Cohort Studies


A cohort is a group of people who share similar characteristics and/or were all exposed to a significant event within a defined period. A cohort study (also referred to as panel study, as commonly done so in anthropology) is a study design where one or more cohorts are followed prospectively or retrospectively. An evaluation is then done with respect to a disease or outcome to determine which participant's risk factors are associated with it. The outcome from participants in each cohort is measured and relationships with specific characteristics determined.[1]


Cohort study is a cross sectional research design that introduces a time element. In a cohort study, the researcher aims to track if and how changes in a dependent variable are affected by one or more independent variables over time. The researcher tries to achieve this goal through the observation of a cohort and the exposures they are subjected to. Subjects with varying exposures to a suspected factor are identified and then observed for the occurrence of certain effects. [2]
Cohort studies are similar to cross sectional experimental design, in which a study group is exposed to a new independent variable and is compared to a control group who does not get exposed. [3] The major difference is that in cohort studies the researcher does not strictly form such groups but rather specifies quasi-experimental and quasi control groups depending on how the results unfold.[4]
Cohort studies became common by 1960s enabled by a growing policy interest on employment and family incomes. As the advances in communications technology and increased knowledge on statistical methods, their growth continues to this day.[5]


Cohort studies are most often conducted prospectively. However, a researcher can conduct a retrospective cohort study, for instance by tracking back the exposures that sick subjects were exposed to compared to healthy subjects of the same cohort. Cohort studies can also be ambidirectional, a researcher can use their data prospectively in order to track short term outcomes and come back to it retrospectively to track long term outcomes.[6]
Cohort studies allow researchers to emulate an experimental model while conducting observational studies. It allows for comparison in certain settings as well. For instance if two similar states decide to adopt different policies on maternal health, a cohort study can help assess the outcomes of the new policies. Cohort studies can utilize both quantitative and qualitative data, or both.[7]
Through cohort studies, researchers can identify incidents and national history of a disease. Analysis of a cohort study uses the ratio of either the risk or rate of disease in the exposed cohort compared with the rate or risk in the unexposed cohort. If follow-up times differ markedly between participants, a rate may be more appropriate. The risk ratio uses as a denominator the entire group recruited at the start of the study while the rate ratio uses as a denominator the person years, which takes account of losses to follow-up.[8]

Cohort Studies in Context

Even though cohort study design has become a cornerstone of public health research, its utilization in anthropology remains relatively low. Gravlee et al found out that the number of papers published using cohort studies remain relatively low in the field of anthropology compared to other social science fields. [9] Unlike other social sciences that can be very static in their analysis at times, anthropology’s main concern is analyzing continuity and change. Hence, cohort study design is well suited for anthropological research. In the subfield of medical anthropology, cohort study method is even more relevant, as it is often utilized in medical and public health research. The use of panel data would slightly change the way research is done in cultural anthropology, as it would require systematic comparisons to be done on the factors (such as attributes of people, households, etc.) being studied.
In anthropological research, the most significant challenge of researchers has been utilizing data produced by ethnographic research in a cohort study design. When developing instruments for ethnographic work, it is essential to have a balance between participant driven procedures and the analytic needs of hypothesis testing. Anthropologists need to develop methods that are capable of examining the impact of culture on human health. Anthropology has largely failed to do this, and needs to develop ethnographically based methods that are capable of ruling out alternative explanations. The combination of panel data and conventional ethnographic methods can provide new insights in confirming the value of an anthropological perspective. [10]

A schematic overview of cohort study design.
A schematic overview of cohort study design.

Examples of Use in Medical Anthropology

Cebu Longitudinal Health and Nutrition Survey (CLHNS) [11] is an illuminating example of how cohort study methods could be applied in medical anthropology research. The CLHNS was administered to study infant feeding patterns, the overall sequencing of feeding events, and how feeding patterns affect the infant, mother and household. These topics were aimed to be studied within as natural a setting as possible. Many factors were to be analyzed; including how infant-feeding decisions interacted with social, economic and environmental factors to affect health, nutritional, demographic and economic outcomes. Surveys were administered in the second to third trimester of pregnancy, after birth, and then every 2 months for 24 months. Subsequent follow-up surveys took place in 1991-92, 1994-95, 1998-99, 2002 and 2005. There were different focuses in each period of surveying. In the prenatal period, the focuses were the social, environmental, demographic, health services and nutritional factors that influence birth outcomes. In the first 2 years of life, the focus was on infant feeding, morbidity, and the mother's health. In adulthood, the focus was on factors that predict academic achievement and the development of cardiovascular disease risk. More than 125 scholarly works based on the CLHNS have been published. Many of the early publications used longitudinal structural models to develop inferences about health determinants.
There are some weaknesses of this type of data collection. Although the data was collected often, it was not conducted annually, which leaves gaps for the study of important developmental years. One important challenge for longitudinal studies is to maintain comparability of measures over time. In addition, it is equally important to maintain comparability for additional topics studied and measures that are timely and age appropriate such as more technologically advanced instruments.
The study has been taking place for 25 years. It has provided repeated measures of maternal and child health data over this period. The length of the study and the level of detail and quality of the data are significant strengths of this study. The addition biomarkers to what was originally an interview-based health survey expands the utility of the study to address. The transition of the survey from infectious to chronic diseases expands the utility of the survey. The original children being surveyed are now becoming a parent, which offers the study to observe multiple generations of aging.
Another example is Ryan Brown’s work in Appalachia. [12] Brown incorporated ideas and tools from ethnographic work to utilize in a population study that assesses the role of culture in human behavior and biology. The goal of the six-year long study was to create an instrument that would collect data in a personally meaningful way, yet would also produce quantitative data amenable to epidemiological analysis and population-level generalization.
This study describes lessons learned in the administration of an ethnographically based survey to assess the life course perspectives of Appalachian youth, the Life Trajectory Interview for Youth (LTI-Y). The LTI-Y is then used to hopefully protect the mental health of the youth. In a sample of 319 youth, the association between depressive symptoms and life course barriers and milestones was observed. It was found that the ethnographically based scales of life course barriers and milestones were associated with unique variance in depressive symptoms. The LTI-Y is specifically adapted to the life goals and concerns of Appalachian youths during the transition to adulthood. It asks participants to perform a variety of sorting, ranking, and response tasks regarding life course milestones (get married, finish high school), socio-emotional resources (social and emotional properties considered important for life course success), material goods (fancy car, nice clothes), and life course barriers (going to jail, dropping out of school).
During early focus group sessions, the researchers had decided that 'the things that help you achieve major life goals' and 'the things that make you happy and satisfied in life' were two separate domains of life. But, during the individual free-listing tasks, the participants listed largely overlapping items for each category. In other words, our participants were stating that the same things that make them happy and satisfied in life also help them to attain highly valued life goals. In conclusion, for the Appalachian youth, the means were also the ends. This research was a significant contribution to learning about culture.

Cohort Studies Made Easy

1)Define the cohort(s) based on particular characteristics and exposures and/or interventions you want to investigate.
2)Reach out to the members of the cohort(s) of your choice and get consent.
3)Conduct pre-exposure measurements for the cohort.
4)Wait for time to pass for exposure to show its effect. Track the level of exposure and relevant changes over time through measurements at regular intervals for the same population (cohort.)
5)Conduct post-exposure measurements on your cohort.
6)Use measure of exposure to build quasi-experimental and quasi control groups.
7)Compare the groups based on the change in the effects of exposure.


  • Cohort studies are very useful in tracking change instead of providing a static picture of affairs. Data can be collected at regular intervals so recall error is reduced.
  • They can account for change within and between individuals.
  • They are very useful in conducting comparative and observational studies.
  • They help to establish the direction and magnitude of causal relationships.
  • They also have the capability of improving the accuracy of measurements.
  • Prospective cohort studies are considered to yield most reliable results in observational epidemiology.
  • They provide the ability to calculate incidence rates, relative risks and confidence intervals. Cohort studies can help illuminate multiple outcomes of the same exposure.


  • Attrition is a major concern, especially in larger group sizes and might end up weakening the data or introduce bias to it.
  • Cohort studies take a long follow up time to generate useful data. Follow up can be excessively complicated and expensive.
  • Confounding factors might be difficult to control for. Cohort studies do not allow for the kind of rigorous control associated with randomized control trials.
  • Selection bias is built into cohort studies (the sample is not random) and therefore generalizability of results might not always be possible, at least not with full confidence.
  • Where testing is utilized, test conditions might not be similar for all the cohorts involved and/or might be altered between the pre and post-tests.

Multimedia Resources

University of Maryland Associate Professor Terry Shaneyfelt talks about cohort design, its advantages and limitations

An overview of the Panel Study of Income Dynamics (PSID), which has followed American families and measured aspects of economic and demographic behavior for more than 40 years:

An introductory video for the Cohort Study “Real Choices, Real Lives” which tracks the lives of young girls from developing nations in conjunction with UN Development Goals

Further Reading

Grimes, David A., and Kenneth F. Schulz. "Bias and causal associations in observational research." Lancet 359.9302 (2002): 248-252.
Meirik, O. "Cohort and case-control studies." (2008).
Willett, Walter C., and Graham A. Colditz. "Approaches for conducting large cohort studies." Epidemiologic reviews 20.1 (1998): 91-99.
Manolio, Teri A., Joan E. Bailey-Wilson, and Francis S. Collins. "Genes, environment and the value of prospective cohort studies." Nature Reviews Genetics 7.10 (2006): 812-820.
Koro-Ljungberg, Mirka, and Regina Bussing. "Methodological Modifications in a Longitudinal Qualitative Research Design." Field Methods (2013).
Jones, Eric C., et al. "Design for data quality in a multisite cross-sectional and panel study." Field Methods 22.3 (2010): 250-269.

Works Cited

  1. ^
  2. ^ Johnson, Janet and Reynolds, Ht. “Political Science Research Methods.” CQ Press. 7th Ed. (2011) pp, 152-8
  3. ^
  4. ^ Wilson, MJ and Miller PC. “A Dictionary of Social Science Methods.” Wiley. 1st Ed (1983) pp. 19-20
  5. ^ Gravlee, Clarence C., et al. "Methods for collecting panel data: What can cultural anthropology learn from other disciplines?." Journal of Anthropological Research (2009): 453-483.
  6. ^ Grimes, David and Schulz, Kenneth. "Cohort studies: marching towards outcomes." The Lancet 359.9303 (2002): 341-345.
  7. ^ Johnson, Janet and Reynolds, Ht. “Political Science Research Methods.” CQ Press. 7th Ed. (2011) pp, 152-8
  8. ^
  9. ^ Gravlee, Clarence C., et al. "Methods for collecting panel data: What can cultural anthropology learn from other disciplines?." Journal of Anthropological Research (2009): 453-483.
  10. ^ Brown, Ryan A., et al. "Moving from ethnography to epidemiology: Lessons learned in Appalachia." Annals of human biology 36.3 (2009): 248-260.
  11. ^ Adair, Linda S., et al. "Cohort profile: the Cebu longitudinal health and nutrition survey." International journal of epidemiology 40.3 (2011): 619-625.
  12. ^ Brown, Ryan A., et al. "Moving from ethnography to epidemiology: Lessons learned in Appalachia." Annals of human biology 36.3 (2009): 248-260.