Changes in physical activity and sedentary time in the Finnish Schools on the Move program : a quasi-experimental study

This is an electronic reprint of the original article. This reprint may differ from the original in pagination and typographic detail. Author(s): Title: Year: Version: Please cite the original version: All material supplied via JYX is protected by copyright and other intellectual property rights, and duplication or sale of all or part of any of the repository collections is not permitted, except that material may be duplicated by you for your research use or educational purposes in electronic or print form. You must obtain permission for any other use. Electronic or print copies may not be offered, whether for sale or otherwise to anyone who is not an authorised user. Changes in physical activity and sedentary time in the Finnish Schools on the Move program : a quasi-experimental study Haapala, Henna; Hirvensalo, Mirja; Kulmala, Janne; Hakonen, Harto; Kankaanpää, Anna; Laine, Kaarlo; Laakso, Lauri; Tammelin, Tuija H. Haapala, H., Hirvensalo, M., Kulmala, J., Hakonen, H., Kankaanpää, A., Laine, K., Laakso, L., & Tammelin, T. H. (2017). Changes in physical activity and sedentary time in the Finnish Schools on the Move program : a quasi-experimental study. Scandinavian Journal of Medicine and Science in Sports, 27(11), 1442-1453. https://doi.org/10.1111/sms.12790 2017 Changes in physical activity and sedentary time in the Finnish Schools on the Move program: a quasi-experimental study Henna L. Haapala 1,2 , Mirja H. Hirvensalo 2 , Janne Kulmala 1 , Harto Hakonen 1 , Anna Kankaanp€ a€ a 1 , Kaarlo Laine 1 , Lauri Laakso 2 , Tuija H. Tammelin 1 1 LIKES Research Center for Sport and Health Sciences, Jyv€ askyl€ a, Finland, 2 Department of Sport Sciences, University of Jyv€ askyl€ a, Jyv€ askyl€ a, Finland Corresponding author: Henna L. Haapala, LIKES Research Center for Sport and Health Sciences, Rautpohjankatu 8, FI-40700 Jyv€ askyl€ a, Finland. Tel. +358 40 8311 303, Fax: +358 20 762 9501, E-mail: [email protected] Accepted for publication 27 September 2016 The aim of the Finnish Schools on the Move program is to create a more active and pleasant school day through physical activity (PA). In this quasi-experimental design, we compared changes in moderate-to-vigorous-intensity physical activity (MVPA) and sedentary time (ST) during the school day and outside school hours for Grades 1–9 over two academic years in four program schools and two reference schools. Altogether 319 girls and boys aged 7–15 participated in the study between 2010 and 2012. MVPA and ST were measured four times over the 1.5-year follow-up period for seven consecutive days, using a hip-worn ActiGraph accelerometer. Linear growth curve modeling was used to examine the effect of the program on MVPA and ST during follow-up. School day MVPA increased (P=0.010) and school day ST decreased (P=0.008) in program primary schools (Grades 1–6) more compared with the reference schools. The effect sizes (Cohen’s d) for the difference in change (from the first to the last measurement) were small (d=0.18 and d=0.27, respectively). No differences in the changes of leisuretime or whole-day MVPA and ST between the program and reference schools were observed during follow-up. In conclusion, the changes in school day MVPA and ST did not translate into positive effects across the whole day. More effective and longer promotion actions are needed for positive changes in PA and ST, especially in lower secondary schools and for all daily segments. Insufficient levels of physical activity (PA) and a sedentary lifestyle can produce global public health issues and the economic burden of non-communicable diseases, i.e., Type 2 diabetes, cardiovascular diseases, and cancer (Lee et al., 2012), often developing already in childhood (Fernhall et al., 2011; V€ aist€ o et al., 2014). In addition, PA in childhood and adolescence also relates to improved academic performance (Singh et al., 2012), cognitive functioning (Donnelly et al., 2016), and mental health (Biddle & Asare, 2011). However, many children do not meet the daily recommendation of at least 60 min of moderate-tovigorous-intensity physical activity (MVPA) and spend a lot of time in sedentary activities (Verloigne et al., 2012). Studies have shown a decline in MVPA and an increase in sedentary time (ST) during childhood and adolescence (Colley et al., 2011; Ortega et al., 2013). Accordingly, 34% and 47% in 11-yearolds and 13% and 22% in 15-year-old Finnish girls and boys, respectively, meet the recommended levels of daily MVPA (Bucksch et al., 2016). School is an important environment to promote PA due to its potential to reach each age group of children and also the inactive and unfit students. However, effect sizes of school-based interventions for PA have been moderate at best and knowledge on the most effective strategies remains limited (Dobbins et al., 2013). A recent meta-analysis suggested that PA interventions in children ages up to 16 years produced only four additional minutes, on average, of daily MVPA (Metcalf et al., 2012). Moreover, a comparison of regular schools and schools offering increased levels of mandatory physical education in the CHAMPSstudy DK showed that children with increased physical education were more active during school hours, but less active during leisure-time; thus, no increase was observed for overall PA during the day (Moller et al., 2014). School interventions that combine multiple elements from educational, environmental, and policy-based approaches appear to be effective in PA promotion for children and adolescents, and effective intervention components This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes. 1442 Scand J Med Sci Sports 2017: 27: 1442–1453 doi: 10.1111/sms.12790 ª2016 The Authors. Scandinavian Journal of Medicine & Science in Sports Published by John Wiley & Sons Ltd seem to be, e.g., providing active schoolyard and equipment, increasing the number and length of physical education lessons, and giving students PA homework (van Sluijs et al., 2007; Kriemler et al., 2011). The social–ecological model suggests that factors at multiple levels influence health-related behaviors, such as PA (Sallis et al., 2006). Therefore, individual, social, environmental, and policy domains should be considered in the development and implementation of schoolbased interventions. Recently, the conceptual framework Comprehensive School Physical Activity Program (CSPAP) from the United States has been developed based on high-quality physical education, and physical activity is added before, during, and after school with staff, family, and community engagement playing a vital role in PA promotion (Centers for Disease Control and Prevention, 2013). In addition, a curriculum approach to CSPAP called Health Optimizing Physical Education (HOPE) delivers a whole-school approach and community engagement to PA promotion (Metzler et al., 2013). Both of these multi-component programs aim to enable more students to meet the recommended 60 min of daily MVPA and ensure lifelong PA engagement. This study evaluates the results of the Finnish Schools on the Move program, a national action program to create a more active and pleasant school day through PA in Finnish comprehensive schools (Haapala et al., 2014). Physical activity is not only the goal here but also a tool used to improve other schoolrelated outcomes, such as school atmosphere, support for learning, and student empowerment. An important feature of this program is a bottom-up approach; the participating schools and municipalities apply for funding to implement their own individual plans to increase PA and decrease sedentary time (ST) during the school day. Specific actions are not required from the schools; however, the program supports the schools by disseminating best practices and ideas in national seminars and program webpages and providing opportunities for support from experienced local mentors (Haapala et al., 2014; McMullen et al., 2015). The aim of this study was to investigate the changes in objectively measured MVPA and ST for Grades 1–9 students over two academic years in a quasi-experimental design in schools involved in the Finnish Schools on the Move program’s pilot phase compared with the reference schools. Students’ MVPA and ST were analyzed for the school day, leisure-time, and the whole day. Materials and methods Study design and participants The data were collected during the national, ongoing Finnish Schools on the Move program. This action program aims to establish a physically active culture in Finnish comprehensive schools (Tammelin et al., 2012), and it is part of the Finnish Government Programme (Prime Minister’s Office, 2015). The program is funded by the Ministry of Education and Culture and organized by the Board of Education, regional state administrative agencies, and other organizations. In spring 2010, the Ministry of Education and Culture launched an application process for school-based physical activity projects to be included and funded in the program. A total of 58 applications from municipalities were received, and the office holders of the ministry granted 21 projects with funding from autumn 2010 onward. During the pilot phase (2010–2012), these 21 regional local projects included 45 schools for approximately 10 000 Grades 1–9 students throughout Finland (Tammelin et al., 2012). Using a quasi-experimental design, data were collected between 2010 and 2012 from four program schools and two reference schools. Four program schools were selected based on the official funding applications, which included the actions plans to promote physical activity during the school day. Selections were made in a research group meeting, and the selection criteria emphasized variety in geographical location and school levels (both primary and lower secondary schools) and viability, versatility, and comprehensiveness of the project action plans to promote school-based PA. The program schools included two primary schools (A and B) and two lower secondary schools (D and E). One primary school (C) and one lower secondary school (F) from similar areas and school levels as the program schools were invited to act as reference schools, i.e., schools not involved in the program. The study population consisted of 319 girls and boys aged 7– 15. An invitation to participate was sent to 303 primary school (Grades 1–6) and 334 lower secondary school (Grades 7–9) students. From the invited sample, 188 (61%) primary school students and 131 (39%) lower secondary school students agreed to participate and were included in the final study sample. The participation rates of the schools varied between 26% and 79% in the first measurement in autumn 2010 (Table 1). The grade levels that participated in the study from each school in the first measurement are presented in Table 1. Program schools and reference schools The bottom-up approach of the program enabled the participating schools to plan and implement their own individual plans to make the school day more physically active. The program schools apart from School A had not started their promotion actions at the time of the first measurement. Primary School A from an urban area had already sought for a more physically active school day for several years before the start of the program. They continued to offer promotion actions between 2010 and 2012, for example, two longer recess periods (25 and 40 min) during the school day for both organized and unorganized physical activities as well as provision of adequate equipment for PA and development of school sports facilities, especially those outdoors. Physical activities at recess were developed and organized by a hired project worker with the help of students acting as recess activators (peer instructors). Primary School B from an urban area had a longer recess period (30 min) for PA in the first academic year of the program and changed this providing to two longer recess periods (25 and 40 min) during the second academic year of the program. They also built an outdoor equipment area for recess and after-school use, trained older students to be recess activators for younger students, and educated their staff about children’s PA. Lower secondary School D from an urban area focused on developing its school cycling culture by buying bicycles and 1443 Physical activity & sedentary time helmets for school lessons taught outside the school building, motivating the students to commute to school in a physically active way, encouraging students and staff to engage in muscular training by developing the school gym, and educating both students and staff. Lower secondary School E from a rural area included regular physically active morning assemblies and walks during the school day, whole-school events involving sports and PA, lunch break physical activity led by students, and training of its students to be recess activators. Reference School C from an urban area and reference School F from a semi-urban area were not involved in the program. Therefore, only PA measurements were conducted at these schools. Measurements Accelerometer-assessed physical activity and sedentary time MVPA and ST were measured using the ActiGraph GT1M and GT3X accelerometers. The vertical axis output of GT1M and GT3X monitors has been confirmed to be similar (Sasaki et al., 2011). ActiGraph sensors have been widely studied and shown to have adequate reproducibility, validity, and feasibility for both children and adolescents (de Vries et al., 2006). During the regular school day, trained researchers and staff distributed the accelerometers face-toface to the children at the schools, and the children were instructed to wear the accelerometer on the right hip with an elastic band during their waking hours for 7 days, except during water activities. Students were shown how to keep a diary to monitor their school hours, sleeping hours, and the types of activities the accelerometer did not measure (such as cycling, strength training, and swimming). In free-living physical activities, children tend to move in short bursts (Bailey et al., 1995), and thus, a 10-s epoch was chosen for the monitoring. Measurements were conducted four times at each school between 2010 and 2012, and each school had their measurement day the same month in the autumn and spring (Table 1). Data reduction Actilife software (Release 5.0 or later) was used to initialize the accelerometers and download the data. A customized Visual basic macro for Excel software was used for data reduction. Non-wearing time was calculated as periods of more than 30 min of consecutive zero counts. The accelerometer data for school hours (school day) and outside school hours (leisuretime) were filtered from the whole-day data based on the school hours reported in the student diaries. Wearing time during school hours was required to be 80% of full school day. In addition, 500 min of total wearing time was required for a valid day. All school day, leisure–time, and whole-day PA measures were calculated as daily averages for the students that were meeting both conditions for at least two weekdays. PA during the school day, leisure–time, and the whole day was expressed as average counts per minute (cpm). MVPA and ST during the school day were expressed as minutes per hour (min/h) so as to compare the results of school days of varying lengths. MVPA and ST during leisure-time and the whole day were expressed as minutes per day (min/day) and minutes per hour (min/h, adjusted for wearing time), respectively. Cutoff points based on Evenson et al. (2008) were used to calculate ST (<100 cpm) and MVPA (>2295 cpm). A 20 000 cpm upper limit was set to avoid any spurious data (Heil et al., 2012). Anthropometrics Body weight, body height, and waist circumference were measured using standard procedures during the distribution of accelerometers. Body mass index (BMI) was calculated by dividing the weight in kilograms by the square of height in meters and then used as a measure for body adiposity. Data analysis Descriptive statistics for the first measurement in autumn 2010 are presented as mean and standard deviations (Mean SD) for continuous variables and as percentages for categorical variables in each program school and reference school. Student’s t-test and chi-square test were used to investigate the differences in study variables between the program and reference schools at the first measurement point as well as to compare those students with complete data on MVPA (and ST) from all four measurement points with those students with incomplete data. Linear growth curve modeling was used to examine the effect of the program on the development of MVPA (and ST) Table 1. Schools and students participating the follow-up study and measurement points for each school Primary schools (Grades 1–6) Lower secondary schools (Grades 7–9) School A School B School C* School D School E School F* Grades included in the follow-up 1–2, 4–51–32,4–57–87–87–8 Invited to participate † ,n115 59 129 128 81 125 Participants † ,n(%) 90 (79%) 35 (59%) 63 (49%) 58 (45%) 40 (49%) 33 (26%) Measurement points 1st measurement 10/2010 11/2010 11/2010 10/2010 11/2010 2/2011 2nd measurement 3/2011 5/2011 4/2011 3/2011 5/2011 5/2011 3rd measurement 10/2011 11/2011 12/2011 10/2011 11/2011 2/2012 4th measurement 3/2012 5/2012 4/2012 3/2012 5/2012 5/2012 Finnish children are obliged to start school at the age of 7. Therefore, students in Grade 1 are approximately 6–7 years old; in Grade 2, 7–8 years old; in Grade 3, 8–9 years old, in Grade 4, 9–10 years old, in Grade 5, 10–11 years old, in Grade 6, 11–12 years old, in Grade 7, 12–13 years old, in Grade 8, 13–14 years old, and in Grade 9, 14–15 years old. *Reference school. † From the first measurement. 1444 Haapala et al. and the level of MVPA (and ST) during the follow-up period (including four measurement points). The path diagram for the hypothesized model is presented in Fig. 1. A multi-level modeling approach was chosen to fit the growth model. The data were converted to long format and the “student” (ID) was considered as a cluster variable (repeated measures nested within the students). The regression coefficient between time and MVPA time (and ST) was specified as a random slope, meaning the coefficient was allowed to vary among students. The level of MVPA (and ST) was also treated as a random intercept. At between-subject level, the effect of the program on the development (slope) and on the level of MVPA (and ST) was estimated and tested for significance. The model at the between-subject level was controlled for potential confounding variables (background variables that differed between the program and the reference schools at the first measurement). In addition, the model was controlled for background variables that related to missingness. All descriptive statistics were calculated using IBM SPSS Statistics (Version 20.0), and linear growth curve modeling was conducted using the Mplus statistical package (Muth en & Muth en, 2012). The parameters of the models were estimated using the full information maximum likelihood (FIML) estimation method. Missing data were assumed to be missing at random (MAR). MAR means that missingness can be a function of observed covariates and observed outcomes. Unlike the listwise deletion method, FIML uses all available information in a dataset and produces unbiased parameter estimates under MAR assumption. As there were minor violations of normality assumption, maximum likelihood with robust standard errors (MLR) was used. The significance level of the study was set at 0.05. The effect size (Cohen’s d) was calculated as the difference in the mean changes (from the first measurement to the last measurement) between the program schools and the reference schools divided by a pooled standard deviation at the first measurement. The effect size of d=0.2 was considered small; d=0.5, medium; and d=0.8, large (Cohen, 1992). Ethics The study setting for the measurements was approved by the Ethics Committee of the University of Jyv€ askyl€ a. All measurements were carried out in accordance with the Declaration of Helsinki and Finnish legislation. Written informed parental consent was obtained from all participants, and only those students who provided a completed consent form (signed by both the student and the guardian) on their first measurement day were included as participants in the study. Results Missing data Of the total number of participants (n=319), 185 students (58%) provided school day PA data at all four measurement points, 75 students (23.5%) at three measurement points, 37 (11.6%) at two measurement points, and 19 (6%) at one time point. Further, three students (0.9%) did not have valid accelerometer-assessed school day PA data at all. The students with all four measurements of school day PA (n=185) were compared with the students with incomplete data (n=134). Girls had complete data more often than boys (v 2 (1) =9.92, P=0.003). The students with complete data were older and had a slightly higher BMI than those students with one or more missing value [t(317) =2.00, P=0.046 and t(314) =1.91, P=0.057, respectively]. No differences in PA measures in the first measurements were observed, except for school day and whole-day ST [t(284.27) =2.21, P=0.028 and t(301) = 2.89, P=0.004, respectively]. Students with complete data had more school day ST compared with students with incomplete data. Characteristics At the first measurement, primary school students in the reference schools were slightly older, had less total school day and whole-day total PA, and had Fig. 1. The path diagram of a linear growth model carried as a multi-level model. BMI, body mass index. 1445 Physical activity & sedentary time more school day, leisure-time, and whole-day ST than the primary school students in the program schools. No differences in the first measurement were observed in the measures between the program and reference schools at the lower secondary school level (Table 2). Furthermore, primary school students were more physically active and less sedentary during the school day, leisure-time, and the whole day, and they had lower BMI than lower secondary school students at the first measurement (P<0.001). At the primary school level, students at School A had higher levels of total PA and MVPA and lower levels of ST, both during the school day (P<0.001) and the whole day (P=0.001) compared with students at the reference school for the first measurement in 2010. Students at School B had less ST both during the school day (P=0.009) and the whole day (P<0.001) compared with the reference school students. At the lower secondary school level, students at School D had higher school day MVPA levels (P=0.043), while students at School E had less total PA (P=0.035) and MVPA (P=0.015) during the school day than those students at the reference school. Levels and changes in school day, leisure-time, wholeday MVPA, and sedentary time Because there were differences between the program schools and reference schools in terms of age at the first measurement point, and missing data were dependant on sex and BMI, all the models were controlled for these variables. The path diagram for the hypothesized model is presented in Fig. 1, and the mean values for each measurement point and estimation results of the models are presented in Tables 3 and 4, respectively. The regression coefficient (b) between the program (1 =program, 0 =reference school) and level refers to the difference in the overall level of the outcome variable between the Table 2. Descriptive statistics at program schools and reference schools at the first measurement (n=319) School level All schools Program schools Reference schools P-value nMean SD nMean SD nMean SD Primary schools 188 125 63 Sex (boys) 188 50.0% 125 47.6% 63 52.4% 0.643 Age (years) 188 9.3 1.5 125 9.1 1.5 63 9.9 1.2 < 0.001 BMI (kg/m 2 ) 187 17.2 2.6 124 17.1 2.5 63 17.3 2.7 0.622 School day Total PA (cpm) 178 615.2 187.9 119 659.8 194.5 59 525.3 136.0 < 0.001 MVPA (min/h) 178 5.6 2.3 119 6.1 2.4 59 4.5 1.7 < 0.001 ST (min/h) 178 35.8 4.4 119 34.7 4.6 59 37.9 3.2 < 0.001 Leisure-time Total PA (cpm) 178 582.5 161.2 119 588.0 157.1 59 571.5 170.1 0.523 MVPA (min/day) 178 44.3 17.2 119 43.1 15.9 59 46.7 19.3 0.222 ST (min/h) 178 34.9 4.5 119 34.4 4.4 59 35.9 4.6 0.029 Whole-day Total PA (cpm) 178 595.2 144.2 119 614.9 139.7 59 555.6 146.2 0.009 MVPA (min/day) 178 70.2 22.6 119 71.6 21.6 59 67.3 24.3 0.227 ST (min/h) 178 35.6 4.3 119 34.8 3.7 59 37.2 5.0 < 0.001 Lower secondary school 131 98 33 Sex (boys) 131 44.3% 98 48.0% 33 33.0% 0.143 Age (years) 131 14.0 0.6 98 13.9 0.6 33 14.1 0.6 0.164 BMI (kg/m 2 ) 129 20.5 3.4 96 20.4 3.0 33 20.7 4.3 0.730 School day Total PA (cpm) 125 314.2 90.9 93 311.8 92.4 32 321.1 87.4 0.622 MVPA (min/h) 125 2.8 1.1 93 2.8 1.1 32 2.8 0.9 0.945 ST (min/h) 125 45.5 3.2 93 45.7 3.1 32 44.8 3.5 0.187 Leisure-time Total PA (cpm) 125 418.4 143.4 93 414.7 151.8 32 429.2 116.8 0.625 MVPA (min/day) 125 32.8 15.0 93 31.6 14.6 32 36.3 15.8 0.124 ST (min/h) 125 42.4 3.5 93 42.5 3.8 32 42.3 2.6 0.783 Whole-day Total PA (cpm) 125 376.8 100.2 93 372.9 102.3 32 388.3 94.7 0.456 MVPA (min/day) 125 48.5 17.0 93 47.4 16.4 32 51.7 18.3 0.223 ST (min/h) 125 43.7 2.8 93 43.7 2.9 32 43.5 2.4 0.632 SD, standard deviation; BMI, body mass index; PA, physical activity; Cpm, counts per minute; ST, sedentary time; MVPA, moderate-to-vigorousintensity physical activity. P-value for difference between program and reference schools from Student’s t-test or chi-square test. Statistically significant values presented in bold (P<0.05). 1446 Haapala et al. program and the reference school; the regression coefficient between the program and slope refers to the difference in the change that occurred during the follow-up period for the program schools and the reference schools. In the primary schools, the level of school day MVPA was 1.1 min/h higher (P<0.001), while leisure-time MVPA was 5.8 min/day lower (P<0.038) in program schools compared with reference schools (Table 4, level estimates). The level of school day ST Table 3. Mean values (standard deviations) for school day, leisure-time, and whole-day moderate-to-vigorous-intensity physical activity (MVPA) and sedentary time (ST) at each measurement point Measurement points* 1234 Primary schools Program schools School day MVPA (min/h) 6.1 2.4 5.5 1.8 6.2 2.0 6.3 2.0 School day ST (min/h) 34.7 4.6 36.9 3.3 35.9 3.4 36.6 4.1 Leisure-time MVPA (min/day) 43.1 15.9 48.4 20.3 36.5 15.8 45.5 19.9 Leisure-time ST (min/h) 34.5 4.5 34.7 4.9 36.7 5.8 36.3 5.3 Whole-day MVPA (min/day) 71.8 21.4 73.9 25.1 67.5 20.1 76.6 22.4 Whole-day ST (min/h) 34.8 3.7 35.4 3.8 36.8 4.2 36.6 4.1 Reference school School day MVPA (min/h) 4.5 1.7 5.1 2.4 4.3 1.5 4.3 1.8 School day ST (min/h) 37.9 3.2 38.8 4.6 39.9 3.6 40.9 3.3 Leisure-time MVPA (min/day) 47.3 19.0 47.8 21.2 44.2 17.8 45.4 18.2 Leisure-time ST (min/h) 35.8 4.5 35.7 4.6 37.4 4.3 37.9 5.2 Whole-day MVPA (min/day) 68.1 24.0 72.1 27.4 64.3 21.9 66.4 23.5 Whole-day ST (min/h) 37.3 4.9 40.3 9.3 40.9 8.1 40.3 5.0 Lower secondary schools Program schools School day MVPA (min/h) 2.8 1.1 3.6 1.7 3.2 1.7 3.2 1.3 School day ST (min/h) 45.8 3.1 45.9 3.9 46.8 3.0 47.7 2.7 Leisure-time MVPA (min/day) 31.8 14.5 31.8 18.2 26.7 12.8 29.5 14.9 Leisure-time ST (min/h) 42.3 4.4 41.8 5.2 43.4 4.1 42.6 5.3 Whole-day MVPA (min/day) 47.6 16.4 53.2 22.8 45.5 16.3 48.0 18.1 Whole-day ST (min/h) 43.7 2.9 43.9 3.2 45.3 2.4 45.4 2.7 Reference school School day MVPA (min/h) 2.8 0.8 3.3 1.0 2.7 1.4 3.1 1.4 School day ST (min/h) 44.8 3.4 45.6 2.6 46.3 3.7 48.3 2.9 Leisure-time MVPA (min/day) 36.6 15.6 33.9 11.1 36.9 17.0 37.8 20.1 Leisure-time ST (min/h) 42.2 2.6 42.2 3.6 42.4 5.4 42.8 4.1 Whole-day MVPA (min/day) 51.7 17.9 52.8 14.5 51.9 19.4 53.2 23.5 Whole-day ST (min/h) 43.5 2.4 43.6 3.0 44.6 2.9 44.0 3.6 *First measurement 10–11/2010 and 2/2011, second measurement 3–5/2011, third measurement 10–12/2011 and 2/2012, and fourth measurement 3–5/2012. Table 4. The regression coefficients for the effect of the program on the level and slope of school day, leisure–time, and whole-day moderate-tovigorous-intensity physical activity (MVPA) and sedentary time (ST) Level Slope bSE PbSE P Primary schools School day MVPA (min/h) 1.14 0.28 < 0.001 0.27 0.10 0.010 School day ST (min/h) 2.07 0.54 < 0.001 0.52 0.20 0.008 Leisure-time MVPA (min/day) 5.77 2.78 0.038 0.13 0.99 0.894 Leisure-time ST (min/h) 0.10 0.57 0.869 0.04 0.26 0.879 Whole-day MVPA (min/day) 1.40 3.57 0.695 1.33 1.16 0.250 Whole-day ST (min/h) 2.50 0.69 < 0.001 0.19 0.35 0.584 Lower secondary schools School day MVPA (min/h) 0.13 0.19 0.481 0.06 0.11 0.561 School day ST (min/h) 1.52 0.55 0.006 0.48 0.26 0.063 Leisure-time MVPA (min/day) 3.89 2.62 0.137 1.31 1.48 0.376 Leisure-time ST (min/h) 0.39 0.60 0.518 0.10 0.27 0.709 Whole-day MVPA (min/day) 2.78 3.18 0.382 0.65 1.68 0.698 Whole-day ST (min/h) 0.53 0.49 0.285 0.33 0.24 0.162 1=program school, 0 =reference school. b=unstandardized regression coefficient. SE =standard error. The models controlled for sex, age, and body mass index. Statistically significant values presented in bold (P<0.05). 1447 Physical activity & sedentary time was 2.1 min/h lower (P<0.001) and whole-day ST was 2.5 min/h lower (P<0.001) in program schools compared with reference schools at the primary school level. In the lower secondary schools, there was no difference in the levels of MVPA or ST between the program schools and reference schools, with the exception of school day ST being 1.5 min/h higher (P=0.006) in the program schools compared with reference schools (Table 4). In the program primary schools, school day MVPA increased (P=0.010) and school day ST decreased (P=0.008) more than in the reference schools during the follow-up period (Table 4, slope estimates). The effect sizes for these differences were small (d=0.18 and d=0.27, respectively). In the lower secondary schools, there were no significant differences between program and reference schools in the change in different PA and ST measures during the follow-up (Table 4). When the results were analyzed separately for all four program schools, the level of school day MVPA was higher (P<0.001) and school day ST was lower (P<0.001) only in primary School A compared with reference School C (Table 5). However, no significant differences were observed in the change (slope) of school day MVPA (P=0.054) or ST (P=0.526) during the follow-up period for School A and reference School C. School day MVPA increased (P=0.016) and school day ST (P<0.001) and leisure-time MVPA (P<0.001) decreased more in primary School B than in reference School C during the follow-up period. School day ST decreased (P=0.039) more in the lower secondary School D compared with the reference School F during the follow-up period. Leisure-time MVPA and whole-day MVPA decreased more (P=0.001 and P<0.001) in the lower secondary School E compared with the reference School F. Discussion The purpose of this study was to investigate the changes in MVPA and ST over two academic years in two primary and two lower secondary schools Table 5. The regression coefficients for the effect of the program on the level and slope of school day, leisure–time, and whole-day moderate-tovigorous-intensity physical activity (MVPA) and sedentary time (ST) Schools Level Slope bSE PbSE P Primary schools School A School day MVPA (min/h) 1.70 0.29 < 0.001 0.17 0.11 0.135 School day ST (min/h) 2.84 0.58 < 0.001 0.15 0.21 0.486 Leisure-time MVPA (min/day) 4.54 2.87 0.113 0.62 1.05 0.553 Leisure-time ST (min/h) 0.10 0.27 0.717 0.11 0.60 0.717 Whole-day MVPA (min/day) 5.53 3.66 0.131 1.45 1.21 0.231 Whole-day ST (min/h) 2.50 0.74 0.001 0.22 0.52 0.674 School B School day MVPA (min/h) 0.38 0.40 0.350 0.38 0.16 0.016 School day ST (min/h) 0.45 0.72 0.529 1.38 0.27 < 0.001 Leisure-time MVPA (min/day) 6.80 1.46 < 0.001 1.92 0.02 < 0.001 Leisure-time ST (min/h) 0.59 0.85 0.488 0.58 0.32 0.073 Whole-day MVPA (min/day) 8.11 4.86 0.095 0.16 1.60 0.921 Whole-day ST (min/h) 2.45 1.29 0.058 0.14 1.35 0.916 Lower secondary schools School D School day MVPA (min/h) 0.57 0.23 0.012 0.13 0.13 0.327 School day ST (min/h) 1.28 0.61 0.035 0.56 0.27 0.036 Leisure-time MVPA (min/day) 2.42 2.67 0.364 1.26 1.58 0.426 Leisure-time ST (min/h) 0.31 0.64 0.628 0.19 0.27 0.499 Whole-day MVPA (min/day) 0.39 3.27 0.904 0.05 1.81 0.978 Whole-day ST (min/h) 0.31 0.50 0.530 0.19 0.18 0.294 School E School day MVPA (min/h) 0.56 0.15 < 0.001 0.05 0.11 0.672 School day ST (min/h) 1.81 0.57 0.001 0.30 0.25 0.227 Leisure-time MVPA (min/day) 5.55 1.25 < 0.001 1.54 0.47 0.001 Leisure-time ST (min/h) 0.40 0.77 0.600 0.30 0.27 0.262 Whole-day MVPA (min/day) 7.21 6.80 0.289 1.82 0.19 < 0.001 Whole-day ST (min/h) 0.95 0.61 0.118 0.19 0.18 0.313 1=program school, 0 =reference school. Results analyzed separately for four program schools compared with reference school (primary Schools A and B vs reference School C, and secondary Schools E and F vs reference School G). b=unstandardized regression coefficient. SE, standard error. The models controlled for sex, age, and body mass index. Statistically significant values presented in bold (P<0.05). 1448 Haapala et al. involved in the Finnish Schools on the Move program’s pilot phase and to compare those results with those of reference schools not involved in the program. We found that school day MVPA increased and school day ST decreased more in the program primary schools (Grades 1–6) than in the reference schools; however, no differences in changes were observed in whole-day MVPA or ST between the program and reference schools during the follow-up. In lower secondary schools, no differences between the program and reference schools were observed in changes of PA and ST. Previous studies have provided evidence that school-based PA interventions have relatively small effects on the PA levels in children and adolescents (Dobbins et al., 2013). Similarly, we found positive, but small, increases in school day MVPA in the program primary schools compared with the reference schools. We also found that ST increased less in the program primary schools than in the reference schools. However, these changes did not translate into positive effects across the day in our study. Thus, our results suggest that students compensated for the increased school day MVPA and decreased ST by decreasing MVPA in their leisure-time. This finding is supported by studies that indicate that children tend to compensate the increased PA levels by diminishing PA afterward. In a Danish study, PA during school hours was increased by additional physical education lessons, but at the same time, children decreased their PA levels after school hours (Moller et al., 2014). Ridgers et al. (2014, 2015) reported similarly that increases in children’s PA levels were compensated by decreased PA levels within the same day or the following day. One possible solution to avoid this compensation effect could be a more intense focus on a “whole-school approach” (Institute of Medicine, 2013). Instead of only intervening the school time, possibilities before, during, and after the school day should be embraced and examined from both an environmental and social perspective using the school as the center of distribution. Effective examples of these approaches are the “KISS” study with increased physical education classes and PA homework (Kriemler et al., 2010) and the “Active Living” study which tackles PA and ST issues in the school, transportation to school (neighborhood), and out-of-school activity programs during leisure-time (van Kann et al., 2016). The Finnish Schools on the Move program—unlike most interventions—relies on a bottom-up approach in which schools and municipalities have the autonomy to plan and implement actions suitable for their particular situations (Haapala et al., 2014). The concept of the program links to the overall Finnish Education Policy by relying on customization, creativity, encouragement of risk-taking and shared responsibility, and trust (Sahlberg, 2011). As suggested by the social–ecological model (Sallis et al., 2006), the program schools in this study applied these possibilities for physical activity promotion by creating both educational and environmental/policy dimensions. In fact, such multi-component approaches have shown encouraging results in improving PA levels among children and adolescents (van Sluijs et al., 2007; Kriemler et al., 2011). In addition, models based on choice and individualization have presented positive results in both elementary (Naylor et al., 2006) and middle school levels (Hoelscher et al., 2016) for student PA levels and program flexibility. The approach considers each school’s unique features, such as readiness or knowledge of staff, current facilities and equipment, and existing co-operation with other networks. This aspect also embraces participant ownership in the process at the individual, school, and community levels. The challenge with this type of setting is possible selection bias, wherein the study population may be drawn from schools that are more inclined to encourage activity and are motivated to enroll in the program. Program guidelines given to the schools were general and mainly concerned reporting to the funding agent and participation in the evaluation by the research center. Evaluation of the program fidelity in the study schools was based on qualitative data gathered from the overall evaluation of the national program (Tammelin et al., 2012). Local contact persons in each school project participated in interviews (1/2011 and 5/2012) and answered surveys (5/2011 and 1/2012) concerning implemented strategies, benefits, facilitators, and barriers, from the perspectives of students, staff, school community/ culture, and networks (Tammelin et al., 2012). In conjunction with the notes on discussions and observations from school visits, the program schools in this study initially implemented the majority of their planned actions. Overall, program fidelity in the study schools appeared to be high. Similarities were observed between the program school actions despite the autonomy of planning, and some of these strategies have shown effectiveness for PA and ST in other studies: modification of school facilities and equipment (van Sluijs et al., 2007; Kriemler et al., 2011), structural changes in the school day (Kriemler et al., 2011), staff and student schooling on PA, and student participation in planning and implementation (Haapala et al., 2014). However, only one primary program school was able to achieve favorable changes in school day MVPA and ST ultimately compensated outside school hours. Reasons for the ineffectiveness of these strategies may lay in, e.g., insufficient intensity or 1449 Physical activity & sedentary time