Cancer Epidemiology Biomarkers & Prevention Vol. 12, 586-587, June 2003
© 2003 American Association for Cancer Research
Reply
Joel L. Weissfeld and
Robert E. Schoen
Department of Epidemology and Division of Gastroenterology and Hepatology, University of Pittsburgh, Pittsburgh, Pennsylvania 15213-5582
Global tests of general association do not show statistically significant associations between sex-specific VAT quartile and IGF measures, dichotomized as sex-specific median cut-points. Scatter plots do not provide strong evidence for an inverse U-shaped association between IGF-I and VAT, in men (Fig. 1)
or women (Fig. 2)
. The elevated odds of high median IGF-I in women with second quartile VAT relative to women with first quartile VAT probably represents a chance occurrence related to the specific cut-points chosen to convert variables (e.g., IGF-I and VAT) measured on an interval scale into analytic variables measured on an ordinal scale.
View larger version (15K):
[in this window]
[in a new window]
|
Fig. 1. Association between IGF-I and VAT, in men (n = 267), with least square quadratic regression curve (F value = 0.28, 2 degrees of freedom; P = 0.76).
|
|
View larger version (14K):
[in this window]
[in a new window]
|
Fig. 2. Association between IGF-I and VAT, in women (n = 165), with least square quadratic regression curve (F value = 0.59, 2 degrees of freedom; P = 0.56).
|
|
Footnotes
The abbreviations used are: VAT, visceral adipose tissue; IGF, insulin-like growth factor. 
Received 2/28/03;
accepted 3/19/03.
