多分类 logistic 回归在冠心病危险因素研究中的应用

目的：应用多分类logistic回归找出冠心病危险因素与冠状动脉病变程度间的关系，建立冠状动脉病变程度危险因素的“最优”回归方程。方法采取系统抽样的方法，对来自2005年1月至2009年12月期间山东省潍坊市某几所医院心血管内科初步诊断为冠心病并进行冠脉造影的患者病例，抽取256例作为多分类logistic回归分析的样本。冠心病组根据狭窄病变累及血管范围分为1支病变（单支病变组）、2支病变（双支病变组）和3支及以上病变（多支病变组）。通过多分类logistic回归方法分析冠心病与冠脉病变程度相关关系。结果以冠脉病变程度为因变量，各个因素为自变量。根据单因素分析的结果，从20个研究因素中筛选出8个有统计学意义的影响因素，经过数据相关性分析、共线性诊断及专业解释等，筛选出8个影响因素进入多分类logistic回归中，最后得到与冠脉病变程度有关的有统计学意义的影响因素6个，分别是年龄、合并疾病、心率、血糖、脂蛋白（a）；保护因素1个：X17（载脂蛋白A1）。根据筛选出的6个影响因素建立“最优”回归方程。结论应用多分类logistic回归找出与冠脉病变程度有关的危险因素，并定量分析出各危险因素在冠脉病变不同程度上的概率值。

Application of multinomial logistic regressoi n analysis on risk factors of coronary heart disease

Objective To set up the ′best′model by applying multinomial ordinal logistic regression analysis and to se-lect the risk factors and find out the ′best′model suitable to this research.Methods The subjects of this study came from the internal medicine of hospitals in Weifang of Shandong province between January, 2005 and December, 2009, who were suffering from heart disease.Systematic sampling was used to select 256 patients to be studied in our research.The CHD patients were di-vided into 3 rank groups, 1-vessel group, 2-vessel group and multi-vessel group patients.Finally, multinomial ordinal logistic re-gression analysis was used to analyze the correlation between the severity of CHD and the development of coronary artery.Results We took CHD as the response variable and the various factors as the explanatory variables.According to the results of single factor analysis, after carrying out correlation analysis and collinearity diagnostics and the specialized knowledge, 8 factors were selected for multinomial ordinal logistic regression.Six variables （age, combined disease, heart rate, glucose, ApolipoproteinA1 and lipoprotein （a）） were screened to enter the logistic regression.The ′best′model of multinomial ordinal logistic regression was established in the end. Co nclusion Risk factors of severity of CHD were found out by the application of multinomial ordinal logistic regression.The comparison among various risk factors had been made through the quantitative analysis.