Journal of the National Cancer Institute Advance Access originally published online on October 7, 2008
JNCI Journal of the National Cancer Institute 2008 100(20):1482; doi:10.1093/jnci/djn311
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© The Author 2008. Published by Oxford University Press.
CORRESPONDENCE |
Re: Randomized Controlled Trial to Evaluate Transdermal Testosterone in Female Cancer Survivors With Decreased Libido: North Central Cancer Treatment Group Protocol N02C3
Affiliation of authors: Statistical Research and Consulting Centre (BJ) and Urology and Sexual Health (SH), Pfizer Global Research and Development, Sandwich, UK
Correspondence to: Byron Jones, PhD, Statistical Research and Consulting Centre, Pfizer Global Research and Development, Sandwich CT13 9NJ, UK (email: Byron.Jones{at}pfizer.com).
Barton et al. (1) presented the results of the analyses of various endpoints from a two-period crossover trial to compare active drug and placebo for the treatment of loss of libido in female cancer survivors. The authors stated that they had used a methodology for their analyses that "encompasses the state of the science for crossover studies." However, the results given in their table 2, and the subsequent conclusions made, are clearly not based on the correct crossover trial analyses.
The correct methodology is described in Chapter 2 of the book by Jones and Kenward (2). Curiously, it is stated by Barton et al. that they have followed such methodology, referring to it in their article as the "sums and differences" analysis. However, what the authors have reported in their table 2 are P values for the comparison of drug and placebo in each period. Such comparisons are based on the variability between subjects and are therefore not correct for the analysis of a crossover trial. The correct analysis is based on the within-subject differences between the second and first period measurements [see Section 2.3 of (2)].
The authors also advocate using the test for a carryover difference as a preliminary test for deciding if the test for drug vs placebo should be based either 1) on only the data from the first period or 2) on the data from both periods. This procedure is seriously flawed, as pointed out by Freeman (3) and further illustrated in Section 2.7 of (2).
REFERENCES
1. Barton DL, Wender DB, Sloan JA, et al. Randomized controlled trial to evaluate transdermal testosterone in female cancer survivors with decreased libido: North Central Cancer Treatment Group Protocol N02C3. J Natl Cancer Inst (2007) 99:672–679. 9.
2. Jones B, Kenward MG. Design and Analyses of Cross-over Trials (2003) 2nd ed. Boca Raton, FL: Chapman and Hall, CRC Press.
3. Freeman PR. The performance of the two-stage analysis of two-treatment, two-period crossover trials. Stat Med (1989) 8:1421–1432. 12.[Web of Science][Medline]
Response to this Correspondence
CiteULike 
Connotea 
Del.icio.us What's this?
J Natl Cancer Inst 2008 100: 1482.
This article has been cited by other articles:
|
|
 |
|
  S. ANTAL, B. SZENDE, J. LENGYEL, and E. J. HIDVEGI Joint Effects of Cigarette Smoking and Irradiation in Pregnant Mice and their Offspring In Vivo, March 1, 2009; 23(2): 267 - 272. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||