Epidemiology with its broad spectrum of objectives, scope, and applications in clinical research, medicine, and public health, studies the incidence and prevalence of health outcomes and events to control or prevent those problems. It establishes associations between various risk factors and health events and measures the frequency and strength of the associations to determine the causation and patterns of disease dissemination.
Based on the research question, the approach of an epidemiological study begins with defining and setting the hypothesis. Whether a particular risk factor or agent leads to a fatal disease or if the health care and intervention lead to a greater probability of an improved outcome, these concerns are considered while defining a research question. The next step includes deciding the appropriate study design to test the research hypothesis as it directs the investigation process. Researchers can observe or intervene to an extent but cannot manipulate the associations experimentally.
Depending on the method used to determine if there is an association between exposure and the outcome, the epidemiological study designs are broadly categorized as Observational Study Designs and Interventional Study Designs.
In an interventional or experimental study design, the patients are subjected to intervention by the investigators and assigned to control groups with different conditions of exposure. This enables the researcher to identify the effects, causal links, and association between the outcome and experiments. In cases where an experimental study design is not appropriate to use, observational studies are preferred. Observational studies, also called non-experimental studies consist of simply observing patients in an uncontrolled setting without actively intervening or altering any study aspects. Depending upon the type of observational study, it may be current, retrospective, or prospective.
Various kinds of observational and interventional study designs are described below:
Interesting Science Videos
Observational Study Design
It is mostly used to study the association between the causes, exposure, and effects that are known to be harmful. An investigator cannot intervene and ask a group of smokers to stop smoking and another unexposed group to start smoking to study the pattern of lung cancer in smokers. Observational studies do not influence the choice of exposure in human subjects and evaluate the outcomes in subjects who were either exposed or not exposed to the factors of interest. However, one of the limitations of observational studies is the characteristic differences that might occur in the groups under study. Population in different occupations might be exposed to various occupational hazards, similarly, they might differ in lifestyle, health status, and many more. Due to these frequently immeasurable elements, it is more challenging to determine the impact of a particular exposure being studied.
Descriptive Studies
In descriptive studies, the characteristics of a health problem and its occurrence in a population at a given time are described. With simultaneous recordings of exposure and outcome and no follow-ups, researchers cannot establish causal-effect relationships in this type of study. However, it is ensured that the conclusions made from the obtained results are valid. In epidemiology, public health, and social sciences, these studies are commonly used for generating a hypothesis. This study, in turn, is categorized as individual (prevalence studies and case studies) and population (ecological) based on the unit of study involved.
Ecological studies
In public health research where the unit of study is a group and data at the population level is only available, ecological studies are used. In this context, the exposure and the health outcome should be evident in the groups being studied, therefore, the correlations between the exposure and disease rates are obtained. It is also used in cases where studying the impact of exposures on a disease condition requires large-scale comparisons at the population level. Ecological fallacy refers to a type of confounding that occurs when relationships identified at the group level are assumed to be true for individuals. The application of ecological study designs is commonly observed when geographical comparisons are made, and social class, migrant groups, and time trends concerning diseases are studied.
Cross-sectional studies
With its relatively simple-to-conduct approach, this is the most commonly used study design. The basic foundation of this study design is to study and investigate a section of the total population considering the characteristics in a population are similar. As a type of observational study, it details the exposure and health outcome at a single point in time (cross-sectionally). In this study, the exposure and outcome status are identified at one point in time, therefore it does not establish a cause-effect relationship which is considered as a limitation. The most popular example of cross-sectional studies are surveys that collect data on multiple characteristics simultaneously to explore the associations, assessed by sound hypotheses. It is useful in describing the prevalence of the outcome in a population. These studies are widely used in Genetic epidemiology.
Case-control studies
It is a comparative study of the population with the disease called the case, and the population without the disease called the control group. To study and investigate the exposures in a retrospective approach with follow-ups, cross-sectional studies are used to study diseases with longer latency periods. Cases may sometimes exaggerate their exposures compared to controls while assessing them in retrospect, which is known as recall bias. This is also a limitation of this study design. Moreover, it is important to identify and select suitable control groups as they may affect the study findings resulting from selection bias. Since the study is retrospective, the case-control study design is widely used for assessing prevalence but not incidence.
Cohort studies
A cohort or longitudinal study design is an observational epidemiological study that monitors the development of disease in the exposed and unexposed groups of the population over time. The groups are classified based on the exposure and level of exposure. Unlike case-control studies that assess the prevalence of a disease, this type of study directly calculates the incidence of the disease. As multiple outcomes are recorded simultaneously, there is a low chance of recall bias in cohort studies, however, chances of selection bias are high. It can be time-consuming and expensive to study rare diseases using this study design which is considered as one of the disadvantages. One of the measures of association, relative risk is provided only in cohort studies as it calculates the difference in risk in exposed and unexposed groups. It investigates two or more groups from the time of exposure to outcome or vice-versa. Depending upon this, cohort studies are classified as prospective and retrospective studies.
Prospective Cohort Studies
In a prospective cohort study, the population without the disease is categorized based on whether they have a specific risk factor, then followed over time by the researcher to see if they develop the outcome of interest or not.
Retrospective Cohort Studies
In a retrospective cohort study, subjects are classified based on their exposure to a specific risk factor. Unlike a prospective study, both the exposure and the outcome have already occurred at the time of the study period.
Experimental/ Interventional Study Design
Through the use of randomization, in interventional study designs, population groups are subjected to various conditions by the investigator. Researchers can minimize the influence of bias and ensure that the groups being compared are similar at the start of the study using randomized assignments. This helps to increase the validity of the study results because it reduces the likelihood that any observed differences between groups are due to factors other than the intervention being studied. Randomization is a fundamental principle in experimental design and is essential for drawing accurate conclusions from research studies.
Randomized controlled trials
Randomized clinical trials, also known as randomized control trials (RCT) are considered the benchmark in epidemiological study designs. The principle of this study design is the random allocation of subjects to defined population groups. The subjects are randomly assigned by the investigator to the control group. As randomization prevents confounding and reduces selection bias, it is considered advantageous over observational epidemiological studies. As the experimental and control groups are similar, the experimental group receives the exposure and treatment while the control group does not receive any treatment or an inactive treatment depending on the objective of the study. This also means that the intervention is the only difference between the groups, thereby isolating the effect of outcomes as the result of difference in intervention.
Although considered the gold standard of study design in terms of validity and credibility to assess causality, many researchers do not prefer this approach due to ethical reasons, small sample size, and difficulty in randomizing subjects and locations.
Quasi-experimental study
Also known as the nonrandomized, or pre-post intervention studies in medical-informatics literature, quasi-experimental studies fall in between individual randomized clinical trials, where variables are controlled and observational studies with no control on variables. This study design is used to answer the research questions, test hypotheses, and investigate the cause-effect relationships between the intervention and outcome in the study. It is the experimental study design that evaluates interventions without any randomization. Unlike RCTs, quasi-experimental studies aim to increase validity and generate remarkable conclusions while recognizing practical limitations and ethical concerns. The two main types of quasi-experimental studies are quasi-experimental design with a control group and without a control group, where the indicator is observed before and after an intervention.
References
- Munnangi, S. & Boktor, S.W. (2023) Epidemiology Of Study Design. StatPearls Publishing. Available from: https://www.ncbi.nlm.nih.gov/books/NBK470342/
- https://publications.iarc.fr/_publications/media/download/4112/a11889f12e9991702f16a2ebbc796f1a68c0aafa.pdf
- Belbasis, L., & Bellou, V. (2018). Introduction to Epidemiological Studies. Genetic Epidemiology, 1–6. doi:10.1007/978-1-4939-7868-7_1
- https://minnstate.pressbooks.pub/hgantunez/chapter/study-designs-commonly-used-in-epidemiology
- Rivas-Ruiz, F., Expósito-Ruiz, M., & Domínguez-Almendros, S. (2012). Research designs in clinical epidemiology. Allergologia Et Immunopathologia, 40(2), 117–124. https://doi.org/10.1016/j.aller.2011.11.003
- https://edisciplinas.usp.br/pluginfile.php/2066043/mod_resource/content/1/2016-09-21-StudyDesign.pdf
- https://www.appinio.com/en/blog/market-research/quasi-experimental-design
- Wang, J. & Attia, J. (2010). Study Designs in Epidemiology and Levels of Evidence. American Journal Of Opthalmology. 367-370. DOI.1016/j.ajo.2009.08.001
- Harris, A.D., McGregor, J.C., Perencevich, E.N., Furuno, J.P., Zhu, J., Peterson, D.E. & Finkelstein, J. (2006). The use and interpretation of quasi-experimental studies in medical informatics. J Am Med Inform Assoc. 13(1):16-23. DOI: 10.1197/jamia.M1749
- Maciejewski, M. L. (2018). Quasi-experimental design. Biostatistics & Epidemiology. 1–10. DOI:10.1080/24709360.2018.14