Tutorial on Causal Mediation Analysis for Pharmacometricians

Mediation analyses can support biomarker development by quantifying the part of the total treatment effect on clinical outcome that is mediated by the biomarker. Given that pharmacometricians are often involved in the analysis of biomarker data with pharmacokinetic-pharmacodynamic (PK-PD) models, mediation analysis can be a valuable addition to the pharmacometrics toolbox. The generalized nature of causal mediation analysis makes it particularly suited for use in pharmacometrics, where complex models with non-linearities or interaction effects are common. This tutorial covers the concepts of causal mediation analysis and how these might be applied within a pharmacometric context using a simulation-based workflow, including example code and datasets.