Van Voorhis, Bradley J. MD1; Santoro, Nanette MD2; Harlow, Sioban PhD3; Crawford, Sybil L. PhD4; Randolph, John MD3
Click on the links below to access all the Data Supplements for this article.
Please note that Data Supplement files may launch a viewer application outside of your web browser.
Changes in both timing and amount of menstrual bleeding are common in the menopausal transition, but the underlying mechanisms are incompletely understood. In a population-based survey of Australian women between the ages of 45 and 55, women were asked to compare their menstrual periods in the preceding 3 months with periods occurring 12 months before the survey. Thirty percent reported no change, 10% reported a change in bleeding frequency but not in amount, 22% reported a change in amount but not in frequency, 26% reported a change in both frequency and amount, and 12% reported 3 months of amenorrhea.1 Several longitudinal studies have demonstrated that menstrual cycles become more irregular in the years before the final menstrual period.2–4 In contrast to the typically regular cycles experienced by younger women, older premenopausal women often have shorter cycle intervals early in the transition and then longer menstrual cycle intervals later in the transition; this cycle irregularity defines the onset of the menopause transition.5
Changes in the timing of menstrual bleeding, especially longer cycle intervals, are commonly thought to be secondary to anovulation, which becomes more frequent as women age. We hypothesized that changes in amount of bleeding and, particularly, heavy bleeding in the menopausal transition might also have a hormonal basis. A study of daily hormone dynamics found that, compared with younger cycling women, 11 perimenopausal women (age 43–52) had increased urinary excretion of follicle-stimulating hormone (FSH) and estrone conjugates with decreased excretion of progesterone.6 Relative hyperestrogenism, if commonly present in the menopause transition, might contribute to gynecologic morbidity, including heavy bleeding.
The Study of Women’s Health Across the Nation (SWAN) is a multisite, multiethnic longitudinal study of midlife women. Among the goals of SWAN is the characterization of the reproductive hormone patterns that occur in women as they approach and traverse the menopausal transition and correlation of these hormonal changes with menstrual bleeding patterns in these women. By associating hormonal changes with menstrual bleeding patterns, two objectives might be met. First, early hormonal predictors of menopause and the stages of the menopausal transition might be discovered. This might, in turn, justify the clinical measurement of hormone levels in women in the menopausal transition for diagnostic or prognostic purposes. Secondly, the hormonal basis behind the bleeding abnormalities that are so common in the menopausal transition will be better understood.
MATERIALS AND METHODS
SWAN is a cohort study of 3,302 middle-aged women enrolled at seven sites throughout the United States. The design of the main cohort study has been reported previously.7 All women participating in SWAN completed interviewer-administered questionnaires detailing their complete medical history, including prior diagnoses of uterine fibroids, thyroid disorders, and diabetes. For fibroid determination, women were asked “Has a doctor, nurse practitioner or other health care provider ever told you that you had fibroids, benign growths of the uterus or womb?” The Daily Hormone Study (DHS) is a substudy of SWAN in which a subset of women (n=804) collected first morning voided urine samples daily for one complete menstrual cycle or 50 days (whichever came first) once a year. Details on specimen collection have been published previously.8 In the present communication, we report findings from the urine samples collected in the first 3 years of the study. We measured the excreted levels of FSH, luteinizing hormone (LH), estrone conjugates (E1c), and pregnanediol glucuronide (Pdg). We then used these measurements in validated algorithms9 to evaluate the menstrual cycles for features consistent with ovulation and corpus luteum function.
A subset of women from all SWAN clinical sites was enrolled in the DHS; recruitment details have been previously published.6 This study was approved by all of the sites’ institutional review boards, and written informed consent was obtained from each participant. The baseline cohort for SWAN included women of Caucasian (n=1,550), African (n=935), Chinese (n=250), Japanese (n=281), and Hispanic (n=286) ethnic origins who were aged 42–52 years. Inclusion criteria for recruitment into the DHS were: 1) an intact uterus and at least one ovary, 2) at least one menstrual period in the previous 3 months, 3) no use of sex steroid hormones in the previous 3 months, and 4) not pregnant. Women eligible for the DHS were categorized in terms of menopausal status at the onset of the study as well as at each yearly visit before the onset of daily urine collections. Premenopausal status was defined as menses in the past 3 months with no change over the past year in predictability of menstrual periods. Early perimenopausal status was defined as menses in the past 3 months with less predictable periods.1,10 Late perimenopause was defined as two or more skipped cycles and an interval of amenorrhea of 90 days or more.
Luteinizing hormone, FSH, E1c, and Pdg were measured using newly adapted chemiluminescent assays, previously described.8 Data were normalized for creatinine concentration.11 Total-cycle integrated hormone concentrations also were analyzed. Technically, hormone excretion in the urine is being measured. However, this has been shown to correlate closely with hormone production,12 so, for the purposes of clarity, these measures will be considered to reflect LH, FSH, estradiol, and progesterone production in this article.
A significant increase in Pdg concentrations was accepted as evidence of luteal activity, which is consistent with presumed ovulation by a validated algorithm.13 The algorithm locates the 5 nadir days of Pdg in the follicular phase using moving averages throughout the cycle. A threefold increase in Pdg concentrations above this nadir for at least 3 consecutive days was considered evidence of ovulation. Cycles with no evidence of ovulation were further subdivided into those in which the collection ended because of the onset of a bleeding episode or those in which the collection was automatically terminated at 50 days without bleeding. For each of the four hormones, mean daily hormone concentration within a collection was computed as the sum over the collection of a woman’s observed daily hormone divided by the total number of daily observations in the collection.
Cycle interval and days of bleeding were derived from women’s menstrual calendars, which were kept concurrently with urine collection and subsequently. Women were asked to record menstrual bleeding as it occurred, distinguishing spotting, light/moderate bleeding, and very heavy bleeding. Normal menstrual cycle characteristics were defined by values in the literature14 and consisted of cycle intervals of 21–35 days and duration of bleeding of 4–7 days. Hormone production was correlated with menstrual bleeding characteristics of the period immediately after the daily urine collections. Cycles that did not terminate in bleeding within 7 days of the end of daily urine collections were excluded from analysis because of the belief that associations between daily urinary hormones and subsequent bleeding would be unclear in such cases.
Baseline characteristics of the samples were summarized using frequencies for categorical variables and medians and minimum/maximum for continuous variables. Frequency distributions of cycle interval and bleeding characteristics were estimated separately for each annual visit. Longitudinal correlates of cycle interval and bleeding characteristics were identified using separate random effects logistic regression models for each outcome15 to account for within-woman correlation over time. Because proportional odds logistic regression models did not provide an adequate fit to the data—that is, the χ2 test for proportional odds indicated that this assumption was violated—each of the outcomes was modeled using two separate binomial logistic regressions, comparing a reference category to each of the other categories. For example, predictors of short cycles versus normal-length cycles and long cycles versus normal-length cycles were identified in two separate binomial logistic regressions. Predictors included evidence of luteal activity, baseline categorized body mass index (BMI), overactive or underactive thyroid at baseline, self-reported fibroids at baseline, diabetes, age, menopause status, mean daily hormone concentrations (adjusted for creatinine), baseline smoking, ethnicity, and region (Eastern, Midwest, and West Coast United States, included in models using two dummy variables). Backward elimination was used to omit redundant or irrelevant covariates from multivariate models; the resulting statistically significant predictors were similar to those from models including all candidate predictors (results not shown). To handle right-skewness, hormone variables were log-transformed; for ease of interpretation, odds ratios from logistic regressions are presented in terms of comparing the 75th percentile to the 25th percentile, as a 1-unit change in the log-transformed versions would be difficult to interpret. For time-varying predictors (evidence of luteal activity, self-reported fibroids, age, menopause status, and daily hormones), models were adjusted for both baseline values and change in the predictor since baseline (eg, baseline age and time on study).
Of 848 cycles collected at the baseline visit, eight cycles started with a surge of LH and an increase in Pdg, suggesting that periovulatory bleeding occurred and cued women to begin their urine collection at the wrong time; these cycles were not included in subsequent analyses. An additional 36 baseline observations were missing covariate data and also were excluded, yielding an analytic sample of 804 women. We studied an ethnically diverse population of women with a median age of 47 years at the start of the study (range 43.1–54.1 years Table 1). At baseline, a large majority (73%) of these women were in the early perimenopause, characterized by experiencing recent variable cycle lengths. Most women were not smokers, and 28% of women were obese. Twenty-one percent reported that they had ever been diagnosed with fibroids. Because there were study dropouts, characteristics of women participating at visits 2 and 3 also are reported (Table 1).
The most common cycle characteristics included a cycle interval of 21–35 days, a menstrual bleeding duration of 4–7 days, and no days characterized as heavy flow days (Table 2). There was little change in the prevalence of the various cycle characteristics over this 3-year time frame. Twenty percent of all cycles were anovulatory in this population of women. Anovulatory cycles were common in women with short cycle intervals and even more common in women with long cycle intervals (Table 2).
As compared with women with a normal cycle interval, short cycle intervals were more common in older women and as women transitioned into the early perimenopause (Table 3). Anovulation in the concurrent cycle was a risk factor for having a short interval regardless of ovulatory status at baseline. Women with short cycle intervals had lower daily production of FSH and higher daily production of E1c.
Long cycle intervals were associated with anovulation and transition to the perimenopause (Table 3). Longer cycle intervals were seen more commonly later in the transition and were associated with greater production of FSH, lower production of progesterone, and the presence of diabetes (Table 3).
Both short (1–3 days) and long (8 or more days) durations of bleeding in the subsequent menstrual period were associated with anovulation as compared with periods with a normal (4–7 days) duration of menstrual bleeding (Table 4). Women with fibroids were also more likely to have a long menstrual period. After correction for anovulation, daily and integrated hormone production had no independent effect on the number of days of bleeding during subsequent menses.
Women with heavy menstrual flow in the current cycle were less likely to have had an anovulatory cycle regardless of whether they had an ovulatory or anovulatory cycle at the baseline study cycle (Table 5). Other factors associated with heavy periods were the reported presence of uterine fibroids and a BMI of 30 and above. The only hormonal association with heavy menses was elevated production of FSH at the baseline cycle. Conversely, in the concurrent cycle, reduced production of FSH was associated with 1–2 days of heavy bleeding. No other hormonal measures, either daily or integrated, predicted heavy bleeding.
Factors that had no independent effect on menstrual cycle characteristics included ethnicity, smoking history, and medical conditions including thyroid disorders and diabetes.
Menstrual bleeding irregularities are a hallmark of the menopausal transition but frequently result in women seeking gynecologic consultation.16 Although an overwhelmingly probable marker of the normal transition, bleeding irregularity raises concerns about possible pregnancy, endometrial hyperplasia, or gynecologic malignancy and leads to diagnostic and therapeutic interventions including hysterectomy. Hysterectomies are more common as women approach the menopausal transition, with a median age of hysterectomy being 45 in the United States.17 The major indication for hysterectomy is uterine fibroids, although at least 10% of hysterectomies are done for menstrual bleeding disorders as the primary diagnosis.
We have confirmed that anovulation is highly associated with variability in the timing of menstrual bleeding. Fully 20% of cycles were found to be anovulatory in this group of women, predominantly in the early perimenopause. Anovulation was associated with both short and long cycle intervals as well as with both short and long duration of the ensuing menstrual period. Short cycle intervals are seen more often early in the menopausal transition. Thus, short intervals may be the first sign of a woman entering the transition; longer cycle intervals are seen later and are associated with higher FSH production, indicating even fewer functioning ovarian follicles. The presence of fibroids also is associated with a greater chance of having a longer duration of bleeding.
In contrast to timing issues, an increased amount of bleeding is not positively associated with anovulation. In fact, anovulatory cycles were more often associated with lighter bleeding in the following menstrual period. Heavy bleeding was associated with obesity and with the self-reported presence of uterine fibroids. In contrast to our initial hypothesis, we could find no association between heavy bleeding and either estrogen (E1c) or progesterone (Pdg) production when evaluated as mean daily production or cycle-integrated production or when expressed as ratios of E1c to Pdg production.
Given the higher prevalence and symptomatology of fibroids among African-Americans, we expected to see ethnic differences in reports of heavy bleeding. In fact, there were statistically significant unadjusted ethnic differences of 3 or more heavy bleeding days versus no heavy bleeding, with African-American and Hispanic women more likely and Chinese and Japanese women less likely to have this complaint compared with Caucasians (data not shown). Adjustment for prevalent fibroids reduced the ethnic differences slightly, but they remained statistically significant. On further adjustment for BMI, ethnic differences were no longer statistically significant, suggesting that any perceived ethnic differences in reported heavy bleeding are accounted for primarily by BMI differences.
Our findings would suggest that, if a clinician encounters a woman in the early menopausal transition with abnormal timing of menstrual bleeding (short or long cycle intervals, short or long duration of menstrual bleeding), the first thought should be anovulation as the cause. In contrast, if the complaint is only heavy bleeding, anovulation is less likely and careful evaluation for structural lesions including polyps and fibroids is warranted. The association of heavy bleeding with obesity is interesting and also concerning given the epidemic of obesity in the United States.
Strengths of this study include the large and ethnically diverse population being studied. In addition, this is a large prospective study evaluating daily hormone production over an entire menstrual cycle. Hence, we can accurately correlate hormone production with menstrual bleeding patterns.
Weaknesses of the study lie primarily in the determination of heavy bleeding, which was by self-report. Several studies have shown a lack of correlation between the subjective complaint of heavy bleeding and objectively measured blood loss. However, more recent reports have found that, depending on the types of questions asked, there is a better correlation between a woman’s report of heavy bleeding and measured blood loss.18,19 We simply asked women to record days when they experienced “very heavy bleeding.” From a practical standpoint, objective measures of actual blood loss during menses are not performed clinically, so our reliance on a woman’s self-report may mimic clinical practice. Another weakness was the reliance on self-report for uterine fibroids. Because fibroids are known to be somewhat more common than the 21% rate reported in our study, it is likely that some were missed and only the larger and more clinically significant fibroids were discovered and reported by the women. In addition, it is known that polyps can lead to abnormal bleeding, and no systematic means of detecting these lesions was performed.
This large study of hormonal production in the early menopause transition confirms a hormonal basis for changes in menstrual cyclicity and timing. Shortened menstrual cycle intervals are the earliest change in the transition, and these cycles are often anovulatory. Longer menstrual cycle intervals are more common later in the menopause transition and are also associated with anovulation. Both longer and shorter duration of bleeding in a menstrual period are also associated with anovulation. Contrary to our original hypothesis, we found no evidence of a hormonal basis for heavy bleeding in the early menopause transition. Therefore, evaluation of estrogen, progesterone, LH, or FSH levels in women with this complaint is not warranted. Instead, attention should be directed toward obesity or lesions, including polyps and fibroids, as possible causes.
1. Dudley EC, Hopper JL, Taffe J, Guthrie JR, Burger HG, Dennerstein L. Using longitudinal data to define the perimenopause by menstrual cycle characteristics. Climacteric 1998;1:18–25.
2. Treloar AE, Boynton RE, Behn BG, Brown BW. Variation of the human menstrual cycle through reproductive life. Int J Fertil 1967;12:77–126.
3. Treloar AE. Menstrual cyclicity and the pre-menopause. Maturitas 1981;3:249–64.
4. Taffe JR, Dennerstein L. Menstrual patterns leading to the final menstrual period. Menopause 2002;9:32–40.
5. Soules MR, Sherman S, Parrott E, Rebar R, Santoro N, Utian W, et al. Executive summary: Stages of Reproductive Aging Workshop (STRAW). Fertil Steril 2001;76:874–8.
6. Santoro N, Brown JR, Adel T, Skurnick JH. Characterization of Reproductive Hormonal Dynamics in the Perimenopause. J Clin Endocrinol Metab 1996;81:1495–1501.
7. Sowers MF, Crawford SL, 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, Menopause: biology and pathobiology. New York (NY): Academic Press; 2000. 175–88.
8. Santoro N, Crawford SL, Allsworth JE, Gold EB, Greendale GA, Korenman S, et al. Assessing menstrual cycles using urinary hormone assays. Am J Physiol Endocrinol Metab 2003;284:E521–30.
9. Prior JC. Perimenopause: the complex endocrinology of the menopausal transition. Endocr Rev 1998;19:397–428.
10. Brambilla DJ, McKinlay SM, Johannes CB. Defining the perimenopause for application in epidemiologic investigations. Am J Epidemiol 1994;140:1091–5.
11. Taussky HH. A microcalorimetric determination of creatinine in urine by the Jaffe reaction. J Biol Chem 1954;208:835–61.
12. Waller K, Swan SH, Windham GC, Fenster L, Elkin EP, Lasley BL. Use of urine biomarkers to evaluate menstrual function in healthy premenopausal women. Am J Epidemiol 1998;147:1071–80.
13. Kassam A, Overstreet JW, Snow-Harter C, De Souza MJ, Gold EB, Lasley BL. Identification of anovulation and transient luteal function using a urinary pregnanediol-3-glucuronide ratio algorithm. Environ Health Perspect 1996;104:408–13.
14. Wood C, Larsen L, Williams R. Menstrual characteristics of 2,343 women attending the Shepard Foundation. Aust NZJ Obstet Gynaec 1979;19:107–10.
15. Molenberghs G, Verbeke G. Models for discrete longitudinal data. New York (NY): Springer; 2005.
16. Santoro N. The menopause transition: an update. Hum Reprod Update 2002;8:155–60.
17. Farquhar CM, Steiner CA. Hysterectomy rates in the United States 1990–1997. Obstet Gynecol 2002;99:229–34.
18. Warner PE, Critchley HO, Lumsden MA, Campbell-Brown M, Douglas A, Murray GD. Menorrhagia I: measured blood loss, clinical features, and outcome in women with heavy periods: a survey with follow-up data. Am J Obstet Gynecol 2004;190 (5):1216–23.
19. Warner PE, Critchley HO, Lumsden MA, Campbell-Brown M, Douglas A, Murray GD. Menorrhagia II: is the 80-mL blood loss criterion useful in management of complaint of menorrhagia? Am J Obstet Gynecol 2004;190:1224–9.
Figure. No caption available.