About the Number Needed to Treat Calculator
The number needed to treat (NNT) is the number of patients who must receive a treatment instead of the comparator for one extra patient to benefit — or, when the treatment increases events, the number needed to harm (NNH). Introduced by Laupacis, Sackett and Roberts in the New England Journal of Medicine in 1988, it is the reciprocal of the absolute risk reduction and is one of the clearest ways to express how much a treatment actually helps.
Enter the event rate in the control group and in the treatment group as percentages, or switch to raw counts from a trial’s results table. You get the NNT or NNH, the absolute risk reduction (ARR), relative risk reduction (RRR), relative risk (RR) and odds ratio (OR). With counts, the calculator also gives a 95% confidence interval for the ARR and NNT using the standard normal approximation.
This is a tool for clinicians, students, journal-club presenters and anyone reading medical research. Remember that an NNT only applies to the population, outcome and follow-up period of the study it came from: a "25% relative reduction" can mean an NNT of 20 in high-risk patients and 400 in low-risk ones.
With the default inputs, the nnt / nnh (rounded up) is 20. Change any value above to recalculate instantly.
How to use the number needed to treat calculator
- 1Choose whether to enter event rates or raw counts from the trial.
- 2Enter the control (placebo or usual care) group result.
- 3Enter the treatment group result for the same outcome and follow-up period.
- 4Read the NNT or NNH, ARR, RRR, relative risk and odds ratio.
- 5With counts, check the 95% confidence interval before drawing conclusions.
Formula and method
The control event rate (CER) and experimental event rate (EER) are the proportions of patients who had the outcome in each group. Their difference is the absolute risk reduction; its reciprocal is the NNT. When the treatment group has more events the difference is negative and its reciprocal (as a positive number) is the number needed to harm. By convention NNT and NNH are rounded up to the next whole patient.
With raw counts, the 95% confidence interval for the ARR uses the Wald normal approximation, ARR ± 1.96 × √[CER(1−CER)/n₁ + EER(1−EER)/n₂], and the NNT interval is the reciprocal of its limits. When the ARR interval crosses zero, the NNT interval runs from a benefit through infinity to a harm (Altman’s convention), meaning the result is not statistically significant.
- CER
- Control group event rate (proportion)
- EER
- Experimental (treatment) group event rate
- ARR
- Absolute risk reduction
- n₁, n₂
- Number of patients in control and treatment groups
Worked examples
Event rate falls from 20% to 15%
ARR = 20% − 15% = 5%, so NNT = 1 ÷ 0.05 = 20. The relative risk is 0.15 ÷ 0.20 = 0.75 (a 25% relative reduction) and the odds ratio is (0.15/0.85) ÷ (0.20/0.80) = 0.706.
Side effect more common on treatment: 10% vs 14%
The treatment group has 4 percentage points more events, so ARR = −4% and the number needed to harm is 1 ÷ 0.04 = 25. Relative risk is 1.4 and the odds ratio 1.465.
Trial counts: 30/200 vs 18/200
CER = 30/200 = 15% and EER = 18/200 = 9%, so ARR = 6% and NNT = 16.7, rounded up to 17. The 95% CI for the ARR (−0.34% to 12.34%) crosses zero, so the NNT interval runs from 8.1 through infinity to an NNH of 292.4 — not statistically significant.
Larger trial: 60/300 vs 30/300
Event rates of 20% and 10% give an ARR of 10% and an NNT of 10. The 95% CI for the ARR is about 4.3% to 15.7%, so the NNT interval is roughly 6.4 to 23.0 — a statistically significant benefit.
Frequently asked questions
What is a good number needed to treat?+
Lower is better: an NNT of 1 would mean every treated patient benefits. What counts as good depends on the outcome, harms and costs — an NNT of 50 can be worthwhile for preventing death but not for a minor symptom.
Why is NNT rounded up?+
Because you cannot treat a fraction of a patient, the convention is to round the NNT up to the next whole number, which slightly understates the benefit rather than overstating it.
What is the difference between absolute and relative risk reduction?+
Absolute risk reduction is the simple difference in event rates (20% to 15% is 5 points). Relative risk reduction expresses that difference as a share of the control rate (5 ÷ 20 = 25%). The same RRR gives very different NNTs at different baseline risks.
What is number needed to harm (NNH)?+
NNH is the number of patients who need to receive a treatment for one extra patient to experience a particular harm, such as a side effect. It is calculated the same way as NNT when the treatment increases the event rate.
Can I compare NNTs from different trials?+
Only with care. NNT depends on the baseline risk of the population, the outcome definition and the length of follow-up, so NNTs are only comparable when those are similar.
For healthcare professionals and education only. This calculator implements a published formula or score but is not a diagnosis and does not replace clinical judgment. Verify inputs and results against the original source before using them in patient care.