Current issues of ACP Journal Club are published in Annals of Internal Medicine


Review: HRT is not associated with cardiovascular events or cancer in postmenopausal women in recent clinical trials

ACP J Club. 1998 Mar-April; 128:48. doi:10.7326/ACPJC-1998-128-2-048

Source Citation

Hemminki E, McPherson K. Impact of postmenopausal hormone therapy on cardiovascular events and cancer: pooled data from clinical trials. BMJ. 1997 Jul 19;315:149-53.



To determine, using meta-analysis, the effect of hormone replacement therapy (HRT) on cardiovascular disease or cancer in postmenopausal women.

Data sources

Studies were identified with MEDLINE (1985 to November 1995). Bibliographies of review articles, books, and identified trials were also searched.

Study selection

English-, German-, and Scandinavian-language studies were selected if they were controlled trials that compared HRT with placebo, no therapy, or vitamins or minerals in postmenopausal women. HRT was defined as estrogen in any form, alone or combined with progestin or progesterone. Trials < 3 months in duration were excluded.

Data extraction

Data were extracted on medication regimens, including dose; number of women studied; and adverse outcomes, including cardiovascular events (cardiac arrest, cerebrovascular accidents, ischemic attacks, and myocardial infarction), thromboembolic events (pulmonary embolism and deep venous thrombosis), superficial phlebitis, thrombophlebitis, and cancer (including breast; uterine; and other types, including cervical cancer).

Main results

23 trials with 4164 women with breast cancer and 2899 women with cardiovascular disease and types of cancer other than breast cancer were included. Data were pooled, and intention-to-treat analysis was used. No outcome was associated with HRT, although the number of women included and number of events were small (Table). Further, with an odds ratio (OR) of 1.64 for cardiovascular disease (Table), the probability is < 5% that a benefit of the size suggested by observational studies (OR 0.70) was missed in these studies. The ORs were recalculated after exclusion of trials that enrolled women who had had oophorectomy, an outlier trial, and trials with < 1 year of follow-up; none of these subgroup analyses showed decreased risks for cardiovascular events.


Recent clinical trials do not support benefits or risks of short-term HRT as suggested by observational studies.

Source of funding: No external funding.

For article reprint: Professor E. Hemminki, National Research and Development Centre for Welfare and Health, Health Services Research Unit, P.O. Box 220, 00531 Helsinki, Finland. FAX 358-9-3697-2485.

Table. Risk factors for adverse outcomes with hormone replacement therapy (HRT) in postmenopausal women taken from randomized controlled trials

Event Number of studies HRT events Control events Odds ratio 95% CI
Cardiovascular event and thromboembolism 12 17 6 1.64 0.65 to 4.18
Cardiovascular event 10 12 5 1.39 0.48 to 3.95
Thromboembolism 5 5 1 2.89 0.34 to 24.8
Phlebitis 4 21 17 0.71 0.37 to 35
Breast cancer 9 19 9 0.85 0.38 to 1.89
Uterine cancer 4 2 2 0.58 0.08 to 4.1
Other types of cancer 10 12 8 0.86 0.35 to 2.12


For the past 2 decades, prevailing wisdom has suggested that estrogen and perhaps estrogen plus progestin replacement therapy protects against cardiovascular disease in postmenopausal women. More than 30 observational studies have suggested that women receiving HRT have a lower risk for cardiovascular morbidity and mortality. 4 previous meta-analyses estimated this mortality reduction to be between 35% and 50%. These findings have helped make HRT a leading prescription treatment in the United States. However, women receiving HRT and those who are not differ considerably with respect to lifestyle and socioeconomic factors, which raises the possibility of a spurious association.

Such potential bias can be minimized by large, rigorous, randomized controlled trials. Although such trials are currently being done, their results will not be available for several years. In the interim, the meta-analysis by Hemminki and McPherson, which combines smaller controlled trials, is particularly important because it represents a synthesis of the best currently available data from experiments rather than observations. This meta-analysis has strong review methods but tallies adverse events from studies that did not include morbidity and mortality as major end points. The outcomes of interest were not systematically reported in all trials, the total number of events was small, and the effects were limited to those immediately related to HRT.

Evidence is emerging that meta-analyses of smaller clinical trials do not always predict the results of similar larger studies. Whether this will be the case for HRT literature remains to be seen. Very recent departures from the traditionally positive association between HRT and cardiovascular disease include a case-control study by Sidney and colleagues (1) that failed to show a decreased association between myocardial infarction and current HRT. However, at present, the prevailing wisdom remains that HRT is protective.

Roberta B. Ness, MD, MPH
University of PittsburghPittsburgh, Pennsylvania, USA

Roberta B. Ness, MD, MPH
University of Pittsburgh
Pittsburgh, Pennsylvania, USA


1. Sidney S, Pettiti DB, Quesenberry CP Jr. Myocardial infarction and the use of estrogen and estrogen-progestogen in postmenopausal women. Ann Intern Med. 1997;127:501-8.