Causal Mediation Analysis With Time-Varying and Multiple Mediators

Abstract

The assessment of direct and indirect effects with time-varying and multiple mediators is a common but challenging problem, and standard mediation analysis approaches are generally not applicable in this context. This dissertation focused on extending mediation analysis into a setting with time-varying and multiple mediators. An interventional approach has been used to define and identify the direct and indirect effects as well as path specific effects based in a causal inference framework, propose a parametric approach to estimate these effects, and provide an algorithm as well as corresponding software for practical application. In the first paper, we develop a parametric estimation approach to the mediational g-formula, including a feasible algorithm implemented in a freely available SAS macro. In the Framingham Heart Study data, we apply this method to estimate the interventional analogues of natural direct and indirect effects of smoking behaviors sustained over a 10-year period on blood pressure when considering weight change as a time-varying mediator. Compared with non-smoking, smoking 20 cigarettes per day for 10 years was estimated to increase blood pressure by 1.18 (95 % CI: -0.68, 2.69) mm-Hg. The direct effect was estimated to increase blood pressure by 1.52 (95 % CI: -0.25, 2.90) mm-Hg, and the indirect effect was -0.34 (95 % CI: -0.52, -0.13) mm-Hg, which is negative because smoking leads to lower weight which leads to lower blood pressure. These results provide evidence that weight change in fact partially conceals the detrimental effects of cigarette smoking on blood pressure. Our work represents the first application of the parametric mediational g-formula in an epidemiologic cohort study. The second paper proposes an approach to conduct mediation analysis for survival data with time-varying exposures, mediators, and confounders. We identify the direct and indirect effects through a survival mediational g-formula and provide the required assumptions. We also provide a feasible parametric approach along with an algorithm and software to estimate these effects. We apply this method to analyze the Framingham Heart Study data to investigate the causal mechanism of smoking on mortality. The risk ratio of smoking 30 cigarettes per day for ten years compared with no smoking on mortality is 2.34 (95 % CI = (1.44, 3.70)). Of the effect, 7.91 % is mediated by coronary artery disease. The survival mediational g-formula constitutes a powerful tool for conducting mediation analysis with longitudinal data. Finally, the third paper further proposes a method, defining a randomly interventional analogue of path-specific effect, which can always be non-parametrically identified under assumptions of no unmeasured confounding. This method also allows settings with mediators dependent on each other, interaction, and mediator-outcome confounders which are affected by exposure. In addition, under linearity and no-interaction, our method has the same form of traditional path analysis for path-specific effect. Furthermore, under a single mediator without a mediator-outcome confounder affected by exposure, it also has the same form of the results of causal mediation analysis. We also provide SAS code for settings of linear regression with exposure-mediator interaction and perform analysis in Framingham Heart Study dataset, investigating the mechanism of smoking on systolic blood pressure mediated by both cholesterol and body weight. Allowing decomposition of total effect into several analogues of path-specific effects, our method contributes to the investigation of complicated causal mechanisms in settings with multiple mediators.