In prospective cohort study, multi follow up is often necessary for study subjects, and the observed values are correlated with each other, usually resulting in time-dependent confounding. In this case, the data generally do not meet the application conditions of traditional multivariate regression analysis. Sequential conditional mean model (SCMM) is a new approach that can deal with time-dependent confounding. This paper mainly summarizes the basic theory, steps and characteristics of SCMM.
在前瞻性队列研究中,经常需要对研究对象进行多次随访,其产生的多个观测值之间相互关联,常导致时依性混杂,这种情况下的数据一般不满足传统的多因素回归分析的应用条件。序列条件平均模型(SCMM)是一种可以处理时依性混杂的新方法。本文主要对SCMM的基本原理、步骤及特点进行概括。.
Keywords: Generalized estimating equation; Propensity score; Sequential conditional mean model; Time-dependent covariate.