Journal of the New Zealand Medical Association, 29-June-2012, Vol 125 No 1357
Motor vehicle traffic crashes are the leading cause of mortality and morbidity among adolescents and young adults (15–25 years of age) in New Zealand 1,2 and many other developed countries.3 In response to the high motor vehicle crash rate among young drivers in New Zealand a Graduated Driver Licensing System (GDLS) was introduced in 1987.4
The key elements of the GDLS are: a 6 months learner licence stage of supervised driving; a restricted licence stage of 18 months that allows unsupervised driving except at night-time (10pm–5am) or with young passengers in the car; a full licence stage with no restrictions. Further details of the New Zealand GDLS have been described elsewhere.5
Despite a significant improvement since graduated licensing was introduced5, traffic related injury remains the leading cause of death and hospitalisation among young New Zealanders, especially young Māori.1,2 Most New Zealand research on young drivers to date has used routinely collected crash data, such as the police traffic crash reports and the national hospital inpatient records, but has been limited in that these databases do not, and realistically cannot, include the level of detail required to ensure that learner driver policy and programmes are based on sound scientific evidence applicable to young drivers in the current New Zealand context.6
There have been a small number of studies examining the role of parents/caregivers in the learning to drive process. These studies have shown that high parental monitoring of driving by teens can reduce risky driving and crashes, especially during the high risk first few months of driving, and interventions designed to encourage parental monitoring have shown promise.6 None of these studies were undertaken in New Zealand. The lack of relevant information relevant to young drivers in New Zealand lead to the establishment of the New Zealand Driver’s Study (NZDS).
The NZDS is a prospective cohort study designed to explore the relationship between a comprehensive range of driving and traffic safety related factors and subsequent traffic crashes and convictions among newly licensed drivers. The study is designed to capitalise on the unique characteristics of New Zealand’s multistage graduated driver licensing system with the interview stages for collecting exposure data coinciding with the three testing stages of graduated driver licensing (Figure 1)
The protocol for this study has been described in detail elsewhere.6 Briefly, study participants are recruited at the learner licence stage and complete a self-administered baseline questionnaire. They are then followed-up at the restricted and full licence stage, by telephone interview. Official traffic crash and offence outcome data are obtained from the authorities that maintain these databases, and self-reported crashes are obtained at the restricted and full licence interview.
The study was established following a comprehensive pilot study which included: consultation with Māori; consultation with agencies involved in driver licensing; trialling procedures for recruitment; and developing, pre-testing and piloting questionnaires.6,7
In this paper we describe:
Goals and objectives—The goal of the New Zealand Drivers Study is to provide an evidence base that would facilitate efforts aimed at reducing traffic-related injury in New Zealand, and especially among the high-risk adolescent/young adult population. The primary objective of the study is to explore the relationship between a comprehensive range of driving and traffic safety related factors (e.g. driving experience, motivation for driving/licensing, driver training, alcohol use, risk-taking) and subsequent traffic crashes and convictions among newly licensed drivers, and from this identify specific areas that can be targeted to reduce traffic-related injury among this high-risk group.
The secondary objectives are to:
An overview of the study design is provided in Figure 1.
Figure 1. Overview of study design
Study population size—We aimed to recruit 5,000 newly licensed drivers of whom 1500 (minimum 1000) would be Māori, over a 12-month period. In reality we recruited 4282 over a 2-year period but 290 were ineligible (e.g. overseas licence conversions, truck licences, unsigned consent form) leaving a final cohort of 3992, of whom 825 self-identified as Māori.
Although the total cohort, and the number of Māori, were less than our original targets, the power calculations as given in the study protocol6 show that these numbers provide adequate statistical power to show effect sizes of around 1.5 to 2.0 for policy issues (e.g. full cohort) and 2.0 to 2.5 for programme issues (e.g. Māori cohort), which we considered satisfactory.
Recruitment methods—Our aim was to recruit participants from a diverse range of locations throughout NZ. During the pilot study7 recruitment was undertaken by the University research team. Given the differing needs of the local populations, we anticipated that different methods would be required in different regions.
In the larger urban areas, (Auckland, Christchurch, Dunedin), research assistants (RAs) (mostly postgraduate students) were employed part-time to undertake face-to-face recruitment at the local driver licensing centres. For example, in areas with a relatively high proportion of Māori, such as South Auckland (Manukau), and on the East Coast of the North Island (Hastings/Napier, Wairoa, Gisborne and Ruatoria) it was preferred that the recruitment be done by local Māori community groups. Six such groups, with whom we had consulted and established a collaborative partnership during the pilot, were subcontracted to undertake the recruitment among their local people.
All of these community groups had many years of experience delivering learner licence courses, and the source of participants for the NZDS. Five of these community/iwi groups were also subcontracted to undertake recruitment at their local driver licensing centre.
To boost the number of rural participants, postcards were distributed to all NZ Automobile Association (NZAA) licensing centres, except those in the large urban centres. On this postcard was an invitation for all newly licensed learner drivers to take part in the study, a mention of the $20 petrol voucher all participants would receive for each completed interview, and our contact details (email address, a free telephone number, and cell phone number) which they needed to contact if they wished to take part. This recruitment method was reasonably successful, and in particular helped boost the number of rural participants.
The licensing centres chosen for the recruitment were those that had issued the highest number of learner licences in the previous year. However, as there was no booking system for learner licence testing, there was a degree of uncertainty as to how many potential study participants would turn up at any centre, at any given time. Most of the recruitment took place at NZAA licensing centres, and although the recruitment took two years instead of one, the NZAA staff were very accommodating and supportive of the research staff. This highlighted the importance of having established a good rapport with the NZAA head office personnel during the pilot study.
Recruitment commenced on 1 February 2006 and finished on 31 January 2008. Of the 3992 study participants, 2685 (67%) were recruited at a licensing centre, 916 (23%) at learner licence courses, and 391 (10%) by postcards. At the licensing centres, the recruitment rate across all research assistants was approximately 75% and for the learner licence courses it was around 90%. We were unable calculate a recruitment rate for postcards as we did not know the denominator.
Explanatory/exposure measures—Figure 1 provides an overview of the explanatory and the exposure measures, and the stage of the study when this information is collected.
Outcome measures—We intended that official data be sourced for traffic offences (any recorded breach of the Road Code), and injury crashes. We got participants consent to access crash and offence data at learner, restricted and full licence stage for up to 10 years after full licence. Our pilot work indicated a significant portion may not give consent for access to official data, especially for police records, and the period for which we sought it. In reality, this was not a problem. Of the 3992 study participants 98% gave consent to access all the listed traffic and injury records, 1% refused access to all records, and 1% denied access to some records.
For the 98% who gave consent, we have successfully accessed the traffic infringement and offence and convictions notices from the Driver Licence Registry (with assistance from the Department of Justice who hold these records, and not the NZ police as stated in the protocol) using the driver licence number. As at 31 December 2010, 1589 participants had at least one traffic offence, and there were 5751 offences in total.
We have also been successful in accessing crash information from the NZ Crash Analysis System (CAS), which is maintained by the Ministry of Transport. Crashes in which the study participant was a driver are identified using their driver licence number, and verified by personal data such as name and date of birth.
To identify crash records with no licence number name, date of birth, and address variables were used in the linkage. These variables are recorded on the bulk of the crash records. As at 31 December 2010 the participants had been involved in 143 police reported crashes.
Since official crash reports are known to have significant biases associated with them 8 we also sought information from the participants about injury crashes. The intention is to use official and self-report information to derive a single outcome measure of crash experience.
Follow-up of the cohort—Central to the study design was the expectation that participants who had obtained their learner licence would be followed up when they got their restricted and full licences. Critical to achieving this was the establishment of a study specific database by the New Zealand Driver Licence Registry (DLR) that would allow us to track the participants’ progress as they moved through the licensing system. With the cooperation of the DLR we achieved this objective. Each week the DLR provide us with an updated electronic file of any changed records in the database. A changed record indicates progression to the next licence stage, and the eligibility of the participant for the next licence stage interview.
Figure 2 is a snapshot in time of participants’ progress through the licensing stages and demonstrates that our tracking has been very successful. As at 30 April 2011, 74% had passed their restricted licence test and 87% of these completed the restricted licence interview, and 39% had passed the full licence test and 93% of them had completed the full licence interview.
Figure 2. Cohort progression as at 30 April 2011
After a minimum of 2.5 years had elapsed since obtaining a learner licence we conducted telephone interviews to ascertain reasons for non-progression to the restricted licence. Up to six attempts were made to contact non-progressors. As at 30 April 2011 we had attempted to contact all non-progressors. Of these 548 (46%) have been interviewed, 101 (8%) refused, and 540 (45%) could not be contacted.
Sociodemographic and behavioural characteristics of the cohort—One of the objectives of the recruitment process was to recruit a cohort that reflected the geographic and cultural diversity of the newly licensed driving population of New Zealand, and to recruit sufficient Māori to be able to conduct a separate analysis for Māori. As noted above, to achieve this, recruitment was undertaken in regions that represented both rural and urban areas, North and South Islands, in ethnically diverse communities, and included several regions with a relatively high proportion of Māori in the population.
Table 1 shows the sociodemographic characteristics of the study population at the learner licence stage. Gender was evenly distributed and, the majority were young with 77% aged 17 years or younger. The results show we successfully recruited an ethnically diverse cohort, with 825 (21%) and 516 (13%) identifying as Māori and Pacific origins, respectively. It should be noted that participants could identify with as many ethnic groups as they wished.
Table 1. Sociodemographic characteristics of the NZDS participants at learner licensing stage
The residential address of each study participant was classified according to Statistics New Zealand “Urban/rural profile”.9 The majority (89%) lived in an urban location.
The NZDep2006 score10 is a measure of socioeconomic deprivation created by combining nine variables, which reflect eight dimensions of deprivation, from the 2006 census. The scores in Table 1 show a reasonably even distribution (8%–10%) across all levels, except for the level of highest (16%) and lowest (12%) deprivation.
Table 2 shows a similar degree of diversity within behavioural factors. Scores for impulsivity/sensation seeking and aggression/hostility were derived using Zuckerman’s personality measure.11 Of note, there are substantial numbers at the extremes on measures of impulsivity/sensation seeking and aggression/hostility. The cohort also includes significant numbers of hazardous drinkers (measured by the AUDIT-C12 ) and occasional and regular drug users. Also of significance is that half the cohort had driven (illegally) on a public road prior to obtaining their learner licence and approximately a quarter have either been in a crash themselves or know someone who had.
Table 2. Behavioural characteristics of the NZDS participants at learner licensing stage
Parents’ study—One objective of this study was to examine the role of parents/caregivers as supervisors of newly licensed drivers. As planned a parent or caregiver was eligible for the parents’ study if their child:
In the protocol we proposed to recruit the first 1000 eligible parents. This, however, ran the risk of recruitment bias as 15–17 year olds recruited in the latter part of the recruitment phase would have had insufficient time to progress to the restricted licence stage, and therefore their parents would not be eligible for the study. Accordingly recruitment for the parent study continued until August 2008, seven months after the NZDS cohort recruitment phase finished. This meant that every 15–17 year old held their learner licence for the minimum length of time (6 months), and thus could progress to the restricted licence stage, should they have chosen to.
Figure 3 shows the numbers that met the successive recruitment criteria. In all, 85% of the parents invited to take part in the study agreed to do so, resulting in 1200 parent participants (773 mothers, 427 fathers).
Cohort studies provides excellent means of providing insight into the factors associated with adverse outcomes as young drivers commence their driving careers. There are major challenges to mounting such a study, nevertheless we have been successful in mounting such study, albeit with some limitations.
While the study population size was less than we had planned for, as indicated above, it has not seriously compromised the statistical power of the study. It was particularly pleasing that we were able to achieve sufficient numbers of Māori to undertake a separate Māori analyses. This is important for the credibility of the study in the New Zealand context. Internationally, given the dearth of injury epidemiology which is focused on indigenous populations we are well placed to add significantly to knowledge in this area.13 This process has already commenced with an examination of unlicensed driving among Māori 14 and another on attitudes and opinions of newly licensed drivers. 13
The greatest challenge to the study to date has been the recruitment. By extending the recruitment period to two years instead of one, we were able to achieve an adequate number of study participants. This was especially important for Māori, as an analysis solely for Māori, was a specific objective of this study. The Māori community/iwi groups that undertook the recruitment must be given credit, for making this possible. Our experiences confirm those of others who have shown community engagement and local assistance is critical to successful recruitment of indigenous people. 16,17
Critical to the success of this study is our ability to track outcomes for all participants, via official data sources. Our very high levels of consent to access outcome data (traffic crashes, infringements and offences) during the licensing years, and for ten years after obtaining a full licence, means we can access outcomes for 98% of the cohort, on an ongoing basis. Also, as we are able to link the official outcome data to our study data using driver’s licence numbers, because the majority of crashes do have the driver licence number recorded on the file, we are confident we can access complete data. The risk of bias in terms of those we do not have outcomes information for is minimal.
A major threat to validity of the findings of a cohort study such as this is attrition bias due to failure to monitor some participants through the licensing stages. Figure 2 shows that we have been very successful in tracing participants and subsequently obtaining follow-up information from them. It should also be noted that the response rates to interviews at the restricted and full licence stages have been very high at 87% and 93%, respectively. Failure to follow-up at the restricted licence stage should not be interpreted as precluding follow-up at the full licence stage.
Currently, we have successfully followed up at the full licence stage two thirds of those we failed to follow-up at the restricted stage. In this context it is important to note that we have followed-up approximately half of those who have failed to progress from the learner to the restricted licence stage, to ascertain their reasons for failing to progress through the graduated licensing system and to obtain information on driving behaviour comparable to that obtained from those who have progressed to the restricted licence stage. .
The characteristics of the study population showed substantial heterogeneity in demographic characteristics at the learner licensing stage. For example, the range of ages potentially allows us to determine the independent contributions of age (maturity) and driving experience to crash risk. The relatively small number of participants from rural settings will restrict our ability to determine rurality as a factor for various outcomes. Aside from the sociodemographic factors, the results show we have heterogeneity in terms of impulsivity/sensation seeking, aggression, alcohol and other drug use, and road safety experience. Of particular note is that we have a significant number of the cohort who had been driving (illegally) on public roads prior to obtaining their learner licence.
Our results also show high response rates, and thus low risk of bias, for all the items in Table 2. Even the more sensitive issues, such as drug use, had less than 4% who did not complete the item.
The recruitment for the parent study was successful with 85% of parents agreeing to participate. This is a unique feature of this cohort study of young drivers and will enable us to determine how central parents are in how young drivers progress, or do not, through the graduated licensing system.
A limitation of this study is that we have not recruited a representative cohort. This means that the prevalences we report cannot be considered to represent young drivers in New Zealand. This decision was deliberate since recruitment of representative cohort would have required us to have a far larger cohort to ensure, for example, we had sufficient Māori to allow separate analyses. This would have been very expensive and not an efficient means of addressing our primary objectives. Irrespective we considered the recruitment of a representative cohort would have been extremely difficult, if not impossible, given high refusal rates for certain sub-populations of young drivers. As has been demonstrated by Dunedin Multidisciplinary Health and Development Study non-representative cohorts have the potential to make a substantial contribution to our knowledge.
Although analyses addressing the primary objectives of the study have only recently commenced other results from the NZDS are achieving our main aim, that is to influence young driver policy in New Zealand. Results from the NZDS have been used by the Ministry of Transport in the development of New Zealand’s young driver policy18, and learner driver policy currently under development.
Results from the parents’ study have been requested by the New Zealand Transport Agency to assist in the development of a young driver programme for parents. We are also currently preparing a number of papers of direct relevant to policy. For example, the government recently indicated that as one of its supporting actions to increase the safety of young drivers it intended to investigate the introduction of maximum licence time limits for learner and restricted licences.19
Despite the fact that staged progression is central to graduated licensing schemes there is limited information on the sociodemographic and behavioural characteristics which are associated with non-progression, whether non-progression results in negative traffic outcomes, and why non-progressor choose not to progress. We have recently addressed this issue and the findings have been forwarded to the Ministry of Transport.20
A significant investment was made in pretesting and piloting for the NZDS. This has paid off as there have only been minor variations to the protocol for the main study. Moreover the study population displays significant variation in range of factors at the learner licensing stage. The NZDS study has already made a contribution to the development of young driver policy and, as has been demonstrated here with the example of non-progressors, it is well placed to significantly increase that contribution.
Competing interests: None declared.
Author information: John Langley, Professorial Research Fellow1; Dorothy Begg Senior Research Fellow1; Rebecca Brookland, Research Fellow1; Shanthi Ameratunga, Professor2; Anna McDowell, Statistical Analyst3; John Broughton, Associate Professor4
Acknowledgements: This study was funded by the Health Research Council of New Zealand, Accident Compensation Corporation, the Road Safety Trust (2005–2009) and is currently funded by the Health Research Council of New Zealand and the Road Safety Trust (2009–2013). The authors also gratefully acknowledge the assistance of all the research assistants and community organisations that recruited the study cohort, and the study participants for their ongoing contribution to this study.
Correspondence: Professor John Langley, Injury Prevention Research Unit, Dept of Preventive and Social Medicine, Dunedin School of Medicine, University of Otago, PO Box 56, Dunedin 9054, New Zealand. Fax: +64 (0)3 4798337; email: John.Langley@otago.ac.nz
issue | Search journal |
Archived issues | Classifieds
| Hotline (free ads)
Subscribe | Contribute | Advertise | Contact Us | Copyright | Other Journals