top of page
Search

Relative Risk vs Hazard Ratio for MRCP Part 1: Key Differences Explained

TL;DR

For MRCP Part 1, relative risk (RR) and hazard ratio (HR) are not interchangeable. Relative risk compares the probability of an event between two groups, while a hazard ratio compares the rate of an event over time using survival analysis. If time-to-event, censoring, or Kaplan–Meier curves are mentioned, the correct interpretation almost always involves a hazard ratio.


Why this topic matters for MRCP Part 1

Biostatistics questions in MRCP Part 1 are designed to test interpretation, not arithmetic. Candidates commonly lose marks by confusing measures of risk with measures of rate. Relative risk and hazard ratios appear frequently in questions on clinical trials, cohort studies, and survival analysis. Understanding what each statistic represents—and when it should be used—is essential for accurate exam answers and for real-world appraisal of medical literature.

This article supports revision from the official MRCP Part 1 syllabus and aligns with how statistics are tested in College questions rather than how they are presented in textbooks.


Core definitions (exam-focused)

Relative Risk (RR)

Relative risk compares the probability of an outcome in an exposed group with that in a control group.

Relative Risk=Risk in exposed groupRisk in control group\text{Relative Risk} = \frac{\text{Risk in exposed group}}{\text{Risk in control group}}Relative Risk=Risk in control groupRisk in exposed group​

What it tells you: “How much more or less likely is the event in the exposed group?”

Typical use:

  • Randomised controlled trials

  • Prospective cohort studies

  • Binary outcomes (event vs no event)

Hazard Ratio (HR)

A hazard ratio compares the instantaneous event rate over time between two groups. It is derived from survival analysis, most commonly using a Cox proportional hazards model.

What it tells you:“At any given point in time, how quickly is the event occurring in one group compared with another?”

Typical use:

  • Time-to-event data

  • Survival outcomes

  • Studies with censoring or variable follow-up


Relative Risk vs Hazard Ratio: high-yield comparison

Feature

Relative Risk

Hazard Ratio

Type of data

Binary outcome

Time-to-event

Considers time

❌ No

✅ Yes

Handles censoring

❌ No

✅ Yes

Typical analysis

Risk comparison

Survival analysis

Common exam clue

“Risk of event”

“Event over time”, KM curve

Value = 1

No difference

No difference

Exam tip: If the question stem mentions follow-up duration, Kaplan–Meier curves, or censoring, relative risk is almost certainly the wrong answer.


The 5 most tested subtopics in MRCP Part 1

1. Interpretation of values

  • RR or HR = 1 → no difference between groups

  • RR or HR >1 → increased risk or rate

  • RR or HR <1 → protective effect

The College often asks for the most accurate interpretation, not a vague statement of “benefit”.

2. Confidence intervals

For both RR and HR, a 95% confidence interval that crosses 1 indicates the result is not statistically significant.

Example: HR 0.82 (95% CI 0.60–1.12) → not significant

3. Survival curves and follow-up

Relative risk ignores when events occur. Two treatments may have the same cumulative risk at 5 years but very different early event rates—this difference is captured by the hazard ratio.

4. Censoring

Patients who are lost to follow-up or who do not experience the event by study end are censored. Hazard ratios account for this; relative risk does not.

5. Proportional hazards assumption

Hazard ratios assume that the ratio of hazards between groups remains constant over time. If survival curves cross, this assumption may be violated—an important subtlety sometimes hinted at in exam stems.


Mini MRCP-style question

A cohort study follows patients with atrial fibrillation for 4 years. Stroke-free survival is analysed using a Cox proportional hazards model. The hazard ratio for stroke with a new anticoagulant is 0.70 (95% CI 0.52–0.94). Which interpretation is correct?

A. The absolute risk of stroke is reduced by 30%B. The probability of stroke is 30% lower in the treatment groupC. At any given time, the stroke rate is 30% lower with the new drugD. The number needed to treat is 3E. The result is not statistically significant

Correct answer: C

Explanation: A hazard ratio refers to the rate of events over time. The confidence interval does not cross 1, so the result is statistically significant.


MRCP Part 1 biostatistics study setup with clinical trial graphs

Common exam traps (know these)

  • Treating hazard ratio as a probability

  • Confusing relative risk with absolute risk reduction

  • Ignoring time-to-event wording in the stem

  • Forgetting the significance of censoring

  • Over-interpreting small numerical differences without checking confidence intervals


Practical study checklist for MRCP Part 1

Use this rapid checklist during revision and in the exam:

  1. Is time explicitly mentioned? → think hazard ratio

  2. Is the outcome purely binary? → think relative risk

  3. Does the CI cross 1? → not statistically significant

  4. Is there a Kaplan–Meier curve? → interpret HR, not RR

  5. Are follow-up times unequal? → RR is unreliable

Practising these patterns repeatedly is far more effective than memorising formulas.


FAQs

Is hazard ratio the same as relative risk?

No. Relative risk compares probabilities, whereas hazard ratio compares event rates over time.

Can hazard ratios be used without survival data?

No. Hazard ratios require time-to-event data and usually involve censoring.

Does MRCP Part 1 require calculations of RR or HR?

No. The exam focuses on interpretation and appropriate usage, not mathematical derivation.

What does a hazard ratio of 0.5 mean clinically?

At any given moment, the event rate in the treatment group is half that of the control group.


Ready to start?

Ready to lock in biostatistics for MRCP Part 1 the exam-smart way?👉 Practise high-yield statistics questions with detailed explanations in our curated MRCP question bank:https://crackmedicine.com/qbank/

Then simulate real exam pressure with full-length practice papers here:👉 https://crackmedicine.com/mock-tests/


Sources

 
 
 

Comments

bottom of page