GLM-Univariate和Logistic回归分析方法筛选影响血糖水平和高血糖的主要因素  

Stepwise selection of main influencing factors on the level of FBG and prevalence of hyperglycemia with GLM Univariate and Logistic regression model

作　　者：金永哲[1] 李跃[1] 方今女[1] 

机构地区：[1]延边大学医学部,吉林延吉133000

出　　处：《延边大学医学学报》2010年第3期196-198,共3页Journal of Medical Science Yanbian University

基　　金：国家自然科学基金资助(项目号:30260098;30860244);吉林省教育厅基金资助[2008(2);2006(12)]

摘　　要：[目的]筛选延边地区朝鲜族及汉族居民血糖水平和高血糖的主要危险因素.[方法]采用单变量分析的一般线性模型(GLM-Univariate)逐步筛选出主要影响血糖水平的因素和协变量,并采用逐步Logistic回归分析方法筛选高血糖的主要影响因素.[结果]通过GLM-Univariate分析方法所筛选出的影响血糖水平的因素是TG,HDL-C,WHtR,SBP,DBP,BMI,TC及性别;逐步Logistic回归分析结果显示,高血糖与民族、性别、中心性肥胖、高胆固醇血症、低高密度脂蛋白血症、高甘油三酯血症及年龄密切相关.[结论]采用不同统计分析方法所筛选出的影响血糖水平和高血糖的因素有所不同.OBJECTIVE To be screening the main influencing factors on the level of FBG and prevalence of hyperglycemia in Korean-Chinese and Han nationality of Yanbian area.METHODS Main influencing factors on the level of FBG and prevalence of hyperglycemia were screened with GLM Univariate and Logistic regression model.RESULTS Main influencing factors of blood glucose level were TG,HDL-C,WHtR,SBP,DBP,BMI,TC and sex with GLM Univariate,and main influencing factors of prevalence of hyperglycemia were nationality,sex,centrally obesity,hypercholesterolemia,hypo-HDL-C,hypertriglyceridemia and age with Logistic regression model.CONCLUSION The main influencing factors on the level of FBG and prevalence of hyperglycemia are some of different screened by using different statistical analysis.

关 键 词：LOGISTIC回归分析 方法筛选 血糖水平 高血糖 主要因素 Logistic regression model GLM FBG level influencing factors main 统计分析方法 低高密度脂蛋白血症 高甘油三酯血症 主要影响因素 一般线性模型 高胆固醇血症 中心性肥胖 单变量分析 逐步筛选 

分 类 号：R[医药卫生]