Villareal DT, Binder EF, Williams DB, Schechtman KB, Yarasheski KE, Kohrt WM. Bone Mineral Density Response to Estrogen Replacement in Frail Elderly WomenA Randomized Controlled Trial. JAMA. 2001;286(7):815–820. doi:10.1001/jama.286.7.815
Author Affiliations: Washington University Claude Pepper Older Americans Independence Center, Division of Geriatrics and Gerontology, Department of Internal Medicine (Drs Villareal, Binder, and Yarasheski), Department of Obstetrics and Gynecology (Dr Williams), and Division of Biostatistics (Dr Schechtman), Washington University School of Medicine, St Louis, Mo; and Department of Medicine, Division of Geriatric Medicine, University of Colorado Health Sciences Center, Denver (Dr Kohrt).
Context Although hormone replacement therapy (HRT) is an established approach
for osteoporosis prevention, little is known about the osteoprotective effects
of HRT in frail elderly women.
Objective To determine whether HRT increases bone mineral density (BMD) in frail
Design and Setting Randomized, double-blind, placebo-controlled trial conducted in a US
university-based research center from September 1995 to August 2000.
Participants Sixty-seven women aged 75 years or older with mild-to-moderate physical
Intervention Participants were randomly assigned to receive conjugated estrogens,
0.625 mg/d, plus trimonthly medroxyprogesterone acetate, 5 mg/d for 13 days
(n = 45), or matching placebo (n = 22), for 9 months.
Main Outcome Measures The primary outcome measure was 9-month change in BMD of the lumbar
spine and hip, measured by dual-energy x-ray absorptiometry. Secondary outcomes
were changes in markers of bone turnover.
Results Based on intention-to-treat analyses, HRT resulted in significantly
larger increases in BMD of the lumbar spine than placebo (mean change, 4.3%
vs 0.4%; between-group difference, 3.9%; 95% confidence interval [CI], 3.5%-4.3%)
and total hip (mean change, 1.7% vs −0.1%; between-group difference,
1.8%; 95% CI, 1.5%-2.1%). Compared with placebo, HRT resulted in significant
decreases in serum bone-specific alkaline phosphatase levels (mean change, −24%
vs 6%; between-group difference, −30%; 95% CI, −26% to −33%)
and urine N-telopeptide levels (mean change, −48%
vs 4%; between-group difference, −52%; 95% CI, −47% to −55%).
Conclusions In physically frail elderly women, 9 months of HRT significantly increased
BMD compared with placebo in clinically important skeletal regions. Further
studies are needed to determine whether these osteogenic effects of HRT in
elderly women are associated with a reduction in osteoporotic fractures.
The highest incidence of osteoporotic fractures is in women older than
75 years, who constitute a rapidly expanding segment of the US population.
Bone mineral density (BMD), which is a strong risk factor for osteoporosis,
continues to decline with age. Indeed, recent prospective studies indicate
that bone loss not only continues from middle age into old age but may in
fact accelerate in elderly persons.1 Given
the progressive increase in physical frailty that typically occurs in elderly
women, an osteoporotic fracture, particularly of the hip, often contributes
to the loss of functional independence.
Although estrogen-based hormone replacement therapy (HRT) is the foundation
of osteoporosis prevention it is rarely initiated in elderly women, despite
accumulating evidence that estrogen therapy may be as effective in attenuating
bone loss in elderly women as in younger postmenopausal women.2- 8
In fact, it has been suggested that the response of bone to estrogens is greatest
in those women furthest from menopause.2,3
However, most previous studies of the effects of estrogens on bone in older
women have been observational.1,3- 5,9- 11
Because socioeconomic factors are the chief determinants of estrogen use by
elderly women,12 the benefits of HRT in observational
studies may have been related to other lifestyle factors.
Although prospective studies have shown an increase in BMD in response
to HRT,2,6- 8
they typically have included women younger than 75 years. The effects of HRT
on the BMD of very old, frail women at high risk for osteoporotic fractures
have not been reported. Therefore, we performed a randomized, double-blind,
placebo-controlled trial of the effects of HRT on BMD in women aged 75 years
or older with mild-to-moderate physical frailty. We hypothesized that HRT
would increase the BMD of the lumbar spine and proximal femur in this population.
The study was conducted through the Washington University Older Americans
Independence Center (OAIC) from September 1995 to August 2000. Recruitment
was directed to elderly women, aged 75 years or older, from the community
at large and from congregate living sites. Volunteers provided written informed
consent to participate in the study, which was approved by the institutional
review board of the Washington University School of Medicine.
The research focus of the Washington University OAIC was on the amelioration
of physical frailty in elderly persons. Therefore, eligible volunteers were
women who had mild-to-moderate physical frailty, as defined by meeting at
least 2 of the following 3 criteria: (1) low peak aerobic power (VO2peak) of 11 to 18 mL/min per kg of body weight,13
(2) self-reported difficulty or need for assistance with 2 instrumental activities
of daily living (ADLs) or 1 basic ADL, and (3) modified physical performance
test score of 18 to 22 (score range, 0-36).14
The screening procedures have been described previously.14- 16
Exclusion criteria included use of estrogens within the past year, history
of breast cancer or other estrogen-dependent neoplasia, history of cancer
within the previous 5 years, recent history (<5 years) of thromboembolic
disease, use of drugs that affect bone metabolism in the previous year, or
active serious illness. Women taking thyroid hormone who were not on a stable
dose for at least the previous 3 months were excluded. Bone mineral density
levels were not exclusionary.
Figure 1 shows the results
of recruitment and randomization. Of the 292 women who underwent screening
evaluations, 67 were randomized to receive either HRT (n = 45) or placebo
(n = 22) in a 2:1 ratio, using a computer-generated block random permutation
procedure.17 Twice as many women were assigned
to the HRT group because it was anticipated that attrition would be higher
in this group and because at the end of the 9-month trial, women taking HRT
were invited to continue HRT and were further divided into 2 exercise study
groups. Measures to ensure blinding included: (1) an investigator who did
not interact with the participants after screening assessments
maintained the randomization log, (2) all investigators involved with outcome
data were blinded to treatment assignment, (3) participants who complained
of vaginal bleeding were instructed to report to the gynecologist member of
the research team who did not interact with other trial staff and had no role
in the ascertainment of the main outcome variables, and (4) participants and
clinicians did not review the results of BMD or bone turnover until the end
of the study.
The 9-month HRT regimen consisted of conjugated estrogens, 0.625 mg/d,
and cyclic medroxyprogesterone acetate (MPA), 5 mg/d for 13 consecutive days
every third month. Placebo and active tablets were identical in appearance.
Women without a uterus were not provided MPA or placebo MPA. Women with a
uterus took active MPA if they were in the HRT group and placebo MPA if they
were in the placebo treatment group. Adherence to treatment was evaluated
using pill counts.
Bone mineral content (BMC) and BMD of the total body, lumbar spine,
and proximal femur were measured at 3-month intervals during the study using
dual-energy x-ray absorptiometry on a QDR-1000/W instrument (Hologic Inc,
Waltham, Mass). Coefficients of variation for these measures in our laboratory
in older women have been reported previously.18
Serum bone-specific alkaline phosphatase (BAP) activity (Metra Biosystems
Inc, Mountain View, Calif), a marker of bone formation, and urinary cross-linked N-telopeptide of type I collagen (NTX) (Ostex International,
Seattle, Wash), a marker of bone resorption, were measured by enzyme-linked
immunosorbent assay. Coefficients of variation for these measurements were
4% to 7%.
Participants completed 3-day food records at the beginning and end of
the study period under supervision by a registered dietitian. Records were
analyzed using Nutritionist IV (First Databank, San Bruno, Calif). Based on
the initial dietary assessment women were provided supplemental calcium and
cholecalciferol to adjust intake to approximately 1200 mg/d and approximately
800 U/d, respectively.
Based on our previous studies of the effects of HRT on BMD of postmenopausal
women,19,20 the mean (SD) differences
in the changes in BMD of the lumbar spine and femoral neck between the placebo
and HRT groups were projected to be 3.9% (3.4%) and 1.5% (1.8%), respectively.
Thus, for the projected sample sizes, the estimated power to detect significant
effects of HRT was 98% for lumbar spine BMD and 83% for total hip BMD.
The primary analyses of BMC and BMD data were carried out in an intention-to-treat
fashion. Independent t tests were used to determine
whether the percent change in outcomes was significantly different in response
to HRT compared with placebo. When follow-up data were not available, the
last observation was carried forward, which yields conservative results. Secondary
analyses were conducted in those women who adhered to treatment and provided
follow-up data. Results are presented as mean (SE) unless otherwise stated.
SAS version 6.08 (SAS Institute, Cary, NC) was used for all statistical analyses.
Of the 67 women enrolled, 54 underwent follow-up evaluations (Figure 1). Two women in the placebo group
dropped out and refused final testing for personal reasons; 11 women in the
HRT group also dropped out and did not undergo final testing. Compared with
the completers, those who dropped out were older (84  vs 81  years, P = .03) and had a lower VO2peak (13  vs
15  mL/min per kg of body weight, P = .04), but
there were no significant differences in baseline BMC and BMD values.
Participants who were unable to tolerate the prescribed estrogen due
to adverse effects (7 HRT, 2 placebo) had the HRT modified. The dosage was
reduced to one 0.625-mg tablet every other day for a period of 1 to 4 weeks.
The dosage was then increased to 1 tablet daily. Women who continued to be
bothered by adverse effects (2 HRT, 1 placebo) returned to a regimen of 3
to 4 tablets per week for the remainder of the study period.
The percentage of prescribed doses taken by women in the placebo group
who completed the study averaged 85% (11%) and 99% (4%) for estrogens and
MPA, respectively. Compliance in the HRT group was 86% (13%) for estrogens
and 86% (33%) for MPA.
The only significant difference between the groups at baseline was the
age at menopause, which was younger in the HRT group (Table 1). Total body BMC and BMD of all regions measured tended
to be higher in the HRT group but differences were not statistically significant.
Based on BMD values for the femoral neck, 91% of women in the placebo group
and 93% in the HRT group were osteopenic or osteoporotic according to the
criteria of the World Health Organization.
There were no significant differences between the groups at baseline
in calcium, vitamin D, or energy intake. Calcium intake averaged 735 (316)
and 671 (231) mg/d in the placebo and HRT groups, respectively, at baseline
and 1353 (256) and 1366 (266) mg/d, respectively, at the end of the study
period. Vitamin D intake averaged 151 (162) and 138 (81) U/d at baseline in
the placebo and HRT groups, respectively, and 756 (107) and 784 (97) U/d,
respectively, at the end of the study period.
Bone density measurements are shown in Table 2 and Figure 2 as
the percentage change from baseline. Based on the intention-to-treat analyses,
increases in total BMC and in BMD of the total body, lumbar spine, total hip,
and trochanter were significantly larger in response to HRT than placebo treatment.
The statistical results remained the same when absolute changes rather than
percentage changes in bone measures were used, as well as when baseline bone
measures were included in the models as covariates.
Secondary data analyses included women who provided final evaluations
and were adherent to treatment. Adherence was defined as taking more than
80% of the prescribed dose over the period of study. There were 29 adherent
women in the HRT group, including 27 who received standard treatment and 2
in whom treatment was modified for only 1 to 2 weeks. The increases in BMC
and BMD tended to be larger in the subset of women who were adherent to HRT
than in all women randomized to HRT (Figure
2). In addition, women adherent to HRT had a significant increase
in femoral neck BMD when compared with placebo treatment (2.5% vs −0.1%,
respectively; between-group difference, 2.6%; 95% confidence interval [CI],
2.1%-3.1%; P = .04).
Urine samples were obtained from 25 women in the HRT group and 16 women
in the placebo group for the assessment of NTX. Serum samples for the measurement
of BAP were obtained from 30 women in the HRT group and 19 women in the placebo
group. Baseline urinary NTX concentrations were 53.2 (6.3) and 52.6 (3.7)
nmol/L of bone collagen equivalents per mmol/L of creatinine and baseline
serum BAP concentrations were 19.6 (1.7) and 20.5 (1.2) U/L for the placebo
and HRT groups, respectively (P>.05 for all). Compared
with placebo, HRT resulted in significant decreases in NTX (−48% vs
4%; between-group difference,−52%; 95% CI,−47% to −55%; P<.001) and BAP (−24% vs 6%; between-group difference,−30%;
95% CI,−26% to−33%; P = .001) (Figure 3). The change in NTX was inversely
correlated with the change in total body BMC (r = −0.33; P = .02). In addition, the changes in BAP and NTX over
the study period were each inversely correlated with changes in total body,
lumbar spine, total hip, and trochanter BMD (correlation coefficients from r = −0.28 to r = −0.46; P<.05 for all).
Because of the prevalence of low BMD and physical frailty in elderly
women, osteoporotic fractures pose a major threat to functional independence.
Although HRT is one of the most effective means of reducing risk for osteoporosis,
there have been few randomized studies of HRT in elderly women,2,5- 8
and none that has focused specifically on women older than 75 years. Our study
therefore provided novel information on the skeletal response to the initiation
of HRT in physically frail elderly women. Hormone replacement therapy resulted
in significant increases in BMD of the lumbar spine and hip regions in these
women who were at high risk for falls and fractures. This finding corroborates
results from previous studies of HRT in younger postmenopausal women.21,22
Results from both the intention-to-treat and the adherence paradigms
indicated that increases in BMD in response to HRT were significantly larger
than the changes that occurred in placebo-treated women. It should be noted
that the placebo group had stable BMD levels over the 9-month period of study.
It is likely that bone loss in this group was attenuated by the calcium and
vitamin D supplementation, as has been observed by others.23,24
The magnitude of increase in lumbar spine BMD (4.3%) after 9 months
of HRT in these elderly women was similar to the increases reported in previous
randomized studies of HRT in women younger than 75 years (3%-8% after 1 year).2,5- 8
Importantly, the increases in total hip BMD (1.7%) and trochanter BMD (2.3%)
after 9 months of HRT in the intention-to-treat analyses were similar to or
greater than those observed in previous studies of older women.2,7,8
Although a significant increase in femoral neck BMD (2.5%) occurred only in
women who were adherent to HRT, this provides evidence that HRT is effective
at this region when taken as prescribed. Because hip BMD is a strong predictor
of hip fractures,25 our findings of positive
effects of HRT on hip BMD have potentially important clinical implications
for preserving the independence of frail elderly women.
The increase in lumbar spine BMD in elderly women after 9 months of
HRT tended to be larger than the increase (3%-3.6%) that occurred in younger
women (45-64 years) in the Postmenopausal Estrogen/Progestin Intervention
(PEPI) trial after the first year of the 3-year study.21
Although comparisons of findings across studies must be approached cautiously,
both the PEPI trial and others2,3
demonstrated that older age and lower initial BMD were associated with a greater
BMD response to HRT. The positive effects of HRT on BMD are presumably due,
in part, to filling of the remodeling space consequent to the suppression
of bone turnover.26,27 Thus, the
robust response of elderly women to HRT is likely related to the high rate
of skeletal turnover in elderly women. It was once commonly believed that
bone turnover remained elevated for only a few years after menopause28- 30 and that bone loss
subsequently slowed or ceased in older women.31
However, recent studies have provided evidence that bone turnover remains
elevated into old age32 and that bone loss
may accelerate rather than slow in elderly persons.1
Indeed, our subjects had high rates of bone turnover, as indicated by elevated
serum BAP and urine NTX concentrations that decreased significantly in response
The standard replacement dose of conjugated estrogens (0.625 mg/d) was
used in the current study. Recently, Recker et al8
reported that a lower dose (0.3 mg/d) coupled with adequate calcium and vitamin
D intake increased BMD in women older than 65 years. Using an intention-to-treat
approach, they found an increase in spine BMD of about 2% after 12 months
of HRT that continued to increase through year 3, peaking at 4%. We observed
an increase in spinal BMD of 4.3% after only 9 months of standard HRT. Among
women who were at least 90% adherent to HRT in the study by Recker et al,
femoral neck BMD increased by 1.6% in 3 years. In the present study, women
who were at least 80% adherent to HRT had an increase in femoral neck BMD
of 2.5% after 9 months. It will be important to further evaluate the dose-response
effects (benefits, risks, and adherence) of HRT in elderly women in a larger
Although there was some early intolerance to HRT, symptoms were substantially
decreased by temporary dose reduction in most cases. Furthermore, medical
reasons unrelated to HRT were primarily responsible for study discontinuation
in our sample. The HRT attrition rate in our study was 24% compared with 8%
to 20% in previous trials of HRT in younger postmenopausal women.2,7,8,21
Our study contributes to the growing evidence suggesting the positive
effects of HRT on the skeletal health of women in late life. Traditional thought
has been that the estrogen-dependent compartment of bone becomes depleted
approximately 15 years after menopause. Riggs et al33
have challenged this concept by proposing a "unitary model," in which estrogen
deficiency is also primarily responsible for the decline in bone mass that
previously had been attributed to the aging process. Although their hypothesis
still must be proven, the effects of estrogens on calcium conservation through
extraskeletal organs may be a plausible mechanism for the positive effects
of HRT on BMD in elderly women. For example, estrogen treatment has been shown
to preserve intestinal responsiveness to vitamin D,34
increase parathyroid hormone (PTH)-independent tubular reabsorption of calcium,35 and decrease PTH secretion during hypocalcemia.36 Other potential mechanisms include the suppressive
effects of estrogens on bone-resorbing cytokines37
associated with the chronic inflammation accompanying aging.38
Our trial had some limitations. The sample size was small, the duration
of HRT was limited to 9 months, and we used a physiological (ie, BMD) rather
than a clinical (ie, fractures) end point. Because we included only women
with physical frailty, it is possible that our findings are relevant only
to frail elderly women willing to take HRT and not to the general population
of elderly women. Although our study suggests that physically frail women
are candidates for the osteogenic benefits of HRT, their shorter life expectancy
may limit the period over which the benefits would accrue.
In summary, HRT has significant osteogenic effects in very old, physically
frail women. However, fracture risk in very old women is due to multiple factors
in addition to low BMD, including sensory and neuromuscular impairments, medications,
and environmental hazards.39,40
Further research is therefore necessary to elucidate the effectiveness of
HRT, alone and in combination with fall-prevention measures, in reducing fracture
rates and postponing disability in elderly women.