MULTISTATE SURVIVAL MODELS FOR PARTIALLY CENSORED-DATA

Multistate survival models for partially censored data are of great interest for investigating factors attributing to transitions from one state of survival to other states of survival or death. Kay (1982) extended the proportional hazards model for a multistate framework with several transient states. However, in his framework, Kay did not incorporate reverse transitions. This paper discusses an extension of the proportional hazards model for several transient states under a competing risk framework for transitions and reverse transitions. The relationship with the multistate product-limit method is also shown in this paper.