The Women’s Health Initiative (WHI) was published in 2002. I presented a critique of the study in 2003. I gave a lecture in different parts of the country at that time debunking its statistical significance.
No one paid attention to the statistical results. The media was the message and the message has lasted despite good data refuting the message.
I published these thoughts once before in a 2007 blog post. I was stimulated to republish my comments last week because of two excellent lectures I heard at a recent American Association of Clinical Endocrinologists (AACE) meeting reviewing the recent literature as to the value of estrogen replacement therapy in post-menopausal women.
Holly L Thacker M.D. presented “Women Getting Older-and Even Better”.
Dr. Thacker is the Director of the Center for Specialized Women’s Health at Cleveland Clinic.
- Mitchell Harman, M.D. presented “Hormonal Treatment of Menopausal Women: What Are The Data Telling Us (and Not Telling Us)?”
Hopefully these two presentations will help us move forward in getting to the truth about hormone replacement therapy in the post menopausal woman.
Neither presentation stated outright that the Women’s Health Initiative conclusions were invalid because of the fact that none of WHI conclusions were statistically significant.
However, several important new points were added to the discussion using the WHI’s own data. I will present the discussion.
First, I want to present an email exchange a reader had with a retired statistician about my last blog.
Mon, Jun 6, 2016,
The blogger writing below is a now retired endocrinologist, academician and oft critic of ObamaCare.
In his post below he takes on many of the reports that were published based upon the WHI and related.
Passing it along to you FYI owing to his assertions as presented below.
Perhaps instructive after all.
I sent your post on the flaws of WHI data and interpretation to a retired statistician for comments.
He applauds your evaluation and comments. As do I.
Sent: Monday, June 06, 2016
Subject: Re: FW: Repairing the Healthcare System
The points made are good ones. The major point missed is that the study was done with horse piss estrogen, not human biologically identical estrogen which have been shown to produce none of the negative effects of horse derived estrogen.
None the less, the scientific criticism of the reported study are very accurate and very good.
In 2011 AZ LaCroix using the WHI’s own data analyzed the myocardial event risk. Women talking estrogen between ages 50-59 had less risk of having a myocardial infarction than women starting to take estrogen replacement therapy from age 60-69 and from age 70-79.
The WHI lumped conclusions lumps all three groups together in its analysis.
It is well known and accepted that the incidence of myocardial disease increases in women as they age and increases in post menopause to the same incidence as males.
The timing hypothesis for starting estrogen therapy was proposed as a result of this new interpretation of WHI data.
The effects of HRT on CVD are dependent upon time HRT is initiated relative to menopause and/or age (the “timing hypothesis”).
The upper two curves reflect the thickness in the carotid artery tested with ultrasound in women starting to take estradiol ten years after menopause. There is not a significant difference in the increase in intimal thickness between the placebo group and the treatment group (p==.029). Significant p values are 0.05 or less (i.e. 0.001).
Whereas the lower curves where estrogen started in less than 6 years post menopause the placebo thickness increased greater than the estradiol thickness for a very significant p value of 0.007.
This data reflect the influence of estrogen on vessels when taken shortly after the onset of menopause. The result presumes the protective effect of estrogen in post-menopausal women as opposed to the conclusions of the WHI.
I believe that WHI stimulated fright that estrogen causes breast cancer has been exaggerated. Only 4% of the causes of death in women are the result of breast cancer.
It is true the WHI was published in 2002. Prior to 2002 early detection of breast cancer with mammogram and treatment with surgery and chemotherapy was prevalent. It certainly resulted in a decrease in the incidence of death from breast cancer. Today the incidence of death from breast cancer might even be lower.
There is a sense that the breast cancer death rate might be even lower if estrogen is used immediately in postmenopausal women. There has been an 80% decrease in women being treated with hormone replacement immediately post menopause.
The goal should be to lower the death rate from breast cancer to zero.
The huge death rate from cardiac disease in women easily surpasses the death rate from breast cancer.
There is evidence that estrogen replacement might serve to lower the 4% death rate from breast cancer if given appropriately as opposed to the message that estrogen raises the breast cancer death rate as the WHI concluded.
Eliminating patients with Breast Cancer genetic makers from receiving estrogen replacement would lower the death rate even further.
The subgroups in the WHI Estrogen only arm were analysized in 2004. Please note that all subgroups at all the diseases catagories crossed 1 and is not statistically significant. Only the incidence of stroke in the 60 – 69 group did not cross 1.
Therefore none of the subgroups except the 60-69 stroke subgroup was statistically significant. However that group did not attain a hazards ratio above 2 required for it to be statistically significant.
Importantly, the Breast Cancers subgroups all crossed 1 indicating there was no difference statistically significant between estrogen treated age groups and placebo age groups. It cannot be concluded from the WHI study that estrogen causes breast cancer.
Colditz,in the Nurses Health study did not use nominal confidence limits as did the WHI in its conclusions. Colditz considered all of the confounding variables such as age, age of menopause, BMI, hysterectomy or normal onset of menopause, BMI, blood pressure, cholesterol level, smoking, oral contraceptive use, and family history of cardiac disease, or breast cancer.
He used adjusted confidence limits required to be used with confounding variables.
The confidence interval for the occurrence of breast cancer in nurses currently using estrogen was 0.59- 1.00 and using estrogen in the past was 0.63-1.09 both touching or crossing 1. Therefore the results were not statistically significant.
The game changer after all the evidence that the WHI data was misinterpreted was the Danish study DOPS published in the British Medical Journal in 2012.
I have presented this data to my readers to ponder after I heard these two excellent reviews at the American Association of Clinical Endocrinologists (AACE) meeting.
I wanted to point out once again that the media is the message. The media without proper peer review of data has changed the way women are treated post-menopausal forever.
Hopefully disseminating this data will help remove some of the emotional stigma that has influence the thinking and use of estrogen replacement therapy since 2002.
It might stimulate the medical profession, the government, the malpractice legal system and women to start re-thinking their recommendations and conclusions.
This is especially true when women are living longer and estrogen therapy can alleviate some of the emotional, and physical effects women suffer in menopause when estrogen is prescribed appropriately.
The opinions expressed in the blog “Repairing The Healthcare System” are, mine and mine alone.
All Rights Reserved © 2006 – 2015 “Repairing The Healthcare System” Stanley Feld M.D.,FACP,MACE