Cohort studies

Abstract

The simplest cohort design is to obtain exposure data at baseline and follow-up individuals to obtain data up to the point when the event of interest occurs. A richer design includes regularly scheduled visits at which data on exposures are updated. The exposures can be either fixed over time (e.g. race), change directly with time (e.g. age and calendar), or change at their own pace (e.g. biological markers). According to the scientific aims of a cohort study, disease occurrence can be measured as an event in person-time, time-to-endpoint of interest, or change in a biomarker repeatedly measured at follow-up visits. Analytical methods include survival analyses to handle censored observations and late entries due to incomplete observation of the development of events and origin, and longitudinal data analyses for the trajectories of markers of disease progression. Stratification, multivariate regression, and causal inference methods are key tools to accomplish comparability among exposed and unexposed groups. Identification of exposures and risk factors for disease provides a basis for prevention strategies. Data from cohort studies can be used to assess the effects of interventions by using data at the individual level to determine individual effectiveness or by comparing occurrence of disease in the population when typically none or only a few are intervened to determine population effectiveness.