Procedure
The COMPASS study is a prospective cohort study that collects data from a convenience sample of students in grades 9 to 12 in British Columbia, Alberta, Ontario, and Quebec, Canada (28). Schools are recruited based on their use of an active information passive-consent protocol which is essential in self-report research to produce more robust results that limit self-selection and response bias, particularly for measures of substance use behaviours (28–30). The student-level data collected are self-reported via a questionnaire administered to students during class time. The COMPASS study also collects school-level information on policies, programs, and the built environment. The hierarchical nature of the data can allow for more robust statistical analyses and evaluations of programs over time. A full description of the COMPASS study methods can be found in print (28) or online (www.compass.uwaterloo.ca). All procedures were approved by the University of Waterloo Office of Research Ethics (reference number 30118) and appropriate school board committees.
Participants
In Year 6 (September 2017-June 2018 school year), 65,892 students participated in the COMPASS study. Student response rate was 81.8% and the primary reason for non-response being absenteeism at the time of data collection. Students with missing values (n = 5,257; 8%) were excluded from this study. The Supplemental File includes information on missing data (Table 1) for each variable and a flow diagram (Fig. 1). The Supplemental File Table 2 presents a chi-square comparison of students missing intramural participation data (i.e., outcome of interest) by demographic characteristics. We found significant differences between students with complete versus missing data for measures of gender, grade, and ethnicity. The current analyses used complete data from 60,635 students from 121 schools in Alberta (n = 8), British Columbia (n = 16), Ontario (n = 60), and Quebec (n = 37).
Instrumentation
School-level data (SPP Questionnaire and Statistics Canada Data). School-level data were collected using the Schools Policies and Programs (SPP) questionnaire, which is completed annually by a school contact. Additional socioeconomic data was collected from the 2016 census.
At the school level, a contact at each school (typically a principal, guidance counsellor, or gym teacher) completed the SPP questionnaire. To assess the number of intramurals offered at each school, school contacts were asked to “Please select the intramural programs/club activities involving physical activity that were offered to students at your school during the past 12 months.” Program selection included traditional competitive sports such as soccer and badminton as well non-competitive activities such as yoga or fitness class. School contacts were additionally asked to indicate whether the offerings were for males/females only or co-ed. There was also space to indicate other activities not listed. The number of co-ed intramurals offered were categorized based on four quartiles: Q1 (0–4 intramurals), Q2 (5- 7intramurals), Q3 (8–10 intramurals) and Q4 (11–17 intramurals). School contacts also provided school enrollment numbers to determine school size.
Data on schools’ urbanicity was determined by using Geosearch lookup on city name based on 2016 census data (31). Urban/rural classifications were as follows: large urban (populations from 100,000 and greater and a population density of at least 400 per square kilometre), medium urban (populations between 30,000 to 99,999 and a population density of at least 400 per square kilometre), small urban (populations between 1,000 to 29,000 and a population density of at least 400 per square kilometre), and rural (population less than 1,000 or population density less than 400 per square kilometre). School neighbourhood median family income was determined using school postal code to identify household median income in this area, data from the 2016 census (32). School neighbourhood median family income was categorized into 4 groups: less than $50,000, $50,001-$75,000, $75,001-$100,000, and greater than $100,000.
Student level data (Student Questionnaire). To determine student participation in intramurals students were asked “Do you participate in before-school, noon hour, or after-school physical activities organized by your school? (e.g., intramurals, non-competitive clubs)” with response options “Yes, No, or None offered at my school” Students who indicate that none were offered will be grouped with students who responded “No.”
Consistent with other youth health research (33), the following covariates (with their response values in brackets) were included in the analyses: grade, (9,10,11,12), gender (female, male), ethnicity (white, other) weekly spending money (Zero, $1 to $20, $21 to $100, More than $100, Don’t know).
Data Analysis
Descriptive characteristics at the school (n = 121) and student (n = 60,635) levels were examined by gender. An empty Generalized Linear Mixed Model (GLMM) was used to calculate the Intraclass Correlation (ICC) to determine the variability of student intramural participation between schools. Due to the hierarchical nature of these data (students nested within schools), three hierarchical Generalized Estimating Equation (GEE) models via PROC GENMOD were used to create three models. First, examining the association between co-ed intramural quartiles and total student intramural participation. The second model examined if offering female-only intramural programs was associated with intramural participation among females. Finally, the third model studied the relationship between offering male-only intramurals and participation among males. All models controlled for the covariates listed in the above section.