A number without a timeline can mislead you
Imagine a friend tells you a new medication has a “number needed to harm” (NNH) of 50 — meaning one extra person out of every 50 treated experiences a specific side effect. That sounds precise. But precise about what? Fifty people treated for how long? Compared to what baseline risk? Without those two details, an NNH is close to meaningless, even though it looks like hard data.
This guide explains how to read an NNH estimate the way researchers intend it to be read — as a snapshot tied to a specific population, a specific follow-up period, and a specific comparison — so you can bring sharper questions to a conversation with a clinician instead of treating the number as a personal commitment.
What “number needed to harm” actually means
NNH is a way of expressing how often an unwanted effect happens in a treated group compared with a control or comparison group. It answers a narrow question: out of a defined number of people who took a treatment for a defined period, how many experienced one extra case of a defined harm, above and beyond what happened in the comparison group?
It is the mirror image of “number needed to treat” (NNT), which estimates how many people need to receive a treatment for one extra person to benefit. Cochrane’s guidance on presenting results places both measures inside the same family of tools used to translate relative differences between groups into an absolute, real-world quantity.
Stage 1: Check the follow-up period before anything else
The single most common way an NNH gets misread is by ignoring how long the study followed people. An NNH of 100 over one year is a very different risk picture than an NNH of 100 over ten years — the shorter the window, the more concentrated the harm.
- Look for the exact duration the study tracked participants: weeks, months, or years.
- Ask whether the harm being counted could appear early, late, or only with cumulative exposure.
- Be cautious of any summary that reports an NNH without stating the time period — that omission strips the number of its meaning.
Stage 2: Find the baseline risk the comparison was built on
An NNH is only informative next to the baseline risk of the population studied. A harm that is rare in a low-risk population may be far more common in a high-risk population, even at the same relative rate.
- Identify who was enrolled in the study — their age range, health status, and any risk factors specified.
- Note the absolute event rate in the comparison group, not just the relative increase.
- Recognize that an NNH calculated in one population does not automatically transfer to a different population, including yours.
Cochrane’s chapter on completing summary-of-findings tables describes converting relative measures into absolute terms specifically so readers can see how a baseline risk changes the real-world impact of a treatment effect — the same logic applies whether the effect being measured is a benefit or a harm.
Stage 3: Look for the uncertainty around the number
An NNH is a point estimate, not a fixed fact. It typically comes with a confidence interval — a range that reflects how much the estimate could shift if the study were repeated. A wide confidence interval means the true NNH could be considerably higher or lower than the headline number.
- Check whether a confidence interval or range is reported alongside the NNH.
- A narrow range from a large, well-conducted study carries more weight than a wide range from a small one.
- If no uncertainty range is given at all, treat the single number as a rough approximation rather than a precise prediction.
Warning boundary: what an NNH cannot tell you
An NNH describes a group-level pattern, not an individual outcome. It cannot tell you whether you personally will experience the harm, how severe it would be for you, or how it weighs against a specific benefit you might also gain from the same treatment. Two treatments with the same NNH can carry very different real-world consequences if one harm is minor and reversible and the other is serious and lasting.
This article does not diagnose conditions, recommend for or against any treatment, or suggest starting, stopping, or changing a medication. If you are experiencing a medical emergency, contact local emergency services immediately.
A worksheet for reading any NNH claim
Use these five questions any time you encounter an NNH — in a news article, a product summary, or a conversation with a clinician:
- Over what time period was this measured? Ask for the exact follow-up duration.
- Who was studied? Ask about the age, health status, and risk profile of the population.
- What was the baseline rate of harm without treatment? Ask for the absolute risk, not just the relative one.
- How much uncertainty surrounds the number? Ask whether a confidence interval was reported.
- How does this compare with the expected benefit? Ask what the number needed to treat looks like for the same population and timeframe.
A decision path for your next conversation
You do not need to become a statistician to use this worksheet well. The goal is to walk into a care conversation with sharper questions rather than a memorized number:
- Write down the NNH exactly as it was reported, including the source.
- Bring the five worksheet questions above to your clinician or pharmacist.
- Ask them to translate the number into what it would mean for someone with your specific health background, since published NNH figures come from study populations that may not match your situation.
- Ask how the harm being counted would be monitored or managed if it occurred.
Where to go next
If you want the broader picture of how to evaluate a study before you even get to a specific number, start with our Start Here guide. For a fuller explanation of how we vet the sources behind articles like this one, see How We Research.
Sources and limits of this article
This article draws on the Cochrane Handbook for Systematic Reviews of Interventions, specifically the chapters covering completion of summary-of-findings tables and interpretation of results, which describe how absolute measures like NNH and NNT are constructed and why baseline risk and follow-up period change their meaning. It does not report on, endorse, or evaluate any specific medication, product, or brand, and it is not a substitute for professional medical advice, diagnosis, or treatment. Always talk with a qualified healthcare provider about your own situation.
Last updated: September 2026. By Everyday Prevention Evidence Editorial Team.
Leave a Reply