A prospective cohort study measures exposures in a sample of individuals (a cohort) and then follows the cohort for a period of time; disease outcomes are monitored.
The following figure illustrates a basic prospective cohort study design. In the beginning, the investigator assembles a well-defined study population and measures each individual’s exposure characteristics. The study population is subsequently followed up over a period of time, typically years, during which time a fraction will develop the disease of interest. The investigators then compare the occurrence of disease in groups with different degrees of exposure within the cohort, thereby determining if the exposure is a risk factor for the disease.
To examine if dietary fiber intake was associated with the risk of coronary heart disease, Rimm et al (JAMA. 1996 Feb 14;275(6):447-51) first identified a cohort of 43,757 healthy men and measured their fiber intake using dietary questionnaires. During the six years of follow-up, 734 of them developed heart disease. Comparing the risk of heart disease of those who consumed high dosages of dietary fiber (exposed group, median value: 28.9 g/day) to those who consumed little of it (unexposed group, median value: 12.4 g/day), an inverse association between fiber intake and fatal heart disease was observed (risk ratio: 0.45). The result suggested that high fiber intake may be protective against coronary heart disease.
Exposures in a cohort study can be assigned into multiple categories (none, a little, moderate amounts, lots). Furthermore, exposure data on each individual can be measured repeatedly through the follow-up period. There is usually less opportunity for bias in prospective cohort studies than in case-control studies because exposure data are collected during the investigation (not recalled from the past) before the disease has occurred. A prospective cohort study is considered the strongest observational study. Bias in cohort studies still exists, however, and many examples of excellent cohort studies have yielded results that differ from the gold standard of human studies — the randomized clinical trial.
Design of Cohort Studies (Columbia University School of Public Health)
Fundamentals of epidemiology (Univ. of North Carolina, Chapel Hill)
Concept of analyzing longitudinal data (Institute for Clinical and Epidemiologic Research)