Premium
The number needed to treat: a useful clinical measure or a case of the Emperor's new clothes?
Author(s) -
Grieve Andrew P.
Publication year - 2003
Publication title -
pharmaceutical statistics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.421
H-Index - 38
eISSN - 1539-1612
pISSN - 1539-1604
DOI - 10.1002/pst.33
Subject(s) - number needed to treat , emperor , bayesian probability , confidence interval , measure (data warehouse) , medicine , clinical trial , actuarial science , computer science , relative risk , artificial intelligence , data mining , history , economics , ancient history
Abstract The number needed to treat (NNT) was introduced into the medical literature as an easily understood and useful measure of treatment effect for clinical trials in which the main outcome variable is binary. It has been argued that it is more easily understood by practising physicians than more statistically based measures. In this paper we review the claims made for the NNT and question whether it is truly understandable to physicians, and look at issues around determining a confidence interval, or a Bayesian interval, for the NNT. Copyright © 2003 John Wiley & Sons, Ltd.