NNT Calculator (number needed to treat or harm, with 95% CI)

Understanding the NNT calculator — concept and principle

The number needed to treat (NNT) tells you how many people must receive a treatment instead of the comparison for one extra person to avoid a bad outcome. This calculator works it out from the event counts in two groups of a trial, together with absolute and relative risk reduction and 95% confidence intervals. If the treatment does harm, it gives the number needed to harm (NNH).

1 ÷ absolute risk cut: with the default inputs the result is NNT 17 (treat 17 to prevent 1).

How it is calculated

CER = control events ÷ control group size; EER = treatment events ÷ treatment group size

ARR = CER − EER

NNT = 1 ÷ ARR, rounded up to the next whole person (NNH = 1 ÷ |ARR| if ARR is negative)

RRR = ARR ÷ CER; relative risk = EER ÷ CER

95% CI of ARR = ARR ± 1.96 × √(CER(1 − CER)/n₁ + EER(1 − EER)/n₂); the NNT interval is the reciprocal of its limits.

The principle

Relative measures can make small effects sound large: halving a 2% risk and halving a 40% risk both give a 50% relative reduction. The absolute difference shows how many events are actually prevented. Turning it into "how many people to treat for one to benefit" makes trade-offs with cost and side effects easy to discuss. Rounding up is conventional because you cannot treat a fraction of a person.

Worked example

Default inputs: control 60 of 400, treatment 36 of 400.

CER = 60 ÷ 400 = 15%; EER = 36 ÷ 400 = 9%.
ARR = 0.15 − 0.09 = 6%; RRR = 0.06 ÷ 0.15 = 40%; relative risk = 0.6.
NNT = 1 ÷ 0.06 = 16.7 → 17.
SE term = 1.96 × √(0.15 × 0.85 ÷ 400 + 0.09 × 0.91 ÷ 400) = 0.0449 → ARR CI 1.5% to 10.5%.
NNT CI = 1 ÷ 0.105 to 1 ÷ 0.015 = 9.5 to 66 → Benefit across the whole CI.

Each point is the calculator re-run with only “Control group: total people” changed; the gold point is the default example. Change the inputs above and this chart follows your numbers.

Control group: total people	NNT (benefit) or NNH (harm)
100	2
200	5
300	10
400	17
500	34
600	100
800	67

How to read the result

A smaller NNT means a more effective treatment for that outcome and time frame. Always quote the baseline risk and follow-up time: the same treatment has a larger NNT in low-risk people. Compare the NNT with the NNH for side effects to judge the balance.

Common mistakes

Quoting an NNT without its time period.
Using a relative risk reduction from one population with a baseline risk from another without noting it.
Presenting an NNT when the confidence interval includes no effect.

Keep in mind: This is a statistical tool. Treatment decisions should consider the full evidence and individual circumstances with a health professional.

Frequently asked questions

What is a good NNT?

It depends on the outcome, cost and harms; an NNT of 50 can be worthwhile for preventing death with a cheap, safe drug.

Why is the NNT CI sometimes not shown?

When the ARR interval crosses zero, the NNT interval runs through infinity and is hard to express.

What is NNH?

The number needed to harm: how many people must be treated for one extra person to have a harmful outcome.

Sources: Centre for Evidence-Based Medicine, University of Oxford, "Number Needed to Treat (NNT)" · Laupacis A, Sackett DL, Roberts RS, "An assessment of clinically useful measures of the consequences of treatment", N Engl J Med 1988;318:1728–1733

Checked 9 Nov 2026. Centre for Evidence-Based Medicine (University of Oxford), “Number Needed to Treat (NNT)” — ARR = CER − EER, NNT = 1/ARR rounded up, RRR = (CER − EER)/CER; 95% interval by the standard normal-approximation (Wald) method for a difference of proportions.

Your numbers