Online Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 460 - Causal Methods for Discovery, Confirmation and Mechanistic Evaluation
Type: Contributed
Date/Time: Thursday, August 6, 2020 : 10:00 AM to 2:00 PM
Sponsor: Section on Statistics in Epidemiology
Abstract #313317
Title: Dynamic Semiparametric Regression Models for Continuous Time Causal Mediation Analysis
Author(s): Jeffery M Albert* and Tanujit Dey and Jiayang Sun and Wojbor Woyczynski and Meeyoung Min
Companies: Case Western Reserve University and Harvard Medical School and Brigham and Women's Hospital and USDA and George Mason University and Case Western Reserve University and The University of Utah
Keywords: causal inference; externalizing behavior; longitudinal data; mediation formula; potential outcomes

Causal mediation analysis may be of interest when both the mediator and final outcome are measured repeatedly, but limited work has been done for this situation. Available methods are based on parametric models and thus sensitive to model assumptions. We have developed a more flexible and robust approach to causal mediation analysis for longitudinal data via semiparametric continuous time models. Our approach uses linear mixed-effects spline models for the mediator and the final outcome, and fits the models in sequence. The predicted mediator from the mediator model is used as a covariate in fitting the outcome model. The models allow flexible functions for both the mean and individual response functions, as well as lagged effects of the mediator. An extended mediation formula and sequential ignorability assumption is employed to estimate natural direct and indirect effects, both overall and as a function of time. The method is applied to data from a cohort study to assess attention as a mediator of the effect of prenatal tobacco exposure on externalizing behavior in children.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2020 program