Hello all,

I am trying to find out, whether (and how) I can implement the Cox proportional hazard models into mediation analysis? In my study, I would like to test whether lung function mediated air pollution-induced cardiopulmonary mortality.

First, I estimated Kaplan-Meier survival curves as well as long-rank test to study differences in survival between the two groups (with normal and reduced lung function). Second, I fitted Multivariable Cox proportional hazards (PH) models to determine the association between long-term air pollution exposures (IV) and mortality (DV) – adjusted for potential covariates. The results are expressed as hazard ratios (HR).

Now, I would like to test mediation analysis of air pollution through reduced lung function (mediator) on mortality.

I was trying to use Monte Carlo method with implementing logistic regression, but my reviewers recommended to me to implement Cox models instead of it to keep results consistent.

Can you give me an advice how can I conduct the implementation of Cox PH model in mediation analysis in R?

Thank you in advance for your suggestions.