Number Needed to Treat Calculator
The number needed to treat turns a trial result into a single clinically meaningful figure: how many patients you would treat to prevent one bad outcome. Enter the events and totals for the treatment and control arms to get the number needed to treat, the absolute and relative risk reduction, and a confidence interval.
Enter the number of bad outcomes (events) and the total in each arm. The treatment arm is Group 1 and the control arm is Group 2.
How to use the number needed to treat calculator
You need the events and totals from the treatment arm and the control arm of a trial. From those the calculator derives the risk in each group and works forward to the number needed to treat.
- 1. Enter the treatment arm. Type the number of events and the total number of patients who received the intervention.
- 2. Enter the control arm. Add the events and total for the comparison or placebo group.
- 3. Read the absolute risk reduction. This is the control risk minus the treatment risk, the gap the number needed to treat is built from.
- 4. Read the number needed to treat. It is one divided by the absolute risk reduction, shown with its confidence interval and the relative risk reduction beside it. When the treatment increases the bad outcome, the result is labelled number needed to harm instead.
A worked example
Suppose 12 of 200 treated patients had the event and 20 of 200 control patients had it. The risk is 0.06 in the treatment arm and 0.10 in the control arm, so the absolute risk reduction is 0.10 minus 0.06, which is 0.04, or four percentage points. The number needed to treat is one divided by 0.04, which equals 25. You would treat 25 patients to prevent one additional event.
Notice how the relative and absolute pictures differ. The relative risk reduction here is 0.04 divided by the control risk of 0.10, which is 40 percent, a figure that sounds far more impressive than treating 25 people for one to benefit. Both numbers are correct, but the number needed to treat is the one that tells a patient what to expect.
Why absolute risk reduction is the honest figure
A treatment can boast a large relative risk reduction while delivering only a tiny absolute benefit, and the number needed to treat exposes that gap. Halving a risk that was already small, say from two percent to one percent, is a 50 percent relative reduction but only a one percentage point absolute reduction, which translates to a number needed to treat of 100. That is why absolute risk reduction, not relative risk reduction, is the figure that should drive treatment decisions, and why the number needed to treat is so useful for communicating benefit to patients.
The explainer on relative versus absolute risk reduction unpacks this difference in detail, and the guide to interpreting the number needed to treat covers how to read a confidence interval that crosses into harm. To work directly with the underlying ratios, use the underlying relative risk, and to pool effects across trials use the pooled-effect tool before drawing them as a forest plot of the combined trials.
Common mistakes to avoid
- Quoting the relative reduction as the benefit. A 40 percent relative reduction can sit on top of a number needed to treat of 25 or of 250, depending on the baseline risk. The number needed to treat keeps the conversation honest.
- Reporting a single value with no interval. The number needed to treat has a confidence interval, and when the risk difference is not significant that interval runs through infinity into the harm range. A point estimate alone hides that.
- Rounding down. The number needed to treat is always rounded up to the next whole patient, because you cannot treat a fraction of a person and rounding down would overstate the benefit.
- Comparing across different follow-up times. A number needed to treat is tied to the trial's time horizon. A value over one year is not comparable to one over five years, so always state the period.
Need the absolute benefit reported across a whole review?
Translating pooled effects into a number needed to treat your readers will trust takes a careful hand. A methodologist can extract the data, pool it correctly, and write the clinical interpretation that survives peer review.
Get help with your analysisFrequently asked questions
How do you calculate the number needed to treat?
The number needed to treat is the reciprocal of the absolute risk reduction. First find the risk in the control group and the risk in the treatment group, then subtract to get the absolute risk reduction. Dividing 1 by that figure gives the number of patients you would need to treat to prevent one additional bad outcome. This calculator computes the risk in each arm, the absolute risk reduction, and the number needed to treat together, and it adds a confidence interval derived from the interval on the risk difference.
How do you interpret a number needed to treat of 10?
A number needed to treat of 10 means that, on average, you would have to treat 10 patients with the intervention to prevent one additional bad outcome compared with the control. Lower numbers indicate more effective treatments, because fewer patients need to be treated for one to benefit. A value of 10 is generally considered a useful effect for many conditions, but it has to be weighed against the cost, the burden, and the harms of treating the other nine who do not benefit.
What is a good number needed to treat?
There is no fixed threshold, because a good value depends on the seriousness of the outcome and the safety of the treatment. A number needed to treat of 2 or 3 for a low-risk intervention preventing a serious event is excellent, while a value of 50 might still be worthwhile for a cheap, safe drug preventing something severe. Always read it next to the number needed to harm: a treatment is attractive only when the number needed to treat is comfortably smaller than the number needed to harm.
What do number needed to harm and number needed to treat mean?
The number needed to treat counts how many patients must receive a beneficial treatment for one to gain the desired outcome, while the number needed to harm counts how many must be treated for one to suffer an additional adverse event. They are computed the same way, as the reciprocal of an absolute risk difference, and which one applies depends on whether the intervention increases or decreases the outcome. This calculator labels the result as treat or harm automatically based on the direction of the effect.
Do you round the number needed to treat up or down?
Always round up to the next whole number. If the reciprocal of the absolute risk reduction comes to 12.3, you report 13, because you cannot treat a fraction of a patient and rounding down would make the treatment look more effective than the data support. Rounding up is the conservative, conventional choice in clinical reporting.
What does a high number needed to treat mean?
A high number needed to treat means many patients must be treated for one to benefit, so the treatment has a small absolute effect. A value of 100, for instance, says only one in every hundred treated patients avoids the outcome because of the intervention. A high value is not automatically bad if the treatment is cheap and safe and the outcome is serious, but it does signal that the benefit per patient is modest.