Post Menopause Female Hormones               Main Menu              Health Library

USE of FEMALE HORMONES after MENOPAUSE and HEART DISEASE - A Supplementary Analysis

This section provides a Supplementary discussion of the Life Ahead analysis of the  risk of Female Hormones on coronary heart disease. Please see the main review of female hormones on all major disease risks for the picture on overall hormone risks.  It is hoped that this  more technical review will be useful because the confusion that developed after publication of the WHI study of hormones in 2002. This is included at  www.lifeahead.net/female-hormones.htm

Before discussing the research results of hormones on heart disease it is useful to review the overall or macro BioChemical process by which estrogen probably affects the risk of this all important disease. Estrogen does not change risk of the heart disease immediately but acts over many years to gradually SLOW the development of the atherosclerosis that contributes to this disease. The potency of LDL cholesterol in producing atherosclerosis appears reduced by the antioxidant property of estrogen. This slowing can explain why women’s risk of disease is much lower than that of men during the pre-menopausal years, and may be the key reason why women usually live longer than men. Two columns in the table following shows risk of Heart Disease for non-smoking US women and men by year of age. The right column shows women’s risk as a percentage of that of men.  

The coronary disease risk of boys and girls is similar and tiny.  But after the first menstrual cycle at about age 12 the risk of women as percent of that of men moves down steadily during the child-bearing years from near unity to 56% at age 25 to only 25% at age 45. The observation here is that Estrogen operates - as expected from chemistry - in combination with years of exposure.  It can be shown from this comparison that each year of exposure to Estrogen reduces the risk of heart disease by a factor of about 4% or a factor of 0.96. This effect of years of exposure is confirmed further by the fact that both earlier date of first period and later date of menopause - each of which increases duration of exposure of estrogen - reduces risk of heart disease.   Thus the effect of Estrogen becomes the factor 0.96 to the power of estrogen-years of use. Using this simple formula for the 20 years of Estrogen exposure between age 25 and age 45 would compute a reduction in risk of 0.96 ^ 20 or 0.44.  Multiplying the 56% value at age 25 by 0.44 brings the estimate for age 45 to the 25% actual in the above table at age 45.  This simple formula explains the difference in risk between men and women at each age or age interval before menopause.  After menopause at about age 50 the level of Estrogen in women is greatly reduced. The coronary risk of women then starts moving back up toward the level of men.

This “Times-Years” effect of Estrogen is a key to understanding how Estrogen affects the risk of heart disease. This Times-Years effect also explains how Estrogen increases the risks of some causes of cancer, how smoking changes risk, and how antioxidants reduce risk of heart disease.  And this same behavior as will be shown explains the effect of Estrogen in reducing risk of heart disease during the post-menopausal years. Studies H1 through H16 in appended Table 1 are the results of the 15 key published studies of post-menopausal use of Estrogen on heart disease and one meta or summary analysis of 27 different mostly earlier studies. The time of hormone use was not documented in many of these studies, but information available suggests that women users taken from the population usually had used Estrogen for 7-10 years at start of the studies, and then used it for additional time during most studies. Nearly every study where Estrogen use probably was for 5 or more years showed a substantial reduction in risk of heart disease.  The average risk ratio in this extensive mass of research data on many thousands of women where Estrogen was used for 5 or more years was about 0.60, a reduction in risk of 40%. The earlier Meta study (see H6) of Stampfer found a similar average risk ratio of 0.56 for use of Estrogen – and use in most of these studies probably was well over 5 years. No difference in results for estrogen alone vs Estrogen plus Projesterone were found for coronary heart disease.

Also note that every study that involved shorter term use of Estrogen obtained much differing and higher risks for its use. (These are noted by * ) Further, every study that examined the effect of time of estrogen exposure – as studies H5, H9, H15, and H16 showed that coronary risk dropped steadily with increase in time of Estrogen exposure.

Studies H11 and H16 were randomized clinical studies that involved testing different groups with a medication and a placebo.  Other studies were observation studies that measured results on groups that either used or did not use hormones over some extended time. Studies H1 through H11 were for people free of heart disease, studies H12 through H15 were for people that had and survived heart disease.  A statistical analysis was made to answer objectively the question “What factor caused these risk factors to differ. Was it 1. Estrogen time of exposure; 2. Study type, clinical or observation; or 3. Coronary free women vs coronary survivors?  Formula (1) appended provides an answer. The effect of time of exposure was highly significant, with the ‘t’ statistic of over 4 identifying a significance of above 1000 to 1. There was no difference in results between the clinical and observation type studies, or for the coronary free or coronary surviving individuals. The ‘t’ values of only 0.4 indicate near zero significance for study type or presence of previous disease.  This effect of time of exposure was found usual in many other Life Ahead analyses of health factors.

A refinement of Formula (1) suggests that the time of exposure effect of estrogen is not linear.  Rather it approximates as in Formula (2) appended the factor 0.93 ^ years-of-use (or 0.93 to the power of years of use) in early years but develops as a lesser compounding value of about 0.96  after five years of use.  This latter value of 0.96 per year is the same value found for its ‘times-years’ effect during the most of the pre-menopausal years.   Thus estrogen operates to gradually reduce risk of CVD diseases to a similar extent during post-menopausal years as it does in the pre-menopausal years. This value may decrease further for women's years beyond the menopause.

Studies A1 thru A4 in Table 1 show the rate at which atherosclerosis developed in women exposed to and not exposed to post-menopausal hormones. The first three studies that involved women that used hormones over a usual longer time found that hormones slowed the rate of atherosclerosis about in half – a risk factor averaging about 0.5.  Study A4 done for shorter time did not find an effect within its fairly substantial margin of error. The important observation here is this:  The reduction of about 50% in atherosclerosis from hormones is nearly the same as the usual reduction in coronary risk from the population studies.  This confirms again the hormones were accomplishing their benefit via reducing atherosclerosis, a biochemical process that moves slowly over decades of time.

Although the long term benefit of hormone use appears clear, there is evidence that during their first 2-3 years of use hormones will increase risk of heart disease. This is suggested by studies H11, H15, and H16.  This is not entirely surprising because postmenopausal hormones also have been noted to increased venous thrombosis or the clotting tendency of blood. Any enhancement of blood clotting increases risk of heart disease near immediately.  Thus female hormones appear to cause a harmful short term effect of increased blood clotting that increases coronary risk.  This negative effect eventually is erased and more than offset by the gradually increasing benefit from reduction of atherosclerosis.  Formula (2) suggests an average risk factor of 1.30 and 1.07 during a first and second year of hormone use, with net benefit starting in year three.  But this early-years behavior may depend on type of hormone use.

During 2002 a new study called WHI was given major media attention that that upset doctors and millions of their patients.  It failed to confirm the key health benefit of hormones of reducing risk of heart disease.  Heralded by the media including Time magazine as “The largest, most important of studies now shows hormones to increase rather than decrease risk of heart disease”, this media report simply was not accurate.  First, WHI was not the largest study.  It actually was a rather small one. Compared with its only 164 events (later extended to 188 events) of heart disease for the key group and 122 in the control group study H4 included 1800 events, study H8 had 770, study H9 had 2800, and the Nurses study H10 had 1258 events..  Second, the risk value from the WHI (Study H11) was too widely different from the results of at least 30 other studies to be credible without some further supporting evidence..  Compare the risk value of 1.29 with values usually closer to 0.5-0.6 from other studies of more than 5 years duration and their margins of error. . Please study the results of the actual research in Table 1. The statistical theory supporting WHI was that the ‘Clinical’ type study is the definitive one, others are suspect.  But note that another clinical study called HERS showed the same time-depended effect found from all other studies, with a substantial reduction in coronary risk to the usual value of about 0.6 by year 5 of use.  And when subjected to specific objective statistical analysis no difference was found between Clinical and Observation study results either here or for studies on vitamins and other factors after recognition of the key effect of duration of exposure. 

A further problem with this WHI study was that it studied women of average age 63 at outset.  This was not typical of the usual practice of women that consider its use at end of menopause at a typical age of about 50. Thus the study was for women that probably had much lower levels of natural estrogen than was usual for women using hormones. These lower levels of estrogen at start may have contributed to the higher thrombosis noted in the study.

Appended is a rough plot of the research results on risk of hormones on heart disease.  This shows the clear trend of risk values with probable estrogen time of use. A comparison of the square and circle points shows no clear difference in results between clinical and observation type studies once the time trend is recognized. The plots show no difference in risk levels of coronary free or coronary survivors. And the departure of the WHI study point (lined square) from the average correlation line is quite substantial.

Although short duration of Estrogen exposure explains much of the difference from previously measured risks found in the WHI study, two problems remain.  First, no effect of duration was cited for this study.  Second, WHI measured risk is more than 50% higher than that expected after recognizing the effect of duration.  In answer to the first, it seems likely that the usual effect of duration was obtained in the WHI study.  The small numbers of cases within individual years produce a problem of "Statistical noise".  A method for eliminating this 'Statistical noise" is to plot the events in the hormone and placebo groups vs time and take values vs. time from the smoothed plotted risks to simulate results of a far larger study.  This method suggests that risks probably were declining rapidly with time.  These computed risks shown in the the notes of WHI Study H11 could have been 2.0 in year 1; 1.57 in year 2; 1.39 in year 3; 1.17 in year 4; and 0.90 in year 5. But these risks confirm further the second problem that WHI risks of heart disease were higher than expected - and that 5+ years of use, and  not just the usual 3 would be needed for risks to move into benefit.  It also is possible that as noted before the specific estrogen Premarin used in this study had increased clotting behavior. 

Life Ahead attempts to develop best valuations from all research evidence found, with emphasis on most probable global causes of the development of major disease. Thus Formula (2) based on all of the major research evidence now is used to value the effect of post-menopausal hormones on heart disease. A minimum risk of 0.45 is used for long periods of hormone use.   WHI results on cancer are generally consistent with those from other research as is noted in the general review of post-menopausal hormones..

Formulas for Heart Disease:

(1)  Log of risk ratio = 0.0388 - 0.0844 * years of use + 0.07 * had CVD + 0.059 * Clinical study

      t  of years = 4.4  (coeff 0.0844 (5%-95% limits= -0.046 to -0.122) .  Other t values = 0.4 each   

   Shows no difference between clinical and observation type studies, and no difference for coronary and coronary

   free women.  Key effect is and only is duration of hormone use.  See plot following

    (2)  Log of risk ratio = 0.3735 – 0.2225 * years of use + 0.011 * (Years of Use) ^2

                                                      t = 4.8                             t = 2.7 

                                       Table 1:   Female Hormones and Coronary Heart Disease