Among young Swedish drivers, injury in an RTC as an unlicensed driver occurs not only prior to the age of licensing eligibility but also thereafter, and at a rather stable rate until the age of 27. The study is restricted to unlicensed drivers and does not consider those driving while having a permit suspended or revoked. The rationale for this focus is that not being licensed at all is more a reflection of people not engaging in - or completing - the driver training process even after several years of becoming eligible. This is intriguing, seldom reported, and deserves attention. Further, the RTCs involving unlicensed drivers differ from those involving their licensed counterparts with regard to both crash circumstances and injury severity. As more risky driving practices have been associated with severe and fatal RTCs, it is not surprising that crash severity is higher among unlicensed drivers [18, 22]. Whether those risks are specific to crash circumstances or reflect a trait more typical of unlicensed drivers in general remains to be determined. Yet, self-reported studies of young people also indicate that unlicensed driving tends to occur as "joy-riding" and without purpose, typically at night and weekends, and while under the influence of alcohol [16, 17].
In addition, the study shows that being involved in a severe crash as an unlicensed driver is more common among young people who are not from families of the highest socioeconomic position and who live outside metropolitan areas. The former finding echoes an earlier Swedish study on young licensed drivers,[7–10] but we still lack information as to whether the mechanisms are comparable. Is driving unlicensed more prevalent among young people not from the highest socioeconomic position (e.g. licensing barriers) ? Are they more prone to risk-taking (e.g. crash likelihood differences) [21, 24]? Or are the consequences of the crashes they are involved in more severe (e.g., protection differences) ? All three hypotheses are plausible.
The licensing process provides the driver with the minimum skills and experience needed to operate a motor vehicle safely. There are indications that the opportunity to prepare for a license are unequal as a result of less access to vehicles and poorer driving opportunities for young people from families of lower socioeconomic position [10, 23]. The fact that unlicensed drivers are twice as likely to be involved in a single-vehicle crash in which they lost control of the vehicle suggests their lack of formal driving preparation is a factor.
Unlicensed drivers from self-employed families had a higher risk estimate for severe injury than has been reported in other studies [7, 9]. We found that they are more likely to be licensed earlier indicating a need to be mobile perhaps as part of a family business . It seems as though - and not surprisingly -younger-age driving combined with increased exposure increases the risk.
Finally, unlicensed drivers living in rural areas, compared to those living in metropolitan areas, showed a much higher risk for RTCs with severe injury outcome. The imbalance is possibly due to driving exposure because of, inter alia, the necessity of travelling greater distances in higher speed areas, and the lack of commuting alternatives such as public transport . Even factors like inadequate pre-hospital care after a crash may influence the unequal geographical distribution of the most severe injuries in rural areas [27, 28]. Whether the latter applies to the distribution of unlicensed driving in Sweden is not known.
This study contributes important and new information regarding RTCs involving unlicensed drivers. The combination of linking multiple databases containing population, socioeconomic, and crash data provides important insight into the social stratification of RTCs and RTIs. Our study population consists of a large cohort from the Swedish Population and Housing Census Database that is fully representative of the Swedish population and continually updated. License status was gathered from the National Driver's License Register and covers all licenses issued in Sweden. One limitation is that we did not have information regarding revoked licenses, implying possible misclassification of some young people as licensed, and a slight underestimation of the incidence of unlicensed RTCs.
The Swedish National Road Administration Register covers all police-reported RTCs during the seven-year follow-up. It is, however, well known that police crash reports do not give an exhaustive picture of the number of RTCs, especially underestimating RTCs that do not give rise to serious injuries. The police may pay closer attention to specific persons in a crash, especially if that person is a young driver suspected to be under the influence of alcohol/drug. Accuracy and completeness of crash data are also restricted to the reporting and subjective assessment of the police at the scene .
In the cohort analyses, all exposures were assessed through registers implying reduced risk for information bias. However, exposure was assessed at the time of inclusion. The young people were classified according to the socioeconomic position of their parents. For the early birth cohorts, aged 20 at inclusion, this might be misleading as during the seven-year follow-up period, they may establish their own socioeconomic position independent of the family's. Upward social mobility for young people in this study would lead to an underestimation of the relative risk among lower socioeconomic groups.
Confounding in population-based studies of road traffic safety is difficult to control even under the best of circumstances. Our estimation of person-years at risk, based on time of license status, does not take into account the extent to which young people from different socioeconomic groups and levels of urbanicity have similar driving profiles in terms of conditions, types of vehicles, and distances driven. Included in the rate are licensed and unlicensed drivers who may have zero driving exposure. The commitment to road traffic safety in Sweden is supported by culturally and socially defined norms of acceptance and compliance with traffic safety measures that possibly contribute to decreased exposure among some socioeconomic positions .
Young people's access to a vehicle is highly dependent on the availability of a family car and household disposable income. However, car ownership is high in Sweden with 86% of the subjects- families having a registered car during 1998. Adjusting our analyses and taking into account household disposable income and car ownership did not alter our conclusions.
The results can be generalized to other settings in high-income countries with similar socioeconomic differences and motor traffic systems. Even though fatal crashes for both licensed and unlicensed young drivers are relatively rare in Sweden, the social patterning and area distribution of RTCs among unlicensed drivers may be similar in other countries. Access to population-based socioeconomic and crash data in Sweden is important in understanding the mechanisms of unlicensed driving.