- 15 Downloads
A cohort study is an observational study design in which subjects are usually selected based on their risk factor exposure and followed over time to evaluate whether they develop the outcome of interest (usually disease).
Cohort studies are commonly used in behavioral medicine research to investigate associations in which experimental designs are unethical or too costly. In a cohort design, participants who have not experienced the outcome of interest are selected, usually based on whether or not they have been exposed to the risk factor of interest. Therefore, a cohort study design is efficient when the exposure is relatively rare but the outcome of interest is common. For example, a cohort design was used in a study of the association of prenatal polychlorinated biphenyl (PCB) exposure with behavior issues and cognitive disability (Lai et al. 2002). A cohort design is also appropriate when the exposure is common.
The most common type of cohort study, a prospective cohort study, identifies subjects without the outcome of interest (such as disease-free participants) at the outset of the study and then follows them forward through time to assess their outcome (or disease) status. Because the subjects have not experienced the outcome at the outset of the study, this prospective design is less susceptible to many types of bias compared to other observational study designs, such as case-control studies. Included in the prospective cohort study design are large studies such as the Framingham Study in which participants were selected for logistical reasons. By recruiting a large number of residents from the single community of Framingham, Massachusetts, follow-up was simplified, and investigators were able to study prospectively the associations between multiple risk factors and outcomes among the participants (Dawber et al. 1963).
Not all cohort studies are conducted prospectively. In a retrospective cohort study, both the exposure and outcome may have occurred at the time of the initiation of the study. These retrospective, sometimes called historical, cohort studies are often conducted using data previously collected for other purposes. For example, pregnant women drivers involved in motor vehicle crashes were identified by linking Washington State Patrol records to birth and death certificates (Wolf et al. 1993). The exposure of interest, seat belt use at the time of the crash, was determined using the police reports. Investigators determined pregnancy outcomes (including low birth weight and fetal death) based on the birth and fetal death certificate data. This retrospective cohort study found the risk of a low-birth-weight infant was higher among unrestrained female drivers compared to those wearing a seat belt at the time of the crash.
In a cohort study, investigators must follow both the exposed and unexposed subjects equally carefully to avoid detection bias. If the exposed subjects are followed more closely, then an excess number of outcomes may be detected within the exposed group resulting in an overestimate of the exposure effect. Also, loss to follow-up may result in biased results when that loss is associated with the exposure and outcome.
Usually more expensive and time-consuming than case-control designs; less expensive than an experimental design
Often used when the exposure is rare
Not practical when outcome of interest (disease) is rare or has a long-latency period
Appropriate when studying multiple outcomes
Usually can only address a single risk factor
When information collected prospectively, reduces potential for bias
Can be impacted by loss to follow-up
Can compute incidence and relative risk of outcome directly
Often considered stronger study design compared with case-control studies, but weaker study design compared to randomized trials that investigate analogous associations