Health Behaviors
Screen Time. Participants were asked to indicate the amount of time they usually spend each day on the following activities: i) watching/streaming TV shows or movies, ii) Playing video/computer games, iii) surfing the Internet, and iv) texting/messaging/emailing. Response options for each type of sedentary behavior included separate estimates for hours (0-9) and minutes (0, 15, 30, 45). Scores were added across each modality to represent the total screen time participants engaged in per day, on average.
Homework. Time spent on homework was assessed with a single item asking participants to indicate the amount of time spent on homework each day, on average. The response options followed the same format as the screen time items.
Sleep. Sleep duration was assessed using the same scale as the sedentary behaviors. Participants indicated how much time they usually spent sleeping each day with response options ranging from 0-9 hours and the number of minutes in 15-minute intervals.
Physical Activity. Time spent in MVPA was assessed using two items modified from the School Health Action, Planning and Evaluation System (SHAPES) physical activity questionnaire which was designed for multiple large-scale school-based data collections.24 Participants were asked to indicate the amount of time spent in moderate and vigorous physical activity each day over the past week. Response options included hours (0-4) and minutes in 15-minute increments (0 - 45). Participants were provided with written examples of moderate activities (e.g., walking, biking to school, and recreational swimming) and vigorous activities (jogging, team sports, fast dancing, jump-rope) prior to providing their response. Time spent in each intensity was combined to reflect total time spent in MVPA and was averaged across the 7 days to reflect average time spent in MVPA on a given day in the past week. The validity of the scores from the items are consistent with other self-reported assessments of physical activity among adolescents.25
Mental Health
Depression. The 10-item Center for Epidemiologic Studies Depression Scale Revised−10 (CESD-R-10)26 was used to assess depressive symptoms. Participants were asked to indicate the number of days over the past week they experienced a number of symptoms (e.g., “I was bothered by things that usually don't bother me”). Response options ranged from 1 (None or less than 1 day) to 4 (5-7 days). Scale values were mean-imputed for students missing one or two responses. Internal consistency of the scores was α = 0.82.
Anxiety. Anxiety symptoms were assessed using the Generalised Anxiety Disorder 7-item Scale (GAD-7).27 Participants indicated how often they had been bothered by problems such as difficulty controlling feelings of worry, trouble relaxing, nervousness, restlessness and irritability over the previous 2 weeks. Scale values were mean-imputed for students missing one or two responses. The internal consistency of the scores was α = 0.90.
Flourishing. The Flourishing Scale28 consists of 8-items designed to assess positive human functioning. Items assess dimensions of psychological well-being including positive relationships and feeling competent and meaningful in life (sample item: “I lead a purposeful and meaningful life”). The response scale was modified from the original 7-point Likert-scale to a 5-point Likert scale ranging from 1 (strongly agree) to 5 (strongly disagree). A single score was derived by summing participant responses across the 8-items. For ease of interpretation, the overall score was reversed such that higher scores reflect greater flourishing. Support for the construct validity of test scores and estimates of score reliability (coefficient α)29 have been reported in the literature.28,30 Scale values were mean-imputed for students missing one or two responses. The internal consistency of the scores was α = 0.90.
Covariates
Covariates measured included sex (female, male), grade (9–12), province, self-described ethnicity (White, Black, Asian, Latin American/Hispanic, Other/Mixed), weekly spending money as a proxy for socioeconomic status (SES), and self-reported height and weight to calculate weight status using age- and sex-adjusted cut points (recoded as underweight, normal-weight, overweight, obese, missing).31
Data Analysis
The distributions of the physical activity, screen time, homework, and sleep scores were examined. Based on the observed distributions of responses, students with missing data as well as unrealistic or outlier responses were removed. Sample characteristics were calculated and linear mixed models with random intercept were used to examine the impact of changes in MVPA, sleep, homework, and screen time on anxiety, depression and flourishing scores. All models controlled for sex, province, grade, ethnicity, weekly spending money, and BMI classification. Models with anxiety and flourishing outcomes controlled for depressive symptoms, while models with depressive symptoms controlled for anxiety. Random intercepts for each school were used to account for intraclass correlation (i.e. that students from the same school are more alike than students from different schools). All models were run using the GLMMIX procedure in SAS 9.4.
Consistent with previous methods,32 isotemporal substitution models were used to examine the impact of replacing 15 minutes of either screen time, homework, MVPA or sleep with another activity. In isotemporal substitution models, total activity time is held constant and a series of models are run removing one activity at a time. The covariates for the remaining activities represent the effect of replacing the removed activity with an equivalent amount of time of the remaining activity. For example, to examine the impact of replacing 15 minutes of screen time with either MVPA or sleep, the isotemporal substitution model is expressed as:
The coefficient for total time represents the removed activity (i.e. screen time), while the coefficient represents the impact of replacing 15 minutes of screen time with 15 minutes of MVPA, the coefficient represents the impact of replacing 15 minutes of screen time with 15 minutes of sleep, and the coefficient represents the impact of replacing 15 minutes of screen time with 15 minutes of homework.
Partition models were also run to examine the impact of increasing MVPA, screen time, homework, or sleep, while holding all other activity types constant. Partition models represent the effect of adding an activity, as opposed to substituting it with another activity. Unlike isotemporal models, total time is not held constant.
Based on previous research, non-linear associations between sleep, homework, screen time, and MVPA and each outcome variable were examined. Quadratic and square-root terms were tested using random intercept models. Because isotemporal models require linear associations, when single or partition models showed evidence of a non-linear relationship, the isotemporal models were stratified so that associations could be examined separately.