[Durgical](https://durgical.com/en/) › [DurgiPedia](https://durgical.com/en/durgipedia/) › [Epidemiology](https://durgical.com/en/epidemiology/) › [Study designs](https://durgical.com/en/epidemiology/areas/designs/) › Case-control study

# Case-control study

- Study design

Start with sick people (cases) and healthy people (controls) and compare how many were exposed earlier. Ideal for rare diseases.

Updated Nov 5, 2026

## Key points

- Selection is by outcome; exposure is looked at backwards.
- Measure: odds ratio, OR = ad/bc; incidence and RR cannot be calculated directly.
- When the disease is rare, the OR approximates the RR.
- Ideal for rare diseases and long latency; many exposures can be studied.
- Controls must come from the same population that gave rise to the cases.
- Typical biases: recall bias and selection of controls.

On this page

- The basics Key points
- What it is
- How it is built
- Worked example
- Strengths and weaknesses
- Markers
- When to use it
- Common pitfalls
- Memory aid

- Study Test yourself

- Connections Connections

## What it is

In a case-control study people are selected by outcome: cases who have the disease and controls who do not, drawn from the same population. Investigators then look back at how many in each group were exposed. Because the investigator fixes the number of cases and controls, incidence cannot be measured: the measure is the odds ratio (OR). It is efficient for rare diseases or those with long latency.

## How it is built

- A clear case definition is set and cases are recruited, ideally incident (newly diagnosed).
- Controls without the disease are chosen from the same source population.
- Each case may be matched with 1 to 4 controls on age, sex or another variable.
- Past exposure is measured the same way in both groups, preferably from records.
- The OR is calculated from the 2×2 table (or the matched OR if matching was used).

## Worked example

- The study includes 150 women with COPD (cases) and 300 without COPD from the same neighbourhood (controls), 2 controls per case.
- Cooked with firewood: 90 of the 150 cases and 75 of the 300 controls.
- Table: a = 90 exposed cases, b = 75 exposed controls, c = 60 unexposed cases, d = 225 unexposed controls.
- Odds of exposure in cases: 90/60 = 1.5; in controls: 75/225 ≈ 0.33.
- OR = (a × d) / (b × c) = (90 × 225) / (75 × 60) = 20 250 / 4 500 = 4.5.
- Reading: the odds of having cooked with firewood are 4.5 times higher in women with COPD than in controls.

## Strengths and weaknesses

| Strengths | Weaknesses |
| --- | --- |
| Efficient for rare diseases | Does not measure incidence or absolute risk |
| Good for long latency | Recall bias when remembering exposures |
| Studies many exposures at once | Choosing good controls is hard |
| Faster and cheaper than a cohort | Poor for rare exposures |

## Markers

- OR
- ad/bc

## When to use it

- It is the usual design in outbreak investigation: what the sick and the well ate is compared.
- Matching controls the matched variable, but its own effect can no longer be studied and a matched analysis is needed.

## Common pitfalls

- Calculating a risk as 90/150: that percentage depends on how many cases the investigator chose to recruit.
- Taking controls from a hospital where the exposure is more common than in the community (Berkson bias).
- Overmatching: matching on a variable tied to the exposure pulls the OR towards 1.

## Memory aid

Case-control starts at the end: first the disease, then the search for the footprints of exposure.

## Test yourself

Think of the answer, then tap the question to see it.

## Why can a case-control study not calculate incidence?

Because the investigator fixes how many sick and healthy people to include; the share of sick people does not reflect the population.

## When does the OR approximate the relative risk?

When the disease is rare in the population (rule of thumb: under 10%).

## Which bias is typical when exposure is measured by asking participants?

Recall bias: cases tend to remember their exposures in more detail than controls.

## Where should controls come from?

From the same population that produced the cases, so that if they had fallen ill they would have been included as cases.

[Add to my cards (4)](https://durgical.app/estudiar/?add=/en/epidemiology/case-control-study/)

## Connections

## Related 6

- [Odds ratio](https://durgical.com/en/epidemiology/odds-ratio/)
- [The 2×2 table](https://durgical.com/en/epidemiology/the-2-2-table/)
- [Nested case-control study](https://durgical.com/en/epidemiology/nested-case-control-study/)
- [Recall bias](https://durgical.com/en/epidemiology/recall-bias/)
- [Selection bias](https://durgical.com/en/epidemiology/selection-bias/)
- [Cohort study](https://durgical.com/en/epidemiology/cohort-study/)

[← All of epidemiology](https://durgical.com/en/epidemiology/)

Educational content on epidemiological and statistical methods; the examples use made-up numbers to teach the calculation. Not a clinical or legal guide: current Colombian rules are in the official sources.

Sources: [Principles of Epidemiology | Lesson 1 - Section 7](https://archive.cdc.gov/www_cdc_gov/csels/dsepd/ss1978/lesson1/section7.html) (Centers for Disease Control and Prevention, 2012); [Principles of Epidemiology | Lesson 3 - Section 5](https://archive.cdc.gov/www_cdc_gov/csels/dsepd/ss1978/lesson3/section5.html) (Centers for Disease Control and Prevention, 2012); [Chapter 24: Including non-randomized studies on intervention effects](https://www.cochrane.org/authors/handbooks-and-manuals/handbook/current/chapter-24) (Cochrane, 2024).

## **Spotted a mistake?** We are always improving for the community. Tell us what is wrong and our team will review it. Report a mistake

Representative model, for educational purposes only.

---
Source: https://durgical.com/en/epidemiology/case-control-study/
License: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Cite as: DurgiPedia (durgical.com), https://durgical.com/en/epidemiology/case-control-study/