|
| |||||||||||||||||||||||||||||||||||||||
Background Estrogen therapy prevents bone loss in postmenopausal women who take it early in the postmenopausal period. The risk of fracture is highest much later in life, however. We studied whether bone mass in elderly women was affected by earlier estrogen use and how long women needed to take estrogen for it to have a beneficial effect on bone density later in life.
Methods In 1988 and 1989, we measured bone mineral density at the femur, spine, shaft of the radius, and ultradistal radius in 670 white women in the Framingham Study cohort (mean age, 76 years; range, 68 to 96). These women had been followed prospectively through menopause and had been asked repeatedly about estrogen therapy. After excluding women who began taking estrogen after a fracture, we investigated whether postmenopausal estrogen therapy affected bone density; in these analyses we adjusted for age, weight, height, cigarette smoking, physical activity, and age at menopause.
Results A total of 212 women (31.6 percent) had received estrogen therapy (mean estimated duration of treatment, 5 years). Only women who had taken estrogen for 7 to 9 years or for 10 or more years had significantly higher bone mineral density than women who had not taken estrogen (7 to 9 years of treatment, P<0.05 at sites in the femur and the spine;
In the women less than 75 years of age who had taken estrogen for seven or more years, the bone density was, averaging all sites, 11.2 percent greater than in women who had never received estrogen. Among women 75 years of age and older in whom the duration of therapy was comparable, bone density was only 3.2 percent higher than in women who had never taken estrogen.
Conclusions For long-term preservation of bone mineral density, women should take estrogen for at least seven years after menopause. Even this duration of therapy may have little residual effect on bone density among women 75 years of age and older, who have the highest risk of fracture.
10 years, P<0.05 at all sites except the spine).
Estrogen therapy begun soon after the menopause slows or even reverses the loss of bone that occurs normally during those years8,9. The long-term effect of postmenopausal estrogen therapy on bone density, however, is not known. Epidemiologic studies10,11,12 have reported that postmenopausal estrogen therapy protects women against later hip fracture, suggesting a residual effect of estrogen on bone density, but there are few data on the effects of early postmenopausal estrogen therapy on bone density in elderly women.
How long should women take estrogen to obtain a prolonged effect on bone density? Although five years of estrogen therapy has been suggested, on the basis of epidemiologic studies of estrogen therapy and hip fracture in which women who had taken estrogen for different periods of time were grouped together,11,12,13 there are insufficient data to support this recommendation8. Women considering estrogen therapy and their physicians may wish to know the minimal duration of therapy necessary to achieve a prolonged effect on bone density.
The long-term effect of estrogen on bone may be modified by other factors. Age, independent of estrogen loss, may contribute to bone loss in elderly women14; estrogen therapy may become increasingly irrelevant as women approach the age at which the risk of fracture is highest. Other factors such as smoking15,16,17,18 may affect the benefit conferred by long-term estrogen therapy.
In previous studies of women in Framingham, Massachusetts,10 we documented that estrogen therapy protected women against hip fracture. In this study, we examined whether postmenopausal estrogen therapy affected bone density in elderly women in the same cohort.
Methods
The Framingham Study began in 1948 with the primary goal of evaluating risk factors for heart disease. The participants have been examined every two years since that time, and by 1988, over half the original cohort had died. The Framingham Osteoporosis Study, a component of the biennial examination performed in 1988 and 1989, involved 1164 surviving subjects (448 men and 716 women), all of whom were white.
Bone-Mineral-Density Measurements
The bone mineral density of the lumbar spine, femur (neck, trochanter, and Ward's triangle), shaft of the radius, and ultradistal radius were measured during the 20th biennial examination in 1988 or 1989. The bone density of the lumbar spine (L2 through L4) and femur was measured by dual-photon absorptiometry (LUNAR DP3) and the bone density of the shaft of the radius and the ultradistal radius by single-photon absorptiometry (LUNAR SP2). The density of the shaft of the radius was measured at the junction of the proximal two thirds and the distal one third of the radius, and the density of the ultradistal radius near the wrist where the radius and ulna meet. The coefficients of variation for these measurements in young normal subjects ranged from 2.6 percent at the femoral neck to 5.7 percent at the ultradistal radius19.
Estrogen Therapy
From the 7th biennial examination (performed in 1960 through 1963) through the 16th examination (1979 through 1981), all women were asked about hormone use since the previous examination. Starting with the 17th examination (1981 through 1983), the women were asked specifically about estrogen therapy. We characterized therapy received for less than one year as one year of therapy and therapy for one year or more as two years of therapy. Menopause was defined as occurring one year after the last menstrual period or at the time of ovariectomy. Total estrogen therapy included the two years before menopause because symptoms of estrogen deficiency leading to the initiation of estrogen therapy may begin at that time. Restricting the analyses to therapy received after menopause did not alter the results.
To confirm that hormone use referred to estrogen therapy, to verify the women's reports that they were taking estrogen, to verify that at least one year of therapy usually reflected two years of therapy, and to identify women who started estrogen therapy after being given a diagnosis of osteoporosis, we reviewed the charts of all the women in the cohort who were identified by computer as having taken estrogen for three or more years. These reviews led to the exclusion of three women in whom hormone use was identified as corticosteroid therapy. Almost all estrogen therapy was confirmed by the appearance of the medication name in the medical records. If the name of the medication was not recorded, we accepted an appropriate time of therapy (for example, therapy for several years just after menopause), an appropriate indication (such as hot flashes), or the report of estrogen therapy at the 17th examination or later as evidence that the women had taken estrogen. All women who took estrogen for at least five years had at least one examination during which it was noted that they had had two years of therapy. All those who had taken estrogen for at least seven years had at least two examinations during which two years of therapy were noted, and most had taken it continuously.
With respect to the type and dose of estrogen, almost all women took oral conjugated estrogens at a dose of 0.625 mg or more daily, without concomitant progestin therapy.
Other Variables
Other variables that affect bone density were also assessed. They included age, weight at the 20th examination, and height at the first examination. Age at menopause was ascertained prospectively, and it was documented by a review of the operative notes in the cases of women who had undergone ovariectomy. Cigarette smoking was measured as the mean number of cigarettes smoked per day after menopause. Habitual physical activity was assessed with use of the Framingham Physical Activity Index20 at the 4th and 12th examinations, and the results were averaged.
Statistical Analysis
We used multiple regression analysis to assess the effect of estrogen therapy on bone mineral density at different sites, adjusting for confounding factors including age (modeled as a quadratic function), age at menopause, weight, height, and cigarette smoking. Physical-activity assessments were missing for some women, so that analyses reported here did not include this variable; analyses including this variable did not change the estimates of the effect of estrogen on bone. The least-squares means from these linear models and their 95 percent confidence intervals were used to measure the association between estrogen therapy and bone mineral density. These least-squares means, calculated with the PROC GLM program in SAS software, represent the mean bone mineral density for each estrogen-exposure group, assuming mean values for all other covariates (weight, height, and so on). All P values are two-tailed.
We tested the length of time since estrogen therapy was discontinued as an independent factor predicting bone density at all sites. The analyses were limited by the strong inverse correlation (r = -0.60) between the duration of estrogen therapy and the years since therapy was discontinued. Therefore, we analyzed the time since therapy was discontinued as an independent variable in equations in which bone mineral density was the dependent variable and other independent variables (such as age) were included, but without terms for the duration of estrogen therapy. For the women who had taken estrogen for at least seven years, among whom the length of time since the discontinuation of therapy was more weakly correlated with the duration of therapy (r = -0.36), we carried out regression analyses including the duration of estrogen therapy.
To test whether age or other factors (such as smoking) affected a woman's response to estrogen, we added interaction terms to the regression models. In addition, we assessed the effect of perimenopausal and postmenopausal estrogen therapy in women above and below the age of 75 years, the median age of the study subjects at the 20th examination.
We reviewed the records of women who had clinically recognized fractures of the forearm, humerus, pelvis, hip, and vertebral bodies to determine whether estrogen therapy was begun after the fracture, suggesting preexisting osteoporosis. We also identified women with a diagnosis of osteoporosis by chart reviews. We excluded from the analyses women in whom fractures or a diagnosis of osteoporosis preceded estrogen therapy.
Results
Of the 684 women in whom bone mineral density was measured and for whom information about estrogen therapy was available, 14 had fractures or had been given a diagnosis of osteoporosis at one of the two examinations before they began estrogen therapy and were therefore excluded. The mean age of the remaining 670 women was 76 years (range, 68 to 96). The 212 women (31.6 percent) who had been treated with estrogen were younger than the rest of the group and, since many started estrogen therapy after surgical menopause, they had a younger average age at menopause (Table 1). The mean duration of therapy was five years.
|
The density of the shaft of the radius and the ultradistal radius increased with an increasing duration of estrogen therapy and was significantly higher among women who had taken estrogen for at least 10 years than among those who had never taken estrogen (Figure 1). The women who had taken estrogen for 7 to 9 years or 10 or more years generally had higher values for femoral bone density than the untreated women (Figure 2). The results were similar for the spine (Figure 3), where the bone mineral density was significantly increased only among women who had taken estrogen for seven to nine years.
|
|
|
More than two thirds of the women who had taken estrogen (67.5 percent) had not been treated long enough (at least seven years) to have long-term preservation of bone density. Of the entire cohort, only 69 women (10.3 percent) had been treated for at least seven years.
Bone Mineral Density and Age
Among the women less than 75 years old, the 45 who had taken estrogen for at least 7 years (mean [±SD], 11 ±3) had significantly higher bone mineral density than the 209 women who had never taken estrogen, after adjustment for age, weight, and other factors that might affect bone mineral density (Table 2). The bone-mineral-density values among women with seven or more years of treatment were higher at every site; the mean bone-density values were 7.6 to 18.8 percent higher than in the never-treated women.
|
Among women less than 75 years old, the bone mineral density was positively correlated with the duration of estrogen therapy. Considering the mean values at all sites, the mean bone density was 5.3 percent higher among the women treated for five to six years than among the women who had never taken estrogen. The comparable values for the women treated for 7 to 9 years and the women treated for 10 or more years were 8.0 percent and 13.4 percent, respectively. Among the women 75 years old and older, the effect of prolonged therapy was slight. As compared with the women who were never treated, the mean bone density among the women treated for 5 to 6 years was 3.8 percent higher, that among the women treated for 7 to 9 years was 8.0 percent higher, and that among women treated for 10 years or more was 2.7 percent higher. Among the women who received estrogen for at least 7 years, therapy was discontinued 10 ±4 years before this study began in the women less than 75 years old and 12 ±6 years before the study began in those 75 years old or older (P = 0.04). After adjustment for variables affecting bone density, the length of time since the discontinuation of therapy was not an independent predictor of bone density at any site either among all women or among those treated for at least seven years. Nonetheless, among the women who had taken estrogen for at least 10 years, those in the highest third of the group in terms of the time since therapy was discontinued (
14 years) had a lower average bone density than the women in the lowest third, a group of women who had discontinued therapy less than 9 years earlier. The differences in mean bone density between these groups ranged from 13.0 percent at the femoral neck to 23.0 percent at the ultradistal radius.
Effects of Smoking
Among the nonsmokers, the average difference in bone density between estrogen-treated women and women who had never received estrogen was 9.1 to 24.5 percent, depending on the site; the women treated with estrogen had higher bone density at all sites. Among smokers the differences ranged from only 5.6 percent to 14.3 percent, with the estrogen-treated women again having higher bone densities. Although these results suggest a weaker effect of estrogen among smokers, the interaction between smoking and estrogen was not statistically significant at any site.
Discussion
The results of this cross-sectional study of bone density in postmenopausal women suggest that at least seven years of estrogen therapy is necessary for a persistent long-term effect on bone density. Only 10.3 percent of all the women we studied and only 32.5 percent of those treated with estrogen actually took estrogen this long. Thus, for many women treatment may be too short to have a long-term effect on bone mineral density. The long-term benefit of at least seven years of estrogen therapy extended to the age of 75 years, but it may not persist after that age. A strength of this study is that information about estrogen therapy was obtained repeatedly and prospectively, minimizing the likelihood of recall bias.
Nordin et al.14. have suggested that there are estrogen-dependent and age-dependent components of bone loss and that, by the age of 70, the age-dependent component predominates. Even so, one could theoretically prevent the estrogen-dependent component of bone loss if women took estrogen for long enough after the menopause. In one study,13 women whose average age was 73 years who had taken estrogen almost continuously since menopause had higher bone-mineral-density values than those who were never treated, suggesting that the estrogen-dependent component of bone loss may be delayed as long as estrogen is taken. Unfortunately, the effect of estrogen does not persist long after the discontinuation of therapy. Even 10 or more years of past estrogen therapy among women 75 years old or older did not have a significant effect on bone density. Our findings confirm the predictions of Heaney,21 who suggested that 5 or even 10 years of early postmenopausal estrogen therapy would have a trivial residual effect on bone mineral density at age 75. It is also possible that women may be able to reverse some of their estrogen-dependent bone loss or prevent further loss by starting estrogen therapy late in life22.
These results extend and confirm our previous findings that estrogen therapy protects against hip fracture10. Our earlier study was structured differently from this one and involved women whose mean age was in the lower 70s; in it we examined the effect of recent and past estrogen therapy on the risk of hip fracture in the subsequent two years. We found a stronger protective effect against hip fracture among women less than 75 years of age than among those 75 or older. Other studies that found that estrogen therapy protects against hip fracture included only women 75 years of age or younger11 or women up to the age of 8012,23,24. Among women older than 75 or 80, the protective effect of estrogen (usually taken earlier in life) is negligible25,26.
In our study, among women under the age of 75, bone density was higher in the women treated with estrogen. How is that likely to affect the risk of fracture? Cummings et al.27. reported that the risk of hip fracture was 2.6 times higher for every decrease of 1 SD in the density of the femoral neck and 2.7 times higher for every decrease of 1 SD in the density of the trochanter. On the basis of these results, women less than 75 years old in the Framingham Study who took estrogen for 10 or more years would have an approximate reduction of 44 percent in the risk of hip fracture at the femoral neck or of 52 percent at the trochanter. This reduction in risk is similar to that reported in epidemiologic studies of the effect of estrogen in protecting against hip fracture. Because the precision of the measurements obtained by dual-photon absorptiometry in this study was less than the precision of currently used machines for dual-energy x-ray absorptiometry, our estimate of the risk reduction is conservative.
We had difficulty distinguishing the effect of the duration of estrogen therapy from that of the number of years since therapy was discontinued, because the two were strongly inversely correlated. Although our multivariate analyses did not reveal an independent effect on bone density of the length of time since therapy was discontinued, a small number of women who had received long-term estrogen therapy but discontinued it many years earlier had much lower average bone density than women who had taken estrogen recently.
In conclusion, at least seven years of estrogen therapy after menopause are necessary for a long-term protective effect on bone mineral density. Even therapy of this duration may be insufficient to protect women 75 years old and older from fracture.
Supported by grants (Multipurpose Arthritis Center grant AR20613 and R01AG09300) from the National Institutes of Health.
We are indebted to Ms. Gaynell Walker, Ms. Nilsa Carrasquillo, and Ms. Karen Neuhauser for their assistance and to Ms. Cherlyn Mercier and Ms. Mimi Brodsky for the measurements of bone density.
Source Information
From the Boston University Arthritis Center, Boston (D.T.F., Y.Z., M.T.H., J.J.A.); the Departments of Medicine, Boston City Hospital and University Hospital, Boston (D.T.F.); Hebrew Rehabilitation Center for Aged, Boston (D.P.K.); and the Framingham Study, Framingham, Mass. (P.W.F.W.).
Address reprint requests to Dr. Felson at A203, Boston University School of Medicine, Arthritis Center, 80 E. Concord St., Boston, MA 02118.
References
| |||||||||||||||||||||||||||||||||||||||
Related Letters:
Effect of Estrogen Therapy on Bone Density in Elderly Women
Kanis J. A., Stevenson J. C., Seeman E., Felson D. T., Zhang Y.
Extract |
Full Text
N Engl J Med 1994;
330:715-716, Mar 10, 1994.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |