Please note: In 2003, the CTF updated its Grades of Recommendations to include an "I Recommendation" for situations where insufficient evidence exists to allow a recommendation to be made.  (Formerly, these situations were captured under a "C Recommendation".)  This change is not retroactive, and all "C Recommendations" made prior to 2003 have not been reevaluated in light of the new "I" recommendation grade.  For a discussion of these recommendation grades, please link to the 2003 article in the Canadian Medical Association Journal here.

Prevention of Osteoporotic Fractures in Women by Estrogen Replacement Therapy

Overview
There is good evidence to suggest that estrogen therapy slows bone loss in perimenopausal women and fair evidence that estrogen therapy will decrease fractures. In addition, there is fair evidence that estrogen therapy leads to decreased cardiovascular mortality although recent evidence suggests a small increase in the risk of breast cancer. It is therefore recommended that all women be counselled concerning the benefits and possible risks of estrogen replacement therapy (B Recommendation). Decreasing levels of bone mineral density have been associated with increased risk of fracture, however, these predictions remain preliminary. Therefore, widespread bone mineral density screening to identify those at increased risk of fracture is not advised at this time. There may be some merit in performing bone mineral density measurement in selected individual women to assist in decision-making regarding ERT. Note that osteoporosis and diet are discussed briefly in Chapter 49 and the benefits of exercise in Chapter 47.

Burden of Suffering
The most common age-related fractures are those of the distal forearm, vertebrae, and hip. Lifetime risk of Colles’ fracture has been estimated to be 15% in white women. These fractures rarely cause death or long-term disability and most need no rehabilitation. Vertebral fractures are the most common of the osteoporotic fractures. The estimated lifetime risk for a 50 year old woman of sustaining a vertebral fracture is 32%. Vertebral collapse is often asymptomatic and found incidentally on x-ray. In others, vertebral fracture may cause back pain which generally lasts a few months and can be managed with bed rest and analgesics. Progressive vertebral collapse can lead in some cases to kyphosis ("Dowager’s Hump") and chronic pain. The course of spinal osteoporosis is very unpredictable. The proportion of those who are symptomatic with vertebral deformity or collapse is not known.

Hip fractures are associated with more death, morbidity and medical costs than all other osteoporotic fractures combined. The incidence begins to rise after age 50 but rises dramatically after age 70. A 50 year old white woman whose average life-expectancy is 80 years, has a lifetime hip fracture risk of 15% compared with a 5% risk in men. Hip fracture rates are high in American and European whites, intermediate in oriental populations and low in American Blacks. This may be due to differences seen in peak bone mass achieved.

Mortality rates in the first year following a hip fracture are 12-20% higher than rates in those of similar age and sex who have not sustained a fracture. However, much of the increased mortality may be accounted for by concomitant illness and interventions to prevent hip fracture may not decrease this high mortality. Morbidity following hip fracture is high as well. Of those living at home at the time of fracture who survive the first year, 50% require assistance with walking or with activities of daily living and 15-25% become confined to nursing homes. In the U.S. the cost of acute care attributable to osteoporosis was estimated at U.S. $7-10 billion in 1984.

Maneuver
In order to prevent osteoporotic fractures all perimenopausal women could be treated with estrogen replacement therapy (ERT) or screening strategies could be used to identify those at greatest risk for osteoporosis, using either historical risk factors or bone mineral density measurement.

Effectiveness of Preventive Treatment (ERT)
There is good evidence from randomized controlled trials that in the short-term (£2 years) use of percutaneous estrogen,<1> oral estrogen<2> or estrogen and progesterone therapy<3> prevents bone loss or may even increase bone mass in perimenopausal women. There is fair evidence that estrogens retard bone loss up to 10 years.<4,5> However, the important issue is whether this decrease in bone loss translates into decreased fracture rates.

Ert and Incidence of Osteoporotic Fractures
There is fair evidence from case-control,<6-10> cohort<11-13> and one randomized controlled study<5> to suggest that ERT prevents osteoporotic fractures, including, most importantly, fractures of the hip (point estimates of relative risk; 0.65-0.79; p<.05).<11,12> Current users and those starting estrogen therapy within 3-5 years of the menopause seem to benefit most.

The optimal duration of use is not known. A recent study found that only women taking ERT for a minimum of 7 years had significantly higher bone mass compared with women who had never used estrogen. There is little information regarding when to discontinue therapy, if at all. Withdrawal of estrogen therapy, even after 10 years of use, leads to the accelerated loss of bone seen in perimenopausal women. The benefit of starting ERT in older women (for example, over 70 years) is unknown since most studies have been based on younger women.

Endometrial Cancer
The increased risk of endometrial cancer due to unopposed estrogen has been demonstrated in both case-control and cohort designs.<14> This risk increases with increasing doses of estrogen and increasing duration of use. This risk appears to be eliminated by the addition of progesterone and in fact, one study showed a significant decrease in risk (incidence in estrogen users 359.1 per 100,000; in estrogen-progesterone users 56.4 per 100,000 and in untreated women 248.3 per 100,000).<15> Many of these studies, however, were based on small numbers of cases.

Breast Cancer
Although recent evidence suggests a small increased risk in breast cancer in women on estrogen replacement, data are conflicting. A recent meta-analysis<16> of case-control studies found an increased relative risk after 15 years. The relative risk of breast cancer in estrogen users was 1.3 (95% confidence interval (CI): 1.2-1.6). Among women with a family history of breast cancer, those who had ever used estrogen replacement had a significantly higher risk (3.4%; 95% CI: 2.0-6.0) than those who had not (1.5%; 95% CI: 1.2-1.7).

Three other meta-analyses have been published to date.<14,17,18> All three found no significant increased risk of breast cancer in women who ever used estrogens. However, Grady et al<14> pooled relative risk estimates from case-control and cohort studies in which women used ERT for 8 years or more and found a summary relative risk of 1.25 (95% CI: 1.04-1.51).

Up until 1987, most cohort studies suggested no increased risk for breast cancer in estrogen users. Since 1987, several large cohort studies have observed a small increased risk of breast cancer.<19-22> Some studies suggest increasing risk with increasing duration of use<20,21> and current use.<21,22> The increased incidence of breast cancer in all these studies may reflect a certain degree of detection bias.

Although there is an increased incidence of breast cancer, breast cancer mortality rates may be lower for estrogen users<22,23> but this may be due to surveillance bias. Further evidence is needed to confirm this finding.

The estimated potential years of life lost (PYLL) to age 85 for women taking estrogen therapy alone for 15 years post-menopause has been calculated assuming that the relative risk of breast cancer was 2.0 (in American studies the RR was 1.3).<24> This was compared with the PYLL from hip fracture. The potential years of life lost from breast cancer was calculated to be 33,000 compared with only 1,200 from hip fracture. However, if breast cancer mortality is in fact unchanged, PYLL from breast cancer would be less than that calculated above.

In conclusion, there is fair evidence to suggest that there is no increased risk of breast cancer when estrogens are taken for a short period (5 years or less). However, ERT may lead to a small increased risk in breast cancer, if taken for more than 10 years. Current users are at higher risk than past users. There is insufficient evidence at present to draw any conclusions regarding the risk of combination therapy although several studies suggest that adding progesterone has no protective effect.<20,25,26>

Cardiovascular Mortality
Five of six prospective cohort studies of cardiovascular mortality in relation to estrogen use published since 1985<27-32> found a decrease in the relative risk of cardiovascular death or morbidity. A recent quantitative overview<33> noted that 15 of the 16 prospective studies published up to 1990 found a decreased relative risk of coronary heart disease among estrogen users, with a combined relative risk of 0.58 (95% CI: 0.48-0.69). In the meta-analysis published by Grady et al<14> a decreased summary relative risk was found for both coronary heart disease (0.65; 95% CI: 0.59-0.71) and cardiac death (0.63; 95% CI: 0.55-0.72) among women who ever used estrogen. Similar studies have shown a decreased overall mortality, likely due to the decreased CV disease. In order to counteract the increased loss of lives from breast cancer one would need a minimum relative risk reduction in ischemic heart disease mortality of at least 0.8 (in women aged 65-85 years).<24> The available evidence indicates that there does appear to be a reduction of at least this magnitude in cardiovascular mortality.

In summary, there is fair evidence to suggest that ERT leads to a decreased rate of cardiovascular (CV) mortality. As much as possible, these studies have made adjustments for baseline characteristics which might confound the findings. A randomized trial would help eliminate possible sources of bias. Another important caveat is based on the fact that most women in these studies used unopposed estrogen therapy. There is reason to believe that the addition of progesterone may lower the HDL and thereby negate the beneficial effects seen on CV mortality.<34> Estrogen therapy is known to have other adverse effects but not with the magnitude of risk of those mentioned above.

Targeting High-risk Groups for Preventive Therapy
Complex models for risk factor assessment (fractures and decreased bone density) have poor sensitivity and specificity as screening procedures.<35,36> Some studies have examined the possibility of using risk factors as a proxy for bone density measurement. Again, sensitivity and specificity for predicting bone mass were poor.

Bone Mineral Density Measurements to Target High Risk Groups
Case-control studies of bone mineral density (BMD) and fractures have yielded variable results. In general there is considerable overlap between non-fractured and fractured groups.<37,38>

Despite the seemingly poor characteristics of the BMD measurements many experts in the field have compared BMD with procedures such as blood pressure measurement and serum cholesterol determination. They have posed the question, ’can bone density measurement predict RISK of fracture much like hypertension predicts RISK of stroke or cholesterol predicts RISK of cardiovascular events?’ To answer this question several investigators, using mainly peripheral bone density measurements, followed patients prospectively over 1.6-10 years and documented non-spine, spine or hip fractures.<39-43> An increased relative risk of fractures was found when comparing those with the highest to those with the lowest bone density measurements. However, in order to predict risk each investigator used different criteria. Where should the "fracture threshold" be? There is no universally accepted level of osteopenia or fracture risk at which it is agreed treatment should be initiated.

Ross and colleages (1987) suggest that the fracture threshold for vertebral fracture should be based on the level at which the fracture risk doubles. This level corresponds to an annual fracture probability of 0.5% and the 95th percentile for fracture incidence cases. At this level approximately 37.5% of perimenopausal women (aged 50-55) would be at increased risk of fracture. For women aged 60-64, over 50% would be at increased risk. This threshold was defined using the Japanese-American population of Hawaii, which may limit its generalizability. As well, the best risk predictor was the density of the os calcis, a measurement unavailable in most Canadian centres.

Most manufacturers and users of bone density technology take the value of 2 SD below the mean for premenopausal women aged 30-45 as the fracture threshold. This level would encompass approximately 15% of women aged 50-55 years old. This threshold has been chosen arbitrarily and is not derived from epidemiological data. Also, most operators of bone densitometers do not relate the relative risk prediction of their result to the current age and thus the fracture risk of the individual.

Other authors have developed estimates of lifetime fracture risk based on age and the level of femoral bone mineral density. While this model makes intuitive sense, it needs to be evaluated further through longitudinal studies. To date no randomized controlled trials have screened asymptomatic women and demonstrated the efficacy of screening in decreasing fracture rates.

Recommendation of Others
The recommendations of the U.S. Preventive Services Task Force on hormone prophylaxis are currently under review. In 1989, the Task Force recommended against routine screening for decreased bone mass in asymptomatic women but stated that this may be considered in the patient where such information would be used to make decisions regarding ERT.<44> A Consensus Development Conference sponsored by the European Foundation for Osteoporosis and Bone Disease, the National Osteoporosis Foundation and the National Institute of Arthritis and Musculoskeletal and Skin Diseases of the U.S.,<45> recommended that all women at risk for osteoporosis be considered for estrogen therapy, barring contraindications. They also suggested that bone density measurements could aid in predicting risk.

Conclusions and Recommendations
The benefits of estrogen replacement therapy must be weighed against its risks. There is good evidence that estrogen replacement therapy slows the rate of bone loss and fair evidence that estrogen replacement therapy decreases the incidence of fractures. There is also fair evidence that ERT leads to a decreased rate of cardiovascular mortality. This benefit would certainly outweigh the benefit achieved by decreasing fracture risk since the incidence of cardiovascular disease is so great. However, it is not yet known whether this benefit will continue to be seen with combination therapy. The other caveat is the small increase in risk of breast cancer noted in recent studies. It is recommended, therefore, that all women be counselled with regard to the benefits and possible risks of estrogen replacement therapy (B recommendation). Chapters on circulatory disorders (Chapters 53 to 57), tobacco (Chapter 43), physical activity (Chapter 47), and on breast cancer (Chapter 65) may provide useful background information.

Widespread bone mineral density screening is inadvisable at present (D recommendation). There are no universally accepted criteria using bone mineral density measurement to establish a level at which treatment should be instituted. Decreasing levels of bone mineral density are associated with increased risk of fracture. However, true estimates of efficacy can only be obtained by studying perimenopausal women who have been followed for a lengthy period of time. Presently there may be some merit in using bone mineral density measurements in individual women to assist in decision-making regarding ERT.

Unanswered Questions (Research Agenda)
Better quality evidence in terms of a randomized clinical trial would enhance our knowledge regarding the benefits of estrogen therapy in reducing the incidence of both fractures and of cardiovascular deaths. More data are needed regarding the effects of adding progesterone to ERT on breast cancer risk and cardiovascular mortality. More research is needed into the identification of women in the perimenopausal period who are at greatest risk of developing hip fractures. More research is needed in assessing hormone replacement therapy (HRT) benefits in various ethnic groups and in older women. Finally, more research is needed which would assist physicians to assess each individual patient’s health risks and benefits so that they could make a more informed decision regarding HRT.

Evidence
A MEDLINE search was performed for articles from 1987 forwards, using the following search terms: osteoporosis, estrogen replacement therapy, synthetic estrogens, evaluation studies, random allocation, double-blind method, drug evaluation, random, cohort studies, clinical trial, menopause, postmenopausal and english. Other evidence was obtained from reference material and content experts. Although men suffer from non-traumatic fractures as well, only studies of women have been included.

This review was initiated in January 1992 and recommendations were finalized in October 1993. Other Task Force recommendations on postmenopausal osteoporosis and related fractures were published in 1988.<46>

Acknowledgements
The author thanks Elaine Wang, MD, CM, FRCPC, Associate Professor, Department of Pediatrics and of Preventive Medicine and Biostatistics, Faculty of Medicine, University of Toronto; Jonathan D. Adachi, BSC, MD, FRCPC, Associate Professor of Medicine, McMaster University, Hamilton, Ontario; Nancy Kreiger, MPH, PhD, Assistant Professor, Department of Preventive Medicine and Biostatistics, University of Toronto, Toronto, Ontario; and Anthony B. Hodsman, MB, FRCPC, Professor, Department of Medicine, University of Western Ontario, London, Ontario.

Full Citation
Feig D.S. Prevention of osteoporotic fractures in women by estrogen replacement therapy. In: Canadian Task Force on the Periodic Health Examination. Canadian Guide to Clinical Preventive Health Care. Ottawa: Health Canada, 1994; 620-31.