top of page
Search

Kaplan–Meier Curves Explained for MRCP Part 1

TL;DR

Kaplan–Meier curves are a high-yield survival analysis topic in MRCP Part 1, testing your ability to interpret graphs rather than perform calculations. You must recognise censoring, identify median survival, and understand how curves are compared using the log-rank test. This guide focuses on exam-relevant interpretation, common traps, and a mini-MCQ to help you secure easy marks.


Why this matters

Survival analysis frequently appears in MRCP Part 1 questions embedded within oncology, cardiology, infectious diseases, and clinical trials. Kaplan–Meier (KM) curves are especially popular because they test data interpretation under time pressure, not mathematical skill.

Many candidates lose marks by over-interpreting curve separation, confusing median and mean survival, or misunderstanding censoring. A structured approach to reading these graphs can convert a traditionally weak area into a scoring opportunity.

This article supports the MRCP Part 1 hub👉 https://crackmedicine.com/mrcp-part-1/


Scope: what MRCP expects you to know

You are not expected to calculate survival probabilities. Instead, MRCP focuses on:

  • Understanding what a Kaplan–Meier curve represents

  • Correct interpretation of censoring

  • Reading median survival from the graph

  • Knowing how survival curves are compared statistically

  • Recognising limitations of survival data

Core high-yield points (exam essentials)

1. What a Kaplan–Meier curve shows

A Kaplan–Meier curve plots:

  • X-axis: time

  • Y-axis: probability of survival

Survival begins at 1.0 (100%) and falls in a stepwise fashion as events occur. Each vertical drop corresponds to an event such as death or relapse.

2. Censoring (very commonly tested)

Censoring occurs when:

  • A participant is lost to follow-up

  • A participant withdraws

  • The study ends before the event occurs

Censored individuals are shown as tick marks on the curve and do not cause a step down.

Exam pearl: Kaplan–Meier analysis assumes censored patients have the same future risk as those who remain under observation.

3. Median survival (not mean)

Median survival is the time at which 50% of participants have experienced the event.

It is:

  • Read horizontally from 0.5 on the y-axis

  • Preferred to mean survival because it is less affected by incomplete follow-up

Mean survival is rarely reliable and is not usually tested.

4. Comparing two Kaplan–Meier curves

Visual separation between curves does not automatically imply statistical significance.

The correct statistical test used to compare survival curves is the log-rank test.

If a p-value is provided (e.g. p < 0.05), the difference in survival is statistically significant.

5. Early vs late separation

  • Early separation: suggests early treatment benefit

  • Late separation: suggests delayed benefit

  • Crossing curves: indicates that treatment effect changes over time and may violate assumptions of proportional hazards

6. Relationship to hazard ratio

Kaplan–Meier curves display cumulative survival, whereas hazard ratios describe instantaneous risk over time.

The hazard ratio is derived from a Cox proportional hazards model, not directly from the KM curve.

Related reading:Relative risk vs hazard ratio (MRCP-focused)👉 https://crackmedicine.com/blog/relative-risk-vs-hazard-ratio-mrcp/

7. Confidence intervals

Some KM curves include confidence intervals:

  • Wide intervals suggest imprecision

  • Overlapping intervals do not exclude statistical significance

Do not infer significance without a stated statistical test.

8. Number at risk tables

Often displayed beneath the graph, these tables show how many participants remain under observation at different time points.

Late curve interpretation becomes unreliable when numbers at risk are very small.


Practising MRCP Part 1 MCQs on survival analysis and Kaplan–Meier curves

Kaplan–Meier essentials at a glance

Feature

Exam relevance

Y-axis

Probability of survival

X-axis

Time

Stepwise drops

Events

Tick marks

Censored patients

Median survival

Time at 50% survival

Statistical comparison

Log-rank test


Mini-MCQ (MRCP style)

Question A randomised controlled trial compares Drug X with placebo in advanced cancer. The Kaplan–Meier curve shows earlier separation favouring Drug X. Median survival is 16 months with Drug X and 11 months with placebo. The log-rank p-value is 0.02.

Which statement is correct?

A. Drug X reduces absolute risk by 5 monthsB. Mean survival is 16 months in the treatment groupC. The difference in survival is statistically significantD. The hazard ratio is 0.02E. Censoring caused the curve separation

Correct answer: C

Explanation: A p-value of 0.02 from the log-rank test indicates a statistically significant difference in survival. Median survival difference does not equal absolute risk reduction, and hazard ratios are not the same as p-values.


Five most tested subtopics

  1. Meaning of censoring

  2. Reading median survival correctly

  3. Role of the log-rank test

  4. Relationship between KM curves and hazard ratios

  5. Interpretation of number-at-risk tables


Common pitfalls (exam traps)

  • Confusing median survival with mean survival

  • Assuming visual curve separation equals significance

  • Ignoring censoring marks

  • Over-interpreting late survival with few patients at risk

  • Thinking Kaplan–Meier curves calculate hazard ratios


Practical study-tip checklist

  • Identify censoring marks before reading the question

  • Locate median survival early

  • Look for a stated statistical test

  • Avoid over-interpreting the curve tail

  • Practise graph-based questions regularly

You can practise exam-standard questions here:Free MRCP MCQs 👉 https://crackmedicine.com/qbank/

Or simulate exam conditions:Start a mock test 👉 https://crackmedicine.com/mock-tests/


FAQs (People Also Ask)

What does a Kaplan–Meier curve show?

It shows the probability of survival over time while accounting for censored participants.

How is median survival determined?

It is the time point at which 50% of participants have experienced the event.

Which test compares Kaplan–Meier curves?

The log-rank test is used to compare survival distributions between groups.

Can Kaplan–Meier curves give hazard ratios?

No. Hazard ratios come from Cox proportional hazards models, not directly from KM curves.


Ready to start?

Ready to turn Kaplan–Meier curves into easy marks in MRCP Part 1?


Sources

 
 
 

Comments

bottom of page