Reader Comments

Post a new comment on this article

Correction to my earlier response

Posted by plosmedicine on 31 Mar 2009 at 00:19 GMT

Author: Frank Hummer
Position: Mathematician
Submitted Date: January 24, 2008
Published Date: January 25, 2008
This comment was originally posted as a “Reader Response” on the publication date indicated above. All Reader Responses are now available as comments.

I mispoke in my earlier response (early January). The vegetable and fruit intake in the fifth quintile, Q5, for the women, was the same as the vegetable and fruit intake in Q2 for the men (not Q5 for the men as I had earlier said.) Also, on re-examining the paper, I do see that the authors do refer to the fact that they only reported the difference between men and women where the difference was significant, and I had not acknowledged the fact that the authors had mentioned this. Nevertheless, when I graph the cancer occurances, for each qunitile, for each type of cancer, for the genders separately, the difference in the trends is striking. Eating red meat almost appears to have a protective effect in women, against cancer. I have not tested this for statistical significance. It wasn't clear to me from the paper (and this may be my fault, not the authors' ) precisely what constituted a statistically significant difference between men and women. I speculated that perhaps they limited their reporting of gender differences to only those specific instances in which the following three conditions were met: (1) the connection between meat intake and cancer was significant for men, (2) the connection between meat intake and cancer was significant for women, and (3) the meat-cancer "connections" were in the opposite directions between the two genders. If these were the requirement for reporting gender differences, then that is a tougher test for significance than merely testing for a gender difference per se -- it would be possible for the meat-cancer relationship to be statistically insignificant for each gender separately, and yet for there to be a statistically significant gender difference in the meat-cancer trend (say, a positive correlation for men and negative correlation for women)between the genders. Also, even if the gender differences in correlation was insignificant for particular cancers, it could nevertheless be significant when all cancers are pooled. In order to see what is motivating my questions, you could make two separate excel spreadsheet line graphs, one for men, one for women, where the five quintiles for red meat consumption are represented on the horizontal axis, and the incidence of cancer is represented on the vertical axis. I used a separate line graph for each type of cancer.

No competing interests declared.