Abstract:Causal mediation analysis can improve understanding of the mechanisms underlying epidemiologic associations. However, the utility of natural direct and indirect effect estimation has been limited by the assumption of no confounder of the mediator-outcome relationship that is affected by prior exposure---an assumption frequently violated in practice. We build on recent work that identified alternative estimands that do not require this assumption and propose a flexible and double robust semiparametric targeted minimum loss-based estimator for data-dependent stochastic direct and indirect effects. The proposed method treats the intermediate confounder affected by prior exposure as a time-varying confounder and intervenes stochastically on the mediator using a distribution which conditions on baseline covariates and marginalizes over the intermediate confounder. In addition, we assume the stochastic intervention is given, conditional on observed data, which results in a simpler estimator and weaker identification assumptions. We demonstrate the estimator's finite sample and robustness properties in a simple simulation study. We apply the method to an example from the Moving to Opportunity experiment. In this application, randomization to receive a housing voucher is the treatment/instrument that influenced moving to a low-poverty neighborhood, which is the intermediate confounder. We estimate the data-dependent stochastic direct effect of randomization to the voucher group on adolescent marijuana use not mediated by change in school district and the stochastic indirect effect mediated by change in school district. We find no evidence of mediation. Our estimator is easy to implement in standard statistical software, and we provide annotated R code to further lower implementation barriers.

Estimating Causal Mediation Effects under Correlated Errors

Causal Mediation Analysis in the Context of Clinical Research.

Causal Mediation Analysis in the Multilevel Intervention and Multicomponent Mediator Case

Identifiability and Estimation of Causal Mediation Effects with Missing Data

Non-linear Mediation Analysis with High-dimensional Mediators whose Causal Structure is Unknown

Causal mediation analysis in presence of multiple mediators uncausally related

Mediation pathway selection with unmeasured mediator-outcome confounding

On estimating causal controlled direct and mediator effects for count outcomes without assuming sequential ignorability

On Efficient Inference of Causal Effects with Multiple Mediators

High-dimensional confounding in causal mediation: a comparison study of double machine learning and regularized partial correlation network

Functional Mediation Analysis with an Application to Functional Magnetic Resonance Imaging Data

Causal Mediation Analysis with Hidden Confounders

Causal Mediation Analysis with Sure Outcomes of Random Events Model

Identification, Inference and Sensitivity Analysis for Causal Mediation Effects

Identification Of Causal Mediation Models With An Unobserved Pre-Treatment Confounder

De-confounding causal inference using latent multiple-mediator pathways

General targeted machine learning for modern causal mediation analysis

Robust and Flexible Estimation of Stochastic Mediation Effects: A Proposed Method and Example in a Randomized Trial Setting

Causal Mediation Analysis with Multiple Treatments and Latent Confounders

Clinically Relevant Mediation Analysis using Controlled Indirect Effect

Causal Mediation Analysis with a Three-Dimensional Image Mediator