Semiparametric analysis of correlated recurrent and terminal events.

In many studies, survival data involve several types of failure. This is commonly referred as competing risk data. In other situations, failures or events can recur on the same subject. My research is focused on providing new methodologies to analyze data in which these complicated situations occur. Different approaches have been proposed to model competing risk data. One is to model directly the cause-specific hazard function. Fine and Gray (1999) suggested modeling the type-specific subdistribution (or cumulative incidence) function, which has the advantage of directly assessing covariate effects on this function. We compare and contrast the two modeling strategies using Cox type models, and argue that the cause-specific hazard models provide a simpler and clearer picture of the situation, and at least in the examples we presented seem to provide a better approach to competing risk data, analysis. In clinical and observational studies, an event can recur on the same subject, and in a more complicated situation, there exist a terminal event (e.g. death), which stops the recurrent event process. In many instances, the terminal event is strongly correlated with the recurrent event process. I propose a semiparametric method to jointly model the recurrent and terminal event processes. The dependence is modeled by a shared gamma frailty that is included in both the recurrent event rate and terminal event hazard function. Marginal models are used to estimate the regression effects on the terminal and recurrent event processes. First we propose a Poisson model to estimate the dispersion of the frailty variable. A sandwich estimator is used to achieve additional robustness. Then, we relax the Poisson assumption, and extend the analysis to model multiple types of recurrent events and terminal events. An analysis of hospitalization data for patients in the peritoneal dialysis study is presented to illustrate the proposed method.Ph.D.Biological SciencesBiostatisticsUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/126334/2/3238123.pd