Number Needed to Treat (NNT) Calculator
Enter control and experimental event counts to calculate NNT, NNH, absolute risk reduction, and relative risk reduction with confidence intervals.
The number needed to treat (NNT) is calculated as 1 divided by the absolute risk reduction (ARR), where ARR = control event rate (CER) minus experimental event rate (EER). For example, if CER is 15% (30/200) and EER is 9% (18/200), ARR is 6%, so NNT = 1/0.06 = 16.7, rounded up to 17. This means 17 patients need to be treated to prevent one additional bad outcome.
Free, runs in your browser. Nothing you enter is uploaded.
How to use it
- Enter the number of events in the control group and the total number in the control group.
- Enter the number of events in the experimental group and the total number in the experimental group.
- Read the absolute risk reduction (ARR), number needed to treat (NNT), relative risk reduction (RRR), and 95% confidence intervals.
What you enter
- Number of events in control group
- Total number in control group
- Number of events in experimental group
- Total number in experimental group
What you get
- Number Needed to Treat (NNT)
- Number Needed to Harm (NNH)
- Absolute Risk Reduction (ARR)
- Relative Risk Reduction (RRR)
- Odds Ratio
- 95% CI for ARR
- 95% CI for NNT
Formula
ARR = CER − EER | NNT = 1 ÷ ARR (round up) | RR = EER ÷ CER | RRR = 1 − RR | 95% CI for ARR = ARR ± 1.96 × √(CER(1−CER)/n₁ + EER(1−EER)/n₂)
The number needed to treat (NNT) is a measure of treatment effect that indicates how many patients must be treated to prevent one additional adverse event. It is calculated from the absolute risk reduction (ARR), which is the difference between the control event rate (CER) and the experimental event rate (EER). CER is the proportion of patients in the control group who experience the event (events divided by total number in control), and EER is the same for the experimental group. ARR = CER - EER; a positive ARR means the treatment reduces risk.
NNT is computed as 1 divided by ARR, and the result is rounded up to the nearest whole number. If ARR is negative, the treatment may cause harm, and the number needed to harm (NNH) is calculated as 1 divided by the absolute value of ARR. NNT is always tied to a specific time period and depends on the baseline risk: for the same relative risk reduction (RRR), a higher baseline risk yields a lower NNT. For example, a 30% RRR gives an NNT of 33 at a 10% baseline risk but 333 at 1%.
Other related measures include relative risk (RR), which is EER divided by CER, and RRR, which is ARR divided by CER, or equivalently 1 - RR. The odds ratio is another metric: [EER/(1-EER)] divided by [CER/(1-CER)]. Unlike ARR, relative measures do not directly convey baseline risk and can be misleading.
Confidence intervals (CI) for ARR are calculated using the Wald method: ARR ± 1.96 × √(CER(1–CER)/n₁ + EER(1–EER)/n₂), where n₁ and n₂ are the total numbers in control and experimental groups. The CI for NNT is obtained by taking the reciprocals of the ARR CI limits. If the ARR CI includes zero, the NNT CI is expressed as "NNT a to ∞ to NNH b" (Altman 1998), indicating the treatment may be beneficial or harmful. In the example with control 30/200 (15%) and treatment 18/200 (9%), ARR is 6% (CI –0.3% to 12.3%), so NNT is 17 with CI 8.1 to ∞ to NNH 292, which is not statistically significant.
Worked example
Input: Control 30/200 (15%), treatment 18/200 (9%)
Result: ARR 6%, NNT 17, RRR 40%; NNT 95% CI 8.1 to ∞ to NNH 292
Not statistically significant, because the CI for ARR crosses zero.
Questions
What is the number needed to treat (NNT)?
NNT is the number of patients who must receive a treatment to prevent one additional bad outcome compared to a control. It equals 1 divided by the absolute risk reduction (ARR), rounded up to the next whole number. NNT is always linked to a specific time period and depends on the baseline risk of the population.
How do you calculate NNT?
First, compute the control event rate (CER) and experimental event rate (EER) as events divided by total subjects. Subtract EER from CER to get ARR. Then NNT = 1 / ARR, rounded up. If ARR is negative, the number needed to harm (NNH) is 1 / |ARR|.
What is the difference between NNT and NNH?
NNT measures benefit (treatment prevents events), while NNH measures harm (treatment causes additional events). NNH is calculated when the experimental event rate is higher than the control rate, giving a negative ARR. The formula is NNH = 1 divided by the absolute value of ARR.
Why does NNT depend on baseline risk?
Because ARR is influenced by the baseline risk (CER). The same relative risk reduction (RRR) yields different ARRs. For instance, an RRR of 30% gives an ARR of 3% at a 10% baseline risk (NNT ≈ 33) but only 0.3% at 1% baseline (NNT ≈ 333). So NNT varies with the population's underlying risk.
How do you interpret a confidence interval for NNT that includes infinity?
When the 95% CI for ARR includes zero, the treatment effect is not statistically significant. The NNT CI is then expressed as "NNT benefit to infinity to NNT harm" (e.g., NNT 8.1 to ∞ to NNH 292). This means the treatment could be beneficial, have no effect, or cause harm.
What does it mean if the absolute risk reduction is negative?
A negative ARR occurs when the event rate in the experimental group exceeds that in the control group. This suggests the treatment may be harmful. The number needed to harm (NNH) is then calculated by taking the reciprocal of the absolute ARR, indicating how many treated patients lead to one extra harmful event.
Limitations
- NNT is specific to the time period and baseline risk of the studied population; it cannot be applied to different contexts without adjustment.
- The confidence interval calculation assumes a large sample approximation (Wald method) and may be inaccurate with small sample sizes.
- If the ARR is small, small changes in event rates can greatly change NNT, making it sensitive to measurement error.
- NNT does not convey the magnitude of benefit on an individual level; it is a population-averaged measure.
References
- Laupacis A, Sackett DL, Roberts RS (1988) An assessment of clinically useful measures of the consequences of treatment. NEJM 318:1728–1733
- Altman DG (1998) Confidence intervals for the number needed to treat. BMJ 317:1309–1312
- Cook RJ, Sackett DL (1995) The number needed to treat: a clinically useful measure of treatment effect. BMJ 310:452–454
Related reading
- Critical Appraisal of Randomized Controlled Trials in Veterinary Medicine
- Evidence-Based Veterinary Medicine: Principles and Practice
- Critical Appraisal Tools for Veterinary Research: A Comparative Review
- Absolute Risk Reduction vs Relative Risk Reduction: Meaning, Formulas and Examples
- Critical Appraisal of Observational Studies
- How to Calculate Number Needed to Treat (NNT) With a Confidence Interval
More Statistics
- Standard Deviation Calculator
- Mean, Median & Mode Calculator
- Variance Calculator
- Z-Score Calculator
- P-Value Calculator
- Confidence Interval Calculator
- T-Test Calculator
- Chi-Square Test Calculator
- Correlation Coefficient Calculator
- Linear Regression Calculator
- Normal Distribution Calculator
- Probability Calculator
All free tools · By Dr. Zubair Khalid, DVM, MS, PhD.