top of page
Search

NNT and NNH Calculations Made Easy for MRCP Part 1

TL;DR:

NNT (Number Needed to Treat) and NNH (Number Needed to Harm) are high-yield statistics repeatedly tested in MRCP Part 1. They are simple to calculate once you focus on absolute risk, not relative risk. This article explains the concepts clearly, shows you how to calculate them step by step, highlights common exam traps, and includes an exam-style MCQ.


Why NNT and NNH matter in MRCP Part 1

In MRCP Part 1, statistics are not tested in isolation. NNT and NNH are usually embedded in clinical trials, guideline summaries, or treatment comparison questions. Examiners want to know whether you can:

  • Identify absolute risk correctly

  • Calculate benefit and harm

  • Interpret numbers in a clinical context

Candidates often lose easy marks by confusing relative risk reduction with absolute risk reduction, or by rounding incorrectly.

A solid grasp of NNT and NNH also supports evidence-based medicine, which underpins UK clinical guidance and postgraduate exams.

Core concepts you must know

Absolute risk (AR)

The probability of an event occurring in a group over a defined time period.

Example: 10 strokes in 100 patients over 5 years → absolute risk = 10% (0.10)

Absolute risk reduction (ARR)

The reduction in risk achieved by treatment compared with control.

ARR = AR (control) − AR (treatment)

This is the key value used to calculate NNT.

Absolute risk increase (ARI)

The increase in risk caused by a treatment.

ARI = AR (treatment) − AR (control)

This is used to calculate NNH.

Number Needed to Treat (NNT)

The number of patients who need to receive a treatment to prevent one additional adverse outcome.

NNT = 1 ÷ ARR

  • Always rounded up to the nearest whole number

  • Must include a time frame

Number Needed to Harm (NNH)

The number of patients who need to receive a treatment for one additional harmful outcome to occur.

NNH = 1 ÷ ARI

  • Also rounded up

  • Higher NNH = safer treatment

Medical postgraduate revision setup for MRCP Part 1 statistics and evidence-based medicine

One-table summary (exam essential)

Measure

Formula

Key exam point

Absolute risk

Events ÷ total

Convert % to proportions

ARR

AR(control) − AR(treatment)

Ignore relative risk

ARI

AR(treatment) − AR(control)

Used for harm

NNT

1 ÷ ARR

Always round up

NNH

1 ÷ ARI

Always state time period


Most tested MRCP Part 1 subtopics

  1. Cardiovascular prevention (statins, antihypertensives)

  2. Anticoagulation (stroke prevention vs bleeding risk)

  3. Diabetes treatments (microvascular outcomes)

  4. Oncology trials (survival benefit vs toxicity)

  5. Screening programmes (absolute benefit vs overdiagnosis)

Worked example (step by step)

Scenario In a 5-year trial, myocardial infarction occurred in:

  • 10% of the placebo group

  • 7% of the treatment group

Calculation

  • AR(control) = 0.10

  • AR(treatment) = 0.07

  • ARR = 0.10 − 0.07 = 0.03

  • NNT = 1 ÷ 0.03 = 33.3 → 34

Interpretation34 patients must be treated for 5 years to prevent one myocardial infarction.


Exam-style MCQ

A drug reduces stroke risk from 4% to 2% over one year but increases major bleeding from 1% to 2%.

What are the NNT and NNH?

Answer

  • ARR = 0.04 − 0.02 = 0.02 → NNT = 50

  • ARI = 0.02 − 0.01 = 0.01 → NNH = 100

Key insight: The benefit occurs more frequently than harm.


Common MRCP Part 1 traps (5)

  • Using relative risk reduction instead of absolute risk

  • Forgetting to convert percentages to proportions

  • Rounding NNT or NNH down

  • Ignoring the time period stated in the question

  • Confusing which group represents harm vs benefit


Practical study tips for revision

  • Always write ARR first, then calculate NNT or NNH

  • Practise mixed benefit–harm questions under timed conditions

  • Revise statistics alongside cardiology and public health topics

  • Use question banks early to identify pattern recognition

You can practise high-yield exam questions using the Crack Medicine MRCP Part 1 Qbank:https://crackmedicine.com/qbank/

For exam simulation, timed practice is essential:https://crackmedicine.com/mock-tests/


FAQs

What is a good NNT in exams?

There is no universal “good” NNT. The exam tests calculation and interpretation, not whether the number seems impressive.

Is NNT always rounded up?

Yes. NNT and NNH are always rounded up to the nearest whole number.

Can NNT be negative?

No. A negative value indicates harm, and you should calculate NNH instead.

Do MRCP questions ask about confidence intervals?

Occasionally. You are usually expected to interpret them, not calculate them.


Ready to start?

For systematic practice of statistics questions, revise this topic alongside cardiology and public health using the MRCP Part 1 overview hub /mrcp-part-1/. Pair this article with our sibling post on relative risk vs absolute risk in the blog [/blog/].


Sources

 
 
 

Comments

bottom of page