Marginal Structural Illness-Death Models for Semi-Competing Risks Data

The three-state illness death model has been established as a general approach for regression analysis of semi-competing risks data. In this paper, we apply it to a class of marginal structural models for observational data. We consider two specific such models, the usual Markov illness-death structural model and the general Markov illness-death structural model which incorporates a frailty term. For interpretation purposes, risk contrasts under the structural models are defined. Inference under the usual Markov model can be carried out using estimating equations with inverse probability weighting, while inference under the general Markov model requires a weighted EM algorithm. We study the inference procedures under both models using extensive simulations and apply them to the analysis of mid-life alcohol exposure on late-life cognitive impairment as well as mortality using the Honolulu Asia Aging Study data set. The R codes developed in this work have been implemented in the R package semicmprskcoxmsm that is publicly available on CRAN.