The question

Questions of aetiology and risk factors commonly arise in relation to public health. For example:

• Does the evidence support a likely causal effect of a factor (e.g. obesity) on a particular disease (e.g. breast cancer)?

Clearly, in public health terms, you may want to know the whole array of health effects of an exposure, but the evidence for each causal or preventive influence has to be first assessed separately, along the lines we suggest here. Largely, such evidence will come from case-control and cohort studies, although in some instances randomized controlled trials (RCTs) provide critical tests of causal hypotheses.

In a striking recent example, RCTs showed β-carotene to be an ineffective preventive of lung cancer, contrary to deductions made from a large number of observational studies that evaluated diet and laboratory data. This role of RCTs needs to be borne in mind when constructing a search strategy, as indicated below.

Getting the right balance in the question being addressed may be straightforward (e.g. ‘Do oral contraceptives cause breast cancer?’), especially when it derives from a clear clinical or public health question. But should a review of body size and breast cancer include:

• all measures of body size (height, weight, skinfolds, circumferences, derived variables);

• only those that are modifiable (removing height, which is of biological interest); or

• only the most direct estimates of adiposity, such as skinfolds? Such issues usually make systematic reviews of aetiology and risk factors more complex than systematic reviews of interventions.

Study design

Epidemiological studies of a etiology (often called observational studies) treat individual characteristics, personal behaviours, environmental conditions and treatments as ‘exposures’ that may modify risk of disease.