Marginal Partial Likelihood Approach in the Cox Model with Non-ignorable Missing Covariates

Marginal partial likelihood approach is used for estimating the parameters for the Cox model with missing covariates and a non-ignorable missing data mechanism. An efficient algorithm based on Markov chain Monte Carlo stochastic approximation is proposed to solve the resulting estimating equations. Simulation studies show that the proposed estimation procedure works well and gives accurate estimates and their variance estimates. We also illustrate the method with a melanoma data set.