Abstract
Objective:
Previous studies have assessed potential risk factors for vasomotor symptoms (VMS) beginning in midlife. We examined whether early adulthood risk factors predict VMS trajectories over time.
Methods:
We performed a secondary data analysis of the Coronary Artery Risk Development in Young Adults (CARDIA) study, a population-based cohort. We included women who answered questions about VMS at ≥3 exams (n=1966). We examined whether risk factors at baseline (when participants were ages 18–30, average age 25 years) and the Year 15 (Y15) exam (at ages 33–45; average age 40 years) were associated with VMS trajectories from Y15 through Y35. Logistic regression models were used to evaluate the associations with VMS trajectories.
Results:
We identified 3 trajectories of VMS presence: minimal (40%), increasing over time (27%), and persistent (33%). Baseline factors associated with persistent VMS over time included Black race, less than a high school education, depressive symptoms, migraines, cigarette use, and at Y15, hysterectomy. Baseline factors associated with increasing VMS over time included Black race and lower body mass index. Risk factors for bothersome VMS were similar and also included thyroid disease, although thyroid disease was not associated with persistence of VMS over time. Associations were similar among women who had not undergone hysterectomy and in Black and White women.
Conclusions:
Risk factors for VMS may be identified in early adulthood. Further examination of risk factors such as migraines and depressive symptoms in early adulthood may be helpful in identifying therapies for VMS.
Keywords: vasomotor symptoms, hot flashes, night sweats, menopause
Introduction
Vasomotor symptoms (VMS), consisting of hot flashes and night sweats, affect almost 80% of midlife women.1 Approximately one-half have moderate to severe symptoms,2 which are linked with decreased quality of life and work productivity.3 Risk factors for VMS are likely present years before menstruation ceases: genetic variants linked with reproductive aging are also associated with VMS,4 and studies of midlife populations have reported that VMS are often present over a decade prior to the cessation of menses.5 However, the timing of VMS risk factors in relation to symptoms is not well-understood, since studies of VMS typically begin in midlife. Because VMS may begin earlier in reproductive life, while women are still menstruating regularly, assessment of potential VMS risk factors would ideally begin earlier in adulthood. Such an analysis could inform interventions to prevent or reduce VMS, which may be undertreated6 due to concerns regarding patient-perceived risks of existing therapies.7
The Coronary Artery Risk Development in Young Adults (CARDIA) study is a population-based cohort study originally designed to identify risk factors for cardiovascular disease beginning when participants were aged 18–30 years of age until the most recent exam when women were aged approximately 60 years.8 CARDIA also has collected VMS characteristics beginning at approximately 40 years of age and every 5 years thereafter, as well as information on gynecologic surgeries, menstrual cycle length, hormone use, and reproductive histories.9 We have several a priori hypotheses based upon previous reports of risk factors for VMS collected in later life and at a single point in time. We hypothesized that risk factors associated with persistent VMS would differ from those associated with women who reported increasing VMS over time vs. women who have minimal VMS in midlife. We also hypothesized that risk factors for VMS were present at baseline but that these would no longer be significant after adjustment for risk factors in later life. Finally, we also hypothesized that the strongest risk factors for persistent VMS would be adverse social determinants of health, tobacco use, and obesity. We also examined associations between persistent VMS with migrainous disorders10 as well as depressive symptoms.11
Methods
Study population
The Coronary Artery Risk Development in Young Adults (CARDIA) Study is an ongoing multicenter longitudinal study conducted at four US communities (Birmingham, Alabama; Chicago, Illinois; Minneapolis, Minnesota; and Oakland, California) to study cardiovascular disease risk trends and clinical sequelae from young adulthood. At baseline or Year 0 (Y0, 1985–1986), 5,115 healthy adults were recruited from the general population to be balanced on sex, race (White or Black), age (18–24 or 25–30 y) and education (high school or less, or more than high school). Data collection and follow-up protocols were approved by the Institutional Review Boards of each field center with all participants providing written informed consent. Details of the study design, recruitment, methodology, and baseline characteristics are described else-where.8 Since then, nine follow-up examinations have occurred. Of the original cohort of 2787 women, the number of women who attended at least 3 exams between Y15 and Y35 was 1966 (71%). The percent of women who attended at each subsequent exam year was: Y2 (91%), Y5 (86%), Y7 (81%), Y10 (79%), Y15 (74%), Y20 (72%), Y25 (71%), Y30 (69%), Y35 (46%); (Y35 examination was concurrent with the COVID-19 pandemic). For the purposes of this analysis, we included any woman who responded to at least three assessments regarding the presence of VMS (n=1966 women) between and including Y15, when questions regarding VMS were first asked, and Y35, the last CARDIA examination conducted. Since the prevalence of hormone use was low at Y15, because hormones are used more frequently among women who undergo gynecologic surgery, and also because hormones may not completely resolve VMS, these women were included in the analysis. Compared to women without these characteristics, women who were excluded were slightly younger (24 years vs. 25 years), more likely to be Black (39% vs. 49%), and more likely to have disadvantageous social determinants of health and health behaviors (less than a high school education, more difficulty paying for basic necessities, higher depressive symptom score, current cigarette use, higher BMI, and higher systolic blood pressure) although prevalence of oral contraceptive pill use, hormone use at Y15, perimenopausal symptoms, and any gynecologic surgery was lower among excluded women.
Measures
At every CARDIA examination, standardized protocols were used to collect information on demographics, medical history, lifestyle, behavioral factors, anthropometrics, and cardiovascular risk factors. No single tool is currently used to validate whether VMS are due to estrogen withdrawal as opposed to systemic illness or medications.12, 13 Previous cohort studies indicate that hot flashes and night sweats are highly correlated, and the presence of either of these has been termed “vasomotor symptoms.”14 Beginning at the Y15 examination, CARDIA women were asked, “Have you experienced hot flashes or night sweats in the past 3 months?” Gynecologic surgery and use of oral contraceptives or exogenous estrogen therapy were also assessed. Women reporting no menstrual cycles within the previous 12 months in the absence of gynecologic surgery were defined as having natural menopause,15 and women who reported cessation of menstrual bleeding preceded by hysterectomy or bilateral oophorectomy were defined as having surgical menopause. Women were also asked how their periods had changed over the past year; women who responded that their periods had become farther apart, closer together, or occurred more variably were classified as having perimenopause, which we have previously reported corresponds with cessation of menopause at the next exam.16
At baseline (Y0) and every exam thereafter, women were asked about their medical history, including whether they had thyroid disease and whether a doctor or nurse had told them they had a nervous, emotional, or mental disorder. Beginning at the Y5 exam, depressive symptoms were assessed by the Center for Epidemiologic Studies Depression scale (CES-D).17 The CES-D is a 20-item, self-report, global measure of depressive symptoms over the past week. Each symptom is assessed on a scale from 0 (never) to 3 (nearly every day), with total scores ranging from 0 (none/low depressive symptoms) to 60 (high depressive symptoms). Scores ≥16 points are considered to be consistent with dysthymia.17 Beginning at the Y7 exam, women were asked whether a doctor or nurse had told them they had migraine headaches (aura was not assessed). Cigarette smoking was assessed by means of an interviewer-administered tobacco questionnaire and classified as current, former or never. Body mass index (BMI) was calculated by dividing measured weight in kilograms by height in meters squared. Blood pressure was measured with participants seated and after 5 minutes of rest using a random-zero sphygmomanometer. The average of the second and third consecutive measurements was used for analysis.
First, to characterize the patterns of missings and presence or absence of VMS at each exam, we examined the patterns of VMS across CARDIA exams using heat maps, which were constructed based on the responses given by participants to questions regarding the occurrence of VMS, spanning a period from Y15 to Y35 (Figure 1). Missing responses were coded as 0.5, ‘no’ responses as 1, and ‘yes’ responses as 2 to facilitate the creation of the heatmap. A dendrogram of individuals was then created using hierarchical clustering, an agglomerative (bottom-up) unsupervised learning algorithm that groups similar observations together based on specific distance metrics.18 Initially, each individual response was treated as its own cluster. These were then progressively merged into larger clusters, until all responses were contained within a single comprehensive cluster.
FIGURE 1.
Heat maps for vasomotor symptoms (VMS) among CARDIA women showing the presence of VMS at each exam among the 1966 women who responded to this question during at least 3 exams. Black indicates VMS present; Gray indicates VMS not present; White indicates no response at that exam year. All of the women had at least 3 responses consisting of “VMS present” or “VMS absent” but may have had missing responses at other exam years.
Second, we used latent class models (fitted by SAS Proc Traj)19, 20 to model VMS trajectory, using chronologic age as the x-axis. These analyses included only visits with an observed response for VMS (yes or no) and did not include missing values. We analyzed trajectory models with 2 through 10 groups, and model fit was assessed using the Bayesian Information Criterion (BIC) and the Akaike Information Criterion (AIC).21 The optimal number of trajectory groups is determined when the BIC and AIC were maximized or when adding more groups did not substantially differentiate between trajectory patterns. We first ran each model with a cubic function; however, the VMS trajectories did not reach a global maximum until we assigned lower polynomial function (quadratic). Next, because trajectory models often find only the local maximum (as opposed to the global maximum) when using default start values, we ran each model using the recommended start parameters and polynomial function for each trajectory group to achieve a model that reached a global maximum (i.e., best-fit polynomial function for each trajectory group within the trajectory model). The posterior predictive probability of group membership was calculated for each model, and participants were assigned to the trajectory group for which they had the greatest posterior predictive probability. The average posterior predictive probabilities were assessed for adequacy of model fit.
We identified three trajectory groups which were characterized by minimal symptoms over time (trajectory 1), increasing symptoms over time (trajectory 2), and persistent symptoms over time (trajectory 3) (Figure 2). Participant characteristics by VMS trajectory group were defined by means, medians, and proportions as appropriate (Table 1), and differences between participants in each trajectory were tested using t-tests, Wilcoxon tests, and χ2 analyses for continuous and categorical characteristics, respectively. Polytomous logistic regression was conducted to assess the associations between the outcome of VMS trajectories and independent variables grouped into demographics at baseline (Table 2, Model 1), health conditions and behavior at baseline (Table 2, Model 2), reproductive history at baseline (Table 2, Model 3), and all of these factors together (Table 2, Model 4). A parallel set of models was created with characteristics at the Y15 exam (Table 3), and then a final set of models including both baseline and Y15 factors was created. Variables were selected based upon significant associations with VMS category in unadjusted comparisons shown in Table 1. Multivariable models adjusted for age, race, and center.
FIGURE 2.
Probability of vasomotor symptoms (VMS) by chronologic age among women with at least 3 visits (n=1966): 40.2% of women are in trajectory 1 or “low probability of VMS over time,” 26.7% of women are in trajectory 2 or “increasing VMS over time,” and 33% of women are in trajectory 3 or “persistent VMS over time.”
Table 1.
Risk factors for vasomotor symptoms among CARDIA women at baseline (year 0) and Year 15.a
| Presence of vasomotor symptoms (n=1966) | ||||
|---|---|---|---|---|
| Minimal or infrequent VMS (Trajectory 1) |
Increasing VMS (Trajectory 2) |
Persistent VMS (Trajectory 3) |
p-value | |
| N = 839 | N = 525 | N = 602 | ||
| Baseline socioedemographic factors | ||||
| Age (years) | 24.9 (3.7) | 25.7 (3.5) | 24.8 (3.7) | <0.0001 |
| Black (n, %) | 354 (42%) | 243 (46%) | 377 (63%) | <0.0001 |
| Less than high school (n, %) | 246 (29%) | 156 (30%) | 271 (45%) | <0.0001 |
| Income < $50,000/year (n, %) | 576 (69%) | 350 (67%) | 448 (74%) | 0.010 |
| Difficulty paying for basics (n, %) | 270 (32%) | 165 (31%) | 235 (39%) | 0.008 |
| Year 15 socioedemographic factors | ||||
| Less than high school (n, %) | 196 (23%) | 113 (22%) | 200 (33%) | <0.0001 |
| Income < $50,000/year (n, %) | 298 (36%) | 194 (37%) | 273 (45%) | 0.0005 |
| Difficulty paying for basics | 141 (17%) | 90 (17%) | 154 (26%) | <0.0001 |
| Baseline health conditions and health behaviors | ||||
| Hyperthyroidism (n, %) | 7 (0.8%) | 6 (1%) | 5 (0.8%) | 0.82 |
| Hypothyroidism (n, %) | 7 (0.8%) | 6 (1%) | 9 (1.5%) | 0.50 |
| CES-D score (Y5) | 10.6 (8.2) | 10.9 (8.1) | 13.35 (9.1) | <0.0001 |
| History of mental disorders (n, %) | 44 (5%) | 27 (5%) | 25 (4%) | 0.61 |
| Migraine headaches (n, %) | 91 (11%) | 55 (10%) | 93 (15%) | 0.012 |
| Tobacco use (n, %) | 183 (22%) | 131 (25%) | 188 (31%) | 0.0003 |
| Alcohol use (ml/day) | 6.8 (14.9) | 7.1 (13.0) | 8.1 (15.3) | 0.27 |
| Year 15 health conditions and behaviors | ||||
| Hyperthyoidism (n, %) | 16 (2%) | 7 (1%) | 15 (2%) | 0.37 |
| Hypothyroidism (n, %) | 41 (5%) | 21 (4%) | 23 (4%) | 0.56 |
| CES-D score | 8.4 (7.5) | 8.6 (7.7) | 11.5 (9.3) | <0.0001 |
| History of mental disorders (n, %) | 59 (7%) | 37 (7%) | 57 (9%) | 0.18 |
| Migraine headaches (n, %) | 120 (14%) | 72 (14%) | 137 (23%) | <0.0001 |
| Tobacco use (n, %) | 112 (13%) | 78 (15%) | 152 (25%) | <0.0001 |
| Alcohol use (ml/day) | 5.8 (10.9) | 6.6 (12.2) | 8.7 (29.4) | 0.021 |
| Baseline cardiovascular risk factors | ||||
| BMI (kg/m2) | 24.5 (5.8) | 23.5 (4.3) | 25.0 (5.8) | <0.0001 |
| BMI (n, %) | 0.0001 | |||
| <25 kg/m2 | 565 (68%) | 378 (72%) | 370 (62%) | |
| 25–29.9 kg/m2 | 165 (20%) | 97 (19%) | 118 (20%) | |
| ≥30 kg/m2 | 107 (13%) | 48 (9%) | 110 (18%) | |
| Systolic blood pressure (mm Hg) | 106.3 (9.7) | 105.8 (9.0) | 106.78 (10.2) | 0.23 |
| Diastolic blood pressure (mm Hg) | 66.7 (8.6) | 66.8 (8.4) | 66.72 (9.3) | 0.97 |
| Year 15 cardiovascular risk factors | ||||
| BMI (kg/m2) | 29.0 (8.0) | 27.7 (6.7) | 30.2 (7.9) | <0.0001 |
| BMI (n, %) | <0.0001 | |||
| <25 kg/m2 | 284 (37%) | 210 (43%) | 168 (31%) | |
| 25–29.9 kg/m2 | 198 (26.0%) | 138 (28%) | 131 (25%) | |
| ≥30 kg/m2 | 281 (37%) | 141 (29%) | 235 (44%) | |
| Systolic blood pressure (mm Hg) | 109.7 (14.5) | 110.1 (13.9) | 113.8 (16.4) | <0.0001 |
| Diastolic blood pressure (mm Hg) | 71.7 (10.3) | 72.0 (11.0) | 74.1 (12.8) | 0.0003 |
| Baseline reproductive risk factors | ||||
| Number of pregnancies | 2.0 (1.1) | 2.1 (1.3) | 2.2 (1.3) | 0.021 |
| Oral contraceptive pill use (n, %) | 276 (31%) | 186 (34%) | 211 (33%) | 0.50 |
| Preeclampsia (n, %) | 21 (2%) | 19 (3%) | 22 (3%) | 0.35 |
| Hypertension during pregnancy (n, %) | 57 (6%) | 30 (5%) | 52 (8%) | 0.15 |
| Hysterectomy (n, %) | 17 (2%) | 6 (1%) | 27 (5%) | 0.0021 |
| Oophorectomy (n, %) | 22 (3%) | 5 (1%) | 21 (4%) | 0.027 |
| Year 15 reproductive risk factors | ||||
| Number of pregnancies | 3.2 (2.0) | 3.1 (1.6) | 3.3 (1.8) | 0.34 |
| Oral contraceptive pill use (n, %) | 127 (15%) | 65 (12%) | 54 (9%) | 0.0023 |
| Hormone use (n, %) | 36 (4%) | 7 (1%) | 31 (5%) | 0.0021 |
| Perimenopausal (n, %) | 205 (24%) | 140 (27%) | 185 (31%) | 0.029 |
| Hysterectomy (n, %) | 50 (6%) | 35 (7%) | 91 (16%) | <0.0001 |
| Oophorectomy (n, %) | 34 (4%) | 17 (3%) | 48 (8%) | 0.0003 |
BMI = body mass index, CES-D = Center for Epidemiologic Studies Depression Scale, VMS = vasomotor symptoms
N(%) or mean (SD) noted.
Table 2.
Association between baseline risk factors with trajectory of vasomotor symptoms (VMS). Odds ratios and 95% confidence intervals (OR, 95% CI) shown, with significant associations in bold type.
| Trajectory 2 or VMS increasing over time (reference = Trajectory 1 or minimal VMS) |
Trajectory 3 or Persistent VMS (reference = Trajectory 1 or minimal VMS) |
|
|---|---|---|
| Model 1: Sociodemographic characteristics (adjusts for center) | ||
| Age | 1.07 (1.03 – 1.1) | 1.02 (0.99 – 1.05) |
| Race (Black v. White) | 1.3 (1.03 – 1.64) | 2.01 (1.6 – 2.52) |
| Less than high school education | 1.09 (0.85 – 1.4) | 1.75 (1.38 – 2.21) |
| Household income <$50,000 per year | 0.92 (0.72 – 1.17) | 1.12 (0.88 – 1.43) |
| Difficulty paying for basic necessities | 0.95 (0.74 – 1.21) | 1.24 (0.99 – 1.57) |
| Model 2: Health conditions and behaviors (adjusts for center) | ||
| Age | 1.01 (0.98 – 1.05) | 1.08 (1.04 – 1.11) |
| Race (Black v. White) | 2.25 (1.76 – 2.87) | 1.55 (1.20 – 1.99) |
| CES-D score (at year 5) | 1.01 (0.99 – 1.02) | 1.03 (1.01 – 1.04) |
| Migraines | 0.94 (0.65 – 1.36) | 1.49 (1.07 – 2.07) |
| Tobacco use | 1.14 (0.87 – 1.5) | 1.52 (1.17 – 1.97) |
| BMI (kg/m2) | 0.94 (0.92 – 0.97) | 0.99 (0.97 – 1.01) |
| Model 3: Reproductive history (adjusts for center) | ||
| Age | 1.05 (0.997 – 1.11) | 0.95 (0.9 – 0.99) |
| Race (Black v. White) | 1.30 (0.93 – 1.82) | 2.09 (1.51 – 2.88) |
| Gravid vs. Nulligravid | 1.04 (0.91 – 1.20) | 1.14 (1.00 – 1.29) |
| Hysterectomy (at year 7) | 0.64 (0.21 – 1.91) | 1.73 (0.78 – 3.86) |
| Partial or total oophorectomy (at year 7) | 0.48 (0.17 – 1.38) | 0.83 (0.37 – 1.85) |
| Model 4: Combined model consisting of significant factors (adjusts for center) | ||
| Age | 1.08 (1.04 – 1.12) | 1.02 (0.99 – 1.06) |
| Race (Black v. White) | 1.53 (1.18 – 1.97) | 2.10 (1.63 – 2.69) |
| Less than high school education | 1.12 (0.85 – 1.46) | 1.64 (1.28 – 2.10) |
| CES-D score (at year 5) | 1.00 (0.99 – 1.02) | 1.02 (1.01 – 1.04) |
| Migraines | 0.94 (0.65 – 1.35) | 1.45 (1.04 – 2.02) |
| Tobacco use | 1.12 (0.85 – 1.48) | 1.40 (1.08 – 1.82) |
| BMI (kg/m2) | 0.94 (0.92 – 0.97) | 0.98 (0.96 – 1.00) |
BMI = body mass index, CES-D = Center for Epidemiologic Studies Depression Scale, VMS = vasomotor symptoms
Table 3.
Association between risk factors at year 15 with trajectory of vasomotor symptoms. Odds ratios and 95% confidence intervals (OR, 95% CI) shown, with significant associations in bold type.
| Trajectory 2 or VMS increasing over time (reference = Trajectory 1 or minimal VMS) |
Trajectory 3 or Persistent VMS (reference = Trajectory 1 or minimal VMS) |
|
|---|---|---|
| Model 1: Sociodemographic characteristics (adjusts for center) | ||
| Age | 1.06 (1.03 – 1.09) | 1.01 (0.98 – 1.04) |
| Race (Black v. White) | 1.28 (0.99 – 1.65) | 2.08 (1.62 – 2.66) |
| Less than high school education | 0.96 (0.71 – 1.28) | 1.42 (1.09 – 1.85) |
| Household income <$50,000 per year | ||
| Difficulty paying for basic necessities | 0.94 (0.68 – 1.31) | 1.52 (1.13 – 2.05) |
| Model 2: Health conditions and behaviors (adjusts for center) | ||
| Age | 1.07 (1.03 – 1.1) | 1.00 (0.97 – 1.04) |
| Race (Black v. White) | 1.58 (1.19 – 2.08) | 1.99 (1.52 – 2.62) |
| CES-D score | 1.00 (0.98 – 1.01) | 1.02 (1.01 – 1.04) |
| Migraines | 0.90 (0.64 – 1.27) | 1.80 (1.33 – 2.44) |
| Tobacco use | 1.04 (0.73 – 1.49) | 1.87 (1.36 – 2.56) |
| Alcohol use (ml/day) | 1.00 (0.99 – 1.01) | 1.01 (1.00 – 1.02) |
| BMI (kg/m2) | 0.96 (0.94 – 0.98) | 1.00 (0.98 – 1.01) |
| Systolic blood pressure (mm Hg) | 1.01 (0.99 – 1.02) | 1.01 (1.00 – 1.03) |
| Diastolic blood pressure (mm Hg) | 1.00 (0.98 – 1.02) | 1.00 (0.98 – 1.01) |
| Model 3: Reproductive history (adjusts for center) | ||
| Age | 1.05 (1.01 – 1.09) | 0.97 (0.94 – 1.01) |
| Race (Black v. White) | 1.16 (0.88 – 1.53) | 1.8 (1.37 – 2.36) |
| Parous vs. nulliparous | 0.90 (0.74 – 1.09) | 0.84 (0.69 – 1.02) |
| Hormone use | 0.24 (0.09 – 0.63) | 0.69 (0.35 – 1.35) |
| Oral contraceptive pill use | 0.71 (0.48 – 1.06) | 0.67 (0.45 – 1.00) |
| Perimenopause | 0.91 (0.67 – 1.24) | 1.49 (1.12 – 1.99) |
| Hysterectomy | 1.29 (0.70 – 2.37) | 3.20 (1.91 – 5.37) |
| Partial or total oophorectomy | 0.76 (0.36 – 1.6) | 1.17 (0.64 – 2.14) |
| Model 4: Combined model consisting of significant factors (adjusts for center) | ||
| Age | 1.07 (1.04 – 1.11) | 1.00 (0.96 – 1.03) |
| Race (Black v. White) | 1.59 (1.21 – 2.1) | 1.89 (1.44 – 2.48) |
| Less than high school education | 0.99 (0.71 – 1.39) | 1.19 (0.88 – 1.63) |
| Difficulty paying for basic necessities | 0.97 (0.70 – 1.36) | 1.21 (0.89 – 1.64) |
| CES-D score | 1.00 (0.98 – 1.01) | 1.02 (1.003 – 1.04) |
| Migraines | 0.88 (0.62 – 1.25) | 1.69 (1.24 – 2.29) |
| Tobacco use | 1.08 (0.76 – 1.54) | 1.90 (1.39 – 2.60) |
| BMI (kg/m2) | 0.96 (0.94 – 0.98) | 1.00 (0.98 – 1.01) |
| Hormone use | 0.29 (0.12 – 0.67) | 0.96 (0.53 – 1.74) |
| Perimenopause | 1.07 (0.81 – 1.42) | 1.59 (1.21 – 2.09) |
| Hysterectomy | 1.29 (0.76 – 2.18) | 2.72 (1.74 – 4.25) |
BMI = body mass index, CES-D = Center for Epidemiologic Studies Depression Scale, VMS = vasomotor symptoms
We conducted several sensitivity analyses. First, women who answered the question about presence of VMS were then asked, “Did the symptoms bother you?” Of the women who responded to this question (n=699), we conducted similar analyses described to those outlined above. We identified two trajectories of women with bothersome VMS: persistent bothersome VMS vs. non-bothersome VMS. We identified risk factors associated with persistent bothersome VMS. We also examined whether Black women had different patterns of associations compared to White women by examining patterns of associations stratified by race. We also excluded women who had any gynecologic surgery (hysterectomy or oophorectomy) by the Y15 exam. All analyses were conducted using SAS, Version 9.4. A threshold of p<.05 (two-sided) was used to determine statistical significance.
Results
Figure 1 shows the heat map displays of VMS across exams. The majority of women who answered questions about the presence or absence of VMS at ≥3 timepoints reported having VMS at least once between the Y15 and Y35 exams (Figure 1). Figure 2 shows VMS trajectories across exams, with chronologic age along the x-axis. Of the women who answered questions about presence of VMS at ≥3 timepoints, 43% (n=839) reported minimal VMS at any timepoint compared with women in the other trajectories, with frequency peaking at around 50 years of age and declining thereafter (trajectory 1, Figure 2). Twenty-seven percent (n=525) reported increasing VMS beginning at approximately 45 years of age (trajectory 2, Figure 2), whereas 33% (n=602) reported persistent VMS (trajectory 3, Figure 2) beginning with the Y15 exam until the Y35 exam.
Table 1 shows the characteristics of participants by VMS trajectory, as shown in Figure 2, in unadjusted comparisons. At baseline and Y15, women with persistent VMS (trajectory 3) were more likely to be Black, and to report greater difficulty paying for basic necessities. They were also more likely to report depressive symptoms, migraine headaches, cigarette use, and to have obesity than women with minimal VMS (trajectory 1). At Y15, although not at baseline, women with persistent VMS drank slightly more alcohol and had higher levels of systolic and diastolic blood pressure. At baseline, women with persistent VMS had more pregnancies and by Y7 were more likely to report hysterectomies and oophorectomies. At Y15, they were less likely to report OCP use and more likely to report hormone use, to be perimenopausal, and to report gynecologic surgery. Table 1 also shows the characteristics of women with increasing VMS, who tended to be similar to women with minimal VMS regarding their demographic factors, health conditions and behaviors, and cardiovascular risk factors except they had the lowest BMI out of the three categories of women.
Table 2 shows the multivariable associations between baseline risk factors and VMS persistence as shown in Figure 2. In multivariable models, factors associated with persistent VMS (trajectory 3) as opposed to low probability or infrequent VMS (trajectory 1) included Black race and lower levels of education, along with higher depressive symptom score, presence of migraines, and tobacco use. BMI was not associated with persistent VMS (Table 2, Models 1 and 2). After adjustment for age, race, and center, reproductive factors were not associated with VMS trajectory (Table 2, Model 3). Factors associated with increasing VMS (trajectory 2) as opposed to infrequent VMS (trajectory 1) included Black race; lower, rather than higher, BMI was associated with increasing VMS.
These patterns were similar to those observed for Year 15 risk factors and VMS trajectory as shown in Figure 2 (Table 3). In multivariable models, factors associated with persistent VMS (trajectory 3) as opposed to infrequent VMS (trajectory 1) included Black race, lower levels of education, difficulty paying for basic necessities, along with higher depressive symptom score, presence of migraines, and tobacco use. Perimenopause and hysterectomy were associated with prolonged VMS symptoms. Factors associated with increasing VMS (trajectory 2) as opposed to infrequent VMS (trajectory 1) included Black race. Hormone use was associated with lower odds of having increasing VMS. As with baseline risk factors, lower, rather than higher, BMI was associated with increasing VMS.
In multivariable models considering factors from both baseline and Y15 as well as age, race, and center, the factors associated with persistent VMS as opposed to infrequent VMS included less than a high school education (OR 1.70, 95% CI 1.22 – 2.38), migraines at Y15 (OR 1.75 – 95% CI 1.19, 2.58), tobacco use at Y15 (OR 2.06, 95% CI 1.37 – 3.11), perimenopause at Y15 (OR 1.71, 95% CI 1.29 – 2.27), and hysterectomy at Y15 (OR 2.47, 95% CI 1.56 – 3.92). Lower BMI at baseline (OR 0.95, 95% CI 0.92 – 0.99) although not at Y15 (OR 1.02, 95% CI 0.99 – 1.05) was also associated with membership in trajectory 2, or increasing VMS over time.
In sensitivity analyses, when we examined associations with bothersome VMS in the 699 women who responded to this question, risk factors for bothersome VMS were generally similar to those noted above, except that history of thyroid disease (OR 3.26, 95% CI 1.28 – 8.29) was associated with bothersome symptoms as well. When we excluded women who had undergone any gynecologic surgery by the Y15 exam, we found that the patterns of associations were similar. That is, Black race, less than a high school education, higher depressive symptom score, the presence of migraines, and cigarette use were still associated with persistent VMS at baseline and Y15, and perimenopause at Y15 was also associated with persistent with VMS. Lower BMI at baseline and Y15 was still associated with increasing VMS compared to minimal VMS. In models including Black women only, factors most strongly associated with persistent VMS included migraines, tobacco use, and hysterectomy at Y15. Again, lower BMI at baseline and Y15 were associated with increasing VMS as opposed to minimal VMS. Among white women only, factors associated with persistent VMS included less than a high school education from baseline, higher depressive symptom burden, tobacco use, and perimenopause, whereas lower BMI was associated with increasing VMS versus minimal VMS.
Discussion
Although VMS are estimated to affect over three-quarters of midlife women and to negatively impact quality of life, work performance, and healthcare utilization,22 our understanding of the etiology and risk factors for VMS is incomplete. Using data from a prospective population-based cohort study beginning in early adulthood, we found that risk factors for VMS could be identified in early adulthood, when women were in their twenties. These findings are unique in the prospective and comprehensive assessment of risk factors, beginning early in reproductive life.
Other reports, most notably those from the Study of Women’s Health Across the Nation (SWAN), have reported that women’s patterns of VMS vary in their temporality, specifically in their age at onset in the relationship to the final menstrual period (FMP) and persistence afterwards.23 In SWAN, a large cohort of midlife women, 4 VMS groups were identified among naturally menopausal women: persistent VMS, lower probability VMS, VMS declining after the FMP, and VMS increasing after the FMP. Our identification of 3 trajectories may be due to our inclusion of women with gynecologic surgery as well as not centering upon FMP, as we wished to characterize VMS in women experiencing these surgeries. We also included women using estrogen for similar reasons.
Like SWAN, we found that Black race and poor social determinants of health were robust predictors of VMS pattern.5 Specifically, we found that Black women and women with less than a high school education, cigarette use, depressive symptoms, and migrainous disorders during early adulthood were more likely to experience persistent VMS as opposed to infrequent VMS. To some extent, this was explained by the fact that women with adverse risk factors at baseline were more likely to have adverse risk factors at approximately 40 years of age, when CARDIA began ascertaining VMS. For example, when both tobacco use at baseline as well as later in life were considered, tobacco use at Y15 was more strongly associated with persistent VMS than tobacco use earlier in life. Several risk factor associations were stronger later in life rather than earlier life, such as the presence of changing menstrual cycle length and hysterectomy, because these factors were not present in early adulthood. However, several associations were stronger with baseline risk factors, including high school education, perhaps because this risk factor is upstream from other unspecified risk factors for VMS. Although Black women were also more likely to have increasing VMS as opposed to infrequent VMS, social determinants of health did not have significant associations with increasing VMS over time. However, hormone use at Y15 was associated with lower odds of increasing VMS over time, and lower BMI at baseline were associated with greater odds of increasing VMS over time.
Midlife cohorts including SWAN, the Australian Longitudinal Study on Women’s Health, and the Penn Ovarian Aging Study have previously reported on these risk factors and associations with persistent VMS.24–28 Our findings add to existing studies in our examination of risk factors when women were younger than 40 years. We also found that migrainous disorders also predict future risk of persistent VMS, an association that has been previously reported in cross-sectional studies of menopause.10 This suggests that the increased vasoreactivity that characterizes both disorders may be present in early adulthood, or that some other common pathway exists. In addition, we found that hysterectomy by the age of approximately 40 years is significantly linked with future VMS trajectory; unlike other reports, we did not exclude women who had undergone gynecologic surgery prior to cessation of menstruation, and we were also able to examine hysterectomy at younger ages.
Previous studies conflict regarding the association between BMI and VMS. Some cohorts report that higher BMI and obesity are linked with VMS.26–28 It is possible that we found a different pattern of associations with BMI due to how we characterized VMS, with emphasis upon prolonged VMS over time, or through our adjustment for other risk factors for VMS, such as tobacco use or OCP use in early adulthood, or confounding use of OCPs which are contraindicated for migraine with aura and in tobacco use, or other unmeasured confounders. However, as in our report, other studies have not found an association with persistent VMS.29–32 Of note, as with our report, SWAN also found that obesity was less common among women whose VMS symptoms increased over time i.e. late-onset VMS. One study of Black and White women in North Carolina noted that thin women who smoke in the premenopause are the most likely to experience VMS, and BMI in this cohort did not appear to be associated with VMS.32 BMI was also not associated with clinically significant VMS after multivariable adjustment in a in a cohort of midlife women in Maryland,29 a cohort of approximately ten thousand women in the United Kingdom,30 and among older women with an average of sixty-seven years.31
Early adulthood depressive symptoms and VMS, even in the absence of a formal diagnosis of any mental disorder, suggests that there may be shared pathways between depressive symptoms and VMS. The Midlife Women’s Health Study noted cross-sectional associations between depression and VMS.29 The nature of these pathways is speculative; SWAN investigators have reported that inflammatory markers reported to be associated with depressive symptoms do not predict onset of VMS,33 although it is not clear whether such markers might have significant associations when assessed earlier in adulthood, or with prolonged or more bothersome VMS. Both depressive symptoms and VMS have been linked with higher cortisol and norepinephrine, suggesting that both may be linked with central sympathetic activation and abnormalities in cortisol response.34 Ovarian steroids including estradiol and testosterone are associated with VMS35, 36 as well as depressive symptoms,37 but studies conflict regarding the direction of these associations and best practices regarding assays and characterization of fluctuations, and therefore more research is needed. Finally, hysterectomy, one of the most commonly performed surgeries for women, is a particularly strong risk factor for VMS. Although hysterectomy is generally not believed to alter sex steroid levels, at least one report notes that women with ovarian conservation who had undergone hysterectomy still experienced a significant decline in sex steroid levels.31 Others have reported that women undergoing hysterectomy may inaccurately report oophorectomy,38 and thus it is possible that the associations between hysterectomy and VMS may have been due to misclassification. Thus, systematic assessment is needed to determine if hormone levels change with hysterectomy, whether this varies with surgical technique, and to what extent factors predisposing to hysterectomy also increase risk of VMS.
We also found that risk factors for bothersome VMS were largely similar to persistent VMS in sensitivity analyses. Our findings add to existing studies by noting that hypothyroidism, reported even in early adulthood prior to the onset of VMS, can predict future VMS severity, even if thyroid disease was not associated with persistence of VMS over time. A report in the French E3N cohort study (where the women were aged approximately 46 years at baseline) noted that benign thyroid disease was associated with greater risk of future VMS.39 Women with and without hot flashes had similar thyroid stimulating hormone levels in a randomized trial of raloxifene in older women, suggesting that the association may not be significant if thyroid levels are within normal limits.31 Both hypothyroidism and VMS can be characterized by perimenopausal symptoms, suggesting that at least part of the reason for the association is due to greater vasoreactivity from two pathophysiologic processes. Whether systemic assessment for thyroid disease among midlife women experiencing VMS might be effective for guiding screening and treatment has not been studied.
Strengths of this report are that it is the first prospective study to examine early adulthood risk factors and their relationship with VMS, which we characterized by persistence as well as bothersomeness. The cohort is diverse and the list of potential risk factors extensive. There are several limitations, however. The ascertainment of VMS varies between reports, and there is a lack of consensus on how to distinguish between VMS due to estrogen withdrawal versus other etiologies and how to categorize persistence and severity of VMS. Health conditions including migraine headaches and thyroid disease were obtained by self-report and may be imprecise compared to validation through medical record review; in particular, migraine with aura was not assessed which is relevant for estrogen use. We could not examine the potential roles of mechanistic pathways, including sex steroid levels and inflammatory markers. Our results may not extend to women of other racial and ethnic groups. Whether the experience of VMS has changed across cohorts is unknown, particularly because VMS data were collected upon women who were aged approximately 40 years of age in the year 2000, and thus may not reflect the menopausal experience of women currently; such women may have access to other therapies such as serotonin reuptake inhibitors and have greater use of exogenous sex steroids. Finally, CARDIA did not collect thyroid stimulating hormone levels or thyroxine levels, and thus could not determine whether thyroid disease was controlled or not; this would bias associations to the null.
Conclusions
Risk factors for VMS, a condition that limits quality of life and also predicts other morbidity, may be identified in early adulthood. Further examination of shared risk factors, particularly migraines and depressive symptoms beginning in early adulthood, may be helpful in identifying therapies for VMS. Specifically, therapies for these risk factors could be examined for their impact upon VMS, particularly calcitonin gene-related peptide antagonists used for migraine and pharmacologic therapies used for mood disorders. Among women with thyroid disease, the impact of thyroid therapy upon VMS should be examined further. Inter-relationships between sex steroid milieu, gynecologic surgeries, depressive symptoms, and VMS should be examined further beginning in reproductive age.
Sources of funding:
The Coronary Artery Risk Development in Young Adults Study (CARDIA) is conducted and supported by the National Heart, Lung, and Blood Institute (NHLBI) in collaboration with the University of Alabama at Birmingham (HHSN268201800005I & HHSN268201800007I), Northwestern University (HHSN268201800003I), University of Minnesota (HHSN268201800006I), and Kaiser Foundation Research Institute (HHSN268201800004I). This work was supported by R56HL169167. This manuscript has been reviewed by CARDIA for scientific content.
Financial disclosures/Conflicts of interest:
Abbi Lane received past institutional funding from Nutrasource. Richard J Auchus did contracted research for Neurocrine Biosciences/Diurnal LTD, Spruce Biosciences, Corcept Therapeutics, and Sparrow Pharmaceuticals; was a consultant for Quest Diagnostics, Corcept Therapeutics, Xeris Pharmaceuticals, Crinetics Pharmaceuticals, Adrenas Therapeutics, PhaseBio Pharmaceuticals, Novo Nordisk, Neurocrine Biosciences/Diurnal LTD, Recordati Rare Diseases, H Lundbeck A/S, and Sparrow Pharmaceuticals. The other authors have nothing to disclose.
REFERENCES
- 1.Kronenberg F Hot flashes: epidemiology and physiology. Ann N Y Acad Sci. 1990;592:52–86; discussion 123–33.doi: 10.1111/j.1749-6632.1990.tb30316.x [DOI] [PubMed] [Google Scholar]
- 2.Thurston RC, Bromberger JT, Joffe H, et al. Beyond frequency: who is most bothered by vasomotor symptoms? Menopause. 2008;15(5):841–7.doi: 10.1097/gme.0b013e318168f09b [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Todorova L, Bonassi R, Guerrero Carreño FJ, et al. Prevalence and impact of vasomotor symptoms due to menopause among women in Brazil, Canada, Mexico, and Nordic Europe: a cross-sectional survey. Menopause. 2023.doi: 10.1097/gme.0000000000002265 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Zhao W, Smith JA, Yu M, et al. Genetic variants predictive of reproductive aging are associated with vasomotor symptoms in a multiracial/ethnic cohort. Menopause. 2021;28(8):883–92.doi: 10.1097/gme.0000000000001785 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Tepper PG, Brooks MM, Randolph JF Jr., et al. Characterizing the trajectories of vasomotor symptoms across the menopausal transition. Menopause. 2016;23(10):1067–74.doi: 10.1097/gme.0000000000000676 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Duralde ER, Sobel TH, Manson JE. Management of perimenopausal and menopausal symptoms. Bmj. 2023;382:e072612.doi: 10.1136/bmj-2022-072612 [DOI] [PubMed] [Google Scholar]
- 7.DePree B, Houghton K, DiBenedetti DB, et al. Practice patterns and perspectives regarding treatment for symptoms of menopause: qualitative interviews with US health care providers. Menopause. 2023;30(2):128–35.doi: 10.1097/gme.0000000000002096 [DOI] [PubMed] [Google Scholar]
- 8.Friedman GD, Cutter GR, Donahue RP, et al. CARDIA: study design, recruitment, and some characteristics of the examined subjects. J Clin Epidemiol. 1988;41(11):1105–16.doi: 10.1016/0895-4356(88)90080-7 [DOI] [PubMed] [Google Scholar]
- 9.Kim C, Catov J, Schreiner PJ, et al. Women’s Reproductive Milestones and Cardiovascular Disease Risk: A Review of Reports and Opportunities From the CARDIA Study. J Am Heart Assoc. 2023;12(5):e028132.doi: 10.1161/jaha.122.028132 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Faubion SS, Smith T, Thielen J, et al. Association of Migraine and Vasomotor Symptoms. Mayo Clin Proc. 2023;98(5):701–12.doi: 10.1016/j.mayocp.2023.01.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Worsley R, Bell R, Kulkarni J, Davis SR. The association between vasomotor symptoms and depression during perimenopause: a systematic review. Maturitas. 2014;77(2):111–7.doi: 10.1016/j.maturitas.2013.11.007 [DOI] [PubMed] [Google Scholar]
- 12.Reed SD, LaCroix AZ, Anderson GL, et al. Lights on MsFLASH: a review of contributions. Menopause. 2020;27(4):473–84.doi: 10.1097/gme.0000000000001461 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.The 2022 hormone therapy position statement of The North American Menopause Society. Menopause. 2022;29(7):767–94.doi: 10.1097/gme.0000000000002028 [DOI] [PubMed] [Google Scholar]
- 14.Avis NE, Stellato R, Crawford S, et al. Is there a menopausal syndrome? Menopausal status and symptoms across racial/ethnic groups. Soc Sci Med. 2001;52(3):345–56.doi: 10.1016/s0277-9536(00)00147-7 [DOI] [PubMed] [Google Scholar]
- 15.Harlow SD, Gass M, Hall JE, et al. Executive summary of the Stages of Reproductive Aging Workshop + 10: addressing the unfinished agenda of staging reproductive aging. Menopause. 2012;19(4):387–95.doi: 10.1097/gme.0b013e31824d8f40 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Whitham HK, Maclehose RF, Harlow BL, Wellons MF, Schreiner PJ. Assessing the utility of methods for menopausal transition classification in a population-based cohort: the CARDIA Study. Maturitas. 2013;75(3):289–93.doi: 10.1016/j.maturitas.2013.04.015 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Radloff LS. The CES-D Scale:A Self-Report Depression Scale for Research in the General Population. Applied Psychological Measurement. 1977;1(3):385–401.doi: 10.1177/014662167700100306 [DOI] [Google Scholar]
- 18.Nielsen F Hierarchical clustering, Introduction to HPC with MPI for Data Science: Springer; 2016. [Google Scholar]
- 19.Jones BL, Nagin DS. Advances in Group-Based Trajectory Modeling and an SAS Procedure for Estimating Them. Sociological Methods & Research. 2007;35(4):542–71.doi: 10.1177/0049124106292364 [DOI] [Google Scholar]
- 20.Nagin D Group-based modeling of development over the life course. Cambridge, MA: Harvard University Press; 2005. [Google Scholar]
- 21.Akaike H A new look at the Bayes procedure. Biometrika. 1978;65(1):53–9.doi: 10.1093/biomet/65.1.53 [DOI] [Google Scholar]
- 22.Faubion SS, Enders F, Hedges MS, et al. Impact of Menopause Symptoms on Women in the Workplace. Mayo Clin Proc. 2023;98(6):833–45.doi: 10.1016/j.mayocp.2023.02.025 [DOI] [PubMed] [Google Scholar]
- 23.Tepper P, Brooks M, Randolph J Jr., et al. Characterizing the trajectories of vasomotor symptoms across the menopausal transition. Menopause. 2016;23(10):1067–74 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Gold EB, Colvin A, Avis N, et al. Longitudinal analysis of the association between vasomotor symptoms and race/ethnicity across the menopausal transition: study of women’s health across the nation. Am J Public Health. 2006;96(7):1226–35.doi: 10.2105/ajph.2005.066936 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Avis NE, Crawford SL, Greendale G, et al. Duration of menopausal vasomotor symptoms over the menopause transition. JAMA Intern Med. 2015;175(4):531–9.doi: 10.1001/jamainternmed.2014.8063 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Avis NE, Crawford SL, Green R. Vasomotor Symptoms Across the Menopause Transition: Differences Among Women. Obstet Gynecol Clin North Am. 2018;45(4):629–40.doi: 10.1016/j.ogc.2018.07.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Freeman EW, Sammel MD, Lin H, Liu Z, Gracia CR. Duration of menopausal hot flushes and associated risk factors. Obstet Gynecol. 2011;117(5):1095–104.doi: 10.1097/AOG.0b013e318214f0de [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Wilson LF, Pandeya N, Byles J, Mishra GD. Hot flushes and night sweats symptom profiles over a 17-year period in mid-aged women: The role of hysterectomy with ovarian conservation. Maturitas. 2016;91:1–7.doi: 10.1016/j.maturitas.2016.05.011 [DOI] [PubMed] [Google Scholar]
- 29.Ziv-Gal A, Smith RL, Gallicchio L, Miller SR, Zacur HA, Flaws JA. The Midlife Women’s Health Study - a study protocol of a longitudinal prospective study on predictors of menopausal hot flashes. Womens Midlife Health. 2017;3:4.doi: 10.1186/s40695-017-0024-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Hunter MS, Gentry-Maharaj A, Ryan A, et al. Prevalence, frequency and problem rating of hot flushes persist in older postmenopausal women: impact of age, body mass index, hysterectomy, hormone therapy use, lifestyle and mood in a cross-sectional cohort study of 10,418 British women aged 54–65. Bjog. 2012;119(1):40–50.doi: 10.1111/j.1471-0528.2011.03166.x [DOI] [PubMed] [Google Scholar]
- 31.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):840–6.doi: 10.1001/archinte.168.8.840 [DOI] [PubMed] [Google Scholar]
- 32.Schwingl PJ, Hulka BS, Harlow SD. Risk factors for menopausal hot flashes. Obstet Gynecol. 1994;84(1):29–34 [PubMed] [Google Scholar]
- 33.Gold EB, Xing G, Avis NE, et al. The longitudinal relation of inflammation to incidence of vasomotor symptoms. Menopause. 2022;29(8):894–904.doi: 10.1097/gme.0000000000002005 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Gordon JL, Rubinow DR, Thurston RC, Paulson J, Schmidt PJ, Girdler SS. Cardiovascular, hemodynamic, neuroendocrine, and inflammatory markers in women with and without vasomotor symptoms. Menopause. 2016;23(11):1189–98.doi: 10.1097/gme.0000000000000689 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Santoro N, Roeca C, Peters BA, Neal-Perry G. The Menopause Transition: Signs, Symptoms, and Management Options. J Clin Endocrinol Metab. 2021;106(1):1–15.doi: 10.1210/clinem/dgaa764 [DOI] [PubMed] [Google Scholar]
- 36.El Khoudary SR, Thurston RC. Cardiovascular Implications of the Menopause Transition: Endogenous Sex Hormones and Vasomotor Symptoms. Obstet Gynecol Clin North Am. 2018;45(4):641–61.doi: 10.1016/j.ogc.2018.07.006 [DOI] [PubMed] [Google Scholar]
- 37.Hemachandra C, Islam RM, Bell RJ, Sultana F, Davis SR. The association between testosterone and depression in postmenopausal women: A systematic review of observational studies. Maturitas. 2023;168:62–70.doi: 10.1016/j.maturitas.2022.11.001 [DOI] [PubMed] [Google Scholar]
- 38.Phipps AI, Buist DS. Validation of self-reported history of hysterectomy and oophorectomy among women in an integrated group practice setting. Menopause. 2009;16(3):576–81.doi: 10.1097/gme.0b013e31818ffe28 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Sabia S, Fournier A, Mesrine S, Boutron-Ruault MC, Clavel-Chapelon F. Risk factors for onset of menopausal symptoms: results from a large cohort study. Maturitas. 2008;60(2):108–21.doi: 10.1016/j.maturitas.2008.04.004 [DOI] [PubMed] [Google Scholar]