Study designs explained: case-control, cohort, cross-sectional and RCT
Three questions sort the main study designs: was the exposure assigned, is there a comparison group, and were people picked by exposure or by outcome. Then name the measure each design gives.
Chapters (12)
Distinguish cross-sectional, case-control, cohort and randomised controlled trial designs by how participants were selected and when data were collected, and identify the measure each yields.
The short version
- Three questions sort the designs: did the researchers assign the exposure, is there a comparison group, and were people picked by exposure, by outcome or by neither.
- Cross-sectional gives prevalence, case-control gives the odds ratio, and cohort studies and trials give incidence and relative risk.
- A rare outcome points to a case-control study; a rare exposure points to a cohort study.
- Retrospective describes when the data were collected, not the design: a retrospective cohort study still starts with exposure.
- Only a trial randomises, and a suspected harmful exposure cannot ethically be assigned.
Three questions that sort the study designs
Three patients come to one clinic in a single month with the same rare liver injury, and all three took the same herbal supplement. Did the supplement cause it? Nobody can say yet. What the researchers do next depends on who they pick first, and when.
Study designs explained in one idea: the design is defined by how people are picked and whether the researchers assigned the exposure. Three questions sort the main designs.[6][2]
- Did the researchers assign the exposure?If yes, it is an experiment. If they only watch, it is observational.[6]
- Is there a comparison group?If not, the study is descriptive.[1][2]
- Were people picked by exposure, by outcome, or by neither?This applies to observational studies with a comparison, and it separates cohort, case-control and cross-sectional designs.[3][4][5]
Put together: an experiment means the researchers assigned the exposure. An observational study without a comparison group is descriptive. An observational study with a comparison is then sorted by how people were picked.[2][6]
Case report and case series
The three patients are a case series: a description of a few patients, with no comparison group. One patient alone would be a case report.[1][2]
A case series is good for spotting an unusual pattern and generating a hypothesis. It cannot show causation.[1]
Cross-sectional study: prevalence
Option one is a cross-sectional study. Survey 2,000 adults today, recording supplement use and liver injury at the same moment, then compare injury rates in users and non-users.[3][9]
It measures prevalence, a snapshot of how common something is. But exposure and outcome are recorded together, so we cannot tell which came first.[3]
Comparing injury rates in users and non-users inside that one-time sample is still possible, but because it is a single time point the study cannot show which came first.[9][3]
Case-control study: odds ratio
Option two is a case-control study. Start with the outcome: 100 patients with liver injury and 200 without. Then look back at who took the supplement.[4]
It suits rare diseases. Because we chose how many cases and controls to recruit, it gives an odds ratio. Recall bias is its weak spot, which the next lesson covers.[4]
It does not give incidence. Cases and controls are chosen by the researchers, which is why the odds ratio is the measure it yields.[4]
Cohort study: incidence and relative risk
Option three is a cohort study. Start with exposure: 5,000 supplement users and 5,000 non-users, followed forward for five years.[5]
We count who develops the disease, so we get incidence and relative risk. Timing is clear, and one exposure can be followed to several outcomes. Losing people during follow-up is its weak spot.[5][9]
Participants are free of the outcome at the start. They are selected by exposure and followed forward, so the order of events is clear.[5]
Prospective vs retrospective cohort study
A cohort study may be prospective, collecting new data, or retrospective, using existing records. Either way, it starts with exposure and follows to outcome.[5]
Retrospective describes when the data were collected. It does not turn a cohort study into a case-control study.[5]
A rare outcome points to a case-control study. A rare exposure points to a cohort study.[4][5]
Randomised controlled trial and the measure each design gives
Option four is a randomised controlled trial: the researchers assign the exposure by chance. Randomisation tends to balance known and unknown confounders across groups, so a well-run trial is among the strongest designs for testing a treatment.[6]
A trial is the one design in which the researchers decide who gets the exposure. In every other design they only observe.[6]
But for a supplement suspected of harm, we cannot ethically assign it. That question stays with cohort and case-control designs.[7][2]
| Design | People picked by | Measure |
|---|---|---|
| Cross-sectional | Neither: one time point | Prevalence |
| Case-control | Outcome | Odds ratio |
| Cohort | Exposure | Incidence and relative risk |
| Randomised controlled trial | Exposure assigned by chance | Incidence and relative risk |
Cross-sectional gives prevalence; case-control gives the odds ratio; cohort studies and trials give incidence and relative risk.[3][4][5][8]
Practice question
Original · USMLE styleResearchers use the electronic health records of a large bakery chain to study whether workplace flour dust exposure is associated with adult-onset asthma. From records covering 2008 to 2018, they identify 600 bakery workers with flour dust exposure documented at the time of hire and 600 workers without such exposure. None of the 1,200 workers had asthma at hire. The researchers then review each worker's record to determine who received a new diagnosis of asthma during the following years.
Common questions
What is the difference between a cohort study and a case-control study?
A cohort study picks people by exposure and follows them to the outcome, giving incidence and relative risk. A case-control study picks people by outcome and looks back at exposure, giving an odds ratio.
Does retrospective mean case-control?
No. Retrospective describes when the data were collected. A retrospective cohort study uses existing records but still starts with exposure and follows to outcome.
Which study design suits a rare disease, and which a rare exposure?
A rare outcome points to a case-control study. A rare exposure points to a cohort study.
What measure does each study design give?
Cross-sectional gives prevalence; case-control gives the odds ratio; cohort studies and trials give incidence and relative risk.
Why can a harmful exposure not be studied with a randomised trial?
Researchers cannot ethically assign an exposure suspected of harm, so that question stays with cohort and case-control designs.
Transcript
Read the full narration, by chapter688 words
0:00Three patients, one supplement: who do we pick first?
Three patients come to one clinic in a single month with the same rare liver injury, and all three took the same herbal supplement. Did the supplement cause it? Honestly, nobody can say yet. So what would we do next? The answer depends on who the researchers pick first, and when. Today: tell the main study designs apart by how people are picked, and name the measure each one gives. Three questions sort the main study designs.
0:27Three questions that sort the main study designs
First: did the researchers assign the exposure? If yes, it is an experiment. If they only watch, it is observational. Second: is there a comparison group? If not, the study is descriptive. Third, for observational studies with a comparison: were people picked by exposure, by outcome, or by neither? Our three patients are a case series:
0:55Case report and case series
a description of a few patients, with no comparison group. One patient alone would be a case report. It is good for spotting an unusual pattern and generating a hypothesis. It cannot show causation. To compare, option one is a cross-sectional
1:12Cross-sectional study: prevalence
study: survey 2,000 adults today, recording supplement use and liver injury at the same moment, then comparing injury rates in users and non-users. It measures prevalence, a snapshot of how common something is. But exposure and outcome are recorded together, so we cannot tell which came first.
1:36Case-control study: odds ratio
Option two is a case-control study. Start with the outcome: 100 patients with liver injury and 200 without. Then look back at who took the supplement. It is quick, cheap and suits rare diseases. We chose how many cases and controls to recruit, so it gives an odds ratio. Recall bias is its weak spot, which we meet next lesson.
2:01Cohort study: incidence and relative risk
Option three is a cohort study. Start with exposure: 5,000 supplement users and 5,000 non-users, followed forward for 5 years. We count who develops the disease, so we get incidence and relative risk. Timing is clear, and one exposure can be followed to several outcomes. Losing people during follow-up is its weak spot.
2:25Prospective vs retrospective cohort
A cohort study may be prospective, collecting new data, or retrospective, using existing records. Either way, it starts with exposure and follows to outcome. Retrospective describes when the data were collected. It does not turn a cohort study into a case-control study. A rare outcome points to a case-control study. A rare exposure points to a cohort study.
2:52Randomised controlled trial
Option four is a randomised controlled trial: the researchers assign the exposure by chance. Randomisation tends to balance known and unknown confounders across groups, so a well-run trial is among the strongest designs for testing a treatment. But for a supplement suspected of harm, we cannot ethically assign it. That question stays with cohort and case-control designs.
3:18The measure each design gives
Measures: cross-sectional gives prevalence; case-control gives the odds ratio; cohort studies and trials give incidence and relative risk.
3:29Practice question
Your turn. Using employee health records from 2008 to 2018, researchers identify 600 bakery workers with documented flour dust exposure at hire, and 600 without. None had asthma at hire. The researchers review the records to see who was later diagnosed with asthma. Which study design is this? A, case series. B, case-control study. C, cohort study. D, cross-sectional study. E, randomised controlled trial. Pause here and decide.
4:08The answer, one option at a time
A is out: a case series has no comparison group, and here there are two. B is out: a case-control study picks people by outcome, and here they were picked by exposure. D is out: a cross-sectional study measures exposure and outcome at one time, not across 10 years of records. E is out: the researchers did not assign the exposure. So the answer is C. Groups were defined by exposure and followed to outcome. Using old records only makes it a retrospective cohort study.
4:44Recap and the next lesson
Three things to keep. One: ask whether the researchers assigned the exposure, whether there is a comparison group, and whether people were picked by exposure or by outcome. Two: cross-sectional gives prevalence; case-control gives the odds ratio; cohort studies and trials give incidence and relative risk. Three: only a trial randomises, and retrospective describes the timing of the data, not the design. Next, bias: selection, information and confounding.
Sources
Numbers in the text, e.g. [1], point here- Writing a Case Report or Case Series. Clin Case Rep 2026. Europe PMC PMC13451630. pmc.ncbi.nlm.nih.gov/articles/PMC13451630/
- BJA Educ 2025, study design overview. Europe PMC PMC11951191. pmc.ncbi.nlm.nih.gov/articles/PMC11951191/
- Cross-sectional studies and prevalence. Europe PMC PMC4885177. pmc.ncbi.nlm.nih.gov/articles/PMC4885177/
- Case-control studies and the odds ratio. Europe PMC PMC4817437. pmc.ncbi.nlm.nih.gov/articles/PMC4817437/
- Cohort studies, prospective and retrospective. Europe PMC PMC4763690. pmc.ncbi.nlm.nih.gov/articles/PMC4763690/
- Randomised controlled trials and observational designs. Europe PMC PMC4966397. pmc.ncbi.nlm.nih.gov/articles/PMC4966397/
- Ethics and randomisation. Europe PMC PMC6302384. pmc.ncbi.nlm.nih.gov/articles/PMC6302384/
- Cochrane Handbook, chapter 15. training.cochrane.org/handbook/current/chapter-15
- STROBE explanation and elaboration (cohort, case-control and cross-sectional studies). Europe PMC PMC2020496. pmc.ncbi.nlm.nih.gov/articles/PMC2020496/
The patient scenario, group sizes and dot grids are illustrative teaching models, not real studies.
The practice question is original and its patient is synthetic. Every fact is checked against the sources above in independent accuracy reviews. Always follow your local guidelines and your own laboratory’s ranges.
Image credits. Illustrations: Servier Medical Art (smart.servier.com), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/), adapted (recoloured, outlined): hospital, three clothed adults, liver, herb pot, hourglass, inflamed bronchus. Icons: Health Icons (healthicons.org), CC0; Document icon by OpenClipart via bioicons.com, CC0. The decision tree, timelines, dot grids, coin, tables, question card, pocket card and all handwriting are drawn in code (original). No AI-generated teaching images.
Corrections. None so far. Spotted an error? Email info@medservatory.com with the lesson code MS-S1-BIOS-005.
For exam study only, not medical advice. Read the medical disclaimer.
In this series
Playlist on YouTube- EP 01Sensitivity vs specificity: the 2×2 table, SnNout and SpPin5:34
- EP 02PPV, NPV and prevalence explained5:27
- EP 03Relative risk vs odds ratio: the 2x2 table explained5:15
- EP 04Number needed to treat: ARR, RRR and NNH from trial data5:00
- EP 05Study designs explained: case-control, cohort, cross-sectional and RCT5:22
- EP 06Types of bias: selection, recall, lead-time and confounding5:30
- EP 07Type I and type II errors: alpha, beta, power and the p-value5:16
- EP 08Confidence intervals and significance: is the null value inside?5:12