Results from accelerometers showed children in the intervention schools being more active compared to those in control schools after one year of intervention. The difference at that time point was driven by boys in the intervention group, who at that point had increased their amount of MVPA significantly more than the intervention girls. No group difference was detected at the end of the intervention period a year later. Boys in both groups were consistently more active than girls at all three time points, and those children with higher BMI were less active during school hours and after school hours, but there was no association between BMI and physical activity during after-school hours.
It is important to evaluate the intervention process of every intervention study conducted before the efficacy of the program on other biological or social variables is assessed. Was the intervention conducted as planned, and if so, to what extent were the objectives met? To our knowledge no study has used the combination of methods described herein to assess changes in school-related physical activity during a two-year school-based intervention program. It strengthens the results that there seems to be harmony between both objective assessment of physical activity and subjective assessment of physical activity implementation, to the extent that they can be compared. Both measurements show a parabolic curve-shape when the three time points where physical activity was objectively assessed are contrasted. Perhaps these findings are positive because objective and subjective measurements have often showed inconsistent, even contradictory, results, both in observational and experimental studies [11, 30].
The purpose of the extra PE lesson in the second year of the intervention was to considerably increase the amount of time where all participating children would get an opportunity to partake in physical activity of greater intensity. However, it seems clear that the increase in number of minutes engaged in MVPA during school hours in the fall of 2007 is driven by the boys being considerably more active at moderate-to-vigorous intensity, by on average about 10 minutes during school hours. These results are somewhat in line with results from the M-SPAN study, which conducted a two-year intervention focused on an environmental and policy-driven approach in middle schools in San Diego County, California where they did not see positive intervention effects on physical activity in girls . The authors claimed that challenges were anticipated since girls are generally less active than boys but the reasons for these differential effects were nevertheless unclear. Otherwise, there is little evidence for boys and girls responding differently to school-based interventions as well as to different components of the interventions . Our results nonetheless do suggest a future effort be made to test various gender-specific strategies given the gender-specific differences in the determinants of physical activity [32–34] and the results presented here. These results are nonetheless disappointing because the teachers reported during interviews that they had emphasized activities that they intended to be equally suitable for both genders.
Studies have previously reported conclusive evidence for gender specific differences concerning the amount of physical activity performed by children and adolescents [35–37], which is also confirmed at all time points in this study. In light of this fact we emphasized the importance that activities performed during PE and during regular class hours would suit boys and girls equally. None of the classroom teachers (all female) reported they had experienced gender differences in participation in the numerous activities performed during the intervention phase when asked specifically about it during the group interview sessions in the spring of 2008. However, there are several plausible explanations for this seemingly different response between the genders. First, the teachers' assessment may be wrong! Secondly, perhaps girls were not as active as boys during recess periods, when the children were not under surveillance by their teachers, in the fall of 2007. Another explanation may partly lie in how differently boys and girls may perceive barriers and facilitators of physical activity. A recent study of 350 adolescents in Maryland in the US, showed that when it came to performing physical activity, adolescent girls were more sensitive to their environment and perceived more barriers than boys . If this holds true for younger children then perceived environmental barriers may have contributed to this differential rate of increase in MVPA during school hours in fall of 2007. It is also worth mentioning that results from other intervention studies that have used objective measures to assess MVPA have shown significant increase in MVPA after and during the intervention, but do not show differential intervention effects on intensity nor volume between boys and girls [14, 15].
There are several possible explanations for the drop in school-related physical activity recorded by the teachers at the end of the study. We believe that the primary reason for this is that only two of the initial eight general teachers were still part of the study team at that time, while six of the teachers were either on maternity leave or had started teaching a different class. The majority of the new teachers had only received minimal training during a mere one meeting prior to the measurements being conducted in the fall of 2008. Another contributing factor to this drop could be that the extra PE lesson/week introduced in the fall of 2007 was no longer available to the children at the intervention schools. This may have significantly affected the amount of MVPA the children received during school hours. The intervention did not seem to have any effect on physical activity during after-school hours because no significant difference in volume (cpm) or intensity (minutes of MVPA) of physical activity was detected during this time of the day. This seems to be consistent with what Dobbins et al. concluded in their review that there is no evidence for positive effects of school-based physical activity interventions on leisure time physical activity in children .
Some of the strengths related to the implementation of this intervention are in line with strengths of comparable intervention studies . First, the results from the teacher interviews during the intervention phase may suggest that empowering the teachers to become effective implementers of positive change in physical activity during school hours may partly explain the increase. Second, the progressive intervention allowed for the possibility of on-site trial and error while the implementers slowly built up their skill set, enabling them to increase school-based physical activity throughout the school day. Third, this method respects a teacher's independence, as it allows individual teachers to adjust the physical activity-related activities at their own will. Finally, it is worth mentioning that the use of accelerometers as an objective measure of school-based physical activity eliminated the possibility of self-report bias, and using hierarchical models to analyze these results is an appropriate technique taking into account the clustered data structure.
There are methodological limitations to this study that must be addressed. Measuring school-based physical activity at the end of each school year, similar to what was done at the beginning of each school year, would likely have provided stronger indications of the intervention progression. Thus, while having three objectively obtained physical activity data points can be viewed as strength, an additional two in the spring of 2007 and spring of 2008 would have strengthened our inferential ability. Despite all the children being a similar age, the study population was relatively small, thus limiting the potential to generalize the findings. Further, we cannot state which specific part of the intervention contributed more than others to the overall increase in physical activity during school hours in the fall of 2007, i.e. if it was the extra PE lesson or integrated physical activity within the various general school subjects. It has also recently been pointed out that having the accelerometers record 60 s epoch is likely to have resulted in a less accurate estimation of physical activity than using shorter epoch like 15 s . This may have underestimated the amount of MVPA the children performed during school hours. Finally, the lack of consensus on where to place accelerometer cut-points defining moderate-to-vigorous physical activity limits our ability to accurately classify physical activity intensity.