STATISTICS

Number needed to harm (NNH)

Glossary · Published 2026-05-18 · Last reviewed 2026-09-23

Share this

EmailXText

Plain-language definition

The number needed to harm, or NNH, is how many people would need to receive a treatment for one extra person to experience a particular harm, such as a side effect.

This term is specific to research interpretation and is defined on AnxietyResearch. Shrinktionary, the network's language layer, doesn't currently carry a canonical entry for it.

In more detail

NNH is the mirror image of the number needed to treat. NNT counts benefit. NNH counts harm. A high NNH is good, because it means harm is rare. A low NNH is concerning, because it means harm is common. Reading the two together gives a balanced picture of a treatment.

Why it matters in anxiety research

Every anxiety treatment has trade-offs. Medications can cause side effects. NNH puts a number on how often. Comparing NNT and NNH helps a clinician and patient weigh whether a treatment's likely benefit is worth its likely harm.

An example

A medication has an NNT of 6 for a meaningful response and an NNH of 30 for a bothersome side effect. Benefit is more common than this particular harm, which supports its use for many patients, though not all.

What to watch for

NNH depends entirely on which harm is being counted. A drug can have a high NNH for a rare serious harm and a low NNH for a common mild one. Check which harm the number describes. A single NNH doesn't capture every risk.

Related terms

Related reports or summaries

Related on this site

Benzodiazepines, benefits and risks; SSRIs and SNRIs for anxiety

Source

Cochrane Handbook for Systematic Reviews of Interventions, current edition. https://training.cochrane.org/handbook

Share this

EmailXText

Your next step in The Shrink Network

You are here: AnxietyResearch, the evidence layer of The Shrink Network.

Each site in the network has one job. No matter where you enter, we help you find the next step that makes sense.

Get clinical care at shrinkMD

Want to understand more first?