- Email: [email protected]
Passenger carriage and car crash injury: a comparison between younger and older drivers
Accident Analysis and Prevention 35 (2003) 861–867
Passenger carriage and car crash injury: a comparison between younger and older drivers Lawrence T. Lam a,b,∗ , Robyn Norton b , Mark Woodward b , Jennie Connor c , Shanthi Ameratunga c a
The Royal Alexandra Hospital for Children, Locked Bag 4001, Sydney, Westmead NSW 2145, Australia b Institute for International Health, University of Sydney, Sydney, Australia c Department of Community Health, University of Auckland, Auckland, New Zealand Received 15 April 2002; received in revised form 9 July 2002; accepted 11 July 2002
Abstract This study was conducted to investigate the effects of passenger carriage, including the number of passengers and the ages of passengers, on the risk of car crash injury. The study utilised data obtained from a case–control study conducted in the Auckland region of New Zealand between 1998 and 1999. Cases were car drivers who involved in crashes in which at least one occupant was hospitalised or killed. Controls were selected from a cluster random sample of car drivers on the roads in the same region. Self-report information on the numbers of passengers carried and their ages at the time of crash or at the time of the roadside survey, as well as potential confounding factors, was obtained from the drivers, or a proxy, using an interviewer-administered questionnaire. A total of 571 cases (93% response rate), including 195 younger drivers (aged <25 years), and 588 controls (79% response rate), including 94 younger drivers participated in the study. After adjusting for other risk factors, the odds of car crash injury among younger drivers was 15.55 times (95% CI 5.76–42.02) for those who carried two or more same age passengers, and 10.19 times (95% CI 2.84–36.65) for those who carried two or more other age passengers, compared with unaccompanied drivers. In comparison, no increase in risk was observed for older drivers who carried two or more passengers regardless of age. The carriage of two or more passengers, irrespective of the ages of passengers, significantly increases the risk of car crash injury among younger drivers. Passenger restriction as part of the graduate licensing system was discussed in the light of these results. © 2003 Elsevier Science Ltd. All rights reserved. Keywords: Passenger carriage; Young drivers; Risk factors; Car crash injury
1. Introduction Much effort has been invested in the search for risk and protective factors for car crashes and injuries among young people in the last few decades. Among these, the influence of passenger carriage has been suggested as an important risk factor. There is a growing body of literature examining the relationship between passenger carriage and the risk of car crash and injury (Foldvary and Lane, 1969; Drummond and Triggs, 1991; Williams and Wells, 1995; Williams et al., 1995; Ulmer et al., 1997; Preusser et al., 1998; Doherty et al., 1998; Chen et al., 2000). A close examination of the more recent epidemiological studies (Preusser et al., 1998; Ulmer et al., 1997; Doherty et al., 1998; Chen et al., 2000) indicates that, though increasingly sophisticated statistical techniques and analyti∗
Corresponding author. Tel.: +61-2-9845-3055; fax: +61-2-9845-3082. E-mail address: [email protected] (L.T. Lam).
cal methodologies were used, they were all studies utilising routinely collected official reports on road fatalities and injuries. Among these, three studies (Williams et al., 1995; Preusser et al., 1998; Chen et al., 2000) investigated the risk of fatal crashes only. The remaining two (Ulmer et al., 1997; Doherty et al., 1998) included other crashes resulting in personal injuries and property damages. The relative risks of passenger carriage and car crashes or crash-related injuries were calculated in three studies (Preusser et al., 1998; Doherty et al., 1998; Chen et al., 2000) using comparison groups consisted of crashed drivers who were in the lowest risk group (Preusser et al., 1998), or crashed drivers of the same ages who did not carry any passengers (Doherty et al., 1998; Chen et al., 2000). Although, all these studies suggested that passenger carriage increases the risk of car crash injury particularly among young people, there were methodological problems that pose limitations on the risk estimation as well as the interpretation of results obtained. First, the risk estimations obtained from
0001-4575/$ – see front matter © 2003 Elsevier Science Ltd. All rights reserved. PII: S 0 0 0 1 - 4 5 7 5 ( 0 2 ) 0 0 0 9 1 - X
862
L.T. Lam et al. / Accident Analysis and Prevention 35 (2003) 861–867
the comparison between deaths and injured cases against those non-injured crash cases, as in most of these studies, did not represent the risk in the population. They were risk estimations over and above the amount of risk experienced by drivers who had crashed when the particular risk factor (in this case, passenger carried) was absent. Thus, the research hypothesis tested in these studies was not whether passenger carriage would increase the risk of car crash or crash-related injury in the population. Second, data utilising routinely collected information is averaged over individuals, thus, the degree of association between exposure and outcome might not represent individual-level associations. Moreover, as reflected in the above-mentioned studies, adjustment for potential confounding factors at the individual-level was inadequate. This would further distort the already-imprecise estimation of association. Studies of this design are appropriate for generating research hypothesis, but not for testing pre-defined hypothesis (Rothman and Greenland, 1998). Third, due to the limitation of being non-population-based studies, it rendered any calculation of population attributable risk (PAR) inappropriate. Other study designs, such as the case–control study, need to be employed in order to produce more reliable relative risk estimation. No case–control study has been conducted, so far, in this important area of young driver road safety. The aim of the present study is to present the results obtained from a case–control study conducted in the Australasian region on the effects of the numbers and ages of passengers carried, on the risk of car crash injury, with specific focus on the comparison between younger and older drivers.
2. Methods The method has been detailed previously by Connor et al. (2002). In brief, a population-based case–control study was conducted in the Auckland region of New Zealand, between April 1998 and July 1999. The region included urban, suburban and rural areas and a population of approximately 1 million, and the source population for study participants comprised drivers of light vehicles driving on public roads. Eligible vehicles included cars, vans and light utility vehicles, but excluded all vehicles licensed as heavy vehicles, taxis, and emergency vehicles. The definitions of geographical boundaries, time period, eligible vehicles, and eligible roads were applied in an identical manner to cases and controls. The study was approved by the Northern Regional Health Authority Ethics Committee. Informed consent was obtained from all participants for personal participation and use of records. 2.1. Selection of cases All drivers or passengers in eligible vehicles who were hospitalised or died as the result of a car crash in the study region were identified prospectively through daily surveil-
lance and case-finding in the region’s four trauma hospitals and single Coroner’s office. Hospital trauma teams, emergency department staff, and the New Zealand Police collaborated to ensure comprehensive case-finding. In each case, the driver of the vehicle was the key informant, whether or not the driver themselves was injured. 2.2. Selection of controls Controls were a sample of car drivers representative of all time spent driving on the region’s roads during the study period. They were identified by cluster sampling of drivers at 69 randomly selected sites on the road network. The day of week, time of day and direction of travel for each survey site were randomly assigned. Eligible vehicles passing each survey site during a 2-h period were randomly sampled in proportion to the traffic volume at the site, which had been determined by earlier measurements. Cars were selected one at a time, from traffic that had been slowed as it approached the site. On the busiest roads, we selected the first car identified after the previous one had been processed. At less busy sites, sampling was slowed by waiting for a predetermined period (2, 5, or 15 min depending on traffic volume) between finishing with one car and selecting the next. Residual differences in sampling fractions between survey sites were adjusted for in the analysis, by weighting the clusters of controls by the inverse of the sampling fraction from their site. Surveys were carried out at an average of one per week and recruitment approximately matched accrual of cases. When vehicles selected as controls did not stop, or when it was not possible to stop traffic safely (e.g. motorway sites), vehicle owners were identified through registration plates and telephone directories, so that drivers could be invited to participate in the study. 2.3. Data collection Interviews with case drivers were conducted face-to-face in hospital, or by telephone at home. Proxy interviews were sought for drivers who sustained fatal injuries, or were too ill to participate. For control drivers, contact details, suitable interview times and a breath test for alcohol were obtained at the roadside recruitment sites, and interviews conducted by telephone. The majority of case and control interviews were carried out in the first 48 h after the crash or survey, although follow-up calls were sometimes necessary. The highly structured interview was based on a questionnaire covering the circumstances of the current trip and many usual behaviours and background characteristics of the drivers. In particular, respondents were asked to record the number of passengers carried in the car, and their ages at the time of crash or survey. Information on a range of other variables including demographics, time of crash/interview, self-reported alcohol consumption 6 h within crash/interview, the average kilometres driven per week, and drivers’ sleepiness as measured by Stanford Sleepiness Scale was also collected. These
L.T. Lam et al. / Accident Analysis and Prevention 35 (2003) 861–867
variables were considered as potential confounders in the relationship between passenger carriage and the risk of car crash injury. They were selected as confounders based on the information suggested from the literature.
863
were conducted by subgroups of younger (<25 years) and older (≥25 years) drivers. Logistic regression analyses were conducted to calculate the unadjusted and adjusted odds ratios (OR) and their corresponding 95% confidence intervals (95% CI) for the associations between passenger carriage and the case–control status. All analyses were conducted using the SUDAAN statistical analysis software (Shah et al., 1997) to handle the cluster sampling effect of data collected on the controls. The population attributable risks (PAR) were also calculated using the method detailed by Bruzzi et al. (1985). This method provided a means to calculate the PAR after adjusting for other potential risk factors.
2.4. Data analyses For the purpose of comparison, the age groups of drivers were classified into two, the younger driver (<25 years) and the older driver (≥25 years). In the analyses, the number of passengers was classified as none, one and two or more. The passenger ages were defined by the ages of passengers carried in relation to the ages of drivers. A “same age” passenger was defined as any passenger whose age was in the same age group as the driver, otherwise the passenger was classified as “other age” passenger. Thus, the “other age” passenger category included passengers whose ages were older than the drivers or a mixture of older and same age passengers. As it was anticipated that the number of passengers carried and passenger ages are highly correlated, a composite variable, namely passenger carriage, was created for both cases and controls, to include both the numbers of passengers and passenger ages. This included single drivers, one same age passenger only, two or more same age passengers, one other passenger only, two or more other passengers. For the comparison between younger and older drivers, data analyses
3. Results A total of 571 cases including 195 (34.2%) younger drivers and 588 controls including 94 (14.1%) younger drivers participated in the study. The response rate for the cases was 93% and that for the controls was 79%. For both younger and older control drivers, nearly 63% were driving alone at the time of survey (Table 1). In comparison, 59% of older drivers involved in crashes were driving alone whereas only 34% of younger drivers who involved in crashes were single drivers (Table 1). The frequency
Table 1 Frequency distributions of passenger carriage, demographic and potential confounding variables by age, and case and control status <25 years Cases (n = 195) Passenger carriage Single driver One same age passenger only ≥2 same age passengers One other passenger only ≥2 other passengers Potential confounding variables Sex Female Male
66 43 40 13 33
(33.9) (22.1) (20.4) (6.7) (16.9)
≥25 years Controls (n = 94)
Cases (n = 376)
Controls (n = 494)
59 15 6 6 7
220 53 6 32 65
297 84 9 36 66
(64.3) (16.2) (6.8) (6.6) (6.1)
(58.5) (14.1) (1.6) (8.5) (17.3)
(62.0) (15.9) (1.6) (7.4) (12.2)
53 (27.2) 142 (72.8)
38 (44.0) 56 (56.0)
145 (38.6) 231 (61.4)
188 (40.9) 306 (59.1)
81 (41.5) 114 (58.5)
17 (10.0) 77 (90.0)
95 (25.3) 281 (74.7)
50 (7.0) 444 (93.0)
Self-reported alcohol consumption (6 h prior) No/1 drink 148 (75.9) ≥2 47 (24.1)
88 (95.9) 6 (4.1)
309 (82.2) 67 (17.8)
474 (97.4) 20 (2.6)
Average driving per week <100 km 100–300 km 300–500 km >500 km
20 33 21 10
51 190 55 40
75 204 120 81
Night-time driving 9:00 p.m.–5:59 a.m. Other
Driver sleepiness No Yes
37 94 24 8
(22.7) (57.7) (14.7) (4.9)
141 (72.7) 53 (27.3)
Column percentage of controls was adjusted for sampling design.
(18.1) (44.0) (26.4) (11.5)
90 (97.2) 4 (2.8)
(15.2) (56.6) (16.3) (11.9)
306 (81.6) 69 (18.4)
(15.7) (46.7) (22.5) (15.1)
488 (98.5) 6 (1.5)
864
L.T. Lam et al. / Accident Analysis and Prevention 35 (2003) 861–867
distributions of the potential confounders were also presented in Table 1. Table 2 summarises the relationships between passenger carriage, potential confounders, and the risk of car crash injury by age groups. There was a significant unadjusted association between passenger carriage and the risk of car crash injury for younger drivers. The odds of car crash injury increased by 2.5 times (crude OR = 2.51 (1.22–5.15)) for younger drivers carrying one same age passenger only, 5.5 times (crude OR = 5.57 (2.25–13.78)) carrying two or more same age passengers, and more than five times (crude OR = 5.18 (1.61–16.61)) carrying two or more other passengers, each compared to unaccompanied driving. On the other hand, the unadjusted association between passenger carriage and the risk of crash injury for older driver was insignificant. Results obtained from the univariate analyses also indicted that sex of driver, night-time driving, self-reported alcohol consumption, and driver sleepiness are significantly related to car crash injury among younger drivers (Table 2). In comparison, alcohol consumption and driver sleepiness were associated significantly with crash injury among older drivers. To elucidate the confounding status of those potential confounding variables, further analyses were conducted on the relationship between these variables and passenger carriage (Table 3). Results indicated that most of these variables were associated with passenger carriage except gender and self-reported alcohol consumption for younger drivers and night-time driving for older drivers. All these variables were considered as potential confounding factors to be adjusted for in following analyses. The results obtained from the multiple logistic regression were in agreement with those obtained from the univariate
Table 2 Associations between passenger carriage, potential confounder, and car crash injury by ages of drivers Crude OR (95% CI)
Passenger carriage Single driver One same age passenger only ≥2 same age passengers One other passenger only ≥2 other passengers
<25 yearsa
≥25 years
1.00 2.59 5.77 1.91 5.19
1.00 0.96 1.06 1.24 1.53
(1.26–5.33) (2.32–14.33) (0.64–5.68) (1.66–16.18)
Potential confounding variables Sex Female 1.00 Male 2.18 (1.17–4.05) Night-time driving Other 9:00 p.m.–5:59 a.m.
(0.58–1.59) (0.34–3.30) (0.68–2.26) (0.93–2.50)
1.00 1.10 (0.81–1.48)
1.00 1.00 6.21 (1.31–29.56) 4.53 (0.90–22.79)
Self-reported alcohol consumption previous 6 h No/1 drink 1.00 1.00 ≥2 7.28 (2.06–25.71) 8.14 (3.59–18.47) Average driving per week <100 km 100–300 km 300–500 km >500 km
1.00 1.07 (0.52–2.23) 0.44 (0.21–0.90) 0.34 (0.11–1.06)
Driver sleepiness No Yes
1.00 1.00 13.22 (3.32–52.64) 15.15 (5.78–39.72)
a
1.00 1.25 (0.81–1.93) 0.75 (0.44–1.31) 0.82 (0.45–1.52)
Crude ORs were adjusted for sampling design.
Table 3 Associations between potential confounding variables and passenger carriage by ages of drivers Passenger carriage <25 years
≥25 years
χ12 = 2.98, P > 0.05
χ12 = 8.36, P < 0.05
Night-time driving Other 9:00 p.m.–5:59 a.m.
χ12 = 4.88, P < 0.05
χ12 = 2.49, P > 0.05
Self-reported alcohol consumption previous 6 h No/1 drink ≥2
χ12 = 3.26, P > 0.05
χ12 = 4.40, P < 0.05
χ32 = 12.53, P < 0.05
χ32 = 23.69, P < 0.05
χ12 = 5.29, P < 0.05
χ12 = 5.21, P < 0.05
Potential confounding variables Sex Female Male
Average driving per week <100 km 100–300 km 300–500 km >500 km Driver sleepiness No Yes
L.T. Lam et al. / Accident Analysis and Prevention 35 (2003) 861–867 Table 4 Adjusteda odds ratios (95% CI) for car crash injury by passenger carriage for younger (<25 years) and older (≥25 years) drivers Passenger carriage
Single driver One same age passenger only ≥2 same age passengers One other passenger only >2 other passengers
Adjusted OR <25 years
≥25 years
1.00 2.39 (0.91–6.29) 15.55 (5.76–42.02) 3.490 (0.69–17.79) 10.19 (2.84–36.65)
1.00 0.78 0.30 1.19 1.33
(0.42–1.46) (0.06–1.44) (0.63–2.22) (0.78–2.27)
a ORs were adjusted for sex, night-time driving, self-reported alcohol consumption, average driving per week and driver sleepiness.
analyses. As shown in Table 4, the model was significant (P < 0.001) with passenger carriage and five potential confounding variables. Similar to the unadjusted associations, younger drivers who carried more than one passenger tended to have an increased risk of car crash injury for both same age and other passengers when compared with single drivers (Table 4). However, no equivalent increase of risk was observed among older drivers. The odds of car crash injury increased by about 16 times for younger drivers who carried two or more young passengers (OR = 15.55 (5.76–42.02)). The odds was also increased by 10 times for those who carried two or more other type of passengers (OR = 10.19 (2.84–36.65)) as compared with single drivers of the same age group. The population attributable risk associated with passenger carriage with two or more passengers, disregarding their ages, was 33.3% (95% CI 27.8–40.0) as compared to those driving alone or carrying one passenger for younger drivers. Calculation of the PAR for older drivers was not attempted since it had been shown that passenger carriage is not associated with crash injury.
4. Discussion The results obtained in this study have demonstrated that the number of passengers carried is an important risk factor for car crash injury among younger drivers. Carrying more than one passenger has significantly increased the risk of car crash injury for younger drivers, but not for older drivers. Furthermore, such an increase in the risk occurred not just for carrying same age passengers, but also passengers of other ages. These results, in general, are consistent with the finding in other studies (Doherty et al., 1998; Chen et al., 2000). Moreover, they suggest that, after adjusting for the risk factors identified in the literature, the effect of the number of passengers carried, rather than the ages of passengers, is more significant on the risk of car crash injury among younger drivers. This is the first study utilising a case–control study design in investigating the relationship between passenger carriage and the risk of car crash injury. All other studies are
865
ecological in nature. As mentioned, the study design has posed a restriction on the accuracy of risk estimations on the association between passenger carriage and car crash injury. The results obtained in this study provided odds ratios calculated after adjusting for important potential confounding risk factors suggested in the literature (Rothman and Greenland, 1998). Another strength of the study is that selection bias has been minimised. All cases occurring within the study region and study period have been identified. The cluster sampling methodology employed to recruit controls has provided a random sample of the driving population at risk in the same geographic location of the cases. The controls are comparable to the cases in terms of the distributions in the demographic and other variables of interest, except that the proportion of controls is lower in the younger group than that in the older. Reasonably, high response rates were achieved for both cases and controls. Information bias has also been minimised by using a standardised questionnaire, interview protocol and trained interviewers. The outcome of each case has been accurately reported and documented with detailed information gathered. Potential bias, particularly recall bias on the exposure measure, may exist for both cases and controls. The recall on the number of passengers carried in the vehicle at the time of crash or interview is likely to be accurate, except for those severely injured cases that required proxy respondents. However, the recall of the passengers’ ages is subjected to bias due to unreliable memory or inaccurate estimation. Such a bias may not be series for those passengers whose ages are obviously older or younger. The ages of passengers between 20- and 30-year-old are more likely to be either under or over estimated. Other recall biases may also occur in the assessment of other variables, such as self-reported alcohol consumption. This is a common source of bias in most epidemiological studies that rely upon self-reported information. The importance of the results obtained in this study is in its implication in the theoretical conceptualisation of the relationship between passenger carriage and the risk of car crash injury among young drivers. It has been suggested that the key to understanding such a relationship lies in a better comprehension of the social interaction or some other dynamics between the passenger and the driver (Regan and Mitsopoulos, 2001). It has been noted that the drivers carrying passengers will adjust their driving behaviour according to the social expectations of the passengers perceived by the drivers (Baxter et al., 1990). Baxter et al. (1990) further observed that young drivers, particularly males, tended to drive at excessive speeds in the presence of male passengers. This observation has been strengthened by the results obtained from the study by Mckenna et al. (1998) that young drivers, both male and female, drove faster and accept shorter gaps at junctions when accompanied by young male passengers. These results are consistent with the findings of the Chen’s study mentioned above (Chen et al., 2000). Thus, it seems that carrying passengers, possibly
866
L.T. Lam et al. / Accident Analysis and Prevention 35 (2003) 861–867
young passengers, particularly young men has an detrimental effect on the road safety of younger drivers such that the presence of these passengers induces greater risky driving behaviour. Another school of thought proposes that the link between passenger carriage and car crash is that the presence of passengers may be a distraction to the driver. These distractions, as suggested by Williams (2000), might result from the greater verbal interaction, music playing or even physical interactions between younger drivers and their passengers, especially same age passengers. This seems to be supported by the finding of the recent study that passengers of all age groups perceived themselves to be more likely to talk to younger than older drivers (Regan and Mitsopoulos, 2001). This results in a reduction of attention on driving and, thus, compromises safety. This is probably due to the fact that as drivers are distracted there is an increase in the load of their cognitive functioning and processing speed. As reported by Lamble et al. (1999) an increase in cognitive load impairs the ability of drivers to detect changes in the environment, thus, increasing the chance of a collision. This may further interact with inexperience in employing compensatory speed-reducing action for the loss of attention among younger drivers. This results in a greater increase in the risk of crash among younger drivers. The results obtained in this study that the number of passengers carried, rather than passenger types, tended to be a more significant factor in influencing the risk of car crash injury among young drivers has lent further support to this view. Increased number of passengers, despite their ages, may have increased in-vehicle distractions and, thus, increased the chance of collision and injury. The findings of this study also have practical implication in terms of road safety policy making. It has been demonstrated that graduated licensing systems, in various forms, are effective measures in reducing car crash among novice young drivers (Langley et al., 1996; Ulmer et al., 2000; Foss et al., 2001; Shope et al., 2001). However, it remains uncertain as whether several components, such as night-time curfew and passenger restrictions, should be included in such licensing systems. At present, the graduated licensing systems in New Zealand and Australia pose no restrictions on passenger carriage. The results obtained in this study have indicated carrying more than one passenger is detrimental to younger drivers’ safety. These results can most probably be applied to those younger novice drivers who are still going through the GLS even though the driving experience has not been included in the analyses. Thus, passenger restrictions could be considered as part of the graduated licensing system. Young drivers who have just obtained their provisional licences might only be allowed to carry only one passenger, and preferably an older passenger. This view has been advocated by some researchers (Chen et al., 2001), and has already been implemented in some jurisdictions (Frith and Perkins, 1992; Williams, 2000). To ascertain confirmative supportive evidence, further study should be
conduct to investigate the differences of risk estimation due to passenger carriage for younger novice and experienced drivers.
References Baxter, J.S., Manstead, A.S.R., Stradling, S.G., Campbell, K.A., Reason, J.T., Parker, D., 1990. Social facilitation and driver behaviour. Brit. J. Psychol. 81, 351–360. Bruzzi, P., Green, S.B., Byar, D.P., Brinton, L.A., Schairer, C., 1985. Estimating the population attributable risk for multiple risk factors using case–control data. Am. J. Epidemiol. 122, 904–914. Chen, L., Baker, S.P., Braver, E.R., Li, G., 2000. Carrying passengers as a risk factor for crashes fatal to 16- and 17-years old drivers. JAMA 283, 1578–1582. Chen, L., Braver, E.R., Baker, S.P., Li, G., 2001. Potential benefits of restrictions on the transport of teenage passengers by 16- and 17-year-old drivers. Injury Prev. 7, 129–134. Connor, J., Norton, R., Ameratunga, S., Robinson, E., Civil, I., Dunn, R., Bailey, J., Jackson, R., 2002. Driver sleepiness and the risk of serious car occupant injury: a population-based case–control study. Brit. Med. J. 324, 1125–1128. Drummond, A.E., Triggs, T.J., 1991. Driving as Skilled Performance: A Perspective for Improving Young Driver Safety. Accident Research Centre, Monash University, Melbourne, Australia. Doherty, S.T., Andrey, J.C., MacGregor, C., 1998. The situational risks of young drivers: the influence of passengers, time of day and day of week on accident rates. Accid. Anal. Prev. 30, 45–52. Foldvary, L.A., Lane, J.C. 1969. Car crash injuries by seating position and miles travelled. In: Proceedings of the 13th Annual Conference of the American Association for Automotive Medicine. American Association for Automotive Medicine, Minneapolis, MN, America. Foss, R.D., Feaganes, J.R., Rodgman, E.A., 2001. Initial effects of graduated driver licensing on 16-year-old driver crashes in North Carolina. J. Am. Med. Assoc. 286, 1588–1592. Frith, W.J., Perkins, W.A., 1992. The New Zealand Graduated Licensing System. Land Transport, Wellington, New Zealand. Lamble, D., Kauranen, T., Laakso, M., Summala, H., 1999. Cognitive load and detection thresholds in car following situations: safety implication for using mobile (cellular) telephones whole driving. Accid. Anal. Prev. 31, 617–623. Langley, J.D., Wagenaar, A.C., Begg, D.J., 1996. An evaluation of the New Zealand graduated driver licensing system. Accid. Anal. Prev. 28, 139–146. Mckenna, F.P., Waylen, A.E., Burkes, M.E., 1998. Male and female driver: how different are they? AA Foundation for Road Safety Research, Baisingstoke, UK. Preusser, D.F., Ferguson, S.A., Williams, A.F., 1998. The effect of teenage passengers on the fatal crash risk of teenage drivers. Accid. Anal. Prev. 30, 217–222. Regan, M.A., Mitsopoulos, E., 2001. Understanding passenger influences on driver behaviour: implications for road safety and recommendations for countermeasure development. Report no. 180. Accident Research Centre, Monash University, Victoria, Australia. Rothman, K.J., Greenland, S., 1998. Modern Epidemiology. Lippincott Williams & Wilkins, Philadelphia, USA. Shah, B.V., Barnwell, B.G., Bieler, G.S., 1997. SUDAAN User’s Manual, Release 7.5. Research Triangle Institute, Research Triangle Park, NC. Shope, J.T., Molnar, L.J., Elliott, M.R., Waller, P.F., 2001. Graduate driver licensing in Michigan: early impact on motor vehicle crashes among 16-year-old drivers. J. Am. Med. Assoc. 286, 1593–1598. Ulmer, R.G., Williams, A.F., Preusser, D.F., 1997. Crash involvement of 16-year-old drivers. Accid. Anal. Prev. 28, 97–103.
L.T. Lam et al. / Accident Analysis and Prevention 35 (2003) 861–867 Ulmer, R.G., Preusser, D.F., Williams, A.F., Ferguson, S.A., Farmer, C.M., 2000. Effect of Florida’s graduated licensing program on the crash rate of teenage driver. Accid. Anal. Prev. 32, 527–532. Williams, A.F., 2000. Teenage passengers in motor vehicle crashes: a summary of current research. Available at: http://www.highwaysafety. org/safety facts/teens/teen passenger.pdf. Accessed October 2001.
867
Williams, A.F., Wells, J.K., 1995. Deaths of teenagers as motor vehicle passengers. J. Safety Res. 26, 161–167. Williams, A.F., Preusser, D.F., Ulmer, R.G., Weinstein, H.B., 1995. Characteristics of fatal crashes of 16-year-old drivers: implications for licensure policies. J. Public Health Policy 16, 347–360.
Passenger carriage and car crash injury: a comparison between younger and older drivers Lawrence T. Lam a,b,∗ , Robyn Norton b , Mark Woodward b , Jennie Connor c , Shanthi Ameratunga c a
The Royal Alexandra Hospital for Children, Locked Bag 4001, Sydney, Westmead NSW 2145, Australia b Institute for International Health, University of Sydney, Sydney, Australia c Department of Community Health, University of Auckland, Auckland, New Zealand Received 15 April 2002; received in revised form 9 July 2002; accepted 11 July 2002
Abstract This study was conducted to investigate the effects of passenger carriage, including the number of passengers and the ages of passengers, on the risk of car crash injury. The study utilised data obtained from a case–control study conducted in the Auckland region of New Zealand between 1998 and 1999. Cases were car drivers who involved in crashes in which at least one occupant was hospitalised or killed. Controls were selected from a cluster random sample of car drivers on the roads in the same region. Self-report information on the numbers of passengers carried and their ages at the time of crash or at the time of the roadside survey, as well as potential confounding factors, was obtained from the drivers, or a proxy, using an interviewer-administered questionnaire. A total of 571 cases (93% response rate), including 195 younger drivers (aged <25 years), and 588 controls (79% response rate), including 94 younger drivers participated in the study. After adjusting for other risk factors, the odds of car crash injury among younger drivers was 15.55 times (95% CI 5.76–42.02) for those who carried two or more same age passengers, and 10.19 times (95% CI 2.84–36.65) for those who carried two or more other age passengers, compared with unaccompanied drivers. In comparison, no increase in risk was observed for older drivers who carried two or more passengers regardless of age. The carriage of two or more passengers, irrespective of the ages of passengers, significantly increases the risk of car crash injury among younger drivers. Passenger restriction as part of the graduate licensing system was discussed in the light of these results. © 2003 Elsevier Science Ltd. All rights reserved. Keywords: Passenger carriage; Young drivers; Risk factors; Car crash injury
1. Introduction Much effort has been invested in the search for risk and protective factors for car crashes and injuries among young people in the last few decades. Among these, the influence of passenger carriage has been suggested as an important risk factor. There is a growing body of literature examining the relationship between passenger carriage and the risk of car crash and injury (Foldvary and Lane, 1969; Drummond and Triggs, 1991; Williams and Wells, 1995; Williams et al., 1995; Ulmer et al., 1997; Preusser et al., 1998; Doherty et al., 1998; Chen et al., 2000). A close examination of the more recent epidemiological studies (Preusser et al., 1998; Ulmer et al., 1997; Doherty et al., 1998; Chen et al., 2000) indicates that, though increasingly sophisticated statistical techniques and analyti∗
Corresponding author. Tel.: +61-2-9845-3055; fax: +61-2-9845-3082. E-mail address: [email protected] (L.T. Lam).
cal methodologies were used, they were all studies utilising routinely collected official reports on road fatalities and injuries. Among these, three studies (Williams et al., 1995; Preusser et al., 1998; Chen et al., 2000) investigated the risk of fatal crashes only. The remaining two (Ulmer et al., 1997; Doherty et al., 1998) included other crashes resulting in personal injuries and property damages. The relative risks of passenger carriage and car crashes or crash-related injuries were calculated in three studies (Preusser et al., 1998; Doherty et al., 1998; Chen et al., 2000) using comparison groups consisted of crashed drivers who were in the lowest risk group (Preusser et al., 1998), or crashed drivers of the same ages who did not carry any passengers (Doherty et al., 1998; Chen et al., 2000). Although, all these studies suggested that passenger carriage increases the risk of car crash injury particularly among young people, there were methodological problems that pose limitations on the risk estimation as well as the interpretation of results obtained. First, the risk estimations obtained from
0001-4575/$ – see front matter © 2003 Elsevier Science Ltd. All rights reserved. PII: S 0 0 0 1 - 4 5 7 5 ( 0 2 ) 0 0 0 9 1 - X
862
L.T. Lam et al. / Accident Analysis and Prevention 35 (2003) 861–867
the comparison between deaths and injured cases against those non-injured crash cases, as in most of these studies, did not represent the risk in the population. They were risk estimations over and above the amount of risk experienced by drivers who had crashed when the particular risk factor (in this case, passenger carried) was absent. Thus, the research hypothesis tested in these studies was not whether passenger carriage would increase the risk of car crash or crash-related injury in the population. Second, data utilising routinely collected information is averaged over individuals, thus, the degree of association between exposure and outcome might not represent individual-level associations. Moreover, as reflected in the above-mentioned studies, adjustment for potential confounding factors at the individual-level was inadequate. This would further distort the already-imprecise estimation of association. Studies of this design are appropriate for generating research hypothesis, but not for testing pre-defined hypothesis (Rothman and Greenland, 1998). Third, due to the limitation of being non-population-based studies, it rendered any calculation of population attributable risk (PAR) inappropriate. Other study designs, such as the case–control study, need to be employed in order to produce more reliable relative risk estimation. No case–control study has been conducted, so far, in this important area of young driver road safety. The aim of the present study is to present the results obtained from a case–control study conducted in the Australasian region on the effects of the numbers and ages of passengers carried, on the risk of car crash injury, with specific focus on the comparison between younger and older drivers.
2. Methods The method has been detailed previously by Connor et al. (2002). In brief, a population-based case–control study was conducted in the Auckland region of New Zealand, between April 1998 and July 1999. The region included urban, suburban and rural areas and a population of approximately 1 million, and the source population for study participants comprised drivers of light vehicles driving on public roads. Eligible vehicles included cars, vans and light utility vehicles, but excluded all vehicles licensed as heavy vehicles, taxis, and emergency vehicles. The definitions of geographical boundaries, time period, eligible vehicles, and eligible roads were applied in an identical manner to cases and controls. The study was approved by the Northern Regional Health Authority Ethics Committee. Informed consent was obtained from all participants for personal participation and use of records. 2.1. Selection of cases All drivers or passengers in eligible vehicles who were hospitalised or died as the result of a car crash in the study region were identified prospectively through daily surveil-
lance and case-finding in the region’s four trauma hospitals and single Coroner’s office. Hospital trauma teams, emergency department staff, and the New Zealand Police collaborated to ensure comprehensive case-finding. In each case, the driver of the vehicle was the key informant, whether or not the driver themselves was injured. 2.2. Selection of controls Controls were a sample of car drivers representative of all time spent driving on the region’s roads during the study period. They were identified by cluster sampling of drivers at 69 randomly selected sites on the road network. The day of week, time of day and direction of travel for each survey site were randomly assigned. Eligible vehicles passing each survey site during a 2-h period were randomly sampled in proportion to the traffic volume at the site, which had been determined by earlier measurements. Cars were selected one at a time, from traffic that had been slowed as it approached the site. On the busiest roads, we selected the first car identified after the previous one had been processed. At less busy sites, sampling was slowed by waiting for a predetermined period (2, 5, or 15 min depending on traffic volume) between finishing with one car and selecting the next. Residual differences in sampling fractions between survey sites were adjusted for in the analysis, by weighting the clusters of controls by the inverse of the sampling fraction from their site. Surveys were carried out at an average of one per week and recruitment approximately matched accrual of cases. When vehicles selected as controls did not stop, or when it was not possible to stop traffic safely (e.g. motorway sites), vehicle owners were identified through registration plates and telephone directories, so that drivers could be invited to participate in the study. 2.3. Data collection Interviews with case drivers were conducted face-to-face in hospital, or by telephone at home. Proxy interviews were sought for drivers who sustained fatal injuries, or were too ill to participate. For control drivers, contact details, suitable interview times and a breath test for alcohol were obtained at the roadside recruitment sites, and interviews conducted by telephone. The majority of case and control interviews were carried out in the first 48 h after the crash or survey, although follow-up calls were sometimes necessary. The highly structured interview was based on a questionnaire covering the circumstances of the current trip and many usual behaviours and background characteristics of the drivers. In particular, respondents were asked to record the number of passengers carried in the car, and their ages at the time of crash or survey. Information on a range of other variables including demographics, time of crash/interview, self-reported alcohol consumption 6 h within crash/interview, the average kilometres driven per week, and drivers’ sleepiness as measured by Stanford Sleepiness Scale was also collected. These
L.T. Lam et al. / Accident Analysis and Prevention 35 (2003) 861–867
variables were considered as potential confounders in the relationship between passenger carriage and the risk of car crash injury. They were selected as confounders based on the information suggested from the literature.
863
were conducted by subgroups of younger (<25 years) and older (≥25 years) drivers. Logistic regression analyses were conducted to calculate the unadjusted and adjusted odds ratios (OR) and their corresponding 95% confidence intervals (95% CI) for the associations between passenger carriage and the case–control status. All analyses were conducted using the SUDAAN statistical analysis software (Shah et al., 1997) to handle the cluster sampling effect of data collected on the controls. The population attributable risks (PAR) were also calculated using the method detailed by Bruzzi et al. (1985). This method provided a means to calculate the PAR after adjusting for other potential risk factors.
2.4. Data analyses For the purpose of comparison, the age groups of drivers were classified into two, the younger driver (<25 years) and the older driver (≥25 years). In the analyses, the number of passengers was classified as none, one and two or more. The passenger ages were defined by the ages of passengers carried in relation to the ages of drivers. A “same age” passenger was defined as any passenger whose age was in the same age group as the driver, otherwise the passenger was classified as “other age” passenger. Thus, the “other age” passenger category included passengers whose ages were older than the drivers or a mixture of older and same age passengers. As it was anticipated that the number of passengers carried and passenger ages are highly correlated, a composite variable, namely passenger carriage, was created for both cases and controls, to include both the numbers of passengers and passenger ages. This included single drivers, one same age passenger only, two or more same age passengers, one other passenger only, two or more other passengers. For the comparison between younger and older drivers, data analyses
3. Results A total of 571 cases including 195 (34.2%) younger drivers and 588 controls including 94 (14.1%) younger drivers participated in the study. The response rate for the cases was 93% and that for the controls was 79%. For both younger and older control drivers, nearly 63% were driving alone at the time of survey (Table 1). In comparison, 59% of older drivers involved in crashes were driving alone whereas only 34% of younger drivers who involved in crashes were single drivers (Table 1). The frequency
Table 1 Frequency distributions of passenger carriage, demographic and potential confounding variables by age, and case and control status <25 years Cases (n = 195) Passenger carriage Single driver One same age passenger only ≥2 same age passengers One other passenger only ≥2 other passengers Potential confounding variables Sex Female Male
66 43 40 13 33
(33.9) (22.1) (20.4) (6.7) (16.9)
≥25 years Controls (n = 94)
Cases (n = 376)
Controls (n = 494)
59 15 6 6 7
220 53 6 32 65
297 84 9 36 66
(64.3) (16.2) (6.8) (6.6) (6.1)
(58.5) (14.1) (1.6) (8.5) (17.3)
(62.0) (15.9) (1.6) (7.4) (12.2)
53 (27.2) 142 (72.8)
38 (44.0) 56 (56.0)
145 (38.6) 231 (61.4)
188 (40.9) 306 (59.1)
81 (41.5) 114 (58.5)
17 (10.0) 77 (90.0)
95 (25.3) 281 (74.7)
50 (7.0) 444 (93.0)
Self-reported alcohol consumption (6 h prior) No/1 drink 148 (75.9) ≥2 47 (24.1)
88 (95.9) 6 (4.1)
309 (82.2) 67 (17.8)
474 (97.4) 20 (2.6)
Average driving per week <100 km 100–300 km 300–500 km >500 km
20 33 21 10
51 190 55 40
75 204 120 81
Night-time driving 9:00 p.m.–5:59 a.m. Other
Driver sleepiness No Yes
37 94 24 8
(22.7) (57.7) (14.7) (4.9)
141 (72.7) 53 (27.3)
Column percentage of controls was adjusted for sampling design.
(18.1) (44.0) (26.4) (11.5)
90 (97.2) 4 (2.8)
(15.2) (56.6) (16.3) (11.9)
306 (81.6) 69 (18.4)
(15.7) (46.7) (22.5) (15.1)
488 (98.5) 6 (1.5)
864
L.T. Lam et al. / Accident Analysis and Prevention 35 (2003) 861–867
distributions of the potential confounders were also presented in Table 1. Table 2 summarises the relationships between passenger carriage, potential confounders, and the risk of car crash injury by age groups. There was a significant unadjusted association between passenger carriage and the risk of car crash injury for younger drivers. The odds of car crash injury increased by 2.5 times (crude OR = 2.51 (1.22–5.15)) for younger drivers carrying one same age passenger only, 5.5 times (crude OR = 5.57 (2.25–13.78)) carrying two or more same age passengers, and more than five times (crude OR = 5.18 (1.61–16.61)) carrying two or more other passengers, each compared to unaccompanied driving. On the other hand, the unadjusted association between passenger carriage and the risk of crash injury for older driver was insignificant. Results obtained from the univariate analyses also indicted that sex of driver, night-time driving, self-reported alcohol consumption, and driver sleepiness are significantly related to car crash injury among younger drivers (Table 2). In comparison, alcohol consumption and driver sleepiness were associated significantly with crash injury among older drivers. To elucidate the confounding status of those potential confounding variables, further analyses were conducted on the relationship between these variables and passenger carriage (Table 3). Results indicated that most of these variables were associated with passenger carriage except gender and self-reported alcohol consumption for younger drivers and night-time driving for older drivers. All these variables were considered as potential confounding factors to be adjusted for in following analyses. The results obtained from the multiple logistic regression were in agreement with those obtained from the univariate
Table 2 Associations between passenger carriage, potential confounder, and car crash injury by ages of drivers Crude OR (95% CI)
Passenger carriage Single driver One same age passenger only ≥2 same age passengers One other passenger only ≥2 other passengers
<25 yearsa
≥25 years
1.00 2.59 5.77 1.91 5.19
1.00 0.96 1.06 1.24 1.53
(1.26–5.33) (2.32–14.33) (0.64–5.68) (1.66–16.18)
Potential confounding variables Sex Female 1.00 Male 2.18 (1.17–4.05) Night-time driving Other 9:00 p.m.–5:59 a.m.
(0.58–1.59) (0.34–3.30) (0.68–2.26) (0.93–2.50)
1.00 1.10 (0.81–1.48)
1.00 1.00 6.21 (1.31–29.56) 4.53 (0.90–22.79)
Self-reported alcohol consumption previous 6 h No/1 drink 1.00 1.00 ≥2 7.28 (2.06–25.71) 8.14 (3.59–18.47) Average driving per week <100 km 100–300 km 300–500 km >500 km
1.00 1.07 (0.52–2.23) 0.44 (0.21–0.90) 0.34 (0.11–1.06)
Driver sleepiness No Yes
1.00 1.00 13.22 (3.32–52.64) 15.15 (5.78–39.72)
a
1.00 1.25 (0.81–1.93) 0.75 (0.44–1.31) 0.82 (0.45–1.52)
Crude ORs were adjusted for sampling design.
Table 3 Associations between potential confounding variables and passenger carriage by ages of drivers Passenger carriage <25 years
≥25 years
χ12 = 2.98, P > 0.05
χ12 = 8.36, P < 0.05
Night-time driving Other 9:00 p.m.–5:59 a.m.
χ12 = 4.88, P < 0.05
χ12 = 2.49, P > 0.05
Self-reported alcohol consumption previous 6 h No/1 drink ≥2
χ12 = 3.26, P > 0.05
χ12 = 4.40, P < 0.05
χ32 = 12.53, P < 0.05
χ32 = 23.69, P < 0.05
χ12 = 5.29, P < 0.05
χ12 = 5.21, P < 0.05
Potential confounding variables Sex Female Male
Average driving per week <100 km 100–300 km 300–500 km >500 km Driver sleepiness No Yes
L.T. Lam et al. / Accident Analysis and Prevention 35 (2003) 861–867 Table 4 Adjusteda odds ratios (95% CI) for car crash injury by passenger carriage for younger (<25 years) and older (≥25 years) drivers Passenger carriage
Single driver One same age passenger only ≥2 same age passengers One other passenger only >2 other passengers
Adjusted OR <25 years
≥25 years
1.00 2.39 (0.91–6.29) 15.55 (5.76–42.02) 3.490 (0.69–17.79) 10.19 (2.84–36.65)
1.00 0.78 0.30 1.19 1.33
(0.42–1.46) (0.06–1.44) (0.63–2.22) (0.78–2.27)
a ORs were adjusted for sex, night-time driving, self-reported alcohol consumption, average driving per week and driver sleepiness.
analyses. As shown in Table 4, the model was significant (P < 0.001) with passenger carriage and five potential confounding variables. Similar to the unadjusted associations, younger drivers who carried more than one passenger tended to have an increased risk of car crash injury for both same age and other passengers when compared with single drivers (Table 4). However, no equivalent increase of risk was observed among older drivers. The odds of car crash injury increased by about 16 times for younger drivers who carried two or more young passengers (OR = 15.55 (5.76–42.02)). The odds was also increased by 10 times for those who carried two or more other type of passengers (OR = 10.19 (2.84–36.65)) as compared with single drivers of the same age group. The population attributable risk associated with passenger carriage with two or more passengers, disregarding their ages, was 33.3% (95% CI 27.8–40.0) as compared to those driving alone or carrying one passenger for younger drivers. Calculation of the PAR for older drivers was not attempted since it had been shown that passenger carriage is not associated with crash injury.
4. Discussion The results obtained in this study have demonstrated that the number of passengers carried is an important risk factor for car crash injury among younger drivers. Carrying more than one passenger has significantly increased the risk of car crash injury for younger drivers, but not for older drivers. Furthermore, such an increase in the risk occurred not just for carrying same age passengers, but also passengers of other ages. These results, in general, are consistent with the finding in other studies (Doherty et al., 1998; Chen et al., 2000). Moreover, they suggest that, after adjusting for the risk factors identified in the literature, the effect of the number of passengers carried, rather than the ages of passengers, is more significant on the risk of car crash injury among younger drivers. This is the first study utilising a case–control study design in investigating the relationship between passenger carriage and the risk of car crash injury. All other studies are
865
ecological in nature. As mentioned, the study design has posed a restriction on the accuracy of risk estimations on the association between passenger carriage and car crash injury. The results obtained in this study provided odds ratios calculated after adjusting for important potential confounding risk factors suggested in the literature (Rothman and Greenland, 1998). Another strength of the study is that selection bias has been minimised. All cases occurring within the study region and study period have been identified. The cluster sampling methodology employed to recruit controls has provided a random sample of the driving population at risk in the same geographic location of the cases. The controls are comparable to the cases in terms of the distributions in the demographic and other variables of interest, except that the proportion of controls is lower in the younger group than that in the older. Reasonably, high response rates were achieved for both cases and controls. Information bias has also been minimised by using a standardised questionnaire, interview protocol and trained interviewers. The outcome of each case has been accurately reported and documented with detailed information gathered. Potential bias, particularly recall bias on the exposure measure, may exist for both cases and controls. The recall on the number of passengers carried in the vehicle at the time of crash or interview is likely to be accurate, except for those severely injured cases that required proxy respondents. However, the recall of the passengers’ ages is subjected to bias due to unreliable memory or inaccurate estimation. Such a bias may not be series for those passengers whose ages are obviously older or younger. The ages of passengers between 20- and 30-year-old are more likely to be either under or over estimated. Other recall biases may also occur in the assessment of other variables, such as self-reported alcohol consumption. This is a common source of bias in most epidemiological studies that rely upon self-reported information. The importance of the results obtained in this study is in its implication in the theoretical conceptualisation of the relationship between passenger carriage and the risk of car crash injury among young drivers. It has been suggested that the key to understanding such a relationship lies in a better comprehension of the social interaction or some other dynamics between the passenger and the driver (Regan and Mitsopoulos, 2001). It has been noted that the drivers carrying passengers will adjust their driving behaviour according to the social expectations of the passengers perceived by the drivers (Baxter et al., 1990). Baxter et al. (1990) further observed that young drivers, particularly males, tended to drive at excessive speeds in the presence of male passengers. This observation has been strengthened by the results obtained from the study by Mckenna et al. (1998) that young drivers, both male and female, drove faster and accept shorter gaps at junctions when accompanied by young male passengers. These results are consistent with the findings of the Chen’s study mentioned above (Chen et al., 2000). Thus, it seems that carrying passengers, possibly
866
L.T. Lam et al. / Accident Analysis and Prevention 35 (2003) 861–867
young passengers, particularly young men has an detrimental effect on the road safety of younger drivers such that the presence of these passengers induces greater risky driving behaviour. Another school of thought proposes that the link between passenger carriage and car crash is that the presence of passengers may be a distraction to the driver. These distractions, as suggested by Williams (2000), might result from the greater verbal interaction, music playing or even physical interactions between younger drivers and their passengers, especially same age passengers. This seems to be supported by the finding of the recent study that passengers of all age groups perceived themselves to be more likely to talk to younger than older drivers (Regan and Mitsopoulos, 2001). This results in a reduction of attention on driving and, thus, compromises safety. This is probably due to the fact that as drivers are distracted there is an increase in the load of their cognitive functioning and processing speed. As reported by Lamble et al. (1999) an increase in cognitive load impairs the ability of drivers to detect changes in the environment, thus, increasing the chance of a collision. This may further interact with inexperience in employing compensatory speed-reducing action for the loss of attention among younger drivers. This results in a greater increase in the risk of crash among younger drivers. The results obtained in this study that the number of passengers carried, rather than passenger types, tended to be a more significant factor in influencing the risk of car crash injury among young drivers has lent further support to this view. Increased number of passengers, despite their ages, may have increased in-vehicle distractions and, thus, increased the chance of collision and injury. The findings of this study also have practical implication in terms of road safety policy making. It has been demonstrated that graduated licensing systems, in various forms, are effective measures in reducing car crash among novice young drivers (Langley et al., 1996; Ulmer et al., 2000; Foss et al., 2001; Shope et al., 2001). However, it remains uncertain as whether several components, such as night-time curfew and passenger restrictions, should be included in such licensing systems. At present, the graduated licensing systems in New Zealand and Australia pose no restrictions on passenger carriage. The results obtained in this study have indicated carrying more than one passenger is detrimental to younger drivers’ safety. These results can most probably be applied to those younger novice drivers who are still going through the GLS even though the driving experience has not been included in the analyses. Thus, passenger restrictions could be considered as part of the graduated licensing system. Young drivers who have just obtained their provisional licences might only be allowed to carry only one passenger, and preferably an older passenger. This view has been advocated by some researchers (Chen et al., 2001), and has already been implemented in some jurisdictions (Frith and Perkins, 1992; Williams, 2000). To ascertain confirmative supportive evidence, further study should be
conduct to investigate the differences of risk estimation due to passenger carriage for younger novice and experienced drivers.
References Baxter, J.S., Manstead, A.S.R., Stradling, S.G., Campbell, K.A., Reason, J.T., Parker, D., 1990. Social facilitation and driver behaviour. Brit. J. Psychol. 81, 351–360. Bruzzi, P., Green, S.B., Byar, D.P., Brinton, L.A., Schairer, C., 1985. Estimating the population attributable risk for multiple risk factors using case–control data. Am. J. Epidemiol. 122, 904–914. Chen, L., Baker, S.P., Braver, E.R., Li, G., 2000. Carrying passengers as a risk factor for crashes fatal to 16- and 17-years old drivers. JAMA 283, 1578–1582. Chen, L., Braver, E.R., Baker, S.P., Li, G., 2001. Potential benefits of restrictions on the transport of teenage passengers by 16- and 17-year-old drivers. Injury Prev. 7, 129–134. Connor, J., Norton, R., Ameratunga, S., Robinson, E., Civil, I., Dunn, R., Bailey, J., Jackson, R., 2002. Driver sleepiness and the risk of serious car occupant injury: a population-based case–control study. Brit. Med. J. 324, 1125–1128. Drummond, A.E., Triggs, T.J., 1991. Driving as Skilled Performance: A Perspective for Improving Young Driver Safety. Accident Research Centre, Monash University, Melbourne, Australia. Doherty, S.T., Andrey, J.C., MacGregor, C., 1998. The situational risks of young drivers: the influence of passengers, time of day and day of week on accident rates. Accid. Anal. Prev. 30, 45–52. Foldvary, L.A., Lane, J.C. 1969. Car crash injuries by seating position and miles travelled. In: Proceedings of the 13th Annual Conference of the American Association for Automotive Medicine. American Association for Automotive Medicine, Minneapolis, MN, America. Foss, R.D., Feaganes, J.R., Rodgman, E.A., 2001. Initial effects of graduated driver licensing on 16-year-old driver crashes in North Carolina. J. Am. Med. Assoc. 286, 1588–1592. Frith, W.J., Perkins, W.A., 1992. The New Zealand Graduated Licensing System. Land Transport, Wellington, New Zealand. Lamble, D., Kauranen, T., Laakso, M., Summala, H., 1999. Cognitive load and detection thresholds in car following situations: safety implication for using mobile (cellular) telephones whole driving. Accid. Anal. Prev. 31, 617–623. Langley, J.D., Wagenaar, A.C., Begg, D.J., 1996. An evaluation of the New Zealand graduated driver licensing system. Accid. Anal. Prev. 28, 139–146. Mckenna, F.P., Waylen, A.E., Burkes, M.E., 1998. Male and female driver: how different are they? AA Foundation for Road Safety Research, Baisingstoke, UK. Preusser, D.F., Ferguson, S.A., Williams, A.F., 1998. The effect of teenage passengers on the fatal crash risk of teenage drivers. Accid. Anal. Prev. 30, 217–222. Regan, M.A., Mitsopoulos, E., 2001. Understanding passenger influences on driver behaviour: implications for road safety and recommendations for countermeasure development. Report no. 180. Accident Research Centre, Monash University, Victoria, Australia. Rothman, K.J., Greenland, S., 1998. Modern Epidemiology. Lippincott Williams & Wilkins, Philadelphia, USA. Shah, B.V., Barnwell, B.G., Bieler, G.S., 1997. SUDAAN User’s Manual, Release 7.5. Research Triangle Institute, Research Triangle Park, NC. Shope, J.T., Molnar, L.J., Elliott, M.R., Waller, P.F., 2001. Graduate driver licensing in Michigan: early impact on motor vehicle crashes among 16-year-old drivers. J. Am. Med. Assoc. 286, 1593–1598. Ulmer, R.G., Williams, A.F., Preusser, D.F., 1997. Crash involvement of 16-year-old drivers. Accid. Anal. Prev. 28, 97–103.
L.T. Lam et al. / Accident Analysis and Prevention 35 (2003) 861–867 Ulmer, R.G., Preusser, D.F., Williams, A.F., Ferguson, S.A., Farmer, C.M., 2000. Effect of Florida’s graduated licensing program on the crash rate of teenage driver. Accid. Anal. Prev. 32, 527–532. Williams, A.F., 2000. Teenage passengers in motor vehicle crashes: a summary of current research. Available at: http://www.highwaysafety. org/safety facts/teens/teen passenger.pdf. Accessed October 2001.
867
Williams, A.F., Wells, J.K., 1995. Deaths of teenagers as motor vehicle passengers. J. Safety Res. 26, 161–167. Williams, A.F., Preusser, D.F., Ulmer, R.G., Weinstein, H.B., 1995. Characteristics of fatal crashes of 16-year-old drivers: implications for licensure policies. J. Public Health Policy 16, 347–360.