Response to Senn Paper (Drug Information Journal, 2001;35:1479-1494)

摘要

In a recent paper (1), Senn is very critical of three analyses of clinical data sets purported to make statements about assessments of benefit from therapy at the individual level. Included in these examples is a report of ours (2), in which we discuss the estimation of the number needed to treat (NNT) index. Senn argues that our calculations are inappropriate because they do not take into account the possibility of subject-by-treatment interaction. Using data from a crossover study (3), Senn argues that our method of calculating NNT leads to inconsistent and inappropriate results. We wish to make two brief rebuttals to Senn's paper.