Causal Mediation Analysis for Multivariate Longitudinal Data and Survival Outcomes

This study proposes a joint modeling approach to conduct causal mediation analysis that accommodates multivariate longitudinal data, dynamic latent mediator, and survival outcome. First, we introduce a confirmatory factor analysis model to characterize a time-varying latent mediator through multivariate longitudinal observable variables. Then, we establish a growth curve model to describe the linear trajectory of the dynamic latent mediator and simultaneously explore the relationship between the exposure and the mediating process. Finally, we link the mediating process to the survival outcome through a proportional hazards model. In addition, we use the mediation formula approach to assess the natural direct and indirect effects and prove the identifiability of the causal effects under sequential ignorability assumptions. A Bayesian approach incorporating the Markov chain Monte Carlo algorithm is developed to estimate the causal effects efficiently. Simulation studies are conducted to evaluate the empirical performance of the proposed method. An application to the Alzheimer's Disease Neuroimaging Initiative study further confirms the utility of the proposed method.