Study Designs: Case-Control, Cohort, RCTs (MRCP Part 1)
- Crack Medicine

- 3 hours ago
- 3 min read
TL;DR;
For MRCP Part 1, you must rapidly identify study designs from short vignettes and choose the correct measure of association. Case-control studies suit rare diseases and report odds ratios; cohort studies suit rare exposures and report risk; randomised controlled trials (RCTs) test treatment efficacy with the strongest causal inference. This article distils examiner favourites, common traps, and a worked MCQ to help you score reliably.
Why study design questions matter in MRCP Part 1
Epidemiology questions are frequent and time-pressured. Errors usually arise from mixing up directionality (exposure → outcome vs outcome → exposure), confusing odds ratio (OR) with risk ratio (RR), or overlooking bias. Examiners reward candidates who can match the clinical question to the best design and interpret outputs accurately. This post supports the core MRCP Part 1 hub and directs you to targeted practice.
The three core study designs (high-yield overview)
Feature | Case-control | Cohort | Randomised controlled trial |
Direction | Outcome → exposure | Exposure → outcome | Assigned intervention → outcome |
Best use | Rare diseases, long latency | Rare exposures, prognosis | Treatment efficacy, causality |
Measures | Odds ratio | Risk ratio, incidence | RR, ARR, NNT |
Time & cost | Faster, cheaper | Slower, costlier | Most expensive |
Key biases | Recall, selection | Loss to follow-up | Performance, attrition |
Ethics | Observational | Observational | Requires equipoise |
Case-control studies
When to choose: Rare diseases (e.g., mesothelioma), outcomes with long latency. How they work: Identify cases with the outcome, then select controls from the same source population and compare prior exposure. Key output: Odds ratio (approximates RR when disease is rare).Exam traps:
Reporting RR from a case-control study.
Poor control selection (controls must represent the population that produced the cases).
Recall bias when exposure relies on memory.
Exam pearl: Stems mentioning “matched controls” or “retrospective exposure assessment” usually indicate a case-control design.
Cohort studies
When to choose: Studying incidence, prognosis, or effects of rare exposures (e.g., occupational hazards).How they work: Classify participants by exposure and follow for outcomes (prospective or retrospective).Key outputs: Risk ratio, incidence rate ratio, attributable risk. Strengths: Clear temporality; multiple outcomes from one exposure. Limitations: Time, cost, and loss to follow-up bias.
Exam pearl: If incidence or prognosis is central to the question, a cohort design is often correct.

Randomised controlled trials (RCTs)
When to choose: Testing treatment efficacy. Why they’re strongest: Randomisation balances known and unknown confounders. High-yield concepts:
Intention-to-treat analysis (preserves randomisation).
Blinding and allocation concealment (prevents selection bias).
Absolute risk reduction (ARR) and number needed to treat (NNT)—often tested.
Exam pearl: Randomisation reduces confounding but does not eliminate performance or attrition bias.
Most-tested subtopics (revision checklist)
Choosing the correct design for a given clinical question.
OR vs RR—and when OR approximates RR.
Design-specific biases.
Confounding and its control (randomisation, restriction, stratification).
Association vs causation (RCTs provide strongest inference).
Mini-case (exam-style MCQ)
Question: Investigators study a rare cancer. They compare prior asbestos exposure in patients with the cancer to exposure in matched individuals without the cancer. Which design and measure are most appropriate?
Answer: Case-control study; odds ratio.Explanation: The disease is rare and cases are identified first. Incidence cannot be calculated, so OR is the correct measure.
Common pitfalls to avoid (5)
Calling a retrospective cohort a case-control study.
Quoting RR from a case-control design.
Ignoring loss to follow-up in cohorts.
Assuming randomisation removes all bias.
Overstating causality from observational studies.
How to revise efficiently for MRCP Part 1
Memorise one-line indications for each design.
Practise rapid recognition using Free MRCP MCQs: https://crackmedicine.com/qbank/
Drill OR vs RR with small 2×2 tables until automatic.
Simulate timing with mock tests: https://crackmedicine.com/mock-tests/
Revisit the MRCP Part 1 overview hub for integrated revision: https://crackmedicine.com/mrcp-part-1/
FAQs
How do I quickly identify a study design in MRCP Part 1?
Check directionality. Outcome-first suggests case-control; exposure-first suggests cohort; assigned treatment indicates an RCT.
Why is odds ratio used in case-control studies?
Because incidence and risk cannot be calculated without a defined population at risk.
Can cohort studies be retrospective?
Yes. The defining feature is exposure → outcome, not calendar time.
Does randomisation guarantee no bias?
No. It reduces confounding, but performance and attrition bias may remain.
Ready to start?
Consolidate this topic with timed practice on our MRCP Part 1 overview hub and test retention using the Free MRCP MCQs and mock tests.
Sources
MRCP(UK). Sample Questions and Regulations. https://www.mrcpuk.org/mrcpuk-examinations/mrcp-part-1
BMJ Statistics Notes. https://www.bmj.com/specialties/statistics-notes
Oxford Handbook of Medical Statistics. Oxford University Press. https://global.oup.com/academic/product/oxford-handbook-of-medical-statistics-9780199236557



Comments