You are here

  • Home
  • Archive
  • Volume 64, Issue 11
  • Socioeconomic status and risk of car crash injury, independent of place of residence and driving exposure: results from the DRIVE Study

Article Text

Article menu

Download PDFPDF

Research report
Socioeconomic status and risk of car crash injury, independent of place of residence and driving exposure: results from the DRIVE Study
  1. H Y Chen1,
  2. R Q Ivers1,
  3. A L C Martiniuk1,
  4. S Boufous1,
  5. T Senserrick1,
  6. M Woodward1,2,
  7. M Stevenson1,
  8. R Norton1
  1. 1The George Institute for Global Health, The University of Sydney, Sydney, Australia
  2. 2Mount Sinai School of Medicine, New York, New York, USA
  1. Correspondence to Rebecca Ivers, The George Institute for Global Health, The University of Sydney, PO Box M201, Missenden Road, Sydney, NSW 2050, Australia; rivers{at}georgeinstitute.org.au

Abstract

Background Previous studies that found increased crash risks for young drivers of low socioeconomic status (SES) have failed to adjust for factors such as driving exposure and rural residence. This aim of this study is to examine the independent effect of SES on crash risk, adjusting for such factors, and to examine the relationship between injury severity following a crash and SES.

Methods Information on risk factors for crash collected from 20 822 newly licenced drivers aged 17–24 years in New South Wales, Australia, as part of the DRIVE Study was prospectively linked to hospitalisation data. SES was classified as high, moderate or low based on the Australia 2001 Socio-Economic Index for Areas. Poisson regression was used to model risk of crash-related hospitalisation by SES, adjusting for confounders. Two measures of injury severity—urgency of treatment and length of hospital stay—were examined by SES.

Results Results of multivariable analysis showed that drivers from low SES areas had increased relative risk (RR 1.8, 95% CI 1.1 to 3.1) of crash-related hospitalisation compared to drivers from high SES areas. This increased risk remained when adjusting for confounders including driving exposure and rurality (RR 1.9, 95% CI 1.1 to 3.2). No significant association was found between injury severity and SES.

Conclusion The higher risk of crash-related hospitalisation for young drivers from low SES areas is independent of driving exposure and rural–urban differences. This finding may help improve and better target interventions for youth of low SES.

  • Socioeconomic status
  • traffic injury
  • young driver
  • the DRIVE study
  • cohort ME
  • injury
  • road accidents
  • social inequalities

https://doi.org/10.1136/jech.2009.091496

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Young people from lower socioeconomic status (SES) backgrounds have significantly higher risks of traffic injury than those from higher SES backgrounds.1–9 A similar pattern has been found for children as pedestrians, motor vehicle passengers and bicyclists9 as well as for adults as drivers.10 However, factors underlying this increased risk remain unclear.

    Various hypotheses have been proposed, including that drivers from low SES backgrounds exhibit more risky behaviours11 including impaired driving,12 have longer driving exposure2 or have a poorer driving environment.13 As previous studies also show that rural young drivers are overinvolved in fatal or severe crashes,14 15 and SES and rural residence are strongly correlated,16 17 place of residence (eg, urban, inner regional or rural) is likely to be an important determinant of crash-related injury. However, previous studies focusing on the effect of SES among young drivers' crash risk have failed to either measure or adjust for place of residence or other important risk factors including driving exposure, which may overestimate the risk.

    Previous research has shown that socially disadvantaged groups use more medical services compared to less disadvantaged persons18; however, potential SES disparities in injury severity following a crash have not been examined. Examining proxy measures of injury severity, such as urgency of treatment triaged in the emergency room and length of stay,19 20 by SES may help to clarify whether any such disparity exists.

    The objective of this study is to differentiate the risk of crash-related hospitalisation among young drivers by SES, adjusting for multiple confounders that are well-known risk factors for crashes, including driving exposure and place of residence. This study also aims to examine variations in severity of crash-related injury by SES. The DRIVE Study,21 with comprehensive information on known risk factors for crash, a large sample size and a prospective cohort design offered the potential to elucidate these associations.

    Methods

    Young drivers aged 17–24 years old and holding a first-year independent (provisional) licence were invited to participate in the DRIVE Study between June 2003 and December 2004 in New South Wales, Australia. Information on known and hypothesised crash risk factors was collected via online questionnaire from 20 822 drivers, with consent given for subsequent linkage to routinely collected data. In addition to demographic information, details on driving experience (such as experience in driving car or other motor vehicle while holding Learner license or prelicenced) and training (hours of driving training under professional/non-professional instructor), risky driving behaviours (using a composite measure of 14 risky driving behaviours),22 sensation-seeking behaviour,23 and alcohol use and mental health were collected. The mean follow-up period of this cohort was 2 years (median 2.1 years). Probabilistic linkage was used to link 1996–2005 crash-related hospitalisations, from the New South Wales (NSW) Health, to survey responses. Detailed methods and characteristics of study participants of the DRIVE Study have been previously described.21 This study was approved by the University of Sydney Human Research Ethics Committee and the NSW Health Ethics Committee.

    The Education and Occupation Index, one of the 2001 Socio-Economic Indexes for Areas (2001 SEIFA) in Australia,24 was used as a proxy for SES. The index uses statistical techniques to summarise the information from a variety of SES variables, calculating weights that will give the best summary for the underlying variables.24 As most participants were still studying in school and did not have a formal occupation, parental postcodes were used for the index. The SES index was classified into tertiles to indicate drivers from high (mean 1117.9, SE 47.83), moderate (mean: 995.4, SE 23.95) or low SES (mean 928.9, SE 26.77) areas. The average mean and median score of the Education and Occupation Index in 2001 for the general population in NSW was 1009 and 996, respectively.24

    Residential postcodes of young drivers were also grouped into three levels of urbanisation—urban (metropolitan cities), regional (country towns and surrounds) and rural (including remote areas) based on guidelines developed by the Australian Bureau of Statistics, indicating distances to public services.25

    Crash-related hospitalisations occurring before participants joined the study were excluded. Injury severity was determined by clinical urgency of treatment in the emergency room and length of stay (LOS) for stays longer than 1 day, as well as day-only length of stay (DOLOS). Clinical urgency of treatment received in the emergency room was derived from the Australian Triage Scale,20 a valid and reliable indicator of injury severity that has been used throughout Australia and New Zealand, which comprises five categories representing “resuscitation” (immediate), “emergency” (<10 min), “urgent” (10–30 min), “semi-urgent” (31–60 min) and “non-urgent”(61–120 min). As only a small number of injuries were coded as “resuscitation” in the present study, they were combined with the “emergency” category. LOS and DOLOS information was provided by NSW Health. As several drivers were admitted to hospital more than once but with all admissions for the same crash (identified via the same ICD-10 external cause code), LOS and DOLOS were summed for patients with multiple admissions. None of the crash-related hospitalisations had a fatal outcome.

    Statistical analyses

    Crash-related hospitalisation rates by SES and by rurality were examined by a χ2 trend test. Continuous variables were categorised before testing in models. Poisson regression was selected to determine the relative risk (RR) and corresponding 95% CIs of crash-related hospitalisation (one or more hospitalisations due to crash vs none) by SES. Age, gender, driving exposure (average driving hours per week) as well as other potential risk factors (the p value for the association between the variable and the main study outcome was <0.2 in the univariate analysis) were adjusted in multivariable models, and place of residence was entered as a final adjustment. Length of time of participation (days) in the DRIVE Study was included as an offset to estimate person-days at risk. Final multivariable models were created using backward elimination to take out variables if variable removal did not change the direction, or the estimate, of the main study factor (SES) by greater than 10% or the precision of the estimated effect to a significant level. Behavioural factors significantly associated with an increased risk of crash-related hospitalisation in univariate models (p<0.05) included marijuana usage, illicit drugs usage, high risk taking, previous crash history, moderate and high sensation seeking, average sleep hours per night less than 7 h, history of traffic offences, previous driving experience before licensure, driving a motor vehicle before learner licence, raced go-karts or similar vehicles, repeated attempts at the driving test and amount of time spent driving per week after licensure.

    The associations between the presence of an emergency visit, urgency of treatment at the time of arrival by SES and rurality were assessed by Mantel–Haenszel χ2 test due to most cell counts being less than 5. The variations in total LOS and the total DOLOS by SES and rurality were examined by a non-parametric Kruskal–Wallis test. The statistical software package SAS was used for all analyses.26

    Results

    Demographics and risk factors of participants by SES are presented in table 1. The majority of participants were aged 17 years, and 54% were women. Of participants from high SES areas, 95% lived in urban areas, compared to 65% and 62% of people from moderate and low SES areas, respectively. There was little variation in driving-related factors across SES groups, except that drivers with low SES reported a higher number of weekly driving hours and less professional driving lessons as a learner compared to other young drivers.

    View this table:
    Table 1

    Sample demographics and risk factor percentages by socioeconomic status

    The distribution of crash-related hospitalisation rates by SES and rurality is presented in table 2. There were 127 (0.61%) young drivers admitted to hospital due to a traffic crash. Crash-related hospitalisation rate significantly increased with decreasing SES (χ2=10.01, p<0.01), as well as increasing rurality (χ2=6.42, p<0.01).

    View this table:
    Table 2

    Distribution of crash-related hospitalisation rate by socioeconomic status and rurality

    Table 3 shows the results of the regression models. The RR of having a crash-related hospitalisation for drivers of low SES was almost double that for drivers of high SES (RR 1.9, 95% CI 1.24 to 3.03). After adjusting for multiple confounders including driving exposure, the risk of having a crash-related hospitalisation for drivers of low SES was 80% higher than that of drivers of high SES (RR 1.8, 95% CI 1.05 to 3.06). This association held when adjusting for place of residence, albeit with a wider CI (RR 1.9, 95% CI 1.07 to 3.24). No single adjustment changed the RRs for drivers of low SES by more than 9%.

    View this table:
    Table 3

    Relative risk of crash-related hospitalisation by SES

    More than 50% of drivers of high SES were triaged as “emergency”, the most urgent treatment category, decreasing to 40% for drivers of moderate SES and to 30% for drivers of low SES; however, these differences were not statistically significant (table 2). There was a significant association between urban residence and treatment urgency (χ2=4.13, p<0.05), with more than 40% of urban drivers triaged as “emergency” on arrival, compared to 11% only in rural areas.

    Table 4 summarises the statistics of duration of stay in hospital by SES and rurality. Young drivers from low SES areas had longer LOS and fewer DOLOS compared to drivers from high SES areas, but neither difference was statistically significant. There was a non-significant trend for fewer DOLOS for rural drivers compared to that of urban and regional drivers, and for shorter LOS for rural drivers compared to urban drivers. A comparison of the median LOS and DOLOS showed similar trends to the analysis of the mean LOS and DOLOS, with no significant differences found by either SES or rurality of place of residence.

    View this table:
    Table 4

    Duration of stay incurred from crash-related hospitalisations by SES and remoteness

    Discussion

    This study examined the risk of crash-related hospitalisation by residential SES of young drivers. Hospitalisation rates increased significantly with decreasing area-level SES and also with increasing rurality of place of residence. The results of multivariable models showed that, compared to drivers from high SES areas, young drivers from low SES areas were twice as likely to be involved in a crash-related hospitalisation, which could not be explained by driving exposure or place of residence. The urgency of treatment in the emergency room and duration of hospital stay were not significantly different by SES in the present study, although the treatment urgency was significantly different by place of residence.

    While the observed direction and strength of association between SES and risk of crash-related hospitalisation among young drivers is similar to findings from another young driver cohort in Sweden,1–4 7 8 12 this is the first study that has been able to adjust for individual risk factors hypothesised to explain the social gradient in traffic injury, particularly driving exposure, and is thought to be the best explanation for the high traffic injury risks for children and youths of low SES.9

    Place of residence has been regarded as a key confounder for the relationship between young driver SES and crash risk for two reasons. First, as patterns of risky health behaviours have been found to be aggregated in both rural residents and low SES individuals,27 28 rural and low SES drivers may share the same risky behaviours, such as high alcohol consumption and failure to wear seatbelts while driving.29 30 Second, drivers of low SES may drive longer distances or use less safe cars, which is also likely to be true for rural residents.31 The present study did not observe marked differences in risk behaviours by SES, in line with previous studies,9 although a higher level of driving exposure was reported by drivers of low SES compared to other SES groups, lending some support to this second proposition.

    One previous study was able to take place of crash (rural/urban) into account while examining the effect of SES on young driver crash8; however, the influence of place of residence on health is not only contextual (poor rural roads) but also compositional (lower use of protective devices) and therefore is a stronger measure than place of crash.32 The failure to adjust for place of residence in previous studies therefore may have confounded the effect of SES of young driver on risk of crash-related injuries. In this study, the higher risk of crash-related hospitalisation for young drivers of low SES remained after further adjustment for residence in rural areas, implying that there are other important underlying risk factors for crash unique to young drivers of low SES areas, which needs to be elucidated.

    The differences in SES or rurality of place of residence with respect to crash-related injury severity have not been examined simultaneously in previous research. Although the difference in clinical urgency of treatment as assessed upon triage in the emergency department by SES was not significant, in this study treatment urgency was significantly correlated with place of residence, with a greater proportion of urban drivers triaged in the highest urgency category. Previous studies have found that occupants with severe injuries following a crash are more likely to be transferred to urban, tertiary care hospitals that are better prepared to address the most severe injuries,33 34 thus masking the known higher crash severity in rural areas.35 In contrast to these findings, we found that more drivers living in urban areas were triaged as having serious injuries at the time of arrival than drivers from rural areas, which may require further examination. Accounting for occupants who died at the crash scene or before arriving at hospital may reverse the trend we found; however, we were not able to assess this in the current study since only seven crash-related deaths occurred in our study during the follow-up period, which made comparison of deaths by SES impractical. Future studies would need to investigate death registry data or coroner records, or follow-up a larger cohort with longer period to obtain sufficient numbers of fatal crashes to examine this potential link.

    The comparison of LOS, another common indicator of injury severity,19 showed that differences by SES were not significant, although there was a trend indicating that the mean LOS increased with decreasing area-level SES. The role of SES on LOS caused by other diseases is ambiguous as one previous study found that patients of lower SES had longer hospital stays,36 whereas another study showed that the SES of patients had limited effects on discharge decisions once patients accessed the medical care system.37 The difference of the mean LOS between rural and urban patients was also not significant in this study. Such rural/urban differences in LOS from the previous literature are also inconsistent. Whereas one study showed that paediatric patients admitted to rural hospitals had significantly shorter LOS and lower odds for prolonged stay compared to those admitted to urban hospitals,38 another study showed that adolescents who suffered head injuries and lived in rural areas had significantly longer LOS in hospital.39 The small number of observed hospitalisations and the difficulties measuring injury severity in the present study may, however, mask the differentials in injury severity by SES or rurality. Examination of population-based hospitalisations and selection of well-verified indicators for injury severity, such as a standardised injury severity score, in future research may help clarify this issue.

    We used a measure of parental education and occupation as a proxy of SES, which has been deemed the best approximate measure for SES in young people.40 Although this proxy measure of SES is derived from area levels, it represents an average estimate of education and occupation levels for people living in the area. However, exactly how parental education and occupation might influence the crash risk of their children has been subject to ongoing debate. Young drivers may adopt parental driving behaviours,41 42 but young drivers from low SES households may only access older cars that offer lower occupant protection or others' cars that are unfamiliar.8 43 Young drivers of low SES may also crash more frequently because of living in a community with poor traffic environment.13 It was not possible to examine these issues in this study, but they are worthy of further investigation.

    The SES measure of the present study differs from that used in previous studies focusing on associations between SES of young drivers and road crashes.1–5 7 8 12 As the Australian SEIFA index is designed for measuring the SES of an area,24 the SES measures used in this study reflects the area SES rather than individual SES of the young driver.44 This measure of SES may therefore represent variables not otherwise measured in the DRIVE Study, or may act as a proxy for individual SES.45–47 However, we believe that many of the individual-level factors measured in the DRIVE Study could be considered surrogates for SES—for example, risky behaviour and driving exposure.2 11 This study therefore finds a significant effect of area SES on crash risk while also adjusting for SES factors measured at an individual level, which has not been done previously.9

    The strengths of this study include the prospective cohort design and collection of comprehensive information on known and hypothesised risk factors that enabled examination of relative risk for SES after adjusting for multiple risk factors. Further, this is the first study to look at the SES differentials in injury severity following a crash.

    Nonetheless, the limitations of this study need to be considered when interpreting the findings. First, as the DRIVE Study is not, and did not attempt to be, representative of any specific population, the results may not be generalisable. However, a representative cohort is not necessary to obtain an unbiased and generalisable estimate of association between an outcome and potential risk factor.48 Second, the composite nature of the SEIFA index may mask variation between areas (two areas with the same score may differ in the values that contributed to that score), and sometimes increase the difficulty of identifying the focus of intervention (areas with low education attainment or areas with low occupation class).49 This needs to be examined in future work. Further, as there was a very small number of people of high SES in rural regions, mainly because of the method used to measure the SES in this study, the multivariable model may have underadjusted for the effect of remoteness. Finally, average driving hours per week may not be the best measure of driving exposure.50 However, Chipman et al51 concluded that driving time is a better measure of exposure than distance travelled to explain crash risks among drivers and regions with very different driving patterns and environments, as is the case in the present study.

    Conclusions

    After adjusting for driving exposure, rurality of residence and other well-known risk factors of crash, young drivers from low SES areas were twice as likely to have a crash-related hospitalisation. As individual-level factors were well controlled for in the analysis, it is possible that this increased risk may be due to area level effects of SES; however, further work is needed to examine this hypothesis. Elucidating the unique aspects of low SES that independently contribute to increased crash injuries may lead to improved and better targeted intervention to reduce the over-representation of youth in road traffic injuries. Furthermore, an increased risk of crash-related hospitalisation in NSW for those of low SES irrespective of place of residence indicates the needs of drivers from low SES areas to receive priority or specific targeting in future intervention initiatives, regardless of where they live. Improving road safety in low SES areas, such as general road infrastructure, has the potential to benefit young drivers in these communities, as well as other drivers with whom they share the roads.

    What is already known on this subject

    Young drivers of low SES backgrounds have a significantly higher risk of crash-related injury; however, factors underlying this increased risk are unknown. Factors including driving exposure and rural residence are theorised to explain the increased risk but have not yet been examined. Associations between injury severity following a crash and SES have not been previously examined.

    What this study adds

    Adjusting for driving exposure and rural residence did not explain the increased crash risk for young drivers of low SES, indicating that there are some underlying risk factors for crash unique to young drivers of low SES areas. Injury severity of crash-related hospitalisation was not found to differ by SES.

    Acknowledgments

    We thank all those who initiated and contributed to the DRIVE study: RQ Ivers, SJ Blows, MR Stevenson, RN Norton, A Williamson, M Eisenbruch, L Lam, P Palamara and J Wang.

    References

    1. ↵
      1. Hasselberg M,
      2. Vaez M,
      3. Laflamme L
      . Socioeconomic aspects of the circumstances and consequences of car crashes among young adults. Soc Sci Med 2005;60:287–95.
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      1. Hasselberg M,
      2. Laflamme L
      . Socioeconomic background and road traffic injuries: a study of young car drivers in Sweden. Traffic Inj Prev 2003;4:249–54.
      OpenUrlCrossRefPubMed
    3. ↵
      1. Hasselberg M,
      2. Laflamme L
      . The social patterning of injury repetitions among young car drivers in Sweden. Accid Anal Prev 2005;37:163–8.
      OpenUrlCrossRefPubMedWeb of Science
    4. ↵
      1. Hasselberg M,
      2. Laflamme L
      . Road traffic injuries among young car drivers by country of origin and socioeconomic position. Int J Public Health 2008;53:40–5.
      OpenUrlCrossRefPubMedWeb of Science
    5. ↵
      1. Hasselberg M,
      2. Laflamme L,
      3. Weitoft GR
      . Socioeconomic differences in road traffic injuries during childhood and youth: a closer look at different kinds of road user. J Epidemiol Community Health 2001;55:858–62.
      OpenUrlAbstract/FREE Full Text
    6. ↵
      1. Laflamme L,
      2. Engstrom K
      . Socioeconomic differences in Swedish children and adolescents injured in road traffic incidents: cross sectional study. BMJ 2002;324:396–7.
      OpenUrlFREE Full Text
    7. ↵
      1. Laflamme L,
      2. Hasselberg M,
      3. Kullgren A,
      4. et al
      . First car-to-car crashes involving young adult drivers: main patterns and their relation to car and driver characteristics. Int J Inj Contr Saf Promot 2006;13:179–86.
      OpenUrlCrossRefPubMed
    8. ↵
      1. Laflamme L,
      2. Vaez M
      . Car crash and injury among young drivers: contribution of social, circumstantial and car attributes. Int J Inj Contr Saf Promot 2007;14:5–10.
      OpenUrlCrossRefPubMed
    9. ↵
      1. Laflamme L,
      2. Diderichsen F
      . Social differences in traffic injury risks in childhood and youth—a literature review and a research agenda. Inj Prev 2000;6:293–8.
      OpenUrlAbstract/FREE Full Text
    10. ↵
      1. Whitlock G,
      2. Norton R,
      3. Clark T,
      4. et al
      . Motor vehicle driver injury and socioeconomic status: a cohort study with prospective and retrospective driver injuries. J Epidemiol Community Health 2003;57:512–16.
      OpenUrlAbstract/FREE Full Text
    11. ↵
      1. Oksuz E,
      2. Malhan S
      . Socioeconomic factors and health risk behaviors among university students in Turkey: questionnaire study. Croat Med J 2005;46:66–73.
      OpenUrlPubMedWeb of Science
    12. ↵
      1. Vaez M,
      2. Laflamme L
      . Impaired driving and motor vehicle crashes among Swedish youth: an investigation into drivers' sociodemographic characteristics. Accid Anal Prev 2005;37:605–11.
      OpenUrlCrossRefPubMedWeb of Science
    13. ↵
      1. Rivara FP
      . Child pedestrian injuries in the United States. Current status of the problem, potential interventions, and future research needs. Am J Dis Child 1990;144:692–6.
      OpenUrlCrossRefPubMed
    14. ↵
      1. Moller J
      . The spatial distribution of injury deaths in Australia: urban, rural and remote areas? Aust Inj Prev Bull 1994;8:8.
      OpenUrl
    15. ↵
      1. Chen HY,
      2. Ivers RQ,
      3. Martiniuk ALC,
      4. et al
      . Risk and type of crash among young drivers by rurality of residence: findings from the DRIVE Study. Accid Anal Prev 2009;41:676–82.
      OpenUrlCrossRefPubMedWeb of Science
    16. ↵
      1. Vos T,
      2. Begg S,
      3. Chen Y,
      4. et al
      . Socioeconomic differentials in life expectancy and years of life lost in Victoria, 1992–1996. N S W Public Health Bull 2001;12:126–30.
      OpenUrlCrossRefPubMed
    17. ↵
      1. Rupnik C,
      2. Thompson-James M,
      3. Bollman RD
      . Measuring economic well-being of rural Canadians using income indicators. 2001, Ottawa, ON, Canada: Statistics Canada, Agricultural Division, 2001. Cat no 21-601-MIE01045.
    18. ↵
      1. Moorin RE,
      2. Holman CD
      . The effects of socioeconomic status, accessibility to services and patient type on hospital use in Western Australia: a retrospective cohort study of patients with homogenous health status. BMC Health Serv Res 2006;6:74.
      OpenUrlCrossRefPubMed
    19. ↵
      1. Rosman DL,
      2. Knuiman MW,
      3. Ryan GA
      . An evaluation of road crash injury severity measures. Accid Anal Prev 1996;28:163–70.
      OpenUrlCrossRefPubMedWeb of Science
    20. ↵
      Australian College for Emergency Medicine. Policy on The Australian Triage Scale. http://www.acem.org.au/media/policies_and_guidelines/P06_Aust_Triage_Scale_-_Nov_2000.pdf (accessed Jun 2010).
    21. ↵
      1. Ivers RQ,
      2. Blows SJ,
      3. Stevenson MR,
      4. et al
      . A cohort study of 20,822 young drivers: the DRIVE study methods and population. Inj Prev 2006;12:385–9.
      OpenUrlAbstract/FREE Full Text
    22. ↵
      1. Ivers R,
      2. Senserrick T,
      3. Boufous S,
      4. et al
      . Novice drivers' risky driving behaviour, risk perception and crash risk: findings from the DRIVE study. Am J Public Health 2009;99:1638–44.
      OpenUrlCrossRefPubMedWeb of Science
    23. ↵
      1. Zuckerman M,
      2. Kuhlman DM,
      3. Joireman J,
      4. et al
      . A comparison of the three structural models for personality: the big three, the big five, and the alternative five. J Pers Soc Psychol 1993;65:757–68.
      OpenUrlCrossRefWeb of Science
    24. ↵
      1. Trewin D
      . Socio-economic indexes for areas. Census of population and housing. Canberra, ACT: Australian Bureau of Statistics, 2001.
    25. ↵
      1. ABS
      . Australian Standard Geographical Classification (ASGC). Canberra, ACT: Australian Bureau of Statistics, 2001.
    26. ↵
      SAS Institute Inc. SAS: statistical software. Version 9.1. Cary, NC. 2006.
    27. ↵
      1. Hartley D
      . Rural health disparities, population health, and rural culture. Am J Public Health 2004;94:1675–8.
      OpenUrlCrossRefPubMedWeb of Science
    28. ↵
      1. Turrell G,
      2. Mathers CD
      . Socioeconomic status and health in Australia. Med J Aust 2000;172:434–8.
      OpenUrlPubMedWeb of Science
    29. ↵
      1. Steinhardt DA,
      2. Watson BC
      . Night-time seatbelt non-use in serious crashes: a comparison of contribution factors in rural and urban areas of the United States and Queensland. In: Road Safety: Research, Policing, Education Conference. Melbourne, 2007.
    30. ↵
      1. Lerner EB,
      2. Jehle DV,
      3. Billittier AJ 4th.,
      4. et al
      . The influence of demographic factors on seatbelt use by adults injured in motor vehicle crashes. Accid Anal Prev 2001;33:659–62.
      OpenUrlCrossRefPubMedWeb of Science
    31. ↵
      1. Keen J
      . High gas prices threaten to shut down rural towns. California: USA TODAY, 2008.
    32. ↵
      1. Phillips CD,
      2. McLeroy KR
      . Health in rural America: remembering the importance of place. Am J Public Health 2004;94:1661–3.
      OpenUrlPubMedWeb of Science
    33. ↵
      1. Grossman DC,
      2. Hart LG,
      3. Rivara FP,
      4. et al
      . From roadside to bedside: the regionalization of trauma care in a remote rural county. J Trauma 1995;38:14–21.
      OpenUrlPubMedWeb of Science
    34. ↵
      1. Stella J,
      2. Sprivulis P,
      3. Cooke C
      . Head injury-related road crash mortality in rural Western Australia. ANZ J Surg 2001;71:665–8.
      OpenUrlCrossRefPubMedWeb of Science
    35. ↵
      1. Muelleman RL,
      2. Wadman MC,
      3. Tran TP,
      4. et al
      . Rural motor vehicle crash risk of death is higher after controlling for injury severity. J Trauma 2007;62:221–5; discussion 5–6.
      OpenUrlCrossRefPubMedWeb of Science
    36. ↵
      1. Epstein AM,
      2. Stern RS,
      3. Tognetti J,
      4. et al
      . The association of patients' socioeconomic characteristics with the length of hospital stay and hospital charges within diagnosis-related groups. N Engl J Med 1988;318:1579–85.
      OpenUrlPubMedWeb of Science
    37. ↵
      1. Weingarten JP,
      2. Clay JC,
      3. Heckert DA
      . Impact of socioeconomic status on health care utilization: factors influencing length of stay. J Health Hum Serv Adm 1997;19:384–409.
      OpenUrlPubMed
    38. ↵
      1. Lorch SA,
      2. Zhang X,
      3. Rosenbaum PR,
      4. et al
      . Equivalent lengths of stay of pediatric patients hospitalized in rural and nonrural hospitals. Pediatrics 2004;114:e400–8.
      OpenUrlAbstract/FREE Full Text
    39. ↵
      1. Chiang MF,
      2. Chiu WT,
      3. Chao HJ,
      4. et al
      . Head injuries in adolescents in Taiwan: a comparison between urban and rural groups. Surg Neurol 2006;66:S14–19.
      OpenUrlCrossRefPubMedWeb of Science
    40. ↵
      1. Huurre T,
      2. Aro H,
      3. Rahkonen O
      . Well-being and health behaviour by parental socioeconomic status: a follow-up study of adolescents aged 16 until age 32 years. Soc Psychiatry Psychiatr Epidemiol 2003;38:249–55.
      OpenUrlCrossRefPubMedWeb of Science
    41. ↵
      1. McKay MP,
      2. Coben JH,
      3. Larkin GL
      . Driving beliefs and behaviors of novice teen drivers and their parents: implications for teen driver crash risk. Annu Proc Assoc Adv Automot Med 2003;47:197–214.
      OpenUrlPubMed
    42. ↵
      1. Bingham CR,
      2. Shope JT
      . Adolescent developmental antecedents of risky driving among young adults. J Stud Alcohol 2004;65:84–94.
      OpenUrlPubMedWeb of Science
    43. ↵
      1. Williams AF,
      2. Leaf WA,
      3. Simons-Morton BG,
      4. et al
      . Vehicles driven by teenagers in their first year of licensure. Traffic Inj Prev 2006;7:23–30.
      OpenUrlCrossRefPubMed
    44. ↵
      1. Walker AE,
      2. Becker NG
      . Health inequalities across socio-economic groups: comparing geographic-area-based and individual-based indicators. Public Health 2005;119:1097–104.
      OpenUrlCrossRefPubMedWeb of Science
    45. ↵
      1. Amir LH,
      2. Donath SM
      . Socioeconomic status and rates of breastfeeding in Australia: evidence from three recent national health surveys. Med J Aust 2008;189:254–6.
      OpenUrlPubMedWeb of Science
    46. ↵
      1. Pickett KE,
      2. Pearl M
      . Multilevel analyses of neighbourhood socioeconomic context and health outcomes: a critical review. J Epidemiol Community Health 2001;55:111–22.
      OpenUrlAbstract/FREE Full Text
    47. ↵
      1. Woodward M
      . Small area statistics as markers for personal social status in the Scottish heart health study. J Epidemiol Community Health 1996;50:570–6.
      OpenUrlAbstract/FREE Full Text
    48. ↵
      1. Rothman KJ,
      2. Greenland S
      . Modern epidemiology. Philadelphia, PA: Lippincott, 1998.
    49. ↵
      1. McCracken K
      . Into a SEIFA SES cul-de-sac? Aust N Z J Public Health 2001;25:305–6.
      OpenUrlPubMedWeb of Science
    50. ↵
      1. Chipman ML,
      2. MacGregor CG,
      3. Smiley AM,
      4. et al
      . Time vs. distance as measures of exposure in driving surveys. Accid Anal Prev 1992;24:679–84.
      OpenUrlCrossRefPubMedWeb of Science
    51. ↵
      1. Chipman ML,
      2. MacGregor CG,
      3. Smiley AM,
      4. et al
      . The role of exposure in comparisons of crash risk among different drivers and driving environments. Accid Anal Prev 1993;25:207–11.
      OpenUrlCrossRefPubMedWeb of Science

    Footnotes

    • Funding This work was supported by National Health and Medical Research Council of Australia, Roads and Traffic Authority of New South Wales (NSW), NRMA Motoring and Services, NRMA-ACT Road Safety Trust, NSW Health and the Motor Accidents Authority. RQI, TS, SB and MS received salary funding from the National Health and Medical Research Council of Australia. ALCM was supported by a fellowship from the Canadian Institutes of Health Research.

    • Competing interests None.

    • Ethics approval This study was conducted with the approval of The University of Sydney Human Research Ethics Committee and the NSW Health Ethics Committee.

    • Provenance and peer review Not commissioned; externally peer reviewed.