Online Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 414 - Making Questions Relevant and Assumptions Realistic: New Strategies for Causal Mediation
Type: Topic Contributed
Date/Time: Wednesday, August 5, 2020 : 1:00 PM to 2:50 PM
Sponsor: Mental Health Statistics Section
Abstract #312379
Title: Variable Selection for Causal Mediation Analysis Using LASSO-Based Methods
Author(s): Yeying Zhu* and Zhaoxin Ye and Donna L Coffman
Companies: University of Waterloo and University of Waterloo and Temple University
Keywords: Covariate balancing; Electronic health record; Marginal structural model; Outcome-adaptive LASSO; Regularization

Unbiased causal mediation effect estimates can be obtained from marginal structural models using inverse probability weighting with appropriate weights. In order to compute weights, treatment and mediator propensity score models need to be fitted first. If the covariates are high-dimensional, parsimonious propensity score models can be developed by regularization methods including LASSO and its variants. Furthermore, in a mediation setup, efficient direct or indirect effect estimators can be obtained by using outcome-adaptive LASSO to select variables for propensity score models by incorporating the outcome information. A simulation study is conducted to assess how different regularization methods can affect the performance of estimated natural direct and indirect effect odds ratios. Our simulation results suggest that bias reduction can be achieved with sufficient covariate balancing and propensity score models regularized by outcome-adaptive LASSO can be used to improve the efficiency of the natural direct effect estimator. The regularization methods are then applied to MIMIC-III database, an ICU database developed by MIT.

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

Back to the full JSM 2020 program