前瞻性研究非独立性二分类结局的Poisson回归方法  

Prospective study of non-independent binary classification by Poisson regression

作　　者：王俊苹 沈灵智[2] 卢肇骏 寇硕 郑卫军[1] Wang Junping;Shen Lingzhi;Lu Zhaojun;Kou Shuo;Zheng Weijun(Public Health School,Zhejiang Chinese Medical University,Hangzhou 310000,China;Zhejiang Provincial Center for Disease Control and Prevention,Hangzhou 310000,China)

机构地区：[1]浙江中医药大学公共卫生学院,浙江杭州310000 [2]浙江省疾病预防控制中心,浙江杭州310000

出　　处：《中国医院统计》2022年第5期397-400,共4页Chinese Journal of Hospital Statistics

基　　金：浙江省医药卫生科技计划(2020KY091)。

摘　　要：目的基于SAS软件结合修正Poisson回归方法分析非独立性前瞻性数据。方法通过实例,比较常见方法和非独立Poisson回归法分析观察性研究的性能,并运用SAS软件分析数据。结果以普通Poisson回归法为标准参照,logistic回归得到的OR值明显偏离真实的RR值,非独立Poisson回归的RR区间估计最为精确。结论非独立Poisson回归法更适合于处理常见结局事件的非独立性前瞻性研究资料。Objective To analyze non-independent prospective data based on SAS software combined with modified Poisson regression.Methods The cluster data of observational studies were analyzed by comparing common methods and modified central effect Poisson regression analysis with SAS software.Results According to the standard of general Poisson regression,the OR value obtained by logistic regression deviates from the real RR value,and the modified central effect Poisson regression is more accurate.Conclusion Modified central effect Poisson regression model is more suitable for the treatment of common outcome events of non-independent prospective research data.

关 键 词：POISSON回归 非独立性数据 RR值 SAS软件 

分 类 号：R195.1[医药卫生—卫生统计学]