Many of us do our own health research, either because we have a specific question or simply to keep up with the news. If we don’t read the original scientific articles, we rely on experts to provide summaries in newspapers, magazines, or on a variety of websites. It seems as though by choosing sources judiciously we should be able to count on finding information that is accurate. However, relying on authority, whether this authority derives from a writer having scientific or medical training, or the writer being a professional journalist, or some other reason, is not enough.
I thought about this recently when I saw an article on Medscape, a website for health professionals, especially physicians, called “Early menopause doubles CVD risk regardless of race.” A summary of a new journal article, it was highlighted on the Medscape home page for many days. It began: “Women who experience early menopause–before their 46th birthday–are twice as likely to suffer from coronary heart disease and stroke as women who don’t enter menopause prematurely, and this finding is independent of traditional risk factors.” Johns Hopkins University, where one of the authors is employed, issued a press release entitled “Early Menopause Associated With Increased Risk Of Heart Disease, Stroke” which also begins: “Women who go into early menopause are twice as likely to suffer from coronary heart disease and stroke, new Johns Hopkins-led research suggests.” Similar articles appeared in Medline+ (a National Institutes of Health and National Library of Medicine website), a Blue Cross Blue Shield healthcare news website, and many print newspapers.
So, what was in the original scientific article? The article was published in Menopause, which, like many journals, does not post its articles free online for non-subscribers. Many academic libraries do not carry this journal. However, if a reader does get the original article, these are some of the details: The women in this research were studied for a number of years. The researchers collected information about many predictors of circulatory problems (smoking, diabetes, etc.). The women were also asked at what age they had reached menopause. If this was when they were younger than 46, they were classified as having an “early menopause” whether menopause was caused by surgery (ovaries removed) or occurred naturally. The researchers looked at whether the women developed heart problems or strokes, and created mathematical models to study which predictors of these problems were important.
Twice the number of women with “early menopause” had heart problems compared with women who reached menopause later. This is what is called “relative risk.” The “absolute risk” numbers were: 3% of the women with early menopause had heart problems compared with 1.4% of those who did not; for stroke, the numbers were 2.6% vs. 1%. This is still a difference, but not as dramatic as a twofold increase. In addition, the way the strength of the association was mathematically computed was to first predict heart problems and stroke with more usual predictors: age, risk factors like diabetes. The difference in risk due to menopause was in the uncertainty left after all these other factors had already been taken into account. Further, we don’t know whether the “early menopause” group had other associated characteristics leading to a health difference—if they were unhealthy in other ways. The authors, for example, state that if a woman had a family history of heart problems, and if this was mathematically taken into account before looking at menopause, then early menopause was no longer a predictor of her having a problem. In accounting for results, the article cannot distinguish between surgical and natural menopause, which differ in many ways.
It is true that, in the media accounts of this research, if a reader reads the entire article, qualifiers do appear embedded in the article in some of the sources. Some do say that the number of women in the study who developed heart problems or strokes was small; that this was a correlation, not a cause-and-effect association; or that when family history of cardiovascular disease was taken into account the relationship disappeared (although in Medscape, the author of the study was quoted as saying that “the pattern was still similar”). A piece of misinformation that reappeared in some of the sources was that the increased risk was similar whether the women had early menopause naturally or because their ovaries had been surgically removed. The research article clearly states that the authors did not have sufficient power (in research this means, basically, enough subjects to get an accurate answer to the question) to determine this.
I was puzzled why so much publicity was given to this study. In my opinion, it did provide some interesting, suggestive results and contributed information about women from a range of ethnic groups (who have been understudied in the past), but the study’s results were modest and inconclusive. However, what the article did do was to claim to support the underlying assumption that menopause and heart disease are related, an idea that keeps re-occurring in the professional literature, even stated as though it is a fact, although the evidence for it has been at the very best arguable and weak. A recent SMCR blog post by Chris Hitchcock analyzed media misreporting of the results of another research project intended to test this relationship. In the study I am discussing, highlighting weak data that seems to suggest a relationship between menopause and ill health, blurring the distinction between natural and surgical menopause, contribute to this meta-message. Ages 40 to 45 would be considered within the normal age range for menopause by many professionals, but is here defined as creating health risks. I would hate to think that meta-messages promoting ideas that menopause is unhealthy and causes risk of heart disease contributed to the perceived importance of the article.
Reference
Early menopause predicts future coronary heart disease and stroke: the Multi-Ethnic Study of Atherosclerosis. Melissa Wellons, MD, NCMP, Pamela Ouyang, MBBS, Pamela J. Schreiner, PhD, David M. Herrington, MD, and Dhananjay Vaidya, PhD, Menopause: The Journal of The North American Menopause Society, 2012. Vol. 19, pp. 1081-1087
This is a very valuable commentary in that it provides important information on the topic of menopause while putting it in the context of media selectivity and slant when the original research gets reported and filtered through the sieve of second-hand sources, many of which are inadequately informed or putting it in the context of social assumptions, fears and sterreotypes.
Thanks!
Exactly, David. I see this happening with media stories on every aspect of women’s health. The people reporting don’t have deep enough knowledge to do much more than report what the researchers want them to report.
Good summary, Paula.
The other thing to know is that the person writing the headline is generally an editor, not the author of the article, and sometimes the title truly misrepresents both the content and the emphasis of the article itself. Headlines are marketing strategies to draw the reader in, even more so now that people scan headlines before they click, and success is determined by clicks.
In a study such as this, it is also important to remember that an association is just that, an association. An association between A and B may be because:
– A causes B
– B causes A
– A and B are causally unrelated, but both are caused by some third factor, C
In this case, both medical articles and media coverage tend to argue for a causality from early menopausal age to cardiovascular disease.
However, I suspect that women in poor health may go through menopause at an earlier age than they would otherwise do if their health were poorer. That is, I suspect that the cardiovascular disease may drive the earlier age at menopause.
Also, I believe that most ovariectomy happens with hysterectomy for preventative reasons (you won’t be using them any more, and if we take them you want get cancer), and the major cause of hysterectomy is to resolve issues with heavy menstrual flow. This is rcognized as an unnecessary surgery. But the peak age for hysterectomy tracks the age of final menstrual period, just s fe years earlier. So mowt likely women who have hysterectomies tend to be a few years away from their final menstrual period.
Anyway, interesting article, and here are some more thoughts to add.
Thanks, all, for your comments. I think, especially, that the confusion between correlation and causality wasn’t laid out clearly enough in my blog, but all of the other factors you’ve all pointed to that contribute to less-than-objective reporting are important.