 1998). However, many cognitively impaired drivers believe they are as good, if not better, drivers than other people their age or younger (Dobbs 1999). Poor cognition also enhances driving restrictions. Such restrictions, especially elimination of long highway trips (Forrest et al. 1997), may actually increase crash risk. Driving in cities involves a disproportionate number of intersections, traffic congestion, and other situations that require a high degree of information processing and decision making (Eberhard 1996).
Alternative Transportation.
Why some individuals with cognitive impairment continue to drive may be explained, in part, by the independence driving provides in meeting mobility needs. Indeed, older adults, regardless of health status, continue to rely on the automobile as the primary means of transportation (Jette and Branch 1992). However, access to other drivers may promote driving restriction or cessation. Kington, Reuben, Rogowski, and Lillard 1994 and Jette and Branch 1992 reported that drivers with declining health were less likely to drive, provided there were other drivers in the household. Driving cessation can lead to a heavy dependence on informal support systems. Taylor and Tripodes 2001 found that some older persons who stopped driving had difficulty accessing social and recreational services. This was particularly a problem when there was a lack of available licensed drivers to provide transportation services. Former drivers who did not live with at least one licensed driver reported the greatest difficulty matching availability of transportation with need and desire to travel. In addition, some caregivers missed work or gave up working to care for and provide transportation services for cognitively impaired former drivers. Consequences such as these may promote reluctance to stop driving among some cognitively impaired drivers.
Although some cognitively impaired individuals may not have driving partners, they may receive help from nondriving spouses. Especially nondriving wives are reported to assist their driving partner as copilots (Burkhardt, Berger, and McGavock 1998). Copilots were relied on by 10% of still-driving men with incident dementia (Foley et al. 2000). Although copiloting has been the subject of recent attention (Bedard, Molloy, and Lever 1998; Foley et al. 2000; Shua-Haim and Gross 1996), this strategy has not been systematically studied.
Physical Health and Driving Involvement.
In addition to cognition, driving restriction and cessation are also influenced by selected physical health problems. Functional disability contributes to changes in driving patterns, such as refraining from highway driving, driving to fewer and more essential and selective destinations (e.g., grocery store and church; Kline et al. 1992), driving fewer miles, and driving cessation (Marottoli et al. 1993). In a survey of surviving members of the New Haven Established Population for Epidemiologic Studies of the Elderly cohort, high mileage drivers tended to be active men who still worked, whereas mileage reduction was associated with increasing disability (Marottoli et al. 1993). These findings are similar to those of an analysis of data from the Iowa 65+ Rural Health Study (Colsher and Wallace 1993). Colsher and Wallace 1993 found that men with gross physical functional impairments were more likely than men without functional impairments to report driving less than they had 5 years earlier, as were men with impaired self-care ability. However, some physically impaired individuals continue to drive. Carr, Jackson, and Alguire 1990 reported that 26% of drivers referred to an outpatient geriatric assessment center required assistance with bathing and dressing. Because functional physical impairments reduce motor ability and reaction time, both integral to driving performance, physically impaired older drivers who continue to drive may be at risk for injury (Marottoli and Drickamer 1993). Declines in strength also can adversely affect driving performance (Retchin, Cox, Fox, and Irwin 1988; Stock, Light, and Douglass 1970).
Additional evidence has linked driving involvement in older people to general health and selected health problems. For example, Blomqvist and Wahlstrom 1998 study of Finnish men and women aged 70 and older who renewed or did not renew their driver's license showed that decline in health is the primary reason men did not renew their driver's license. Women did not renew their license because they were more likely to have already stopped driving prior to the license expiration. Other studies have also demonstrated that selected medical conditions (including cardiovascular disorders and Parkinson's disease) are a commonly reported reason for driving cessation. Exdrivers tended to have a higher prevalence of illnesses such as heart disease, neurological disorders, and rheumatism (Campbell, Bush, and Hale 1993; Dellinger, Sehgal, Sleet, and Barrett-Conner 2001), and women with a history of fractures, heart disease, or diabetes reduced their mileage, avoided long trips, or stopped driving altogether (Forrest et al. 1997). However, arthritis has been associated with increased driving involvement (Kington et al. 1994; Tuokko, Beattie, Tallman, and Cooper 1995). Although health conditions such as these have been cited as a primary reason for driving cessation, they have not specifically been implicated in crash risk in older drivers. The possible exception is arthritis, which may place drivers at risk for crashes (Tuokko et al. 1995).
Driving cessation or restriction has further been associated with visual impairments. Persons with no visual impairments are more likely to continue driving (Foley et al. 2000). In contrast, visually impaired older persons either cease to drive or avoid night driving, bad weather (Ball et al. 1998; Hennessy 1995; Owsley, Ball, Sloane, Roenker, and Bruni 1991), and left turns (Hennessy 1995). Continuation of driving among those with visual impairments has been linked to enhanced crash risk. Studies indicate that drivers with visual impairments are more likely to have a history of crashes (Ball et al. 1993; Ball et al. 1998; Ball & Reebok, 1994; Johnson and Keltner 1983; Lundberg et al. 1998) and are more likely to have future crashes than are older drivers without visual impairments (Ball and Owsley 1994). In fact, older drivers with a reduction of 40% or greater in the useful field of view (UFOV) are twice as likely to be involved in a crash within 3 years than are those with less than 40% reduction in UFOV (Owsley et al. 1998), primarily because of visual attentional deficits and diminished ability to divide attention.
Driving performance has further been tied to selected personal characteristics such as age (Marottoli et al. 1993; Quillian, Cox, Kovatchev, and Phillips 1999) and gender (Marottoli et al. 1993). Increasing age is associated with driving cessation (Dellinger et al. 2001; Marottoli et al. 1993) and reported as one of the most common reasons for giving up driving (Dellinger et al. 2001). Older drivers tend to be more conservative in decision making, particularly with respect to gap choice in making turns (Andrea, Fildes, and Triggs 1999; Guerrier, Manivannan, and Nair 1999). However, age is also associated with increased crash risk. Drivers 70 years old and older are more than twice as likely as middle-aged drivers to be in a fatal crash even after self-regulating driving activities (National Research Council 1988) and are involved in more crashes and fatalities per mile driven than most other age groups except the youngest (National Highway Traffic Safety Administration 1989).
The link between gender and driving involvement is shown, for example, in Stutts and colleagues 1998 study of driving exposure among older drivers. Cognitively impaired men were six to seven times more likely to report mileage reduction compared with unimpaired men, and women with low cognition were one and one half to two times more likely to report driving less than were women with higher scores. Men have been found to have more crashes than women (Lundberg et al. 1998; Sims, Owsley, Allman, Ball, and Smoot 1998; Zhang, Lindsay, Clarke, Robbins, and Mao 2000). This may be due to gender roles that enhance exposure (i.e., men drive more frequently or longer distances than women do; Gallo, Rebok, and Lesikar 1999; Marottoli et al. 1993) and that may promote more aggressive driving techniques. However, female drivers have been reported to be involved in more intersection accidents than male drivers have (Guerrier et al. 1999).
A few studies further linked driving involvement to socioeconomic status. Individuals in the lower socioeconomic status groups (e.g., those with low incomes) may not be able to afford driving or driving long distances (Jette and Branch 1992; Kington et al. 1994; Marottoli et al. 1993; Persson 1993).
Taken together, these studies suggest that both driving performance and driving cessation are contingent on cognitive and physical impairments, as well as on gender and other personal and socioeconomic characteristics. What remains to be determined is the extent to which individuals with different levels of cognitive impairment continue or stop driving and the conditions under which their driving involvement is altered.
|
Hypotheses |
|---|
|
TOPAbstractLiterature ReviewHypothesesMethodsResultsDiscussionReferences |
|---|
|
Methods |
|---|
|
TOPAbstractLiterature ReviewHypothesesMethodsResultsDiscussionReferences |
|---|
). Identification of households with persons under age 80 was based exclusively on an area-probability approach, whereas a dual-frame approach was used for households with persons born prior to 1914. One half of these persons were identified through area-probability sampling and the other half from Medicare files. However, multistage probability sampling was used in both cases. Interviews were conducted with one age-eligible respondent from the identified households and, if married, their spouses (regardless of the spouse's age). The baseline response rate was 80.4%. The final sample consists of 6,047 households and 8,222 individuals. Weights adjust for stratification, nonresponses, and oversampling of certain groups (e.g., minorities) thus ensuring generalizability of the data to the targeted cohorts. The weight used in the analysis is centered and yields approximately the number of interviewed respondents. This weight is appropriate for use with probability statistics. Weights are only available for age-eligible individuals (n = 7,447). In addition, sampling error stratification codes and stratum-specific half-sample codes permit estimation of sampling errors for complex survey sample designs. The subsample for the present study consisted of age-eligible respondents (primary respondents and spouses) who had some driving experience (i.e., those who never drove were excluded; n = 897). Also excluded were the very few respondents with missing data on driving involvement (n = 58). These exclusions yielded a sample size of 6,492. In addition, 9.9% of this sample was based on proxy interviews. Cognition is not available for proxy interviews. An additional 3.9% refused to complete the cognition tests. Both groups (proxy interviews and cognition refusals) were also excluded from the analyses (n = 900). Individuals with proxy interviews were much less likely to drive than were other groups (60.4% compared with 20.9% of those with valid cognitive scores). Thus, their exclusion may reduce the association between cognition and driving involvement and lead to overly conservative tests of our hypotheses. Missing data for the other variables (including missing data for spouses) were very low (n = 132) so that it was not deemed necessary to impute missing data. The final sample size for all analyses is, therefore, 5,460 (2,261 men and 3,199 women).
Measures
Dependent Variable.
The dependent variable for our analyses is respondents' driving involvement. This variable was derived from answers to the questions "Are you able to drive?" and "Do you limit your driving to nearby places or do you also drive on longer trips?" Responses to these two questions were combined into one three-category variable (does not drive, drives short distances, and drives long distances).
Independent Variables.
The main independent variable for the analyses is cognitive functioning. Cognition was measured in the AHEAD study in several ways in an effort to capture the major dimensions of cognitive functioning and to provide differentiation across the full range of cognitive abilities (Herzog and Wallace 1997; Ofstedal, McAuley, and Herzog 2001). These included measures of memory, working memory, and mental status. An immediate free-recall test and delayed free-recall test were used to assess memory. In the immediate test, short, concrete, high-frequency nouns were read to the respondent, who was then asked to recall as many as possible. After 5 min of survey questions, the respondent was asked to complete the delayed test, recalling the nouns previously presented. The number of correctly recalled nouns was scored for both immediate and delayed recall performance. Working memory was assessed by the Serial 7 test. In this test, the respondent was asked to begin at 100 and subtract by increments of 7 for five trials. Each correct trial scored 1 point. Additional tests to measure knowledge, language, and orientation included the following: counting backward from 20 for 10 continuous numbers (2 points for correct answer at the first start, 1 point for the correct answer at the second start); naming the day of the week and the day, month, and year (1 point each for correct response); naming the objects that "people use to cut paper" and a "prickly plant that grows in the desert" (1 point for each correctly named); naming the President and Vice-President of the United States (1 point for each correct last name). Points for these tests are combined (Telephone Interview for Cognitive Status). AHEAD formed an aggregate index of cognitive functioning by summing the raw scores of the two memory tests and the mental-status–type items. We use this index in most analyses. Herzog and Wallace 1997 further suggested cut points for the aggregate cognition score to distinguish among severely cognitively impaired and other individuals. This cut point was set at 8 of the 35-point cognition score, that is, 2 standard deviations below the mean. On the basis of this recommendation, we created a three-category cognition score. Respondents scoring 8 or fewer points were classified as severely cognitively impaired, and respondents with scores of 9 to 12 points (1 standard deviation below the mean) were classified as mildly cognitively impaired. Respondents scoring 13 or more points were considered unimpaired. This variable was used exclusively for the bivariate analysis shown in Table 2 .
|
Controls.
Numerous other factors have been identified as affecting driving involvement in later life. These include sociodemographic background, which relates to access to reliable cars along with costs of driving, as well as physical limitations and other health conditions that may reduce one's ability to drive.
The sociodemographic characteristics in the analyses were race, income and assets, and education. Race was coded into three dummy variables (Black, Hispanic, other racial/ethnic background but not Caucasian). Caucasians served as a reference category. The two economic variables were income (personal or, for married individuals, couple's income) and net worth. Net worth constitutes the individual's or couple's total net assets. Both variables were logged to reduce their skewed distribution. Education reflects years of formal schooling (range = 0–17).
The second set of controls consisted of selected health conditions that may constrain individuals' ability to drive. We use limitations both in activities of daily living (ADLs) and selected other health conditions that may affect driving ability. The AHEAD survey contains several assessments of limitations in ADLs. One set of questions pertains to severe limitations (ability to dress, bathe, walk across a room, eat, get in and out of bed, use the toilet), another set refers to less severe limitations (walking several blocks, climbing stairs, pushing large objects, lifting weights of 10 pounds or more, picking up a dime). Because persons with severe restrictions in ADLs most likely are unable to drive as well (indeed, some of the measured severe ADLs predict driving behavior nearly completely—e.g., 97% of those who have difficulty eating and 95% of those unable to use the toilet without help do not drive), we focus on the less severe ADLs. Limitations in instrumental ADLs such as using the telephone or taking medicine tap both cognitive and physical abilities. Because of the potential overlap with cognitive functioning, we chose not to include limitations in instrumental activities as controls in the analyses. However, sensitivity analyses including number of limitations in instrumental activities as controls yield the same results for the main independent variables as the models without this variable. For the limitations in ADLs used in the analyses (walking several blocks, climbing stairs, pushing large objects, lifting weights of 10 pounds or more, picking up a dime), individuals were asked whether they had "any difficulty" performing the activity and, if they had some difficulty, whether it was "a little" or "a lot." On the basis of these responses we computed a score of limitations in ADLs that combines number of limitations and extent of difficulty. Specifically, for each activity we assigned a score of 0 if a person had no difficulty, a score of 1 if she or he had a little difficulty, and a score of 2 if she or he had a lot of difficulty with the task. These scores were then summed across all five activities. The resulting score could range from 0 to 10.
The AHEAD surveys further include measures of numerous other health conditions. For some conditions, individuals were simply asked whether or not they had the condition (e.g., arthritis), but for other conditions they were asked whether they ever had a problem (e.g., a stroke or heart attack) with follow-up questions about whether or not they currently had problems resulting from these events (e.g., whether they currently had problems from their stroke or whether they currently had angina or chest pains). Clearly, it is current problems that are likely to impinge on driving behaviors. We thus used only current health problems as controls. The only exception was the occurrence of a stroke. Persons who ever had a stroke may have experienced some permanent impairment which may not be perceived as a serious problem. Consequently, we used two dummy variables for stroke: ever had a stroke but no current problems (1 = yes), and has current problems related to a past stroke (1 = yes). Persons who never had a stroke served as reference. All other conditions were coded as single dummy variables (1 = currently has the condition, 0 = does not have the condition). The conditions used in the analyses are those that have been shown to affect driving behaviors, namely, heart problems, arthritis, diabetes, and respiratory problems. In addition, we included a self-assessment of vision based on a question on how individuals judged their eyesight (with correction, if any). Answer categories ranged from 1 (excellent) to 5 (poor). Thus the scoring reflects perceived vision problems. Means and standard deviations of all variables are shown in Table 1 .
|
|
). Because interaction terms are often difficult to interpret, we estimated probabilities of driving involvement for relevant subgroups (i.e., by gender, driver in the household, partner's driving involvement, residence in a metropolitan area) from the full multivariate models. For these estimates, cognition was set to the mean and 1 standard deviation above and below the mean (Aiken and West 1991; Demaris 1992).
All analyses rely on weighted data. Sample selection for AHEAD relied on a complex survey design in which selection into the sample is not independent of other respondents. This can lead to an underestimation of error variances and thus biased results. We consequently estimated the regressions using the complex survey design methods provided in Stata (Stata 1999). Specifically, Stata uses pseudo-maximum likelihood estimates for error variances, adjusted for sample weights, clusters (primary sampling units), and strata. Model fit is based on F tests. Because standard errors in these models are estimated, the exponentiated coefficients are relative risk ratios rather than the standard odds ratios (Stata 1999).
Initial analyses revealed a significant gender interaction, and they also indicated that the effects of other independent variables on cognition varied by gender. We consequently estimated all models (except those with the Gender x Cognition interaction term) separately by gender.
|
Results |
|---|
|
TOPAbstractLiterature ReviewHypothesesMethodsResultsDiscussionReferences |
|---|
That gender moderates the impact of cognition on driving is also confirmed by multivariate analyses that include all independent and control variables (data not shown, the variables in the model were the same as those shown in Table 3 ). Because some men and women in our sample are married to each other, these analyses were performed using multinomial logistic regressions with clusters for households and robust standard errors (Stata 1999). This approach corrects for nonindependence of observations. As shown in Fig. 1, the data indicate a significant Gender x Cognition interaction for driving short distances versus not driving (b = .06, p < .01). Specifically, among women, poorer cognitive functioning is primarily linked to driving cessation but has only a limited impact on short-distance driving, whereas among men it is foremost associated with driving short distances but has less effect on driving cessation. Long-distance driving is positively associated with better cognitive functioning among both men and women. This suggests that women adjust to cognitive impairments through driving cessation, whereas men adapt through driving restriction.
|
Having another person of driving age in the household enhanced the chances of not driving for both men and women. This effect did not differ by gender. This suggests that the availability of an alternate driver in the household encourages driving cessation.
In addition, the presence of other persons of driving age in the household did moderate the effect of cognition on driving involvement, and this interaction differed significantly by gender (see Fig. 2 and Fig. 3). Among women, cognition has a stronger impact on driving if there is another driver in the household than if there is no other driver (the interaction terms are b = -.07, p = .053, for not driving versus driving long distances and b = -.06, p < .05, for driving short versus long distances). Thus, women with low cognition are much less likely to drive long distances if there is another driver in the household than if there is no other driver, and they are more likely to cease driving altogether if another driver is available (see Fig. 2). This suggests that the lack of an alternate driver in the home keeps cognitively impaired women on the road. Among men, the relationship between cognition and driving involvement is also moderated by having another person of driving age in the household (the interaction effects are b = .13, p < .01, for not driving versus driving long distance and b = .07, p < .05, for driving short distances versus driving long distances). Cognition has little influence on driving among men with alternate drivers in the household, and it has a pronounced effect among men with no alternate drivers in the household (see Fig. 3). Specifically, men with low cognitive functioning are somewhat more likely to drive short distances if no other driver is in the household than if another driver is present, and men with high cognitive functioning are more prone to continue long-distance driving when no other driver is in the household than when another driver is available. Thus, having an alternative driver in the household enhances the effect of cognition on driving among women, and it reduces this effect among men.
|
|
Furthermore, there is evidence for an effect of partner's driving involvement. Having a partner who drives long distances decreases the probability of driving short rather than long distances, an effect that is similar for men and women. However, individuals whose partners are long-distance drivers also tend to cease driving altogether rather than driving short distances. This latter effect is significantly more pronounced for men than for women. These results suggest a complex scenario concerning the impact of partner's driving. If individuals drive at all, then they seem to adjust the extent of their driving to that of the partner, that is, they are more likely to drive long distances if their partner drives long distances and vice versa. On the other hand, the decision to cease driving altogether seems to be furthered by having a long-distance driving partner, especially among men.
We hypothesized that the effect of cognition on driving would be contingent on partner's driving involvement. Initial analyses revealed, first, that the moderating effect of partner's driving is restricted to women. Among men, partner's driving involvement did not moderate the relationship between cognition and driving (the gender difference is significant). Furthermore, the main moderating condition is whether or not the partner drives at all, whereas the extent of partner's driving (short versus long distances) has little impact. Also, the association between cognition and driving is similar among those women whose partners drive and those who have no partners.
On the basis of these initial findings, we further explored the impact of having a nondriving partner versus having no partner or a driving partner for women. These analyses demonstrate that the association between cognition and driving is much stronger among women with nondriving partners (b = -.28 for not driving versus driving long distance and b = .13 for driving short distance versus not driving) than among women with driving partners or no partners (bs = -.09 and .06, respectively). Thus, women with low cognition are more likely not to drive if they have a nondriving partner than if they have a driving partner or no partner, whereas women with high cognition are particularly unlikely to cease driving if their partner doesn't drive (see Fig. 4). Apparently, partner's driving or not having a partner keeps cognitively impaired women driving at least short distances, whereas women with high cognitive functioning seem to compensate for their partner's nondriving by driving themselves.
|
In addition to the hypothesized effects, we also examined the impact of the control variables on driving involvement (see Table 3 ). Age generally contributes to both driving restriction and cessation. Minorities (Black, Hispanic) are more likely not to drive at all or to restrict their driving to short distances. However, Black men are somewhat less likely not to drive than to drive short distances. Examination of the impact of the socioeconomic variables (income, net worth, education) suggests a trend toward more driving involvement among the higher socioeconomic status groups. Selected health conditions also influence driving behaviors. Persons who experienced a stroke with lasting problems tend not to drive, as do men (but not women) who experienced a stroke but report no lasting symptoms. Driving restriction and cessation are also more common among individuals with impaired vision and those with limitations in ADLs. On the other hand, heart disease, arthritis, and diabetes have no impact on driving when other health conditions are controlled.
|
Discussion |
|---|
|
TOPAbstractLiterature ReviewHypothesesMethodsResultsDiscussionReferences |
|---|
Our analyses confirm findings from earlier studies (Colsher and Wallace 1993; Marottoli et al. 1993; Stutts et al. 1998; Trobe et al. 1996) that demonstrated an association between cognition and driving. Generally, driving involvement increases with cognitive functioning. However, over one third of persons with low cognitive functioning (37%) continue to drive though mostly short distances. The size of the group still driving was previously unreported and, given its size, indicates a significant need to better understand decision processes about driving behavior among those with impaired cognitive functioning. Given the difficulty in measuring cognitive functioning in large surveys, future research will also need to address to what extent cognition measures like those used in the AHEAD surveys adequately distinguish between competent and noncompetent drivers.
The results further indicate distinct driving behaviors among men and women. Both driving restriction and cessation predominate among women, and the association between cognition and driving involvement differs by gender. Specifically, men with low cognitive functioning tend to restrict their driving to short distances, whereas women with similarly low cognitive functioning tend to cease driving altogether. That men with mild and severe cognitive impairments are less likely to cease driving than women are lends support to the contention that the ability to drive is an important contributor to men's self-image, whereas women most likely view its importance more in terms of mobility and functional independence. Another explanation may be that men of this generation are able to maintain competence using a copilot spouse (Foley et al. 2000), an activity not available to many women who have outlived their spouses. It is notable, however, that the effect of cognition on driving involvement is quite similar for men and women in the analyses performed separately by gender. These latter analyses (the results shown in Table 3 ) imply a different model than the analyses testing for a Gender x Cognition interaction, namely, they imply gender interactions with all variables in the model. This suggests that it is not gender per se that moderates the effect of cognitive functioning on driving involvement but rather gender differences in the characteristics (sociodemographic, health conditions, alternate transportation opportunities) that impinge on cognition and driving behaviors. Our analyses reveal some gender differences in the moderating effect of alternative transportation opportunities on the association between cognition and driving involvement, but future research will need to explore other potential moderators.
Additional hypotheses addressed the impact of alternative transportation opportunities on driving involvement per se and on the association between cognitive functioning and driving involvement. We first explored the effect of having another person of driving age (other than one's partner) in the household. As expected, individuals with potential drivers in the household are more likely not to drive than are those persons who have no alternate driver in the household. However, having a driver in the household does not distinguish among short- and long-distance drivers. Thus, alternate drivers in the household seem to encourage driving cessation but not driving restriction, and this trend is similar for men and women.
Having an alternate driver in the household not only affects driving involvement per se but also moderates the association between cognitive functioning and driving. This moderating effect differs by gender. Among women, having an alternate driver in the household tends to discourage long-distance driving and to encourage driving cessation and restriction among those with low cognitive functioning. For men, on the other hand, having no alternate driver in the household furthers continued long-distance driving especially among those with relatively high cognitive functioning. In other words, the need to drive (lacking alternate drivers) keeps cognitively impaired women on the road, and it keeps men with high cognitive functioning driving long distances. Although we only have information on the age of household members but not on their actual driving competency, this finding is unlikely to reflect differences in the driving competency of men's and women's coresidents. We found no significant gender differences in type of household composition, that is, in whether respondents lived with children or with other relatives or nonrelatives.
Another alternate transportation opportunity derives from partner's driving involvement. The findings indicate quite complex relationships between respondents' driving involvement and that of their partners. Men without partners tend not to drive at all, a result that is not replicated among women. Men without partners constitute a distinct social group, as most older men remain married until their deaths (Fields and Casper 2001). However, our analyses control for main health conditions and sociodemographic background. Thus, further research is needed to establish what conditions lead older single men to cease driving. It also is conceivable that being alone decreases men's need to drive. Women without partners are less likely than women with live-in partners to restrict their driving to short distances, most likely a function of lesser need to drive long distances when a partner is available.
Among those with live-in partners, driving involvement is contingent on that of the partner. On the one hand, the extent of both men's and women's driving involvement parallels that of the partner, that is, they are more prone to drive long distances if their partner drives long distances and more prone to drive short distances if their partner drives short distances. On the other hand, men, but not women, with long-distance driving partners are more likely not to drive than to drive short distances, suggesting that their wives' ability to drive enables men to give up driving. For men, these findings provide some support for the assumption that partner's driving can constitute an alternative transportation opportunity that decreases the need to drive oneself. However, there is also evidence of similarity in partners' driving involvement, a finding consistent with research showing homogeneity in spouses' health behaviors, such as smoking, dietary practice, and substance abuse (Graham and Braun 1999; Knuiman, Divitini, Welborn, and Bartholomew 1996; Macken, Yates, and Blancher 2000). As surveys such as AHEAD only provide information on trends but not on intricate decision-making processes, more research is needed to explore how couples decide on each partner's driving involvement and whose driving ability and/or inclination to drive is more influential in the relationship.
Having a partner and partner's driving involvement were further expected to moderate the association between cognitive functioning and driving involvement. Our data support such a moderating effect for women but not men. Generally, the association between cognition and driving involvement is more pronounced among women whose partner does not drive than among those with no partners or with driving partners. Specifically, women with low cognitive functioning are particularly prone to cease driving if their partner does not drive, another indication of the tendency toward similarity in partners' driving involvement. On the other hand, women with medium and high cognitive function tend to maintain short-distance driving if their partner does not drive, which may again be a function of their greater need to drive. This again suggests that adjustment of driving to cognitive competence derives from complex decision processes that reflect not only the need to drive but also adaptations to partners' behaviors and perhaps partners' ability to assist a cognitively impaired person in her or his driving efforts. The gender difference in these findings may also reflect women's inclination to adjust to their partners' needs (Gilford 1986; Niederfranke 1991).
Taken together, our findings suggest that decisions to drive in the later years derive from complex decisions that reflect not only competence but also the availability of alternate transportation opportunities as well as issues of self-image and partner relations. In view of the limitations of our study, each of these potential influence factors will require further research. Even though our analyses overcame some problems of earlier research, they are also limited in several ways. The exclusion of persons with proxy interviews reduced generalizability of the data and may have led to overly conservative estimates, especially of the association between cognitive functioning and driving involvement. Furthermore, the validity of cognitive tests used in large surveys (without confirming clinical data) requires further investigation, and more detailed information on specific driving habits is needed to derive more precise conclusions about adjustments of driving behaviors to cognitive functioning as well as to other circumstances such as alternative transportation opportunities. In addition, cross-sectional data typically do not allow causal attributions. Although longitudinal analyses are certainly needed to assess whether cognitive impairment and changes in cognitive functioning predict driving cessation, it is unlikely that driving cessation is a primary cause of cognitive impairment. Thus, we felt justified in applying a causal interpretation to the data. We also lack information on driving performance or recent history of driving performance (crashes, traffic violations). Such information would be necessary to ascertain whether cognitively impaired drivers pose a threat to themselves or to other drivers.
Despite these limitations, our findings offer some insights useful for interventions with older drivers. Most importantly, we find that a significant minority of cognitively impaired persons continue to drive. Even if our measure of cognition may be wanting, this raises serious questions about the efficiency of licensing and other procedures in weeding out incompetent drivers. Furthermore, interventions also need to address the motives that keep cognitively impaired individuals on the road. Among men, identification of driving with the male role may lead to unsafe driving behaviors. Thus, successful interventions may have to address the association of driving cessation with men's self-image. Similarly, spouses apparently engage in complex decision processes concerning each partner's driving involvement. It thus seems essential to involve both partners in interventions and licensing determinations even if only one partner's driving is problematic. Especially among women but also among some men, the need to drive seems to override adjustments of driving to cognitive impairment. This suggests the need for interventions that target older individuals' ability to maintain independence without having to drive. More generally, efforts to keep potentially risky drivers off the road need to address more than driving competence. They must deal with the complex decision processes surrounding driving restriction or cessation and the consequences of driving cessation for individuals' self-worth, partner relationships, and independent living.
|
Acknowledgments |
|---|
Received for publication June 28, 2001. Accepted for publication March 22, 2002.
|
References |
|---|
|
TOPAbstractLiterature ReviewHypothesesMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  L. Keay, B. Munoz, K. A. Turano, S. E. Hassan, C. A. Munro, D. D. Duncan, K. Baldwin, S. Jasti, E. W. Gower, and S. K. West Visual and Cognitive Deficits Predict Stopping or Restricting Driving: The Salisbury Eye Evaluation Driving Study (SEEDS) Invest. Ophthalmol. Vis. Sci., January 1, 2009; 50(1): 107 - 113. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. D. Edwards, L. A. Ross, M. L. Ackerman, B. J. Small, K. K. Ball, S. Bradley, and J. E. Dodson Longitudinal Predictors of Driving Cessation Among Older Adults From the ACTIVE Clinical Trial J. Gerontol. B. Psychol. Sci. Soc. Sci., January 1, 2008; 63(1): P6 - P12. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. A. Perkinson, M. L. Berg-Weger, D. B. Carr, T. M. Meuser, J. L. Palmer, V. D. Buckles, K. K. Powlishta, D. J. Foley, and J. C. Morris Driving and Dementia of the Alzheimer Type: Beliefs and Cessation Strategies Among Stakeholders Gerontologist, October 1, 2005; 45(5): 676 - 685. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
