Analytical Study Designs

Where descriptive epidemiology describes occurrence of disease (or of its determinants) within a population, the analytical epidemiology aims to gain knowledge on the quality and the amount of influence that determinants have on the occurrence of disease. The usual way to gain this knowledge is by group comparisons. Such a comparison starts from one or more hypotheses about how the determinant may influence occurrence of disease.

For example, the hypothesis may be "people who have eaten home preserved green olives in restaurant X in August 2006 have an increased risk of developing botulism than those who have not eaten such olives".

We can test this hypothesis in an analytical epidemiological study where the risk of developing botulism is studied in 2 comparable populations; one group consists of people that have visited restaurant X in August 2006 and who did eat green home preserved olives. The other group consists of guests of restaurant X in August 2006 that have not eaten those olives. In both groups the risk of developing botulism is measured (by counting botulism cases that occurred in each group within 30 days after visiting the restaurant). Then those two risks are compared to see if they are significantly different.

Observational studies

In the above example of a simple analytical epidemiological study, a traditional cohort study design was chosen. Another group of traditional study designs that belongs to analytical epidemiology are case control studies. Other less traditional analytical study designs include case-case studies and case-cross over design. In each of these analytical studies, observations in one group in the population are compared to another group (also called 'reference group'). Choosing the appropriate reference group is one of the challenging aspects of analytical epidemiology.

The examples above belong to the category of 'observational studies' in analytical epidemiology. In such studies, the investigator observes systematically how exposure and outcome are distributed in the populations, and the comparison of those observations is made.

Experiments

Another category of analytical studies are 'experimental studies', for example in which the investigator is able to randomly assign exposure to individuals from a particular population after which the occurrence of disease is measured in exposed and unexposed groups. Such experiments are called 'randomised controlled trials (RCT)' and are usually considered the gold standard in analytical epidemiology since the amount of bias is usually very limited. However RCT are not an option if the exposure is known to be very dangerous to humans, in which case it would not be ethical to conduct a RCT. In our example above, it is very clear that a RCT would be completely unacceptable (i.e. deciding randomly which guest should eat green home preserved olives, and then to count botulism cases among exposed and unexposed).

Therefore in Field Epidemiology we are usually left with observational study designs, to observe the 'experiments that nature has created for us'. This often creates challenges in finding appropriate comparison groups.

EPIET Lectures:

Case Control and Cohort Studies

Choice of a reference group

Alternative study designs

FEM PAGE CONTRIBUTORS 2007

Editor

Arnold Bosman

Contributors

Vladimir Prikazsky

Arnold Bosman