To the Editor: We have noted that the Journal currently publishes articles on cancer treatment that include comparisons of the survival of patients who respond to treatment with that of patients who do not respond. A recent example is an article by Kiang et al. (Nov. 14 issue).1 However, various authors have recently shown that the usual method of comparing survival according to response category produces survival plots that are biased, statistical tests of significance that are invalid, and conclusions that are misleading.2 3 4 5 6 7 8 The usual method of comparing patients who respond with those who do not produces biased survival plots.
ASJC Scopus subject areas
- General Medicine