摘要
目的拟利用我市2013年居民健康档案相关数据,探讨logistic回归和分类树模型在高血压危险因素中的应用前景,并分析高血压的相关危险因素。方法抽取在我市25岁以上且居住5年以上的普通人群的代表性样本9 950例,按照预设调查内容开展问卷调查,利用logistic回归模型和分类树模型分析高血压危险因素。结果本次调查共抽取居民健康档案9 950例,调查问卷经筛查后有效问卷9 778份,有效率98.27%,满足研究条件。logistic回归分析显示,女性及轻中度职业人群是高血压的保护因素,而BMI高、未婚(独居)、大于25岁年龄组、有高血压家族史是高血压的危险因素。分类树分析显示,其危险因素主要有年龄、性别、高血脂、吸烟、饮酒、中心型肥胖、超重。高危人群主要分布在第4、6、9、11、12共5个节点内:终点6表现为中心型肥胖+超重+饮酒者;终点12表现为高血脂+超重者;终点9和11表现为中心型肥胖+超重+高龄及男性烟民;终点4表现为吸烟+饮酒+中心型肥胖者。logistic回归与分类树分析预测效果中等。结论中心型肥胖、超重、饮酒、高龄、高血脂症是高血压的危险因素,分类树模型和logistic回归模型都适合于高血压危险因素的判断,且前者的判断能力更好、更直观。
Objective To explore the application prospect of logistic regression and classifcation tree model in the risk factors of hypertension, and to analyze the related risk factors of hypertension by using the related data of the residents＇ health records in 2013. Methods 9 950 representative cases of the general population sample over 25 years old in our city at least 5 years of living were sampled and surveyed according to the preset questionnaire. Logistic regression model and classifcation tree model were used to analysis of risk factors of hypertension. Results In this survey, 9 950 cases of residents＇ health records were selected and 9 778 valid questionnaires were taken after screening with the effective rate 98.27%. Logistic regression analysis showed that female and mild to moderate occupational population were the protective factors of hypertension, while high BMI, unmarried, older than 25 years old, and the family history of hypertension were the risk factors of hypertension. Classification tree analysis showed that the main risk factors were age, gender, hyperlipaemia, smoking, alcohol drinking, central obesity and overweight. High risk population was mainly distributed in 4, 6, 9, 11, 12, a total of 5 nodes： the end of 6 showed central obesity＋overweight＋drinkers; the end of 12 showed hyperlipidemia＋overweight; the end of the 9 and 11 showed the central obesity＋overweight＋elderly and male smokers; the end of 4 showed smoking＋alcohol＋central obesity. The efficacy of predict the logisticegression and classifcation tree analysis were medium. Conclusion The isk factors for hypertension include central obesity, overweight, drinking, old age and hyperlipidemia. Classifcation tree model and logistic regression model are suitable for judgment of the risk factors for hypertension and the ormer model is better at judgment ability and more intuitive than the late model.
出处
《中国卫生标准管理》
2017年第24期7-10,共4页
China Health Standard Management
