Clinical research has taught us vital public health lessons over the years. We now know that smoking is a significant risk factor for emphysema, being overweight can lead to pre-diabetes, and drinking a glass of red wine with dinner offers benefits beyond winding down after a tough day.
While randomized clinical trials (RCTs) are considered the “gold-standard” for medical research, there are phenomena that simply cannot be studied using this method. For example, consider a study involving adolescents on the effects of vaping on pulmonary function. It would be unethical for a researcher to assign study participants to groups using vaping products when it is suspected they can damage the respiratory system.
How then can a researcher determine whether vaping is a risk factor for pulmonary problems? Enter the “observational study” design – an efficient and practical way to study exposures and risk factors in relation to health and other outcomes.
Observational studies are “non-interventional” – meaning that the researcher merely observes and measures variables on naturally assembled cohorts, without any manipulation. There is no random assignment and no clinical intervention other than the existing standard of care.
There are three major types of observational studies:
Cross-sectional studies estimate the prevalence of a disease or other outcome, and enable researchers to study multiple exposures and outcomes at the same time. Survey research typically employs a cross-sectional design. These studies can be conducted relatively quickly. They also tend to be inexpensive relative to other research
A considerable limitation of cross-sectional research is its inability to determine the temporal sequence of events. In other words, while associations among study variables can be explored, cause and effect cannot be established. Nonetheless, this study design can lay the groundwork for the identification of exposures and outcomes that are related, and these associations can be more rigorously explored in a cohort study, and in some instances, an experimental design.
Case-control studies involve data that are already in existence. Medical records, imaging, and laboratory reports are just a few examples of the type of data used within this design. The researcher first identifies a group with the disease or outcome of interest (cases) and a group of persons without the disease under study (controls). For example, to study an outbreak of foodborne illness (Listeria, L. monocytogenes), a researcher would first identify persons with the illness and collect demographic information. They would then identify healthy, demographically comparable individuals in the same area, and collect and compare retrospective data to identify predictors of Listeria. This study approach was applied to a multi-state Listeria outbreak in 2011, in which the consumption of contaminated whole cantaloupes from a specific produce supplier was deemed the culprit.
A major challenge of case-control study design is the difficulty in determining an adequate control group for comparison to cases. In addition, the retrospective nature of the design is susceptible to a number of types of bias. Despite this, when a disease outbreak occurs, requiring immediate study and intervention, a case-control study is the “go to” approach among epidemiologists and other public health professionals.
Cohort studies are considered the most powerful of the three types of observational
designs. Prospective cohort studies, in particular, are unparalleled for estimating the incidence and natural history of a disease. Risk factors or predictors are closely examined in relation to specified outcomes. One or more cohorts are assembled. Participants in the cohort must be free of the outcome under study. For example, to study factors related to the development of Type II diabetes, individuals must be
free of the disease at study onset. Data are collected on all variables thought to be related to diabetes on a regular basis at predetermined times. Since events are measured in temporal sequence, the researcher can identify predictors or risk factors for the study outcome, and cause-effect can, at the very least, be inferred.
Prospective cohort studies can take a long time to complete and be costly. However, they are hands-down more definitive than cross-sectional and case–control designs. Perhaps most importantly, this design can circumvent the ethical concerns that arise when a scientific inquiry involves exposures thought to be harmful.
Observational Research Training
Learning more about observational research designs will help researchers design studies to best answer their research questions. CITI Program’s new Observational Research Protocols: An Introduction course covers the basics of observational research and how it is commonly used. It also dives deep into the three major types of observational studies mentioned earlier. Finally, the course concludes with a module designed to help researchers apply their knowledge and design their own observational research study. The course is filled with examples, helpful graphics, and case studies to illustrate key points and help explain concepts.