药品不良反应信号检测中混杂因素的控制方法  

Methods of Controlling Confounding Factors in Adverse Drug Reaction Signal Detection

作　　者：钱维[1] 叶小飞[1] 王超[1] 贺佳[1] 

机构地区：[1]第二军医大学卫勤系卫生统计学教研室,上海200433

出　　处：《中国药物警戒》2010年第3期142-144,共3页Chinese Journal of Pharmacovigilance

基　　金：国家自然科学基金资助项目(No.30872186);上海市优秀学科带头人计划(A类)(No.09XD1405500)

摘　　要：上市后药品不良反应监测无法实现用药人群的随机化,因此无法控制药品不良反应信号检测中数据的混杂偏倚。通过介绍三种运用较为广泛的混杂因素控制方法,简单分析了各个方法的优缺点和在药品不良反应数据中的应用情况,以期为上市后药品不良反应的准确监测提供借鉴。It is impossible to randomize the allocation of people using the drug under the post-marketing condition, so hard to control the confounding bias in signal detection of adverse drug reaction（ADR） databases. This article briefly introduces three widely-used methods of controlling confounding bias. It give a brief description of the advantages and limitations of each method, and their applications in ADR signal detection in order to provide some effective information for the post-marketing surveillance.

关 键 词：信号检测 混杂 分层 LOGISTIC回归 倾向性评分法 

分 类 号：R95[医药卫生—药学] R994.11