Background material for this section can be found in Diggle (2002), Elliott et al. (1992), Elliott et al. (2000) and Coggon et al. (2003).
A popular method of investigating the causes of a disease in a population is to take a random sample from the population and follow them up over time to see which individuals develop the disease. Having done this, one can then estimate the effects of putative covariates on the risk of disease. Such a study is termed a cohort study. The cohort study is a powerful study design when a sizeable proportion of the population are likely to develop the disease in the time-frame of the study. However, for rare diseases the sample size, or study time-frame, required to attain sufficient statistical power to be able to detect any effects may be prohibitive.