广义相加模型和广义线性模型在糖尿病相关因素分析中的应用

目的 利用广义相加模型和广义线性模型探讨糖尿病与相关因素的关系。 方法 2010-2012年在广西5市/县采用分层整群抽样的方法,选取18岁及以上常住居民作为研究对象。3 827名被调查者均接受问卷调查,测量身高、体重、血压、腰围(WC),检测空腹血糖(FPG)。 结果 糖尿病患病率为9.4%,男性和女性患病率分别为10.3%、8.8%,差异无统计学意义(χ²=2.629,P=0.105)。单因素结果显示年龄、城乡、民族、文化程度、婚姻状况、饮酒、肥胖类型(OBPH)7个因素与糖尿病有关(P＜0.01)。多因素logistic回归分析结果显示,农村相对城市有降低患糖尿病的风险(OR=0.633,95%CI:0.499~0.802,P=0.000);60岁及以上人群与35岁以下人群相比患糖尿病风险高(OR=14.037,95%CI:6.538~30.134,P=0.000);中心型肥胖+超重、中心型肥胖+肥胖分别与正常体重比较,患糖尿病风险高(分别OR=2.259, 95%CI:1.705~2.994, P=0.000;OR=2.068, 95%CI:1.368~3.125, P=0.001)。广义相加模型和广义线性模型结果显示饮酒与糖尿病呈现J型非线性关系(χ²=7.712,P=0.019),饮酒＜1次/周和≥6次/周增加患糖尿病的风险;肥胖类型与糖尿病呈现为平躺∽型曲线关系(χ²=13.547,P=0.008),中心肥胖和低体重患糖尿病的风险增加。 结论 广义相加模型和广义线性模型能直观呈现饮酒、肥胖类型与糖尿病的非线性关系。

Generalized additive and generalized linear models in analysis of diabetes-related factors

Objective To explore the correlation between diabetes mellitus and its related factors based on generalized additive model and generalized linear model. Methods A stratified cluster sampling method was used to select permanent residents aged 18 years and above in 5 cities/counties of Guangxi from 2010 to 2012, and 3,827 selected residents served as the research subjects. A questionnaire survey was conducted, and the residents’ height, weight, blood pressure, waist circumference (WC)and fasting plasma glucose(FPG)were measured. Results The prevalence rate of diabetes mellitus was 9.4%, and no statistically significant difference was found in the prevalence rate between males and females (10.3% vs. 8.8%, χ²=2.629, P=0.105). Single factor analysis showed that 7 factors, including age, urban and rural areas, ethnic origin, educational background, marital status, drinking, and obesity phenotypes (OBPH), were associated with diabetes mellitus (P<0.01). Multivariate logistic regression analysis indicated that the risk of diabetes mellitus was higher in urban areas than in rural areas (OR=0.633,95%CI:0.499-0.802, P=0.000), higher in residents aged 60 years and above than in ones aged 35 years and below (OR=14.037, 95%CI:6.538-30.134, P=0.000) as well as higher in the residents with central obesity & overweight or central obesity & obesity than in ones with normal-weight (OR=2.259, 95%CI:1.705-2.994, P=0.000; OR=2.068, 95%CI:1.368-3.125, P=0.001). Generalized additive model and generalized linear model revealed that a type-J non-linear relationship was found between drinking and diabetes mellitus (χ²=7.712, P=0.019), and residents with drinking χ²=13.547, P=0.008), and residents with central obesity or low weight were at an increased risk of developing diabetes mellitus. Conclusions Generalized additive and generalized linear models can directly show the non-linear relationship between diabetes mellitus and alcohol consumption, obesity phenotypes.