Topics on Analyzing Recurrent Event Data with Sparsely Observed Longitudinal Information

Recurrent event data are quite common in biomedical and epidemiological studies. A significant portion of these data also contain sparsely observed longitudinal information on surrogate markers. Previous studies have shown that popular methods using a Cox model with longitudinal outcomes as time-dependent covariates may lead to biased results, especially when longitudinal outcomes are measured with error. Thus, it is important to incorporate longitudinal information into the analysis properly. To achieve this, we model the correlation between longitudinal and recurrent event processes using latent random effect terms. We then propose a two-stage conditional estimating equation (CEE) approach to model the rate function of recurrent event process conditional on sparsely observed longitudinal information.;Longitudinal data are typically sparse in many applications. Such sparsity may lead to unstable estimates for parameters defining the correlation structures of random effects. Under a joint modeling framework, we propose a second-order analysis (SOA) approach to estimate the correlation parameters by using information from the recurrent event processes. We then develop a three-stage conditional estimating equation (CEE) approach based on the estimated correlation parameters.;The performance of our proposed approaches is evaluated through simulation. We also apply the approaches to analyze cocaine addition data collected by Yale Stress Center (YSC) and by the University of Connecticut Health Center (UCHC). Both data include recurrent event data on cocaine relapse and longitudinal cocaine craving scores.