Endogeneity problem in recurrent event data analysis

In this paper, we analyze recurrent event data in the framework of mixed-effects models to control the between and within subject variabilities among observations. We assume that longitudinal variables are informative for the analysis of recurrent data and thus are considered as covariates in the structure of underlying mixed-effects model. Since longitudinal variables are stochastic, they may be correlated with the random effects. This correlation causes biased estimates of regression coefficients. To solve this problem, we propose jointly modelling of longitudinal and recurrent event data in the framework of shared-random effects models. Bayes estimates of model parameters are achieved by the use of Gibbs sampling algorithm. A simulation study is conducted to show the performance of the proposed model