摘要目的:探究生活环境因素对我国儿童青少年BMI的影响。方法:利用中国健康与营养调查数据，研究对象为2000、2006、2011和2015年4轮调查中参与调查、且数据完整的7~17岁儿童青少年共6 626人（男性3 473人，女性3 153人）。首先在三水平随机截距混合效应模型中进行环境因素单变量分析，然后利用偏最小二乘回归拟合线性模型进一步筛选环境因素，最后分城乡拟合三水平随机截距混合效应模型进行分析。结果:三水平随机截距混合效应模型的空模型拟合结果对数似然估计值-17 034.68，随机效应似然比检验 χ 2=483.06， P<0.001；社区分层组内相关系数（ ICC）=9.97%，个人和社区水平总 ICC=39.38%；三水平模型显示城市模型中城市化指数（ β=-0.05，95% CI：-0.09~-0.01， P<0.05）、公园所在位置在本市另一居委会（ β=-0.88，95% CI：-1.72~-0.04， P<0.05）、15~17岁组（ β=-1.04，95% CI：-1.78~-0.30， P<0.05）与BMI值呈负相关，社区到最近健身房的距离（ β=0.12，95% CI：0.02~0.22， P<0.05）、家庭彩电数（ β=0.50，95% CI：0.08~0.92， P<0.05）、家长提示电视内容虚构的频率为3~4次/周（ β=1.85，95% CI：0.70~3.00， P<0.05）与BMI值呈正相关；农村儿童青少年模型中的城市化指数（ β=-0.04，95% CI：-0.07~-0.01， P<0.05）、农村人均收入（ β=-7.29e -4，95% CI：-1.00e -3~-6.77e -5， P<0.05）、家长规定看电视的频率≥5次/周（ β=-1.29，95% CI：-2.36~-0.21， P<0.05）、脂肪供能比（ β=-0.03，95% CI：-0.06~-1.00e -3， P<0.05）与BMI值呈负相关，家长提示电视内容虚构的频率≥5次/周（ β=3.01，95% CI：0.03~6.00， P<0.05）、调查时间为2015年（ β=4.83，95% CI：1.96~7.69， P<0.05）与BMI值呈正相关。 结论:环境因素从多水平多方面间接地影响着儿童青少年BMI的变化，城市化指数和农村人均收入对儿童青少年BMI的升高有微弱的保护作用，社区层面需要关注儿童青少年生活环境周边的活动场所的设置，而在家庭中应当正确引导儿童青少年看电视行为。

abstractsObjective:To investigate the influence of environmental factors on body mass index of children and adolescents in China.Methods:Using data from the China Health and Nutrition Survey, the research object to 7 - 17 years old children and adolescents who participated in 2000, 2006, 2011, and 2015 round of survey with complete data, a total of 6 626 children and adolescents (male 3 473, female 3 153) were investigated under univariate analysis for each environmental factor variable before using Partial Least Square Regression fitting a linear model for further screening. Finally, we fitted a three-level linear mixed-effects model distinct by urban and rural area for analysis.Results:The three-level null model, log likelihood=-17 034.68, χ 2=483.06, P<0.001. Intern-class correlation coefficient ( ICC) showed that community-level was 9.97%, and both community and individual were 39.38%. The three-level model also showed that urban model's urbanization index ( β=-0.05, 95% CI: -0.09--0.01, P<0.05), the park location ( β=-0.88, 95% CI: -1.72 - -0.04, P<0.05), 15 - 17 age group ( β=-1.04, 95% CI:-1.78 - -0.30, P<0.05) were negatively correlated with BMI. The distance to the gym ( β=0.12, 95% CI: 0.02 - 0.22, P<0.05), the number of home TV sets ( β=0.50, 95% CI: 0.08 - 0.92, P<0.05) and the frequency of parents' alarm of fiction TV program contents ( β=1.85, 95% CI: 0.70 - 3.00, P<0.05) were correlated with BMI. Rural urbanization index ( β=-0.04, 95% CI:-0.07 - -0.01, P<0.05). Rural per capita income ( β=-7.29e -4, 95% CI:-1.00e -3 - -6.77e -5, P<0.05), parents' restricted frequency of watching TV ( β=-1.29, 95% CI:-2.36 - -0.21, P<0.05), adipo-energy ratio ( β=-0.03, 95% CI:-0.06 - -1.00e -3, P<0.05) were negatively correlated with the BMI. Factors as the frequency of parents' alarm of fiction TV program contents ( β=3.01, 95% CI: 0.03 - 6.00, P<0.05), the survey time was 2015 ( β=4.83, 95% CI: 1.96 - 7.69, P<0.05) were correlated with BMI. Conclusions:Environmental factors could indirectly influence the change of BMI of children and adolescents to different degrees and various aspects. Urbanization index and rural per capita income had a slight protective effect on increasing BMI of children and adolescents. At the community level, attention should be paid to the setting of activity places around the living environment of children and adolescents. Family members should also guide their children and adolescents to develop nice behavior in watching TV.