VerifiedEvidence: highv1.0.0

Number Needed to Harm

The number of patients who would need exposure to a treatment or risk factor for one additional patient to experience an adverse outcome.

Last reviewedDarrin Baines IP Ltd

Concept Architecture

Concept


Theoretically, Number Needed to Harm (NNH) is an epidemiological and clinical measure that quantifies the number of individuals who must receive an intervention for one additional adverse event to occur compared with a control or alternative intervention over a specified period. It represents the reciprocal of the absolute increase in risk attributable to treatment and is founded on absolute risk comparison. The concept exists to express treatment-related harm in an intuitive and clinically meaningful manner for benefit-risk assessment and healthcare decision-making.

Mathematically, the Number Needed to Harm is represented as the reciprocal of the Absolute Risk Increase (ARI). Because it is based on an absolute difference in event risk between two groups, NNH is expressed as the number of treated individuals required to produce one additional harmful outcome. Smaller NNH values indicate greater treatment-related harm, whereas larger values indicate lower absolute harm.

In practice, Number Needed to Harm is calculated from randomised controlled trials, observational studies and meta-analyses using absolute event risks. It is widely applied in clinical guideline development, health technology assessment and health economic evaluation to compare the safety profiles of interventions alongside measures such as the Number Needed to Treat.

Purpose


Used to quantify the absolute frequency of treatment-related adverse events, compare intervention safety, support benefit-risk assessment and inform clinical and health economic decision-making.


Mathematical Formulae

Primary Formula

NNH = 1 / ARI

where:

  • NNH = number needed to harm
  • ARI = absolute risk increase

Supporting Formulae

ARI = Risk? ? Risk??

Risk = Events / Population

NNH = 1 / (Risk? ? Risk??)

where:

  • Risk? = risk in the treatment group
  • Risk?? = risk in the control group

Related Mathematical Methods

  • Absolute Risk Increase
  • Absolute Risk
  • Risk Difference
  • Number Needed to Treat
  • Relative Risk
  • Attributable Risk
  • Benefit-Risk Assessment

Example


A clinical trial reports that a serious adverse event occurs in 8% of patients receiving a new treatment and 4% of patients receiving standard care.

Absolute Risk Increase:

ARI = 0.08 ? 0.04 = 0.04

Number Needed to Harm:

NNH = 1 / 0.04 = 25

On average, treating 25 patients with the new intervention results in one additional serious adverse event compared with standard care.


Excel Implementation

FunctionExample FormulaHealth Economics Application
Subtraction=B2-C2Calculates the Absolute Risk Increase from treatment and control event risks.
Division=1/(B2-C2)Calculates the Number Needed to Harm.
ABS=1/ABS(B2-C2)Calculates NNH when only the magnitude of the absolute risk difference is required.
IF=IF(B2>C2,1/(B2-C2),"No treatment-related harm")Calculates NNH only when the treatment increases adverse event risk.

VBA (Optional)


A VBA macro can automatically calculate Number Needed to Harm values, confidence intervals and comparative safety summaries across multiple clinical studies.


Sources

  • Altman DG. Confidence intervals for the number needed to treat. BMJ. 1998;317:1309?1312.
  • Cook RJ, Sackett DL. The number needed to treat: a clinically useful measure of treatment effect. BMJ. 1995;310:452?454.
  • Altman DG, Andersen PK. Calculating the number needed to treat for trials where the outcome is time to an event. BMJ. 1999;319:1492?1495.
  • Drummond MF, et al. Methods for the Economic Evaluation of Health Care Programmes. 4th ed.
  • NICE. Health Technology Evaluation Manual.

Library

Publications

1
  • Book

    Statistical Analysis of Cost-Effectiveness Data — Willan & Briggs, 1st Edition ed., 2006 (John Wiley & Sons)

    A synthesis of statistical methods for analysing cost-effectiveness data, including net-benefit regression, confidence intervals for the ICER, cost-effectiveness acceptability curves, and covariate adjustment. Part of the Wiley Statistics in Practice series.

Frequently Asked Questions (6)

  • What is the number needed to harm?

    The number of patients who would need exposure to a treatment or risk factor for one additional patient to experience an adverse outcome.

    Source: Cook & Sackett 1995

  • What does the number needed to harm express about a treatment's risk?

    The number needed to harm expresses how many patients would have to be exposed to a treatment or risk factor for one extra person to suffer a particular adverse outcome. A small number means harm is common, arising after only a few are treated, while a large one means it is rare. Set beside the number needed to treat for benefit, it lets the good a treatment does be weighed against the harm, in the same intuitive units of people affected. Framing risk as people harmed is its purpose. Sackett and colleagues (1991) describe this measure.

    Source: Sackett et al. 1991

  • How is the number needed to harm calculated?

    The number needed to harm is calculated as the reciprocal of the absolute risk increase, that is one divided by the difference in the risk of the adverse outcome between the exposed and unexposed groups. If exposure raises the risk of harm by a certain absolute amount, the number needed to harm is one divided by that amount. So the number needed to harm is calculated from the absolute risk increase in the adverse outcome, taking its reciprocal, which converts the difference in harm between groups into the number who must be exposed for one additional harmful event.

    Source: Cook & Sackett 1995

  • How is the number needed to harm interpreted?

    The number needed to harm is interpreted as the number of patients who must be exposed for one extra adverse outcome to occur: a small number means harm arises after relatively few exposures and is therefore more frequent, while a large number means harm is rare. So the number needed to harm is interpreted as how many exposures produce one additional harm, with smaller values indicating greater harm, and it is weighed against the number needed to treat for benefit, since comparing how many must be treated to help one patient with how many to harm one patient informs the balance of benefit and risk.

    Source: Cook & Sackett 1995

  • How does the number needed to harm relate to the number needed to treat?

    The number needed to harm relates to the number needed to treat as its counterpart for adverse outcomes: the number needed to treat is the number who must be treated for one to benefit, and the number needed to harm the number who must be exposed for one to be harmed, both being reciprocals of an absolute risk difference. So the number needed to harm and the number needed to treat together express a treatment's benefits and harms on a common, interpretable scale, allowing clinicians and patients to weigh how many must be treated to help one against how many would be harmed.

    Source: Cook & Sackett 1995

  • What are the limitations of the number needed to harm?

    The limitations of the number needed to harm include that it depends on the baseline risk and the time period, so it varies between populations and durations and is not a fixed property of a treatment; that it requires a reliable estimate of the absolute risk increase; and that it summarises one harm at a time. So the number needed to harm is interpreted with attention to the baseline risk, time frame, and population to which it applies, since a single number needed to harm does not capture all harms or their severity, and comparing across settings requires comparable baseline risks.

    Source: Cook & Sackett 1995

Trust Record

Verified by Dr Darrin Baines

British health economist

Professional identity: darrinbaines.org

Verification date: 9 Dec 2025

Content version: 1.0.0

Canonical Identity

Term code
HE-ES-RM-016

Stable URI · Machine-readable · Resolvable · CC BY 4.0