| For many years, methodological research in biostatistics has been dominated by further development of Cox's (1972) regression model for life tables and of Nelder and Wedderburn's (1972) formulation of generalized linear models. Similarities of these two streams of work traces back to the fact that both have the logistic model as a key source. In both of these areas, the need to address the problems introduced by subject level heterogeneity for data containing recurrent events has provided a major motivation for additional research and has been widely discussed.; Analysis of recurrent event data is an important issue in the assessment of disparities in health experience and health care. In many medical studies, the time to occurrence of some event and its relation to explanatory variables is the primary interest. In particular, the spatial component in a recurrent event distribution can lead to important insights into the local health environment. Many analyses in epidemiological and prognostic studies of event history data require methods that allow for unobserved covariates or "frailties". For clustered data commonly observed in biomedical survival studies, frailty models have become increasingly popular.; There has been much work with Bayesian models, since Gibbs sampling has simplified the analysis. Advances in Bayesian analysis of hierarchical models have been carried over into frailty models. Banerjee et al. (2003) examined single endpoint data using a parametric frailty model for spatially correlated survival data. Sinha (1993) created a Bayesian semiparametric model for multiple event time data.; A discrete time multilevel model, which looks at recurrent events along with the novel incorporation of a spatial component and a Gaussian random walk transition is proposed. Dependence is accounted for between intervals within an individual by incorporating a Gaussian random walk transition. A spatial component is also included to model the effect of individuals' location on visits. |
PDF Full Text Request