B0909
Title: A competing risks analysis with cause-specific cure
Authors: Eni Musta - University of Amsterdam (Netherlands) [presenting]
Tijn Jacobs - Leiden University (Netherlands)
Marta Fiocco - Leiden University (Netherlands)
Abstract: In survival analysis, a competing risk is an event whose occurrence precludes the occurrence of the event of interest and needs to be accounted for in the estimation procedure. Standard methods to deal with competing risks assume that all subjects are susceptible to both events and only a few recent papers try to accommodate the possibility of being immune ('cured') to one of the risks or all of them simultaneously. A general model with two competing events and a cause-specific cure is considered for each event. Then, the previously mentioned settings considered in the literature become particular cases of this general model. The research was mainly motivated by the question: can we identify the relation between the two cure statuses without making any a priori restrictions? A logistic model is considered for the cure probabilities and a semiparametric Cox model for the cause-specific hazards. First, quantities which can be identified from the data and under what assumptions are discussed. In addition, an estimation procedure is proposed based on the EM algorithm and both asymptotic and finite sample performance of the method is investigated. The approach is illustrated through an application to consumer loan data for which the competing events are default and prepayment.