There is some evidence that children who attended Lifeskills at the age of 10–11 years had enhanced safety skills and attitudes in adolescence. Attenders were more likely to report that they always/mostly used pedestrian crossings on their way to school, were more likely to report that binge drinking was very harmful to physical health, and were less likely to report recent smoking or occasional cannabis use. However, for many of the outcomes, we did not detect differences between those who had and had not attended Lifeskills.

Of the road safety related outcomes, the use of pedestrian crossings was the only one to be associated with Lifeskills attendance. Choosing where to cross the road is something most of the children had some autonomy over (the majority walked with friends, siblings, or on their own; few walked with adults). Improving road safety on the school commute is a key priority as most road accidents involving children occur during school commuting times: an average of 16 children per week were killed or seriously injured between 8–9 am and 4–7 pm in England between 2008 and 2012, with those from more deprived neighbourhoods being most at risk (Public Health England, 2014). Lifeskills training aims to influence a child's road crossing behaviour through teaching children to recognise the risk posed by cars, that cars have long stopping distances, and that pedestrian crossings are the safest place to cross. These safety messages are pertinent to the school commute as research suggests that car drivers frequently do not comply with speed limits in school zones (Strawderman et al., 2015, Ellison et al., 2011). In the earlier evaluation of Lifeskills, knowledge of car stopping distances was one of the best acquired and retained pieces of information (Lamb et al., 2006).

Children aged 10–15 years are at greater risk of having a cycling accident than any age group other than adults aged over 60 years (Royal Society for the Prevention of Accidents, 2014), and accidents are more common in children from deprived neighbourhoods (Public Health England, 2014). In those who have an accident, head injuries are common and the majority of fatalities involve a head injury (Royal Society for the Prevention of Accidents, 2014), thus signalling the importance of helmet use at this age. In the current study, reported use of cycle helmets was low and use of reflective clothing very low amongst children whether or not they had attended Lifeskills. Only 37% of helmet owners reported wearing a helmet on their most recent cycle ride, and less than 4% of cycle owners had worn reflective or fluorescent clothing on their most recent ride. Cycle helmet ownership, and helmet use amongst those who owned a helmet, were more common in children from higher SEP families, in general concordance with previous research (Williams et al., 1997, Lang, 2007). It could be argued that it is difficult for Lifeskills to have a direct effect on these kinds of safety behaviours. For example, a child who did not own a helmet or reflective clothing prior to their Lifeskills visit would need not only to learn the benefits of such items and retain that knowledge, but also to articulate that knowledge to a parent in order to persuade them to acquire the items. Limited financial resources to buy safety equipment are obstacles to a child successfully adopting safety behaviours taught to them.

Health behaviour models have long indicated that knowledge, whilst necessary, may not be sufficient to result in behaviour change as it competes with inertia and other barriers, such as a belief that a given risk is too low for concern, or that the behaviour change required is at variance with social and behavioural norms (Ajzen, 2011, Noar and Zimmerman, 2005). For example, a study of teenagers in Oxford found high levels of knowledge about the benefits of cycle helmets, with 96% believing wearing a helmet reduces the risk of head injury, but also high levels of perceived barriers to their use with 71% thinking they ‘looked ridiculous’, and many saying their friends ‘discouraged them from wearing one’ (Joshi et al., 1994). Children's attitudes towards helmets, for example in terms of their benefits and how they look, are strong determinants of whether they wear one (Lang, 2007). To be successful, interventions need to address not only knowledge but social attitudes to the behaviour in question. They also need to be relevant to young people irrespective of social background to ensure social inequalities in health and safety behaviours are not inadvertently increased (White et al., 2009).

The results for substance perceptions and use are in agreement with the need for safety education to address more than knowledge. The vast majority of ALSPAC adolescents reported that alcohol, smoking and cannabis were ‘harmful’ or ‘very harmful’ to health when aged 14, yet many were soon using these substances themselves. By 15.5 years almost one in five reported frequent drinking, almost 10% were weekly smokers, and a similar proportion reported occasional cannabis use.

This is the first evaluation of Lifeskills with relatively long-term outcomes. We have been able to examine differences in road safety behaviours, and the perceptions of the health effects, and use, of substances, between those who did and did not attend Lifeskills using a large, longitudinal population-based cohort. We were able to adjust for some potential confounders and to take into account clustering at the school and neighbourhood level. Furthermore, we have examined both self-reported outcomes and outcomes from linkage to hospital records.

Our study does however have limitations. None of the ALSPAC measures were designed a priori to evaluate Lifeskills. For the cycle safety measures, use of a cycle helmet and fluorescent/reflective clothing refer only to the last time cycled. We have no information on the purpose of that cycle ride, or the time of day. The main objective of the Lifeskills set on substances was for the child to be able to distinguish between drugs that do good (i.e. medicines) and drugs that harm, between prescription and over-the-counter medicines, and between illegal and legal drugs. Our outcome measures are not a close match to this objective, although it can be assumed that the long-term goal of teaching children about drugs that harm is ultimately to reduce their use.

Lifeskills attendance is not recorded at the individual level and so there is a risk of misclassification in attendance status (e.g. if a child moved school part way through Year 6), which would attenuate results towards the null. The analysis of A&E attendance was compromised by the lack of data for the years immediately following the ALSPAC children's visit to Lifeskills, when we would assume Lifeskills would have the greatest potential for accident reduction. Available codes in the A&E data did not make it possible to identify causes directly related to a Lifeskills outcome. As the study sample was relatively affluent, and several outcomes socially patterned, the overall prevalences could be under-estimates of risky behaviours, or over-estimates of safe behaviours, compared to adolescents in the general population. As the section of the population most likely to have accidents was under-represented in our sample (Towner et al., 2005), this will have reduced our power to detect small effects. Finally, with the exception of cycle proficiency training, we have no information on other relevant training the children may have received (e.g. drug safety lessons in school); such training may have contributed to, or masked, differences in the outcomes by Lifeskills attendance status if the likelihood of receiving other training was associated with Lifeskills attendance.