纵向匹配在滚动入组干预因果推断中的应用

Application of longitudinal matching in causal inference for rolling enrollment interventions

滚动入组是医学实践中常见的参与者招募方式。在纵向数据中,由于研究者更感兴趣治疗后一段时间的效果,因果效应定义与截面数据不同,这给纵向数据中匹配的应用带来新的挑战。纵向匹配是匹配方法在滚动入组干预等静态处理的纵向研究中的拓广。目前纵向匹配方法已在疗效比较研究中得到较为广泛的应用。本研究通过理论讲解结合实例分析介绍4种纵向匹配方法的基本原理、适用条件、代码实现和应用实例,可为纵向数据的因果效应估计提供方法学参考。

Rolling enrollment is a common method for participant recruitment in medical practice.In the longitudinal data,where researchers are often interested in outcomes occurring after a certain period of treatment,the definition of causal effects differs from that in the cross-sectional data.It poses new challenges for the application of matching methods in the longitudinal studies.Longitudinal matching is an extension of matching methods in longitudinal studies involving static interventions such as rolling enrollment.Currently,longitudinal matching methods are widely applied in the comparative effectiveness research.This article elucidates the fundamental principles,applicable conditions,code implementation,and application instances of four longitudinal matching methods through theoretical discussions and empirical illustrations.It provides methodological references for estimating causal effects in longitudinal data analysis.