[Retrospective mixed-longitudinal study on the growth trajectory of height among children and adolescents]

Longitudinal studies in height for school-age children and adolescents mainly focused on the velocity, spurt and peak of growth, and currently no changing growth trajectory or channel around the baseline were reported except the shift of percentiles among infants. This study aimed to analyze general characteristic of the growth trajectory of height among children and adolescents so as to provide scientific evidence for early detecting height deviation and evaluating treatment interventions for pediatric clinical and health care professionals.

Height measurement data of 4 632 school-aged children and adolescents were retrospectively collected from two middle schools in urban Beijing between 2006 and 2012, with physical examination records varying from 1 to 8 times. Corresponding Z-score curve of the first height data of each study subject was defined as the baseline growth trajectory. Interval censored data were determined by calculating and comparing the amount of up or down floating deviation of those follow-up measurements around the baseline trajectory at different time points, and the LIFEREG procedure was used to fit parametric regression model. Defined 0.67 s as isometric growth channel, proportions of follow-up measurements along the baseline or off-baseline channel were calculated at different baseline Z-score interval channel.

Totally 3 308 subjects with height examination records of 3 times or more were included in this study. Interval censored data of height abide by the exponential distribution based on parametric regression model. The exponential distribution model with two covariates of sex and baseline Z-score interval channel showed that sex was not statistically significant (P = 0.186 6). After removing factor of sex, the model only with baseline Z-score channel displayed that those Z-score channels over +2.00 s (P = 0.946 7) and -2.00-1.34 s (P = 0.091 1) were not statistically significant with the control of the channel below -2.00 s. Refined exponential distribution model (P < 0.000 1) was constructed after the Z-score channel reduced to six consecutive isometric growth channels. There were district differences in height deviations around the baseline trajectory in different baseline Z-score interval channels, floating downward in upper channel and floating upward in lower channel. Overall, the range of deviation of 4.06 cm was observed around the baseline trajectory in 90% of individuals (from P5 to P95), with the floating down 2.60 cm and the floating up 1.46 cm. The proportions of the individuals growing along the original channel were 43.8%-47.4% in the upper Z interval (0-2.00 s) and 33.0%-37.9% in the lower Z interval (-2.00-0 s); the proportions not shifting ± 1 channel were 94.0%-94.8% in the upper part and 79.1%-91.0% in the lower part.

This study obtained general characteristics of growth trajectory of height among children and adolescents: the deflection range of shifting up and down is about 4 cm around the baseline trajectory for 90% individuals, with the downward float 2.00-3.00 cm and the upward 1.00-2.00 cm; and 30%-50% individuals go along the baseline channel and 80%-90% individuals do not exceed ± 1 channels.