Secular changes in height, weight and body mass index in Hong Kong Children

The study population included all school children in Hong Kong aged 6 to 18 y. Children were classified into 26 half-year age groups. As body weight can be assumed to be normally distributed, the sample size for each age group was calculated in terms of the standard deviation of the 100αth centile (c _100α, alpha = 0.05 for 95^th percentile) and the age-specific standard deviation (SD), using the sample size planning formula developed by Healy [10]. Using the figures of the 1993 growth survey as inputs and applying the above formula, a sample size of 200 boys and 200 girls for each half-year age group in the range 6 to 9 y was estimated to keep the 95% confidence interval (CI) within ± 1.5 kg. A sample size of 300 per group for those children aged 10 y or above was estimated to keep the 95% CI within ± 3 kg. Using these estimates, it was planned to recruit 14,000 children 6 to 18 y old.

A two-stage cluster sampling method was used. Data from the Education and Manpower Bureau was used to compile a sampling frame of all schools in Hong Kong. International and English Schools Foundation schools, which partly cater for expatriate non-Chinese children, were excluded. In the first stage of the sampling, one primary school and one secondary school were randomly selected from each of the 18 Districts in Hong Kong. Selection of a school was based on computer-generated random numbers according to numbers of schools per district, and if the selected school declined to participate, the next randomly selected school was invited to participate. In the second stage, two classes in each grade were selected in collaboration with the school principal based on timetables and operational needs. All students of the selected classes were invited to join the study. This method allowed the data to be comparable with those obtained in the 1993 Growth Survey. Parents were informed about the study through a letter distributed by the school and asked to inform the school if they did not wish their child to participate. Parents were told that participating children would be given a record of their body measurements, and that any child with high blood pressure or other abnormal findings would be referred for further assessment. Parents were invited to complete a questionnaire providing demographic information including gestation and birth weight, and family or personal history related to risk factors for obesity. Verbal consent was obtained from the students who were then asked to completed a Self-Administered Physical Activity Checklist [11], and a locally developed and validated rapid diet assessment questionnaire. Students were asked if they had attended the SHS for a physical examination during the current or previous academic year.

A team of eight trained research staff visited each selected school on a pre-arranged date to collect the anthropometric data. All instruments were validated following the standard methods of the manufacturers, and the balances were zero calibrated. Standing height without shoes was measured using a Harpenden Stadiometer (Holtain, UK) to the nearest 0.1 cm. Body weight was measured with the lightest clothing to the nearest 0.1 kg by an electronic weighing scale (Tanita BF-522, Japan).

The reported weights of the school children in the 1993 study were adjusted by subtracting the average weight (median 0.3 kg, inter-quartile range 0.3 – 0.4 kg) of the lightest school uniform in each school from the observed weight [6]. To ensure comparability of the 1993 data with our data, we used the original unadjusted measured weights from the 1993 data to compare with our unadjusted weights.

Percentile curves were constructed using LMS method. The sex-specific percentile values (10^th, 50^th, and 90^th) of weight-for-age, height-for-age, and BMI-for-age were compared with those of the 1993 data. The LMS method using maximum penalized likelihood was used to perform model fitting of the anthropometric centiles for the physical parameters [12]. The LMS method estimates the measurement centiles in terms of three age-sex-specific cubic spline curves: the L curve (Box-Cox power to transform the data that follow a Normal distribution), M curve (median) and S curve (coefficient of variation). In brief, if Y(t) denotes an independent positive data (e.g. height) at t age(year), the distribution of Y(t) can be summarized by a normally distributed SD score (Z) as follows:

Once the L(t), M(t), and S(t) have been estimated for each age t, the 100α th centile at t gestation weeks could be derived from

C _100α (t) = M(t) [1 + L(t)S(t)Z _α ]^1/L(t)

where Z_α is the α centile of the Normal distribution (for example for the 97^th centile, α = 0.75 and Z_α = 1.88).

The study was approved by the Joint The Chinese University of Hong Kong and New Territories East Cluster Clinical Research Ethics Committee and the Ethics Committee of the Department of Health of the Hong Kong Government.

For each age group and for both genders, height increased significantly both from 1963 to 1993 and again from 1993 to 2005/6 (Table 1). During both periods, the height increases were greater in the younger children, and particularly in the boys, who were 3 cm taller at ages 12 and 13 y, while girls' maximum height differences were 2 cm at age 8 y and again at ages 12 and 13 y. To better compare changes over the two different time periods, 1963 to 1993 and 1993 to 2005/6, the annual increases in height were also calculated (Table 1, Figure 3).

For all age and gender groups, weight and BMI increased from 1963 to 1993 and again from 1993 to 2005/6 (Tables 2 and 3, Figure 3). However for the girls older than 13 y, neither the weight nor the BMI increases were significant in the latter period. For boys, however, although the weight increased significantly at all ages during the latter period, BMI increased significantly only up through age 15 y during that time. Again, therefore as for height, during both periods, the weight and BMI increases were greater in the younger children, and particularly in the boys. The annual increases in weight and BMI tended to be greater in the earlier period, 1963 to 1993, than in the latter period, 1993 to 2005/6 (Tables 2, 3).

The 1993 and 2005/6 datasets were analysed according to different cut-offs to define overweight and obesity to reveal the percentages of overweight or obese children and adolescents in those two surveys (Table 5) [7, 9]. Further detailed comparisons of BMI by age group were made with recently published data from mainland China, which showed similar proportions of children having BMI above the 85^th and 95^th centiles (Table 4) [14]. Of the sub-sample of 8199 students who had attended the SHS during the current or previous year, 14.4% were >= 120% median weight-for-height, the SHS criteria to define obesity (Table 5).