An extension of multistate hazards models for transitions and reverse transitions

摘要: The multistate hazards model, where more than one intercommunicating states are taken into consideration to explain the transition from one state to another over time, provide more realistic interpretations of the data from a prospective study. A multistate hazard model for transition and reverse transition for longitudinal data is illustrated in this paper. Simple test procedures for the parameters and the different transition probabilities of the proposed model is also calculated. This proposed model is flexible enough to include unequal number of covariates in different transitions and provide us more realistic interpretations of the data. The potential risk factors at the transitions and reverse transitions can be identified from this model, where the results can be compared in relative terms. As an example the proposed model is applied to the longitudinal data from Bangladesh Institute of Research and Rehabilitation in Diabetes, Endocrine and Metabolic Disorders (BIRDEM). The results based on proposed models, revealed some important features of the patients in terms of their status regarding the disease which could not be obtained by using the existing models.