Case-Control Study

A case-control study is an observational study design that starts from outcome status: it compares a group of people who have a disease or outcome (cases) with a group who do not (controls), looking backward to compare how often each group was exposed to a suspected risk factor. Because it samples on the outcome rather than the exposure, it is efficient for rare diseases and is a foundational design in epidemiology.

Definition

A case-control study identifies subjects by outcome status (diseased = cases, non-diseased = controls) and then ascertains and compares prior exposure between the two groups, estimating the odds ratio as the measure of association.

Scope

The entry covers the logic of outcome-based sampling, the selection of cases and controls, the odds ratio as the design's natural measure of association, and the principal biases (selection and recall) that the design is prone to. It treats the case-control study as a methodological topic within epidemiologic study designs, not as clinical guidance.

Key concepts

• Outcome-based (retrospective) sampling
• Cases and controls
• Odds ratio
• Selection bias
• Recall bias
• Matching and confounding control
• Rare-disease assumption

Mechanisms

Subjects are enrolled on the basis of outcome status, then their history of exposure is reconstructed and compared. The design's natural effect measure is the odds ratio — the odds of exposure among cases divided by the odds of exposure among controls — which, when the outcome is rare, approximates the risk ratio that a cohort study would yield (the rare-disease assumption). Jerome Cornfield gave the formal argument that the exposure odds ratio estimates the disease odds ratio, making the design quantitatively interpretable. Controls must be sampled from the same source population that produced the cases so that they represent the exposure distribution of the population at risk.

Clinical relevance

Case-control studies provide much of the observational evidence behind known risk factors for disease, and reading them critically is part of evidence appraisal in the health sciences. They are a reference design for understanding how exposure-outcome associations are estimated; they describe how evidence is generated and are not a basis for individual diagnostic or treatment decisions.

Epidemiology

The design is especially suited to rare outcomes and to diseases with long latency, where a cohort study would require very large samples or long follow-up. The landmark application is the Doll and Hill (1950) study linking smoking to lung cancer, which compared lung-cancer patients with non-cancer hospital controls and helped establish the design as a tool of modern epidemiology.

History

Outcome-based comparison has roots in nineteenth-century medicine, but the modern case-control study was consolidated in mid-twentieth-century cancer epidemiology. Doll and Hill's 1950 investigation of smoking and lung cancer is the canonical early example, and Cornfield's 1951 paper supplied the statistical justification for estimating relative risk from the odds ratio. Breslow and Day's 1980 monograph then codified the analytic methods, and the design became a workhorse of chronic-disease epidemiology.

Debates

How should controls be selected?

Valid inference depends on controls representing the exposure distribution of the population that generated the cases; hospital, population, and neighbourhood controls each carry different selection-bias risks, and the choice remains a central methodological judgement.

How much does recall bias threaten validity?

Because exposure is ascertained after the outcome is known, cases may recall or report past exposures differently from controls, a bias the design cannot fully eliminate and that must be addressed by design and analysis.

Key figures

• Richard Doll
• Austin Bradford Hill
• Jerome Cornfield
• Kenneth Schulz
• David Grimes

Seminal works

• doll-hill-1950
• cornfield-1951
• breslow-day-1980
• schulz-grimes-2002

Frequently asked questions

Why does a case-control study report an odds ratio instead of a risk ratio?

Because subjects are sampled by outcome status, the design cannot directly estimate incidence or absolute risk; the odds ratio is the measure it can estimate, and it approximates the risk ratio when the outcome is rare.

What is the main weakness of a case-control study?

It is vulnerable to selection bias (in how controls are chosen) and recall bias (in how past exposure is reported), because outcome status is already known when exposure is measured.