Design of Epidemiologic Studies

Experimental studies have active exposure assigned by the researchers. An important component to this active assignment is randomization. Randomization uses chance to assign a particular person their exposure. This helps control a bias called confounding. Another feature of experiments that helps control another type of bias is blinding. Clinical trials are experimental studies that study therapeutic interventions. They usually include randomization and double blinding. Clinical trials are conducted on patients whereas field trialsField trials are conducted on individuals in the population. Field trials tend to be very large studies of a common or very serious disease. Community intervention trials randomize communities rather than individuals. They are used when an exposure cannot be assigned on an individual basis. Cluster randomized trials are like community intervention trials except they randomize groups smaller than an entire community. In observational studies, the researcher passively determines what exposure has been assigned by the subjects themselves or by nature. Ethically, observational studies can investigate exposures that cannot be assigned by the researcher. There are several types of observational studies. Cohort studies are longitudinal studies in which the temporal sequenceTemporal sequence between exposure and disease is observed. In cohort studies, the researcher assigns participants into groups according to their exposure and then, observes occurrence of disease. They can be prospective in which the researcher lives through the temporal sequence, or they can be retrospective in which the researcher is a historian. Case-control studies are longitudinal studies in which the researcher assigns participants into groups according to newly diagnosed disease and then determines frequency of exposures in the past. This is a design created by epidemiologists to study rare diseases without the necessity of having large studies. One of the most difficult features of case-control studies is selection of the source of controls. Cross-sectional studies are ones in which the temporal sequenceTemporal sequence between exposure and disease cannot be ascertained. They are often done on exposures that are ongoing like gender.