ABSTRACT
Many studies of chronic diseases, or other processes, involve the selection of individuals whose observed process history satisfies particular conditions. This chapter considers a population of individuals whose processes are independent. It presents a small simulation study to illustrate differences in information from retrospective, cross-sectional, and prospective data on multistate processes sampled subject to different state-dependent selection conditions. Prevalent cohort studies are ones where individuals are required to have experienced onset prior to their time of selection. More realistic disease processes featuring age-dependent Markov intensities and age-dependent selection criteria are routinely employed. The chapter considers a population of individuals whose processes are independent and a situation with an illness-death process. The pros and cons of alternative sampling plans can be investigated with the expressions, or extensions of them when covariates are present. There is often more heterogeneity across processes than can be accounted for by the incorporation of known covariates.
