人体测量学指标与多重代谢异常的非线性典型相关分析

Non-linear canonical correlation analysis between anthropometric indicators and multiple metabolic abnormalities

目的明确人体测量学指标整体与多重代谢异常症状整体的关系,以期对代谢综合征(metabolic syndrome,MS)的预防提出针对性的建议。方法采用随机整群抽样的方法,以河南省某县为研究现场对常住成年居民进行问卷调查、体格检查和血生化指标检测。采用SPSS 13.0的OVERALS过程进行非线性典型相关分析,并计算典型相关系数、多元相关系数,并绘制各指标关联强度的重心图。结果共获得调查对象2 914人,其中男性1 134人(38.9%)、女性1 780人(61.1%),平均年龄(50.58±13.70)岁。非线性典型相关的拟合结果为:损失量为0.577,占总变异的28.8%,相对较小,说明两组变量,即生化指标总胆固醇、总甘油三酯、高密度脂蛋白胆固醇、低密度脂蛋白胆固醇和空腹血糖与性别、年龄、体质指数及腰围之间的相关程度较大,常有同步变化趋势;多元相关系数显示各生化检测指标之间和性别、年龄及人体测量学指标之间均存在较强的内部相关,即常出现整体异常现象。各指标关联强度的重心图表明随着年龄增长代谢异常症状增多;女性代谢异常症状较多;超重及肥胖人群伴有多代谢异常;腰围与代谢异常呈正相关。结论 (1)生化指标、人体测量学指标组合变化的几率高;(2)年龄较大人群,特别是中老年男性及老年女性是一级预防的重点人群;(3)超重和腹型肥胖可作为预测多代谢异常的敏感指标;(4)非线性典型相关和重心图在分析多组变量时具有优势。

Objective To define the general correlation between anthropometric indicators and multiple metabolic abnormalities, and to put forward some particular suggestions for the prevention of multiple metabolic abnormalities. Methods A random cluster sampling was carried out in one county of Henan Province. Questionnaire, physical examination and biochemical tests were admitted to the adult inhabitants. Non-linear canonical correlation analysis （NLCCA） was applied with OVERALS of SPSS 13.0. The coefficients of canonical correlation and multiple correlation were calculated. The plot of centroids labeled by variables showed the correlation among various indicators. Results In total, 2 914 objects were investigated. It included 1 134 （38.9%） males and 1 780 （61.1%） females （60. 0% ）. The average age was（50. 58 ± 13.70）years old. The fitting result of NLCCA were as follows： the loss of 0. 577 accounting for 28.8% of the total variation was relatively small, and indicated that the two sets of variables of this study, namely sets of biochemical indicators （including serum total cholesterol, total triglyceride, high-density lipoprotein cholesterol, low density lipoprotein cholesterol and fasting plasma glucose） and sets of others （including gender, BMI and waist circumference ） were closely related and often changed synchronously. Multivariate correlation coefficient showed that internal indicators of the above two sets were closely related respectively and often showed the multiple anomalies of the same set. The diagram of the center of gravity of the association of various indicators showed that the symptoms of metabolic abnormalities increased with age. Women were more liable to have metabolic abnormalities. Overweight and obese people often suffer multiple metabolic disorders. Waist circumference was positively correlated with metabolic abnormalities. Conclusion （ 1 ） Biochemical indicators and anthropometric often change in combination. （2）Much attention should be paid to older people especially middle-aged or older men and older women in primary prevention. （3） Overweight and abdominal obesity can be considered the sensitive predictive indicator of multiple metabolic abnormalities. （4）Nonlinear canonical correlation and center of gravity Figure had the advantage of analyze the correlation between multiple sets of variables.