π NNT Calculator (Number Needed to Treat)
How many must be treated for one to benefit?
For informational purposes only, not medical advice. Consult a healthcare professional.
Number needed to treat is the clearest efficacy number in medicine: how many patients must receive a treatment for one extra patient to benefit. It is simply one divided by the absolute risk reduction. An NNT of 13 means treating 13 prevents one event; the other 12 get no benefit from that outcome (though they may benefit in other ways).
Enter the event rate in the control group and in the treatment group as percentages. The calculator returns the NNT rounded up (you cannot treat half a patient), plus absolute and relative risk reductions for comparison.
How to Use This Calculator
Find the event rates in the trial: percent of control patients and percent of treated patients who had the outcome over the study period. Enter both. Read the NNT. Always note the timeframe: an NNT of 20 over 5 years is a different proposition from 20 over 5 weeks.
Enter the control group event rate as a percent.
Enter the treatment group event rate as a percent.
Read the NNT (rounded up) and risk reductions.
Note the trial timeframe and outcome when quoting it.
One Over Absolute Reduction
NNT = 1 / ARR (round up)
Absolute risk reduction is the honest efficacy measure: the actual percentage-point difference treatment makes. NNT inverts it into people: 1/0.08 = 12.5, rounded up to 13, because fractional patients do not exist. Relative risk reduction (here 40%) describes the same effect against the baseline and systematically sounds better, which is why marketing prefers it and informed consent should include the absolute numbers. NNT also has a harm twin, NNH (number needed to harm), computed the same way from adverse event rates: the benefit-harm comparison that actually guides prescribing.
Worked Example
Control event rate 20%, treatment 12%: absolute reduction is 8.0%, so NNT is 13, meaning 1 in 13 treated benefits. The relative reduction of 40% sounds more impressive than it is: relative risks flatter treatments, absolute numbers inform patients.
What Counts as a Good NNT
Under 10 is a strong effect (many antibiotics for bacterial infections live here). Ten to 50 is moderate and typical of preventive cardiology: statins for secondary prevention run NNTs around 20-30 over 5 years. Above 50 is modest per patient but can still matter at population scale. Context rules: an NNT of 100 to prevent a fatal event over a year beats an NNT of 10 to prevent a mild rash, and NNT without NNH is half the story.
| NNT | Strength | Example territory |
|---|---|---|
| <10 | Strong | Antibiotics, acute treatments |
| 10-50 | Moderate | Statins, BP drugs (years) |
| >50 | Modest | Some primary prevention |
| No benefit | None | ARR zero or negative |
Trial Numbers Need Clinical Judgment
NNTs come from trial populations over fixed timeframes and do not automatically apply to any individual: baseline risk, adherence, and competing risks all modify the real number. Treatment decisions weigh NNT against NNH, cost, and patient preference with a clinician. For informational purposes only, not medical advice. Consult a healthcare professional.
Related tools: incidence rate calculator and mortality rate calculator.
For informational purposes only, not medical advice. Consult a healthcare professional.
Frequently Asked Questions
How do you calculate NNT?
One divided by the absolute risk reduction: NNT = 1/(control rate - treatment rate), rounded up to the next whole number.
What is a good NNT?
Under 10 is strong, 10-50 moderate, over 50 modest. The outcome's severity matters as much as the number.
What is the difference between NNT and relative risk?
NNT uses absolute difference (honest scale); relative risk reduction describes the proportional change and usually sounds more impressive.
What is NNH?
Number needed to harm: the same math applied to adverse events. Comparing NNT with NNH is the core benefit-harm assessment.
Can NNT be negative?
A negative ARR means the treatment caused more events: it becomes a number needed to harm instead.
Why round NNT up?
You cannot treat a fraction of a patient, so 12.5 becomes 13 by convention.
Does NNT depend on timeframe?
Yes, strongly. Always quote the trial duration: NNT 20 over 5 years differs from NNT 20 over 5 weeks.
Where can I find published NNTs?
Resources like TheNNT.com summarize NNTs and NNHs for common treatments from the trial literature.
What is a good NNT?
Depends on the stakes. An NNT of 2 for curing an infection is excellent; an NNT of 50 to prevent one heart attack over five years can still be worthwhile if the drug is cheap and safe. Compare NNT against the harm equivalent (NNH): a treatment with NNT 20 and NNH 100 is attractive, while NNT 20 with NNH 10 is a hard sell.
Why do NNTs from different trials disagree?
Because NNT depends on baseline risk, which varies between trial populations. A drug that cuts risk by 25 percent has NNT 40 in a high-risk group (10 percent baseline) but NNT 400 in a low-risk group (1 percent baseline). Same drug, same relative effect, tenfold different NNT. Always check whose risk the trial studied before applying it to a patient.