Marginal Structural Models with Counterfactual Effect Modifiers.

Wenjing Zheng,Zhehui Luo,Mark J Van Der Laan

doi:10.1515/ijb-2018-0039

Abstract

2000;11:561-570), and a non-targeted G-computation estimator (Robins JM. A new approach to causal inference in mortality studies with sustained exposure periods - application to control of the healthy worker survivor effect. Math Modell. 1986;7:1393-1512.). The comparative performance of these estimators is assessed in a simulation study. The use of the projected TMLE estimator is illustrated in a secondary data analysis for the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) trial where effect modifiers are subject to missing at random.

Highlights

In social and health sciences, research questions often involve systematic comparison of the effectiveness of different longitudinal exposures or treatment strategies on an outcome of interest
This paper aims to fill these gaps in the literature by establishing the efficiency theory for CHA-Marginal Structural Models (MSMs) and providing substitution-based, semi-parametric efficient and robust estimators
We studied causal effect modification by a counterfactual modifier

Summary

Introduction

In social and health sciences, research questions often involve systematic comparison of the effectiveness of different longitudinal exposures or treatment strategies on an outcome of interest. Consider a study where individuals are followed over time. In addition to their baseline covariates, we record their time-varying treatments, time-varying covariates, and the outcomes of interest. Time-varying confounding is ubiquitous in these settings: the treatment of interest depends on past covariates and in turn affects future covariates. Marginal Structural Models (MSMs), introduced by [4], are well-established and widely used tools to address the problem of time-varying confounding. These models estimate the marginal expectation of an intervention-specific counterfactual outcome, i.e. the mean outcome of a subject in an ideal experiment where he/she was assigned to follow a given intervention

Objectives

Results

Conclusion