广义相对危险度模型法在探索人类流行性疾病发病危险性因素中的应用  

Application of general relative risk models on the analysis of risk factors of epidemic diseases

作　　者：裘炯良[1] 郑剑宁[1] 陈坤[2] 王建跃[3] 

机构地区：[1]宁波出入境检验检疫局卫生检疫处,浙江宁波315012 [2]浙江大学公共卫生学院,浙江杭州310031 [3]舟山市疾病预防控制中心,浙江舟山316000

出　　处：《疾病控制杂志》2006年第5期496-498,共3页Chinese Journal of Disease Control and Prevention

基　　金：卫生部基金项目(WKZ-2001-1-17)

摘　　要：目的为人类流行性疾病病因学研究提供一种科学的分析方法。方法建立涵盖相加、相乘等一系列模型通用的相对危险度模型———广义相对危险度模型,用偏差度统计量来衡量模型拟合效果并以χ2检验对模型的拟合优度进行分析,通过尺度参数λ的改变及相应偏差度的变化来选择最佳模型。结果以对舟山某流行性疾病发病资料的广义相对危险度模型分析验证该方法的科学性。当尺度参数λ等于1时,偏差度D值最低,即logistic回归模型为最优模型,并分析得出与该病发生相关的3大类14种危险性因素。结论应采用正确的分析方法来研究具有复杂病因的人类流行性疾病,才可以得出科学、正确、可靠的结论。Objective To study a kind of scientific method in the exploration of the risk factors of epidemic diseases. Methods General relative risk models including a series of adding and multiplication models was established. Deviance was used for measuring the fitting effects and goodness of fittest tested by chi-square analysis. The optimal multifactor model was selected according to deviance with the changes of λ values. Results The method was confirmed by the analysis on the data of some epidemic disease in Zhoushan Islands. While λ was one, deviance was the lowest. So the logistic regression model was recognized as the optimal model for this study. Three category and fourteen risk factors were found of being related to that epidenlic disease. Conclusions The most scientific, exact and credible conclusions was gotten only when right methods were applied for studying multifactor epidemic diseases.

关 键 词：危险性评估 流行病学 偏差 

分 类 号：R181.2[医药卫生—流行病学]