Fig. 4
Temporal trend from 2018 to 2024 of height, weight, and BMI by age (year) and by gender. Temporal trend slopes and 95% confidence intervals (95%CI) of (a) height. (b) weight. (c) body mass index (BMI). The trend slopes (coefficients) and 95%CI estimated from Generalized Estimating Equations (GEE) models were plotted by age (in year from 2 to 18) and by gender together with observed values and were summarized in the adjacent table. The correlation structure for GEE models was selected based on the smallest Quasi Information Criterion (QIC). For height, weight, BMI of both males and females, “independent” correlation structure had the smallest QIC and thus was selected (detailed results regarding QIC of GEE models with different correlation structures was summarized in Supplementary Table 6). The results of the GEE models with “independent” correlation structure were displayed here. The calendar year from 2018 to 2024 was standardized by minus 2018 (i.e. after standardization, the year 2018 became 0, and the year 2024 became 6 and so on). The GEE models for height, or weight, or BMI contained height, or weight, or BMI as dependent variable, respectively, and standardized calendar year, age in years and interaction term between calendar year and age in years as independent variables. The GEE models were fitted separately for males and females. Result interpretation: The positive slopes with 95%CI which did not include zero were interpreted as significant increase (e.g. the height of males at 18 years old significantly increased by 0.32 cm (95%CI = 0.08 cm to 0.57 cm) per year from 2018 to 2024). The negative slopes with 95%CI which did not include zero were interpreted as significant decrease (e.g. the BMI of males at 8 years old significantly decreased by −0.14 kg/m2 (95%CI = −0.17 kg/m2 to -0.1 kg/m2) per year from 2018 to 2024). The slopes with 95%CI which included zero were not significant (no significant increase nor decrease trend).
