Longitudinal Mediation Analysis Using Natural Effect Models

Murthy N. Mittinty, Stijn Vansteelandt

Research output: Contribution to journalArticlepeer-review

12 Citations (Scopus)


Mediation analysis is concerned with the decomposition of the total effect of an exposure on an outcome into the indirect effect, through a given mediator, and the remaining direct effect. This is ideally done using longitudinal measurements of the mediator, which capture the mediator process more finely. However, longitudinal measurements pose challenges for mediation analysis, because the mediators and outcomes measured at a given time point can act as confounders for the association between mediators and outcomes at a later time point; these confounders are themselves affected by the prior exposure and outcome. Such posttreatment confounding cannot be dealt with using standard methods (e.g., generalized estimating equations). Analysis is further complicated by the need for so-called cross-world counterfactuals to decompose the total effect. This work addresses these challenges. In particular, we introduce so-called natural effect models, which parameterize the direct and indirect effect of a baseline exposure with respect to a longitudinal mediator and outcome. These can be viewed as a generalization of marginal structural mean models to enable effect decomposition. We introduce inverse probability weighting techniques for fitting these models, adjusting for (measured) time-varying confounding of the mediator-outcome association. Application of this methodology uses data from the Millennium Cohort Study, a longitudinal study of children born in the United Kingdom between September 2000 and January 2002.

Original languageEnglish
Pages (from-to)1427-1435
Number of pages9
Issue number11
Early online date27 May 2020
Publication statusPublished - Nov 2020
Externally publishedYes


  • Counterfactual
  • Decomposition
  • Longitudinal mediation
  • Mediation


Dive into the research topics of 'Longitudinal Mediation Analysis Using Natural Effect Models'. Together they form a unique fingerprint.

Cite this