This follow-up evaluation of the GF2R programme showed that the positive changes in physical activity and body composition observed at the end of the intervention  were not sustained, and almost 2 years after the intervention there were no significant differences between the two groups for physical activity (both groups had increased physical activity) and some evidence for poorer outcomes in body composition within the intervention group. The general increase in physical activity across both groups may reflect seasonal differences in physical activity as the baseline measures were conducted in autumn/winter and the follow-up measures were conducted in spring/summer.
At post intervention both groups had increased their physical activity but the increase was significantly greater in the intervention group (e.g., 1532 steps/day greater increase in the intervention group ). At follow-up the difference in increase was smaller and non-significant (e.g., 678 steps/day greater increase in the intervention group). Given that the difference between groups is much lower at follow-up the non-significant result is more likely to be a true reflection of no difference between the groups rather than a result of loss of power due to diminishing sample size. There are a number of possible explanations for the lack of sustained behaviour change observed.
The intervention itself lasted for only 10 months, which may have been insufficient. Additionally the intervention may not have been intensive enough. Two primary school interventions that have demonstrated sustained intervention effects over 3  and 4  year follow-ups were longer interventions in the first place (3 school years  and 6 school years ) and also involved a much more intensive programme of intervention. In the current study teachers were free to use their professional judgement to choose which parts of the intervention would best work for them and their pupils, this may have resulted in low levels of exposure to different components of the intervention. Additionally, by conducting the intervention over a longer period of time participants would have been repeatedly exposed to key messages and intervention activities potentially resulting in a longer lasting effect. Even in the two studies [4, 5] which continued to show a significant intervention effect for physical activity, the differences between intervention and control groups were narrowing in magnitude over time  suggesting that more research is needed to investigate the best way to create sustained change as children progress through to adolescence and adulthood.
Long-term interventions may be particularly important in children as the type and purpose of physical activity undertaken varies with age. At young ages basic movement patterns are developed which form the foundation for activity at later stages . With growth, maturation, and experience, these basic movements are coordinated into more complex movement patterns that characterise the free play, games and sports of older children . Malina  suggested that until approximately 8 - 10 years the main emphasis is on greater physical activity and particularly motor skills. After 8 - 10 years, the emphasis becomes increasingly focused on prescriptive physical activity, with an emphasis on health, fitness and behavioural outcomes. These changes, alongside other physical, social and cognitive changes occurring through childhood and adolescence perhaps suggest that long-term interventions that adapt to the changing needs of the young person are required to support sustained engagement with physical activity, and that it is perhaps unrealistic to expect long-term impact of a one year intervention within such a dynamic system. It is likely that the nature and content of the interventions will need to vary as children develop, and there is evidence of programme evolution in both Manios  and Nader . van Sluijs et al.  suggested that traditional cognitive approaches, potentially combined with environmental approaches, may increase activity among adolescents and older children (> = 10 years), but more structural environmental or policy changes might be needed to change younger children's physical activity.
The variance estimates for the physical activity measures reported in Table 2 demonstrate that over time the variance within an individual is greater that the variance between individuals. This is most likely to be a reflection of the developmental changes discussed previously. This suggests that behaviour changes as children develop, and for example gain greater autonomy and independence, are greater within an individual than the heterogeneity between subjects at any one point in time.
As with other studies  the focus groups in this study highlighted the importance of parents in promoting physical activity in children. The role of parents may be particularly important in maintaining change through the provision of ongoing encouragement and tangible support for participation. Greater emphasis on engaging and supporting parents within school-based interventions may be required to facilitate long-term change.
By far the majority of teachers in this study had not continued to use the resources provided during the intervention period. While the teachers provided many potential explanations for this, an important explanation may lie in the philosophy of the intervention itself. The intervention did not mandate the use of any resource and teachers were free to choose what to use. In an environment where teachers constantly seek to meet changing curriculum demands and emphasis from regulatory authorities (see comments in focus groups) it may be that teachers do not have the time to embed successful intervention strategies long-term because they have been pulled off in a different direction. Continued support to teachers and emphasis on the outcomes of strategies may be needed to make sure that successful strategies are not overlooked in the future, or a strategy of phased support may be required. For example, Haerens et al [23, 24] reported on a 2-year long school-based intervention in which the support offered to teachers reduced over time. In the first year the teachers were provided with guidance and support from the research team to help get the intervention started, but in the second year this external support was decreased with the intention of increasing the autonomy of schools. It was hypothesised that the second year would not lead to additional positive changes but it was hoped that the original changes would be sustained. Results showed significant positive intervention effects for physical activity at year 1  which were sustained during year 2 . While the long-term effects of this intervention have not yet been evaluated, the strategy of phased support may provide one avenue to ongoing intervention success. This would support the comments from the teachers in the current study who suggested that continued input from the intervention team would have been helpful and welcomed.
Teachers and pupils alike recognised the importance of the highlight events within the intervention but opportunities like this were not subsequently provided by the schools. This is not surprising given the many time demands and responsibilities of teachers. Ways to continue to provide highlight events without increasing the demands of teachers need to be explored. Likewise, pupils liked the holiday materials and other resources provided and reported that they thought they were useful, however, very few children had taken the idea of planning forward and continued to use some sort of action planner. Further understanding of how best to facilitate long-term use of such approaches is required.
The body composition results at follow-up are difficult to explain from the data available, particularly in light of the general increase in physical activity in both groups. Nader et al  also reported no significant intervention effect for body composition variables at 3-year follow-up in the CATCH trial. Singh et al  reported on the results of an 8-month multi-component health promotion intervention aimed at preventing excessive weight gain in young adolescents (12-14 years at baseline). Intervention effects at the end of the program and at 4-month and 12-month follow-up were presented. At the 12 month follow-up intervention effects remained for sum of skinfolds in girls. However, no intervention effects were observed in sum of skinfolds at any time point in boys, and no intervention effects were observed for BMI at any time point in both boys and girls. It is obvious that challenges remain in identifying effective strategies that result in long-term positive changes in body composition among youth. The possibility of negative rebounds when interventions are removed needs further investigation and may have important implications for the maintenance of a healthy body composition.
Although this study has several strengths (e.g., objective measures of physical activity, multiple measures of body composition) several methodological limitations should be acknowledged. Although 70% of participants were still present at follow-up the loss to follow-up was greater in the intervention schools. Overall, there was no difference (p > .05) at baseline in age, body composition, steps/day, or minutes of MVPA between those present at follow-up and those absent. Due to the local media content it was not possible to conduct a randomised control trial. However, schools were matched on key variables and there is debate as to the appropriateness of randomised control trials for evaluating health promotion interventions [26, 27]. The group level matching was not reflected at the individual level resulting in the intervention group being of lower socioeconomic status than the control group. However, including socioeconomic status in the analytical models did not improve the fit of the models suggesting that these differences did not influence the outcomes.
The initial evaluation of the GF2R programme showed that the strategies employed within the intervention were effective in producing short-term changes in physical activity and body composition; however, this follow-up evaluation shows that the changes were not sustained. Thus questions remain as to how to effect long-term favourable changes in health behaviours in young people. Longer term interventions, with greater links with families are most likely required but the exact nature and contribution of this involvement remains unclear . Further support to schools and teachers is also likely to be required but the best way to provide this within an already busy curriculum needs further attention.