Number needed to treat
- Measure
How many people must be treated, over a given time, to prevent one more event than without treatment.
What it is
The number needed to treat (NNT) is the inverse of the absolute risk reduction. It turns a trial result into an intuitive figure: how many patients must receive the intervention, over the study period, for one of them to benefit. A low NNT means a large effect. NNTs can only be compared between studies with the same outcome, the same period and similar populations.
Formula and reading
- NNT = 1 / (CER − EER). With the ARR as a percentage: NNT = 100 / ARR.
- If the CI of the ARR includes 0, the CI of the NNT includes infinity and crosses over to harm.
- From the RRR: NNT = 1 / (RRR × CER).
Worked example
- Data: in a care home, an exercise programme is tested for one year. Falls: 18% in controls and 12% with the programme.
- ARR = 0.18 − 0.12 = 0.06.
- NNT = 1 / 0.06 ≈ 16.7.
- Round up: NNT = 17 over one year.
- Another programme achieves an ARR of 0.025 over one year: NNT = 1 / 0.025 = 40.
- Reading: 17 residents must take part for a year to prevent one fall; the first programme does more than the second.
Markers
- NNT
- 1/ARR
How to read it
- The NNT helps explain benefits to patients in concrete terms.
- It is set beside the number needed to harm (NNH) to weigh benefit against harm.
Common pitfalls
- Calculating the NNT from the relative reduction instead of the absolute one.
- Rounding down, which exaggerates the benefit.
- Comparing NNTs from studies with different periods or outcomes.
Test yourself
Think of the answer, then tap the question to see it.
Control risk 10% and treated risk 6%. What is the NNT?
ARR = 0.04; NNT = 1 / 0.04 = 25.
Why is the NNT rounded up?
To avoid overstating benefit: nobody can treat a fraction of a person, and rounding down overestimates the effect.
What is missing from the statement that the NNT is 50?
The period and the outcome: for example, 50 treated for 5 years to prevent one myocardial infarction.
Connections
Educational content on epidemiological and statistical methods; the examples use made-up numbers to teach the calculation. Not a clinical or legal guide: current Colombian rules are in the official sources.
Sources: Chapter 15: Interpreting results and drawing conclusions (Cochrane, 2024); Chapter 6: Choosing effect measures and computing estimates of effect (Cochrane, 2024).
Spotted a mistake?We are always improving for the community. Tell us what is wrong and our team will review it.Report a mistake
Representative model, for educational purposes only.