Number Needed to Treat

The number needed to treat (NNT) is the number of people who would need to receive a treatment for one additional person to benefit, compared with a control. It is the reciprocal of the absolute risk reduction and re-expresses an absolute measure of association in terms patients and clinicians find intuitive.

Definition

The number needed to treat is the reciprocal of the absolute risk reduction between a treatment and a control group, interpreted as the number of patients who must be treated for one additional beneficial outcome.

Scope

This entry covers the definition of the NNT, its derivation from the absolute risk reduction, its dependence on baseline risk and time horizon, the related number needed to harm, and the cautions around its confidence intervals. It treats the NNT as a methodological measure, not as clinical guidance.

Key concepts

• Reciprocal of the absolute risk reduction
• Dependence on baseline risk
• Time horizon of the estimate
• Number needed to harm
• Confidence intervals for the NNT
• Patient-facing communication of effect

Mechanisms

The NNT is computed as 1 divided by the absolute risk reduction, the difference in the probability of a good (or bad) outcome between the treatment and control groups. A smaller NNT indicates a more effective treatment, because fewer people must be treated for one to benefit. Because it derives from an absolute difference, the NNT inherits that difference's dependence on baseline risk and on the length of follow-up over which the outcome is measured, so an NNT is meaningful only for a stated comparator, outcome, and time horizon. The analogous number needed to harm is the reciprocal of an absolute risk increase. The NNT requires care in interval estimation: when the risk difference is small or its confidence interval crosses zero, the NNT scale behaves awkwardly, which is why methods for confidence intervals on the NNT were developed.

Clinical relevance

The NNT translates an absolute measure of association into a single figure intended to make the size of a treatment effect easier to grasp when appraising trial evidence. It summarises evidence about a treatment in a population for a defined outcome and horizon; it characterises study results and is not by itself a prescription for any individual patient.

Epidemiology

The NNT is used to summarise and compare treatment effects from randomised trials and meta-analyses, and the parallel number needed to harm summarises adverse effects. Because it is tied to baseline risk and follow-up time, an NNT computed in one population or over one horizon does not transfer automatically to another, and it is reported with a confidence interval and the underlying absolute risk reduction.

History

The number needed to treat was introduced by Laupacis, Sackett, and Roberts in 1988 as a more clinically interpretable expression of treatment effect than relative measures alone, and was popularised by Cook and Sackett in 1995 within the evidence-based-medicine movement. Altman's 1998 work then addressed how to attach confidence intervals to the NNT, a recurring difficulty given its reciprocal scale.

Debates

Confidence intervals and unstable NNTs

When the absolute risk reduction is small or not statistically significant, the NNT's reciprocal scale produces awkward or seemingly infinite intervals, so the NNT must be reported with appropriately constructed confidence limits and read together with the underlying risk difference.

Transportability across baseline risks

Because the NNT depends on baseline risk and time horizon, an NNT from one trial or population does not apply unchanged to patients with different baseline risk, which limits naive comparison of NNTs across studies.

Key figures

• Andreas Laupacis
• David Sackett
• Richard Cook
• Douglas Altman

Seminal works

• laupacis-1988
• cook-sackett-1995
• altman-1998

Frequently asked questions

How is the number needed to treat calculated?

It is the reciprocal of the absolute risk reduction; for example, an absolute risk reduction of 0.05 (5 percentage points) gives a number needed to treat of 20, meaning 20 people must be treated for one additional beneficial outcome over the stated period.

Does a single number needed to treat apply to every patient?

No; the number needed to treat depends on the baseline risk, the comparator, the outcome, and the follow-up time it was derived from, so it describes a study population and does not transfer unchanged to patients with different baseline risk.