5 Comments

1. Tori on November 23, 2010 at 7:38 am

   I’m a little confused by the 2 groups described in the study abstract. There’s “a history of irregular cycles” and “a regular cycle length of 27 to 29 days.” Which are both fine and good groups, but they don’t seem to account for all of the population; there are a fair number of people who have regular cycle lengths but who are outside that (pretty narrow) 27-29 day window.

   And while I understand that the researchers may have had to simply things for the purposes of their study, I feel like that simplification may limit the practical applications this study has for some of us.

2. Pam on November 23, 2010 at 3:35 pm

   This is so frustrating. Irregular menstrual cycles associated with polycystic ovaries are no doubt correlated with diabetes, and there may be a larger pattern there not fully understood. But that there is a correlation for some people with irregular cycles doesn’t mean that there is a universal correlation for women with irregular cycles. Being two standard deviations from the norm by itself doesn’t mean something is wrong–variation is normal! So many women are on the pill that irregular cycles now seem weird, but they used to be common. I went through my reproductive life with cycles from 45 to 60 days, and got pregnant the first month of trying at 34 years of age.

3. Elizabeth Kissling on November 23, 2010 at 3:52 pm

   Pam and Tori, I think you’re right – regular doesn’t have to mean the mythical average 28-day cycle! The actual study does seem to acknowledge both short, regular cycling and long, regular cycling.

   Would either of you like to write a more comprehensive review of this article for re:Cycling? I have a PDF I can send. It’s not very long, but the statistical analysis is beyond my expertise.

4. Paula Derry on November 23, 2010 at 4:40 pm

   This raises issues for me about general issues in making sense of research. It seems like we should be able to read a research report and take what is said at face value. Unfortunately, this is not the case. Making accurate inferences about what research shows is a complicated business. Here, the authors report a “tendency” towards an increase in diabetes risk–a “tendency” is at best a suggestion, not valid data. That is, the authors say in the article itself that the findings for type 2 diabetes were not “statistically significant”; that means, by definition, the results might have occurred purely by chance. This study, also, used what is called relative risk rather than absolute risk of CHD and diabetes. An increase of a third in CHD risk may sound like a big effect. However, the vast majority of women did not have a CHD problem, and factors other than cycle irregularity were first taken into account before looking at cycle effects, so that an increase of a third affects a very small percentage of women with cycle irregularity. On the one hand, this still means there was for CHD a statistically significant association. On the other hand, what this does mean is not clear. This is a prospective cohort study, which means it shows a correlation, not causation; the cycle is “associated with” health outcomes, but we don’t know what this association means, if anything. An example of correlation (vs. causation) is that eating ice cream has increased over historical time, as has crime rates, but this doesn’t necessarily mean that eating ice cream is a window into crime. There are other issues—when researchers do a large number of comparisons, as was the case here, a few will end up statistically significant purely by chance; the authors relied on subjects’ memory of cycle characteristics, etc. Do I believe that the menstrual cycle is part of a complex, whole-body system? Yes. However, studies like these raise more questions than they answer, and provide, at best, food for thought.

5. Tori on November 23, 2010 at 6:28 pm

   Elizabeth, I’d love to, but I’ll also admit that I’m no expert on statistics. If you can send me the full text of the study, I can send you a draft, though.