Reproductive aging in women is an endocrine process that extends for a number of years before the cessation of menses. 1 The World Health Organization defined the perimenopause as the period immediately before the menopause when endocrinological, biological, and clinical features of approaching menopause commence 2 but was unable to provide an operational definition of the onset of this period. At present, there are no hormone or symptom markers for the onset of the perimenopause, and clinical definitions are based on observable changes in menstrual regularity. 2 However, evidence that biochemical and hormonal changes may occur before observable menstrual changes is increasing, 3–6 and definitions of the perimenopause that depend on changes in menstrual regularity do not account for observable symptoms and menstrual changes that can be hormonally characterized as different from those of premenopausal women. 7
The purpose of this report is to describe the symptoms in a cohort of generally healthy women in their late reproductive years to establish a symptom baseline for longitudinal study of changes commonly associated with the menopause. A second aim is to compare African American and white women to determine whether there are racial differences in symptom reports, which is an important question that is largely unanswered at this time. Previous research has shown that variables, such as education, income, marital status, and body mass index (BMI) have racial differences and are also associated with mood symptoms such as depression. We therefore hypothesized that any symptom differences between African American and white women would be negligible after controlling for variables known to differentiate these racial groups. The third aim is to show the extent to which variables in reproductive health, mood and behavior, lifestyle, and demographic background are associated with the reported symptoms. Based on the existing literature, we hypothesized that previous mood problems predicted the mood symptoms, but that physiological symptoms commonly associated with the menopause were not predicted by previous mood problems.
The subjects in this report were identified through random digit dialing and asked to participate in a cohort study of psychological, behavioral, and hormonal factors in the late reproductive years. Recruitment was stratified by race to enroll equal numbers of African American and white women. The study enrolled 436 women, with 218 women in each racial group. A total of 1,420 women (563 African American, 633 white, and 224 race unknown) were identified through random digit dialing as potentially eligible for the study. There were 402 women who refused screening, and 1,018 women were fully screened to yield 580 eligible and 438 ineligible subjects. Of the eligible women, 436 participated (218 in each racial group) and 144 declined to participate. The study was approved by the University Institutional Review Board, and written informed consent was obtained.
Eligibility criteria for the study of the late reproductive years included ages of 35 to 47 years, menstrual cycles in normal range (22–35 days) for the previous 3 months, intact ovaries and uterus, and no serious health problem that compromised ovarian function. The age range was selected to study hormone levels and symptoms in the late reproductive years before observable changes in menstrual regularity. The lower age limit of 35 years was selected as the age when follicular depletion accelerates, resulting in hormone changes. 8 The upper age limit of 47 years was selected as an accepted cutoff for women entering the menopausal period, based on the median age for the onset of the perimenopause as 47.5 years. 9 Exclusion criteria included current use of hormonal medications (including oral contraceptives), pregnancy or breast feeding, current use of psychotropic medications, current alcohol or drug abuse, illnesses that could affect hormonal function (e.g., diabetes, liver disease, breast or endometrial cancer), non-English-speaking, or knowledge that the subject would be leaving the area during the study period. Further details of the subject enrollment are described elsewhere. 10
This is a cross-sectional analysis of symptom reports at the enrollment baseline in a 4-year prospective cohort study.
Study visits were scheduled to occur within the first 6 days of the menstrual cycle to control for menstrual cycle effects on symptoms and hormones of the study. Nearly all visits occurred in the subjects' homes. A trained research interviewer administered a structured interview questionnaire, subjects completed standard self-report questionnaires, and blood samples were obtained for later hormone analyses.
The interview questionnaire was constructed to collect information on menstrual cycle characteristics, reproductive health, medical history, current medications, health practices and behaviors, and demographic background factors. Measures of height, weight, and waist and hip circumferences were obtained; BMI was calculated from the weight in kilograms divided by height in meters squared. 11 The standard self-report measures to assess mood and behavior included the following:
- The CES-D Scale, which is a well-established epidemiological measure of depression, 12 assessed current depressive symptoms. Subjects rated 20 items that relate to depressed mood from 0 (not at all) to 4 (most of the time); the ratings were summed for a total score. Scores greater than 16 were classified as indicating depression. 13
- The Zung Anxiety Scale, which is a validated measure that is sensitive to the frequency of anxiety symptoms, 14 assessed current anxiety. Subjects rated 5 affective and 15 somatic symptoms on four-point scales ranging from none (0) to all of the time (4). The ratings were summed for a total score, with higher scores signifying more anxiety.
- The Perceived Stress Scale, which is a 14-item validated measure of the degree to which situations are appraised as stressful, 15 assessed current stress. Subjects rated the items on five-point scales from 0 (never) to 4 (very often). Total scores were obtained by reverse-scoring the seven positive items and summing all ratings. The Perceived Stress Scale has been correlated with depression, physical symptomatology, and anxiety and has been shown to measure a different and independently predictive construct of appraised stress. 15
- The SF-12 Health Survey, 16 which is a short form developed from the SF-36 Health Survey to monitor health outcomes, 17 was used to evaluate the subjects' physical and mental health. The summary scales have demonstrated equivalence to the SF-36. 16 The 12 items yield two reliable summary scales for physical and mental health, with higher scores signifying better health.
- Quality of life was assessed with the Q-LES-Q short form, 18 which is a self-report measure of various aspects of daily living plus a global assessment of life satisfaction during the past week. The 14 items are the summary scales of the Q-LES-Q, which was shown to be a reliable and valid measure of a construct of quality of life that is not redundant with measures of depression. 18 Subjects rated each item from very poor (1) to very good (5); the items were summed for a total score, with higher scores indicating better quality of life.
Whether the patient experienced menopausal symptoms was asked in the structured interview administered at study enrollment. The subjects were asked “Have you ever had menopausal symptoms, such as hot flashes, night sweats, or vaginal dryness?” and responses were coded as “yes” or “no.”
The subjects rated a daily symptom report (DSR) throughout one menstrual cycle. The daily symptom rating form was constructed to include 20 common menopausal complaints based on the literature. Subjects rated each of the DSR items from 0 (not present) to 4 (very severe) (Appendix 1). The DSR was initially used and results were reported for a pilot study for this project. 19 The reliability of the DSR was estimated at 0.94 for Cronbach's coefficient alpha, indicating high internal consistency for the 20 items. Statistically derived factors of the DSR are described below.
Daily ratings on the DSR started at the first visit (scheduled within days 1–6 of the cycle), when the ratings were completed for the previous day. Subjects were then asked to complete the ratings at the end of each day until the next visit, which was scheduled for the same time in the next menstrual cycle. DSR scores were calculated by summing item ratings for 6 days after menses (days 6–12) and 6 days before the next menses (days 23–28, with day 28 anchored at the day before bleeding). This approach was selected because it provided scores within the same menstrual cycle for all subjects. The scores averaged potentially high and low symptom levels to describe the overall symptom status of the subjects, which was the objective of this report. The method standardized the cycle scores to a 28-day cycle and allowed for comparisons of follicular and luteal phase scores in other analyses.
The factor structure of the DSR was examined using principal components analysis of the total scores for each of the 20 items. The DSR produced three factors with eigen values >1, and the three factors were retained after analysis with varimax rotation. The factor loadings were clear-cut for all items (cutoff >0.45) except headache, which loaded similarly on factors 1 and 2. The decision was to include headache in factor 2 with other somatic symptoms. The items in each factor were as follows: factor 1 (psychological symptoms)—depression, irritability, anxiety, concentration difficulties, feeling out of control, decreased interest, mood swings, fatigue, insomnia, decreased libido; factor 2 (somatic symptoms)—swelling and weight gain, appetite changes, breast tenderness, aches, headache; factor 3 (physiological symptoms)—urine leaks, dizziness, hot flashes, vaginal dryness, poor motor coordination/clumsiness. The standardized coefficient alpha was 0.78 for factor 1, 0.79 for factor 2, and 0.85 for factor 3.
Before factor analysis of the total DSR score, preliminary analyses of the item scores were conducted to identify the extent of cycle phase differences in symptom severity. Although the mean luteal scores were significantly higher than the mean follicular scores for all DSR items, the effect sizes of these differences were very low, and the items had only a slight scale change that was not clinically meaningful. The physiological symptoms had almost no change over the menstrual cycle. For example, the DSR items of hot flashes, vaginal dryness, and urine leaks had very low effect sizes of 0.16, 0.13, and 0.12, respectively.
To further examine our concern that the associations of the variables in this report might be distorted because of the non-responders to the DSR, using the DSR subsample, we examined the set of variables that were previously reported for the full cohort, 10 which examined the associations of race, age, education, depression, dysmenorrhea, and BMI with subjects' global report of menopausal symptoms. We did not find any meaningful differences in the associations of these variables in the DSR subsample compared with the full cohort. We then created an indicator variable for “completed DSR” versus “not completed DSR.” Each variable in the previously reported model was then tested for the interaction between that variable and the “completed DSR” variable. There were no statistically significant differences, indicating that the associations appeared the same in the DSR cohort as in the non-DSR group. These results suggest that the associations observed in the DSR cohort may not be different from those that would be seen in the full cohort.
The associations of race and age with each DSR factor were tested in two-factor analysis of variance models. Additional variables, selected on the basis of previously reported associations with menopause, 10,20,21 were examined in relation to the DSR scores and included the following domains:
- Mood and behavior: depression; anxiety; perceived stress; history of depression [subject report of no (0) or yes (1)]; history of postpartum depression [subject report of no (0) or yes (1)]; premenstrual symptoms that disrupt functioning [subject rating of the degree to which premenstrual symptoms interfered with daily activities on a scale of 0 (none) to 4 (a lot), answer 4 was then scored as 1 and all others scored 0]; and quality of life.
- Reproductive health: dysmenorrhea [subject rated none (0) to severe (3); moderate/severe was coded as 1, and none/mild was coded as 0]; cycle length (subject report for overall length in past year); short cycles (subject report of any cycles less than 21 days in the past year); long cycles (subject report of any cycles over 35 days in the past year); age at menarche; and uterine fibroids (subject reported yes or no).
- Physical health: BMI, physical health status as assessed by the SF-12, current smoking, and alcohol use.
- Demographic background: employment, education, and marital status.
The following four variables had no significant correlation with DSR scores (p > 0.20): marital status, alcohol use (number of drinks/week), age at menarche, and uterine fibroids (yes/no). These factors were not examined further.
All variables with p values ≤ 0.20 in the three-variable DSR models were entered in a general linear model adjusted for race and age to determine independent contributions to each of the DSR factor scores. The significance level for inclusion in the models was selected to evaluate and control for all potential confounding variables in the multivariate analyses. Logistic regression was used for DSR factor 3 (with “any vs. no symptoms” as the outcome) because of the number of zero scores (38% of the subjects) on this factor. To check the consistency of the results as yielded by the general linear model, the same sets of variables also were examined in stepwise regression models. The results of each pair of stepwise regression models (with and without forcing age and race) were nearly identical to each other and to the general linear models shown in this report. In all analyses, tests were two-tailed with p values ≤ 0.05 considered significant. The statistical analyses were performed using SAS version 6.12 (SAS Institute, Inc., Cary, NC, USA). 22
Of the 436 women who enrolled in the study, 308 (126 African American women, 182 white women;p < 0.001) completed 1 month of daily symptom ratings on the DSR at the initial assessment period. Comparisons of major study variables between the subjects who completed the DSR and the subjects who did not showed no differences in age or menstrual cycle length. The subjects who completed the DSR were more likely to have a high school diploma than those who did not (92% vs. 83%;p < 0.01). The subjects who completed the DSR were less likely to report symptoms compared with those who did not as measured by the following: 36% of those who completed the DSR reported menopausal symptoms in the interview versus 54% of those who did not (p < 0.001); CES-D [mean, 15 (SD, 11) vs. mean, 17 (SD, 11)] (p = 0.04); Perceived Stress Scale [mean, 20 (SD, 8) vs. mean, 24 (SD, 7)] (p = 0.0006); and Zung Anxiety Index [mean, 34 (SD, 8) vs. mean, 37 (SD, 9)] (p = 0.004).
Table 1 shows subject characteristics by racial group. The mean age of the subjects was 41 (SD, 3) years (median, 41 years; range, 35–47 years). The two racial groups did not differ in age, cycle length, age at menarche, completion of a high school education, current smoking status, number of alcoholic drinks per week, or history of depression. The white women were significantly more likely to be employed, more likely to be married, and had a lower BMI (Table 1).
Table 2 shows the DSR scores and other symptom measures for the two racial groups. The African American group consistently had higher scores on the symptom measures, indicating higher symptom levels. Thirty-seven percent of the healthy women at a mean age of 41 years with normal menstrual cycles responded that they experienced menopause-type symptoms (46% African American and 30% white;p < 0.001).
The prospective symptom ratings on the DSR indicated that overall the subjects rated the symptoms from none to mild, as would be expected in nonclinical subjects. The symptoms most frequently reported by these nonclinical subjects were fatigue (reported by 83% of the subjects), swelling and weight gain (78%), anxiety/tension (75%), headaches (75%), and appetite changes (74%), but the severity was generally rated from minimal to mild. The physiological symptoms of DSR factor 3 were the least frequently reported by the subjects who had regular menstrual cycles. These included central nervous system symptoms of hot flashes (30%), dizziness (30%), poor coordination/clumsiness (24%), and reproductive system symptoms of urine leaks (22%) and vaginal dryness (21%).
Associations of race and age with symptom scores
Race was significantly associated only with the physiological symptoms (factor 3), which the African American women rated more severe than the white women (p < 0.04). The primary symptoms in this factor that contributed to the racial differences were hot flashes (p < 0.008) and poor coordination/clumsiness (p < 0.04), with the African American women having the higher scores.
Age was significantly associated only with the somatic symptoms (factor 2), which included swelling/weight gain, appetite changes, breast tenderness, aches, and headache. The oldest age group (45–47 years) had the highest scores on this factor and were more likely to report aches/joint pains (p = 0.03).
Associations of other variables with DSR scores
Tables 3 and 4 show the associations of all study variables with the DSR factor scores. After examining each of these variables individually in the DSR models adjusted for race and age, multivariate analyses were conducted (Tables 5–7). The strongest predictors of psychological symptoms (factor 1) after adjusting for race and age were depression (p = 0.0001), perceived stress (p = 0.005), and poorer physical health (p = 0.04). Race and age remained nonsignificant (Table 5).
The strongest predictors of the somatic symptoms (factor 2) after adjustment for race and age were poorer physical health (p = 0.008) and longer cycle length (p = 0.04). The association of race with factor 2 remained nonsignificant, whereas the association of age increased because of correlations of the added variables with both age and outcome (Table 6).
Analyses of the same set of variables with the physiological symptoms (factor 3) showed that race remained an independent predictor of the physiological symptoms after adjustment for the other variables in the model. Other significant predictors were depression (p = 0.03), anxiety (p = 0.05), previous postpartum depression (p = 0.003), and severe premenstrual symptoms (p = 0.04). The interaction of race and age was significant (p = 0.01), and the results are therefore shown for each racial group (Table 7). In the African American group, women in their forties were more symptomatic compared with women of 35 to 39 years of age. Previous postpartum depression (p = 0.03) significantly predicted factor 3 symptoms adjusted for the other variables in the model. In the white group, women in their forties were less symptomatic than women of 35 to 39 years of age. Anxiety (p = 0.04) and poorer physical health, as assessed by the SF-12 (p = 0.03), were significant predictors of factor 3 symptoms after adjustment for the other variables in the model. Interactions of race with each variable in the full-sample model were significant only for age and dysmenorrhea (p = 0.05).
This study of DSRs in a community-based sample of generally healthy women was conducted to provide a baseline in the late reproductive years for longitudinal study of symptom changes that subsequently occur in the approach to menopause. Overall, psychological and somatic symptoms were minimal, as would be expected in a nonclinical group. However, 37% reported in the interview that they experienced menopausal symptoms. In prospective reports of daily symptom ratings that followed the interview, 30% of the women reported hot flashes, which are commonly viewed as “true” menopausal symptoms. 23 It is these reports of menopause-type symptoms that are of particular interest inasmuch as the mean age of the women was 41 years (approximately 10 years before the mean age at menopause 9), and the symptoms preceded the onset of menstrual irregularities, which constitute the traditional definition of perimenopause. 2,24 The findings suggest that menopausal symptoms may occur earlier in the late reproductive years than has been commonly recognized and may not be due entirely to menopausal declines in estradiol levels.
Another unexpected result of this study was the finding of racial differences in DSRs after adjusting for variables known to differentiate these groups, such as education, income, and BMI. More African American women than white women reported that they experienced “menopausal symptoms” in the interview and also rated higher scores for hot flashes and other physiological symptoms as assessed by the prospective DSRs. Race was associated only with the physiological symptom factor (and not with the psychological or somatic symptom factors) and remained after adjusting for age, BMI, smoking, and the other demographic background and clinical variables of the study. Furthermore, the physiological symptom ratings increased with age in the African American group, whereas the reverse was observed in the white group. Further longitudinal study is needed to determine whether these racial differences are maintained, increase, or diminish as the women approach menopause.
There are few studies of the associations of race with menopausal symptoms, and existing reports are contradictory, in part, because of differing definitions of menopausal status. Several reports showed no racial differences in menopausal symptoms. 24,25 However, a recent study reported more estrogen-related symptoms in African American women compared with white women, 26 which is similar to the findings of this study. African American women also were more likely to experience hot flashes compared with white women after surgical menopause, independent of weight and hormone replacement status. 27 Another study found that African American women had an earlier median age at menopause, 28 an important finding that requires further confirmation. Further study of the present cohort will determine whether the observed racial differences in symptoms are maintained with continued ovarian aging and whether African American women experience hormonal changes of the menopause earlier than white women.
The self-reported depression ratings as assessed by the CES-D seemed relatively high inasmuch as a score of 16 has long been considered to denote the depression threshold. 13 More importantly, the African American women reported significantly higher depression scores than the white women. Seventeen percent of the African American women and 13% of the white women had CES-D scores ≥ 16. Further examination of depressive diagnoses that were obtained from administration of the Primary Care Evaluation of Mental Disorders (PRIME-MD) 29 in this cohort supported these results. Twenty-one percent of the African American women versus 9% of the white women met Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) criteria for depressive diagnoses, as assessed by the PRIME-MD. 30 Thirteen percent of the African American women and 4% of the white women met criteria for major depression as assessed by the PRIME-MD. This is within the range of rates for major depression as identified in the National Comorbidity Study, in which the rates for major depression ranged from 2% to 16% in the primary care outpatient population. 31 The greater prevalence of depressive disorders among African American women was unexpected but was also found to be notably similar to another recent urban community-based study in Britain, in which depressive disorders were significantly more common among African Caribbean women (19%) than white women (11%). 32
Age was significantly associated with the somatic symptom cluster (aches, headache, swelling/weight gain, breast tenderness, appetite changes). The oldest age group had the highest somatic symptom levels, with aches and joint pains being the primary complaint. The association of somatic symptoms with longer cycle lengths was of interest inasmuch as cycle lengths increase approximately 2 to 4 years before menopause, 33,34 suggesting a possible association of these symptoms with ovarian aging.
Limitations of this study include self-selection in the recruitment of the cohort as well as in the subset who completed the daily symptom ratings. The sampling was stratified to obtain equal numbers of African American and white women, and the participation rate was reasonable for population-based recruitment methods. 35,36 The education level and employment status of the cohort were representative of women of this age group in Philadelphia County. The subset of subjects who completed the DSR were more likely to be white, to have a high school diploma, and to be employed. Potential bias of the DSRs may be in the direction of less symptomatology inasmuch as the comparisons of all other standard symptom measures obtained at enrollment showed that the subset of subjects who completed the DSR had lower symptom scores than those who did not in both the African American and white groups. Another limitation is that women who were known to have chronic physical or psychiatric illnesses were excluded from enrollment in the cohort. Although the cohort can provide much-needed information on reproductive aging in generally healthy women, it is limited for evaluation of physical and psychiatric disorders that may have strong associations with menopausal symptoms. This report of baseline symptom status does not include hormonal changes, which may have important effects on symptom expression but are not available until follow-up assessments when changes from baseline can be measured.
In conclusion, these data provide a baseline in the late reproductive years for symptoms commonly associated with the menopause. Although the physiological symptoms, such as hot flashes may be signals of ovarian aging, they have been largely ignored in research and clinical practice when they occur before observable menstrual cycle changes. Others have suggested that the perimenopause may begin much earlier than is commonly defined and noted that biological changes that herald the onset of the perimenopausal transition, such as decreased bone density, decreased fertility, and increased embryonic aneuploidy, also may occur earlier than is generally accepted. 1 Further longitudinal study of this cohort of late reproductive-age women will 1) examine changes in the observed symptoms and the associations of these changes with hormonal measures of ovarian aging, 2) determine whether racial differences are maintained, and 3) determine whether African American women reach an earlier menopause.
This work was supported by a grant from the National Institutes of Health, RO1-AG-12745.
1. Seifer DB, Naftolin F. Moving toward an earlier and better understanding of perimenopause. Fertil Steril 1998; 69:387–8.
2. WHO Scientific Group 1996 Research on the Menopause in the 1990's. A report of the WHO Scientific Group. World Health Organization, Geneva, Switzerland. 1996;866:1–79.
3. Welt CK, McNicholl DJ, Taylor AE, Hall JE. Female reproductive aging is marked by decreased secretion of dimeric inhibin. J Clin Endocrinol Metab 1999; 84:105–11.
4. Santoro N, Tovaghgol A, Skurnick JH. Decreased inhibin tone and increased activin A secretion characterize reproductive aging in women. Fertil Steril 1999; 71:658–62.
5. Danforth DR, Arbogast LK, Mroueh J, et al. Dimeric inhibin: a direct marker of ovarian aging. Fertil Steril 1998; 70:119–23.
6. Soules MR, Battaglia DE, Klein NA. Inhibin and reproductive aging in women. Maturitas 1998; 30:193–204.
7. Prior JC. Perimenopause: the complex endocrinology of the menopausal transition. Endocr Rev 1998; 19:397–428.
8. Speroff L, Glass RH, Kase NG. Clinical gynecologic endocrinology and infertility
, 5th ed. Baltimore: Williams & Wilkins, 1994.
9. McKinlay, Brambilla DJ, Posner JG. The normal menopausal transition. Maturitas
10. Grisso JA, Freeman E, Maurin E, Garcia-Espana B, Berlin JA. Racial differences in menopause information and the experience of hot flashes in the late reproductive years. J Gen Int Med 1999; 14:98–103.
11. Garrow JS, Webster J. Quetelet's Index (W/H2
) as a measure of fatness. Int J Obes 1985; 9:147–53.
12. Radloff LS. The CES-D scale: a self-report depression scale for research in the general population. Appl Psychol Measurement 1977; 1:385–401.
13. Myers JK, Weissman MM. Use of a self-report symptom scale to detect depression in a community sample. Am J Psychiatry 1980; 137:1081–4.
14. Zung WWK. A rating instrument for anxiety disorders. Psychosomatics 1971; 12:371–9.
15. Cohen S, Kamarck T, Mermelstein R. A global measure of perceived stress. J Health Soc Behav 1983; 24:385–96.
16. Ware JE, Kosinski M, Keller SD. A 12-item short-form health survey: construction of scales and preliminary tests of reliability and validity. Med Care 1996; 34:220–33.
17. McHorney CA, Ware JE, Raczek AE. The MOS 36-item short-form health survey (SF-36) 0.2: psychometric and clinical-tests of validity in measuring physical and mental-health constructs. Med Care 1993; 31:247–63.
18. Endicott J, Nee J, Harrison W, Blumenthal R. Quality-of-life enjoyment and satisfaction questionnaire: a new measure. Psychopharmacol Bull 1993; 29:321–6.
19. Pham K-T, Grisso JA, Freeman EW. Ovarian aging and hormone replacement therapy: hormonal levels, symptoms, and attitudes of African-American and white women. J Gen Intern Med 1997: 12: 230–6.
20. Collins A, Landgren B-M. Reproductive health, use of estrogen and experience of symptoms in perimenopausal women: a population-based study. Maturitas 1995; 20:101–11.
21. Hunter M, Battersby R, Whitehead M. Relationships between psychological symptoms, somatic complaints and menopausal status. Maturitas 1986; 8:217–28.
22. SAS/STAT User's Guide, Version 6
, 4th ed. Cary, NC: SAS Institute, Inc., 1989.
23. Kahn SA, Pace JE, Cox ML, et al. Climacteric symptoms in healthy middle-aged women. Br J Clin Pract 1994; 48:240–2.
24. Wilbur J, Miller AM, Montgomery A, Chandler P. Sociodemographic characteristics, biological factors, and symptom reporting in midlife women. Menopause 1998; 5:43–51.
25. Schwingl PJ, Hulka BS, Harlow SD. Risk factors for menopausal hot flashes. Obstet Gynecol 1994; 84:29–34.
26. DeAngelis T. Menopause symptoms vary among ethnic groups. APA Monitor
[serial online]. Available at http://www.apa.org.
27. Langenberg P, Kjerulff KH, Stolley PD. Hormone replacement and menopausal symptoms following hysterectomy. Am J Epidemiol 1997: 146:870–80.
28. Bromberger JT, Matthews KA, Kuller LH, Wing RR, Meilahn EN, Plantinga P. Prospective study of the determinants of age at menopause. Am J Epidemiol 1997; 145:124–33.
29. Spitzer RL, Williams JBW, Kroenke K, et al. Utility of a new procedure for diagnosing mental disorders in primary care: The PRIME-MD 1000 study. JAMA 1994; 272:1749–56.
30. Morrison MF, TenHave T, Freeman EW, et al. The relationship of DHEA-S levels and depressive symptoms in a cohort of African American and Caucasian women in the late reproductive years. Submitted for publication.
31. Frasure-Smith N, Lesperance F, Talajic M. Depression following myocardial infarction: impact on 6-month survival. JAMA 1993; 270:1819–25.
32. Shaw CM, Creed F, Tomenson B, Riste L, Cruickshank JK. Prevalence of anxiety and depressive illness and help seeking behaviour in African Caribbeans and white Europeans: two phase general population survey. BMJ 1999; 318:302–5.
33. Lenton EA, Sexton L, Lee S, Cooke ID. Progressive changes in LH and FSH and LH: FSH ratio in women throughout reproductive life. Maturitas 1988; 10:35–43.
34. Treloar AE. Menstrual cyclicity and the pre-menopause. Maturitas 1981; 3:249–64.
35. Hunninghake DB, Darby CA, Probstfield JL. Recruitment experience in clinical trials: literature summary and annotated bibliography. Control Clin Trials 1987; 8:6S–30S.
36. Bjornson-Benson WM, Stilbolt TB, Manske KA, Zavela KJ, Youtsey DJ, Buist AS. Monitoring recruitment effectiveness and cost in a clinical trial. Control Clin Trials 1993; 14(Suppl 2):52S–67S.