Participants
Students were recruited from all three grades (7th, 8th, and 9th) of all eight public junior high schools located in Kikugawa City and Kosai City, both of which are suburban areas within Shizuoka Prefecture, located in central Japan (N = 2968). The 7th, 8th, and 9th grades in Japan are generally for students aged 13, 14, and 15 years, respectively. In the Japanese education system, junior high school is a lower-secondary school that comes after elementary school and before senior high school. The survey was conducted between December 2012 and January 2013. Homeroom teachers guided students in filling out the surveys during class time, and completed questionnaires were collected in sealed envelopes. Completion of the anonymous self-assessment questionnaire was considered consent to participate in the study. Completed questionnaires were then collected and placed in sealed envelopes individually. Questionnaire of students who rejected this survey were also collected remains blank, in the same way as other students. (Additional file 1).
Ethics statement
Written explanations of the study were provided to students, parents, and guardians, as well as homeroom teachers and principals. Those who did not wish to participate in the study could decline to respond. Written informed consent was assumed by voluntary response of the anonymous questionnaire. In Japan, according to ethical guidelines for epidemiological research, written informed consent is not necessary for observational research that does not collect human biological specimens, such as blood or urine.
Measurement
Weight and height were self-reported in the questionnaire in kilograms and centimeters respectively, and recorded to the first decimal place. Previous studies in Japan have shown that self-reported child weight and height is highly accurate [19]. BMI (kg/m2) was calculated as weight (kg) over height squared (m2). To adjust for child age and gender, BMI z-scores were calculated based on WHO standards [20].
Because it would be difficult to obtain exact figures of students’ annual household income, we assessed economic status using a subjective measurement of economic status as follows: “How do you rate your household living-conditions as financial situation?” This item was scored on a 5-point Likert scale: (1) “difficult,” (2) “somewhat difficult,” (3) “normal,” (4) “somewhat comfortable,” and (5) “comfortable.” This subjective measurement of economic status was used in the 2010 Comprehensive Survey of Living Conditions conducted by the Japanese Ministry of Health, Labour and Welfare [21] and validated by comparing the question with annual household income reported in another study [22]. To focus on childhood poverty, we divided economic status into two categories: “low” (difficult: boys 4.28 %, girls 3.35 %) and “non-low” (somewhat difficult: boys 14.88 %, girls 19.06 %; normal: boys 63.56 %, girls 60.72 %; somewhat comfortable: boys 12.92 %, girls 10.33 %; and comfortable: boys 4.36 %, girls 6.53 %).
Covariates
The questionnaire included questions about sociodemographic factors such as gender, date of birth, grade in school, lifestyle topics including eating, activity, sleeping patterns and smoking status, self-rated health status, family members’ smoking status and, family structure. Breakfast skipping and the number of meals from convenience stores and fast-food outlets (convenience stores and fast-food outlets) were rated if the frequency of skipping breakfast and eating meals from convenience stores or fast-food outlets were ≥ 1 time per week. We also assessed participation in club activities at school (club activities at school) (“physical,” “non-physical,” “both physical and non-physical,” or “no participation”), and participation in physical and non-physical activities outside of school (activities outside of school) with a yes/no. In the Japanese education system, public junior high schools provide non-compulsory club activities that are either physical (e.g., baseball club and soccer club) or non-physical (e.g., watching movies, brass band, and chorus). For sleeping patterns, we assessed the number of times students got up and went to bed each night. Sleeping patterns were then divided by two, using a less than first quintile cut-off (less than 6.5 h per night). Self-rated health status was scored on a 5-point Likert scale of (1) “excellent,” (2) “good,” (3) “fair,” (4) “somewhat poor,” and (5) “poor,” which was then reorganized into two categories of “good” (excellent, good, and fair) and “not good” (somewhat poor, poor). We used a self-rated question for health status obtained from the 2010 Comprehensive Survey of Living Conditions [21], which was correlated with the Birleson depression self-rating scale (DSRS) (boys: Spearman’s r = 0.43, p < 0.001; girls: Spearman’s r = 0.48, p < 0.001). Smoking status was assessed using the question “Have you ever smoked before (even if only once)?” with a yes/no response. As for family members who lived together, we divided students into family structure of “living with both of their biological parents” and “those living in a single-parent family or stepfamily. Finally, family members’ smoking status (father, mother, and other) was rated and classified into two categories of “yes” (≥1 person) and “no” (none).
Statistical analysis
We examined the association between economic status and BMI z-scores using linear regression analysis. Univariate regression analysis was conducted in Model 1, and school, grade, and family structure were added to Model 2 as confounding variables. In Models 3, 4, 5, 6, and 7, we added the following possible mediators: family members’ smoking status (Model 3), breakfast skipping (Model 4), activities outside of school (Model 5), sleeping patterns (Model 6), and self-rated health status (Model 7), owing to estimate change of coefficient. Finally, multivariate regression analysis was performed for economic status using explanatory variables and an objective variable of BMI z-scores, and was adjusted for all confounders and all possible mediators (Model 8). We used STATA version 13.0 (STATA Corp LP., College Station, TX, USA) for statistical analysis.