RelativeVersusAttributable Risk of Breast Cancer from Estrogen Replacement Therapy

Estrogen deficiency causes vasomotor instability and urogenital atrophy in a large fraction of menopausal women. Short term use of estrogen replacement therapy (ERT) can effectively relieve symptoms related to these conditions (1, 2). ERT may also relieve neurocognitive symptoms of mood changes, irritability, and depression resulting from acute estrogen deprivation (3). When given long term, estrogens prevent the development of osteopenia, osteoporosis, and bony fractures (4, 5) and may be effective for primary prevention and reduction of mortality from cardiovascular disease (6, 7). Observational studies suggest that the incidence of colon cancer and Alzheimer’s disease may be reduced by hormone replacement therapy (8–12). The Nurses Health Study, a large prospective cohort study, reported that all causes of mortality among nurses who use postmenopausal hormones are lower than those among nonusers (13). The overall benefits of estrogen therapy to relieve symptoms and to prevent osteoporosis and cardiovascular disease have led to recommendations that the majority of women should consider taking ERT. However, surveys in the United States indicate that an average of only 20% of postmenopausal women take ERT (14–16). Even among nurses, only 50% report ever using ERT, and those choosing to take ERT generally discontinue it after a period of 1–3 yr (6). With substantial evidence supporting the use of ERT, why do most women choose not to start or to discontinue estrogen replacement? One reason is the concern about breast and uterine cancer. Although these risks may be real, the lay public and physicians generally do not fully understand the implications of studies reporting the risks of estrogen causing cancer. One reason for this misunderstanding is the widespread use of relative risk (RR) statistics that are often interpreted as attributable risk. This treatise will clarify the meaning of relative and attributable risk, review data regarding the effects of ERT on these two parameters, and provide a practical approach to aid women in the decision making process. Concepts underlying calculation of RR