三种统计分析方法在婴儿生长发育随访资料中的比较研究

A Comparative Study of Three Statistical Analysis Methods on Baby Growth and Development Follow-up Data

目的运用重复测量方差分析、混合线性模型、广义估计方程分别拟合生长发育随访资料,对三种统计方法分析的结果进行比较,探讨婴儿体重增长的影响因素。方法采用回顾性出生队列资料,运用SAS9.2等统计软件进行分析。结果重复测量方差分析得到婴儿体重的增长与出生体重、月龄、是否就医有关,且后两者存在交互作用;混合线性模型得到出生体重、月龄、母亲文化水平、喂养方式、睡眠时长,以及月龄与喂养方式的交互作用有统计学意义;广义估计方程分析得到出生体重、月龄、喂养方式、睡眠时长、以及月龄与喂养方式的交互作用具有统计学意义。结论三种统计方法均可应用于生长发育随访资料分析,各有优缺点,但分析结果存在一定差异,应根据具体的适用情况,判断何种方法分析的结果更为可靠。

Objective Using the method of repeated measure analysis of variance,mixed linear model, generalized esti- mating equations to fit the growth and development data respectively, comparing the results of three statistical analysis methods and exploring the influence factors of baby weight gain. Methods We used the SAS9. 2 software to analyze the retrospective birth cohort study. Results The method of repeated measure analysis of variance revealed that birth weight,month age,health seeking and the interactions of month age and whether or not go to a doctor had a significant effect on baby weight gain. The method of mixed linear model revealed that birth weight, month age, level of maternal education, feeding pattern, sleeping time, and the interactions of month age and feeding pattern had a significant effect on baby weight gain. We got the same results with the mixed linear model except for level of the maternal education by method of generalized estimating equations. Conclusion Three statistical methods can be applied to analyze the growth and development data and each method has advantages and disad- vantages respectively, though there are some differences between the results. We should consider the specific conditions and determine to choose which kind of method for more reliable results.