Data Interpretation in Clinical Research

The Number Needed to Treat: 25 Years of Trials and Tribulations in Clinical Research

Samy Suissa


The number needed to treat (NNT) is a simple measure of a treatment’s impact, increasingly reported in randomized trials and observational studies. It has been found to be incorrectly calculated in several studies involving varying follow-up times. We discuss the NNT in these contexts and illustrate the concept using several published studies. The computation of the NNT is founded on the cumulative incidence of the outcome. Instead, several published studies use simple proportions that do not account for varying follow-up times, or use incidence rates per person-time. We show how these approaches can lead to erroneous values of the NNT and misleading interpretations. For example, a trial of 3,845 very elderly hypertensives randomized to a diuretic or placebo reported a NNT of 94 treated for 2 years to prevent one stroke, though the correct approach results in a NNT of 63. We also note that meta-analyses involve trials of differing lengths, but often report a single NNT. For example a meta-analysis of 22 trials of the anticholinergic tiotropium in chronic obstructive pulmonary disease reported a NNT of 16 patients “over one year,” even if the trials varied in duration from 3 to 48 months, with the more specifically computed NNTs varying widely from 72, 15, and 250 for the 3-month, 12-month, and 48-month trials, respectively. Finally, we describe the value of the NNT in assessing benefit–risk, such as low-dose aspirin use in secondary prevention, where prevention of mortality was assessed against the risk of gastrointestinal bleeding. As the “number needed to treat” becomes increasingly used in the comparative effectiveness and safety of therapies, its accurate estimation and interpretation become crucial to avoid distorting clinical, economic, and public health decisions.

Rambam Maimonides Med J 2015;6(3):e0033