Case-Control and Cohort Studies in Outbreak Investigation

Once descriptive epidemiology has generated a hypothesis about the source of an outbreak, that hypothesis is tested with an analytic study. In outbreak settings the two workhorse designs are the retrospective cohort study, used when the population at risk is well defined and enumerable, and the case-control study, used when it is not. Both compare exposure between people who became ill and those who did not, to identify the likely vehicle or source.

Definition

In outbreak investigation, a retrospective cohort study compares disease attack rates across exposure groups within a defined at-risk population, while a case-control study compares prior exposure between ill (case) and well (control) people drawn from the same population, each aiming to identify the exposure most strongly associated with illness.

Scope

This topic explains how the standard analytic designs are adapted to the speed and constraints of a field investigation: choosing between a cohort and a case-control approach, computing attack rates and relative risks or odds ratios, and interpreting the resulting associations to point at a source. It treats these as methodological tools within outbreak investigation and assumes familiarity with the general designs covered elsewhere; it is not statistical instruction or operational guidance.

Core questions

• Is the population at risk defined and enumerable, favouring a cohort design, or not, favouring a case-control design?
• Which exposure is most strongly associated with becoming a case?
• What is the appropriate measure of association, a relative risk or an odds ratio, for the design chosen?
• Could confounding, selection, or recall problems explain the observed association?

Key concepts

• Retrospective cohort study
• Case-control study
• Defined population at risk
• Attack rate
• Relative risk and odds ratio
• Control selection from the source population
• Recall and selection bias in outbreaks

Mechanisms

When the population at risk is enumerable, for example everyone who attended a wedding, a retrospective cohort study tallies who was exposed to each item and computes attack rates and relative risks, identifying the exposure with both a high attack rate among the exposed and a strong relative risk. When the population is open or unknown, a case-control study instead compares the exposure histories of ill people with those of well controls sampled from the same source population, estimating odds ratios. Both designs depend on a clear case definition, careful exposure ascertainment, and attention to confounding and to the biases, especially recall and selection, that observational designs carry.

Clinical relevance

The associations these studies produce direct the control measures of an outbreak, such as recalling a contaminated product or closing a source of exposure. For health professionals, understanding the design helps in interpreting why a particular vehicle was implicated and how strong that evidence is. This entry describes how outbreak hypotheses are tested and is not guidance for managing an individual case.

Epidemiology

Retrospective cohort studies are typical of point-source events with a defined guest or attendee list, such as foodborne outbreaks, whereas case-control studies dominate when cases arise from a dispersed or undefined population. In the 2003 Hong Kong SARS outbreak, detailed comparison of exposed and unexposed groups within an affected hospital and housing estate helped characterise who was at risk and how transmission occurred, illustrating analytic comparison in an emergency.

History

The case-control and cohort designs were consolidated in mid-twentieth-century chronic-disease epidemiology, but their rapid retrospective application to acute outbreaks became a staple of field epidemiology as training programmes such as the Epidemic Intelligence Service spread the method. Field-epidemiology manuals codified the choice between cohort and case-control approaches according to whether the population at risk can be enumerated.

Debates

Cohort or case-control design in a field investigation?

When the population at risk is small and fully enumerable a retrospective cohort study is preferred because it yields attack rates and relative risks directly; when it is large or undefined a case-control study is more practical, and choosing well under time pressure is a recurring methodological judgement.

Key figures

• Michael Gregg
• Kenneth Rothman
• Sander Greenland

Seminal works

• gregg-2008
• cdc-fetp-2012

Frequently asked questions

When is a retrospective cohort study used instead of a case-control study in an outbreak?

A retrospective cohort study is used when the whole population at risk can be listed and contacted, such as attendees of a single event, so that attack rates and relative risks can be computed; a case-control study is used when the population at risk is large or undefined and cannot be fully enumerated.

Why is the odds ratio used in outbreak case-control studies?

Because a case-control study samples on illness rather than exposure, it cannot directly estimate attack rates, so the odds ratio is the measure it can compute, approximating the relative risk when the outcome is uncommon.