The Number Needed to Treat (NNT) is an important concept in evidence-based medicine (EBM). It represents the average number of patients who must be treated with a specific therapy or intervention for one additional patient to benefit compared to a control group (often a placebo or standard of care). The NNT is derived as the inverse of the Absolute Risk Reduction (ARR), providing a practical measure of clinical effectiveness.
In short, NNT represents the number of patients who must be treated to prevent one additional negative outcome or achieve one additional positive outcome.
Absolute Risk (AR): The proportion of patients experiencing an event in a group (e.g., 10% of patients have a heart attack in a control group).
Absolute Risk Reduction (ARR): The difference in the absolute risk between the treatment group and the control group.
Formula: ARR = AR_control - AR_treatment
Number Needed to Treat (NNT): The inverse of the absolute risk reduction.
Formula: NNT = 1 / ARR
For example, if the incidence of a particular outcome (e.g., myocardial infarction) in a control group is 10% and the incidence in the treatment group is 5%, the ARR is 0.05 (or 5%).
Thus, the NNT is:
NNT = 1 / 0.05 = 20 In this scenario, twenty patients must be treated for one patient to avoid a myocardial infarction who would otherwise have had one under the control condition.
Lower NNT: Generally indicates a more effective treatment, as fewer patients need to be treated to observe one beneficial outcome.
Higher NNT: Suggests that the intervention is less effective on average, or that the baseline risk is low, requiring more patients to be treated before seeing one additional benefit.
It is critical to interpret NNT values in context. Even a high NNT might be considered acceptable if the condition is serious, the intervention is low-risk, or the cost is minimal. Conversely, an extremely low NNT for a treatment with significant side effects or high cost may still require careful risk-benefit analysis.
Below are illustrative examples (approximate values) of interventions with relatively low NNTs:
Antibiotics for Bacterial Pneumonia: NNT can be as low as around 5 for preventing serious complications in susceptible populations. The high impact of appropriate antibiotic therapy in an acute setting leads to a favourable NNT.
Smoking Cessation Interventions in High-Risk Patients: Intensive behavioural therapy combined with pharmacotherapy (e.g., nicotine replacement, varenicline) might have an NNT between 5 and 10 for achieving long-term abstinence. The benefit of reducing morbidity is substantial, given the high baseline risk of smoking-related diseases.
Vaccinations (e.g., Influenza in Elderly or High-Risk Populations): Depending on the population, seasonal influenza vaccination can have an NNT ranging from around 12 to 40 to prevent one case of influenza, though these numbers vary widely by season, vaccine match, and patient comorbidities. During severe flu seasons or in particularly vulnerable groups, the NNT can be pretty low.
Some therapies or preventive measures show benefits only when administered broadly or for long durations; these often come with higher NNTs:
Statins for Primary Prevention of Cardiovascular Disease: In individuals without established cardiovascular disease (low to moderate risk), the NNT for statins to prevent one cardiovascular event over a period of five years can be in the range of 50 to 200, depending on risk factors. Despite a high NNT, statins may still be recommended due to their relatively favourable safety profile and the severity of cardiovascular events.
Aspirin for Primary Prevention of Cardiovascular Events in Low-Risk Patients: For individuals at low cardiovascular risk, the NNT to prevent one heart attack or stroke can be over 100. However, this must be weighed against potential harms such as gastrointestinal bleeding.
Screening Mammography for Women Aged 40–49: Estimates for the NNT (to prevent one breast cancer death) can be in the hundreds when screening is applied to younger, lower-risk populations. As age and risk factors increase, the NNT typically decreases.
Context-Dependent: NNT depends heavily on the baseline risk of the population. An intervention that yields a high NNT in a low-risk population can have a significantly lower NNT in a high-risk population.
Time Horizon: The timeframe over which the benefit is observed affects the NNT. Longer durations can change both the ARR and the overall interpretation of an intervention’s efficacy.
Benefit vs. Harm: Treatments with low NNT but high side-effect profiles may be less favorable than treatments with higher NNT but minimal adverse effects.
Patient Preference: Shared decision-making is essential. Different patients may weigh benefits and risks differently, and an NNT value alone does not fully account for individual preferences or values.
Composite Outcomes: Studies sometimes report composite endpoints (e.g., “major cardiovascular events” that include several conditions). Such composites can alter ARR and thus the NNT. Breaking down endpoints is often necessary for accurate clinical interpretation.
Cook RJ, Sackett DL. “The number needed to treat: a clinically useful measure of treatment effect.” BMJ. 1995;310(6977):452-454.
Walter SD. “Number needed to treat (NNT): estimation of a measure of clinical benefit.” Stat Med. 2001;20(24):3947-3962.
Furukawa TA, Leucht S. “How to obtain NNT from cochrane, other systematic reviews and meta-analyses.” Evid Based Ment Health. 2011;14(3):67-71.
Helfand M, Peterson K. “Preventive Services in the Age of Evidence-Based Medicine: The Case of Aspirin.” American Journal of Preventive Medicine. 2009;36(2):138-141.