- Email: [email protected]
Elderly drivers’ everyday behavior as a predictor of crash involvement—Questionnaire responses by drivers’ family members
Accident Analysis and Prevention 50 (2013) 397–404
Contents lists available at SciVerse ScienceDirect
Accident Analysis and Prevention journal homepage: www.elsevier.com/locate/aap
Elderly drivers’ everyday behavior as a predictor of crash involvement—Questionnaire responses by drivers’ family members Yoshinori Nakagawa a,∗ , Kaechang Park b , Yasuhiko Kumagai b a b
Department of Management, Research Organization for Regional Alliance, Kochi University of Technology, Japan Research Organization for Regional Alliance, Kochi University of Technology, Japan
a r t i c l e
i n f o
Article history: Received 13 January 2012 Received in revised form 26 April 2012 Accepted 4 May 2012 Keywords: Elderly drivers Crash risk Cognitive impairment Driver Behavior Questionnaire (DBQ) Everyday Behavior Questionnaire (EBQ)
a b s t r a c t When or whether elderly drivers stop driving is concerning not only to the drivers themselves but also to their family members. Therefore, it is important for family members to take the initiative if they wish to obtain information on the likelihood of the drivers’ involvement in crashes. On the basis of the older drivers’ Everyday Behavior Questionnaire (EBQ) developed in this paper, we attempt to predict drivers’ involvement in crashes using the responses given by their family members. The results revealed that this 14-item questionnaire has a sufficient level of internal consistency as well as a significant correlation (r = 0.29) with the experience of involvement in crashes in the last three years (p < 0.01). Although the EBQ is a proxy-reported questionnaire and does not include items directly related to driving behaviors, the correlation between the EBQ and crash involvement is stronger than that of the self-reported Driver Behavior Questionnaire reported in deWinter and Dodou (2010), who conducted a meta-analysis and estimated the overall correlation among samples of earlier studies. In addition, logistic regression analysis showed that the EBQ score and the exposure to driving risks, measured by the frequency of driving, are significant predictors of involvement in crashes. © 2012 Elsevier Ltd. All rights reserved.
1. Introduction 1.1. Background Previous studies have shown that older drivers have higher crash rates per vehicle mile of travel (McGwin and Brown, 1999; Retchin and Anapolle, 1993; Massie et al., 1995). With the projected increase in the elderly driving population worldwide (Parker et al., 2000; Lyman et al., 2002), including Japan, the problem of an increasing number of crashes involving elderly drivers is attracting the attention of many researchers and practitioners. When addressing this problem, it is important to note that driving restrictions and cessation threaten older adults’ lives. According to Stutts and Wilkins (2003) and other authors, these conditions lead to consequences such as (i) loss of mobility (Evans, 2001; Marottoli et al., 2000; Rosenbloom, 2001), (ii) loss of identity and increased dependency (Burkhardt et al., 1998; Carp, 1988; Culter, 1975; Eisenhandler, 1990), (iii) physical and mental problems such as increased depression, heart disease, fractures, and stroke (Bassuk et al., 1999; Fonda et al., 2001; Marottoli et al., 1997),
∗ Corresponding author at: 185 Miyanokuchi, Kami City, Kochi Prefecture, Japan. Tel.: +81 887 57 2768; fax: +81 887 57 2124. E-mail address: [email protected] (Y. Nakagawa). 0001-4575/$ – see front matter © 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.aap.2012.05.014
and (iv) decreased social integration, as measured by number and frequency of social contacts (Turano et al., 2009). It should also be noted that when or whether elderly drivers stop driving is of great concern to their family members for two reasons. First, if elderly drivers suffer the above-mentioned negative consequences on cessation of driving, their families are likely to be burdened psychologically, physically, and economically. Second, family members do not want their loved ones to be involved in crashes, and often hope for their driving cessation. Thus, it is important that family members take the initiative, if they wish, to obtain information on the likelihood of the elderly relatives getting involved in crashes and to gain a stance on driving cessation.
1.2. Objective Against this background, this study develops an older drivers’ Everyday Behavior Questionnaire (EBQ). This questionnaire, answered by older drivers’ family members, attempts to predict drivers’ involvement in crashes. This study also checks its internal consistency as well as the validity by calculating its correlation with the drivers’ involvement in crashes in the last three years and with two factors of the proxy-reported Driver Behavior Questionnaire (DBQ), lapses and errors. The EBQ has two characteristics. First, this questionnaire is answered by older drivers’ family members who live with the
398
Y. Nakagawa et al. / Accident Analysis and Prevention 50 (2013) 397–404
drivers and observe their everyday lives. Thus, family members can independently answer the questionnaire without any technical know-how and within a short span of time. Second, as the EBQ consists of items not directly related to driving, respondents are not required to be regular passengers in the elderly drivers’ cars. 1.3. Related earlier studies Several lines of earlier research have aimed at predicting the performance of drivers, specifically those who are elderly. One line of research investigated the association between driving performance and requisite cognitive functions for driving which were assessed using various tests, such as the Ergovision test, the Useful Field of View (UFOV) test, a Complex Reaction Time Task, the Benton Line Orientation task, Clock Drawing, and the Trail Making test. Mathias and Lucas (2009) conducted a meta-analysis of these earlier studies and concluded that a variety of tests appear suitable for screening older drivers, the exact choice of which depends on the “gold standard” for determining driving ability (on-road driving, simulator performance, driving problems). It should be noted that although a combination of such ability measures can successfully identify risky drivers (e.g., McKnight and McKnight, 1999), these tests are burdensome to drivers, and may not be easy to adopt, especially when drivers are unwilling to retire from the roadways. A second line of research examined the association between driving performance and neurological disorders that older people are more likely to suffer, such as stroke, dementia, and mild cognitive impairment (e.g., Frittelli et al., 2009; Zuin et al., 2002). Many of these studies investigated the association between driving performance (on-road driving or simulator performance) and commonly used screening tests for detecting dementia and cognitive impairment, such as the Clinical Dementia Rating (CDR) and Mini Mental State Examination (MMSE). However, these two scores are not without limitations: the CDR is time-consuming and is rarely used outside of specialist clinics, and MMSE scores correlate poorly with driving performance (Wagner et al., 2011); the reason for this poor correlation is discussed by Frittelli et al. (2009). A third line of research explored the association between drivers’ involvement in crashes and their driving behavior patterns. Among the typologies of aberrant driving behaviors, the DBQ is one of the most popular instruments. A number of earlier studies have aimed to predict drivers’ involvement in crashes using the questionnaire’s subscales, such as errors (mistakes that have potentially dangerous consequences), lapses (primarily attentional failures), and violations (risky driving behaviors). In a meta-analysis of these studies, deWinter and Dodou (2010) concluded that the violation and error factors of the DBQ predicted crashes with an overall correlation of 0.13 and 0.10, respectively. Parker et al. (2000) applied the DBQ to drivers aged 50 and above, and found that, in an older sample, relatively high scores on the error and lapse factors were predictive of involvement in an active accident. It should be noted that these studies are based on a self-report questionnaire, and it is not clear to the authors of the present paper whether this approach is applicable to drivers who are suspected of having a certain degree of cognitive impairment. It is expected that the proposed instrument will complement the above-mentioned existing measures in predicting aged drivers’ performance. 2. Methods and materials 2.1. Participants Data were collected via an internet research company, Cross Marketing, Inc., whose registered members were invited by to
participate in the preliminary survey. Potential participants were asked if they satisfied the following four conditions. Those who satisfied all four were invited to participate in the main survey. Condition 1: He or she lives in the same house as the elderly family member aged 70 or more. Condition 2: The elderly relative drives at least twice or three times per month. Condition 3: He or she wants the elderly relative to cease driving. Condition 4: He or she rides as a passenger in the car driven by the elderly family member at least once per two or three months. Condition 3 was checked by the item “Do you want him/her to cease driving?”, and the respondents were asked to choose from one of the five alternatives: 1 = Agree, 2 = Agree to some extent, 3 = No opinion, 4 = Disagree to some extent, and 5 = Disagree. Respondents who chose options 1 or 2 were considered to satisfy the given condition. Condition 4 was included to check the validity of the scale developed in this study, although the EBQ proposed in this study was designed such that respondents were not required to be regular passengers in elderly drivers’ cars. If the scale exhibits validity, then scores should be associated with the frequency of lapses and errors reported by the elderly drivers’ family who have been passengers. Among those who qualified for the preliminary survey, 488 people agreed to participate in the main survey. There were 268 (54.9%) males and 220 (45.1%) females, with an average age of 50.0 (SD 14.8). The average age of their elderly family members was 76.3 (SD 4.6). More than half of them (55.7%) were the respondents’ parents and 22.1% of them the respondents’ spouses. The numbers of drivers with no, one, two, and three or more crash experiences in the last three years were 289, 116, 47, and 36, respectively.
2.2. Materials 2.2.1. Characteristics of respondents and their elderly family drivers Socio-demographic and other questions were included in the questionnaire to determine (i) respondents’ age and gender, (ii) age, gender, and former and current employment status of the respondents’ elderly family members, (iii) driving history of the elderly family members (in years), (iv) driving exposure or frequency of driving of the elderly family members (4 = Every day, 3 = Twice or three times per week, 2 = Once per week, 1 = Twice or three times per month), (v) frequency with which the respondents ride as passengers in the cars driven by the elderly family members (4 = Once or more per week, 3 = Once per two or three weeks, 2 = Once per month, 1 = Once per two or three months), and (vi) the elderly family member’s involvement in crashes in the last three years (4 = Three times or more, 3 = Twice, 2 = Once, 1 = None). Involvement in crashes was measured by the item “In the past three years, has your elderly family member caused crashes (incidents with other vehicles, pedestrians, cyclists, or objects, or more serious crashes) while driving a car?” Although the present study could have used a yes or no response, a four-point scale was adopted, ranging from 1 (None) to 4 (Three times or more), and respondents who answered “1” were regarded as having no crash experiences. This was expected to enhance the reliability of the item. With regard to driving exposure, most earlier studies asked drivers about kilometers (miles) driven in a specific period using a self-report questionnaire survey. We did not adopt this method of measuring exposure, because this study involved a proxy-reported survey, and reporting the total distance driven by their family members would be difficult for the respondents.
Y. Nakagawa et al. / Accident Analysis and Prevention 50 (2013) 397–404
2.2.2. Everyday behavior measures We created a total of 100 items describing everyday behaviors of elderly people, some of which were adopted from earlier studies to predict their driving performance. Using a five-point scale (from 5 = Strongly yes to 1 = Not at all), respondents were asked to rate the extent to which each item described their family members. The 100 items were developed in the following manner. Many authors have classified abilities required in driving; among them, McKnight and McKnight (1999) identified fourteen abilities and classified them into five categories: sensory, attentional, perceptual, cognitive, and psychomotor. We considered the implications of impairment of each of the abilities on older people’s everyday lives and, where possible, developed plural items for each ability. For example, in the case of declining attentional abilities, older people may find it difficult to concentrate on tasks in their daily lives and to drive safely. On the basis of this assumption, we created items such as “It is getting difficult for him/her (your elderly family member) to concentrate on a specific task (e.g., hobbies such as reading books) for as long as he/she used to” and “It is getting difficult for him/her to stay in the same place for as long as he/she used to.” Certain items were adopted from earlier studies; for example, with regard to short-term memory (a cognitive ability), items from the Short Memory Questionnaire (SMQ; Riege, 1982; Koss et al., 1993) were utilized after minor re-wording. The SMQ is a scale consisting of 14 items to assess memory problems of older people. The scale is rated by the elderly drivers’ family caregivers, and its scores have a significant association with the MMSE, which is widely used in clinical practice for screening dementia. An example of such items is “He/She always remembers where he/she put his/her keys.” Finally, some items were adapted from the medical literature describing the symptoms of higher brain dysfunction. As detailed later, 64 of the 100 items were correlated with crashes at a significance level of 0.01, and the top 14 items with relatively higher correlations were included in the questionnaire. The procedure for determining the number of items to be included in the questionnaire will be detailed in Sections 2.3.1 and 3.1. The list of the 64 items is shown in Table 1, with the average score and SD of each item as well as the correlation of each item with crash involvement. 2.2.3. Driver Behavior Questionnaire (DBQ) The Manchester Driver Behavior Questionnaire (DBQ; Reason et al., 1990; Parker et al., 1995a,b; Lawton et al., 1997) consists of 20 items, of which 11 items were modified and adapted. The item selection criteria were as follows. First, many earlier studies that applied factor analysis to the items in the DBQ identified three factors: errors, lapses, and violations. We adopted 16 items relating to errors or lapses, because it has been reported that only items relating to these factors were associated with crash involvement, as far as elderly drivers were concerned (Parker et al., 2000). Second, four items among the sixteen were omitted, because it seemed inappropriate to include them in a proxy-reported questionnaire. For example, the item “Fail to check rearview mirror before pulling out, changing lanes, etc” was omitted, because family passengers do not necessarily recognize whether a driver has checked the mirrors. Third, in addition to these four items, the item “Misread signs and took the wrong turning off a roundabout” was omitted, because drivers seldom encounter roundabouts in Japan. As a consequence of this selection procedure, four items on lapses and seven items on errors were adopted in the present study. Finally, minor re-wordings were made for these eleven items: the word “you” was replaced by “he/she,” denoting the respondent’s elderly family member. The list of the eleven items is shown in Table 2, with mean scores and SD for the 488 respondents. Cronbach’s alpha coefficients of the two factors (i.e., errors and lapses) were 0.87 and 0.88,
399
respectively. Using a six-point scale (from 6 = Nearly all the time to 1 = Never), respondents were asked to rate the extent to which each item described their family members. 2.3. Analysis 2.3.1. Development of Everyday Behavior Questionnaire (EBQ) As stated above, a total of 100 items were used to measure how effectively the elderly family members carry out activities of daily living. We compiled the questionnaire by choosing items that distinguish efficiently between safe and risky drivers. Specifically, we calculated the Pearson product–moment correlation coefficient between the score on each item and the frequency of crashes in the last three years. Then, we chose items whose correlation coefficients were larger than the critical value rcr . Further, the EBQ score was defined as the total score of these selected items. The critical value was defined to maximize the correlation coefficient between the total score and the frequency of crashes. If rcr is too small, then many irrelevant items would be included in the questionnaire, and the correlation coefficient would become small. Moreover, if rcr is too large, the correlation coefficient would again become small, in spite of the strong relevance of the selected items, because the number of items would be small. For this reason, it was expected that the critical value rcr should have an optimal value. We will show later that rcr can indeed be optimized. After determining items to be included in the EBQ, we checked the internal consistency of the set of items by calculating Cronbach’s alpha coefficient. The procedure for defining the critical value rcr will be explained in greater detail in Section 3.1. It should be noted that the frequency at which an elderly driver causes crashes depends not only on his/her driving ability but also on the exposure to risk (i.e., the distance driven by him/her during a specific period). Therefore, to establish that the procedure described is a sound method of choosing items that represent driving ability, it is necessary to confirm the assumption that no items are associated more strongly than others with the exposure to risk. Our analysis confirmed that among the 100 items, only two items were significantly associated with exposure (p < 0.05). Moreover, in regard to these items, the absolute values of the correlation coefficients were very small (0.12 and 0.09, respectively). This implies that the assumption described above is valid and our methodology was sound. 2.3.2. Association between Everyday Behavior Questionnaire (EBQ) score and crash frequency To check the validity of the EBQ, we first calculated correlation coefficients to examine the association of this scale with three external criteria: frequency of crashes in the last three years and the two DBQ factors, lapses and errors. Second, we studied how precisely the scale can screen elderly drivers who are likely to cause crashes. Specifically, we established a multivariate logistic regression model to explain the frequency of crashes in terms of the EBQ score and the exposure to risk. In multivariate logistic regression analysis, objective variables must be binary. In the present study, elderly divers with more than one experience of crash involvement in the last three years were coded 1, and the others were coded zero. As it was assumed that demographic variables such as drivers’ gender and age affect the frequency of crashes only through affecting the EBQ score, they were not included in the model as explanatory variables. 3. Results 3.1. Development of Everyday Behavior Questionnaire (EBQ) The correlation coefficients of the 100 items with crash involvement ranged between −0.04 and 0.25, and the average and SD were
400
Y. Nakagawa et al. / Accident Analysis and Prevention 50 (2013) 397–404
Table 1 14 items and 50 candidate items of the Everyday Behavior Questionnaire (EBQ). No. 1 2 3 4
5 6 7 8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32
33 34 35 36 37 38 39 40
Item
Mean
SD
Correlationa
He/she misses or nearly misses a step on the stairs more frequently than he/she used to in his/her house and other facilities. He/she has slower reflexes than he/she used to have. It is becoming difficult for him/her to think or talk logically. It is becoming difficult for him/her to judge appropriately whether to go across pedestrian crossings or stop and wait for a fresh “walk” sign, considering his/her walking speed and the time remaining before the signal changes. It is becoming difficult for him/her to stay in the same place for as long as he/she used to. He/she acts on the basis of delusions more often than he/she used to. It is becoming difficult for him/her to hear/understand what you are talking about, unless you talk loudly. In supermarkets, it is becoming difficult for him/her to pay attention to customers passing by in the background due to his/her attention on commodities in the shelves. It is becoming difficult for him/her to distinguish words of the people he/she is talking with, if there are noises or other conversations going on around him/her. He/she gets bored more easily than he/she used to. It is becoming difficult for him/her to kill mosquitoes flying around or on his/her skin. He/she watches TV turned up louder than he/she used to. It is becoming difficult for him/her to imagine how others feel in response to his/her own behavior. He/she has poorer eyesight than he/she used to have. It is becoming difficult for him/her to judge the right time to walk safely across the street where there is no signal for pedestrians. He/she behaves at his/her own pace more frequently than he/she used to. It is becoming difficult to ask him/her to watch over a kettle or a pot being heated when he/she is watching TV. He/she forgets to turn off the electricity and/or gas more frequently than he/she used to. He/she wrongly fastens buttons or fails to straighten a collar more frequently than he/she used to. It seems more difficult than it used to be to ask him/her to turn off the gas ten minutes after the water in a pot has boiled. It is becoming difficult for him/her to settle troubles between friends or neighbors. It is becoming difficult to pay attention when others talk while he/she is working with his/her hands, e.g., needle work and writing. While walking, he/she does not pay enough attention to cars and bicycles more frequently than he/she used to. He/she does not move as quickly as he/she used to. He/she watches steps more carefully when walking in the dark/outside than he/she used to. He/she behaves in an impromptu manner more frequently than he/she used to. It seems more difficult than before for him/her to judge what to do in his/her house when a natural calamity, such as an earthquake, occurs. In unexpected circumstances, it is becoming difficult for him/her to cope with a problem without panicking. When things do not go as planned, he/she goes blank more often than he/she used to. It seems difficult for him/her to perceive insects flying or crawling around than before. It seems more difficult for him/her to tell whether an insect flying around him/her is a mosquito. When walking around in the house, he/she hits his/her arms or legs on the corner of tables, walls, and posts more frequently than he/she used to. He/she adamantly expects younger people around him/her to behave in conformity with him/her than was formerly the case. It is becoming difficult for him/her to put tasks in an orderly sequence and perform them. He/she is in a foggy state more frequently than was formerly the case. He/she is losing patience. It is becoming difficult for him/her to care about issues not immediately present. He/she sometimes prefers not to be surrounded by younger people who worry about him/her. He/she is becoming indifferent to what others think or feel He/she more frequently hesitate to make risky choices.
2.25
0.95
0.247
3.02 2.35 1.87
0.98 1.01 0.83
0.237 0.210 0.210
2.56
1.1
0.204
2.73 2.69
1.1 1.19
0.199 0.198
2.2
0.97
0.198
1.99
1.06
0.196
2.4 2.48
1.06 0.95
0.195 0.192
2.88 2.62
1.24 1.09
0.190 0.190
2.98 1.91
1.02 0.88
0.187 0.186
2.85
0.98
0.185
2.26
1.09
0.183
2.31
1.03
0.182
1.82
0.87
0.181
2.14
1.01
0.179
2.17
0.97
0.178
2.38
1.01
0.178
2.33
0.95
0.176
3.05 2.68
1.03 1
0.175 0.174
1.92
1.05
0.168
2.37
1
0.167
2.51
0.97
0.166
2.3
1.02
0.164
2.3
0.95
0.164
2.15
1.05
0.163
2.26
0.95
0.159
2.43
0.98
0.158
1.82
0.99
0.156
2.31 2.51 1.98
1.09 1.09 1.06
0.155 0.155 0.152
2.72
1.06
0.151
2.02 2.31
1.11 0.96
0.150 0.149
Y. Nakagawa et al. / Accident Analysis and Prevention 50 (2013) 397–404
401
Table 1 (Continued) No.
Item
Mean
SD
Correlationa
41
It is becoming difficult for him/her to manage money and financial affairs (e.g., paying bills, budgeting). It is becoming difficult for him/her to make necessary arrangements and to perform tasks efficiently. He/she more frequently gets into tangles with people whom he/she walks past. When called out to, he/she more frequently looks this way and that to find identify who is talking to him/her. It is becoming difficult for him/her to turn his/her neck and look quickly behind himself/herself. He/she is becoming less willing to tolerate present annoyances for the sake of a future result It takes longer for him/her to comprehend sentences in books or newspapers. It is becoming difficult for him/her to remember what someone asked him/her to do. It is becoming difficult for him/her to find what he/she needs among many objects in a room. He/she has poorer vision in the dark than previously. Once he/she begins a task, it takes over his/her entire attention He/she more frequently becomes agitated if he/she can’t find something right away. It is becoming difficult for him/her to mount an escalator. He/she increasingly struggles to find the words he/she wants to use. He/she seems increasingly vulnerable to sales solicitations or unable to recognize deceptive appeals. It is becoming difficult for him/her to converse while the TV is on. It is becoming difficult for him/her to follow the motion of quickly moving objects, such as flying insects and TV captions. When walking outside, he/she walks slowly and carefully and annoys passers-by more frequently than before. When walking around, he/she more frequently loses his/her way, even in familiar areas. He/she is increasingly optimistic in his/her interpretation of the situations in which he/she is placed. He/she more frequently forgets where he/she put keys. It is becoming difficult for him/her to shop for groceries without a list and without forbecoming items (5 items). It seems more difficult for him/her to remember birthdays of family members. Sunlight and lights are more disconcerting for him/her than before.
1.65
0.96
0.147
2.01
1.07
0.144
1.78
0.92
0.144
1.88
0.98
0.144
2.43
0.95
0.142
1.96
1.07
0.142
2.43
1.11
0.140
1.95
1.04
0.137
2.03
1.09
0.136
2.67 2.05 2.44
0.98 1.04 1.03
0.135 0.134 0.132
1.9 2.22 2.07
1.1 1.09 0.93
0.128 0.128 0.128
1.9 2.63
1.01 0.94
0.127 0.127
2.12
0.92
0.122
1.75
0.95
0.120
2.94
1.07
0.120
2.11 2.03
1.07 1.05
0.120 0.120
1.71
0.94
0.117
2.53
0.97
0.116
42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 a
Correlation of each item with older drivers’ accident involvement in the last three years. The 14 items highlighted in gray, with relatively high correlations, were included in the questionnaire.
Table 2 Driver Behavior Questionnaire (DBQ) items adopted in the present study. Factor
Item
Lapses
Forget where he/she left your car in a carpark Intending to drive to destination A, you suddenly notice that you are on the road to destination B, perhaps because B is his/her more usual destination Switch on one thing, such as the headlights, when he/she meant to switch on something else, such as the wipers Hit something when reversing that he/she had not previously seen Total
Errors
Underestimate the speed of an oncoming vehicle when overtaking Brake too quickly on a slippery road, or steer the wrong way into a skid Queuing to turn left onto main road, he/she pay such close attention to the main stream of traffic that you nearly hit the car in front Fail to notice pedestrians crossing on turning into a side road Miss give way signs and narrowly avoid colliding with traffic having right of way On turning left, nearly hit a cyclist who has come up on your inside Attempt to overtake someone you had not noticed to be signaling a right turn Total
Mean
SD
1.82 1.70
0.95 0.88
1.66
0.91
2.01
1.00
7.19
3.17
1.82
0.90
1.57
0.80
1.70
0.85
1.74
0.85
1.56
0.77
1.70
0.82
1.60
0.81
11.68
4.46
Cronbach’s alpha
0.87
0.88
402
Y. Nakagawa et al. / Accident Analysis and Prevention 50 (2013) 397–404 Table 4 Results of logistic regression analysis of proxy-reported accidents over three years as a function of EBQ score and exposure to risk.
Correlaon coefficient 0.3
0.28
0.26
0.24
Variable
B
Intercept Exposure EBQ score
−4.428 0.249* 0.067**
a * **
0.22
rcr
0.2 0
0.05
0.1
0.15
0.2
0.25
a b * **
2
1 0.40** 0.42** 0.29** −0.01
3
1 0.90** 0.17** −0.06
4
1 0.18** −0.05
1.28a 1.97a
Upper
1.01 1.55
1.62 2.49
Odds ratio corresponding to 1 SD increase. p < 0.05. p < 0.01.
0.4
Frequency of Driving
0.2
4 “Every day” 3 “Twice or three mes per week” 2 “Once per week” 1 “Twice or three mes per month”
1
Everyday Behavior Questionnaire. Driver Behavior Questionnaire. p < 0.05. p < 0.01.
0
10
20
30
40
50
60
70
EBQ score
Fig. 2. Estimated logistic curves representing ratios of drivers with accident experiences over the last three years for different frequencies of driving.
0.13 and 0.05, respectively. For various rcr values, we defined a score as the sum of the items whose correlation coefficients were greater than or equal to the rcr , and then calculated the correlation of the score with crash involvement. The result is shown in Fig. 1. The maximum correlation of the score was 0.29, which was attained when rcr = 0.187. In this case, 14 items were selected. Cronbach’s alpha of these 14 items was 0.92, demonstrating a sufficient level of internal consistency. The 488 respondents’ scores ranged between 14 and 70. The average score and SD were 35.0 and 10.1, respectively.
correlation with crash involvement, 0.29, is statistically significant (p < 0.01), confirming that the scale has psychometric validity. The correlations with the two DBQ factors (i.e., 0.40 and 0.42) were also statistically significant (p < 0.01). However, because the validity of the proxy-reported DBQ has not been reported, this fact may not necessarily support the validity of the established scale, to the best of the authors’ knowledge. Next, we conducted logistic regression analysis to explain crash involvement in terms of the EBQ score and the exposure to risk as measured by the frequency of driving. The result is shown in Table 4. It was confirmed that a larger EBQ score and higher frequency of driving were significantly (p < 0.01 and 0.05, respectively) associated with a larger ratio of drivers who have experienced crashes, particularly in the last three years. The odds ratio corresponding to a 1 SD increase in the EBQ score was 1.96. The estimated logistic curves for the four categories of driving frequency are shown in Fig. 2.
3.2. Association between Everyday Behavior Questionnaire (EBQ) score and crash frequency The correlation coefficients of the 14-item EBQ with the four variables (i.e., error and lapse factors of the DBQ scale, crash involvement, and driving exposure) are shown in Table 3. The
Number of drivers 90 80 70 60
Drivers Without Crash Experiences
50 40
Drivers With Crash Experiences
30 20 10 0
EBQ score 10-
Lower
0.6
5
1 0.10**
0.592 0.120 0.012
95% CI
0.8
Table 3 Correlations among the EBQ, two DBQ factors, accident involvement, and exposure. 1
**
Odds ratio
Rao of drivers with crash experiences
Fig. 1. Correlation coefficient between accident involvement and total item scores as a function of rcr .
1 EBQa 2 DBQb (error) 3 DBQb (lapse) 4 Accident involvement 5 Exposure
s.e.
15-
20-
25-
30-
35-
40-
45-
50-
55-
60-
65-
70-
Fig. 3. Frequency distributions of EBQ scores among drivers with and without accident experiences.
Y. Nakagawa et al. / Accident Analysis and Prevention 50 (2013) 397–404
Finally, the EBQ score distributions are plotted in Fig. 3 for the two groups of drivers, i.e., with and without crash experiences in the last three years. As would be expected, EBQ scores were higher for drivers with at least one crash experience. Specifically, a score above 35 would identify approximately 71% of the drivers with crash experiences and 42% of the drivers without crash experiences.
4. Discussion This study aimed to develop an instrument that would predict older drivers’ involvement in crashes on the basis of their family members’ responses, which were based on the drivers’ everyday lives. We successfully developed a 14-item questionnaire with a sufficient level of internal consistency and validity. A number of studies have investigated whether it is possible to predict crash involvement using the DBQ factors. The present analysis shows that the correlation of the EBQ with crash involvement (i.e., 0.29) is not weaker than that of the DBQ, as reported in a number of studies, in spite of the fact that the EBQ is not a self-report questionnaire and does not include items directly related to driving. In fact, according to a meta-analysis by deWinter and Dodou (2010), the estimated overall zero-order correlation of error and violation factors of the DBQ among samples of 23 earlier studies were 0.10 and 0.13, respectively. The EBQ is not as predictive of poor driving performance as the battery of computerized tests adopted by McKnight and McKnight (1999). They applied a computerized measure of 22 abilities required by elderly drivers (aged 62 and above), and showed that a total score based on the 22 ability measures correctly identified 80% of drivers with experiences of unsafe incident involvement, while misidentifying only 20% of the incident-free drivers. In contrast, in the present study, if we set a cut-off score that identifies 80% of incident-involved drivers, 55% of incident-involved drivers are misidentified. Inversely, if we set a cut-off score that misidentifies 20% of the incident-free drivers, we can identify only 44%. This fact does not necessarily imply that the questionnaire developed in the present study is of no value. In contrast to many existing measures, the EBQ is easy to use, it does not require special knowledge or equipment, and it can be completed quickly. Therefore, this instrument can serve as a screening tool that can be used in advance of more complex instruments. To determine a cut-off score for identifying older drivers who need to be tested by the more complex instruments, it may be useful to consider the curves shown in Fig. 2. On the basis of the information about the frequency of driving and the EBQ score, we can estimate the probability that the older drivers are involved in crashes in the three-year period. However, it should be noted that these curves are sensitive to the sample. This study is based on a highly selected sample, and thus it will be seminal to repeat the study with a more representative and larger sample. Regarding the social implementation of the EBQ, it is important to note two points. First, this instrument should be used in a spontaneous way by elderly drivers’ family members who are concerned about the safety of the elderly drivers, rather than by the government to identify unsafe elderly drivers. The reason is that family members’ perceptions that their responses could result in the suspension of driving licenses may bias their responses. Practitioners such as medical doctors and professional caregivers, who are likely to be asked for advice by the family members, could use this instrument to judge whether or not a more complex assessment of driving ability is needed. The government might encourage a broad use of this instrument by these practitioners and elderly drivers’ family members. Second, the correlation of 0.29 between EBQ score and crash involvement means that the survey responses account for
403
only 8.4% of the variance in crash involvement. It might be beneficial for existing instruments such as computerized tests to be used in conjunction with the EBQ in order to identify unsafe drivers more accurately than is currently possible. The present study has four important limitations. First, there are various types of crashes, such as collision with other vehicles and hitting cyclists, pedestrians, and other objects. In addition, there are various types of collisions, such as in the same or opposite direction, vehicle turning, and intersecting paths (National Highway Traffic Safety Administration, 2005). This study did not consider this variation. Second, in the present study, the respondents shared various types of relationships with their older drivers, such as sons, daughters, and spouses. Therefore, it is possible that families with different relationships respond to the questionnaire differently. Future studies should investigate whether the respondents’ relationship to the elderly drivers influences the reliability of their responses. The third limitation relates to the generalizability of the questionnaire outside Japan. Certain items in Table 1 seem to be valid only in the context of Japan. For example, item No. 15 “It is getting difficult for him/her to judge the right time to walk safely across a street where there is no signal for pedestrians” may be useful for identifying cognitively impaired individuals only in situations where drivers are unlikely to yield to pedestrians crossing the road, as is the case in Japan. This is the case because if drivers always yield, then even cognitively impaired people can cross the street safely. Fourth, the correlation found in the present study between crash involvement and each item of the EBQ might be biased to some extent. This was because when family members were completing the survey, they were aware of their elderly relative’s history of crash involvement. This bias might have affected the procedure for selecting items to be included in EBQ. Future research might be able to avoid this problem either by excluding items from EBQ that seems more likely to be biased, or by conducting a longitudinal survey and checking the association between each item and crash involvement after, rather than before, completion of the questionnaire. Items that are less likely to be biased are those that describe the behaviors of the elderly people in specific situations that are different from driving situations. An example is item No. 8 “In supermarkets, it is getting difficult for him/her to pay attention to customers passing by in the background due to his/her attention to commodities in the shelves.” In contrast, items such as No. 2 “He/She has slower reflexes than he/she used to have” is likely to be biased because past behaviors of elderly people while driving cars, including crash experiences, may well influence their family members’ responses. Thus, it will be important in a future study to compose items that can predict the frequency of each type of crash and identify respondents to this questionnaire whose answers are the most predictive of crash involvement. Moreover, it will be significant to identify the subset of candidate items in Table 1 that best suits the context or minimizes the bias. Acknowledgments We would like to thank Mr. Hiroshi Miyamoto (deputy director of Nankoku Chuo Hospital, Kochi Prefecture, Japan) for providing us with valuable information on higher brain dysfunction. Some of the items in the EBQ were created on the basis of this information. References Bassuk, S.S., Glass, T.A., Berkman, L.F., 1999. Social disengagement and incident cognitive decline in community-dwelling elderly persons. Annals of Internal Medicine 131, 165–173.
404
Y. Nakagawa et al. / Accident Analysis and Prevention 50 (2013) 397–404
Burkhardt, J.E., Berger, A.M., Creedon, M., McGravock, A.T., 1998. Mobility and independence: Changes and challenges for older drivers. Ecosometrics, Bethesda, MD. Carp, F.M., 1988. Significance of mobility for the well-being of the elderly. Transportation in an Aging Society, vol. II. Transportation Research Board, National Academy of Science, Washington, D.C, pp. 1–20. Culter, S.J., 1975. Transportation and changes in life satisfaction. Gerontologist 15, 155–159. deWinter, J.C.F., Dodou, D., 2010. The driver behavior questionnaire as a predictor of accidents: a meta-analysis. Journal of Safety Research 41, 463–470. Eisenhandler, S.A., 1990. The asphalt identikit: Old age and the drivers’ license. International Journal of Aging & Human Development 30 (1), 1–14. Evans, E.L., 2001. Influences on mobility among non-driving older Americans. Personal travel: The long and short of it. Transportation Research E-Circular, No. E-C026, 151–168. Fonda, S.J., Wallace, R.B., Herzog, A.R., 2001. Changes in driving patterns and worsening depressive symptoms among older adults. Journal of Gerontology: Social Sciences 56B (6), S343–S351. Frittelli, C., Borghetti, D., Iudice, G., Bonanni, E., Maestri, M., Tognoni, G., Pasquali, L., Iudice, A., 2009. Effects of Alzheimer’s disease and mild cognitive impairment on driving ability: a controlled clinical study by simulated driving test. International Journal of Geriatric Psychiatry 24, 232–238. Koss, E., Patterson, M.B., Ownby, R., Stuckey, J.C., Whitehouse, P.J., 1993. Memory evaluation in Alzheimer’s disease: Caregivers’ appraisals and objective testing. Archives of Neurology 50, 92–97. Lawton, R., Parker, D., Manstead, A.S.R., Stradling, S.G., 1997. The role of affect in predicting social behaviors: the case of road traffic violations. Journal of Applied Social Psychology 27, 1258–1276. Lyman, S., Ferguson, S.A., Braver, E.R., Williams, A.F., 2002. Older driver involvements in police reported accidents and fatal accidents: trends and projections. Injury Prevention 8, 116–120. Marottoli, R.A., Mendes de Leon, C.R., Glass, T.A., Williams, C.S., Cooney Jr., L.M., Berkman, L.F., 2000. Consequences of driving cessation: Decreased out-of-home activity levels. Journal of Gerontology: Social Sciences 55B (6), S334–S340. Massie, D.L., Campbell, K.L., Williams, A.F., 1995. Traffic accident involvement rates by driver age and gender. Accident Analysis and Prevention 27, 78–87. Marottoli, R.A., Mendes de Leon, C.R., Glass, T.A., Williams, C.S., Cooney Jr., L.M., Berkman, L.F., Tinetti, M.E., 1997. Driving cessation and increased depressive symptoms: prospective evidence from the New Haven EPESE. Journal of the American Geriatrics Society 45, 202–206.
Mathias, J.L., Lucas, L.K., 2009. Cognitive predictors of unsafe driving in older drivers: a meta analysis. International Psychogeriatrics 21 (4), 637–653. McGwin, G., Brown, D.B., 1999. Characteristics of traffic accidents among young, middle-aged, and older drivers. Accident Analysis and Prevention 31, 181–198. McKnight, A.L., McKnight, A.S., 1999. Multivariate analysis of age-related driver ability and performance deficits. Accident Analysis and Prevention 31, 445–454. National Highway Traffic Safety Administration, 2005. The National Automotive Sampling System (NASS) General Estimates System (GES): The Analytical User’s Manual 1988–2003. National Highway Traffic Safety Administration, U.S. Department of Transportation. Parker, D., Reason, J.T., Manstead, A.S.R., Stradling, S.G., 1995a. Driving errors, driving violations and accident involvement. Ergonomics 38, 1036–1048. Parker, D., West, R.W., Stradling, S.G., Manstead, A.S.R., 1995b. Behavioral characteristics and involvement in different types of traffic accident. Accident Analysis and Prevention 27, 571–581. Parker, D., McDonald, L., Rabbitt, P., Sutcliffe, P., 2000. Elderly drivers and their accidents: the Aging Driver Questionnaire. Accident Analysis and Prevention 32, 751–759. Reason, J.T., Manstead, A.S.R., Strandling, S.G., Baxter, J., Campbell, K., 1990. Errors and violations on the road: a real distinction? Ergonomics 33, 1315–1332. Retchin, S.M., Anapolle, J., 1993. An overview of the older driver. Clinical Geriatric Medicine 9, 279–295. Riege, W.H., 1982. Self-report and tests of memory aging. Clinical Gerontologist 1, 23–36. Rosenbloom, S., 2001. Driving cessation among older people: when does it happen and what impact does it have? Transportation Research Record 1779, 93–99. Stutts, J.C., Wilkins, J.W., 2003. On-road driving evaluations: a potential tool for helping older adults drive safely longer. Journal of Safety Research 34, 431–439. Turano, K.A., Munoz, B., Hassan, S.E., Duncan, D.D., Gower, E.W., Roche, K.B., Keay, L., Munro, C.A., West, S.K., 2009. Poor sense of direction is associated with constricted driving space in older drivers. Journal of Gerontology: Psychological Sciences 64B (3), 348–355. Wagner, J.T., Muri, R.M., Nef, T., Mosimann, U.P., 2011. Cognition and driving in older persons. Swiss Medical Weekly 140, w13136. Zuin, D., Ortiz, H., Doromei, D., Lopez, O.L., 2002. Motor vehicle accidents and abnormal driving behaviours in patients with dementia in Mendoza, Argentina. European Journal of Neurology 9, 29–34.
Contents lists available at SciVerse ScienceDirect
Accident Analysis and Prevention journal homepage: www.elsevier.com/locate/aap
Elderly drivers’ everyday behavior as a predictor of crash involvement—Questionnaire responses by drivers’ family members Yoshinori Nakagawa a,∗ , Kaechang Park b , Yasuhiko Kumagai b a b
Department of Management, Research Organization for Regional Alliance, Kochi University of Technology, Japan Research Organization for Regional Alliance, Kochi University of Technology, Japan
a r t i c l e
i n f o
Article history: Received 13 January 2012 Received in revised form 26 April 2012 Accepted 4 May 2012 Keywords: Elderly drivers Crash risk Cognitive impairment Driver Behavior Questionnaire (DBQ) Everyday Behavior Questionnaire (EBQ)
a b s t r a c t When or whether elderly drivers stop driving is concerning not only to the drivers themselves but also to their family members. Therefore, it is important for family members to take the initiative if they wish to obtain information on the likelihood of the drivers’ involvement in crashes. On the basis of the older drivers’ Everyday Behavior Questionnaire (EBQ) developed in this paper, we attempt to predict drivers’ involvement in crashes using the responses given by their family members. The results revealed that this 14-item questionnaire has a sufficient level of internal consistency as well as a significant correlation (r = 0.29) with the experience of involvement in crashes in the last three years (p < 0.01). Although the EBQ is a proxy-reported questionnaire and does not include items directly related to driving behaviors, the correlation between the EBQ and crash involvement is stronger than that of the self-reported Driver Behavior Questionnaire reported in deWinter and Dodou (2010), who conducted a meta-analysis and estimated the overall correlation among samples of earlier studies. In addition, logistic regression analysis showed that the EBQ score and the exposure to driving risks, measured by the frequency of driving, are significant predictors of involvement in crashes. © 2012 Elsevier Ltd. All rights reserved.
1. Introduction 1.1. Background Previous studies have shown that older drivers have higher crash rates per vehicle mile of travel (McGwin and Brown, 1999; Retchin and Anapolle, 1993; Massie et al., 1995). With the projected increase in the elderly driving population worldwide (Parker et al., 2000; Lyman et al., 2002), including Japan, the problem of an increasing number of crashes involving elderly drivers is attracting the attention of many researchers and practitioners. When addressing this problem, it is important to note that driving restrictions and cessation threaten older adults’ lives. According to Stutts and Wilkins (2003) and other authors, these conditions lead to consequences such as (i) loss of mobility (Evans, 2001; Marottoli et al., 2000; Rosenbloom, 2001), (ii) loss of identity and increased dependency (Burkhardt et al., 1998; Carp, 1988; Culter, 1975; Eisenhandler, 1990), (iii) physical and mental problems such as increased depression, heart disease, fractures, and stroke (Bassuk et al., 1999; Fonda et al., 2001; Marottoli et al., 1997),
∗ Corresponding author at: 185 Miyanokuchi, Kami City, Kochi Prefecture, Japan. Tel.: +81 887 57 2768; fax: +81 887 57 2124. E-mail address: [email protected] (Y. Nakagawa). 0001-4575/$ – see front matter © 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.aap.2012.05.014
and (iv) decreased social integration, as measured by number and frequency of social contacts (Turano et al., 2009). It should also be noted that when or whether elderly drivers stop driving is of great concern to their family members for two reasons. First, if elderly drivers suffer the above-mentioned negative consequences on cessation of driving, their families are likely to be burdened psychologically, physically, and economically. Second, family members do not want their loved ones to be involved in crashes, and often hope for their driving cessation. Thus, it is important that family members take the initiative, if they wish, to obtain information on the likelihood of the elderly relatives getting involved in crashes and to gain a stance on driving cessation.
1.2. Objective Against this background, this study develops an older drivers’ Everyday Behavior Questionnaire (EBQ). This questionnaire, answered by older drivers’ family members, attempts to predict drivers’ involvement in crashes. This study also checks its internal consistency as well as the validity by calculating its correlation with the drivers’ involvement in crashes in the last three years and with two factors of the proxy-reported Driver Behavior Questionnaire (DBQ), lapses and errors. The EBQ has two characteristics. First, this questionnaire is answered by older drivers’ family members who live with the
398
Y. Nakagawa et al. / Accident Analysis and Prevention 50 (2013) 397–404
drivers and observe their everyday lives. Thus, family members can independently answer the questionnaire without any technical know-how and within a short span of time. Second, as the EBQ consists of items not directly related to driving, respondents are not required to be regular passengers in the elderly drivers’ cars. 1.3. Related earlier studies Several lines of earlier research have aimed at predicting the performance of drivers, specifically those who are elderly. One line of research investigated the association between driving performance and requisite cognitive functions for driving which were assessed using various tests, such as the Ergovision test, the Useful Field of View (UFOV) test, a Complex Reaction Time Task, the Benton Line Orientation task, Clock Drawing, and the Trail Making test. Mathias and Lucas (2009) conducted a meta-analysis of these earlier studies and concluded that a variety of tests appear suitable for screening older drivers, the exact choice of which depends on the “gold standard” for determining driving ability (on-road driving, simulator performance, driving problems). It should be noted that although a combination of such ability measures can successfully identify risky drivers (e.g., McKnight and McKnight, 1999), these tests are burdensome to drivers, and may not be easy to adopt, especially when drivers are unwilling to retire from the roadways. A second line of research examined the association between driving performance and neurological disorders that older people are more likely to suffer, such as stroke, dementia, and mild cognitive impairment (e.g., Frittelli et al., 2009; Zuin et al., 2002). Many of these studies investigated the association between driving performance (on-road driving or simulator performance) and commonly used screening tests for detecting dementia and cognitive impairment, such as the Clinical Dementia Rating (CDR) and Mini Mental State Examination (MMSE). However, these two scores are not without limitations: the CDR is time-consuming and is rarely used outside of specialist clinics, and MMSE scores correlate poorly with driving performance (Wagner et al., 2011); the reason for this poor correlation is discussed by Frittelli et al. (2009). A third line of research explored the association between drivers’ involvement in crashes and their driving behavior patterns. Among the typologies of aberrant driving behaviors, the DBQ is one of the most popular instruments. A number of earlier studies have aimed to predict drivers’ involvement in crashes using the questionnaire’s subscales, such as errors (mistakes that have potentially dangerous consequences), lapses (primarily attentional failures), and violations (risky driving behaviors). In a meta-analysis of these studies, deWinter and Dodou (2010) concluded that the violation and error factors of the DBQ predicted crashes with an overall correlation of 0.13 and 0.10, respectively. Parker et al. (2000) applied the DBQ to drivers aged 50 and above, and found that, in an older sample, relatively high scores on the error and lapse factors were predictive of involvement in an active accident. It should be noted that these studies are based on a self-report questionnaire, and it is not clear to the authors of the present paper whether this approach is applicable to drivers who are suspected of having a certain degree of cognitive impairment. It is expected that the proposed instrument will complement the above-mentioned existing measures in predicting aged drivers’ performance. 2. Methods and materials 2.1. Participants Data were collected via an internet research company, Cross Marketing, Inc., whose registered members were invited by to
participate in the preliminary survey. Potential participants were asked if they satisfied the following four conditions. Those who satisfied all four were invited to participate in the main survey. Condition 1: He or she lives in the same house as the elderly family member aged 70 or more. Condition 2: The elderly relative drives at least twice or three times per month. Condition 3: He or she wants the elderly relative to cease driving. Condition 4: He or she rides as a passenger in the car driven by the elderly family member at least once per two or three months. Condition 3 was checked by the item “Do you want him/her to cease driving?”, and the respondents were asked to choose from one of the five alternatives: 1 = Agree, 2 = Agree to some extent, 3 = No opinion, 4 = Disagree to some extent, and 5 = Disagree. Respondents who chose options 1 or 2 were considered to satisfy the given condition. Condition 4 was included to check the validity of the scale developed in this study, although the EBQ proposed in this study was designed such that respondents were not required to be regular passengers in elderly drivers’ cars. If the scale exhibits validity, then scores should be associated with the frequency of lapses and errors reported by the elderly drivers’ family who have been passengers. Among those who qualified for the preliminary survey, 488 people agreed to participate in the main survey. There were 268 (54.9%) males and 220 (45.1%) females, with an average age of 50.0 (SD 14.8). The average age of their elderly family members was 76.3 (SD 4.6). More than half of them (55.7%) were the respondents’ parents and 22.1% of them the respondents’ spouses. The numbers of drivers with no, one, two, and three or more crash experiences in the last three years were 289, 116, 47, and 36, respectively.
2.2. Materials 2.2.1. Characteristics of respondents and their elderly family drivers Socio-demographic and other questions were included in the questionnaire to determine (i) respondents’ age and gender, (ii) age, gender, and former and current employment status of the respondents’ elderly family members, (iii) driving history of the elderly family members (in years), (iv) driving exposure or frequency of driving of the elderly family members (4 = Every day, 3 = Twice or three times per week, 2 = Once per week, 1 = Twice or three times per month), (v) frequency with which the respondents ride as passengers in the cars driven by the elderly family members (4 = Once or more per week, 3 = Once per two or three weeks, 2 = Once per month, 1 = Once per two or three months), and (vi) the elderly family member’s involvement in crashes in the last three years (4 = Three times or more, 3 = Twice, 2 = Once, 1 = None). Involvement in crashes was measured by the item “In the past three years, has your elderly family member caused crashes (incidents with other vehicles, pedestrians, cyclists, or objects, or more serious crashes) while driving a car?” Although the present study could have used a yes or no response, a four-point scale was adopted, ranging from 1 (None) to 4 (Three times or more), and respondents who answered “1” were regarded as having no crash experiences. This was expected to enhance the reliability of the item. With regard to driving exposure, most earlier studies asked drivers about kilometers (miles) driven in a specific period using a self-report questionnaire survey. We did not adopt this method of measuring exposure, because this study involved a proxy-reported survey, and reporting the total distance driven by their family members would be difficult for the respondents.
Y. Nakagawa et al. / Accident Analysis and Prevention 50 (2013) 397–404
2.2.2. Everyday behavior measures We created a total of 100 items describing everyday behaviors of elderly people, some of which were adopted from earlier studies to predict their driving performance. Using a five-point scale (from 5 = Strongly yes to 1 = Not at all), respondents were asked to rate the extent to which each item described their family members. The 100 items were developed in the following manner. Many authors have classified abilities required in driving; among them, McKnight and McKnight (1999) identified fourteen abilities and classified them into five categories: sensory, attentional, perceptual, cognitive, and psychomotor. We considered the implications of impairment of each of the abilities on older people’s everyday lives and, where possible, developed plural items for each ability. For example, in the case of declining attentional abilities, older people may find it difficult to concentrate on tasks in their daily lives and to drive safely. On the basis of this assumption, we created items such as “It is getting difficult for him/her (your elderly family member) to concentrate on a specific task (e.g., hobbies such as reading books) for as long as he/she used to” and “It is getting difficult for him/her to stay in the same place for as long as he/she used to.” Certain items were adopted from earlier studies; for example, with regard to short-term memory (a cognitive ability), items from the Short Memory Questionnaire (SMQ; Riege, 1982; Koss et al., 1993) were utilized after minor re-wording. The SMQ is a scale consisting of 14 items to assess memory problems of older people. The scale is rated by the elderly drivers’ family caregivers, and its scores have a significant association with the MMSE, which is widely used in clinical practice for screening dementia. An example of such items is “He/She always remembers where he/she put his/her keys.” Finally, some items were adapted from the medical literature describing the symptoms of higher brain dysfunction. As detailed later, 64 of the 100 items were correlated with crashes at a significance level of 0.01, and the top 14 items with relatively higher correlations were included in the questionnaire. The procedure for determining the number of items to be included in the questionnaire will be detailed in Sections 2.3.1 and 3.1. The list of the 64 items is shown in Table 1, with the average score and SD of each item as well as the correlation of each item with crash involvement. 2.2.3. Driver Behavior Questionnaire (DBQ) The Manchester Driver Behavior Questionnaire (DBQ; Reason et al., 1990; Parker et al., 1995a,b; Lawton et al., 1997) consists of 20 items, of which 11 items were modified and adapted. The item selection criteria were as follows. First, many earlier studies that applied factor analysis to the items in the DBQ identified three factors: errors, lapses, and violations. We adopted 16 items relating to errors or lapses, because it has been reported that only items relating to these factors were associated with crash involvement, as far as elderly drivers were concerned (Parker et al., 2000). Second, four items among the sixteen were omitted, because it seemed inappropriate to include them in a proxy-reported questionnaire. For example, the item “Fail to check rearview mirror before pulling out, changing lanes, etc” was omitted, because family passengers do not necessarily recognize whether a driver has checked the mirrors. Third, in addition to these four items, the item “Misread signs and took the wrong turning off a roundabout” was omitted, because drivers seldom encounter roundabouts in Japan. As a consequence of this selection procedure, four items on lapses and seven items on errors were adopted in the present study. Finally, minor re-wordings were made for these eleven items: the word “you” was replaced by “he/she,” denoting the respondent’s elderly family member. The list of the eleven items is shown in Table 2, with mean scores and SD for the 488 respondents. Cronbach’s alpha coefficients of the two factors (i.e., errors and lapses) were 0.87 and 0.88,
399
respectively. Using a six-point scale (from 6 = Nearly all the time to 1 = Never), respondents were asked to rate the extent to which each item described their family members. 2.3. Analysis 2.3.1. Development of Everyday Behavior Questionnaire (EBQ) As stated above, a total of 100 items were used to measure how effectively the elderly family members carry out activities of daily living. We compiled the questionnaire by choosing items that distinguish efficiently between safe and risky drivers. Specifically, we calculated the Pearson product–moment correlation coefficient between the score on each item and the frequency of crashes in the last three years. Then, we chose items whose correlation coefficients were larger than the critical value rcr . Further, the EBQ score was defined as the total score of these selected items. The critical value was defined to maximize the correlation coefficient between the total score and the frequency of crashes. If rcr is too small, then many irrelevant items would be included in the questionnaire, and the correlation coefficient would become small. Moreover, if rcr is too large, the correlation coefficient would again become small, in spite of the strong relevance of the selected items, because the number of items would be small. For this reason, it was expected that the critical value rcr should have an optimal value. We will show later that rcr can indeed be optimized. After determining items to be included in the EBQ, we checked the internal consistency of the set of items by calculating Cronbach’s alpha coefficient. The procedure for defining the critical value rcr will be explained in greater detail in Section 3.1. It should be noted that the frequency at which an elderly driver causes crashes depends not only on his/her driving ability but also on the exposure to risk (i.e., the distance driven by him/her during a specific period). Therefore, to establish that the procedure described is a sound method of choosing items that represent driving ability, it is necessary to confirm the assumption that no items are associated more strongly than others with the exposure to risk. Our analysis confirmed that among the 100 items, only two items were significantly associated with exposure (p < 0.05). Moreover, in regard to these items, the absolute values of the correlation coefficients were very small (0.12 and 0.09, respectively). This implies that the assumption described above is valid and our methodology was sound. 2.3.2. Association between Everyday Behavior Questionnaire (EBQ) score and crash frequency To check the validity of the EBQ, we first calculated correlation coefficients to examine the association of this scale with three external criteria: frequency of crashes in the last three years and the two DBQ factors, lapses and errors. Second, we studied how precisely the scale can screen elderly drivers who are likely to cause crashes. Specifically, we established a multivariate logistic regression model to explain the frequency of crashes in terms of the EBQ score and the exposure to risk. In multivariate logistic regression analysis, objective variables must be binary. In the present study, elderly divers with more than one experience of crash involvement in the last three years were coded 1, and the others were coded zero. As it was assumed that demographic variables such as drivers’ gender and age affect the frequency of crashes only through affecting the EBQ score, they were not included in the model as explanatory variables. 3. Results 3.1. Development of Everyday Behavior Questionnaire (EBQ) The correlation coefficients of the 100 items with crash involvement ranged between −0.04 and 0.25, and the average and SD were
400
Y. Nakagawa et al. / Accident Analysis and Prevention 50 (2013) 397–404
Table 1 14 items and 50 candidate items of the Everyday Behavior Questionnaire (EBQ). No. 1 2 3 4
5 6 7 8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32
33 34 35 36 37 38 39 40
Item
Mean
SD
Correlationa
He/she misses or nearly misses a step on the stairs more frequently than he/she used to in his/her house and other facilities. He/she has slower reflexes than he/she used to have. It is becoming difficult for him/her to think or talk logically. It is becoming difficult for him/her to judge appropriately whether to go across pedestrian crossings or stop and wait for a fresh “walk” sign, considering his/her walking speed and the time remaining before the signal changes. It is becoming difficult for him/her to stay in the same place for as long as he/she used to. He/she acts on the basis of delusions more often than he/she used to. It is becoming difficult for him/her to hear/understand what you are talking about, unless you talk loudly. In supermarkets, it is becoming difficult for him/her to pay attention to customers passing by in the background due to his/her attention on commodities in the shelves. It is becoming difficult for him/her to distinguish words of the people he/she is talking with, if there are noises or other conversations going on around him/her. He/she gets bored more easily than he/she used to. It is becoming difficult for him/her to kill mosquitoes flying around or on his/her skin. He/she watches TV turned up louder than he/she used to. It is becoming difficult for him/her to imagine how others feel in response to his/her own behavior. He/she has poorer eyesight than he/she used to have. It is becoming difficult for him/her to judge the right time to walk safely across the street where there is no signal for pedestrians. He/she behaves at his/her own pace more frequently than he/she used to. It is becoming difficult to ask him/her to watch over a kettle or a pot being heated when he/she is watching TV. He/she forgets to turn off the electricity and/or gas more frequently than he/she used to. He/she wrongly fastens buttons or fails to straighten a collar more frequently than he/she used to. It seems more difficult than it used to be to ask him/her to turn off the gas ten minutes after the water in a pot has boiled. It is becoming difficult for him/her to settle troubles between friends or neighbors. It is becoming difficult to pay attention when others talk while he/she is working with his/her hands, e.g., needle work and writing. While walking, he/she does not pay enough attention to cars and bicycles more frequently than he/she used to. He/she does not move as quickly as he/she used to. He/she watches steps more carefully when walking in the dark/outside than he/she used to. He/she behaves in an impromptu manner more frequently than he/she used to. It seems more difficult than before for him/her to judge what to do in his/her house when a natural calamity, such as an earthquake, occurs. In unexpected circumstances, it is becoming difficult for him/her to cope with a problem without panicking. When things do not go as planned, he/she goes blank more often than he/she used to. It seems difficult for him/her to perceive insects flying or crawling around than before. It seems more difficult for him/her to tell whether an insect flying around him/her is a mosquito. When walking around in the house, he/she hits his/her arms or legs on the corner of tables, walls, and posts more frequently than he/she used to. He/she adamantly expects younger people around him/her to behave in conformity with him/her than was formerly the case. It is becoming difficult for him/her to put tasks in an orderly sequence and perform them. He/she is in a foggy state more frequently than was formerly the case. He/she is losing patience. It is becoming difficult for him/her to care about issues not immediately present. He/she sometimes prefers not to be surrounded by younger people who worry about him/her. He/she is becoming indifferent to what others think or feel He/she more frequently hesitate to make risky choices.
2.25
0.95
0.247
3.02 2.35 1.87
0.98 1.01 0.83
0.237 0.210 0.210
2.56
1.1
0.204
2.73 2.69
1.1 1.19
0.199 0.198
2.2
0.97
0.198
1.99
1.06
0.196
2.4 2.48
1.06 0.95
0.195 0.192
2.88 2.62
1.24 1.09
0.190 0.190
2.98 1.91
1.02 0.88
0.187 0.186
2.85
0.98
0.185
2.26
1.09
0.183
2.31
1.03
0.182
1.82
0.87
0.181
2.14
1.01
0.179
2.17
0.97
0.178
2.38
1.01
0.178
2.33
0.95
0.176
3.05 2.68
1.03 1
0.175 0.174
1.92
1.05
0.168
2.37
1
0.167
2.51
0.97
0.166
2.3
1.02
0.164
2.3
0.95
0.164
2.15
1.05
0.163
2.26
0.95
0.159
2.43
0.98
0.158
1.82
0.99
0.156
2.31 2.51 1.98
1.09 1.09 1.06
0.155 0.155 0.152
2.72
1.06
0.151
2.02 2.31
1.11 0.96
0.150 0.149
Y. Nakagawa et al. / Accident Analysis and Prevention 50 (2013) 397–404
401
Table 1 (Continued) No.
Item
Mean
SD
Correlationa
41
It is becoming difficult for him/her to manage money and financial affairs (e.g., paying bills, budgeting). It is becoming difficult for him/her to make necessary arrangements and to perform tasks efficiently. He/she more frequently gets into tangles with people whom he/she walks past. When called out to, he/she more frequently looks this way and that to find identify who is talking to him/her. It is becoming difficult for him/her to turn his/her neck and look quickly behind himself/herself. He/she is becoming less willing to tolerate present annoyances for the sake of a future result It takes longer for him/her to comprehend sentences in books or newspapers. It is becoming difficult for him/her to remember what someone asked him/her to do. It is becoming difficult for him/her to find what he/she needs among many objects in a room. He/she has poorer vision in the dark than previously. Once he/she begins a task, it takes over his/her entire attention He/she more frequently becomes agitated if he/she can’t find something right away. It is becoming difficult for him/her to mount an escalator. He/she increasingly struggles to find the words he/she wants to use. He/she seems increasingly vulnerable to sales solicitations or unable to recognize deceptive appeals. It is becoming difficult for him/her to converse while the TV is on. It is becoming difficult for him/her to follow the motion of quickly moving objects, such as flying insects and TV captions. When walking outside, he/she walks slowly and carefully and annoys passers-by more frequently than before. When walking around, he/she more frequently loses his/her way, even in familiar areas. He/she is increasingly optimistic in his/her interpretation of the situations in which he/she is placed. He/she more frequently forgets where he/she put keys. It is becoming difficult for him/her to shop for groceries without a list and without forbecoming items (5 items). It seems more difficult for him/her to remember birthdays of family members. Sunlight and lights are more disconcerting for him/her than before.
1.65
0.96
0.147
2.01
1.07
0.144
1.78
0.92
0.144
1.88
0.98
0.144
2.43
0.95
0.142
1.96
1.07
0.142
2.43
1.11
0.140
1.95
1.04
0.137
2.03
1.09
0.136
2.67 2.05 2.44
0.98 1.04 1.03
0.135 0.134 0.132
1.9 2.22 2.07
1.1 1.09 0.93
0.128 0.128 0.128
1.9 2.63
1.01 0.94
0.127 0.127
2.12
0.92
0.122
1.75
0.95
0.120
2.94
1.07
0.120
2.11 2.03
1.07 1.05
0.120 0.120
1.71
0.94
0.117
2.53
0.97
0.116
42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 a
Correlation of each item with older drivers’ accident involvement in the last three years. The 14 items highlighted in gray, with relatively high correlations, were included in the questionnaire.
Table 2 Driver Behavior Questionnaire (DBQ) items adopted in the present study. Factor
Item
Lapses
Forget where he/she left your car in a carpark Intending to drive to destination A, you suddenly notice that you are on the road to destination B, perhaps because B is his/her more usual destination Switch on one thing, such as the headlights, when he/she meant to switch on something else, such as the wipers Hit something when reversing that he/she had not previously seen Total
Errors
Underestimate the speed of an oncoming vehicle when overtaking Brake too quickly on a slippery road, or steer the wrong way into a skid Queuing to turn left onto main road, he/she pay such close attention to the main stream of traffic that you nearly hit the car in front Fail to notice pedestrians crossing on turning into a side road Miss give way signs and narrowly avoid colliding with traffic having right of way On turning left, nearly hit a cyclist who has come up on your inside Attempt to overtake someone you had not noticed to be signaling a right turn Total
Mean
SD
1.82 1.70
0.95 0.88
1.66
0.91
2.01
1.00
7.19
3.17
1.82
0.90
1.57
0.80
1.70
0.85
1.74
0.85
1.56
0.77
1.70
0.82
1.60
0.81
11.68
4.46
Cronbach’s alpha
0.87
0.88
402
Y. Nakagawa et al. / Accident Analysis and Prevention 50 (2013) 397–404 Table 4 Results of logistic regression analysis of proxy-reported accidents over three years as a function of EBQ score and exposure to risk.
Correlaon coefficient 0.3
0.28
0.26
0.24
Variable
B
Intercept Exposure EBQ score
−4.428 0.249* 0.067**
a * **
0.22
rcr
0.2 0
0.05
0.1
0.15
0.2
0.25
a b * **
2
1 0.40** 0.42** 0.29** −0.01
3
1 0.90** 0.17** −0.06
4
1 0.18** −0.05
1.28a 1.97a
Upper
1.01 1.55
1.62 2.49
Odds ratio corresponding to 1 SD increase. p < 0.05. p < 0.01.
0.4
Frequency of Driving
0.2
4 “Every day” 3 “Twice or three mes per week” 2 “Once per week” 1 “Twice or three mes per month”
1
Everyday Behavior Questionnaire. Driver Behavior Questionnaire. p < 0.05. p < 0.01.
0
10
20
30
40
50
60
70
EBQ score
Fig. 2. Estimated logistic curves representing ratios of drivers with accident experiences over the last three years for different frequencies of driving.
0.13 and 0.05, respectively. For various rcr values, we defined a score as the sum of the items whose correlation coefficients were greater than or equal to the rcr , and then calculated the correlation of the score with crash involvement. The result is shown in Fig. 1. The maximum correlation of the score was 0.29, which was attained when rcr = 0.187. In this case, 14 items were selected. Cronbach’s alpha of these 14 items was 0.92, demonstrating a sufficient level of internal consistency. The 488 respondents’ scores ranged between 14 and 70. The average score and SD were 35.0 and 10.1, respectively.
correlation with crash involvement, 0.29, is statistically significant (p < 0.01), confirming that the scale has psychometric validity. The correlations with the two DBQ factors (i.e., 0.40 and 0.42) were also statistically significant (p < 0.01). However, because the validity of the proxy-reported DBQ has not been reported, this fact may not necessarily support the validity of the established scale, to the best of the authors’ knowledge. Next, we conducted logistic regression analysis to explain crash involvement in terms of the EBQ score and the exposure to risk as measured by the frequency of driving. The result is shown in Table 4. It was confirmed that a larger EBQ score and higher frequency of driving were significantly (p < 0.01 and 0.05, respectively) associated with a larger ratio of drivers who have experienced crashes, particularly in the last three years. The odds ratio corresponding to a 1 SD increase in the EBQ score was 1.96. The estimated logistic curves for the four categories of driving frequency are shown in Fig. 2.
3.2. Association between Everyday Behavior Questionnaire (EBQ) score and crash frequency The correlation coefficients of the 14-item EBQ with the four variables (i.e., error and lapse factors of the DBQ scale, crash involvement, and driving exposure) are shown in Table 3. The
Number of drivers 90 80 70 60
Drivers Without Crash Experiences
50 40
Drivers With Crash Experiences
30 20 10 0
EBQ score 10-
Lower
0.6
5
1 0.10**
0.592 0.120 0.012
95% CI
0.8
Table 3 Correlations among the EBQ, two DBQ factors, accident involvement, and exposure. 1
**
Odds ratio
Rao of drivers with crash experiences
Fig. 1. Correlation coefficient between accident involvement and total item scores as a function of rcr .
1 EBQa 2 DBQb (error) 3 DBQb (lapse) 4 Accident involvement 5 Exposure
s.e.
15-
20-
25-
30-
35-
40-
45-
50-
55-
60-
65-
70-
Fig. 3. Frequency distributions of EBQ scores among drivers with and without accident experiences.
Y. Nakagawa et al. / Accident Analysis and Prevention 50 (2013) 397–404
Finally, the EBQ score distributions are plotted in Fig. 3 for the two groups of drivers, i.e., with and without crash experiences in the last three years. As would be expected, EBQ scores were higher for drivers with at least one crash experience. Specifically, a score above 35 would identify approximately 71% of the drivers with crash experiences and 42% of the drivers without crash experiences.
4. Discussion This study aimed to develop an instrument that would predict older drivers’ involvement in crashes on the basis of their family members’ responses, which were based on the drivers’ everyday lives. We successfully developed a 14-item questionnaire with a sufficient level of internal consistency and validity. A number of studies have investigated whether it is possible to predict crash involvement using the DBQ factors. The present analysis shows that the correlation of the EBQ with crash involvement (i.e., 0.29) is not weaker than that of the DBQ, as reported in a number of studies, in spite of the fact that the EBQ is not a self-report questionnaire and does not include items directly related to driving. In fact, according to a meta-analysis by deWinter and Dodou (2010), the estimated overall zero-order correlation of error and violation factors of the DBQ among samples of 23 earlier studies were 0.10 and 0.13, respectively. The EBQ is not as predictive of poor driving performance as the battery of computerized tests adopted by McKnight and McKnight (1999). They applied a computerized measure of 22 abilities required by elderly drivers (aged 62 and above), and showed that a total score based on the 22 ability measures correctly identified 80% of drivers with experiences of unsafe incident involvement, while misidentifying only 20% of the incident-free drivers. In contrast, in the present study, if we set a cut-off score that identifies 80% of incident-involved drivers, 55% of incident-involved drivers are misidentified. Inversely, if we set a cut-off score that misidentifies 20% of the incident-free drivers, we can identify only 44%. This fact does not necessarily imply that the questionnaire developed in the present study is of no value. In contrast to many existing measures, the EBQ is easy to use, it does not require special knowledge or equipment, and it can be completed quickly. Therefore, this instrument can serve as a screening tool that can be used in advance of more complex instruments. To determine a cut-off score for identifying older drivers who need to be tested by the more complex instruments, it may be useful to consider the curves shown in Fig. 2. On the basis of the information about the frequency of driving and the EBQ score, we can estimate the probability that the older drivers are involved in crashes in the three-year period. However, it should be noted that these curves are sensitive to the sample. This study is based on a highly selected sample, and thus it will be seminal to repeat the study with a more representative and larger sample. Regarding the social implementation of the EBQ, it is important to note two points. First, this instrument should be used in a spontaneous way by elderly drivers’ family members who are concerned about the safety of the elderly drivers, rather than by the government to identify unsafe elderly drivers. The reason is that family members’ perceptions that their responses could result in the suspension of driving licenses may bias their responses. Practitioners such as medical doctors and professional caregivers, who are likely to be asked for advice by the family members, could use this instrument to judge whether or not a more complex assessment of driving ability is needed. The government might encourage a broad use of this instrument by these practitioners and elderly drivers’ family members. Second, the correlation of 0.29 between EBQ score and crash involvement means that the survey responses account for
403
only 8.4% of the variance in crash involvement. It might be beneficial for existing instruments such as computerized tests to be used in conjunction with the EBQ in order to identify unsafe drivers more accurately than is currently possible. The present study has four important limitations. First, there are various types of crashes, such as collision with other vehicles and hitting cyclists, pedestrians, and other objects. In addition, there are various types of collisions, such as in the same or opposite direction, vehicle turning, and intersecting paths (National Highway Traffic Safety Administration, 2005). This study did not consider this variation. Second, in the present study, the respondents shared various types of relationships with their older drivers, such as sons, daughters, and spouses. Therefore, it is possible that families with different relationships respond to the questionnaire differently. Future studies should investigate whether the respondents’ relationship to the elderly drivers influences the reliability of their responses. The third limitation relates to the generalizability of the questionnaire outside Japan. Certain items in Table 1 seem to be valid only in the context of Japan. For example, item No. 15 “It is getting difficult for him/her to judge the right time to walk safely across a street where there is no signal for pedestrians” may be useful for identifying cognitively impaired individuals only in situations where drivers are unlikely to yield to pedestrians crossing the road, as is the case in Japan. This is the case because if drivers always yield, then even cognitively impaired people can cross the street safely. Fourth, the correlation found in the present study between crash involvement and each item of the EBQ might be biased to some extent. This was because when family members were completing the survey, they were aware of their elderly relative’s history of crash involvement. This bias might have affected the procedure for selecting items to be included in EBQ. Future research might be able to avoid this problem either by excluding items from EBQ that seems more likely to be biased, or by conducting a longitudinal survey and checking the association between each item and crash involvement after, rather than before, completion of the questionnaire. Items that are less likely to be biased are those that describe the behaviors of the elderly people in specific situations that are different from driving situations. An example is item No. 8 “In supermarkets, it is getting difficult for him/her to pay attention to customers passing by in the background due to his/her attention to commodities in the shelves.” In contrast, items such as No. 2 “He/She has slower reflexes than he/she used to have” is likely to be biased because past behaviors of elderly people while driving cars, including crash experiences, may well influence their family members’ responses. Thus, it will be important in a future study to compose items that can predict the frequency of each type of crash and identify respondents to this questionnaire whose answers are the most predictive of crash involvement. Moreover, it will be significant to identify the subset of candidate items in Table 1 that best suits the context or minimizes the bias. Acknowledgments We would like to thank Mr. Hiroshi Miyamoto (deputy director of Nankoku Chuo Hospital, Kochi Prefecture, Japan) for providing us with valuable information on higher brain dysfunction. Some of the items in the EBQ were created on the basis of this information. References Bassuk, S.S., Glass, T.A., Berkman, L.F., 1999. Social disengagement and incident cognitive decline in community-dwelling elderly persons. Annals of Internal Medicine 131, 165–173.
404
Y. Nakagawa et al. / Accident Analysis and Prevention 50 (2013) 397–404
Burkhardt, J.E., Berger, A.M., Creedon, M., McGravock, A.T., 1998. Mobility and independence: Changes and challenges for older drivers. Ecosometrics, Bethesda, MD. Carp, F.M., 1988. Significance of mobility for the well-being of the elderly. Transportation in an Aging Society, vol. II. Transportation Research Board, National Academy of Science, Washington, D.C, pp. 1–20. Culter, S.J., 1975. Transportation and changes in life satisfaction. Gerontologist 15, 155–159. deWinter, J.C.F., Dodou, D., 2010. The driver behavior questionnaire as a predictor of accidents: a meta-analysis. Journal of Safety Research 41, 463–470. Eisenhandler, S.A., 1990. The asphalt identikit: Old age and the drivers’ license. International Journal of Aging & Human Development 30 (1), 1–14. Evans, E.L., 2001. Influences on mobility among non-driving older Americans. Personal travel: The long and short of it. Transportation Research E-Circular, No. E-C026, 151–168. Fonda, S.J., Wallace, R.B., Herzog, A.R., 2001. Changes in driving patterns and worsening depressive symptoms among older adults. Journal of Gerontology: Social Sciences 56B (6), S343–S351. Frittelli, C., Borghetti, D., Iudice, G., Bonanni, E., Maestri, M., Tognoni, G., Pasquali, L., Iudice, A., 2009. Effects of Alzheimer’s disease and mild cognitive impairment on driving ability: a controlled clinical study by simulated driving test. International Journal of Geriatric Psychiatry 24, 232–238. Koss, E., Patterson, M.B., Ownby, R., Stuckey, J.C., Whitehouse, P.J., 1993. Memory evaluation in Alzheimer’s disease: Caregivers’ appraisals and objective testing. Archives of Neurology 50, 92–97. Lawton, R., Parker, D., Manstead, A.S.R., Stradling, S.G., 1997. The role of affect in predicting social behaviors: the case of road traffic violations. Journal of Applied Social Psychology 27, 1258–1276. Lyman, S., Ferguson, S.A., Braver, E.R., Williams, A.F., 2002. Older driver involvements in police reported accidents and fatal accidents: trends and projections. Injury Prevention 8, 116–120. Marottoli, R.A., Mendes de Leon, C.R., Glass, T.A., Williams, C.S., Cooney Jr., L.M., Berkman, L.F., 2000. Consequences of driving cessation: Decreased out-of-home activity levels. Journal of Gerontology: Social Sciences 55B (6), S334–S340. Massie, D.L., Campbell, K.L., Williams, A.F., 1995. Traffic accident involvement rates by driver age and gender. Accident Analysis and Prevention 27, 78–87. Marottoli, R.A., Mendes de Leon, C.R., Glass, T.A., Williams, C.S., Cooney Jr., L.M., Berkman, L.F., Tinetti, M.E., 1997. Driving cessation and increased depressive symptoms: prospective evidence from the New Haven EPESE. Journal of the American Geriatrics Society 45, 202–206.
Mathias, J.L., Lucas, L.K., 2009. Cognitive predictors of unsafe driving in older drivers: a meta analysis. International Psychogeriatrics 21 (4), 637–653. McGwin, G., Brown, D.B., 1999. Characteristics of traffic accidents among young, middle-aged, and older drivers. Accident Analysis and Prevention 31, 181–198. McKnight, A.L., McKnight, A.S., 1999. Multivariate analysis of age-related driver ability and performance deficits. Accident Analysis and Prevention 31, 445–454. National Highway Traffic Safety Administration, 2005. The National Automotive Sampling System (NASS) General Estimates System (GES): The Analytical User’s Manual 1988–2003. National Highway Traffic Safety Administration, U.S. Department of Transportation. Parker, D., Reason, J.T., Manstead, A.S.R., Stradling, S.G., 1995a. Driving errors, driving violations and accident involvement. Ergonomics 38, 1036–1048. Parker, D., West, R.W., Stradling, S.G., Manstead, A.S.R., 1995b. Behavioral characteristics and involvement in different types of traffic accident. Accident Analysis and Prevention 27, 571–581. Parker, D., McDonald, L., Rabbitt, P., Sutcliffe, P., 2000. Elderly drivers and their accidents: the Aging Driver Questionnaire. Accident Analysis and Prevention 32, 751–759. Reason, J.T., Manstead, A.S.R., Strandling, S.G., Baxter, J., Campbell, K., 1990. Errors and violations on the road: a real distinction? Ergonomics 33, 1315–1332. Retchin, S.M., Anapolle, J., 1993. An overview of the older driver. Clinical Geriatric Medicine 9, 279–295. Riege, W.H., 1982. Self-report and tests of memory aging. Clinical Gerontologist 1, 23–36. Rosenbloom, S., 2001. Driving cessation among older people: when does it happen and what impact does it have? Transportation Research Record 1779, 93–99. Stutts, J.C., Wilkins, J.W., 2003. On-road driving evaluations: a potential tool for helping older adults drive safely longer. Journal of Safety Research 34, 431–439. Turano, K.A., Munoz, B., Hassan, S.E., Duncan, D.D., Gower, E.W., Roche, K.B., Keay, L., Munro, C.A., West, S.K., 2009. Poor sense of direction is associated with constricted driving space in older drivers. Journal of Gerontology: Psychological Sciences 64B (3), 348–355. Wagner, J.T., Muri, R.M., Nef, T., Mosimann, U.P., 2011. Cognition and driving in older persons. Swiss Medical Weekly 140, w13136. Zuin, D., Ortiz, H., Doromei, D., Lopez, O.L., 2002. Motor vehicle accidents and abnormal driving behaviours in patients with dementia in Mendoza, Argentina. European Journal of Neurology 9, 29–34.