Background: The estrogen-only arm of the Women's Health Initiative study was a double-blind, placebo-controlled randomized clinical trial that tested the preventive effects of estrogen on chronic disease states in women who had undergone hysterectomy. The study, which randomized 10,739 women to daily therapy with 0.625 mg of conjugated equine estrogen (Premarin) or placebo, was stopped one year early because of an increased risk of stroke. During the intervention phase, women in the intervention group took estrogen for a median of 5.9 years. In this planned postintervention analysis, LaCroix and colleagues followed the participants for an additional mean of 47.2 months to see if short- and long-term risks and benefits persisted after discontinuing estrogen use.

The Study: Of the surviving estrogen-only participants, 77.9 percent (n = 3,778) of the estrogen group and 78.4 percent (n = 3,867) of the placebo group consented to continue in the observation portion. Participants were encouraged to get annual mammograms, and the results were tracked. Between 3.6 and 4.7 percent of women in the estrogen group and 2.7 to 3.0 percent of the placebo group reported using estrogen during the postintervention period. The estrogen and placebo group participants were analyzed on an intention-to-treat basis, and the baseline characteristics of women who gave consent for the continued study period were similar to those of women who declined to continue participation. Main outcome measures included annualized rates of coronary heart disease (CHD), invasive breast cancer, stroke, venous thrombotic event, colorectal cancer, hip fracture, and death.

Results: The hazard ratios (HRs) for CHD, venous thrombotic events, stroke, hip fracture, and invasive breast cancer were compared between the intervention phase and the postintervention phase. The risk of overall CHD was not significantly increased for estrogen users during the intervention phase; that risk did not change in the postintervention phase (HR = 0.95 versus 0.97). The increased risk of stroke in estrogen users during the intervention disappeared in the postintervention phase (HR = 1.36 versus 0.89; P = .05 for the difference). Similarly, the increased risk of deep venous thrombosis or pulmonary embolism found in estrogen users was not sustained when the medication was stopped (HR = 1.32 versus 0.72; P = .01 for the difference).

Approximately 81 percent of women in both groups had at least one mammogram during the postintervention phase. The reduced risk of breast cancer in the estrogen group during the intervention phase (0.28 percent for estrogen group versus 0.35 percent for placebo group) was maintained in the postintervention phase (HR = 0.79 versus 0.75, respectively). A reduced risk of hip fracture in the estrogen group was not maintained in the postintervention phase (HR = 0.67 versus 1.27; P = .01).

When the women were stratified by age (i.e., 50 to 59, 60 to 69, and 70 to 79 years), the risk of CHD was significantly lower in women 50 to 59 years of age who received estrogen compared with placebo; however, there was no difference in the older age groups. There were no age-related risk differences for venous thrombotic events, breast cancer, or stroke. The absolute rates of events per 10,000 women over the 10.7-year follow-up showed a benefit for women 50 to 59 years of age who took estrogen. Conversely, women 70 to 79 years of age who took estrogen had worse outcomes.

Conclusion: The increased risks of stroke and venous thrombotic events among estrogen users in the Women's Health Initiative dissipated after they stopped taking estrogen. The decreased risk of breast cancer persisted.