Study design and participants
The Asia-Fit project was a cross-sectional study conducted in eight Asian metropolises including Bangkok (Thailand), Hong Kong SAR, Kuala Lumpur (Malaysia), Seoul (South Korea), Shanghai (China), Singapore, Taipei (Taiwan), and Tokyo (Japan). The aim of the Asia-Fit study was to investigate and compare the health-related fitness, movement behaviours, and body fat percentage in Asian adolescents. We selected these cities because they (a) could represent the most important metropolises in Asian countries and within-country regions, and (b) differ by geography, culture, and ethnicity. A detailed description of the study design and sampling methodology can be found elsewhere (24). In brief, a stratified random sampling strategy was used to recruit adolescents from secondary schools in each study city. Gender, age, and geographic locations of schools were considered in the sampling process. All students aged 12-15 years with good health status that could participate in physical education classes were eligible. We targeted adolescents aged 12-15 years because they were in a transition period from primary to secondary school, characterised by significant changes in their lifestyle behaviours such as increased recreational screen-time (25).
A sample of 12,590 adolescents aged 13.63 (± 1.01) years provided written consent signed by their legal guardians and participated this study during academic year 2013-2014. Data collection processes, equipment standards, test instructions, scoring were consistent in the eight cities by using the same operational procedures. The ethical committee on the Use of Human & Animal Subjects in Teaching and Research (HASC) in each study city approved this study.
Participants completed an interviewer-administered survey regarding MVPA, recreational screen-time, sleep duration, and covariates, and finished a test of body fat percentage during a physical education lesson under the supervision of trained research assistants. To reduce potential influence of seasonality, data collection was conducted at the same time in each city.
MVPA was measured using the interviewer-administered International Physical Activity Questionnaire-Short Form, IPAQ-SF (26). Participants reported the frequency (times/week) and duration (minutes/time) of walking, moderate activities, and vigorous activities in the last seven days. We only considered the activities that were performed in bouts of at least 10 minutes in duration. The frequency and duration were multiplied to calculate minutes of the three PA types per week. MVPA (minutes/week) was calculated as a sum of the minutes of moderate activities and vigorous activities.
Recreational screen-time was evaluated using the Adolescent Sedentary Activity Questionnaire (27). Participants reported their time spent on recreational screen activities such as watching TV, using internet, and playing video games separately for a typical weekday and weekend. According to previous studies (28), recreational screen-time (hours/day) was calculated by weighting the responses (screen time weekday × 5+screen time weekend × 2)/7.
Sleep duration. Participants reported the bedtime and wake up time separately for a typical weekday and weekend. We calculated the participants’ sleep duration by weighting the responses (sleep time weekday × 5+sleep time weekend × 2)/7.
According to the 24-hour movement guidelines for adolescents (20), meeting the PA guideline was operationalized as ‘average daily MVPA is at least 60 minutes per day’, meeting the recreational screen-time guideline was operationalized as ‘no more than 2 hours’, and meeting the uninterrupted sleep guideline was operationalized as ‘9 to 11 hours for 5-13 years and 8-10 for 14-17 years’. To have a complete profile of compliance with the movement guidelines, two variables were used in this study: (a) number of the guidelines being met as a continuous variable (from 0 = ‘none guideline met’ to 3 = ‘all three guideline met’), and (b) combinations of the guidelines being met as a category variable (‘none’, ‘only the PA guideline met’, ‘only the screen time guideline met’, ‘only the sleep duration guideline met’, ‘both of the PA and screen time guidelines met’, ‘both of the screen time and sleep duration guidelines met’, ‘both of the PA and sleep duration guidelines met’, and ‘all three guidelines met’.
Body fat percentage was assessed using bioelectrical impedance analysis (BIA, Tanita, TBF-543, Japan). BIA is a valid measurement of body fat percentage in adolescents (29). To accurate test results, we followed standard procedures and guidelines such as no beverage intake or engaging in MVPA for at least 12 hours prior to testing (30).
Covariates. Age, gender, perceived health status, life satisfaction, perceived sleep quality, and dietary intake were measured as covariates because they have the potential to influence the adolescents’ adiposity indicators (31, 32). Perceived health status was measured using one item (i.e., ‘How do you think about your health?’) on a 5-point Likert scale ranging from 1 (very bad) to 5 (very good). Life satisfaction was measured using one item (i.e., ‘Are you satisfied about your life?’) on a 10-point Likert scale ranging from 1 (the worst life) to 10 (the best life). Participants reported their sleep quality in the recent month on a 4-Likert scale (i.e., ‘How well your sleep in the recent month?’) ranging from 1 (very bad) to 4 (very good). The Food Frequency Questionnaire-Short Form (FFQ-SF) was used to evaluate the adolescents’ dietary intake (33). Participants were asked to report daily servings of water, fruit, vegetables, dairy products, meat/fish/eggs and carbohydrate on 7-point Likert scale (1= “none”, 7= “six servings or above”).
Data were analysed using IBM SPSS Statistics 25 (Armonk, NY; IBM Corp, 2017). Descriptive statistics including mean, standard division (SD), and percentages were evaluated. City differences in number of the guidelines being met and combinations of the guidelines being met were tested using one-way ANOVA and Chi-square statistics, respectively. Considering the hierarchical nature of the data (individual-level outcomes nested within schools), linear mixed-effects models were used to examine the associations between body fat percentage and meeting the 24-hour movement guidelines. Firstly, a null model that included only the dependent variable (i.e., body fat percentage) and the cluster variable (i.e., school) was tested to evaluate cluster effects. Then, we examined two models that differed by the guidelines met variables: Model 1 included number of the guidelines being met as a continuous variable and Model 2 included combinations of the guidelines being met as a category variable. In both of the two-level mixed effects models, the intercept for the dependent variable (i.e., body fat percentage) was free to vary by school. City, age, gender, perceived health status, life satisfaction, perceived sleep quality, dietary intake were added in the two models 2 as covariates because they have the potential to influence adolescents’ adiposity (23, 34). All the effect sizes were considered to be statistically significant when the p-value was less than 0.05.