Evaluating the effect of a nationwide school policy on physical activity and body mass index

Background: Physical activity is important in health promotion and disease control. Since physical activityaffects energy expenditure, which together with energy intake determine energy balance, physical activity isalso important in weight gain prevention. Despite the well-known effects of physical activity across a lifespan,meeting international guidelines is a growing problem, and primary prevention early in life is essential.Schools are considered an important setting for physical activity promotion since children and adolescentsspent more than half of their waking hours in schools. In 2014, the Danish Government introduced a newschool policy mandating an average of 45 minutes of physical activity daily during teaching. The policy appliedto all public schools across Denmark, and the potential public health benefits of the physical activityrequirement were major. Similar policies have been introduced internationally in Canada, Norway, andHungary, among others, but evidence on the effects of such policies on device-measured physical activityand body-mass index is sparse. The objectives of this thesis were to I) present the study protocol of thePHASAR study, II) examine the effect of the school policy on device-measured school time and full dayphysical activity, and III) examine the effect of the school policy on mean BMI and the 90th BMI percentile. InPaper I, methods of Paper II and III are presented. Thus, Paper I is not presented explicitly.

Methods: No routine-surveillance of device-measured physical activity existed in Denmark before 2018. Thus, pre-policy data in Paper II were composed of device-measured physical activity data from four suitable historical school-based studies completed between 2009-2012. The post-policy population was carefully matched pre-policy study populations, and data were collected at the same schools and age groups during the same seasons as pre-policy. The post-policy data collection was completed in 2017/18 (The PHASAR study). Children were eligible if they had accelerometer measurements and did not have any physical disabilities preventing activity. A total of 4,816 children and adolescents aged 6-17 were included in analyses from the pre- and post-policy populations (2,346 pre-policy and 2,470 post-policy). Physical activity was objectively measured by accelerometry, and main outcome was any bodily movement. Secondary outcomes were moderate to vigorous physical activity and overall movement volume (mean counts per minute). All outcomes were analyzed during a standardized school day (8:10am – 1pm) and a full day. The interrupted time-series approach was used to evaluate any effects of the policy. Time was treated categorical (2009/10, 2010/11, 2012 and 2017/18). Data were analyzed using mixed-effect linear regression that were adjusted for sex, age and season as fixed effects, and individuals, schools, and historical projects as random intercepts. Linearity in pre-policy trends were initially examined in all three outcomes. If pre-policy linearity existed, post estimations were conducted to observe whether linearity continued post-policy or were interrupted by the introduction of the school policy. If no linearity existed pre-policy, a bootstrap method with 1,000 replication was applied to test whether post-policy physical activity in 2017/18 was higher than any levels measured prepolicy.  

In Paper III, body-mass index (kg/m2 ) was the primary outcome. Population representative data on height and weight from the Child database (Børnedatabasen) in the period 2012-2018 was used. The database contains data from preventive health examinations completed by school nurses in pre-preparatory classes (0th -3 rd grade) and lower secondary education (7th -9 th grade). Data from 401,517 children were included in analyses. The effect of the policy on mean BMI and the 90th BMI percentile was evaluated using an interrupted time-series approach. Differences in pre- and post-policy change rates in mean BMI were evaluated using mixed effect linear regression. Differences in the 90th BMI percentile pre- and post-policy change rates were evaluated using quantile regression. All analyses were stratified in two age groups (prepreparatory classes and lower secondary education), and an interaction term between time and sex was included. Post-estimation was used to test differences in pre- and post-policy change rates. All analyses were conducted in StataBE17.

Results: In Paper II, linear negative pre-policy trends in movement, moderate to vigorous physical activity and counts per minute were interrupted during a standardized school day after the introduction of the school policy. Increases of 14.2 minutes movement (95% CI: 11.4; 17.0, p<0.001), 6.5 minutes moderate to vigorous physical activity (95% CI: 4.7; 8.3, P<0.001), and 141ꞏ8 counts per minute (95% CI: 108.5; 175.2, P<0.001) were observed post-policy. During a full day, all outcomes developed non-linear pre-policy, and no postpolicy estimates went beyond pre-policy estimates. In Paper III, post-policy change rates in mean BMI slightly increased in both age groups in girls (0th -3 rd grade: β:0.034 kg/m2 , 95%CI: 0.024; 0.043, P<0.001; 7 th -9 th grade: β:0.066 kg/m2 , 95%CI: 0.028; 0.103, P: 0.001). No changes were observed in post-policy mean BMI in boys in neither age group. Moreover, no changes in post-policy change rates were observed in neither subgroup in the 90th BMI percentile. 

Conclusion: The findings suggest that the school policy succeeded in increasing school time PA. However, no increases were observed in PA during a full day, and the policy was not able to reverse the increasing prepolicy trend in BMI. On the contrary, slight increases in mean BMI were observed post-policy in girls. BMI and leisure time PA (which affect full day PA) are more vulnerable to confounding factors compared to school time PA, and the small increases observed post-policy in mean BMI in girls may be caused by unmeasured confounding.