运用广义线性模型探讨福建省汉族人群原发性肺癌影响因素  

Explore the influence factors on primary lung cancer in Fujian province Han population under the use of generalized finear model

作　　者：何斐[1] 肖仁栋 俞婷婷[1] 张鑫[1] 刘志强[1] 蔡琳[1] 

机构地区：[1]福建医科大学公共卫生学院流行病与卫生统计学系,福州350108 [2]附属第一医院胸外科

出　　处：《中华流行病学杂志》2015年第8期896-899,共4页Chinese Journal of Epidemiology

基　　金：国家自然科学基金（81172766,81402738）

摘　　要：以福建省汉族人群原发性肺癌环境危险因素研究为实例，通过构建广义线性模型，利用九值、偏差度及Pearsonf拟合优度检验等判断方法选择合适模型，进行多因素及因素间交互作用分析，发现自变量之间存在相乘交互作用，进而选用logistic回归模型分析，最终得到吸烟者有7个、非吸烟者有9个单独作用的肺癌危险因素。其中吸烟者中重度吸烟与肺部罹患过炎症性疾病呈正相乘作用，非吸烟者中被动吸烟与少吃新鲜水果呈正相乘作用。运用广义线性模型筛选合适模型并进行交互作用分析，利于全面合理分析流行病学数据。The purpose of this study was to use the data on lung cancer in Han Chinese in Fujian province to explore the value of a generalized, linear model and to investigate the impact related to environment factors on lung cancer as well as the independent and interaction effects on the development of lung cancer. SAS 9.2 was used to build a generalized linear model to evaluate the influence factors and interaction of lung cancer on both smokers and non-smokers. Results showed that the relationship of the factors was multiplied. Under the logistic regression analysis, seven risk factors and nine risk factors were noticed in smokers or in non-smokers, respectively. Heavy smokers and lung diseases appeared a positive multiplying effect on smokers while passive smoking and fresh fruits showed positive multiplying effects on non-smokers. The generalized linear models could filter suitable models thus facilitating further research on the interaction between the two. It seemed easy to carry on the comprehensive and rational analysis on related epidemiological data.

关 键 词：广义线性模型 肺癌 交互作用 影响因素 

分 类 号：R734.2[医药卫生—肿瘤]