Interpreting the number needed to treat for analgesia
The number needed to treat (NNT) calculates how many patients must be treated with an intervention for a given period to achieve a specific effect (eg 30% pain reduction). The higher the NNT, the less effective a treatment is compared to a placebo group. An ideal analgesic would have a calculated NNT of 1 (ie every patient treated achieves the desired outcome compared to none of the placebo-treated patients).
The NNT describes how many patients’ responses are attributable to the intervention rather than the placebo effect.The NNT is affected by the size of the placebo effect; for example, despite no change in efficacy, the NNTs of antidepressants and anticonvulsants have increased over time due to increased placebo responses. This is because more patients expect these treatments to work due to widespread use.
When applying an NNT to an individual patient, consider:
- the calculated NNT and the absolute response rate (ie the total number of patients who respond to the intervention due to the intervention itself plus the placebo effect)
- whether the pain diagnosis, dose used and pain severity in the clinical trial is comparable to the individual patient
- whether the study end-point is clinically significant.
See Example of a calculated NNT compared to absolute response rate for an example of a calculated NNT compared to absolute response rate.
Pregabalin has a calculated NNT of 7.7 (ie approximately 8 patients need to be treated with pregabalin for 1 patient to achieve a 50% reduction in pain that is not attributable to placebo). This corresponds to a response rate of 13%.
In clinical trials, approximately 42% of patients treated with pregabalin report a 50% reduction in pain (ie approximately 2 patients need to be treated with pregabalin for 1 patient to achieve a 50% reduction in pain). This indicates an absolute response rate of 42%. This rate includes both the placebo effect and drug effect experienced in one individual.
The difference in these response rates is attributable to the placebo effect. Large placebo effects in clinical trials will reduce the calculated NNT. Because placebo effects are large in analgesic trials, NNTs are becoming less reflective of absolute response rates.
These considerations are also relevant when interpreting the number needed to harm (ie how many patients must be treated with an intervention for a given period to cause an adverse effect).