Intentions and willingness to engage in risky driving behaviour among high school adolescents: evaluating the bstreetsmart road safety program

Introduction

Serious injuries and road deaths among young novice drivers remain unacceptably high and are a serious public health issue. These road users have some of the highest crash rates in Australia; drivers aged 16-19 years are 6-8 times more likely to crash than those aged 55-59 years¹ and a young novice driver on probationary first year license in Australia is four times more likely to be involved in a fatal crash than a 26 year-old driver². Young novice driver crash and fatality rates are highest in the first year, peaking immediately post-licensure³. This pattern is similar worldwide⁴, contributing to the fact that road traffic injuries are the leading cause of death for adolescents in most developed countries globally⁵.

Young drivers either underestimate the complexity of the driving task, overestimate their skill, or both; the consequence imposing a smaller safety margin than they perceive exists^6,7. Factors influencing novice driver risk include age, gender, maturity⁸, knowledge and attitudes, motivation, and sensation-seeking⁴. These factors impact involuntary and voluntary risky driver behaviour and are highly challenging when developing effective countermeasures. Some research has identified a relationship between the perceived risky driving behaviour of parents and friends and the probationary license holder’s willingness to engage in risky behaviour themselves^6,9. This highlights a need to broaden education beyond the pre-licensed driver, aiming to influence driving behaviour related influences they may have at home or among peers. However, further information regarding these relationships is needed.

Gender differences among young drivers have been predominantly investigated using binary (i.e., male/female) attributed populations, showing males to be more willing to take driving risks⁷, have more traffic violations^7,10 and higher fatality rates than females¹. Sexual minority adolescents (e.g., identifying as non-binary gender), have self-identified as engaging in higher rates of risky health behaviours such as drug/alcohol and intentional self-harm compared with heterosexual adolescents^11,12, yet no peer-reviewed evidence has been found on the relationship between sexual minority adolescents and risky driving behaviours.

Behaviour change techniques (BCT) have been successfully employed to change a variety of health behaviours, such as weight loss. BCTs are defined as observable, replicable component of an intervention that is designed to change behaviour¹³. There have been many initiatives in recent years addressing novice driver behaviour such as graduated licensing schemes, education and training programs for adolescents and their parents, legal enforcements¹⁴, and psychosocial education interventions aiming to impact upon sensation seeking and risk-taking intentions. However, there is little consensus on which approaches, if any, are effective in achieving lasting change in road-user behaviour and relatively few that are theory-led or evidence-based¹⁵. Michie et al¹³ described the concept of willingness to perform a target behaviour which is impacted upon by proposed key concepts for successful BCT such as ‘shaping knowledge’ (assisting BCT recipients to better understand their behaviour), ‘social comparison’ (generally comparing to positive examples of a desired behaviour) and ‘comparison of outcomes’ (allowing exploration of the outcomes of exhibiting or not exhibiting the behaviour). Some of these BCT concepts are incorporated into the bstreetsmart initiative.

‘bstreetsmart’ is an injury prevention program encouraging safer road-user behaviours among New South Wales (NSW) adolescents, established in 2005 by Westmead Hospital Trauma Department, NSW, Australia (https://bstreetsmart.org/). From the ∼450 students witnessing the inaugural event, bstreetsmart has grown to around 25,000 pre-driving adolescents from NSW schools attending bstreetsmart annually, over a 3-day period. Students witness crash re-enactments, potential risk-taking consequences, engage in interactive displays, hear from young drivers permanently injured in road crashes, and learn injury prevention strategies. bstreetsmart’s primary objective is to reduce the fatality and injury rates of young people by promoting safe behaviour; offering students and teachers access to current information injury prevention strategies. Its development has involved significant consultation with NSW Ambulance Service, Fire and Rescue NSW, NSW Police Force and NSW Police Crash Investigation Unit, Transport for NSW and NSW High School Year Advisors. bstreetsmart has undergone several process evaluations (none published), all suffering from small response fractions (< 5%) from students and teachers. However, its capacity to influence the future behaviour of the adolescents transitioning into driving remains unknown.

The Behaviour of Young Novice Drivers Scale (BYNDS) was developed in 2010¹⁶ and has been validated since across various populations^17-19. The BYNDS intended investigation of the underlying dimensions to young novice driver risky behaviour (i.e., probationary drivers), however the authors deemed it equally appropriate to measure stated intentions regarding risk among pre-independent driving youth. Factors measured within the full BYNDS scale comprised those deemed transient that would occur during a driving episode (e.g. speeding), those taken prior to a driving episode classified as fixed for within journey risk (e.g. drinking alcohol before driving), risky driving exposure (e.g. driving drunk friends) and those measuring driver emotions or mood prior to or during a journey. Substantial research describes adolescent development and the impulsivity common among this age group, typical of a young person in the process of creating their identity, values, and opinions²⁰. During this phase, adolescents are deemed more sensitive to social influences and role modelling, and normative aspects of their environment have been shown to assert strong influence over a novice driver’s intentions^4,20. This is described as the young person’s ‘social comparison’, where they may be more likely to do what they think others do. A version of the BYNDS for passengers (BYNDS_P) was developed by Dr Bridie Scott-Parker while conducting this study, to assess the risky driving behaviour of parents and friends with whom a pre-driving adolescent was a passenger (unpublished). The BYNDS_P was included in the pre-event survey to additionally understand the contribution of normative aspects of an adolescent’s road experience and risky driving exposure, to their own stated intentions of willingness to engage in risky driving behaviour.

The aim of this study was to explore self-reported willingness to engage in risky driving behaviour using a validated measure, and the influence of a range of covariates including gender, socio-economic status, and exposure to risky driving behaviour (BYNDS_P). We aimed to explore whether the bstreetsmart event could influence adolescents’ stated intentions to engage in risky driving.

Method

Results

A total of 2360 secondary school students consented to participate in the pre-event survey, 61% were female and 4.0% Indigenous. A total of 25,047 students from 208 secondary schools had registered to attend the event. Table 1 describes characteristics of the study population.

Of the original study cohort of 2360 students, the outcome variable (the modified BYNDS score) was completed by 2342 students (99.2%). Median imputation completed the missing scores. The 3-6 months post survey was completed by 204 students (8.6%) and was considered too small a sample to be used. The post-event survey was completed by 1260 (53.4%) of the 2360 cohort, for whom survey data were linked (Table 2). Comparisons of summary scores of the outcome variable in the pre- and post-event surveys in Table 2 were achieved using t-tests to compare raw scores in gender sub-groups. In the post-event survey (around one week after the bstreetsmart event), females, males and non-binary students showed lower overall scores than prior to the program, indicating lower stated willingness to engage in risky behaviour ‘when I get my P’s’ (probationary license in Australia).

The linear regression using the pre-event data and predictor removal using the AIC, removed the cultural background, SEIFA score and learner permit variables. The linear mixed model included: the ‘risk exposed’ variable, age, sex, and interactions between age and sex. Results are shown in Table 3, identifying that taking into account pre-event risk exposure in parental driving, age and sex, the post-event scores were on average three units lower than the pre-event scores (−2.99, 95% CI -3.418 to -2.466). Exposure to risky driving as a pre-driving teen increased stated intentions to engage in risky driving, with higher modified BYNDS score (1.21, 95% CI 0.622 to 2.011), and compared to females (reference), students identifying as non-binary reported significantly higher BYNDS_M scores (20.8, 95% CI 8.795 to 40.852), after adjusting for other predictors.

Table 4 compares student responses regarding various risky driving behaviours that they quantified in the adult who ‘most commonly drives them around’ (BYNDS_P) with their own stated intentions ‘when I have my probationary license’ (BYNDS_M). Notable differences included intending to “Sometimes drive without a valid licence” (9% versus∼ 2%), “Sometimes drive without a seatbelt” (5.8% versus 2.6%), “Sometimes drive drugged” (2.9% versus 1.1%) and “Sometimes carry too many passengers” (17.7% versus 5.6%) as probationary license holders.

Discussion

This study has demonstrated a significant reduction in the stated intentions of young pre-driving drivers to engage in risky driving behaviour, within the week following attendance at the bstreetsmart event demonstrating risks and consequences of unsafe driving practices. Willingness was measured by the BYND_M, a modification of a validated tool among novice drivers¹⁶, an important aspect when implementing BCT¹³. The impact appeared most effective among females with younger age, which supports existing literature of young males having higher rates of risky or delinquent behaviour compared with females²⁷. The impact appeared less effective among adolescents identifying as non-binary, and those exposed to risky driving behaviours while a passenger.

Behavioural willingness to engage in risk has been previously explored in an Australian driving population aged 17-25 years²⁸, willingness measured as a response to risk-related scenarios using the Prototype Willingness Model²⁹, and comparing probationary drivers (years 1 and 2) with open driver license types. The authors identified younger drivers with higher behavioural willingness was associated with lower perceived risk and higher reported engagement in risky driving behaviour²⁸. They did not, however explore genders other than male and female, nor the influence of other driver modelled behaviour. Gender identity minorities among adolescents are evidenced to have high risk-taking behaviours compared to cis-gender adolescents, for example having higher rates of intentional self-harm including suicide attempt³⁰. There are limited data regarding motor vehicle crashes by sexual orientation, and administrative data collections that document options other than male, female or undetermined sex are rare, and sexual identity does not form part of administrative data collections in general. Reisner et al ³¹ explored the impact of sexual minority status on self-reported seat belt use in American adolescents, finding adolescents of sexual minority orientation to have significantly higher rates of safety belt non-use. Their suggestions as to why this might be focussed on the idea that non-conformity to gender norms as both a pathway to sexual minority orientation and personality characteristics. They proposed that this non-conformity could also include rejection of other ‘conformity’ related behaviours such as seatbelt wearing ³¹. Sexual minorities have similarly been shown to report more risky behaviours than their cisgender counterparts for unhealthy behaviours linked with cancers such as alcohol, tobacco and other drug use, sexual activity, and dietary habits. ²⁷

The American College of Surgeons Committee on Trauma updated published recommendations for the optimal care of the injured patient ³², recommending that “…Trauma centres must implement at least two programs that address one of the major causes of injury in the community”, encouraging trauma services to incorporate injury research and proactive preventive programs into their remit as part of a community. Many programs that have been implemented have focussed on fear appeal approaches, however despite the modest effects of attitudinal change seen in the short term, they have received substantial criticism of the risks of psychological harm with this approach^33,34. Rather than using ‘shock tactics’ or basing prevention strategies on fear appeals, Purtle et al ³⁵recommend trauma-informed programs which acknowledge previous exposures to trauma and offer services that recognise the emotional, social and cognitive impacts of these experiences.

Since its inception in 2005 and due to its popularity, bstreetsmart has seen 207,955 students attending to end 2019. The program has more recently been live-streamed, making exposure counts less certain. The trauma-informed approach includes integration of a trauma-informed approach with the crash re-enactment, speakers with brain and spinal injuries and a family member who has lost a teen driver in a road crash. The potential emotional impact on students and teachers is managed by extensive pre-event communications with schools and provision of preparatory material. During the event multiple trained counsellors are available for assistance, debriefing and support if needed.

This study had several strengths including its sample size and reasonable response fraction for this demographic (53.4%). The population was likely to be representative of the NSW student population, given over 200 schools from greater Metropolitan Sydney area participated. Although around 40% of the study population resided in the socio-economically advantaged area, the less advantaged populations were also evenly represented, with almost one fifth of students(18.1%) living in the least socio-economically advantaged areas, according to the SEIFA scores²⁵. A limitation of comparing the BYNDS_P with BYNDS_M (Table 4) responses concerning passenger carrying, is the potential influence of known passenger restrictions for probationary drivers only. The main study limitation was the restriction to the amount of data permitted to be collected by the second stage of ethical review, limiting the exploration of what may have been important covariates of the outcome of interest. SERAP’s concerns were that parental consent was not obtained. Recent systematic review exploring adolescent consent ³⁶ concluded that most adolescents were able to comprehend the nature and purpose of proposed research, correctly interpret benefits and risks, and thus provide their own consent. Comprehension increased with age and school year level; adolescents aged 15–16 years demonstrating a similar understanding of research to adults.

Conclusion

This study demonstrates a trauma-informed, reality-based injury prevention program as effective in changing short-term stated intentions to engage in risky driving, among pre-independent driving students. Non-binary identifying students showed higher intentions to engage in risky driving and prevention programs should be responsive to these tendencies among sexual minority adolescents. These findings should encourage ongoing research in this field and the continued efforts of prevention programs targeted toward pre-driving adolescents.