A quasi-experimental examination of how school-based physical activity changes impact secondary school student moderate- to vigorous- intensity physical activity over time in the COMPASS study

Background Adolescence is characterized by low moderate- to vigorous- intensity physical activity (MVPA) levels. Targeting the school setting can increase MVPA among a large proportion of adolescents. However, school-based physical activity interventions for adolescents remain largely ineffective. Therefore, the purpose of this study was to examine how naturally-occurring changes to school physical activity policy, recreational programming, public health resources, and the physical environment, impact adolescent MVPA over a 1-year period. Methods Quasi-experimental longitudinal data from 18,777 grade 9–12 students (mean age = 15.1 ± 0.02 years), and 86 principals from 86 schools, participating in year 2 (2013–2014) and year 3 (2014–2015) of the COMPASS study (Ontario and Alberta, Canada) was used. Total MVPA over the previous week was self-reported at both time points using the COMPASS Student Questionnaire and average daily MVPA was calculated. Changes to physical activity policies, recreational programming, public health resources, and the physical environment were self-reported by school principals. Changes to the number and condition of physical activity facilities were objectively measured during school audits using the COMPASS School Environment Application. Multi-level modeling was used to examine change in student MVPA between schools that made changes and schools that did not. Models were adjusted for several student and school level confounders. Results Over the 1-year period, 61 of 86 schools made physical activity related changes. Of these, 9 significantly changed student MVPA. However, only 4 of 9 schools’ changes increased student MVPA, including opening the fitness centre at lunch (β = 17.2, 95 % CI: 2.6–31.7), starting an outdoor club (β = 17.8, 95 % CI:7.4–28.1), adding a bike rack (β–14.9, 95 % CI:0.7–29.1), and adding weightlifting and run/walk clubs, archery, figure skating, increased access to the sports field, and improved condition of the outdoor basketball court (β = 15.5, 95 % CI: 5.2–25.7). Conclusions Changes such as adding or increasing access to facilities, and adding multiple recreational programs, seemed to be effective for increasing student MVPA over the 1-year period. However, given the specificity of results, a one-size fits all approach may not be effective for increasing MVPA. Instead, school principals need to consider the resources within and surrounding their school, and the interests of the students.


Background
Regular moderate-to vigorous-intensity physical activity (MVPA) is associated with several physical, mental health, and cognitive benefits in school-aged children and youth [1][2][3]. According to recent global estimates, approximately 80 % of adolescents failed to meet the recommended amount of MVPA for optimal health benefits [4]. Further, evidence has suggested that there is an actual decline in the amount of time spent in MVPA during adolescence [5,6]. Given the adverse health consequences associated with physical inactivity [4,7], identifying effective strategies to increase MVPA and promote a healthy active lifestyle among adolescents is warranted.
Ample research has identified individual correlates that can be targeted to increase MVPA among adolescents [8]. However, some models suggest that factors outside of the individual also have potential to influence MVPA. For instance, ecological models often recognize that the development of health-enhancing behaviours, such as MVPA, involves interactions between the individual (e.g., self-efficacy, enjoyment, attitudes) [9] and the multiple contexts (e.g., home, school, community) in which they are situated [10,11]. Since most adolescents spend approximately 25 h each week in school throughout the school year [12,13], the school environment represents one important context for shaping MVPA [3,[14][15][16].
Examination of the school environment's influence on student MVPA has been a growing body of research within the last five years [17]. According to a recent systematic review examining aspects of the whole-school environment, it was found that activity setting, perceived teacher support, and intramurals were consistently positively associated with student MVPA [17]. However, the majority of included studies were cross-sectional in nature, and the evidence from the limited longitudinal and experimental studies was mixed [17]. Similar mixed results were found in a separate review, stemming from heterogeneity in terms of frequency, duration, and intensity of school-based PA interventions [18]. While the authors of the review identified several PA intervention successes for increasing MVPA in younger children, it was concluded that PA interventions targeting adolescents were ineffective [18].
Given the combined evidence, one way to strengthen the current literature would be to examine longitudinal associations using quasi-experimental designs [19,20]. Examination of naturally-occurring changes that take place in real-world settings over time could reduce some of the potential external validity issues that have been associated with controlled trials [21], and address issues of causality inherit with cross-sectional studies [22][23][24]. Furthermore, quasi-experimental designs allow for the simultaneous observation of multiple interventions that occur in diverse settings. Such observations can be useful for identifying the most effective interventions for increasing adolescent MVPA. Therefore, the purpose of this study is to examine how naturally-occurring changes to secondary school physical activity (PA) policies, recreational programming, public health resources, and the physical environment, impact secondary student MVPA over a one-year period.

Design
The cohort for obesity, marijuana use, physical activity, alcohol use, smoking, and sedentary behaviour (COM-PASS) is an ongoing quasi-experimental study that collects annual data regarding multiple health behaviours from secondary school students in grades 9 to 12 (aged 13 to 18 years), and the schools they attend in Ontario and Alberta, Canada [25]. COMPASS follows a cyclical process where student and school data is collected and used to generate school health profiles. The school health profiles are given back to the schools with feedback on their students' health status, as well as information and resources that can be used to target identified problem areas. Schools then have the option to make a change themselves or contact a COMPASS knowledge broker for assistance with targeting identified problem areas. For all changes that occur, the COMPASS research team evaluates their impact on student health outcomes to generate practice-based evidence. The current study used longitudinal student-and school-level data from Ontario and Alberta schools in year two (Y 2 : 2013-2014) and year three (Y 3 : 2014-2015) of the COMPASS study. Data was collected using the COMPASS Student Questionnaire (Cq) [26], the COMPASS School Policies and Practices Questionnaire (SPP) (based off of the Healthy School Planner tool; [27]), and the COMPASS School Environment Application (Co-SEA) [28]. A full description of the study methods is available in print [25] or online [29]. Ethical approval was obtained from the University of Waterloo Office of Research Ethics and University of Alberta Research Ethics Board. All school boards and schools approved study procedures. Active-information passive-consent was sought from parents, and assent was obtained from participants on the data collection date. Parents or students could decline to participate at any time.

Procedures
Each year students completed the Cq during class time, school principals filled out the SPP, and a COMPASS staff member performed an audit of the physical environment using the Co-SEA. The Cq collects individual student data on health behaviours, including MVPA, and demographic characteristics [25]. The SPP is a shorter, modified version of the previously validated Healthy School Planner tool [27] and captures information on school policies, programs, resources, and the environment related to student health [25]. The Co-SEA is a software application that can be downloaded to most mobile devices (e.g., cellular phone) and allows for pictures to be taken, stored, and assigned rankings. The Co-SEA was used by a COMPASS staff member to take pictures of PA facilities present within the school [28].

Participants
In Y 2 , data from 79 Ontario and 10 Alberta schools was collected. A total of 57,229 students were enrolled in the 89 secondary schools with 79.15 % (n = 45,298) of eligible students completing the Cq. In Y 3 , data was collected from 78 Ontario and 9 Alberta schools. A total of 53,846 students were enrolled in the 87 secondary schools with 78.66 % (n = 42,355) of eligible students having completed the Cq. Missing respondents due to parental refusal accounted for 1.2 %, and 0.78 %, of the eligible sample in Y 2 , and Y 3 , respectively. The remainder of missing respondents were due to absenteeism or students being on a spare (i.e., unscheduled class) during the data collection, or student refusal. Furthermore, three schools dropped out from Y 2 to Y 3 due to administration changes and questionnaire length. Though one school was added in Y 3 resulting in a sample of 87 secondary schools, this school was not included in the present study because it did not have any Y 2 data. Therefore, 86 schools with complete data were included in this study.
To explore longitudinal changes among respondents, we paired Y 2 and Y 3 student-level data within schools, creating a longitudinal sample of 19,854 students from 86 schools. The paired sample accounted for 35.2 % of eligible Y 2 respondents (n = 56,356), and 37.1 % of eligible Y 3 respondents (n = 53,426). As expected, the 10,233 grade 12 students in Y 2 that graduated were not in school in Y 3 , and the 11,070 grade 9 students that were newly admitted to participating schools in Y 3 were not paired and were excluded. Other reasons for non-paired data included students who transferred schools, students who were had spare classes or were absent during the time of Y 2 or Y 3 data collection in their school, early school leavers, or inaccurate data provided in the data pairing measures on the Cq. Methodological details on the COMPASS data pairing procedures are available [30].

Subjective school-level changes
Changes to PA policies, recreational programming, use of public health units (i.e., a government health agency that carries out community health programs), and environment/equipment were assessed via single items on the SPP. Principals were provided with their previous year's responses and were asked to report if any changes had occurred since the previous school year. If changes were made, they were then prompted to provide additional details about the change.

Objective school-level changes
Changes to quantity and condition of PA facilities were measured using the Co-SEA. Quantity of school PA facilities were recorded in both years by COMPASS research staff performing a school audit. Quantity changes were determined by subtracting the number of facilities present in Y 2 from the number of facilities present in Y 3 . Condition of the PA facilities were measured each year on a 3-point scale (1 = 'poor' , 2 = 'adequate' , 3 = 'good'). Condition changes were determined by subtracting conditions scores in Y 2 from condition scores in Y 3 .

Outcome
Change in student self-reported MVPA MVPA was measured by two questions on the Cq. Students were required to complete the following item "Mark how many minutes of vigorous physical activity you did on each of the last 7 days. This includes physical activity during physical education class, lunch, after school, evenings, and spare time." The same item was used to measure moderate physical activity, but instead of "vigorous," the term "moderate" was substituted into the sentence. Responses were recorded in hours (0-4) and minutes (0, 15,30,45) for each day of the week. To help students better understand the questions they were given examples of vigorous (i.e., jogging, team sports, fast dancing, jump-rope, and any other physical activities that make you breathe hard and sweat) and moderate (i.e., lower intensity activities such as walking, biking to school, and recreational swimming) physical activities. Responses to these questions were added and averaged over the seven days to calculate daily MVPA. These items have demonstrated moderate test-retest reliability (ICC = 0.75); and slight criterion validity for MVPA (ICC = 0.25) against accelerometers [31], which is comparable to other self-reported measures used with adolescents [32][33][34][35][36][37].

Student-level covariates
Age, sex, ethnicity, weekly spending money, and physical education enrollment were considered covariates based on previous research examining their influence on MVPA [8,[38][39][40]. In addition, typical MVPA was also considered as a covariate, given its potential impact on MVPA. These variables were measured via single items on the Cq. There were seven response options for ethnicity. Based on frequency distributions, ethnicity was collapsed into two groups ("White," and "non-White"). There were eight response options for weekly spending money ranging from "zero" to "$100+", and "I do not know how much money I get each week." To maximize sample size, participants who reported "I do not know" or who had missing data for this item were collapsed into one group [41]. There were three response options for physical education enrollment: (1) "Yes, I am taking one this term;" (2) "Yes, I will be taking one or have taken one this school year, but not this term;" and (3) "No, I am not taking a physical education class at school this year." Students who reported different statuses from Y 2 to Y 3 formed one group, and students who reported the same status for each year formed the referent group. There were three response options for typical MVPA: (1) "Yes," (2) "No, I was more active in the last 7 days;" and (3) "No, I was less active in the last 7 days." Students who reported different responses in Y 2 and Y 3 formed one group, and students whose response was the same in each year formed the referent group.

School-level covariates
School size, school area level socioeconomic status, and school location were considered covariates based on previous research examining their influences on school PA facilities and program offerings [23,24,42]. School size was determined via school enrolment records and was entered into the model as a continuous variable. School area level socioeconomic status was constructed using the median household income of census divisions that corresponded with school postal codes, and was collected from 2011 National Household Survey data. School location was determined via school postal code, and Statistics Canada classifications were used to classify schools as "rural," "small urban," "medium urban," and "large urban." Based on frequency distributions, "rural" and "small urban" were collapsed to form one group, and "medium urban" and "large urban" were collapsed to form another group [43].

Analysis
Analyses were completed using SAS version 9.4 (SAS Institute Inc., Cary, NC). Descriptive statistics were calculated for student-level and school-level variables using linear and logistic regressions that accounted for the clustering effect of schools. Likewise, the same procedures were used to compare demographic characteristics between included and excluded participants. To address the main purpose a three level, multi-level growth model was conducted using the MIXED procedure. Data was transposed from wide to long format so that time was nested in students, and students were nested in schools, with random intercepts included for students and schools. Consistent with other quasi-experimental research that looked at change over time [44], each school that made a PA related change between Y 2 and Y 3 was treated as a change group, while schools that made no PA related change between Y 2 and Y 3 were collapsed into one control group and served as the reference group. The multi-level growth model included time (Year), dummy variables for each change group compared to the reference group, and all student-level and school-level covariates. Additionally, to compare the impact of each change group compared to the reference group on change in student-level MVPA, a time*change interaction term was included in the model for each change group. Statistical significance was set a priori at p < 0.05.

Results
Out of 19,854 students with paired data, students with missing variables were excluded (n = 808), and consistent with previous research students with an extreme MVPA change value (±3 SD) were removed (n = 269) [45], resulting in a final sample of 18,777 students. The included sample comprised of significantly older participants (Mean age = 15.07 years versus Mean age =15.01 years), more white participants (73.7 % versus 66.6 %); more female participants, (53.9 % versus 42.71 %), more participants whose typical week of MVPA status remained the same (58.0 % versus 48.5 %), and more participants whose physical education enrolment status remained the same (45.5 % versus 35.6 %) compared to the excluded group. Student and school demographic characteristics are listed in Tables 1  and 2, respectively. In Y 2 mean MVPA for females and males was 107.87 min/day and 132.9 min/day, respectively. In Y 3 mean MVPA for females and males was 100.6 min/ day and 129.53 min/day, respectively. Overall, MVPA declined by 4.86 min/day.
Of the 86 schools included in this study, 61 made PA related changes to at least one feature of the school environment between Y 2 and Y 3 . Detailed descriptions of school changes are presented in Table 3. Briefly, none of the schools made any policy related changes, 15 schools made changes to recreational programming, two schools made changes to their use of public health units, two schools made changes to the subjective environment/equipment, and two schools made changes to both recreational programming and the subjective environment/equipment. Furthermore, 21 schools made changes to the physical environment within their school. Of these 21 schools, quantity changes occurred in five schools, condition changes occurred in 10 schools, and both quantity and condition changes occurred in six schools. Lastly, 19 schools reported multiple changes that encompassed combinations of changes to recreational programming, use of public health units, the subjective environment/equipment (as reported in SPP), and the physical environment (measured by Co-SEA).
As shown in Table 4, of the 61 schools that had PA related changes, a significant change in student MVPA was observed in nine schools. Of these nine schools, four schools' changes resulted in a significant increase in student MVPA, while significant decreases in student MVPA occurred in five schools. Significant increases in MVPA were observed in School 5, 10, 23, and 49.

Discussion
The purpose of this study was to examine how naturally occurring changes to PA policy, recreational programming, public health resources, and the physical environment within schools impacted student MVPA over a oneyear period. We found that changes to some aspects of recreational programming (e.g., PA-related clubs) and the physical environment (e.g., addition of bike rack, fitness room) were associated with a significant change in student MVPA. Changes to public health resources proved not to be significant. Further, no policy changes occurred in any participating schools between Y 2 and Y 3 , suggesting an opportunity for more targeted action moving forward.
To our knowledge this is the first study that looked at how naturally occurring school PA-related changes impacted student MVPA over time. There were nine schools' changes that resulted in significant student MVPA changes

School 24
Co-SEA: Added a bike rack.

School 25
Co-SEA: Added a bike rack.

School 26
Co-SEA: Added a bike rack, tennis court, and outdoor basketball court. Removed the fitness/weight room.

School 27
Co-SEA: Condition of the tennis court improved.

School 28
Co-SEA: Condition of the outdoor track improved.

School 29
Co-SEA: Condition of the gym improved. School 31 (−) Co-SEA: Condition of the fitness/weight room improved.

School 32
Co-SEA: Condition of the fitness/weight room improved, and condition of the gym worsened.

School 33
Co-SEA: Condition of the field worsened.

School 34
Co-SEA: Condition of the outdoor track worsened.

School 35
Co-SEA: Condition of the fitness/weight room worsened.

School 36
Co-SEA: Condition of the fitness/weight room worsened.

Quantity & Condition
School 37 Co-SEA: Added 2 fitness/weight rooms, and the condition of gym improved.

School 38
Co-SEA: Added baseball diamond, removed outdoor basketball court, and the condition of the fitness/ weight room improved.

School 39
Co-SEA: Added dance studio, and the condition of the indoor facility [not specified] improved.

School 40
Co-SEA: Added fitness/weight room, removed yoga room, and the condition of fitness/weight room improved.

School 41
Co-SEA: Added two fitness/weight rooms and the condition of outdoor basketball court improved.

School 42
Co-SEA: Added field, and the condition of the field, and the outdoor track worsened.

School 43
Co-SEA: Added other outdoor facility (crossfit, foobtall equipment), removed a bike rack, and the condition of the field worsened. SPP: Recreational Programming:− Hockey Canada skills academy a 2 credit physical education program. 5 days out of 10 on ice at a recreation center. SPP: Environment/Equipment:− Yes, no bike racks due to school renovations.

School 44
Co-SEA: Added an outdoor basketball court, added a closed road for hockey, biking; and the condition of fitness/weight room improved. SPP: Public Health: − Walking program became defunct over the past year, we are examining ways to get it started up again as well as implementing a house system pedometer walking challenge as recommended in 13/14 COMPASS results. SPP: Recreational Programming: − "house" system has been implemented, gr. 9-12 students now have additional opportunities to participate in friendly grade by grade sports competitions on a monthly basis Grade 7/8 students practiced after school and participated in [community kids marathon]. SPP: Environment/Equipment: − Major renovation in the school allowed the creation of a cross-fit space, being well utilized by phys. ed department and as part of the after school fitness program.  over the one-year period. Considering the current study looked at change over time, it is important to keep in mind that this was not an examination of the presence or absence of policies, recreational programming, public health resources, or features of the physical environment. Therefore, it is quite possible for the control schools (n = 25) to have had existing initiatives in place and did not feel the need to make any PA related changes. As a result, this could explain the mixed impacts on student MVPA from what seemed to be positive PA school-level changes. Given the paucity of longitudinal studies examining the impact of school-led changes over time, it is difficult to directly align these findings with evidence from the current literature. However, ecological models [11] and previously identified associations are available to help support and interpret the results of this study.

School 45
Changing only aspects of the physical environment resulted in increased MVPA in one school, and decreased MVPA in two schools. The addition of a bike rack by school 23 resulted in increased student MVPA and is of particular interest considering two other schools also made this change, while a third school incorporated it among other changes. However, the addition of a bike rack in these three other schools did not significantly impact student MVPA. In previous research, the presence of bike racks alone have not been associated with student MVPA [22][23][24]. However, there is evidence to suggest that bike riding is a popular activity among high school students [46], and that active transportation (e.g., biking to school) is one way for students to engage in MVPA [47]. While this study did account for school-level variability and covariates, it did not observe features of the neighbourhood built environment that may influence MVPA [48]. Therefore, differences within the surrounding community environments of these schools may explain why the addition of a bike rack significantly impacted student MVPA in only one out of four schools. Another physical environment change that significantly impacted MVPA was the addition of a dance studio, which resulted in a decrease in student MVPA. In previous research dance studios typically have not been significantly associated with student MVPA [22][23][24]; therefore, it is unclear why student MVPA decreased in school 22. Restructuring space for dance studios has been suggested as a potential way to increase MVPA among low active groups [49]. However, given the results from this study it appears that more research is needed to determine whether adding a dance studio is a viable solution for schools looking to increase student MVPA. Finally, the last physical environment change that impacted MVPA was improving the condition of the fitness/weight room in school 31, which resulted in decreased student MVPA. Previously, facility condition has been positively associated with MVPA in both males and female students [50]. In the current study a change to facility condition was reported in 10 schools, in which five were specific to the fitness/weight room, yet only one of these significantly impacted student MVPA. Therefore, changing the condition of PA facilities may not be an effective strategy for schools to improve student MVPA as it appears to have a null or negative impact. Given the combined results, it appears that more research is needed examining the impact of objectively measured changes to the physical environment on student MVPA.
Changes to only recreational programming resulted in significant student MVPA changes in three schools. Of these, increased MVPA changes occurred from adding an out-and-abouters club in school 10, and from increasing access to the fitness centre at lunch in school 5. Given that the presence of a room with cardio or weightlifting equipment has previously been associated with increased odds Table 3 Description of physical activity related changes implemented between year 2 (2013-2014) and year 3 (2014-2015) of the COMPASS study (Continued) room during lunch hours, and supervised fitness room after school hours. SPP: Environment/Equipment: − Curtains in female shower stalls.

School 56
Co-SEA: Added a fitness/weight room. SPP: Recreational Programming: − Continued building relationships with community partners (i.e., with senior league golf and the curling club). Hockey academy (a 2 credit physical education package) is now available and is a focused course emphasizing specific activity skill development and conditioning. Soccer academy will be available next year. Healthy active living education courses are now 14 sections. Intramural expansion continues. Badminton club is growing.

School 57
Co-SEA: Removed 2 fitness/weight rooms, and condition of the field improved. SPP: Recreational Programming: − Boarding students have increased access to use fitness studio (now can use without adult supervision) if they go with a buddy.  of being physically active during recess [51], it does not seem out of place that students attending school 5 had increased MVPA. Further, it seems intuitive that offering an outdoor club in school 10 stemming from a student leadership initiative would also have a positive influence on student MVPA. However, it is unclear why the addition of a charitable run (Terry Fox Run) in school 11 resulted in a decrease in student MVPA. Although there is little evidence to suggest that one-time events are effective for increasing MVPA [52,53], it should be noted that it was unknown whether the Terry Fox Run occurred within the recalled week of MVPA (previous week). Therefore, it is possible that some other change not captured in the SPP or Co-SEA was responsible for the observed decrease in student MVPA in this school. Consequently, it appears that more research is needed to examine how one-time events held by schools impact student MVPA over time.
Changes to both the physical environment and recreational programming occurred in three schools. Intuitively, an increase in MVPA was observed in school 49 which added a weightlifting club, a 100 km run/walk club, archery and figure skating, increased access to the sports field at lunch, and improved condition of the outdoor basketball court. However, surprisingly decreased MVPA was observed in school 52 that had built a fitness room complete with spin bikes, bose balls, ping pong, and shuffle board, provided 10 sessions each of spin, yoga, and Zumba, and increased access to the gym at lunch. A decrease in MVPA was also observed in school 58, which added a dance club and improved the condition of their field. Previous research has found that alternate rooms for PA have been positively associated with MVPA [22][23][24]. Therefore, it is interesting that we observed decreased MVPA in school 52 for building a fitness room. One potential explanation could be the timing of data collection, as it was unclear how close the renovations occurred to the actual data collection date. Hence, it will be interesting to see how this change impacts student MVPA in the COMPASS year four data collection. Another potential reason for the difference in results observed between schools could be the extra-curricular activities that were offered. For example, increased MVPA occurred in school 49 from incorporating multiple new activities (e.g., weightlifting, run/walk club, archery, figure skating), which could have appealed to a broader range of male and female students and resulted in a larger proportion of students being active [46]. While there is evidence to suggest that activities offered in school 52, such as spin, yoga, and dance are  [46,54]. Lastly, the observed differences in results could be due to implementation success. Previous research has identified support from school principals, physical space, and scheduling with other school activities (e.g., varsity teams) as factors that can facilitate or impede the ability to offer extra-curricular activities [55,56]. In addition, student hunger, after-school transportation, and other student commitments (e.g., jobs, tutoring, family) were identified in previous research as potential issues that could influence participation in extra-curricular activities [55,56]. Given that the current study did not assess these factors, an examination of the facilitators and barriers experienced by these three schools that made changes to recreational programming could provide a better understanding of the results. Schools often cited policies that were embedded at the provincial level such as the daily physical activity initiative, and the physical education curriculum. However, these are not school-level policies and therefore were not considered an appropriate exposure for this study. Previous research has found that policies implemented beyond the school-level have experienced implementation issues [57,58], and that there seems to be a lack of communication between policy makers, school board officials, principals, and teachers [59]. Therefore, future research should continue to examine PA policies and evaluate the relationship between policies developed at the school level and student MVPA.
Piercy et al., 2015 [60] suggested that public health units can encourage schools to adopt health promoting programs and act as a facilitator between the school and local community resources. However, only 4/61 schools changed the way they collaborated with public health units, with none of these changes significantly impacting student MVPA. Of these four schools, three offered PA programs, and one was in the process of working with their public health unit. One potential reason for the lack of significant results stemming from these schools could be that these were new programs being implemented and it could take a while for them to run efficiently. Future research should continue to assess the relationship between schools and local public health units in order to identify the most effective strategies for increasing the amount of time students spend in MVPA.
While the results of this study are mixed, they solidify previous recommendations in which multiple school contexts such as the physical, social, and policy environments need to be examined concurrently in order to better understand how the complete school environment influences student MVPA [17]. Although this study assessed the physical, and policy environments, the social environment was not addressed. Understanding how previously associated factors such as perceived teacher support [61][62][63] and feelings of school connectedness [64,65], could have aided in the interpretation of the results. Schools may have sufficient facilities and a variety of activities to offer, however students may still need to feel supported by school personnel or have a sense of school connectedness before they engage in the opportunities that are provided. Therefore, more research is needed in order to understand how feelings of school connectedness, and teacher support are associated with the amount of time students spend in MVPA.
Strengths of this study included the quasi-experimental design, longitudinal data, objective and subjective measures of the physical environment, and the use of active-information passive-consent parental protocol. The COMPASS study uses active-information passiveconsent parental protocol for its capability to reduce school-level variance estimates, increase participation rates to obtain a representative sample of the entire student population within a school, reduce the risk for obtaining a biased sample, obtain accurate information regarding substance use, and ensure student confidentiality [66]. Despite these strengths mentioned, there were limitations that need to be considered. The COMPASS study purposely sampled school boards that met a predetermined inclusion criterion and therefore is not a representative sample of all Ontario and Alberta schools. As a result, this could limit the generalizability to larger, English speaking schools [25,67]. Another limitation was the use of a self-reported MVPA, which often results in overestimation of MVPA in adolescents compared to objective measures, such as accelerometers [68]. Further self-report measures are not as accurate in determining different intensities of PA as objective measures. However, given feasibility issues such as cost and time that are associated with objective measurement (e.g., accelerometers), selfreported measures like the Cq are acceptable for use in large samples. Further, if over-reporting did occur, it likely occurred at both times, minimizing the impact on MVPA change over time. Lastly, there are potential limitations regarding the use of the Co-SEA. Although the Co-SEA allows for objective measurement of the quantity of facilities present in physical environment, data collectors are required to subjectively assess the condition of these facilities. While training was provided to ensure a high degree of reliability was achieved, measurement error may still have occurred.

Conclusion
This study provided a quasi-experimental observation of how naturally-occurring school changes to PA policy, recreational programming, public health resources, and the physical environment impacted student MVPA over time. Based on the changes' that resulted in increased MVPA, it appears that providing increased access and multiple opportunities to be active may be an effective strategy for increasing MVPA in secondary school students. However, it was also found that the same schoollevel changes had different impacts on student MVPA. Further, in some schools even intuitively positive changes negatively impacted student MVPA. From an ecological perspective, it is possible that sources of influence at the interpersonal or community level could be interacting with these school-level changes [11]. Considering the dynamic nature of the school and its components (e.g., staff, students, surrounding community), future observations should assess multiple levels of the ecological model. This would be beneficial for understanding how these school components interact with one another, and the surrounding community to influence the amount of time secondary students spend in MVPA. Given the specificity of these results, it may be important for school principals to consider both the internal (e.g., staff, facilities) and external (e.g., community features) resources they have, as well the interests of the students in order to develop and deliver effective strategies for increasing student MVPA.
Abbreviations MVPA, moderate-to vigorous-intensity physical activity; PA, physical activity