基于多水平模型的2型糖尿病影响因素分析

发布时间：2018-06-27 12:28

  本文选题：糖尿病 + 型　； 参考：《中华疾病控制杂志》2017年11期

【摘要】：目的以2型糖尿病为例,探讨具有层次聚集性变量的多水平模型的构建。方法用2015年蚌埠市龙子湖区慢性非传染性疾病调查中的2型糖尿病相关数据作为研究案例,采用R 3.3.1软件对2型糖尿病的影响因素进行单水平和多水平Logistic回归分析。结果单水平Logistic回归和多水平Logistic回归分析筛选的影响因素不完全一致,主要有年龄、糖尿病家族史、高血压病史、水果摄入频率、糖化血红蛋白和体质指数(body mass index,BMI)。非超重(BMI≤23.9 kg/m~2)与超重(BMI≥24.0 kg/m~2)在单水平Logistic回归中有差异(均有P0.05),在多水平Logistic回归中显示无差异(均有P0.05)。赤池信息准则(akaike information criterion,AIC)在单水平和多水平Logistic回归模型的值分别为1 550.2和1 535.3。结论单水平与多水平Logistic回归模型所得出的结论并不完全一致,多水平模型在具有层次聚集性的数据分析中更具有优势。基于2型糖尿病的层次聚集性,应该以社区服务站为基础开展2型糖尿病的防控工作。
[Abstract]:Objective to study the construction of multi-level model with hierarchical aggregation variables in type 2 diabetes mellitus. Methods the data of type 2 diabetes mellitus in the Longzihu District of Bengbu City in 2015 were used as a case study. The single and multilevel logistic regression analysis was carried out on the influencing factors of type 2 diabetes by R3.3.1 software. Results the influencing factors of single level logistic regression and multilevel logistic regression analysis were different. The main factors were age, family history of diabetes, history of hypertension, frequency of fruit intake, glycosylated hemoglobin and body mass index (body mass). Non-overweight (kg/m~2 鈮，

本文编号：2073921