Menstrual Cycle Phase, Menopausal Transition Stage, Self-Reports of PMS and Symptom Severity: Observations from the Seattle Midlife Women’s Health Study

Abstract

Objective:

The aim of this study was to determine the effects of menstrual cycle phases (postmenses and premenses), self-report of PMS, Late Reproductive Stages (LRS1 and LRS2) and Early Menopausal Transition (EMT) Stage (STRAW) on severity of 5 symptom groups.

Methods:

A subset of Seattle Midlife Women’s Health Study participants (n = 290) in either LRS1 or LRS2 or EMT (STRAW+10 criteria), provided daily symptom data for at least one full menstrual cycle during the first year of the study plus reported current PMS. Symptom severity was rated (1–4 least to most severe) in the daily diary for 5 symptom groups (dysphoric mood, neuromuscular, somatic, vasomotor and insomnia) identified earlier with the same sample (Mitchell, 1996). A 3 Way ANOVA was used to test for within and between participants effects on symptom severity.

Results:

Stage had no effect on severity for any of the 5 symptom groups. Dysphoric mood, neuromuscular and somatic symptom severity (but not vasomotor or insomnia severity) differed significantly across menstrual cycle phases, increasing from postmenses to premenses. Current PMS and premenses cycle phase had significant interactive effects on dysphoric mood and neuromuscular symptoms, but there were no significant interaction effects on somatic, vasomotor or insomnia symptom severity.

Conclusions:

Dysphoric mood, neuromuscular and somatic symptoms exhibit cyclicity and are influenced by current PMS. LRS and EMT stages do not have significant effects on the 5 symptom groups. Vasomotor or insomnia symptoms do not exhibit significant cyclicity from postmenses to premenses and are not affected by current PMS. Future study of symptom cyclicity and reproductive aging including daily symptom data across an entire menstrual cycle in samples including women in Late Menopausal Transition Stage are essential to capture the effects of both cyclicity, self-reported PMS to capture symptom severity reports at their peak.

Little is known about women’s symptom experiences as they approach the menopausal transition stages, particularly during the late reproductive stage (LRS) of reproductive aging (1). As defined by the Staging Reproductive Aging Workshop (STRAW+10 framework) (2,3) the late reproductive stage can be divided into substages or phases labelled −3b (the first phase of LRS) and −3a (the second phase of LRS). For the sake of clarity, in this paper these substages will be referred to as LRS1 (−3b) and LRS2 (−3a) to reflect progress toward FMP. During LRS1 women continue menstruating regularly without change in cycle length (3). During LRS2 women also continue menstruating regularly but experience shortening or lengthening of their menstrual cycles by a few days with subtle changes in the amount of menstrual flow, and/or in the number of days of flow (3,4). After LRS2 women enter the Early Menopausal Transition (EMT) when irregular cycles begin with increased variability in cycle length. Irregular cycles, as defined by Mitchell (4) and validated by ReSTAGE (5), is persistent variability of 7 days or more between two consecutive menstrual cycles repeated within 12 months (4) or within the next 10 cycles (5).

Evidence from a recent survey suggests that during the late reproductive stage women begin to notice symptoms often associated with the menopausal transition, such as hot flashes, sleep changes, and mood symptoms (6). Women Living Better Survey participants reported vasomotor symptoms of hot flashes and night sweats prior to experiencing irregular cycles. The prevalence of these symptoms was higher during the menopausal transition stages than during the late reproductive stage (6). In addition, many of the same symptoms reported during the LRS and EMT occurred during the premenses and menses phases of the menstrual cycle among participants in other studies of the menopausal transition (7–8).

Although women continue to menstruate during the LRS and EMT, little is known about cycle phase differences in symptom severity when comparing postmenses (follicular phase) with premenses (late luteal phase) during these stages. To date there are no reports of symptom severity patterns from longitudinal studies of women studied daily during LRS and EMT during an entire menstrual cycle. Consequently, it is unclear whether women have cyclic symptoms in association with phases of the menstrual cycle during LRS or EMT and if symptom severity differs between menstrual cycle phases.

Menstrual cycle phase difference (CPD) in symptoms is denoted by an increase in symptom severity from the follicular or postmenses phase to the late luteal or premenses phase with a decrease after menses. Many different labels have been used to describe this type of symptom pattern such as molimina (9), perimenstrual symptoms (10), premenstrual syndrome or PMS (10,11), premenstrual magnification or PMM (10), and premenstrual dysphoric disorder or PMDD (11), each with different criteria.

Few studies of cyclic symptoms in midlife women, especially those in LRS or EMT, have been reported. Woods and colleagues found that women who were screened for PMS symptom patterns before they reached 40 years of age continued to experience the PMS symptom pattern documented in a daily symptom diary after they reached 40 years of age (12). Freeman and colleagues with the Penn Ovarian Aging Study (POAS) found that fewer women reported the PMS symptom pattern after they reached 40 years of age, although these women continued to report PMS symptoms including hot flashes, depressed mood, decreased libido, and poor sleep during the menopausal transition stages (13–16). Dennerstein also identified an association between women’s prior history of premenstrual complaints and negative mood symptoms experienced over 6 years of follow-up during the menopausal transition by the Melbourne Midlife Women’s Health Project (MMWHP) participants (17).

Investigators for the POAS and the MMWHP relied on women’s retrospective reports of PMS prior to the onset of the menopausal transition. For example, Freeman assessed PMS with 2 questions inquiring about irritability, mood swings, and emotional distress and the degree to which they interfered with daily activities (11). Dennerstein (17) assessed premenstrual complaints by inquiring about physical or psychological changes women considered problematic, causing personal distress, and affecting work or relationships. The cyclic nature of these symptoms was not confirmed with daily prospective measures of postmenses and premenses symptom severity in these studies.

To date, there are no published reports of cycle phase differences in symptom severity during the LRS and EMT using prospective symptom recordings. In addition, there are no published reports comparing self-reports of current PMS by midlife women with prospective reports of cyclic symptoms.

Prospective data provided by women during LRS and EMT could illuminate the relationship between symptoms that vary by menstrual cycle phase and by Reproductive Aging Stages. Our goal was to enhance understanding of this period of a woman’s reproductive aging by considering the effects of the menstrual cycle, self-reports of current PMS and menopausal transition stages on severity of five symptom groups commonly reported by midlife women (18).

The research questions were:

1. Does symptom severity of each symptom group (dysphoric mood, vasomotor, somatic, neuromuscular, insomnia) differ when accounting for stages of reproductive aging (late reproductive and early menopausal transition stages (LRS1, LRS2, and EMT), menstrual cycle phase (follicular/postmenses and late luteal/premenses menstrual cycle phases, and self-report of current PMS?

2. Is there an interactive effect between reproductive aging stage (LRS1, LRS2, EMT), menstrual cycle phase (postmenses, premenses), and self-report of current PMS on symptom severity for each symptom group?

Methods

Design and Sampling

Data reported here were collected from participants in the population-based Seattle Midlife Women’s Health Study of the menopausal transition and early postmenopause described in detail elsewhere (19). Women (N=508) entered the study from early 1990 through 1992. During their first year of participation in the study women kept a daily health diary for 45–80 days or 1–2 cycles. Eligibility requirements for the parent study were 35–55 years of age, not pregnant, intact uterus and at least one ovary, at least one menstrual period in the past 12 months, and ability to read and understand English. For these analyses, of the 508 women in the parent study, 366 kept a daily health diary for 45–80 days or 1–2 consecutive menstrual cycles and 305 of these women met the criteria during the first 12 months of the study as either in the Late Reproductive Stage, referred to here as LRS1 (STRAW −3b) and LRS2 (STRAW −3a) or the Early Menopausal Transition Stage (EMT). The sampling for this study excluded 51 women who were either in Late Menopausal Transition Stage (17), were postmenopausal (3) or taking hormones such as estrogen and/or progestin (31). An additional 14 women were excluded due to either no menstrual calendar for staging (5), an unreadable calendar (5), or diary data without both cycle 1 postmenses and premenses symptom data (4). Finally 11 women had no data about current PMS status. Thus, a total of 290 women were eligible for these analyses and kept the diary for at least I cycle in year 1 of the SMWHS.

Procedure

Data from the first year of the parent study were utilized in the analyses reported here. The source of these data were daily menstrual calendars, daily diary recordings for 1 to 2 menstrual cycles and a health questionnaire.

Women eligible to participate provided informed consent and were instructed by a trained Registered Nurse interviewer about how to keep the health diary and menstrual calendar. Women completed the diary data at the end of each day and rated symptoms for the past 24 hours. Women were contacted by phone after 2 weeks to answer any questions about data collection. A blank menstrual calendar was mailed at the end of each calendar year for completion during the following year. Any occurrence of bleeding (B) or spotting (S) was recorded. The calendars were returned at the start of the following year and reviewed for completeness. The health questionnaire for year 1 was administered in-person by a trained Registered Nurse interviewer.

Measures

Data Selection.

Symptom severity data for 5 postmenses days (+6 thru +10) and 5 premenses days (−1 thru −5) for the first cycle of data during year 1 for women in LRS1 or LRS2 or EMT were selected for these analyses. Women with fewer than 2 postmenses or 2 premenses days in cycle 1 were excluded from the analyses.

Symptom Selection.

Five symptom groups consisting of 28 symptoms formed the basis of these analyses. Symptoms were rated daily in the diary from 0 to 4 where 0 indicated not at all and 4 indicated extremely severe. The source of the five symptom groups was a principal components analysis with varimax rotation reported in 1996 on this same sample using the same data as described in detail elsewhere (18). At that time 61 symptoms commonly associated in the literature with midlife and the menopause, were analyzed for the 290 women. Symptoms within each factor that loaded at 0.4 or higher were selected as one of the 28 symptoms. The 5 factors were dysphoric mood, vasomotor symptoms, somatic symptoms, neuromuscular symptoms, and insomnia. The dysphoric mood factor included nervousness, panic feelings, irritability, mood swings, depressed mood, difficulty concentrating, and forgetfulness. The vasomotor factor included night sweats, hot flashes, trouble breathing, and palpitations. The somatic factor included nausea, headache, dizziness, and painful breasts. The neuromuscular factor included backache, numbness and tingling, joint pain and skin crawling. Insomnia symptoms included difficulty getting to sleep and awakening at night.

Reproductive Aging Stages.

Stages of Reproductive Aging Workshop (STRAW) criteria for Late Reproductive and Early Menopausal Transition Stages (3, 5) were applied to the first 12 months of the menstrual calendar data and to responses to questions, devised by Mitchell, in the initial questionnaire about menstrual cycle changes (4). The Late Reproductive Stage was divided into two substages: LRS1 and LRS2 (STRAW Stages −3b and −3a respectively). LRS1 was characterized by regularly occurring menstrual cycles (variation between consecutive menstrual cycles of fewer than 7 days for 12 consecutive months) documented in daily menstrual calendars women. LRS2 was also characterized by regularly occurring menstrual cycles. In addition, in the questionnaire, women in LRS2 noted subtle changes in menstrual cycle length (increases or decreases), a change in number of days of flow (more or fewer), and/or a change in amount of flow (more or less) (4). The early menopausal transition stage (EMT), identified from the first 12 months of menstrual calendar data, was defined as irregular menstrual cycles with a differences of >/= 7 days in consecutive cycles with no skipped periods (2–5). Women were also asked about their current experiences of PMS in the initial questionnaire. They responded Yes or No to the question: Do you think you have PMS (Premenstrual Syndrome)? The term PMS (Premenstrual Syndrome) was used in the questionnaire to reflect the label in current use at that time by the public for a group of symptoms that consistently occurred together that were low severity in the postmenses and increased premenses.

Analysis

Mean severity level for each of the 28 symptoms within the 5 factors (called symptom groups) for each cycle phase was calculated for each woman. Mean scores were then obtained for each of the 5 factors and used in subsequent analyses. A 3-way mixed ANOVA was used to determine the effects of reproductive aging stage (LR1, LR2, and ET), menstrual cycle phase (postmenses, premenses), and self-reported current PMS (yes, no) on each of the 5 symptom groups (20). A Simple Effects test was done for each significant interaction in the 3-Way Mixed ANOVA (21).

Results

Sample Characteristics

A total of 290 women completed diaries providing data for this analysis. Fifty-eight women were classified as LR1, 169 as LR2 and 63 as EMT. Of the 290 women, 161 self-reported current PMS and 129 self-reported No PMS. All 290 women provided data for the postmenses and premenses of the same menstrual cycle.

The 290 eligible women had an average age of 41 years (SD=4.0, range 35–54 years). A majority was partnered (70%), employed (86%), and White (79%). The sample included African-Amercians (8.6%), Asian/Pacific Islanders (9.3%), and Hispanic, American Indian and mixed race (2.8%), representative of the underlying population. Participants were well educated with a mean of 15.8 years of education (SD=2.7) and middle income levels (mean of $37,200, SD=$15,360) in 1990–1992 at the start of the study.

Reproductive aging stage, cycle Phase, and Current PMS Effects on Symptoms

Each of the five symptom groups was analyzed with a 3-way Mixed ANOVA for the effects of reproductive aging stage (LR1, LR2 and EMT), cycle phase (postmenses and premenses) and current PMS status (Yes or No). The main effect of reproductive aging stage on dysphoric mood was significant (F (2,284)=3.09, p=.047) with greater Early MT stage severity (M=.42, SD=.59) than that for LR1 (M=.32, SD=.33) or LR2 (M=.30, SD=.34) (See Table 1). However, when a post hoc pairwise comparison was done using Tukey HSD there were no significant stage effects, most likely due to a barely significant main effect. In addition, the main effect of cycle phase on dysphoric mood was significant (F(1,284)=22.08), p=<.001) (See Figure 1), as well as current PMS status (F(1,284)=5.23), p=.02). However, these main effects of cycle phase and PMS status were influenced by a significant interaction effect (F(1,284)=9.67, p=.002). A Simple Effects test showed that PMS status had a significant effect on premenses dysphoric mood (F(1,288)=11.7, p=.001) but not on postmenses dysphoric mood (F(1,288)=0.91, p=.34). Women who reported no PMS had a small increase in dysphoric mood symptoms premenses (M=.28, SD=.40) compared to postmenses (M=.26, SD=.38) while those who reported having current PMS had a significant increase between postmenses (M=.30, SD=.35) and premenses (M=.46, SD=.48) phases.

Table 1.

Severity Scores for Five Symptom Groups by Stage, Cycle Phase and PMS Status (M, SD)

			Severity Scores for Five Symptom Groups by Stage Cycle Phase and PMS Status (MSD)
			DYSPHORIC MOOD		NEUROMUSCULAR		SOMATIC		VASOMOTOR		INSOMNIA
			Posta	Preb	Post	Pre	Post	Pre	Post	Pre	Post	Pre
Stage	PMS	N	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)
LR1	No	26	.29 (.33)	.34 (.33)	.31 (.40)	.35 (.50)	.37 (.37)	.46 (.38)	.56 (.20)	.06 (.15)	.46 (.63)	.60 (.90)
	Yes	32	.27 (.35)	.38 (.32)	.31 (.40)	.33 (.28)	.45 (.39)	.52 (.41)	.09 (.19)	.13 (.25)	.28 (.42)	.33 (.46)
LR2	No	69	.22 (.31)	.22 (.28)	.26 (.32)	.29 (.32)	.35 (.30)	.37 (.32)	.06 (.15)	.07 (.20)	.44 (.62)	.39 (.52)
	Yes	100	.27 (.29)	.43 (.41)	.28 (.37)	.41 (.48)	.42 (.35)	.54 (.48)	.09 (.28)	.15 (.34)	.45 (.55)	.58 (.62)
EMT	No	34	.30 (.52)	.34 (.59)	.36 (.64)	.34 (.62)	.34 (.37)	.39 (.35)	.18 (.39)	.21 (.60)	.47 (.78)	.53 (.85)
	Yes	29	.44 (.50)	.61 (.75)	.40 (.49)	.58 (.68)	.56 (.53)	.75 (.55)	.15 (.25)	.21 (.47)	.63 (.79)	.69 (.78)

^aPost=postmenses

^bPre=premenses M=Mean score SD=Standard deviation

Figure 1.

Dysphoric Mood Symptom Severity by Stage (Late Reprodutive Stages 1 and 2, and Early Menopausal Transition Stage) by Menstrual Cycle Phases (Postmenses, Premenses)

There was no significant main effect of PMS status (F(1,284)=1.5, p=.23) or of reproductive aging stage (F(2,284)=1.7, p=.19) on neuromuscular symptoms. The main effect of cycle phase on the neuromuscular symptom group was significant (F(1,284)=9.04, p=.003) with premenses severity (M=.37, SD=.48) higher than postmenses severity (M=.30, SD=.42) (See Figure 2). However, the main effect of cycle phase was influenced by a significant interaction effect between cycle phase and PMS status (F(1,284)=5.3, p=.02). The simple Effects test showed that PMS status had a significant effect on premenses neuromuscular symptoms (F(1,288)=3.8, p=.051) but no significant effect on postmenses neuromuscular symptoms (F(1,288)=0.04, p=.85). Women who reported no PMS had a small increase in neuromuscular symptoms premenses (M=.30, SD=.44) compared to postmenses (M=.31, SD=.45) while those who reported having PMS had a significant increase between postmenses (M=.31, SD=.40) and premenses (M=.42, SD=49).

Figure 2.

Neuromuscular Symptom Severity by Stage (Late Reprodutive Stages 1 and 2, and Early Menopausal Transition Stage) by Menstrual Cycle Phases (Postmenses, Premenses)

The main effect of cycle phase on the somatic symptom group was significant (F(1,284)=12.47, p=.000) with premenses severity (M=.49, SD=.44) higher than postmenses severity (M=.41, SD=.37) (See Figure 3). Also, the main effect of PMS status on somatic symptoms was significant (F(1,284)=11.76, p=.001): those with current PMS (M=.49, SD=.44) had higher severity premenses than postmenses (M=.41, SD=.37). There was no significant main effect for reproductive aging stage (F(2,284)=1.6, p=.20) and there were no significant interactions among the 3 independent variables.

Figure 3.

Somatic Symptom Severity by Stage (Late Reprodutive Stages 1 and 2, and Early Menopausal Transition Stage) by Menstrual Cycle Phases (Postmenses, Premenses)

The main effect of reproductive stage on the vasomotor symptom group was significant (F (2,284)=3.01, p=.051) with Early MT stage severity (M=.19, SD=.44) higher than LR1 (M=.09, SD=.20) or LR2 (M=.10, SD=.26). (See Figure 4). However, when a pairwise comparison was done using Tukey HSD there were no significant stage effects. This was most likely due to a barely significant main effect. There was no significant main effect for cycle phase (F(1,284)=3.54, p=.06) or for PMS status (F(1,284)=.51, p=.48) for the vasomotor symptom group. Also, there were no significant interactions among the 3 independent variables. Finally, for the insomnia symptom group there were no significant main effects (See Figure 5) or interactions of reproductive aging stage [main effects (F(2,284)=1.4, p=.25)], cycle phase [main effects (F(1,284)=2.63, p=.11)] or PMS status [main effects (F(1,284)=.02. p=.89)].

Figure 4.

Vasomotor Symptom Severity by Stage (Late Reprodutive Stages 1 and 2, and Early Menopausal Transition Stage) by Menstrual Cycle Phases (Postmenses, Premenses)

Figure 5.

Insomnia Symptom Severity by Stage (Late Reprodutive Stages 1 and 2, and Early Menopausal Transition Stage) by Menstrual Cycle Phases (Postmenses, Premenses)

Discussion

These findings contribute to our knowledge about the relationship between LRS and EMT and symptoms. These stages of reproductive aging had no overall effect on any of the 5 symptom groups in this sample of women. Instead, premenses cycle phase significantly contributed to dysphoric mood, somatic and neuromuscular symptom severity, although not to vasomotor or insomnia symptom severity. Self-reports of current PMS also significantly affected dysphoric mood and somatic symptoms but not neuromuscular, vasomotor or insomnia symptoms. Menstrual cycle phase and current PMS had interactive effects on dysphoric mood and neuromuscular symptom severity which suggests that these symptoms are related to current PMS and not to LRS or EMT stages. The results also indicate that vasomotor and insomnia symptoms are not related to either LRS or EMT stages, current PMS or to menstrual cycle phase in this sample of women.

These findings provide a unique opportunity to understand cycle phase differences in symptom severity within the progression of reproductive aging. To date most studies of symptoms during midlife do not include women in the LR stages (STRAW −3a and −3b) and only recently have investigators reported symptom data collected during the LR stage in comparison to later MT stages (1,6). The data presented here provide a first opportunity to assess menstrual cycle phase and stages of reproductive aging from LR to ET as they jointly influence symptoms women rated in a daily diary for a complete menstrual cycle. In addition, these analyses provide insight about women’s self-reports of PMS and their relationship to cycle phase differences in symptom severity.

Menstrual cycle phase differences in symptom severity may be influenced by chronological aging, progression through reproductive aging stages and different time trajectories of these changes in relation to menopause, and the hormonal changes that occur both between menstrual cycle phases and among reproductive aging stages. Cycle phase differences in symptom patterns have been linked to age in studies of midlife women. Prospective ratings of symptom severity across women’s menstrual cycles indicating a PMS symptom pattern persisted from measures obtained while they were younger (18 to 45, M=36, SD=5.1 years) after they reached age 40 years (median age was 44 years) (13). Freeman and colleagues found that PMS, measured by questions about women’s PMS symptom histories, was negatively associated with age, with a declining incidence reported among women 40 years and older in the Penn Ovarian Aging Study cohort (12). In a later report with additional longitudinal follow-up of the cohort, POAS investigators confirmed the decline in self-reported PMS as women aged: women 40–44 years had a 59% chance of experiencing PMS, those 45–49 a 42% chance, and those 50–55 a 20% chance (all p<.001) (8). In contrast, progressing through the stages of reproductive aging, approximated by the STRAW stages (2,3), compared to being classified as premenopausal, was not associated with reports of PMS, except for women in the late transition/early postmenopause whose chance of experiencing PMS was significantly reduced from that reported by younger women (p=.001) (8). The declining prevalence of PMS as women aged was corroborated by Dennerstein and colleagues. Studying reports from a wider age range of women, 7226 women aged 15–49 years from Europe, Central and South America, Asia, and Australia, Dennerstein and colleagues found that PMS reports increased with age until women reached approximately age 35 years, followed by a decrease in reports of PMS which investigators attributed to the decrease in ovulatory cycles as women grew older (22).

Effects of progression through the stages of reproductive aging on menstrual cycle phase differences in symptoms has had limited study. POAS participants’ reports of PMS during the late reproductive years were associated with reporting symptoms often associated with the menopausal transition and early postmenopause, including reports of hot flashes, depressed mood, poor sleep, and lower libido over the next 5 years (8, 23). Fluctuations of estradiol were associated with hot flushes, depressive symptoms, and poor sleep (15,16). In subsequent analyses of data from the POAS cohort, women who met the criteria for PMS were significantly more likely than women without PMS to report symptoms of headache, irritability, mood swings, anxiety, and difficulty concentrating at each reproductive aging stage (defined as premenopause, late premenopause, early transition, late transition, and postmenopause) (8,23). There was no interaction between PMS and reproductive aging stages on symptoms reported as moderate or severe. These findings are consistent with SMWHS findings of effects of self-reported PMS on dysphoric mood and somatic symptoms and the cycle phase effects reported in daily symptom diaries by the SMWHS cohort. In addition, the lack of effect of reproductive aging stages on these symptoms among SMWHS participants is consistent with data from the POAS cohort.

In addition to effects of menstrual cycle phase differences, women experience a variety of symptom trajectories as they progress through the stages of reproductive aging. Mishra (24) reported trajectories of vasomotor symptoms (hot flashes and night sweats) that varied with aging and progression to menopause. One group experienced severe symptoms early in the menopausal transition, e.g.1–4 years before menopause, and others experienced severe symptoms later, in years 1–4 postmenopause. Yet another group constantly experienced mild symptoms with a slight peak around menopause and another a constant experience of moderate severity symptoms with a peak 1–4 years postmenopause (24). Symptom classes identified by latent class analysis of data reported by SWAN participants revealed several classes indicating at least 6 different trajectories of symptoms over time (25). These data suggest that there is no single trajectory characterizing individual women’s symptom experience across stages of reproductive aging and raises the possibility that complex sets of physiological and psychosocial factors may be involved. Although there were no effects of the LR and ET stages on symptom severity reported here in SMWHS analyses, this does not preclude the MT stage effects when data from the Late Transition stage are included. Indeed, many of the symptoms reported here, including vasomotor symptoms, depressed mood, and night-time awakening reached peak severity during the Late Transition Stage (19). Data from studies such as the SWAN Daily Hormone Study that includes observations from women for up to 10 years may prove enlightening in future analyses examining both the menstrual cycle phase effects and reproductive aging stages, including the late menopausal transition stage.

Earlier studies of cyclic changes in estradiol and progesterone (26) indicated that these hormones contribute to variation in symptom severity both across the menstrual cycle phases and as women progress through stages of reproductive aging. Women who were cycling regularly experienced a low severity symptom pattern during both postmenses and premenses (LS), compared to women with a significant rise in symptom severity premenses (PMS pattern) and those with higher postmenses symptom severity and a significant rise in severity premenses (premenstrual magnification or PMM pattern), provided daily urine specimens analyzed for estrone glucuronide (E1G) and pregnanediol (Pdg), as well as cortisol, epinephrine, and norepinephrine across one entire menstrual cycle. Women in the PMS and PMM groups experienced a more gradual decline in E1G levels premenses than the LS group, but there were no differences between symptom groups (LS vs PMS, LS vs PMM) for the rate of the pregnanediol decline (26). When women from the LS and PMM groups were included in the analyses, women with the more gradual estradiol drop also had higher cortisol and epinephrine levels during the luteal phase. In addition, epinephrine levels during premenses were significantly positively and norepinephrine levels premenses were significantly negatively correlated with symptoms of emotional turmoil (e.g. hostility, depression, anger, feeling out of control, tension, feeling guilty, tearfulness, anxiety, mood swings, nervousness, irritability, desire to be alone, feeling lonely, and impatient), and fluid retention symptoms (e.g. abdominal bloating, painful breasts, sensation of weight gain, swelling of hands and feet, and skin disorders across all groups (LS, PMS, PMM) (27). These findings may help explain the relationship of endocrine patterns across the menstrual cycle to effects of menstrual cycle phases on some symptoms SMWHS participants experienced and reported here: the increased severity of dysphoric mood, somatic, and neuromuscular symptoms, but not vasomotor symptoms or insomnia. Of note is that the severity of depressed mood, vasomotor symptoms, and insomnia symptoms (especially night-time awakening) increased significantly among the SMWHS participants during the Late Menopausal Transition stage. Unfortunately there were not adequate numbers of women with menstrual cycle data available during the Late Menopausal transition Stage to support these analyses (19).

Changing hormonal patterns across the menstrual cycle during the menopausal transition have been identified and related to symptoms. SWAN Daily Hormone Study participants provided daily morning urine specimens and reported symptoms during each study cycle for up to 50 days until the FMP or up to 10 years. Participants in the DHS most commonly experienced cycles with evidence of luteal activity (ELA) denoted by a robust rise in progesterone until the year prior to the final menstrual period when only 23% of women met criteria for an ovulatory cycle (28). Changes in menstrual cycle types (based on estrogen levels and LH surge) have been studied in relation to vasomotor, sleep, and pain symptoms in the DHS. A majority (73%) of women experienced vasomotor symptoms prior to menopause, but the proportion experiencing VMS was not associated with cycle type. Women with anovulatory cycles that had absence of both estrogen increase and LH surge did not differ with respect to the number of VMS or percentage reporting VMS when compared to those with estrogen increase and LH surge (type 1) or estrogen increase with LH surge (type 2 cycles). The proportion of symptomatic days among women with VMS was associated with cycle type, with those with type 3 cycles experiencing a higher proportion of days with VMS (29). Complicating these findings was that women with no evidence of luteal activity (NELA) cycles in one year did not progress in an unremitting pattern to menopause (29).

SWAN DHS participants whose cycles were classified as premenopausal or perimenopausal (30) were most likely to record they had trouble sleeping the previous night at beginning or end of their menstrual cycles (perimenstrual), inconsistent with the lack of cycle phase effect noted in the SMWHS participants in analyses reported here. Pregnanediol levels were associated with increased trouble sleeping among women in perimenopause and FSH with trouble sleeping among premenopausal women. Both mood symptom scores and vasomotor symptoms were strongly associated with trouble sleeping in both groups.

Studies of pain symptoms indicated that bodily pain, including aches, joint pain, and stiffness, were associated with estradiol variability in the POAS cohort (15). Joint pain and back pain were not associated with estrogen levels (urinary E1G) among SMWHS participants (31). Participants in the SWAN DHS who had a history of migraine demonstrated a more rapid decline in E1c over 2 days following the luteal peak than women without migraine (32). These findings related to estrogen levels and variability may help explain the cycle phase effects seen with neuromuscular symptoms in the SMWHS data reported here.

Although persistence of estradiol during premenses when progesterone levels decline has been related to symptom severity in studies of cycling reproductive age women (26) and also to cortisol and catecholamine levels (27), greater interest has been devoted to estrogen levels and variability during the MT. Evidence from several cohorts indicates that estrogen gradually increases from the early to the late MT stage, with a substantial drop one year before and during the first year after the FMP (19, 33–36) and progesterone levels decrease during the same period (35). Hale noted erratic patterns of estradiol which she termed Luteal Out of Phase (LOOP) cycles, in which a second increase in estradiol occurred during mid- and late luteal phases that continued to peak during a subsequent menstrual phase (36).

During the menopausal transition, within-woman estradiol variability, likely including the luteal out of phase cycles described by Hale, has been related to symptoms, including hot flashes and aches, joint pain and stiffness, lower libido, and longitudinal measures of depressive symptoms and the first occurrence of depressive disorder (16,31). In a study of nonlinear models of estradiol and symptoms (hot flashes, night sweats, sleep disruption, and vaginal dryness) Dennerstein and colleagues (34) investigated the range and slope of estradiol assessed during days 4–8 of the menstrual cycle among women in the Melbourne Midlife Women’s Health Project who were still cycling, annually for years 1–8 and 11–13 of follow-up as they progressed through the menopausal transition. A greater range of estradiol (more variability) was associated with greater symptom severity for hot flushes, night sweats, sleeping problems, and vaginal dryness. Moreover, estradiol slope was also an important correlate of the same symptoms: more rapid change was associated with greater symptom severity.

Freeman suggested that women who experienced PMS may be vulnerable to the fluctuations of estradiol during the transition to menopause, pointing out that distinguishing PMS from menopausal symptoms may be difficult when women continue to experience menstrual cycles (8). Indeed, the relationship of PMS history to symptoms women report during the MT was evident in the POAS sample. Endocrine fluctuations associated with negative mood may become more influential as women progress into the late menopausal transition and experience more NELA cycles when variability of estradiol becomes more marked as noted by Freeman and colleagues in the POAS (8, 14–16). In a recent study Joffe (37) investigated estradiol variability, ovulatory levels of progesterone, and vasomotor symptom burden in relation to depressive symptoms based on weekly assessments of 50 perimenopausal women with mild to moderate depressive symptoms for an 8 week period. A majority had variable estradiol levels (90%) and nearly half (51%) had ovulatory progesterone levels. Greater estradiol variability and absence of progesterone levels consistent with ovulation were associated with more severe depressed mood symptoms, but not with vasomotor symptom frequency. Moreover, Gordon and colleagues found that estradiol variability was linked to enhanced emotional sensitivity (mood symptoms) in midlife women (45–60 years). Results linked estradiol variability and enhanced emotional sensitivity to a social stimulus, e.g. rejection, when it coincided with major very stressful life events (38). Of interest is that women experiencing greater estradiol variability over 14 months reported more severe depressive symptoms, but only if they had experienced very stressful life events, e.g. death of a loved one, suggesting a link between perturbation of estradiol and stress response to dysphoric mood. The data linking estrogen variability and the lower or unmeasurable levels of progesterone to symptoms may help explain the cycle phase effects we found in SMWHS participants during the late reproductive and menopausal transition stages.

Hormone patterns and their relationship to symptoms has been a research focus during the reproductive years through the remainder of reproductive aging. To date there has been limited but recent increasing effort to understand cyclic hormonal patterns as women experience the menopausal transition. Further longitudinal study of women’s cyclical symptoms may benefit from inclusion of women in the late reproductive stage through the late menopausal transition stage when menstrual cyclicity becomes markedly irregular and changes in the hypothalamic-pituitary-ovarian axis may be more strongly related to symptoms. Given the limited number of prospective studies of both menstrual cycle phase and menopausal transition stage effects on symptoms reported in a daily diary in the same cohort of midlife women using the STRAW stages of reproductive aging, the findings presented here warrant replication in other cohorts.

Conclusions

Further studies of symptom experiences during stages of the menopausal transition are needed to identify whether cyclicity in symptoms, such as we reported here, continues throughout the stages of reproductive aging in other cohorts of midlife women and to support use of an integrated view of menstrual cyclicity and progression through the menopausal transition stages in clinical practice with women during these complex periods of reproductive aging. Currently we know little about persistence of menstrual cycle phase effects on symptoms as women are simultaneously transitioning from cyclic regularity to irregularity, e.g. during LRS1, LRS2, Early Transition and Late Transition. Assuming that symptoms are influenced by either menstrual cycle phase or menopausal transition stage may thus provide an incomplete understanding of women’s symptom experiences. Instead, an integrated framework including the influence of both menstrual cyclicity and progression through the menopausal transition stages is needed.

Further study of symptom experiences during the stages of reproductive aging is needed to identify whether and how cyclicity in symptoms continues and varies through the menopausal transition stages in a larger and diverse cohort of midlife women. Moreover, a lifespan perspective is needed to capture both the influence of menstrual cyclicity and progression from the late reproductive stage through the menopausal transition on symptoms. These efforts will support incorporation of an integrated view of menstrual cyclicity and progression through menopausal transition stages in clinical practice with women as they experience the complex processes of reproductive aging.