Estimation of multistate survival model

The longitudinal data analysis generally involves the problems of censoring and repeated observations. The proportional hazards model (Cox, 1972) deals with the modelling of longitudinal data for partially censored data for a single time observation for each individual. Kay (1982) showed an extension of the proportional hazards model for a number of transient states. Islam and Singh (1992) showed the multistate generalization of proportional hazards model. Islam and Singh (1992) did not provide any application in their paper. In this paper, an outline for solving equation for estimating parameters of the multistate survival model (Islam and Singh, 1992) with an application to diabetes data set from a hospital in Bangladesh is discussed. The multistate survival model for partially censored data involves a number of inter communicating and absorbing states and a large number of parameter to be estimated on the basis of interactive solution of simultaneous equations. For estimating parameters, the Interactive Scientific Processor (ISP), a statistical software package, has been used. The multistate survival model estimated in this can be extensively used for solving real life problems.