Abstract:
|
In longitudinal studies involves assessing the trajectory of ordinal disease outcomes, ignoring the correlation between the disease status and the potentially informative censoring event(s) may lead to bias in the estimation of covariates. To address the problems created by these dependencies, we propose a dynamic joint model of a longitudinal process and survival. This model will include incorporate the previous disease states from the longitudinal data in the survival component. This approach requires an accounting of all informative components of the longitudinal trajectory as it is brought into the survival model. We demonstrate an EM-based estimation procedure as well as a resampling approach to standard error estimation. We will demonstrate this approach using the Hispanic Established Populations for the Epidemiologic Study of the Elderly data, which is a longitudinal survey study to estimate the prevalence of and risk factors for key health conditions in older Mexican Americans.
|