Hormone replacement therapy and the risk of venous thromboembolism. Three studies from the Lancet.

1. Risk of venous thromboembolism in users of hormone replacement therapy

Authors: Daly E, Vessey M, Hawkins M, Carson J, Gough P, Marsh S.
Source: Lancet. 348:977-80. October 12, 1996.
Institution: Universtiy of Oxford.
Financial support: Schering Health Care Ltd, Novo Nordisk Pharmaceuticals, Knott Family Trust.

2. Risk of hospital admission for idiopathic venous thromboembolism among users of postmenopausal oestrogens.

Authors: Jick H, Derby L, Myers M, Vasilakis C, Newton K.
Source: Lancet. 348:981-83. October 12, 1996.
Institutions: Boston University Medical Center; Group Health Cooperative of Puget Sound.
Financial support: Astra Hassle, Bayer, Berlex, Boots Healthcare International, Ciba-Geigy, Glaxo Wellcome, Hoechst, RW Johnson, Merck, NV Organon, Pfizer.

3. Prospective study of exogenous hormones and risk of pulmonary embolism in women.

Authors: Grodstein F, Stampfer M, Goldhaber S, Manson J, Colditz G, Speizer F, Willett W, Hennekens C.
Source: Lancet. 348:983-87. October 12, 1996.
Institutions: Harvard Medical School; Harvard School of Public Health.
Financial support: NIH.


In the October 12, 1996 issue of the Lancet are three articles that report on the risk of venous thromboembolic disease (VTE) in relation to postmenopausal estrogen replacement therapy. Although the use of oral contraceptives is known to increase the risk for venous thromboembolism, this has not been clearly documented for hormone replacement therapy (HRT). The three studies presented here do show such a risk, although the absolute risk is small.

The first two studies, one from Oxford, England, the other from the Group Health Cooperative of Puget Sound in the United States, are case-control studies. The third study, part of the Nurses' Health Study from Harvard, is a prospectively designed cohort follow-up study.

Immediately below are the results of the two case-control studies in tabular form, followed by the results of the Harvard study.

  Oxford study Group Health study
Cases Women aged 45-64 hospitalized for suspected pulmonary embolism (PE) and/or deep vein thrombosis (DVT), recruited by concurrent review of Oxford area hospitals, between February, 1993 and December, 1994. Additional patients recruited by retrospective hospital record review (April, 1990 - March, 1993), in order to increase the power of the study. Women 50-74 years old, enrolled with Group Health Cooperative of Puget Sound (Seattle, Washington), with a first hospital discharge diagnosis of DVT or PE between January, 1980 and December, 1994.
Exclusions History of PE, DVT, stroke, MI; recent surgery, pregnancy, trauma or bedrest (to exclude patients with secondary VTE). Hospital admission for recent trauma or surgery; admissions for previous venous thromboembolism, heart disease, cancer, renal failure, epilepsy or diabetes.
Controls Women admitted to area hospitals with diagnoses judged unrelated to HRT; up to two controls per case. Matched by 5-year age group, admission district, date of admission. Same exclusion criteria applied. Four for each case, selected from the base population of active members, matched for age and index date. Same exclusion criteria applied.
Other Detailed interview of cases and controls in hospital or at home; further exclusions for breast or gyn cancer, other active cancer, severe heart disease, oral contraceptive or anticoagulant therapy. Current HRT use defined as use during the preceding month; non-users defined as past users and never-users. Body mass index categorized into quintiles; socio-economic status into 6 groups; current smoking defined as within past 3 months. Current HRT use defined as use within the previous 6 months.

The medical records for each case and for one matched control (selected at random from the four controls) were reviewed for smoking, body mass index, and history of varicose veins or superficial phlebitis.

Patients 103 cases (65% with DVT). 178 controls.

Potential confounders: cases had higher body mass index (27.6 vs 26.0); more had a history of superficial phlebitis (12.6% vs 0%) and varicose veins; fewer were in the higher socio-economic group (22.4% vs 32.8%).

42 cases (74% with DVT), 168 controls.

Potential confounders: cases had higher body mass index.

Current HRT use 42.7% of cases vs 24.7% of controls were current users.

Unadjusted odds ratio for venous thromboembolism associated with current HRT use: 3.0.

After adjustment for the three potential confounders, odds ratio for venous thromboembolism associated with current HRT use: 3.5 (95% CI: 1.8-7.0).

50% of cases vs 25% of controls were current users.

Unadjusted odds ratio for venous thromboembolism associated with current HRT use: 3.6.

Comparing cases to their one matched control, the unadjusted relative risk of VTE for users vs non-users was 4.5; after adjustment for age and body-mass index: also 4.5 (95% CI 1.3-15.1).

Subgroup analyses The adjusted odds-ratio for VTE was higher for shorter duration of use than for longer use. There was no significant difference for type of HRT or route of administration. There was an increasing effect with increasing estrogen dose and a suggestion of a higher risk for shorter duration of use.
Overall risk Based on these results and on regional population statistics, the authors estimate that HRT would cause 16.5 cases of VTE per 100 000 women aged 45-64 exposed per year. Based on their results (and with a few calculations on my part), in this study HRT would cause 23 cases of VTE per 100 000 women aged 50-74 exposed per year.

The Harvard study

Unlike the two previous studies, this one, part of the Nurses Health Study, examined the risk of pulmonary emboli (not DVT) in users of both oral contraceptives (OC) and HRT. The data were derived from a prospective cohort study, not a case-control study.




Although case control studies are subject to multiple forms of bias, the fact that two such studies conducted in very different settings and a third, prospective cohort study, yielded very similar results argues that post-menopausal hormone replacement therapy leads to an increase in the risk of primary venous thromboembolic events. This relative risk appears to be in the range of 2-4.

The authors of these studies note that previous investigations had not revealed this risk, but these earlier studies were smaller and/or not as well designed as the ones presented here.

Although the risk is real, because of the small absolute number of events its actual importance is limited. The effects of hormone replacement therapy on cardiovascular morbidity and mortality, breast and uterine cancer and osteoporosis are, numerically, much greater and will ultimately be the prime determinants of the risk/benefit profile for post-menopausal HRT.

Nevertheless, the data presented here are not without importance. Women who are at increased risk for venous thromboembolic disorders (from hypercoagulable states, obesity, cancer) may experience an increase in risk related to HRT that is clinically significant. This cannot be stated definitively, since effects on subgroups at higher risk are not always additive or multiplicative. Only further clinical trials or data analysis can answer this question with confidence.

In the absence of such data, however, the results presented here imply that HRT should be prescribed with greater than usual caution in women at higher risk for thromboembolic events.

October 31, 1996

Reader comments

Date: Sat, 02 Nov 1996
From: David Winsemius <dwinsemius@imagine.com>

I hope you will admit that your conclusion is without support by data:

Have you perhaps failed to consider the major competing risks for death for both pre-menopausal and post-menopausal women? For premenopausal women, wouldn't pregnancy be more risky than OC's for women with hypercoagulable states? Coronary events would be expected to be increased in hypercoagulable states and obesity in post-menopausal states, yet those events are precisely the ones reduced by estrogens. I would certainly mention the risks to a woman considering HRT, but am far from certain that we can conclude that the bottom line risks would be increased.


Date: Sat, 09 Nov 1996
From: David Winsemius <dwinsemius@imagine.com>

Thank you for both your hours of effort and good work on the site. My recent critique of your cautionary conclusion stems from my sense that internists and family physicians have a longstanding bias against estrogens stemming primarily from the early OC experience and the flawed Framingham analysis that received widespread attention due to its reputation and publication in the NEJM. I don't think physicians should be seeking out reasons to avoid estrogen use. There are other effective preventive practices that probably fall in the same category. Beta-blockers and ACE inhibitors for the appropriate risk groups leap to mind.



Letters to the editor about these articles from the Lancet's website.

August 13, 1997

A letter to the editor in JAMA (August 13, 1997) reports on preliminary results from the HERS (Heart and Estrogen-Progestin Replacement Study). These results appear to confirm the increased incidence in venous thromboembolic events noted here.

Submit a comment about this article


References related to this summary from the NLM's PubMed database.

Return to Journal Club home page

Site Meter