2027 Number Needed to Treat and Number Needed to Harm: Research Interpretation Concepts
Audited and Verified:
Our Editorial and Data Verification Process
Research.com uses government datasets, official education statistics and primary institutional information to support its career and degree guides. Our review process checks factual accuracy, occupational definitions, source dates and the requirements associated with the credentials discussed.
How This Article Was Fact-Checked
Our editorial team verifies salary estimates, employment projections, education requirements and program information against original data providers and official organizations. Salary figures are checked against the relevant occupational definitions and reporting periods. Information about licensure, accreditation and required in-person training is reviewed separately.
Fact-checked and verified on:
A treatment can sound impressive when a study reports a large relative benefit, yet help only a small fraction of patients. Number needed to treat (NNT) and number needed to harm (NNH) put benefits and adverse effects on an absolute scale. For nursing students and practicing nurses reading clinical research, these measures make it easier to assess a study, discuss its relevance to patients, and recognize when a result is too uncertain to guide practice. This article explains the calculations, the interpretation traps, and the preparation needed to use them responsibly.
Key Things You Should Know
- NNT is 1 divided by the absolute reduction in an undesirable outcome; NNH is 1 divided by the absolute increase in an adverse outcome. A 4-percentage-point reduction gives an NNT of 25.
- Always identify the outcome, comparison group, population, and follow-up period. An NNT of 25 over one year does not describe the same effect as an NNT of 25 over five years.
- Neither measure predicts an individual patient's result. Check study quality and uncertainty before weighing an NNT against an NNH or applying either figure in care.
What Do Number Needed to Treat and Number Needed to Harm Mean?
Number needed to treat is the number of people who would need to receive an intervention, rather than the comparator, for one additional person to avoid a specified undesirable outcome during a specified period. Number needed to harm is the number who would need to receive it for one additional person to experience a specified adverse outcome.
Both describe average differences between groups, not which particular patient will benefit or be harmed. A smaller NNT generally indicates a larger absolute benefit for the stated outcome and timeframe. A smaller NNH indicates a more frequent additional harm. Neither number has a useful meaning without its outcome, comparator, and follow-up period: preventing hospitalization, reducing pain, and causing a serious bleed are different events.
How Are Number Needed to Treat and Number Needed to Harm Calculated?
Start with the proportion of participants who experienced the outcome in each group. Use proportions rather than whole-number percentages when taking the reciprocal: 4% is 0.04.
In a hypothetical trial, 10% of the comparison group and 6% of the treatment group are hospitalized within one year. The absolute risk reduction is 0.10 − 0.06 = 0.04. NNT = 1 ÷ 0.04 = 25 people treated for one year to prevent one additional hospitalization.
Suppose the same trial finds a specified adverse effect in 1% of the comparison group and 3% of the treatment group. The absolute risk increase is 0.03 − 0.01 = 0.02. NNH = 1 ÷ 0.02 = 50 people treated for one year to cause one additional adverse event. Conventionally, round a non-whole-person result up to the next whole person. If the difference is zero, there is no finite NNT or NNH; do not report zero as the answer.

How Should Nurses Interpret NNT and NNH in Clinical Research?
Read the figures as a compact description of a specific comparison, then return to the study before drawing a clinical conclusion. This sequence helps prevent a memorable number from obscuring the outcome or the patient population:
- Name the event and its importance: mortality, hospitalization, symptom improvement, and a mild adverse effect do not carry equal weight.
- Check the comparator and duration: treatment versus usual care over six months answers a different question from treatment versus another drug over several years.
- Compare study participants with the patients in your setting, including baseline risk, comorbidities, and relevant access or adherence barriers.
- Examine uncertainty, study limitations, and patient preferences before discussing how the evidence might inform care.
An NNT of 25 and NNH of 50 does not establish that a treatment is "twice as beneficial as harmful." The two numbers may describe events of very different severity, and a patient may experience more than one outcome. Nurses can use the findings in patient education and team discussions while following current protocols and the licensed clinician's patient-specific assessment.
Relative risk reduction compares the change with the comparison group's starting risk. Absolute risk reduction is the difference in event rates between groups; its reciprocal produces the NNT. The same relative reduction can therefore yield very different NNTs when baseline risk changes.
The hypothetical comparisons below each show a 20% relative reduction in the event rate. Their different absolute reductions show why a relative claim alone cannot tell a nurse how many patients might benefit.
| Comparison-group risk | Treatment-group risk | Absolute reduction | NNT |
| 20% | 16% | 4 percentage points | 25 |
| 2% | 1.6% | 0.4 percentage points | 250 |
For an unwanted effect, use the absolute increase in risk to calculate NNH. When applying a trial to a patient group with a different baseline risk, do not assume its published NNT transfers unchanged; the treatment effect and risks may differ too.
A randomized trial can provide a stronger basis for estimating an intervention's effect than an observational comparison, although randomization does not eliminate problems such as missing outcomes, poor adherence, or an unrepresentative sample. In observational research, differences between treated and untreated patients may reflect confounding rather than the treatment. Review how events were defined, how many participants were followed, and whether the reported analysis fits the question being asked.
Confidence intervals show the estimate's uncertainty. For illustration, an absolute risk reduction estimated at 5 percentage points gives an NNT of 20. If its 95% confidence interval runs from 2 to 8 percentage points, the corresponding NNT bounds are approximately 50 and 13, respectively. Those figures should be read alongside the original risk-difference interval, not as equally likely outcomes.
If a risk-difference interval includes zero, its reciprocals do not form one ordinary finite NNT range: no effect is compatible with the interval, as may be an effect in the opposite direction. A precise-looking point estimate is not proof of benefit. Time-to-event studies also require care; use risks estimated at the same stated time point rather than treating a hazard ratio alone as an NNT.

NNT and NNH help nurses appraise an intervention before incorporating evidence into education, care planning discussions, or quality-improvement work. For example, a team considering a fall-prevention approach would specify whether the study measured patients with at least one fall, total falls, or injurious falls. Each can produce a different result, and potential burdens such as reduced mobility matter alongside any benefit.
A practical next step is to record the studied population, intervention, comparator, outcome, follow-up period, absolute event rates, and uncertainty in an evidence summary. Then compare that evidence with local patient needs, available resources, and policy. Do not substitute a published NNT for a patient assessment or treat a change observed in a local project as a causal treatment effect without an appropriate design and oversight.
Course names vary, but nursing research, evidence-based practice, statistics or biostatistics, epidemiology, and population health commonly provide the foundation. Students may encounter NNT and NNH while appraising intervention studies, examining absolute versus relative effects, or assessing whether evidence applies to a clinical population. A course that requires students to calculate event rates and critique a paper offers more direct practice than one that only asks them to recognize terminology.
To prepare, practice converting percentages to proportions and distinguishing a risk difference from a risk ratio. When comparing programs - including lower-tuition BSN-to-MSN education options - look at the actual syllabus or learning outcomes if research interpretation is a priority; a course title alone does not establish how deeply it is taught.
Prelicensure ADN and BSN pathways introduce evidence-informed clinical judgment, while BSN curricula may offer more explicit research-appraisal or population-health coursework; the depth depends on the program. RN-to-BSN and graduate nursing study can build on clinical experience through more advanced appraisal, implementation, or evaluation. Research-focused PhD study goes substantially further into producing original research; a practice-focused DNP generally emphasizes applying and evaluating evidence in practice.
Choose preparation for the task you intend to perform. A practicing RN who needs to interpret NNT and NNH may be well served by a focused evidence-based-practice course, journal club, or mentored appraisal rather than another degree. If broader advancement is the goal, comparing ADN-to-MSN nursing programs can help an RN examine how coursework builds toward a specific graduate role. A graduate degree by itself does not confer a new license, specialty certification, or research competence.
Institutional accreditation, nursing program accreditation, and state board approval serve different purposes. A school's published curriculum and the applicable nursing program standards can help a student determine whether evidence-based practice is taught and evaluated. For a prelicensure applicant, state board requirements and program approval also matter for the intended licensure pathway; research coursework alone does not establish NCLEX eligibility.
Before enrolling, ask the program where students practice appraising clinical studies, whether they work with absolute effects and uncertainty, and how those skills are assessed. Applicants comparing ABSN entry requirements should separately verify the school's prerequisites and its initial-licensure pathway. Neither an accreditation label nor a course description guarantees that a graduate can independently evaluate research in every clinical context.
What Career Benefits Come From Understanding NNT and NNH?
These skills make an RN more useful in evidence-based-practice discussions, patient education, guideline review, and quality or safety work. They also help nurse educators teach students why a large relative effect may translate into a modest absolute benefit. In research support roles, accurately describing outcomes and uncertainty is more valuable than reporting a striking number without its context.
The benefit is a demonstrable skill, not an automatic promotion or pay increase. An RN can show readiness by producing a concise, correctly sourced study appraisal that includes absolute event rates, NNT or NNH where appropriate, confidence intervals, and limitations - without disclosing protected patient information. Nurses considering doctoral study can examine DNP programs organized by completion time, but should assess the project methods, mentorship, and role fit rather than choosing by speed alone.
Other Things You Should Know About Nursing
No. It describes an average difference between study groups. Individual circumstances and preferences still matter in patient-specific decisions.
Not directly from sensitivity or specificity. NNT concerns the effect of an intervention on an outcome; evaluating a test also requires considering how its results change care and patient outcomes.
A mean change alone does not provide an NNT. Defining a meaningful response threshold can make an event-rate calculation possible, but that threshold and the underlying data must be available and justified.
No. Studies may differ in baseline risk, outcomes, and follow-up. An appropriate evidence synthesis analyzes compatible underlying effect estimates and uncertainty rather than taking a simple average of published NNTs.
- Number needed to treat (NNT) in clinical literature: an appraisal | nih.gov https://pmc.ncbi.nlm.nih.gov/articles/PMC5455127/
- Calculation of NNTs in RCTs with time-to-event outcomes | springer.com https://link.springer.com/article/10.1186/1471-2288-9-21
- Number Needed to Treat (NNT) | ox.ac.uk https://www.cebm.ox.ac.uk/resources/ebm-tools/number-needed-to-treat-nnt
- Relative Risk, Absolute Risk, and Number Needed to Treat | WMJ https://wmjonline.org/123no4/calder/
- Problems with NNT - study interpretation | Datamethods Discussion Forum https://discourse.datamethods.org/t/problems-with-nnt/195
- Number needed to treat (NNT) in clinical literature | d-nb.info https://d-nb.info/1136194452/34
- relative and absolute measures in clinical research | Italian Journal of Medicine https://www.italjmed.org/ijm/article/view/2386/2033
- NNT, Eases understanding of evidence | bpac.org.nz https://bpac.org.nz/BPJ/2006/December/nnt.aspx