[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/) › Study designs: how to tell them apart

# Study designs: how to tell them apart

- Concept

Three questions name almost any design: who assigned the exposure, is there a comparison group, and where in time does the study start?

Updated Nov 5, 2026

## Key points

- If the investigator assigned the exposure, it is experimental (a clinical trial); if not, it is observational.
- With no comparison group the study is descriptive (case report or series, descriptive ecological study).
- Cohort: start from exposure and follow forward to the outcome; measures incidence and the relative risk (RR).
- Case-control: start from the outcome and look back at exposure; measures the odds ratio (OR).
- Cross-sectional: exposure and outcome are measured at the same time; measures prevalence and cannot show which came first.
- Ecological: the unit of analysis is the group (municipality, country), not the person.

## What it is

A study design is the plan for whom to study, what to measure and in what order. Observational studies only watch what already happens; in experimental studies the investigator assigns the intervention. Descriptive studies count cases without a comparison group, while analytic studies compare groups to test a hypothesis. Recognising the design is the first step to knowing which measure of association to use and which biases to look for.

## Worked example

- Vignette: 800 bus drivers without hypertension are recruited, 400 on night shifts and 400 on day shifts, and followed for 5 years.
- Question 1: did the investigator assign the shift? No, it already existed: the study is observational.
- Question 2: is there a comparison group? Yes, day shift: it is analytic.
- Question 3: where does it start? From the exposure (shift) towards a future outcome: a prospective cohort.
- Data: 60 of 400 night-shift and 40 of 400 day-shift drivers develop hypertension.
- Risks: 60/400 = 0.15 (15%) and 40/400 = 0.10 (10%).
- RR = 0.15 / 0.10 = 1.5.
- Reading: in this cohort, night shifts were associated with 1.5 times the 5-year risk of hypertension.

## Designs side by side

| Design | Starting point | Direction in time | Main measure |
| --- | --- | --- | --- |
| Case report or series | Patients with the outcome | None (description) | Counts and percentages |
| Ecological | Population groups | Usually one point in time | Correlation between rates |
| Cross-sectional | A sample of a population | One point in time | Prevalence, prevalence ratio |
| Case-control | The outcome (cases and controls) | Backwards, to exposure | Odds ratio (OR) |
| Cohort | The exposure (exposed and unexposed) | Forwards, to the outcome | Incidence, relative risk (RR) |
| Randomised controlled trial | An intervention assigned at random | Forwards | RR, absolute risk reduction, NNT |

## Markers

- RR
- OR
- RCT

## In practice

- In an exam, first find who decided the exposure, then look for a comparison group, and last check the direction in time.
- A cohort can be retrospective (historical): exposure and outcome have already happened, but the analysis still runs from exposure to outcome.
- A design is named by how participants were selected, not by the date the data were analysed.

## Common pitfalls

- Mixing up case-control and retrospective cohort: case-control selects by outcome; a cohort selects by exposure.
- Calling any study with a control group experimental: without assignment by the investigator it is still observational.
- Computing an RR in a case-control study: there the investigator fixes the number of cases, so there is no incidence.

## Memory aid

Who chose, compared with whom, and which way in time: three questions, one design.

## Test yourself

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

## Researchers enrol 100 patients with a rare cancer and 200 without it and ask about past exposures. What design is this?

A case-control study: it starts from the outcome and looks back at exposure.

## What separates an experimental study from an observational one?

In an experimental study the investigator assigns the exposure or intervention; in an observational study they only record what happens.

## Why is a cross-sectional study weak for showing causation?

Because it measures exposure and outcome at the same time and does not show which came first.

## Which measure of association goes with a cohort study?

The relative risk (or the rate ratio), because a cohort measures incidence in the exposed and unexposed.

[Add to my cards (4)](https://durgical.app/estudiar/?add=/en/epidemiology/study-designs-how-to-tell-them-apart/)

## Connections

## Related 6

- [Case-control study](https://durgical.com/en/epidemiology/case-control-study/)
- [Cohort study](https://durgical.com/en/epidemiology/cohort-study/)
- [Cross-sectional study](https://durgical.com/en/epidemiology/cross-sectional-study/)
- [Randomised controlled trial](https://durgical.com/en/epidemiology/randomised-controlled-trial/)
- [Ecological study](https://durgical.com/en/epidemiology/ecological-study/)
- [Levels of evidence](https://durgical.com/en/epidemiology/levels-of-evidence/)

[← 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 1 - Section 6](https://archive.cdc.gov/www_cdc_gov/csels/dsepd/ss1978/lesson1/section6.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/study-designs-how-to-tell-them-apart/
License: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Cite as: DurgiPedia (durgical.com), https://durgical.com/en/epidemiology/study-designs-how-to-tell-them-apart/