Related Experiment Videos
A note on the number needed to treat.
1Biostatistical Centre, Leuven, Belgium. emmanuel.lesaffre@med.kuleuven.ac.be
Controlled Clinical Trials
|September 30, 1999
Summary
The number needed to treat (NNT) is a useful clinical metric, but simple calculations can yield incorrect results. Calculating a combined NNT from meta-analyses is best done by analyzing the absolute risk reduction first.
Area of Science:
- Clinical Epidemiology
- Biostatistics
- Medical Statistics
Background:
- The number needed to treat (NNT) is increasingly used by clinicians to interpret treatment effects.
- NNT is defined as the inverse of the absolute risk reduction (delta).
Purpose of the Study:
- To discuss the mathematical and statistical properties of NNT.
- To demonstrate that simple aggregations of NNTs can produce nonsensical results.
- To recommend the optimal method for calculating combined NNT in meta-analyses.
Main Methods:
- Analysis of mathematical and statistical properties of NNT.
- Simulations comparing different methods of combining NNT estimates.
- Illustration using data from anti-epileptic trials.
Main Results:
- Simple calculations involving NNTs, such as summation, can lead to invalid conclusions.
- Calculating a combined NNT by first combining estimates of absolute risk reduction (delta) is statistically superior.
- Simulations confirmed the better performance of NNT estimates derived from the delta-scale (NNT(P)) over those derived directly from the NNT-scale (NNT(O)).
Conclusions:
- Clinicians should be cautious when performing simple calculations with NNT values.
- For meta-analyses, the recommended approach is to calculate the combined NNT from the pooled estimate of absolute risk reduction.
- This method ensures more accurate and reliable interpretation of treatment efficacy across studies.