The study sample consisted of 9433 male and female students divided into 3 age groups (10-13 yrs, 14-16 yrs, 17-19yrs), with a male: female ratio of 1.14 to 1 (Table ). The overall prevalence of overweight children (BMI ≥ 85th and <95th percentile) was 15.5%, whereas obese (BMI ≥ 95th percentile) children constituted 21.1% of the study population. Regardless of age or gender, approximately 55% of this population had a BMI-for-age within the normal range of between ≥ 5th and < 85th percentile. Male and female underweight (≤ 5th BMI-for-age percentile) children accounted for 10% and 6.8% of the population respectively. Overweight children (between the 85th and 95th BMI-for-age percentile) accounted for 14.4.% of males and 16.7% of females, with the remainder (25.8% male and 15.7% female) having BMI-for-age values of ≥ 95th percentile (Table and Figure ). There were significantly more boys achieving a BMI-for-age ≥ 95th percentile than girls, indicating a higher prevalence of obesity amongst male students (P < 0.001). The numbers of obese male, but not female children also increased with age, so that there were higher numbers of obese children aged 16-19 years than at 10-13 years (P < 0.001). Waist Circumference (W_C) measurements showed similar gender differences, with a higher percentage of boys achieving W_C scores in the ≥ 90th percentile compared to girls, regardless of age group (Table and Figure , P < 0.001). A greater number of female students also had W_C measurements in the ≤ 10th percentile range compared to males (Table and Figure , P < 0.001).
Anthropometric characteristics of the subjects divided into tertiles by age.
Distribution of Body Mass Index groups by gender and age group.
Distribution of Waist Circumference (W_C) groups by gender and age group.
The mean BMI, W_C and weekly intake of selected food items and macronutrients per age group for male and female children is shown in Table . Amongst the 9433 children surveyed, mean W_C significantly increased with age, with boys having larger W_C measurements than girls (Table , P < 0.001). Mean BMI measurements were higher in boys versus girls only at age 17-19 (P < 0.001). Sugar-sweetened carbonated beverage (SSCB) consumption varied from 5.93 to 9.04 servings a week, and was significantly higher than consumption of non-caloric sweetened "Diet" carbonated beverage (DCB), which varied between 0.92 and 1.52 servings per week (Table , P < 0.001). Whereas there was no significant difference between the total Kcals from the self-reported variables consumed by children aged 10 to 19, the reported frequency of weekly consumption of milk (both full fat and low fat), fruit, vegetables, fish, eggs and cereal, pizza, sweet snacks, ice cream and DCB decreased with advancing age in both genders (Table , P < 0.01). Conversely, the frequency of reported consumption of SSBC, added sugar in hot beverages and total sugar intake increased with age in both males and females, suggesting a trend towards sugar-rich foods and away from healthier food choices with advancing age. Additionally, whereas boys consumed significantly more SSCB than girls, and also more DCB, full-fat milk, eggs, fruit, savory snacks and added sugar in hot beverages; boys did not report consuming more fruit juice, low-fat milk, vegetables, fish, cheese, bread, cereals, fast food meals, pizza, sweet snacks or ice cream than girls, regardless of age group.
Descriptive characteristics of self-reported weekly food intake, exercise and sleep patterns in males and females; n = 5033, 4400 respectively.
Hours of both night-time and day-time sleep were surveyed, together with frequency of exercise occasions per week. The number of children reporting less than 6 hours of night-time sleep increased with advancing age, with a higher percentage of girls reportedly having <6 hours sleep compared to boys of similar ages (Table , P < 0.05). Conversely, more girls reported sleeping for one or more hours during the day compared to boys (P < 0.05). Frequency of exercise decreased with increasing age in both genders (Table , P < 0.001). Additionally, boys exercised more than girls across all age groups, with up to 40% of girls reporting performing no exercise at all (P < 0.05).
Table shows SPSS output tables for Pearson r correlations among male and female BMI, W_C and selected food intake frequencies for each of 21 food items. In order to exclude potential over- and under-reporting, we used a ± 1 standard deviation cut-off for the mean total Kcal intake as recommended by Ventura et al
]. After this exclusion, a total of 7031 data entries (74.5% of the total population) were subjected to correlation analysis, comprising of 3781 boys and 3250 girls. Of the data entries excluded from the correlation analysis, 11.1% of the survey population were found to be under-reporters based on the ± 1SD cut-off values, and 14.4% were over-reporters. Correlation analysis of the main portion of our population indicated that waist circumference (W_C) and BMI were positively correlated with SSCB intake in boys but not girls (r = 0.10 and 0.09 respectively, P < 0.001). SSCB consumption was positively associated with poor dietary choices in both males and females. Fast food meal intake, savory snacks, iced desserts and sugar intake correlated with SSCB intake in both males (r = 0.39, 0.13, 0.10 and 0.52 respectively, P < 0.001) and females (r = 0.45, 0.23, 0.16 and 0.55 respectively, P < 0.001). Full fat milk intake positively correlated with fruit, vegetable, eggs and cheese preferences in both boys (r = 0.20, 0.14, 0.17, 0.14 and 0.12 respectively, P < 0.001) and girls (r = 0.19, 0.14, 0.14, 0.15 and 0.17 respectively, P < 0.001). There was a negative correlation between W_C and full fat milk, fruit, vegetable and fish intake in males only (r = -0.07, -0.1, -0.09 and -0.07 respectively, P < 0.001).
Pearson correlation coefficients, between BMI, W_C, self-reported measures of food intake.
Although both BMI and W_C were inversely correlated with frequency of exercise in males (Table , P ≤ 0.001), this was not the case for females. However, exercise positively correlated with fruit, vegetable and unsweetened cereal intake in both genders, and also with full-fat milk intake in males only. Hours of night-time sleep was negatively correlated with BMI and W_C in both boys and girls, whereas day-time sleep correlated positively with SSCB intake in boys and negatively correlated with SSCB in girls.
Correlation coefficients between day/night-time sleep, exercise frequency, BMI, W_C and self-reported measures of food intake.
Table shows the final multivariate regression model for the correlates of BMI in boys and in girls. BMI positively correlated with male SSCB consumption (β 0.10, P < 0.0001), suggesting that every unit increase in self-reported SSCB consumption is associated with a 10% increase in BMI. BMI was also positively correlated with bread consumption in both genders, (P ≤ 0.0001), and added sugars in beverages also had a significant positive association with BMI. In both genders, hours of night time sleep was negatively correlated with BMI, and in boys, BMI was negatively correlated with number of exercise occasions (P ≤ 0.0001). Similarly, waist circumference was positively correlated with self-reported male SSCB intake in a multivariate regression model (Table , β 0.10, P < 0.0001).
Correlates of BMI in a multivariate regression model. ¥
Correlates of waist circumference in a multivariate regression model. ¥