G methods for handling time-varying confounding in the longitudinal study
10.3760/cma.j.cn112338-20200731-01001
- VernacularTitle:纵向研究中控制时依混杂的G方法
- Author:
Jie LIANG
1
;
Simin HE
;
Shuting CHEN
;
Tong WANG
Author Information
1. 山西医科大学公共卫生学院卫生统计学教研室，太原030012
- Keywords:
Time-varying confounding;
Parametric g-formula;
Inverse probability of weighting;
G-estimation
- From:
Chinese Journal of Epidemiology
2021;42(10):1871-1875
- CountryChina
- Language:Chinese
-
Abstract:
The conventional analytical methods cannot effectively adjust for time-varying confounding that occur in a longitudinal study and thus cannot correctly estimate the causal effects. This study explains the necessity of precisely controlling time-varying confounding and outlines G methods, including parametric g-formula, inverse probability of weighting, and G-estimation. We also compare the methods above to provide a reference for correctly estimating causal effects in the longitudinal study.
