What NNT means
Number needed to treat (NNT) translates an absolute difference between groups into an average number treated for one additional beneficial outcome. Always specify the outcome, comparator and follow-up time [2].
An NNT of 10 means one additional person benefits per 10 treated compared with the control over that period. It does not reveal which person benefits, nor does it show that the other nine had no biological response. Some people would have had the same outcome with either treatment.
How it is calculated
For prevention of an unwanted event, NNT is 1 divided by the absolute risk reduction (ARR), expressed as a proportion. If an event occurs in 10% of a placebo group and 5% of a supplement group during the same one-year follow-up, ARR is 0.05 and NNT is 20 over one year [2].
This needs the absolute difference. A drop from 10% to 5% is a 50% relative reduction but a 5-percentage-point absolute reduction. See relative vs. absolute risk. Continuous outcomes, such as a mean change in a symptom score, do not directly give an NNT without a justified definition of benefit.
Why context matters more than a cutoff
For the same outcome, comparator and period, a lower NNT represents a larger absolute benefit. There is no universal boundary between a worthwhile and a trivial NNT: preventing a death and improving a mild symptom are different decisions. Baseline risk, treatment burden, cost and adverse effects also matter [2].
NNT is an estimate, not a guarantee. Its confidence interval can be very wide or include no benefit and possible harm. When an absolute difference is zero, there is no finite NNT; when treatment increases an unwanted outcome, the corresponding measure is a number needed to harm.
Putting it to use
Look for event rates in both groups, the follow-up time and the confidence interval. Ask whether the study participants resemble the people considering treatment. Relative-risk headlines alone cannot tell you the absolute benefit [1][2].
A statistically significant result is not automatically a clinically important one. Conversely, a large NNT can still matter for a serious outcome when treatment is low-burden and sufficiently safe. Evaluate benefits and harms together.