THE BLOG
07/23/2013 02:48 pm ET | Updated Sep 22, 2013

Estrogen and Evidence

Colleagues and I conducted a careful analysis of data reported in a methodologically robust clinical trial and published in a prestigious journal to reach this conclusion: Failure to take estrogen over the past decade has resulted in tens of thousands of premature deaths among women in the U.S. who have undergone hysterectomy. Estrogen significantly lowers the mortality rate among younger women in this group.

Perhaps it is inevitable when you assert that tens of thousands of women have needlessly lost a decade or two of life, and that the death toll is ongoing and probably rising, you will raise some hackles. And, perhaps it's inevitable that when you identify the causes of this public health crisis, the identified causes will push back.

The causes in this case involve no "bad guys," no malice, and no ill intent, however. Rather, we might invoke a confluence of predictably unfortunate happenstance. The findings of large clinical trials make headlines. Headlines are, traditionally, designed for maximal titillation, meaning they are over-simplified and over-generalized as they were in this particular case. And, alas, in the case of hormone replacement, the headlines stuck -- with all subtlety lost. A cultural aversion to hormone replacement overtook patients and clinicians alike. We are simply reporting the unintended consequences.

Peers pushing back against our paper have no real cause to do so. We are not accusing anyone of anything other than the unintended, adverse effects of medicine-meets-media business as usual. We are not refuting the value of the Women's Health Initiative (WHI). Quite the contrary; our data are from the WHI. The WHI has not done harm -- misapplication of the WHI findings has done harm.

We are certainly not saying that every woman with a hysterectomy should take estrogen. Our argument is exactly against the oversimplification of one-size-fits-all approaches. Headlines invite just such dumbing down -- and our very point is that we need to get past the headlines to the relevant nuances. Good clinical care is not well informed by aversions or knee-jerk reactions. It is well informed only by the thoughtful application of the best possible information to a given individual patient's circumstance.

In all of the commentary I have found, I see only one consistent challenge to our paper: It's based on a mathematical model and the assumptions are uncertain. Perhaps, but consider this: Evidence-based medicine, in its entirety, involves extrapolating the results of the best research to the care of patients. That, of course, is far less robust than a formal attempt to link the two in a mathematical model -- it really is just informed observation. Evidence-based medicine comes down to this: "A clinical trial suggests a benefit from X in some people, and it seems to me that you, my patient, are enough like those people to benefit from X as well -- and on that basis, I recommend/prescribe X for you."

Make no mistake, that's exactly what evidence-based medicine is: the application of results in one group (the trial participants) to all patients in the real world who seem to resemble that group. Examining this issue is not a seat-of-the-pants exercise in my case; I have published a textbook on the subject.

Which brings us back to estrogen. My colleagues and I, including Dr. Philip Sarrel, simply took the published results of the estrogen-vs.-placebo branch of the WHI study conducted among women who had undergone hysterectomy, and asked and answered this question: If these results pertain to all women in the U.S. who meet the description of women in the trial, what would it mean? What it would mean is tens of thousands of premature deaths over a decade, just as we reported.

Peers have no business challenging the application of clinical trial data to the world at large in this one instance just because they don't like the implications. If clinical trial data do not pertain to the world at large, then why on earth do we waste time (and fortunes of money) doing clinical trials in the first place? If what happens in the WHI stays in the WHI, if it only pertains to women in the trial, then someone owes the taxpayers an explanation for a boondoggle to the tune of almost half a billion dollars.

But that's not the case. Clearly, the intent from the start was to generate data in this huge clinical trial with real-world application. Clearly, the intent from the start was to extrapolate the WHI findings to all other women like those in the trial. That's exactly what we did, merely adding the rigor of a cautious mathematical model.

If our very straightforward model translating the results of a clinical trial to the real-world population corresponding to the trial participants is really questionable, then it follows that so is everything about evidence-based medicine. Our peers simply cannot have their evidence-based cake and eat it, too.

My colleagues and I stand confidently, if unhappily, behind our conclusion that "estrogen aversion" has killed tens of thousands prematurely, and may do much worse if we don't start approaching hormone replacement as the subtle, individualized clinical challenge it deserves to be. If there really is a problem with this assessment of estrogen, we have a much bigger problem to confront as well: There may be no such thing as reliably evidence-based medicine. Or maybe we just look the other way when the evidence is telling us something we don't want to hear.

-fin

Dr. David L. Katz; http://www.davidkatzmd.com/

www.turnthetidefoundation.org

http://www.facebook.com/pages/Dr-David-L-Katz/114690721876253

http://twitter.com/DrDavidKatz

http://www.linkedin.com/pub/david-l-katz-md-mph/7/866/479/

For more by David Katz, M.D., click here.

For more health news, click here.