Cohort design is a type of research design where investigators follow subjects over time, tracking their development through a set of health-related metrics [1]. As opposed to experimental investigations where researchers intervene to alter the conditions of studied populations, cohort studies involve no such intervention [2]. Instead, studies begin with identifying subjects to place them into two groups: exposed and non-exposed populations [3]. Over time, these groups are studied to determine the incidence, prevalence, prognosis, and potential causes of the central condition [3].
Depending on what investigators hope to study, cohort studies are either prospective or retrospective. In a prospective cohort study, the cohort is selected and measured for various risk factors and exposures before the outcome occurs [2]. An example of a prospective study would be one where investigators seek to measure the likelihood of psoriasis patients experiencing side-effects due to anesthesia. A group of patients with psoriasis would be identified. These people would be tested to ensure that they do not already have conditions that complicate the administration of anesthesia. Then, the researchers would return to these patients over an extended period, tracking whether they have experienced side-effects during procedures since the start of the experiment.
Conversely, retrospective studies aim to analyze a cohort that has already experienced the given outcome [2]. Experimenters collect all their data from records [1]. They will begin with each patient’s initial exposure and status at baseline and continue to follow up into the future, accumulating further data [1]. One retrospective study tracked the occurrence of pneumonia following the contraction of HIV by 2,628 women [4]. Every six months, the subjects would fill out a survey and provide a blood sample [4].
Although the aforementioned study was retrospective, it was combined with data from a simultaneous prospective study to create a more complete picture of pneumonia incidence in HIV-positive individuals [4]. Neither prospective nor retrospective studies are infallible, so combining them may allow researchers to sidestep individual shortcomings.
Generally, prospective designs are considered more robust sources of valid evidence, but this is not always the case and can often come at a cost [5]. Prospective studies can provide researchers with accurate data collection that accounts for exposures, endpoints, and confounders [5]. But this high level of detail requires a vast investment of time and money. Follow-up periods are typically long, and investigators must follow-up with a large number of people [5]. Additionally, prospective designs run a greater risk of loss to follow-up, leading to lost data and reduced internal validity [5]. Especially in the context of rare diseases, where populations are already small, prospective studies may prove too inefficient and inappropriate for meaningful observation [5].
While retrospective methods offer a more time- and cost-efficient approach to cohort designs, the quality of available data can be a concern in these studies [5]. Because investigators have to work with existing data that may not have taken into account certain risk factors or exposures, these studies are more likely to be affected by confounding variables [2]. Information, recall, and selection biases may also reduce the validity of the results [3].
Despite each approach’s shortcomings, many of these concerns can be avoided if investigators meticulously design their studies with these shortcomings in mind. Researchers believe that either form of cohort design, if designed carefully and thoroughly, can result in generalizable, accurate results with important implications for the study of medicine [5].
References
[1] M. S. Setia, “Methodology Series Module 1: Cohort Studies,” Indian Journal of Methodology, vol. 61, no. 1, p. 21-25, Jan-Feb 2016. [Online]. Available: https://doi.org/10.4103/0019-5154.174011.
[2] I. Oliveira, “Cohort studies: prospective and retrospective designs,” Students 4 Best Evidence via Cochrane, March 2019. [Online]. Available: https://bit.ly/3fN22jE.
[3] X. Wang and M. W. Kattan, “Cohort Studies: Design, Analysis, and Reporting,” Chest Journal, vol. 158, no. 1, p. S72-S78, July 2020. [Online]. Available: https://doi.org/10.1016/j.chest.2020.03.014.
[4] S. R. Cole et al., “Combined analysis of retrospective and prospective occurrences in cohort studies: HIV-1 serostatus and incident pneumonia,” International Journal of Epidemiology, vol. 35, no. 6, p. 1442-1446, Aug 2006. [Online]. Available: https://doi.org/10.1093/ije/dyl176.
[5] A. M. Euser et al., “Cohort Studies: Prospective versus Retrospective,” Nephron Clinical Practice, vol. 113, no. 3, p. c214-c217, Oct 2009. [Online]. Available: https://doi.org/10.1159/000235241.
