Epidemiology | Umweltbundesamt

A disease might thus be described at a particular time in different places (cross-sectional study); alternatively, the development of such a disease in a population group during a specified observation period might be described (cohort study). In both cases, the influencing factors and if possible the exact correlations between causes and disease are investigated.

In the case of sequential analyses, the cohort under observation is not defined on an individual basis; instead, for example, the population of a city or city district is taken as a cohort. During a specified observation period, sequential exposure values (e.g. mean daily particulate concentrations) and effects (e.g. daily fatalities, which are registered continuously and thus being available for large-scale studies without significant costs or time delays) are recorded and set in relation. 

With case-control studies, differences in the frequency and intensity of possible factors of influence are investigated using cases (e.g. sick people) and controls (people free from the illness in question who are otherwise comparable to those affected in terms of age, gender, social status etc.), so that relationships to possible causes may be inferred.

In the case of controlled exposure studies on volunteers or animals, all conditions are kept constant with the exception of the factor of influence under investigation. If the test is properly conducted, any differences observed in effects must be traceable to changes in the factor of influence. However, these tests are complex and cannot be carried without time limits. However, many effects cannot be investigated in this way because of ethical and practical considerations.

Epidemiological studies are carried out on populations in real-life situations. Problems of transfer of results from animals to humans or from selected volunteers to the total population or to particular groups do not arise, nor do those of transferability of effects of higher concentrations to those of lower concentrations, or the effects of short-term exposure at low concentrations to those of long-term exposure. In real life, however, side effects and confounding factors cannot be controlled. It is thus much more difficult than with controlled exposure studies to decide whether a recorded statistical relationship is of a causal nature or not. Epidemiological studies can never prove causality; they can however make statements of probability more likely. The criteria developed by Sir Austin Bradford Hill (the so-called Hill Criteria) in 1965 (e. g. the strength of the statistical relation, specificity of the effects, the existence of multiple observations which concur with one another, dose dependency, and correct temporal sequence of cause and effect) are often applied to describe causality.