Hormones To Prevent Coronary Disease in Women: When Are Observational Studies Adequate Evidence?

• Dose-response relationship, drug
• Cerebrovascular accident
• Coronary disease
• Estrogen replacement therapy
• Menopause
• L00312
• Human immunodeficiency virus-1
• Viral load
• Sex factors

For many years, observational studies have reported that postmenopausal women who use estrogen have a lower rate of coronary heart disease than women who do not. Results of the Nurses' Health Study concerning this question were first published in 1985 (1) and were updated in 1991 (2), 1996 (3), and again in this issue (4). The main findings have not changed: Postmenopausal women who are current users of estrogen or estrogen plus a progestin have about a 40% lower risk for coronary disease compared with women who never used hormones. Many other observational studies have reported similar results (5), but the findings of the Nurses' Health Study carry particular weight, since it is large (70 533 postmenopausal women), is methodologically sound, and provides extended follow-up (20 years).

The association between postmenopausal hormone use and lower coronary risk has most of the elements that epidemiologists look for to suggest causation: a strong association that is consistently observed in different populations and plausible biological mechanisms, including beneficial changes in levels of low-density lipoprotein cholesterol and high-density lipoprotein cholesterol (6), improved endothelial function (7), and less progression of atherosclerosis in animals (8). However, the current report from the Nurses' Health Study documents the absence of some elements that are generally thought to support causation, such as dose–response and duration–response effects (4). No increase was seen in the apparent benefit of postmenopausal hormone therapy with increasing doses of estrogen, although a dose response is clearly observed for the beneficial effects of estrogen on lipids (9). Longer duration of hormone use was associated with less coronary benefit than short-term use in the Nurses' Health Study; the relative risk …