欢迎访问《中国循证儿科杂志》官方网站，今天是
English Version
中国循证儿科杂志

中国循证儿科杂志 ›› 2024, Vol. 19 ›› Issue (1): 24-30.DOI: 10.3969/j.issn.1673-5501.2024.01.005

• 论著 • 上一篇    下一篇

基于上海儿童青少年生长发育追踪研究数据库构建的成熟度偏移值预测公式的队列研究

潘其乐1,2　尹晓峰1,2　朱镕鑫1　蔡广1   

  1. 1 上海体育科学研究所（上海市反兴奋剂中心）　上海，200030；2 共同第一作者
  • 收稿日期:2023-11-17 修回日期:2024-01-24 出版日期:2024-02-25 发布日期:2024-02-25
  • 通讯作者: 蔡广

 Prediction equation of maturity offset constructed based on the Shanghai Longitudinal Growth and Development Study: A cohort study

PAN Qile1,2, YIN Xiaofeng1,2, ZHU Rongxin1, CAI Guang1   

  1. 1 Shanghai Research Institute of Sports Science, Shanghai Anti-doping Agency, Shanghai 200030, China; 2 Co-first author
  • Received:2023-11-17 Revised:2024-01-24 Online:2024-02-25 Published:2024-02-25
  • Contact: CAI Guang

PDF

291

摘要： 　摘要　背景　儿童青少年生长发育趋势被认为是“生活状况的生物标准”，体现基因轨迹、以营养和疾病为核心的环境因素、社会经济情况等综合效应。不同个体间存在生长发育上的速率差异。目的　通过身高、坐高和体重纵向随访数据构建的成熟度偏移值预测公式，评估个体生理成熟度。设计　队列研究。方法　基于上海生长发育追踪研究队列（SLGDS）数据库（简称SLGDS数据库），在6.0~14.0岁儿童青少年纵向监测数据中，采集有≥4次完整的体格测量指标（身高、坐高和体重）且每次采集间隔≥11个月的数据，其中男童10岁前和12岁后至少有1次体格测量指标，女童9岁前和11岁后至少有1次体格测量指标。选择年龄、身高、体重、坐高及其交互或比率的14个变量用于男童和女童成熟度偏移值预测公式的建模，采用相关性分析和Lasso回归模型，分性别进行预测变量的筛选，采用逐步回归构建成熟度偏移值预测公式。并对公式的预测性能，根据R2、估计标准误差（SEE）和Bland-Altman分析进行评估。主要结局指标　成熟度偏移值［测量时实际年龄-生长突增高峰年龄（aPHV）］。结果　在SLGDS数据库中，符合本文纳入和排除标准的儿童580名。建模人群439名（男童180名，女童259名），验证人群141名（男童56名，女童85名）。男童和女童建模组和验证组人群总体测量变量差异均无统计学意义。跟踪时长男童建模组人群较验证组人群短、女童建模组人群较验证组人群长，差异均有统计学意义。14个建模变量中，经相关性分析、lasso惩罚函数和线性回归，男童（年龄、坐高和BMI）和女童（年龄、坐高和克托莱指数）各3个变量分别进入预测公式函数，男童的成熟度偏移值=-15.553+0.705×年龄+0.067×坐高+0.063×BMI［可解释成熟度偏移值中91.6%的差异，SEE为0.625，偏差为-0.02年，95%一致性界限（LoA）为-1.244~1.210］，女童的成熟度偏移值=-14.240+0.668×年龄+0.084×坐高+0.002×克托莱指数（可解释成熟度偏移值中93.5%的差异，SEE为0.542，偏差为0.01年，95% LoA:-1.059~1.072）。基于验证组人群，本文公式男童和女童SEE分别为0.758和0.065；Mirwald公式2002加拿大男童和女童SEE分别为0.960和0.112；Moore1公式2015加拿大女童SEE为0.122，Moore1和Moore2公式男性SEE分别为1.021和1.076。总体显示，SEE值男童高于女童，本文公式的SEE值最低，预测效果更佳。本文公式在男性-1.5 aPHV和女性-1.2 aPHV年龄段，成熟度偏移值呈现剪刀叉但与校准曲线偏离更小；Mirwald公式和Moore公式在男性-3.6 aPHV、女性-6.7 aPHV年龄段成熟度偏移值呈现剪刀叉但与校准曲线相对本文公式偏离大。结论男童基于年龄、坐高和BMI，女童基于年龄、坐高和克托莱指数建立的上海儿童青少年人群的成熟度偏移值预测公式，预测效果优于目前国际常用的成熟度偏移值预测公式。

关键词: 队列研究, 儿童青少年, 成熟度偏移量, 预测公式

Abstract:

Abstract Background Trends in growth and development of children and adolescents are considered to be a "biological standard of living conditions", reflecting a combination of genetic trajectories, environmental factors centered on nutrition and disease, and socio-economic circumstances. Differences in growth and developmental velocity exist between individuals. Objective To construct an equation for predicting maturity offsets from longitudinal data on height, sitting height, and weight to assess the physiological maturity level of individuals. Design Cohort study. Methods According to the Shanghai Longitudinal Growth and Development Study (SLGDS), data were collected from healthy children aged 6.0-14.0 years who had more than four measurements of height, sitting height, and weight with an interval of at least 11 months between each data acquisition. Boys had at least one measurement before 10 years of age and after 12 years of age, and girls had at least one measurement before 9 years of age and after 11 years of age. Fourteen variables of age, height, weight, sitting height and their interactions and ratios were selected for the construction of prediction equations. Correlation analysis and Lasso method were applied to select predictors by gender, and stepwise regression was used to construct the equations. The predictive performance of the equations were evaluated by R2, standard error of estimate (SEE), and Bland-Altman analysis. Main outcome measures Maturity offset［chronological age at the time of measurement-age at Peak Height Velocity (aPHV)］. Results In the SLGDS, 580 children met the inclusion and exclusion criteria. The modeling sample was 439 (180 boys and 259 girls) and the validation sample was 141 (56 boys and 85 girls). Differences in overall measurements between two samples of boys and girls were not significant. The difference was statistically significant in the shorter follow-up duration for boys' modeling sample and the longer duration for girls' modeling sample. Three of the 14 variables were included in the equations for both of boys (age, sitting height, and BMI) and girls (age, sitting height, and quetelet index).Conclusion The maturity offset equations in Shanghai children based on age, sitting height, and BMI for boys, and age, sitting height, and quetelet index for girls were better than the current internationally used maturity offset equations.

Key words: Cohort study, Children and adolescents, Maturity offset, Prediction equation