Hot flushes and night sweats, also known as vasomotor symptoms, are among the most frequent complaints of women during the menopausal transition, affecting more than 80% of women during the late perimenopausal period and persisting for up to 5 years after menopause in nearly a third of women.1,2 The cause of these common menopausal symptoms is currently unknown, although alterations in the thermoregulatory set-point of the hypothalamus are widely thought to play a role.3,4 Several laboratory-based studies have recently suggested that menopausal women with more severe hot flushes may have lower levels of total plasma antioxidant activity.5,6 As a result, there has been growing interest in vasomotor symptoms as a possible marker of risk of postmenopausal complications such as cardiovascular disease and osteoporosis that are known to be associated with oxidative stress.7
Several epidemiologic studies have pointed to a possible relationship between vasomotor symptoms and bone loss during the menopausal transition.8–11 These studies were small, however, and focused primarily on perimenopausal or early postmenopausal women, rather than older postmenopausal women who are at greater risk of developing complications from bone loss. Among women in the age range in which osteoporosis is most likely to cause clinical problems, the clinical significance of persistent vasomotor symptoms has never been examined, even though 10–20% of older postmenopausal women not on hormone therapy report these symptoms.12–14
We sought to assess whether greater severity of hot flushes is associated with either increased bone loss or risk of fracture in older postmenopausal women who are most likely to develop complications from osteoporosis. To that end, we examined the association between baseline hot flush severity, baseline femoral neck and lumbar spine bone mineral density, and history of vertebral and nonvertebral fractures in a subset of women in the Multiple Outcomes of Raloxifene Evaluation (MORE) trial. We also compared the 3-year change in bone mineral density and the 3-year incidence of fractures among women with varying severity of hot flushes at baseline. Our goal was to provide insight into whether clinical assessment of vasomotor symptoms in high-risk older postmenopausal women may be useful in estimating the risk of osteoporotic fracture.
MATERIALS AND METHODS
The MORE trial evaluated the effects of 3 years of treatment with raloxifene in a population of primarily white women under age 80, all of whom were at least 2 years postmenopausal and met the World Health Organization’s definition of osteoporosis (femoral neck or lumbar spine bone mineral density of more than 2.5 standard deviations below the normal or history of at least one moderate or 2 mild prevalent vertebral fractures regardless of baseline bone mineral density). Women were excluded if they had taken androgens, calcitonin, or bisphosphonates within the previous 6 months or if they had used oral estrogen within the previous 2 months. Women were also prevented from enrolling if they had menopausal symptoms of such severity that administration of hormone replacement therapy was required, but women who started estrogen after randomization were allowed to remain in the study. Complete inclusion and exclusion criteria, as well as procedures for subject recruitment and follow-up in MORE, have been described elsewhere.15 A total of 7,705 women from 25 countries were enrolled in MORE, and a subset of 3,167 English-speaking women (1,049 in the placebo group, 2,118 in the raloxifene group) at sites in the United States completed detailed female history questionnaires at baseline. Women were randomly assigned to receive either raloxifene (60 or 120 mg/d) or placebo, and all women were given daily supplements of 500 mg of calcium and 400–600 International Units of vitamin D. The MORE trial was approved by the human research review boards at all clinical sites, and informed consent was obtained from all participants.
Severity of hot flushes was assessed by self-report questionnaire at baseline. Women were asked, “During the past 6 months, did any [hot flushes] bother you or interfere with your life?" with response options including “all of the time,” “most of the time,” “some of the time,” “little of the time,” and “none of the time.” We decided a priori to collapse these response options into three hot flush severity categories to ensure sufficient numbers of women in each category for analysis. Women who reported bothersome hot flushes “none” or “little” of the time were considered to have “minimal” hot flushes, those who reported hot flushes “some” of the time were considered to have “moderate” hot flushes, and those who reported hot flushes “most” or “all” of the time were considered to have “severe” hot flushes.
Bone mineral density of the femoral neck and lumbar spine was assessed by dual-energy X-ray absorptiometry (DXA) in MORE at baseline and annually thereafter. A central reading facility (Osteoporosis and Arthritis Research Group, University of California, San Francisco, CA) performed quality assurance of the longitudinal and cross-calibration adjustments to the DXA machines and patients’ bone mineral density measurement data and provided correction factors to adjust for differences between sites and over time.15–17 The rate of change in bone mineral density was expressed as annualized percentage change from baseline: 100×([year 3 bone mineral density−initial bone mineral density]/ initial bone mineral density).
To assess vertebral fractures, surveillance vertebral radiography was performed in all MORE participants at baseline, 24 months, and 36 months, as well as at interim 6-month intervals for participants who reported symptoms of vertebral fracture. As previously described, radiographs were analyzed by using a combination of semiquantitative readings and quantitative morphometric analysis at a central site.15 A new incident vertebral fracture was defined as a fracture occurring in a vertebra that was not fractured at baseline.
Nonvertebral fractures were determined by direct questioning of participants at baseline and at interim 6-month visits and validated by X-rays or medical record review. The MORE study protocol defined nonvertebral fractures to include fractures in the clavicle, scapula, ribs, sternum, pelvis, sacrum, coccyx, humerus, forearm, wrist, femur, tibia, fibula, ankle, calcaneus, tarsus, and metarsus, but not the skull, face, metacarpals, fingers, or toes. Fractures resulting from a traffic collision, a beating, or having been struck by a falling or moving object were considered traumatic and excluded from analysis.
Data on participants’ demographic characteristics, medical history, and health-related habits were collected by self-report questionnaire at baseline. Current use of all prescription and over-the-counter medications was also recorded. Prior estrogen replacement therapy was assessed by asking women about any previous use of oral or transdermal estrogen and recording the number of years of previous use. Depression was measured with the Geriatric Depression Scale, a screening instrument validated in elderly populations, in which scores greater than 5 are consistent with clinical depression.18 Participants’ height and weight were also measured at a baseline visit, and body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared.
Serum estradiol concentration was also measured at baseline and assayed by SciCor (Covance Central Laboratory Services, Indianapolis, IN) using a double antibody procedure. The assay process has been described in detail elsewhere.19 Serum estradiol concentrations were reported by the laboratory as 5.0 pmol/L or less or as exact values for estradiol levels greater than 5.0 pmol/L. For those 1,321women (42%) with estradiol levels less than 5.0 pmol/L, the value of 2.5 was used in analysis.
Of the 7,705 women enrolled in MORE, the 3,167 women who completed female history questionnaires were eligible for inclusion in our cross-sectional analyses of the associations between baseline hot flush severity, baseline bone mineral density, and prevalent fracture. Of these women, data on 3-year change in bone mineral density was available for 2,418 participants (76% of our baseline cohort), and data on incident fractures were available for 2,770 participants (87% of the cohort).
Baseline characteristics of women in each hot flush severity category were compared by using analysis of variance for continuous variables with normal distributions (evaluated by examining a graphic plot of the data distribution), Kruskal-Wallis tests for those with skewed distributions, and χ2 tests for categorical data. The least square means procedure was used to examine 1) the adjusted mean baseline bone mineral density at the femoral neck and lumbar spine and 2) the adjusted 3-year annualized percentage change in femoral neck and lumbar spine bone mineral density, among women with minimal, moderate, and severe hot flushes at baseline. We then assessed for a linear trend in the relationship between baseline severity of hot flushes and both baseline bone mineral density and 3-year annualized percentage change in bone mineral density, using a test that evaluated for orthogonal contrast in the coefficients corresponding to each level of hot flush severity.20 Years since menopause, BMI, and race were included as covariates in all models, while other variables reported to be associated with bone loss in previous research were included only if they were associated with baseline severity of hot flushes at P<.10 in univariable analysis. One covariate, age, was not included in the models due to collinearity with years since menopause (Spearman rank correlation ρ=0.73), although it was associated with baseline hot flush severity. Because previous analyses have indicated that raloxifene increases bone density, decreases vertebral fracture incidence, and increases the severity of hot flushes,21,22 we initially stratified our longitudinal bone mineral density analyses by treatment assignment. Because statistical tests for interaction with treatment were nonsignificant (P>.20 for all), unstratified longitudinal analyses in which treatment assignment was included as a covariate were also performed.
Multivariate logistic regression was used to compare 1) the baseline prevalence of vertebral fractures and nonvertebral fractures after age 45, and 2) the 3-year incidence of vertebral fractures in women with moderate or severe hot flushes compared with minimal hot flushes at baseline. Proportional hazards analysis was used to assess the 3-year incidence of nonvertebral fracture in women with moderate or severe hot flushes compared with minimal hot flushes at baseline.23 Because there were too few fracture events among women with moderate and those with severe hot flushes, we were not able to examine these two groups separately in our fracture models. Years since menopause, BMI, and race were again included as covariates in all multivariable fracture models. Other variables reported to be associated with bone loss in previous research were included in our multivariable models only if they demonstrated an association with baseline hot flush severity at P<.10 in univariable analysis (with the exception of years since menopause, due to colinearity with age). Although statistical tests for interaction with treatment were nonsignificant (P>.20 for both), we initially stratified our longitudinal fracture analyses by treatment assignment and then conducted unstratified analyses in which treatment assignment was included as a covariate. For our incident fracture models, power calculations using a two-sided α suggested that the data provided 80% power to detect either a 1.9-fold increase or a 0.4-fold decrease in the odds of incident vertebral fracture in multivariable models using the unstratified cohort, along with 80% power to detect either a 1.7-fold increase or a 0.5-fold decrease in the odds of incident nonvertebral fracture. All analyses were preformed with SAS 9 (SAS Institute Inc, Cary, NC). This study was approved by the institutional review board of the University of California San Francisco.
The mean age of the 3,167 participants in our baseline analyses was 67±7 years, and 96% of women (n=3,037) were 5 or more years postmenopausal at the start of the trial. Approximately 2% of women reported severe hot flushes (n=54), 10% reported moderate hot flushes (n=321), and 88% reported minimal hot flushes (n=2,792) at baseline. Baseline characteristics of these participants by hot flush severity are shown in Table 1.
The mean baseline bone mineral densities at the femoral neck and lumbar spine in this cohort were 0.611±0.069 g/cm2 and 0.839±0.133 g/cm2, respectively. After adjusting for years since menopause, race, BMI, years of education, prior estrogen therapy, current selective serotonin reuptake inhibitor (SSRI) use, depression, and serum estradiol level, women with more severe hot flushes at baseline were found to have higher, rather than lower, baseline bone mineral density at both the femoral neck and lumbar spine (Table 2). The adjusted mean femoral neck bone mineral densities of women with moderate and severe hot flushes were 0.01 g/cm2 and 0.02 g/cm2 higher, respectively, than those of women with minimal hot flushes at baseline (P for trend across hot flush categories<.001). Similarly, the adjusted mean lumbar spine bone mineral densities of women with moderate and severe flushes were 0.02 g/cm2 and 0.04 g/cm2 higher, respectively, than those of women with minimal hot flushes at baseline (P for trend across hot flush categories=.008). When we excluded serum estradiol level from our multivariate models as a possible mediator of the relationship between hot flush severity and baseline bone mineral density, the association between greater hot flush severity and baseline bone mineral density was relatively unchanged and still strongly significant (P for trend<.001 for femoral neck bone mineral density and P=.004 for lumbar spine bone mineral density). Results were also substantially unchanged when we excluded the small minority of participants (n=118) who were only 2–5 years postmenopausal.
Nearly a third of women in this cohort (n=1,005) had a prevalent vertebral fracture at baseline, while slightly over 40% of women (n=1,337) had a history of nonvertebral fracture occurring after age 45 years. Women with moderate or severe hot flushes were less likely to have a prevalent vertebral fracture at baseline than women with minimal hot flushes, after controlling for years since menopause, race, BMI, years of education, prior estrogen therapy, current SSRI use, depression, and serum estradiol level (odds ratio [OR] 0.64, 95% confidence interval [CI] 0.48–0.84, P=.002). Women with more severe hot flushes at baseline were also less likely to have a history of nonvertebral fracture occurring after age 45 (OR 0.77, 95% CI 0.60–0.98, P=.035) in multivariable-adjusted models. When we excluded serum estradiol level from our models as a possible mediator of the relationship between hot flush severity and fracture history, the association between hot flush severity and both prevalent vertebral fracture and previous nonvertebral fracture after age 45 was relatively unchanged and still significant. Exclusion of women who were 2–5 years postmenopausal did not substantially change these results.
Of the 2,418 women for whom data on 3-year change in bone mineral density was available, 2,135 women reported minimal hot flushes, 245 women reported moderate hot flushes, and 38 women reported severe hot flushes at baseline. In both the placebo and the raloxifene groups, the 3-year percentage annualized change in bone mineral density at the femoral neck and lumbar spine did not differ significantly among women reporting more severe compared with those reporting less severe hot flushes at baseline, after adjusting for years since menopause, race, BMI, years of education, years of prior estrogen therapy, SSRI use, depression, and serum estradiol level (Table 3). These results were not significantly different when serum estradiol level was not included in the models or when we excluded women who were only 2–5 years postmenopausal.
Of the 2,770 participants for whom data on incident fractures were available, 2,445 had minimal hot flushes, 284 women had moderate hot flushes, and 41 women had severe hot flushes. A total of 73 incident vertebral fractures and 92 incident nonvertebral fractures were recorded over 3 years in women randomized to placebo, compared with 98 new vertebral fractures and 174 new nonvertebral fractures in women randomized to raloxifene. We did not find a significant association between baseline hot flush severity and 3-year risk of incident fractures after adjusting for years since menopause, race, BMI, years of education, years of prior estrogen therapy, SSRI use, depression, and serum estradiol level (Table 4). In the combined cohort, the 95% confidence intervals for our multivariable-adjusted odds ratios excluded a more than 15% increase and a more than 68% decrease in the adjusted odds of incident vertebral fracture among women with moderate-to-severe compared with those with minimal hot flushes. Similarly, confidence intervals excluded a more than 64% increase and a more than 27% decrease in the adjusted odds of incident nonvertebral fracture, indicating that a clinically important association between hot flush severity and incident fracture risk was unlikely. These results were not significantly different when we excluded serum estradiol level from the models or when we excluded women who were 2–5 years postmenopausal.
In this cohort of older women with pre-existing osteoporosis, the presence of moderate or severe hot flushes more than 5 years after menopause was associated with higher, not lower, baseline bone mineral density at the femoral neck and lumbar spine, and there was no association between severity of hot flushes at baseline and either bone loss or incidence of vertebral or nonvertebral fracture during 3 years of follow-up among women randomized to either raloxifene or placebo. Taken as a whole, these findings suggest that greater severity of hot flushes does not convey a substantially increased risk of either progressive bone density loss or clinical fracture among older postmenopausal women who are at greatest risk of developing complications from osteoporosis.
Several previous studies have pointed to a positive association between vasomotor symptoms and bone loss, but these focused primarily on perimenopausal and early postmenopausal women, who would not be expected to be at risk for complications of osteoporosis. One small study of middle-aged Swedish women found that those with frequent “climacteric symptoms” had greater forearm bone loss over 2 years than those without symptoms.9 Another analysis of 290 Australian women aged 44–50 years with ongoing menstrual cycles found that women with hot flushes had significantly lower bone mineral density at the spine and hip compared with those without hot flushes.11 Although two previous studies of menopausal symptoms have examined bone mineral density outcomes in older postmenopausal women,10,24 both required women to recall the severity of symptoms experienced in the early menopausal period. In both studies, the average time since menopause was more than 15 years, raising concerns about the reliability of patients’ subjective recall of their symptoms. Additionally, neither study addressed fracture outcomes, although fracture prevention is the primary reason for addressing bone loss in postmenopausal women.
Our finding that older postmenopausal women with more severe hot flushes had higher rather than lower baseline bone mineral density is striking in light of findings in early postmenopausal women. One explanation is that the observed association between hot flush severity and baseline bone mineral density was confounded by factors such as health status or physical activity level that were inconsistently evaluated in MORE. Higher health status and physical activity level are known to be protective against bone loss in older women, although their relationship to hot flushes is not well understood. In the subset of 1,211 participants for whom physical activity data were available, however, we did not find any significant association between physical activity and hot flush severity. Further research is needed to assess whether there may be biological mechanisms underlying a possible protective association between hot flushes and bone loss after menopause.
This analysis benefits from a large population of older postmenopausal women as well as careful assessment of both bone density loss and fracture outcomes over time. Nevertheless, several limitations of this research should be noted. First, severity of hot flushes was assessed by self-report questionnaire only, without correlation using methods such as daily diaries or sternal skin conductance monitoring. Previous methodological studies have supported the accuracy and consistency of women’s self-report of their hot flushes, however, in the absence of electronic monitoring.25 Furthermore, this study was conducted in women who already had significantly reduced bone mineral density at baseline, in which the rate of ongoing bone density loss was fairly low. It is possible that, in a population of women with average bone mineral density at baseline, a different relationship between vasomotor symptoms and bone density loss would have emerged.
In conclusion, we did not find an association between severity of hot flushes and either progressive bone density loss or clinical risk of fracture among older postmenopausal women with pre-existing osteoporosis. Our findings suggest that clinical assessment of vasomotor symptoms should not be used to identify older postmenopausal women at high risk for osteoporotic fracture.
1. Stearns V, Ullmer L, Lopez JF, Smith Y, Isaacs C, Hayes D. Hot flushes. Lancet 2002;360:1851–61.
2. Vasomotor symptoms. Obstet Gynecol 2004;104 suppl:106S–17S.
3. Freedman RR. Pathophysiology and treatment of menopausal hot flashes. Semin Reprod Med 2005;23:117–25.
4. Freedman RR, Subramanian M. Effects of symptomatic status and the menstrual cycle on hot flash-related thermoregulatory parameters. Menopause 2005;12:156–9.
5. Leal M, Diaz J, Serrano E, Abellan J, Carbonell LF. Hormone replacement therapy for oxidative stress in postmenopausal women with hot flushes. Obstet Gynecol 2000;95:804–9.
6. Leal Hernandez M, Abellan Aleman J, Carbonell Meseguer LF, Diaz Fernandez, J, Garcia Sanchez, FA, Martinez Selva, JM.. Influence of the presence of hot flashes during menopause on the metabolism of nitric oxide: effects of hormonal replacement treatment [in Spanish]. Med Clin (Barc) 2000;114:41–5.
7. van der Schouw YT, Grobbee DE. Menopausal complaints, oestrogens, and heart disease risk: an explanation for discrepant findings on the benefits of post-menopausal hormone therapy. Eur Heart J 2005;26:1358–61.
8. Menon RK, Okonofua FE, Agnew JE, Thomas M, Bell J O’Brien PM, et al. Endocrine and metabolic effects of simple hysterectomy. Int J Gynaecol Obstet 1987;25:459–63.
9. Naessen T, Persson I, Ljunghall S, Bergstrom R. Women with climacteric symptoms: a target group for prevention of rapid bone loss and osteoporosis. Osteoporos Int 1992;2:225–31.
10. Lee SJ, Kanis JA. An association between osteoporosis and premenstrual symptoms and postmenopausal symptoms. Bone Miner 1994;24:127–34.
11. Salamone LM, Gregg E, Wolf RL, Epstein RS, Black D, Palermo L, et al. Are menopausal symptoms associated with bone mineral density and changes in bone mineral density in premenopausal women? Maturitas 1998;29:179–87.
12. Diem S, Grady D, Quan J, Vittinghoff E, Wallace R, Hanes V, et al. Effects of ultralow-dose transdermal estradiol on postmenopausal symptoms in women aged 60 to 80 years. Menopause 2006;13:130–8.
13. Barnabei VM, Grady D, Stovall DW, Cauley JA, Lin F, Stuenkel CA, et al. Menopausal symptoms in older women and the effects of treatment with hormone therapy. Obstet Gynecol 2002;100:1209–18.
14. Barnabei VM, Cochrane BB, Aragaki AK, Nygaard I, Williams RS, McGovern PG, et al. Menopausal symptoms and treatment-related effects of estrogen and progestin in the Women’s Health Initiative. Obstet Gynecol 2005;105:1063–73.
15. Ettinger B, Black DM, Mitlak BH, Knickerbocker RK, Nickelsen T, Genant HK, et al. Reduction of vertebral fracture risk in postmenopausal women with osteoporosis treated with raloxifene: results from a 3-year randomized clinical trial. Multiple Outcomes of Raloxifene Evaluation (MORE) Investigators. JAMA 1999;282:637–45.
16. Lu Y, Mathur AK, Blunt BA, Gluer CC, Will AS, Fuerst TP, et al. Dual X-ray absorptiometry quality control: comparison of visual examination and process-control charts. J Bone Miner Res 1996;11:626–37.
17. Lu Y, Ye K, Mathur AK, Hui S, Fuerst TP, Genant HK. Comparative calibration without a gold standard. Stat Med 1997;16:1889–905.
18. Sheikh JI. Geriatric Depression Scale (GDS): recent evidence and development of a shorter version. In: Brink TL, editor. Clinical gerontology: a guide to assessment and intervention. New York (NY): The Haworth Press; 1986. p. 165–73.
19. Modelska K, Litwack S, Ewing SK, Yaffe K. Endogenous estrogen levels affect sexual function in elderly post-menopausal women. Maturitas 2004;49:124–33.
20. Vittinghoff E, Glidden D, Shiboski S, McCulloch C. Regression methods in biostatistics: linear, logistic, survival, and repeated measures models. New York (NY): Springer; 2005.
21. Ettinger B, Ensrud KE, Wallace R, Johnson KC, Cummings SR, Yankov V, et al. Effects of ultralow-dose transdermal estradiol on bone mineral density: a randomized clinical trial. Obstet Gynecol 2004;104:443–51.
22. Jolly EE, Bjarnason NH, Neven P, Plouffe L Johnston CC, Jr, Watts SD, et al. Prevention of osteoporosis and uterine effects in postmenopausal women taking raloxifene for 5 years. Menopause 2003;10:337–44.
23. Kalbfleisch JD, Prentice RL. The statistical analysis of failure time data. New York (NY): Wiley; 1980.
24. von Muhlen DG, Soroko S, Kritz-Silverstein Barrett-Connor E. Vasomotor symptoms are not associated with reduced bone mass in postmenopausal women: the Rancho Bernardo Study. J Womens Health Gend Based Med 2000;9:505–11.
25. Sloan JA, Loprinzi CL, Novotny PJ, Barton DL, Lavasseur BI, Windschitl H. Methodologic lessons learned from hot flash studies. J Clin Oncol 2001;19:4280–90.