Abstract:Path-specific effects are a broad class of mediated effects from an exposure to an outcome via one or more causal pathways with respect to some subset of intermediate variables. The majority of the literature concerning estimation of mediated effects has focused on parametric models with stringent assumptions regarding unmeasured confounding. We consider semiparametric inference of a path-specific effect when these assumptions are relaxed. In particular, we develop a suite of semiparametric estimators for the effect along a pathway through a mediator, but not some exposure-induced confounder of that mediator. These estimators have different robustness properties, as each depends on different parts of the observed data likelihood. One of our estimators may be viewed as combining the others, because it is locally semiparametric efficient and multiply robust. The latter property is illustrated in a simulation study. We apply our methodology to an HIV study, in which we estimate the effect comparing two drug treatments on a patient's average log CD4 count mediated by the patient's level of adherence, but not by previous experience of toxicity, which is clearly affected by which treatment the patient is assigned to, and may confound the effect of the patient's level of adherence on their virologic outcome.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to estimate the path - specific effect in the presence of confounding of mediator variables and outcome variables. Specifically, the author focuses on how to conduct semi - parametric inference after relaxing the strict assumptions about unmeasured confounding factors. The paper develops a set of semi - parametric estimators for estimating the path effect generated through a certain mediator variable (rather than the confounding factors of some mediator variables caused by exposure). These estimators have different robustness properties, and one of the estimators is particularly prominent for its local semi - parametric efficiency and multiple robustness. ### Specific problem description 1. **Background and motivation**: - Traditionally, research on mediation effects has mainly focused on parametric models and has strict assumptions about unmeasured confounding factors. - This study aims to develop a method to estimate the effect through a specific path when these assumptions are relaxed, especially in the presence of confounding of mediator variables and outcome variables. 2. **Research objectives**: - Develop a set of semi - parametric estimators to estimate the path effect \( P_{AMY} \), that is, the effect from exposure \( A \) to outcome \( Y \) through mediator variable \( M \), excluding the influence of confounding factor \( C_1 \) caused by exposure. - Provide a multiple - robust estimator that can maintain consistency even when some model assumptions do not hold. 3. **Application example**: - The paper takes HIV research as an example and compares the effects of two antiretroviral treatment regimens on the average log CD4 count of patients. The mediator variable is the patient's compliance, without considering previous toxicity experiences (which are obviously affected by the assigned treatment and may confound the effect of compliance on virological outcomes). ### Main contributions - **Theoretical development**: Proposed a non - parametric identification formula for path - specific effects and derived its effective influence function. - **Estimation methods**: Developed multiple semi - parametric estimators, each depending on different parts of the observed data likelihood function and having different robustness properties. - **Multiple robustness**: Proposed a multiple - robust estimator that combines other estimators and can maintain consistency and asymptotic normality even when some model assumptions do not hold. Through these methods, the paper provides new tools and theoretical support for path - effect estimation in complex causal structures.

On semiparametric estimation of a path-specific effect in the presence of mediator-outcome confounding

Nonparametric Estimation of Path-specific Effects in Presence of Nonignorable Missing Covariates

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

Mediation pathway selection with unmeasured mediator-outcome confounding

Causal mediation analysis with mediator values below an assay limit

Identification and estimation of mediational effects of longitudinal modified treatment policies

Efficient and flexible causal mediation with time-varying mediators, treatments, and confounders

A multistate approach for mediation analysis in the presence of semi-competing risks with application in cancer survival disparities

Nonparametric estimators of interventional (transported) direct and indirect effects that accommodate multiple mediators and multiple intermediate confounders

Quantifying an Adherence Path-Specific Effect of Antiretroviral Therapy in the Nigeria PEPFAR Program

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

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

Semiparametric causal mediation analysis in cluster-randomized experiments

Bayesian Sparse Mediation Analysis with Targeted Penalization of Natural Indirect Effects

Identification, Semiparametric Efficiency, and Quadruply Robust Estimation in Mediation Analysis with Treatment-Induced Confounding

Robust Inference for Mediated Effects in Partially Linear Models

Heterogeneous interventional indirect effects with multiple mediators: non-parametric and semi-parametric approaches

Efficiently transporting causal (in)direct effects to new populations under intermediate confounding and with multiple mediators

A Semiparametric Approach to Model Effect Modification

Methods of analysis for survival outcomes with time-updated mediators, with application to longitudinal disease registry data

A Bayesian nonparametric approach for causal mediation with a post-treatment confounder