This lecture relates to the selection of study design for clinical and epidemiological (as opposed to laboratory) questions. Epidemiological studies divide into those which are purely descriptive and those which are analytical/hypothesis testing. The first step in addressing a research question is therefore to establish whether the question can be expressed in the form of a hypothesis. Hypothesis testing studies generally require a comparison group in addition to the disease group. Epidemiological study designs include:
(i) Cross sectional studies - these collect information at a single time point and may be relevant to either descriptive or hypothesis testing studies. Associations between measures can be estimated, but it is not possible to infer causality as it is impossible to tell which of any pair of variables may have come first i.e. which is cause and which is effect.
(ii) Case control studies - these select subjects on the basis of the outcome of interest and then collect data retrospectively with respect to risk factors/predictors. Selection of an appropriate control group is often the most challenging step. Controls should be subjects who, if they had developed the disease in question, would have been included amongst the cases i.e. they should come from the same catchment area, primary care physician etc. Controls may be matched on an individual level with the cases or may be a random selection of subjects without the disease under study.
(iii) Cohort studies - these select subjects either at random or on the basis of certain potential risk factors and then follow them forward for the development of the disease. Alternatively they may recruit subjects with a disease and then follow them forwards for the development of a particular complication or co-morbidity. Clinical trials are an example of cohort studies in which the investigator assigns the exposure status (e.g. trial drug, placebo) to the individual subject. Cohort - or longitudinal observational studies - are also the basis of drug registries and other pharmacoepidemiological studies.
Disclosure of Interest None Declared