Cohort Design

In epidemiology, a cohort design or cohort study is a nonexperimental study design that involves comparing the occurrence of a disease or condition in two or more groups (or cohorts) of people that differ on a certain characteristic, risk factor, or exposure. The disease, state, or condition under study is often referred to as the outcome, whereas the characteristic, risk factor, or exposure is often referred to as the exposure. A cohort study is one of two principal types of nonexperimental study designs used to study the causes of disease. The other is the case–control design, in which cases of the disease under study are compared with respect to their past exposure with a similar group of individuals who do not have the disease.

Cohort (from the Latin cohors, originally a unit of a Roman legion) is the term used in epidemiology to refer to a group of individuals who share a common characteristic; for example, they may all belong to the same ethnic or age group or be exposed to the same risk factor (e.g., radiation or soil pollution).

The cohort study is a relatively recent innovation. The first cohort studies were used to confirm the link between smoking and lung cancer that had been observed initially in earlier case–control studies. Cohort studies also formed the basis for much of the early progress in understanding occupational diseases. Cohort studies based on data derived from company records and vital records led to the identification of many environmental and occupational risk factors. Several major cohort studies with follow-up that spanned decades have made significant contributions to our understanding of the causes of several common chronic diseases. Examples include the Framing-ham Heart Study, the Tecumseh Community Health Study, the British Doctors Study, and the Nurses’ Health Study.

In a classic cohort study, individuals who are initially free of the disease being researched are enrolled into the study, and individuals are each categorized into one of two groups according to whether they have been exposed to the suspected risk factor. One group, called the exposed group, includes individuals known to have the characteristic or risk factor under study. For instance, in a cohort study of the effect of smoking on lung cancer, the exposed group may consist of known smokers. The second group, the unexposed group, will comprise a comparable group of individuals who are also free of the disease initially but are nonsmokers. Both groups are then followed up for a predetermined period of time or until the occurrence of disease or death. Cases of the disease (lung cancer in this instance) occurring among both groups are identified in the same way for both groups. The number of people diagnosed with the disease in the exposed group is compared with that among the unexposed group to estimate the relative risk of disease due to the exposure or risk factor. This type of design is sometimes called a prospective cohort study.

In a retrospective (historical) cohort study, the researchers use existing records or electronic databases to identify individuals who were exposed at a certain point in the past and then “follow” them up to the present. For instance, to study the effect of exposure to radiation on cancer occurrence among workers in a uranium mine, the researcher may use employee radiation exposure records to categorize workers into those who were exposed to radiation and those who were not at a certain date in the past (e.g., 10 years ago). The medical records of each employee are then searched to identify those employees who were diagnosed with [Page 190]cancer from that date onward. Like prospective cohort designs, the frequency of occurrence of the disease in the exposed group is compared with that within the unexposed group in order to estimate the relative risk of disease due to radiation exposure. When accurate and comprehensive records are available, this approach could save both time and money. But unlike the classic cohort design, in which information is collected prospectively, the researcher employing a retrospective cohort design has little control over the quality and availability of information.

...