What is Number Needed to Treat (NNT)?
Number Needed to Treat (NNT) is the number of patients who need to be treated with a specific intervention to achieve one additional good outcome (or prevent one bad outcome) compared to a control group. 1, 2, 3
Mathematical Definition
NNT = 1/absolute risk reduction × 100, where absolute risk reduction is the risk difference between two event rates (experimental and control groups). 1, 2, 3
Alternatively expressed: NNT = 100 divided by the absolute risk reduction expressed as a percentage. 4
Clinical Interpretation
What the Numbers Mean:
- Lower NNT = More effective intervention
- NNT < 50: Generally considered acceptable to physicians 4
- NNT < 30: Generally considered acceptable to patients 4
- Higher NNT = Less efficient intervention (more patients must be treated to benefit one person)
Key Components Required for Calculation:
To properly calculate and interpret NNT, you must know: 4
- Baseline risk of the outcome
- Time horizon over which the outcome is measured
- Amount of risk reduction achieved by the intervention
- Confidence intervals (95% CI) for precision of the estimate
Critical Limitations and Pitfalls
Calculation Issues:
- NNT becomes distorted when placebo effects approach treatment effects, potentially yielding infinite values 5
- Requires dichotomous outcomes - continuous variables must be artificially dichotomized, and the NNT changes depending on where you set the cutoff 6, 5
- Confidence intervals are difficult to estimate accurately 5
Context-Dependent Nature:
NNT is heavily influenced by: 6, 7
- Baseline disease or symptom severity
- Specific population characteristics
- Type and intensity of intervention
- Duration of treatment
- Time period to assessment
- Choice of comparator
The same intervention can have vastly different NNTs in different populations or settings. 7
Reporting Problems:
In published literature, many studies fail to report NNT properly: 8
- 27.5% don't report baseline risk
- 25.5% don't report time horizon
- 37.3% don't report confidence intervals
- 29.4% use calculation methods not aligned with basic methodological recommendations
- Meta-analyses are particularly problematic, with 56.5% applying non-recommended methods 8
Balancing Benefit and Harm
NNT should always be balanced against Number Needed to Harm (NNH), which quantifies adverse effects. 4, 9 The clinical decision depends on whether the benefit (NNT) outweighs the harm (NNH) for a given intervention.
Practical Example
In the IMPROVE-IT trial for cardiovascular disease: 4
- NNT = 50 for 7 years of treatment (or 350 for 1 year)
- This means 50 patients needed treatment for 7 years to prevent one cardiovascular event
NNT is not static - it changes over time but will eventually plateau, not improve indefinitely. 10