Title: Joint model for bivariate zero inflated recurrent event data with terminal event
Authors: Yang-Jin Kim - Sookmyung Women University (Korea, South) [presenting]
Abstract: Multivariate recurrent events arise when a subject has experienced several types of event repeatedly over time. In many situations, a substantial portion of subjects have no events even during a fairly long follow-up time. To reflect such event free group, zero inflated model has been implemented in a content of recurrent event. Similar situation is incorporated as a cure rate model in an ordinary survival analysis and the most commonly used method is a mixture model where two possibilities are simultaneously considered. The motivation comes from Korean long-term care insurance data, where two types of recurrent events occur. The former is a hospitalization and the latter is an out-patient services. Subject can experience two events repeatedly until either death or censoring. However, about 60 percent of participants had not experience these events. A joint model is suggested to model regression models for two recurrent events with zero inflation as well as terminal event. Some simulation studies are performed to evaluate the biases of the suggested models.