ABSTRACT

Large prospective cohorts are widely used in epidemiology and other areas of research to investigate associations between putative risk factors and time-to-event outcomes, for example, disease diagnosis or death. To avoid the collection of expensive explanatory variables, such as biological measurements, on the full cohort a substudy can be sampled, within which all variables of interest for a particular investigation are then obtained. There are two main study designs which are used to obtain a sample within a full cohort; the nested case-control design and the case-cohort design. There is a large literature on the design and analysis of these studies; for overviews see Chapters 16 – 18 and (Keogh and Cox (2014), Chapters 7 and 374 8). In both types of study, measures of explanatory variables of interest are obtained for all cases (individuals who have the event) and a subset of the noncases in the full cohort.