Vasomotor symptoms, or hot flushes and night sweats, are the cardinal symptoms of menopause and are experienced by most midlife women.1 Recently, links between vasomotor symptoms and cardiovascular risk have been of interest. Although findings are not entirely consistent,2 there is indication from the Women's Health Initiative3 and the Heart and Estrogen/Progestin Replacement Study4 that the increased cardiovascular disease risk observed with hormone therapy use was particularly elevated among women with moderate to severe vasomotor symptoms at baseline and, in the Women's Health Initiative, the older women with vasomotor symptoms. We and others subsequently found vasomotor symptoms to be associated with increased subclinical cardiovascular disease indices, including poorer endothelial function,5,6 higher aortic calcification,5 and higher intima media thickness.7
Although there is interest in understanding mechanisms that may underlie any links between vasomotor symptoms and cardiovascular risk, the current understanding of the physiology of vasomotor symptoms is incomplete. It is notable that women with vasomotor symptoms tend to have adverse cardiovascular risk factors such as obesity and smoking and adverse psychosocial factors such as low socioeconomic position, depression, and anxiety.1 However, associations between vasomotor symptoms and cardiovascular risk persist controlling for these factors.3–5,7
One potential link between vasomotor symptoms and cardiovascular risk is lipid abnormalities. Findings from two cross-sectional, population-based studies indicate possible associations with vasomotor symptoms and an adverse lipid profile.8,9 However, findings of links between vasomotor symptoms and lipids have not been universal, with several smaller studies showing null10 or mixed findings.11 Thus, whether vasomotor symptoms are related to lipids remains unclear.
The Study of Women's Health Across the Nation provides a unique opportunity to estimate the relation between vasomotor symptoms and lipids. It allows examination of a range of lipid markers while controlling for relevant confounders in a community-based, multiethnic sample of women followed over time. We hypothesized that vasomotor symptoms, particularly hot flushes, would be associated with an adverse lipid profile. We examined this question among Study of Women's Health Across the Nation participants, who were assessed approximately annually for vasomotor symptoms and low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL) cholesterol, apolipoprotein A-1, apolipoprotein B, triglycerides, and lipoprotein(a) over 8 years. We controlled for relevant confounders estradiol (E2), and, in secondary models, follicle-stimulating hormone (FSH). Finally, we examined interactions between vasomotor symptoms and menopausal stage, race/ethnicity, and body mass index (BMI, calculated as weight (kg)/[height (m)]2), characteristics that may act as modifiers of these relations.7
MATERIALS AND METHODS
The Study of Women's Health Across the Nation is a multiethnic cohort study designed to characterize biological and psychosocial changes over the menopausal transition. Details of the study's design and procedures are reported elsewhere.12 Briefly, each study site recruited non-Hispanic white women and women belonging to a predetermined racial or ethnic minority group (African American women in Pittsburgh, Boston, Michigan, Chicago; Japanese in Los Angeles; Hispanic in New Jersey; Chinese in Oakland area of California). Los Angeles, Pittsburgh, and New Jersey sites used random-digit-dialed sampling from banks of telephone numbers, and Boston, Chicago, Michigan, and Oakland sites selected randomly from lists of names or household addresses. Select sites supplemented primary sampling frames to obtain adequate numbers of racial and ethnic minority women. Study protocols were approved by the institutional review boards at each site. Each participant provided written informed consent.
Baseline eligibility criteria for the Study of Women's Health Across the Nation included being aged 42–52 years, having an intact uterus and at least one ovary, not being pregnant or lactating, not using oral contraceptives or hormone therapy, and having at least one menstrual cycle in the 3 months before the interview. Seventy-three percent of the women selected provided information to determine eligibility; 51% (n=3,302) of eligible women enrolled. Annual clinic assessments began in 1996–1997. This investigation was a longitudinal analysis of associations between vasomotor symptoms and lipids from baseline through the seventh study visit.
Of the 3,302 women enrolled in the Study of Women's Health Across the Nation, 92 women were excluded as a result of having had a reported stroke, angina, or myocardial infarction at baseline, and nine women were excluded as a result of having no data for vasomotor symptoms or lipids. Data were censored during the follow-up period at the time of hysterectomy or oophorectomy (n=221), and data from visits in which pregnancy or hormone use (hormone therapy, oral contraceptives) within the previous year was reported were excluded. The 101 women excluded from this analysis had at baseline a higher BMI, more hot flushes and night sweats, higher anxiety and depressive symptoms, lower HDL cholesterol, higher apolipoprotein B, triglycerides, and lipoprotein(a), and were less educated than women included (P<.05). Primary models included 3,201 women.
Vasomotor symptoms were assessed using a questionnaire at each study visit. Women responded to two questions, which asked separately how often they experienced 1) hot flushes and 2) night sweats in the past 2 weeks (not at all, 1–5 days, 6–8 days, 9–13 days, everyday; categorized as none, 1–5, 6 or more days for analysis). Hot flushes and night sweats were considered separately as a result of their differential pattern of associations with lipids in this analysis.
Phlebotomy was performed in the morning after an overnight fast. Participants were scheduled for venipuncture on days 2–5 of a spontaneous menstrual cycle. Two attempts were made to obtain a day 2–5 sample. If a timed sample could not be obtained (as menstrual cycles became less regular, samples tied to the early follicular phase were less feasible), a random fasting sample was taken within 90 days of the annual visit. Blood was maintained up to 1 hour at 4°C until separated, frozen (−80°C), and sent on dry ice to the Clinical Laboratory Improvement Amendments-certified CLASS laboratory at the University of Michigan (Ann Arbor MI; E2) and Medical Research Laboratories (Highland Heights, Kentucky; lipids, lipoproteins, and apolipoproteins) for analysis. For budgetary reasons, lipid marker assays were completed at study baseline and study years 1, 3, 4, 5, 6, and 7.
Lipid fractions were determined from ethylenediamine tetraacetic acid-treated plasma.13,14 Total cholesterol and triglycerides were determined by enzymatic methods. High-density lipoprotein cholesterol was isolated with heparin-2M manganese chloride.13,14 Low-density lipoprotein cholesterol was calculated by the Friedewald equation.15 Apolipoprotein A-1 and apolipoprotein B were measured by immunonephelometry calibrated with a World Health Organization traceable standard.16 Triglycerides and lipoprotein(a) were natural log transformed for analysis.
Estradiol assays were performed on the ACS-180 automated analyzer using a double-antibody chemiluminescent immunoassay with a solid-phase anti-Ig immunoglobulin conjugated to paramagnetic particles, antiligand antibody, and competitive ligand labeled with dimethylacridinium ester. The E2 assay modifies the rabbit anti-E2–6 ACS-180 immunoassay to increase sensitivity with a lower limit of detection=6.6 pg/mL and inter- and intra-assay coefficients of variation of 10.6% and 6.4%, respectively.17 Duplicate E2 assays were conducted and results reported as the arithmetic mean. Follicle-stimulating hormone assays were performed using a two-site chemiluminometric immunoassay, with inter-and intra-assay coefficients of variation of 11.4% and 3.8%, respectively, and lower limit of detection=1.1 milli-international units/mL. Estradiol and FSH values were natural log transformed for analysis.
Race and ethnicity (determined in response to: “How would you describe your primary racial or ethnic group?”) and education (less than compared with college or higher) were obtained in the study screening interview. Age, smoking status (current compared with past or never), depressive or anxious symptoms, physical activity, alcohol use, menopausal status, and medication use and health conditions were derived from standard questionnaires and interviews administered during annual visits. Age, smoking, alcohol use, physical activity, BMI, menopausal status, and medication use and health conditions were considered as time-varying covariates. Physical activity was assessed at baseline and visits 3, 5, and 6 through a modified Kaiser Permanente Health Plan Activity Survey18 with values for visits 1, 4, and 7 carried forward from the last completed observation. Alcohol use was reported average weekly number of servings of beer, wine, liquor, or mixed drinks. Menopausal status was obtained from self-reported bleeding patterns over the year preceding the visit, categorized as premenopausal (bleeding in the previous 3 months with no change in cycle predictability in the past year), early perimenopausal (bleeding in the previous 3 months with decrease in cycle predictability in the past year), late perimenopausal (less than 12 and more than 3 months of amenorrhea), or postmenopausal (12 months or more of amenorrhea) at each visit. Consistent with our prior work,5,19 baseline depressive symptoms (assessed through the Center for Epidemiologic Studies Depression scale)20 and anxiety symptoms (sum score of number of days in the past 2 weeks, 0=no days to 4=everyday, reporting irritability or grouchiness, feeling tense or nervous, heart pounding or racing, or feeling fearful for no reason) were considered as covariates as a result of their associations with vasomotor symptoms1 and cardiovascular risk.21,22 Body mass index was derived from annual physical measures. Cardiovascular disease status (reported hypertension, angina, stroke, or heart attack) and use of cardiovascular disease medications (reported use of medication for a heart condition, an anticoagulant, or for blood pressure-lowering in the past year) were combined into a single variable and covaried. Hypertension was included with other cardiovascular disease variables to avoid model overfitting given the large number of covariates in models. Reported lipid-lowering medication use was covaried separately.
Baseline differences between included and excluded participants were tested using Wilcoxon rank-sum test for continuous variables and χ2 tests for categorical variables. Univariable analyses of relations between covariates and each outcome were performed at baseline using linear regression. Longitudinal associations between hot flushes and night sweats and each lipid marker were evaluated using linear mixed effects models to handle within-participant correlations and unequal measurement intervals or durations (eg, missing visits).23 Repeated measures of each outcome were modeled as a function of repeated measures of hot flushes or night sweats (considered in separate models) with a random intercept to account for random effects associated with the intercept for each participant. An autoregressive covariance structure, which assumes that a woman's observations more proximal in time will be more highly correlated than those more distal in time, was selected as a result of the longitudinal nature of the data and based on the Akaike information criterion (lower indicating better model fit).24 Models were adjusted for age and site and next additionally for covariates selected based on previously documented associations with lipids and present associations with outcomes at P<.05. Particular attention was paid to include factors that may act as confounders in these analyses (associated with both vasomotor symptoms and lipids; eg, education, smoking, race and ethnicity, menopausal stage, BMI).1,25 Serum E2 was added to covariate-adjusted models along with blood draw timing (in compared with out of cycle day 2–5 window). Follicle-stimulating hormone was also considered instead of E2 in secondary models. Because cycle day of blood draw and menopausal status were collinear (only early perimenopausal and premenopausal women had menstrual cycles to provide timed sample), they were considered as a composite variable (premenopausal or perimenopausal timed sample, premenopausal or perimenopausal untimed sample, late perimenopausal, postmenopausal). For time-varying covariates, values concurrent with the outcome measure time point were used. Linear trends were tested by treating hot flushes or night sweats as a continuous variable.26 Interactions between hot flushes and night sweats and race and ethnicity, menopausal status, and BMI were examined as cross-product terms in univariable and multivariable models. For presentation, interactions for BMI are displayed as predicted means for vasomotor symptoms level by obesity status (BMI less than 25, BMI 25–30, BMI higher than 30). Residual analysis and diagnostic plots (eg, scatterplots, histograms) were conducted to verify model assumptions of normality. Analyses were performed with SAS 9.2. Models were two-sided (α=0.05).
The sample at baseline was, on average, 46 years old, overweight, and nonsmoking (Table 1). At baseline, approximately one third of the women in the sample had vasomotor symptoms (hot flushes: 26%, night sweats: 29%), which increased over the study visits (eg, at visit 7, hot flushes: 54%; night sweats: 38%). Consistent with other Study of Women's Health Across the Nation reports,25,27 factors significantly (P<.05) associated with a more adverse lipid profile at baseline included older age, Hispanic race or ethnicity (relative to non-Hispanic white), smoking, low education, not drinking alcohol, being early perimenopausal (compared with premenopausal), and having a high BMI, low physical activity, low E2, high FSH, and high anxious or depressive symptoms. Japanese and Chinese women had a more favorable lipid profile than non-Hispanic white women.
In age- and site-adjusted models, both hot flushes and night sweats, particularly when experienced frequently (6 days or more, past 2 weeks), were associated with higher LDL cholesterol, HDL cholesterol, apolipoprotein B, apolipoprotein A-1, triglycerides, and to a lesser extent higher lipoprotein(a) over the 8-year study period (Table 2). For example, the age- and site-adjusted estimated means (in mg/dL) from the mixed models for hot flushes were 117.7 (none), 119.6 (1–5 days), and 120.5 (6 or more days) for LDL cholesterol; 57.1 (no days), 57.5 (1–5 days), and 57.9 (6 or more days) for HDL cholesterol; and 121.0 (none), 125.2 (1–5 days), and 131.3 (6 or more days) for triglycerides. In multivariable models, associations remained for all lipid markers except lipoprotein(a) (Table 3). Associations remained with further adjustment for E2 (Table 4).
Additional analyses were conducted. First, given prior evidence of effect modification by BMI,7 we examined interactions between vasomotor symptoms and BMI for each outcome. Significant interactions between BMI and hot flushes for LDL cholesterol, apolipoprotein B, lipoprotein(a), and triglycerides (P<.05; Fig. 1), and between BMI and night sweats for LDL and apolipoprotein B (P<.05) in multivariable models. Associations were strongest among leaner women (Fig. 1). There were no interactions by menopausal stage nor by race or ethnicity with the exception of one interaction (P<.05) between hot flushes and race or ethnicity in relation to apolipoprotein B (with associations strongest among white and Japanese women; data not shown), and thereby associations were broadly similar across racial and ethnic groups. Finally, FSH was substituted for E2 as a covariate in multivariable models. Findings were similar to models with E2 with significant associations between vasomotor symptoms for HDL cholesterol (relative to no hot flushes, 1–5 days: β [standard error]=0.24 [0.18], P=.18; 6 or more days: β [standard error]=0.59 [0.24], P=.02), apolipoprotein B (relative to no hot flushes, 1–5 days: β [standard error]=0.96 [0.41], P=.02; 6 or more days: β [standard error]=1.71 [0.54], P=.002), apolipoprotein A-1 (relative to no hot flushes, 1–5 days: β [standard error]=0.87 [0.47], P=.07; 6 or more days: β [standard error]=1.66 [0.63], P=.008), and triglycerides (relative to no hot flushes, 1–5 days: percent change [95% confidence interval]=2.90 [1.39–4.42], P<.001; 6 or more days: percent change [95% confidence interval]=5.70 [3.65–7.79], P<.001), with the exception LDL cholesterol (relative to no hot flushes, 1–5 days: β [standard error]=0.82 [0.46], P<.10; 6 or more days: β [standard error] 0.77 [0.62], P=.21), for which associations were somewhat reduced. Findings for night sweats were similar.
In this large, community-based, multiethnic study, vasomotor symptoms, particularly frequent vasomotor symptoms, were associated with elevated LDL cholesterol, HDL cholesterol, apolipoprotein B, apolipoprotein A-1, and triglycerides over an 8-year period encompassing the menopausal transition. Associations between vasomotor symptoms and lipid markers persisted controlling confounders and were not explained by differences in E2 or, in most cases, FSH. Associations were broadly similar across ethnic groups and were most pronounced among leaner women. Thus, these data provide the most definitive data to date linking vasomotor symptoms to elevated lipids among women transitioning through menopause.
Our results indicate that vasomotor symptoms are associated with higher triglycerides, LDL cholesterol, and apolipoprotein B. These associations are broadly consistent with other large community-based cross-sectional studies.9 These markers are well known to be associated with elevated cardiovascular risk.28 Thus, positive relations between vasomotor symptoms and triglycerides, LDL cholesterol, and apolipoprotein B are consistent with the findings of elevated cardiovascular risk among women with frequent or severe vasomotor symptoms and may be one mechanism linking these symptoms to cardiovascular risk.
However, we unexpectedly observed significant positive associations between vasomotor symptoms and HDL cholesterol and its protein component apolipoprotein A-1, generally considered cardioprotective.28 The reason for this pattern is not immediately clear, and a chance finding cannot be ruled out. However, it is notable that data from the Study of Women's Health Across the Nation29 and other studies30 indicate that HDL may not universally be cardioprotective, potentially varying by menopausal stage. Whereas higher HDL cholesterol is associated with lower subclinical cardiovascular disease during premenopause and early perimenopause, it may become proatherogenic during late perimenopause and early postmenopause, when vasomotor symptoms peak.29 These changes may be attributable in part to HDL particle size reductions that may occur as women transition through menopause29 and evidence that larger but not smaller HDL particles are associated with cardioprotection.31 Thus, although speculative, changes in the atherogenic potential of HDL as women transition through menopause and symptoms peak may play a role in the positive associations between HDL cholesterol and apolipoprotein A-1 and vasomotor symptoms observed here.
Interactions were observed by BMI with findings most pronounced among the leaner women. Obesity is generally strongly associated with an adverse lipid profile. Thus, it is possible that the effect of BMI on lipids or vasomotor symptoms among heavier women overwhelmed the more modest associations between vasomotor symptoms and lipid markers observed here. Notably, the association between menopausal status and lipids in the Study of Women's Health Across the Nation was also stronger among leaner women, potentially for similar reasons as observed here.25,27
Our findings of increased LDL cholesterol, apolipoprotein B, and triglycerides are broadly consistent with the two other population-based studies. In one large cross-sectional study, vasomotor symptoms were associated with elevated total cholesterol.8 In a second cross-sectional analysis, “sweats” (although not hot flushes) were associated with elevated total cholesterol, LDL cholesterol, and triglycerides.9 Smaller clinical studies tend to show more mixed results. A small study of 32 healthy women found that hot flushes were associated with lower HDL cholesterol and apolipoprotein A-1 but not LDL cholesterol or apolipoprotein B.11 Another study (n=150) of very healthy women found no associations between hot flushes and lipids,10 although very few women (n=23) without vasomotor symptoms were assessed. The reasons for differences between studies are not immediately apparent but may be the result of differences in sample size and selection of women to be low on cardiovascular risk factors yet highly symptomatic. None of these studies examined differences by obesity status.
The physiology linking vasomotor symptoms to lipids is not entirely clear. Women with cardiovascular risk factors such as obesity and smoking tend to have more vasomotor symptoms.1 However, controlling for these factors did not eliminate the observed associations. The menopausal transition, including the changing hormonal milieu, is clearly associated with elevated vasomotor symptoms as well as changes in lipids, particularly LDL cholesterol.27 However, controlling for menopausal stage and E2 did not sizably reduce associations, and controlling for FSH reduced associations only between vasomotor symptoms and LDL cholesterol. Thus, these hormones explained only a small part of the associations observed here. Several other systems may be relevant. There is indication of poorer endothelial function5,6 and a more adverse hemostatic profile19 among women with vasomotor symptoms. Furthermore, we and others found changes in autonomic nervous system balance favoring increased sympathetic and decreased parasympathetic tone with vasomotor symptoms,32 a profile associated with cardiovascular risk. Finally, others have proposed changes in hypothalamic–pituitary–adrenal axis function with vasomotor symptoms.33 Thus, many mechanisms may link vasomotor symptoms to altered lipid profiles.
In this study, although associations were observed for both infrequent (1–5 days in the past 2 weeks) and frequent (6 or more days in the past 2 weeks) vasomotor symptoms, associations were most pronounced for frequent vasomotor symptoms. Notably, some (but not all)5 prior work has shown that it is the more frequent7 or moderate to severe3,4 vasomotor symptoms that are associated with elevated cardiovascular risk, suggesting that higher levels of symptomatology are most relevant. Moreover, although associations were apparent for hot flushes and night sweats, associations for hot flushes were somewhat more robust to adjustment. We have previously found stronger associations with indices of cardiovascular risk for hot flushes compared with night sweats,5,7,19 potentially as a result of a differing underlying physiology between the two or to a lower reliability of reporting night sweats given their occurrence during sleep.
The study results should be interpreted in light of several limitations. First, as this was a large, epidemiologic investigation, the measures of vasomotor symptoms were relatively crude, asking women to recall the frequency of symptoms over 2 weeks. These measures are clearly querying about menopausal vasomotor symptoms, and thereby are an improvement over other epidemiologic work in this area.9 However, there was likely some error introduced by memory in these reports, and these measures did not allow us to examine measures such as a more detailed symptom frequency or severity. Moreover, hormonal measures were assessed yearly, and some women likely had widely fluctuating hormone concentrations. Thus, exposure to these hormones may not have been fully quantified, and further investigation of them in these associations is merited. Furthermore, although a range of medical conditions and medications were controlled, some women may have had other medical conditions or medication use that could have affected the markers assessed here. It is also important to note that the magnitude of the differences in lipids by level of symptoms, although statistically significant, was not large. Finally, although this was a longitudinal investigation, the directionality or causal nature of these associations cannot be inferred.
This study had several strengths. This study is the first to examine longitudinal associations between vasomotor symptoms and lipid markers. It investigated a large, longitudinal, community-based investigation of women transitioning through menopause. It included 8 years of follow-up, a design that stands in contrast to the exclusively cross-sectional research on this topic. It examined associations between vasomotor symptoms and lipids in a sample that included women from several different ethnic groups. Thus, it provides the richest analysis on this topic to date.
In conclusion, vasomotor symptoms were associated with elevated LDL cholesterol, HDL cholesterol, apolipoprotein B, apolipoprotein A-1, and triglycerides in this community-based sample of women transitioning through menopause. These associations could not be fully explained by shared risk factors, E2, or FSH as assessed here. They were strongest among lean women. These findings may contribute to ongoing efforts to better understand the physiology of vasomotor symptoms as well as any implications vasomotor symptoms may have for cardiovascular health.
1. Gold E, Colvin A, Avis N, Bromberger J, Greendale G, Powell L, et al.. Longitudinal analysis of vasomotor symptoms and race/ethnicity across the menopausal transition: Study of Women's Health Across the Nation (SWAN). Am J Public Health 2006;96:1226–35.
2. Szmuilowicz ED, Manson JE, Rossouw JE, Howard BV, Margolis KL, Greep NC, et al.. Vasomotor symptoms and cardiovascular events in postmenopausal women. Menopause 2011;18:603–10.
3. Rossouw JE, Prentice RL, Manson JE, Wu L, Barad D, Barnabei VM, et al.. Postmenopausal hormone therapy and risk of cardiovascular disease by age and years since menopause. JAMA 2007;297:1465–77.
4. Huang AJ, Sawaya GF, Vittinghoff E, Lin F, Grady D. Hot flushes, coronary heart disease, and hormone therapy in postmenopausal women. Menopause 2009;16:639–43.
5. Thurston RC, Sutton-Tyrrell K, Everson-Rose SA, Hess R, Matthews KA. Hot flashes and subclinical cardiovascular disease: findings from the Study of Women's Health Across the Nation Heart Study. Circulation 2008;118:1234–40.
6. Bechlioulis A, Kalantaridou SN, Naka KK, Chatzikyriakidou A, Calis KA, Makrigiannakis A, et al.. Endothelial function, but not carotid intima-media thickness, is affected early in menopause and is associated with severity of hot flushes. J Clin Endocrinol Metab 2010;95:1199–206.
7. Thurston RC, Sutton-Tyrrell K, Everson-Rose S, Hess R, Powell L, Matthews K. Hot flashes and carotid intima media thickness among midlife women. Menopause 2011;18:352–8.
8. Gast GC, Grobbee DE, Pop VJ, Keyzer JJ, Wijnands-van Gent CJ, Samsioe GN, et al.. Menopausal complaints are associated with cardiovascular risk factors. Hypertension 2008;51:1492–8.
9. Gast GC, Samsioe GN, Grobbee DE, Nilsson PM, van der Schouw YT. Vasomotor symptoms, estradiol levels and cardiovascular risk profile in women. Maturitas 2010;66:285–90.
10. Tuomikoski P, Mikkola TS, Hamalainen E, Tikkanen MJ, Turpeinen U, Ylikorkala O. Biochemical markers for cardiovascular disease in recently postmenopausal women with or without hot flashes. Menopause 2010;17:145–51.
11. Sassarini J, Fox H, Ferrell W, Sattar N, Lumsden MA. Vascular function and cardiovascular risk factors in women with severe flushing. Clin Endocrinol (Oxf) 2011;74:97–103.
12. Sowers M, Crawford S, Sternfeld B, Morganstein D, Gold EB, Greendale GA, et al.. SWAN: a multicenter, multiethnic, community-based cohort study of women and the menopausal transition. In: Lobo RA, Kelsey J, Marcus R, Lobo AR, editors. Menopause: biology and pathology. New York (NY): Academic Press; 2000. p. 175–88.
13. Steiner P, Friedel J, Bremner W, Stein E. Standardization of micro-methods for plasma cholesterol, triglyceride and HDL-cholesterol with the Lipid Research Clinics' methodology. J Clin Chem Clin Biochem 1981;19:850.
14. Warnick G, Albers J. A comprehensive evaluation of the heparin-manganese precipitation procedure for estimating high density lipoprotein cholesterol. J Lipid Res 1978;19:65–76.
15. Friedewald W, Levy R, Fredrickson D. Estimation of the concentration of low density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem 1972;18:499–502.
16. Myers G, Cooper G, Winn C, Smith S. The Centers for Disease Control–National Heart, Lung and Blood Institute Lipid Standardization Program: an approach to accurate and precise lipid measurements. Clin Lab Med 1989;9:105–35.
17. England BG, Parsons GH, Possley RM, McConnell DS, Midgley AR. Ultrasensitive semiautomated chemiluminescent immunoassay for estradiol. Clin Chem 2002;48:1584–6.
18. Ainsworth BE, Sternfeld B, Richardson MT, Jackson K. Evaluation of the kaiser physical activity survey in women. Med Sci Sports Exerc 2000;32:1327–38.
19. Thurston R, El Khoudary S, Sutton-Tyrrell K, Crandall C, Gold E, Sternfeld B, et al.. Are vasomotor symptoms associated with alterations in hemostatic and inflammatory markers? Findings from the Study of Women's Health Across the Nation. Menopause 2011;18:1044–51.
20. Radloff LS. The CES-D scale: a self-report depression scale for research in the general population. Applied Psychological Measurement 1977;1:385–401.
21. Kubzansky LD, Kawachi I, Weiss ST, Sparrow D. Anxiety and coronary heart disease: a synthesis of epidemiological, psychological, and experimental evidence. Ann Behav Med 1998;20:47–58.
22. Lett HS, Blumenthal JA, Babyak MA, Sherwood A, Strauman T, Robins C, et al.. Depression as a risk factor for coronary artery disease: evidence, mechanisms, and treatment. Psychosom Med 2004;66:305–15.
23. Fitzmaurice G, Laird N, Ware J Applied longitudinal analysis. Hoboken (NJ): Wiley-Interscience; 2004.
24. Akaike H. A new look at the statistical model identification. IEEE Trans on Automat Contr 1974;19:716–23.
25. Derby CA, Crawford SL, Pasternak RC, Sowers M, Sternfeld B, Matthews KA. Lipid changes during the menopause transition in relation to age and weight: the Study of Women's Health Across the Nation. Am J Epidemiol 2009;169:1352–61.
26. Woodward M. Epidemiology: study design and data analysis New York (NY): Chapman and Hall/CRC; 1999.
27. Matthews KA, Crawford SL, Chae CU, Everson-Rose SA, Sowers MF, Sternfeld B, et al.. Are changes in cardiovascular disease risk factors in midlife women due to chronological aging or to the menopausal transition? J Am Coll Cardiol 2009;54:2366–73.
28. Mosca L, Benjamin EJ, Berra K, Bezanson JL, Dolor RJ, Lloyd-Jones DM, et al.. Effectiveness-based guidelines for the prevention of cardiovascular disease in women—2011 update: a guideline from the American Heart Association. Circulation 2011;123:1243–62.
29. Woodard GA, Brooks MM, Barinas-Mitchell E, Mackey RH, Matthews KA, Sutton-Tyrrell K. Lipids, menopause, and early atherosclerosis in Study of Women's Health Across the Nation Heart women. Menopause 2011;18:376–84.
30. Fan A, Dwyer J. Sex differences in the relation of HDL cholesterol to progression of carotid intima-media thickness: the Los Angeles Atherosclerosis Study. Atherosclerosis 2007;195:e191–6.
31. Asztalos BF, Collins D, Cupples LA, Demissie S, Horvath KV, Bloomfield HE, et al.. Value of high-density lipoprotein (HDL) subpopulations in predicting recurrent cardiovascular events in the Veterans Affairs HDL Intervention Trial. Arterioscler Thromb Vasc Biol 2005;25:2185–91.
32. Thurston R, Christie I, Matthews K. Hot flashes and cardiac vagal control: A link to cardiovascular risk? Menopause 2010;17:456–61.
33. Woods NF, Mitchell ES, Smith-Dijulio K. Cortisol levels during the menopausal transition and early postmenopause: observations from the Seattle Midlife Women's Health Study. Menopause 2009;16:708–18.