NNT Calculator

What is the Number Needed to Treat (NNT)?

The Number Needed to Treat (NNT) is an epidemiological and evidence-based medicine (EBM) metric that expresses the clinical effectiveness of a medical intervention or pharmaceutical therapy. It represents the average number of patients who must receive a specific treatment (rather than a control or placebo) for one additional patient to achieve a positive outcome or avoid a specific adverse event.

Unlike relative risk reduction (RRR), which can exaggerate modest therapeutic effects in rare diseases, NNT directly reflects baseline disease prevalence and real-world clinical impact. An ideal therapy has an NNT of 1, meaning every treated patient benefits. In clinical practice, NNTs between 2 and 50 are common depending on the condition and outcome severity.

Formulas and Calculation Methods

NNT is the mathematical inverse of the Absolute Risk Reduction (ARR):

$$\text{ARR} = \text{CER} - \text{EER} = \frac{c}{n_1} - \frac{e}{n_2}$$

$$\text{NNT} = \frac{1}{\text{ARR}} = \frac{1}{\text{CER} - \text{EER}}$$

Where:

• CER (Control Event Rate): The proportion of subjects experiencing the adverse event in the control or placebo group ($c / n_1$).
• EER (Experimental Event Rate): The proportion of subjects experiencing the adverse event in the treatment group ($e / n_2$).
• ARR (Absolute Risk Reduction): The absolute arithmetic difference between the two rates.

By convention, the calculated NNT value is rounded up to the nearest whole integer ($\lceil 1/\text{ARR} \rceil$) to represent realistic whole-patient clinical encounters.

Number Needed to Harm (NNH)

When an intervention increases the incidence of adverse side effects or negative outcomes ($\text{EER} > \text{CER}$), the risk difference is negative. In this situation, the metric is known as the Number Needed to Harm (NNH):

$$\text{NNH} = \frac{1}{\text{ARI}} = \frac{1}{\text{EER} - \text{CER}}$$

Where $\text{ARI}$ is the Absolute Risk Increase. A higher NNH indicates a safer medication, meaning many patients can take the drug before one adverse reaction occurs.

Related Statistical Measures

• Relative Risk (RR): $\text{RR} = \text{EER} / \text{CER}$. Compares event probability in the intervention group to control.
• Relative Risk Reduction (RRR): $\text{RRR} = (\text{CER} - \text{EER}) / \text{CER} = 1 - \text{RR}$. Measures proportionate risk decrease.
• Odds Ratio (OR): Compares odds of event occurrence between groups, standard in case-control trials and logistic regression models.