Which type of epidemiological study?

Alan J. Silman, Gary J. Macfarlane, and Tatiana Macfarlane

The article below outlines some of the main types of epidemiological studies to help you decide which one is more appropriate for your research.

Assuming there is a scientific hypothesis which is suitably answered by an epidemiological approach, the subsequent decision is which type of study is appropriate. The decision will be based not only on methodological but also on practical considerations. A compromise will need to be made—to undertake a study which can be conducted within the budget and time available and which delivers information which is suitable for answering the hypothesis.

Observational Studies

Cross-sectional study

A cross-sectional study studies a whole population or a random sample of a population and collects information about current disease state (i.e. prevalent disease) and also measures exposures of interest. In some instances, the study may base determination of disease status on a period of time. The principal advantages of the cross-sectional study are normally that it is relatively inexpensive and can be conducted within a short timescale. The principal disadvantage is the lack of information on temporality of disease state and exposures.

The case–control study

In this type of study, persons with the disease of interest are identified (cases) together with a sample of those without the disease (controls). The cases and controls are then compared with respect to their exposure to potential risk factors. The case–control approach is particularly suitable when the disease is rare, and when the aim is to investigate the effect of many exposures on one disease.

The cohort study

A cohort study involves one or more groups of subjects, defined by their exposure status and free of the disease of interest, being followed through time to identify disease onset. The purpose is to determine whether initial exposure status influences risk of subsequent disease. Two particular types of cohort study are the prospective cohort study and the retrospective cohort study. In the prospective approach, cohort(s) are identified by their exposure status presently and are followed-up to determine any future disease onset. The retrospective approach identifies the exposure status of cohort(s) in the past and in a parallel sense they are ‘followed-up’ until the present time, when their disease status is determined. The latter approach will undoubtedly be quicker and less expensive, but may not always be appropriate or possible. The major advantages of a prospective study are that it can be determined which exposure is measured and how.

Randomized controlled trials

All the study designs previously discussed in this article are observational (i.e. they are studying free-living populations who make a choice about their lifestyle, work, and so on). As discussed this means that certain habits tend to cluster in individuals -such as alcohol drinking, smoking- and this means it is difficult to disentangle their individual effects. In randomized controlled trials researchers randomly allocate exposures to individuals. Participants enter a study and then their allocation to an arm of the study is determined randomly, although there may be stratification to ensure certain key characteristics are balanced between studies. What need to be taken into consideration is that persons agreeing to take part in a trial may not be representative of the wider population. It is apparent that many questions can be addressed by several different study designs and thus the decision has to be made as to the most appropriate design to answer a particular question. The choice is often between feasibility, that is, obtaining an answer within a required timescale, and validity (i.e. obtaining the most accurate answer). Common sense dictates that one without the other negates the value of undertaking the study and also that the two are not mutually exclusive. The decision is normally based on an appraisal of both scientific and logistic considerations.

Alan J. Silman is a Professor of Musculoskeletal Health (Botnar Research Institute, Nuffield Department of Orthopaedics, Rheumatology and Musculoskeletal Sciences, University of Oxford).
Gary J. Macfarlane is a Professor of Epidemiology and Deputy Director of the Institute of Applied Health Sciences at the University of Aberdeen.
Tatiana Macfarlane is a Honary Reader at the Institute of Applied Health Sciences (School of Medicine of the University of Aberdeen).

This is an edited extract from the full chapter ‘Which type of epidemiological study?’ from Epidemiological Studies: A Practical Guide (3rd edition.), (OUP, 2018).