Number Needed to Treat (NNT)

Concept and Formula

NNT rests on the absolute risk reduction (ARR), which is the difference between the control event rate (CER) and the experimental event rate (EER). The formulas are ARR = CER - EER and NNT = 1/ARR. For example, if the event rate is thirty percent in the control group and twenty percent in the treatment group, ARR = 0.10 and NNT = 10, meaning ten patients must be treated for one additional good outcome. NNT is always rounded up to the nearest whole number and must always be defined for a specific outcome and a specific time horizon.

How to Read and Interpret NNT

NNT = 1 represents perfect efficiency; every treated patient benefits. In practice this is rarely achieved. As a general rule, a smaller NNT indicates a more efficient treatment, but an acceptable threshold depends on disease severity, treatment cost, and the side-effect profile. NNT must always be reported with a confidence interval; a point estimate alone conceals uncertainty. When the treatment also carries risks of harm, NNT should be interpreted alongside the number needed to harm (NNH), which is computed in an analogous way from the absolute risk increase.

Common Misconceptions

The most common error is overstating treatment benefit by relying on the relative risk reduction (RRR) rather than the NNT. A fifty percent RRR sounds impressive, but if baseline rates are very small, the NNT can be in the hundreds. Another mistake is directly comparing NNT values across different time frames or patient populations; when baseline risk changes, so does the NNT. Finally, because NNT is derived from the baseline risk of the trial population, it cannot be applied directly to individual patients whose baseline risk differs; a patient-specific ARR estimate is then required.

Why It Matters and How to Report It

NNT translates abstract statistics into a language that decision makers can act on. Clinicians use it to counsel patients, guideline panels use it to prioritize resources, and health economists use it as an input to cost-effectiveness analyses. A rigorous NNT report should include the specific outcome and follow-up duration, the absolute event rates in both arms, the ARR with a ninety-five percent confidence interval, and the NNH when relevant harms exist. CONSORT and GRADE reporting guidelines provide formal standards. When a publication reports only relative measures, readers can reconstruct the ARR and NNT from the reported control event rate.

Sources