Huang AJ, Grady D, Jacoby VL, Blackwell TL, Bauer DC, Sawaya GF. Persistent Hot Flushes in Older Postmenopausal Women. Arch Intern Med. 2008;168(8):840-846. doi:10.1001/archinte.168.8.840
To examine the prevalence, natural history, and predictors of hot flushes in older postmenopausal women.
Prevalence, severity, and 3-year change in severity of hot flushes were assessed by questionnaire in 3167 older postmenopausal women with osteoporosis. Logistic regression was used to identify characteristics associated with symptoms at baseline and after 3 years of follow-up.
At baseline, 375 women (11.8%) reported bothersome hot flushes. Women were more likely to have baseline symptoms if they were less educated (odds ratio [OR], 1.28; 95% confidence interval [CI], 1.06-1.53 per 4-year decrease), more recently menopausal (OR, 1.44; 95% CI, 1.34-1.56 per 5-year decrease), had previously used estrogen (OR, 1.57; 95% CI, 1.23-2.00), or had undergone hysterectomy (OR, 1.51; 95% CI, 1.14-1.99). Hot flushes were also associated with higher body mass index (OR, 1.22; 95% CI, 1.08-1.38 per 1 SD), higher follicle-stimulating hormone levels (OR, 1.34; 95% CI, 1.20-1.51 per 1 SD), lower high-density lipoprotein levels (OR, 1.17; 95% CI, 1.03-1.34 per 1 SD decrease), vaginal dryness (OR, 1.52; 95% CI, 1.19-1.93), and trouble sleeping (OR, 2.48; 95% CI, 1.94-3.16), but not estradiol levels. Of the 375 women with baseline symptoms, 278 contributed 3-year data, and 157 (56.5%) of these women reported persistent symptoms after 3 years. Fewer years since menopause (OR, 1.15; 95% CI, 1.01-1.32 per 5-year decrease) and trouble sleeping (OR, 1.97; 95% CI, 1.19-3.26) were associated with symptom persistence.
For a substantial minority of women, hot flushes are a persistent source of discomfort into the late postmenopausal years. Identification of risk factors for hot flushes may help guide evaluation and treatment in this population.
Hot flushes are among the most frequent complaints of women during the menopausal transition, affecting up to 80% of women within the first year of cessation of menses.1,2 These common symptoms have been shown to negatively affect quality of life by disturbing sleep, interfering with work and leisure activities, and exacerbating anxiety and depression.3- 5 With several randomized controlled trials pointing to the negative long-term effects of systemic estrogen therapy in postmenopausal women,6,7 there is growing interest in identifying modifiable risk factors for hot flushes and developing alternate therapies for these symptoms.
In most women, hot flushes resolve within a few years of menopause, but some women report symptoms for many years after they cease to menstruate. Very little attention has been given to the substantial minority of women who continue to have hot flushes 5 or more years after menopause, even though it is these women who appear to be at greatest risk of suffering adverse effects from using estrogen to treat their symptoms.6 At this time, the prevalence of hot flushes in older postmenopausal women has not been well documented, predictors of symptoms in older postmenopausal women have not been examined, and it is not clear why hot flushes resolve in some women, but continue for many years in others.
To address this issue, we analyzed data from 3167 women enrolled in the Multiple Outcomes of Raloxifene Evaluation (MORE) trial, of whom more than 95% were 5 or more years postmenopausal. Our goal was to describe the natural history of hot flushes in older postmenopausal women and to identify factors associated with persistence of hot flushes in this population.
The MORE trial evaluated the effects of 3 years of treatment with the selective estrogen receptor modulator raloxifene hydrochloride in a population of women younger than 80 years, all of whom were at least 2 years postmenopausal and met the World Health Organization criteria for osteoporosis (femoral neck or lumbar spine bone mineral density >2.5 SDs below normal, or history of at least 1 moderate or 2 mild prevalent vertebral fractures regardless of baseline bone density). Women were excluded if they had taken androgens, calcitonin, or bisphosphonates in the previous 6 months, had used oral estrogen in the previous 2 months, or required estrogen for treatment of severe menopausal symptoms. Complete inclusion and exclusion criteria, as well as procedures for subject recruitment and follow-up in MORE, have previously been described.8 A total of 7705 women from 25 countries were randomized to placebo or to 60 or 120 mg of raloxifene daily, and English-speaking women from the 65 participating sites in the United States (n = 3167; 1049 in the placebo group and 2118 in the raloxifene groups) completed detailed female history questionnaires, including questions about menopausal symptoms. The MORE trial was approved by the human research review boards at all sites, and informed consent was obtained from all participants.
Severity of hot flushes was assessed at baseline and after 3 years of follow-up. Women were asked, “During the past 6 months, did any [hot flushes] bother you or interfere with your life?” Response options included “all of the time,” “most of the time,” “some of the time,” “little of the time,” and “none of the time.” For the purposes of analysis, women who reported bothersome hot flushes “some,” “most,” or “all” of the time were considered to have “clinically significant” hot flushes, while those who reported bothersome hot flushes “none” or “little” of the time were considered to have no clinically significant hot flushes. Two other symptoms found to be associated with menopause by the 2005 National Institutes of Health State-of-the-Science Panel on treatment of menopausal symptoms,9 vaginal dryness and trouble sleeping, were also assessed using similarly worded questions.
Data on participants' demographic characteristics, medical history, health-related habits, and medication use were recorded at baseline. Depression was measured using the Geriatric Depression Scale, a screening instrument validated in elderly populations; scores of more than 5 are consistent with clinical depression.10 Participants' height and weight were also measured, and body mass index was calculated as weight in kilograms divided by height in meters squared.
Serum total estradiol level was measured at baseline by SciCor (Covance) Central Laboratory Services (Indianapolis, Indiana) using a double antibody assay process.11 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 1321 women (41.7%) with serum estradiol levels of 5.0 pmol/L or less, a value of 2.5 was extrapolated for analysis (to convert estradiol to picograms per milliliter, divide by 3.671). Follicle-stimulating hormone (FSH) and thyroid-stimulating hormone levels were also measured by SciCor (Covance) using microparticle enzyme immunoassay technology. Fasting blood glucose, high-density lipoprotein (HDL), low-density lipoprotein, and triglyceride levels were also measured.
Multivariable logistic regression was used to identify demographic and clinical characteristics associated with having clinically significant hot flushes among participants at baseline. We evaluated characteristics implicated in previous epidemiologic studies of hot flushes, including years since menopause, race, education level, smoking, alcohol use, body mass index, hysterectomy, oophorectomy, depression, physical activity, serum estradiol levels, and serum FSH levels. In addition, we evaluated associations with medications that might modulate neurohormonal or vascular mechanisms contributing to hot flushes (ie, selective serotonin reuptake inhibitors, tricyclic antidepressants, β-blockers, calcium channel blockers, nitrates, α-blockers, aspirin, statins, and nonsteroidal anti-inflammatory drugs). In light of recent research pointing to an interaction between severity of hot flushes and the effects of estrogen therapy on cardiovascular disease in older postmenopausal women,6 we also examined associations with cardiovascular variables such as coronary heart disease history, serum cholesterol levels, and fasting glucose levels.
Predictor variables that were associated with clinically significant hot flushes at a significance of P≤.10 in univariate analyses were initially entered into a multivariable model. Age was excluded from the regression model despite being associated at P≤ .10 in univariate analysis because of colinearity with years since menopause (Spearman rank correlation ρ = 0.73). Backward elimination was then used to exclude predictor variables associated at P>.10. Those variables that were associated with clinically significant hot flushes at P<.05 in adjusted analysis were considered to be independent risk factors for this outcome.
Among the subset of women who reported clinically significant hot flushes at baseline and contributed hot flush data after 3 years (n = 278), we used backward selection logistic regression to identify characteristics associated with having persistent hot flushes after 3 years. Because the proportion of women with persistent hot flushes after 3 years did not differ significantly for women randomized to raloxifene vs placebo in this population (P = .67), we combined data from both treatment groups for these longitudinal analyses. However, because previous research has suggested that raloxifene increases the incidence of hot flushes,12 we retained treatment assignment as a covariate in our multivariable model to confirm that the associations between our other predictor variables and our outcome were not affected by whether women were randomized to raloxifene vs placebo. All analyses were performed using SAS statistical software, version 9.1 (SAS Institute, Inc, Cary, North Carolina). These analyses were approved by the institutional review board of the University of California–San Francisco.
The mean (SD) age of participants was 67 (7) years, and the mean (SD) number of years since menopause was 19 (9). Of the 3167 participants, 3002 were white (94.8%), 3037 were 5 or more years postmenopausal (95.9%), 361 had previously undergone bilateral oophorectomy (11.3%), and 1379 had previously used oral or transdermal estrogen (43.5%) (Table 1). At baseline, 375 women reported experiencing clinically significant hot flushes that were bothersome some, most, or all of the time in the previous 6 months (11.8%) (Table 2). The prevalence of hot flushes was inversely related to time since menopause. Of the 296 women who were only 2 to 5 years postmenopausal, 52 reported clinically significant hot flushes (44.9%); of the 1601 women who were 20 or more years postmenopausal, 125 reported clinically significant hot flushes (7.8%) (P<.01 for trend by years since menopause). In contrast, there was a more gradual decrease in the prevalence of vaginal dryness with increasing years since menopause, and no statistically significant decrease in the prevalence of trouble sleeping with increasing years since menopause (Table 3).
In multivariate analyses, women were more likely to have clinically significant hot flushes if they were less educated, were more recently menopausal, had undergone hysterectomy, or had previously used systemic estrogen (Table 4). Hot flushes were also associated with higher body mass index, higher serum FSH levels, and lower serum HDL cholesterol levels. The association with lower HDL cholesterol levels remained significant even after additional adjustment for use of statin medications. Women with clinically significant hot flushes were also more likely to concomitantly report vaginal dryness or trouble sleeping (Table 4).
In multivariate analysis, no significant associations were seen with smoking, alcohol use, oophorectomy (independent of hysterectomy), depression, aerobic activity, medications other than estrogen, serum estradiol levels, serum LDL cholesterol levels, fasting blood glucose levels, or serum thyroid-stimulating hormone levels. The association of serum estradiol levels with hot flushes remained nonsignificant (P>.10) even when we eliminated participants whose estradiol levels had to be extrapolated because they were below our minimum assay threshold.
Of 375 women who reported clinically significant hot flushes at baseline, data on hot flush severity at the 3-year follow-up visit were available for 278 women (74.1%), but unavailable for 97 women (25.9%). Compared with the 278 women with 3-year data, the 97 women with missing follow-up data were slightly older (mean [SD] years since menopause, 18  vs 15 ), reported more extensive prior estrogen use (mean [SD] years of estrogen use, 2.8 [4.7] vs 1.7 [3.3]), had higher rates of depression (12 [12.4%] vs 12 [4.3%]), and had lower rates of aspirin use (24 [24.7%] vs 112 [40.3%]) and nonsteroidal anti-inflammatory drug use (43 [44.3%] vs 159 [57.2%]), but higher rates of β-blocker use (14 [14.4%] vs 17 [6.1%]). Of 278 women for whom follow-up data were available, 50% reported hot flushes that were unchanged or worse after 3 years, and this proportion did not differ significantly for women randomized to raloxifene vs placebo (P = .67). Hot flushes were more likely to resolve after 3 years if women were farther away from menopause (P<.01 for trend by years since menopause). Among women who were 2 to 5 years postmenopausal and had hot flushes at baseline, 66% reported persistent hot flushes after 3 years, compared with 46% of women who were 20 years postmenopausal (Table 5).
Adjusting for treatment assignment, the only independent predictors of continuing to have significant hot flushes after 3 years in multivariate analysis were fewer years since menopause (odds ratio [OR], 1.15; 95% confidence interval [CI] , 1.01-1.32 for each 5-year decrease) and trouble sleeping (OR, 1.97; 95% CI, 1.19-3.26). There was also a borderline association with baseline fasting blood glucose level (OR, 0.78; 95% CI, 0.59-1.04 per 1 SD increase in blood glucose).
These analyses of data from a large clinical trial demonstrate that for a substantial minority of women, hot flushes are a significant source of discomfort and distress well into the late postmenopausal years. Among women who were 5 to 9 years postmenopausal, more than 20% reported clinically significant hot flushes, and among those 10 or more years postmenopausal, nearly 10% reported clinically significant hot flushes. Furthermore, more than half of women who complained of significant hot flushes at baseline continued to report persistently bothersome symptoms after 3 years.
Although most research on hot flushes has focused on younger perimenopausal or early postmenopausal women, a few previous studies involving older postmenopausal women have also pointed to a significant prevalence of hot flushes in this population. An analysis of 2763 postmenopausal women aged 55 to 88 years with preexisting coronary artery disease found that 3% had “very frequent” and 13% had “somewhat frequent” hot flushes.13 Among women aged 50 to 79 enrolled in the Women's Health Initiative estrogen plus progestin trial, slightly more than 12% reported “moderate or severe” vasomotor symptoms at baseline. In a survey of women living in Gothenberg, Sweden, 15% of women aged 66 years and 9% of women aged 72 years reported “any” hot flushes, although no assessment of frequency or severity of hot flushes was performed.14
Greater severity of hot flushes was inversely associated with years of education in our study, a finding that has also been noted in studies of perimenopausal and early postmenopausal women.15,16 One theory that has been proposed to explain this association is that highly educated women may be more likely to follow healthier lifestyle habits such as regular physical exercise that may ameliorate the severity of menopausal symptoms15; however, other studies have reported that physical exercise may be detrimental to hot flushes.17,18 Regardless, aerobic activity, alcohol use, and tobacco use were not independently associated with symptoms in our population. Recently, several laboratory-based studies have suggested that hot flushes may be potentiated by increased neurosympathetic activation,19 and it is possible that differences in neurohormonal stress response, which may be increased in individuals of lower socioeconomic status,20 may mediate a relationship between hot flushes and lower educational status.
In our study, women who had undergone hysterectomy were more likely to report hot flushes, regardless of whether they had also undergone oophorectomy. Although some researchers have hypothesized that hysterectomy may cause vascular compromise of the ovaries and earlier ovarian failure,21- 23 this phenomenon would not explain the persistently increased risk of hot flushes in women likely to be remote from hysterectomy. Another possibility is that the association between hysterectomy and hot flushes is mediated by the original indication for hysterectomy rather than any direct effect of the procedure itself. Women who undergo hysterectomy for problems such as dysfunctional uterine bleeding may differ hormonally or physiologically in ways that would make them more prone to hot flushes. Alternatively, it is possible that women who undergo hysterectomy for benign conditions may differ in their subjective experience of and willingness to report symptoms such as hot flushes.
We found a significant association between higher body mass index and hot flushes in our study, even though overweight women tend to have higher estrogen levels as a result of adipoctye-based aromatization of estrone and conversion of androstenedione to estrone.24 Previous researchers noting this same paradoxical relationship in younger menopausal women have hypothesized that higher body mass index may lead to earlier ovarian failure, resulting in more severe hot flushes in midlife.5 Nevertheless, this phenomenon would not explain a persistent association between higher body mass index and hot flushes in women who are 5 or more years past menopause. Another hypothesis is that the greater heat insulation afforded by greater adiposity in overweight women leads to more hot flushes as the thermoneutral zone narrows during menopause.25
A notable finding of our study is that serum FSH levels, rather than estradiol levels, are associated with greater severity of hot flushes in older postmenopausal women. A previous analysis of women enrolled in the Study of Women Across the Nation also found an association between vasomotor symptoms and FSH levels, even after adjusting for serum estradiol and sex hormone binding globulin levels.26 Our findings suggest that nonestrogen feedback systems may be important in modulating severity of hot flushes even among women who are many years away from menopause, when FSH levels have typically stabilized or are on the decline.
Our finding that serum estradiol levels are not significantly associated with hot flush severity is consistent with research involving younger postmenopausal women in the Study of Women Across the Nation, in which the relationship between serum estradiol and hot flushes was found to be mediated by serum FSH. Our study used a sensitive assay capable of detecting an estradiol level as low as 5 pmol/L, and elimination of participants with levels less than 5 pmol/L did not substantially change our analysis results. This, once again, suggests that nonestrogen feedback systems or differences in estrogen activity aside from serum estrogen levels may play a greater role in hot flushes in older postmenopausal women.
We are not aware of any previous studies that have reported an association between serum HDL cholesterol levels and hot flushes in older women, although a previous analysis of 49 women found that those with significant hot flushes had higher levels of lipid peroxidation products.27 It is possible that the observed association between serum HDL and hot flushes in our study reflects biological effects of estrogen that were not adequately captured by serum estradiol levels. Previous studies have documented that exogenous estrogen therapy increases serum HDL levels after menopause,28 but the relationship of endogenous estrogen to serum cholesterol levels in women is unclear.29
Interestingly, in our study, older postmenopausal women with clinically significant hot flushes were more likely to report other symptoms commonly associated with menopause such as vaginal dryness, independent of FSH levels, estradiol levels, or other characteristics. Even though the natural history of hot flushes differs from that of vaginal dryness, with the latter persisting even longer into the late postmenopausal years,16 it is possible that these different clinical manifestations of menopause may share underlying mechanisms other than estrogen deficiency. Previous research has suggested that androgens may affect postmenopausal women's experience of both hot flushes and vaginal dryness30,31; psychological factors may also be important in determining women's subjective distress in response to both types of symptoms.32
The characteristic most strongly associated with hot flushes in our study was trouble sleeping, even though the latter did not tend to improve with increasing time since menopause. Multiple previous studies have found an association between hot flushes and subjective sleeping disturbances, leading some researchers to conclude that hot flushes disrupt sleep; however, several laboratory-based studies using objective measures have documented that hot flushes and night sweats tend to follow, rather than precede, arousals and awakenings from sleep.33,34 Rather than being a cause of sleep disturbances, therefore, hot flushes may be a comorbid symptom of menopause that shares common underlying triggers.
Our analyses benefit from a large population of older postmenopausal women, assessment of a broad spectrum of demographic and clinical characteristics, and sensitive assays to detect the low levels of circulating estradiol characteristic of older women. Nevertheless, several limitations of this research should be noted. First, MORE was a study of women with preexisting osteoporosis, and our results may not be generalizable to women with preserved bone density who may have different hormone profiles or responsiveness to hormonal effects. Second, symptoms were assessed only among English-speaking women at US sites, most of whom were white, and our results may not be generalizable to women of other nationalities and ethnic backgrounds.
Furthermore, severity of hot flushes was assessed by self-report questionnaire only, without correlation using methods such as daily diaries or skin conductance monitoring, and participants' recall of their symptoms may not be completely reliable. Because the question used to assess hot flushes emphasized symptoms that were bothersome or interfered with women's lives, our study may underestimate symptoms that were prevalent but minimally disruptive. Finally, one of the exclusion criteria for the MORE trial was the presence of postmenopausal symptoms of such severity that they required hormone therapy. We do not have data on how many women were excluded for this reason, and our study may underestimate the prevalence of severe hot flushes in the later postmenopausal period.
In conclusion, a substantial minority of women who are 5 or more years postmenopausal have clinically significant hot flushes, and more than half of older postmenopausal women presenting with hot flushes can be expected to have persistent hot flushes after 3 years. Further research is needed to explore the mechanisms underlying the relationship of educational status, obesity, hysterectomy, vaginal symptoms, and trouble sleeping to hot flushes in this population.
Correspondence: Alison J. Huang, MD, MPhil, 1635 Divisadero St, Ste 600, San Francisco, CA 94115 (email@example.com).
Accepted for Publication: November 7, 2007.
Author Contributions: Dr Huang had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Huang, Grady, and Sawaya. Analysis and interpretation of data: Huang, Grady, Jacoby, Blackwell, Bauer, and Sawaya. Drafting of the manuscript: Huang and Sawaya. Critical revision of the manuscript for important intellectual content: Huang, Grady, Jacoby, Blackwell, Bauer, and Sawaya. Statistical analysis: Blackwell. Study supervision: Huang and Grady.
Financial Disclosure: None reported.
Funding/Support: The MORE trial was funded by Eli Lilly and Company. This study was supported in part by a Clinical and Transitional Science Award for Medical Research KL2 grant RR024130 from the National Center for Research Resources, National Institutes of Health (NIH) (Dr Huang) and a Women's Reproductive Health Research Career Development K12 grant HD001262 from the National Institute of Child Health and Human Development and the NIH Office of Research on Women's Health (Dr Jacoby).
Role of the Sponsor: The funders of the MORE trial had no role in the design or conduct of this study; collection, management, analysis, or interpretation of the data; or preparation or revision of this manuscript, other than to approve the manuscript without comment or request for emendation.
Previous Presentation: Results of this study were presented at the Society for General Internal Medicine national scientific meeting; April 27, 2007; Toronto, Ontario, Canada.