Abstract The exposome paradigm aims to characterise the totality of environmental exposures shaping health across the life course, integrating chemical, physical, behavioural, and social domains. While exposome studies have been highly successful in describing complex exposure patterns and mixtures, they often rely on associative analytical frameworks, which can limit the interpretation of results in terms of causal mechanisms and potential intervention targets. Causal mediation analysis offers a natural framework to address these challenges by decomposing total exposure effects into pathway-specific components. However, the diversity of mediation estimands, assumptions, and analytical strategies developed in the causal inference literature may have limited their use in exposome research. This article provides a structured synthesis of modern causal mediation analysis approaches, with a focus on their conceptual foundations and relevance for exposome and life-course epidemiology. We review classical and contemporary mediation frameworks, including controlled, natural, and interventional direct and indirect effects, and discuss their identification assumptions under different causal structures. Particular attention is given to settings encountered in exposome research, such as time-varying exposures, exposure-induced confounding, high-dimensional mediators, and survival outcomes. By clarifying the conceptual landscape of causal mediation analysis and its applicability to exposome research, this work aims to support more interpretable, mechanism-oriented, and causally-informed investigations of how environmental exposures become biologically embodied across the life course.
Lepage et al. (Sat,) studied this question.