Is This Woman Perimenopausal?

For each of the following cases, the clinician may need to determine the probability that the patient is perimenopausal.

Case 1. A 45-year-old woman who has had a hysterectomy at age 42 years for uterine fibroids reports during her annual examination that she has hot flashes and has been feeling irritable for the past month.

Case 2. A 41-year-old woman tells her physician she thinks she is starting menopause. She has smoked 1 pack of cigarettes a day for the past 20 years.

Case 3. A 47-year-old woman who has been taking oral contraceptives for the past 25 years requests information about her menopausal status. She is sexually active and wants to know if she needs to continue taking birth control medication.

The question "Is this woman perimenopausal?" is important for clinicians because patients ask and want to know if they are undergoing a physical and emotional change and if they are experiencing the menopausal transition. Physicians need information to identify perimenopausal women, to be able to reply to women's questions about the changes they may be experiencing, and to offer counseling on symptom relief, contraception, and disease prevention. As women begin the perimenopausal years, clinicians should counsel them on strategies to prevent osteoporosis as well as on evidence-based treatment options for climacteric symptoms such as hot flashes and night sweats. Clincians commonly identify perimenopausal women by their age, by inquiring about their menstrual history and symptoms, and by some ordering laboratory tests to look at hormone levels, such as follicle-stimulating hormone (FSH) and estradiol levels, to confirm their clinical suspicion. It would be useful to know how age, self-assessment, family and medical history, symptoms, physical signs, and laboratory tests affect the probability that the woman is perimenopausal.

In this article, we intend to answer the following questions: (1) What is the value of asking a woman whether she thinks she is starting menopause? (2) How accurate are symptoms and signs in detecting perimenopause? (3) Is there any value in asking about family and medical history in determining menopausal status? and (4) Are laboratory tests more useful than clinical examination in diagnosing perimenopause?

Climacteric is a general term referring to the entire transition from the reproductive to the postreproductive interval in a woman's life.¹ Thus, it includes immediate premenopausal, perimenopausal, and postmenopausal women. All women do not go through the same transition of regular menses to irregular menses to amenorrhea as they approach menopause. In 2001, a panel of experts (from The Stages of Reproductive Aging Workshop) met to discuss a staging system to classify reproductive aging.² This proposed new classification of the transition from reproductive to postmenopausal includes 7 stages, which are based on menstrual cycles and plasma FSH levels. The experts of this system noted that this is a work in progress and it has not been validated in research settings.

The World Health Organization defines natural menopause as "the permanent cessation of menstruation, determined retrospectively after 12 consecutive months of amenorrhea without any other pathological or physiological cause."³ Menstruation ceases as ovarian follicle stores are depleted and ovarian function is diminished, leading to eventual decreased production of estrogen by the ovary and decreased stimulation of the endometrial lining.⁴ Analysis of longitudinal data of women at all ages shows the probability of menstruating spontaneously after 12 months of amenorrhea to be less than 2%.⁵ The accurate diagnosis of perimenopause allows patients and physicians to predict the onset of menopause.

Perimenopause refers to the year before the final menstrual period through the first year after the final menstrual period.^{3 ,6} During perimenopause, ovulation occurs irregularly as a result of fluctuations in the hormones of the hypothalamic-pituitary-ovarian axis.⁶ For example, in early perimenopause inhibin B levels decline, resulting in a rise in FSH levels, with no significant change in inhibin A or estradiol levels. Follicle-stimulating hormone levels may rise during some cycles but return to premenopausal levels in subsequent cycles. Further complicating the determination of FSH concentration is the pulsatile pattern of secretion. Similarly, concentrations of estradiol also may decrease or even increase during perimenopause.⁵ This hormonal variability creates difficulties in interpreting a single laboratory test value.

Based on longitudinal data of women's menstrual cycles, Brambilla et al⁷ and Dudley et al⁸ further refined the definition of perimenopause by considering women perimenopausal if they have not had a period within the previous 3 to 11 months or if they have experienced changes in menstrual regularity (either shortening or lengthening of time between menses) during the past 12 months. In a 5-year population-based study, Brambilla et al⁷ found that 3 to 11 months of amenorrhea or of irregular periods among women aged 45 to 55 years were most predictive of menopause within the following 3 years (sensitivity rate, 72% and specificity rate, 76%). Dudley et al⁸ validated this definition, finding that these 2 characteristics are the best predictors of menopause 4 years after baseline (sensitivity rate, 32% and specificity rate, 99%). The perimenopausal definition of Brambilla et al⁷ was used as our reference standard for this systematic review.

To determine a woman's likelihood of perimenopause, clinicians must first estimate the pretest probability of perimenopause. This estimate should be based primarily on the patient's age, although certain aspects of the medical and family history also may be useful.

In a 30-year study following college women throughout their lifetime until menopause, Treloar et al⁵ reported the mean age of onset of perimenopause as 45.5 years, with a mean duration of 6.25 years. Based on 5-year follow-up data from a population-based study of 5547 women aged 45 to 55 years, McKinlay et al⁹ in the Massachusetts Women's Health Study (1992) reported the median age of onset of perimenopause as 47.5 years, with a mean duration of 3.8 years. Figure 1 shows the prevalence of perimenopause and postmenopause based on age from McKinlay et al.⁹ Unfortunately, estimating the time of onset of perimenopause is very challenging, and data were not available from the literature on the prevalence of perimenopause among women younger than 45 years. McKinlay et al⁹ reported that by age 45 years, 40% of all women have started or completed the menopause transition (32% are perimenopausal and 8% are postmenopausal). By age 50 years, 75% of women have started or completed the transition (38% perimenopausal and 37% postmenopausal). By age 55 years, only 2% of women are premenopausal.

This evaluation can be divided into 5 basic categories: (1) self-assessment, (2) symptoms, (3) family and medical history, (4) physical signs, and (5) laboratory tests.

Self-assessment. Clinicians can ask a woman if she thinks she is starting menopause. Women may base their perceptions of their menopausal status on awareness of the subtle changes taking place in their bodies.^{10 - 11} In a cross-sectional study by Garamszegi et al,¹⁰ self-reported menopausal status was more correlated with symptoms than menstrual cycle characteristics.

Symptoms. Climacteric symptoms typically include vasomotor complaints, such as hot flashes and night sweats. Other symptoms associated with perimenopause in cross-sectional studies are thought to be associated with fluctuating levels of estrogen and progesterone. These include vaginal dryness, variable sexual interest, urinary incontinence, depressed mood, nervous tension and irritability, and sleep disturbances.¹

Hot Flashes. Hot flashes are sudden sensations of heat, sweating, and flushing that most often occur in the face, head, neck, and chest. Chills, clamminess, and anxiety also may accompany hot flashes. They generally last 1 to 5 minutes, and only 6% of women experience hot flashes that last longer than 6 minutes.¹² Most North American, European, and Australian women report that they experience hot flashes (50%-85%)^{9 ,12 - 14} and that they occur periodically during a span of 1 to 5 years.^{15 - 16} There appear to be cultural differences in the reporting or experiencing of hot flashes. For example, only 10% to 20% of Indonesian women¹⁷ and 10% to 25% of Chinese women¹⁸ report experiencing them. The mechanism triggering these episodes is thought to be a combination of fluctuating estradiol levels and a narrowing of the thermoneutral zone.¹⁹

Night Sweats. Night sweats are hot flashes that occur at night, usually while the woman is sleeping. Often, she will awake drenched in sweat. If night sweats interfere with sleeping patterns, this may explain reports of insomnia, fatigue, and irritability among climacteric women.

Vaginal Dryness. Vaginal dryness is sometimes experienced as a result of decreasing estrogen production during perimenopause. This can lead to urogenital atrophy and changes in the quantity or composition of vaginal secretions. Estimates of the prevalence of vaginal dryness among late perimenopausal women range from 18%²⁰ to 21%.²¹

Variable Sexual Interest. Dennerstein et al²² report in a study of Australian women that although the majority of women indicated no change in sexual interest during menopause, 31% experienced a decrease and 7% reported an increase in sexual interest. Only 6% of those reporting a decrease indicated menopause as a reason for the decline in interest.²² This decrease may be caused by physiological factors making sexual relations more difficult (eg, vaginal dryness, hot flashes, urinary incontinence) or social and environmental factors. Several studies have found that menopausal symptoms are but 1 of many different factors affecting sexual interest among women in midlife and later.^{23 - 24}

Urinary Incontinence. In western countries, urinary incontinence effects between 26%²⁵ and 55%²⁶ of middle-aged women. This may be caused by or exacerbated by declining estrogen levels. Lower estrogen levels can lead to atrophy of the urethral mucosa and the trigone, the muscle controlling urination, resulting in less urinary control.⁶ Some studies have found an association between increased prevalence of urinary incontinence and menopause²⁵ while others have not.^{27 - 28}

Depressed Mood. Avis et al²⁹ classified 10% of 45- to 55-year-old women participating in a population-based longitudinal study of women from Massachusetts as experiencing clinical depression. Many studies do not find an association of menopause with depression or find that it can be explained by other menopausal symptoms.^{29 - 34} Evidence from North American²⁹ and British³⁵ cohorts found high rates of depression among perimenopausal women with a previous history of depression, supporting the theory that women with previous affective disorders may be at an increased risk for recurrent depression. Conclusions from other reports have suggested that depression could be increased due to declines in estrogen levels,³⁶ changes in social circumstances,³⁷ and changes in self-concept as women lose reproductive function.³⁸

Nervous Tension and Irritability. Many checklists for menopause symptoms used in epidemiologic studies include nervous tension and irritability.^{21 ,39 - 42} Although the relevance of these symptoms is unclear, they could be caused by lack of sleep due to menopausal symptoms, illness, or stressful life events. Some authors have suggested that they could result from changes in hormone levels, which also occurs during the 10- to 14-day luteal phase of the menstrual cycle.⁴³

Medical and Family History.Age of Mother's Menopause. Genetic factors seem to predispose women to menopause at an earlier age.^{44 - 45} Torgerson et al⁴⁴ reported that women with premature (<40 years) and early (<45 years) menopause report significantly younger maternal menopausal ages than women with normal menopausal ages. In a case-control study of women from the greater Boston area, Cramer et al⁴⁵ found women with a family history (eg, mother, sister, aunt, grandmother) of menopause before age 46 years had a higher risk of early menopause (odds ratio 6.1; 95% confidence interval [CI], 3.9-9.4).

Cigarette Use. Approximately 23% of US adult women smoke cigarettes regularly.⁴⁶ Evidence indicates that women who smoke experience menopause 1 to 2 years earlier than nonsmokers.^{2 ,47 - 54} Cigarette smoking reduces bioavailable estrogen by increasing hepatic metabolism of estrogen,^{55 - 56} decreasing production of estrogen,^{57 - 58} or increasing circulation of androgens.⁵⁹ Several studies support the assertion that quitting smoking can significantly delay menopause.^{48 - 49} Other evidence suggests that the median age of menopause is not statistically different between never smokers and ex-smokers.^{60 - 61} Nevertheless, a majority of research on cigarette smoking and menopause does indicate a dose-response relationship between number of cigarettes currently smoked and age at menopause.^{48 - 49 ,53 ,62} Furthermore, Gold et al^{20 ,54} noted that "past smoking and current smoking were positively associated with prevalence of vasomotor symptoms" in agreement with most previous collected data.⁵⁴

Hysterectomy Status. It is often assumed that women who have had a hysterectomy with conservation of the ovaries should not experience menopausal symptoms earlier or more severely due to their hysterectomy. Nonetheless, evidence shows that women with ovarian conservation following hysterectomy report more vasomotor complaints, vaginal dryness, and other complaints than do women of similar age who did not have a hysterectomy.^{63 - 64} In developed countries, hysterectomy is one of the most frequent operations in adult women⁶³ ; one third of US women will have had a hysterectomy by age 65 years.⁶⁵ Hysterectomy may inhibit blood circulation in ovaries, decreasing ovarian function⁶⁴ and causing more frequent or severe menopausal symptoms.

Signs.Maturation Index. One proposed assessment of vaginal estrogen deficiency is an evaluation of the vaginal epithelium maturation index. This procedure involves obtaining cells from the junction of the upper and middle third of the lateral vaginal wall using a brush. These cells are prepared on a slide using the Papanicolaou technique and the percentages of parabasal, intermediate, and superficial cells are counted.⁶⁶ Although the maturation index changes significantly after estrogen replacement therapy, diagnostic studies have not compared the maturation index with menstrual cycle characteristics.

Vaginal pH. Some investigators have suggested than an elevated vaginal pH (6.0-7.5) in the absence of potentially pathogenic bacteria may be a reasonable marker of decreased estradiol serum levels.⁶⁷ This test is performed by directly applying pH paper to the lateral vaginal wall at the outer third of the vagina. Changes in pH could be due to alterations in the composition of vaginal secretions that accompany atrophy.

Skin Thickness. Estrogen stimulates the epidermal growth rate and promotes the formation of collagen and hyaluronic acid, which increase the turgor and vascularization of the skin.⁶⁸ During climacteric, declining estrogen levels result in the thinning and atrophy of the epidermis.⁶⁸ Investigators have proposed measuring skin thickness using ultrasound at the greater trochanter area to estimate menopausal status, but this procedure has not been supported by research to date.⁶⁸

Laboratory tests.Follicle-stimulating hormone. Measurement of FSH plasma levels has been used to try to identify perimenopausal and postmenopausal women. High FSH levels indicate that menopausal changes are occurring in the ovary. As the ovary becomes less responsive to stimulation by FSH from the pituitary gland (and produces less estrogen), the pituitary gland increases production of FSH to try to stimulate the ovary to produce more estrogen (Figure 1). However, some clinicians and researchers doubt the clinical value of FSH measurements in perimenopausal women as FSH levels fluctuate considerably each month depending on whether ovulation has occurred.^{2 ,69 - 70}

Estradiol. Recent longitudinal studies have reported that early perimenopausal (change in cycle frequency) women maintained premenopausal estradiol levels whereas late perimenopausal (no menses in previous 3-11 months) and postmenopausal women experienced significant declines in estradiol levels.⁷¹ Estradiol can be measured using plasma, urine, and saliva. Like FSH, estradiol levels are highly variable during perimenopause.¹

Inhibins. Inhibin A and inhibin B are secreted by the ovary and, like estradiol, exert negative feedback on the pituitary gland, reducing FSH and luteinizing hormone secretion (Figure 1). Loss of inhibin contributes to the rise in FSH that occurs with ovarian senescence. A recent longitudinal study of hormone levels throughout the menopause transition reported that inhibin B levels decline as women progress through perimenopause while inhibin A levels remain unchanged. Inhibin A levels did decrease around the time of the final menstrual period.⁷¹ Inhibin levels are usually measured in plasma. The ovaries produce less inhibin B as fewer follicles proceed to maturation, and the number of follicles declines with age.⁷²

We searched the MEDLINE database for English-language articles concerning the diagnosis of menopause that were published between 1966 and 2001. The key words used included menopause, perimenopause, premenopause, climacteric, sensitivity and specificity, diagnosis, prospective/cross-sectional studies, health status, and hormones of the hypothalamic-pituitary-ovarian axis. We included articles that used the diagnosis of perimenopause based on menstrual irregularity or 3 to 11 months of amenorrhea, included a premenopausal control group, and presented data that could be extracted to calculate both sensitivity and specificity rates. We included articles on laboratory tests that are available to clinicians for 2 reasons. First, women may ask for laboratory tests to assess their menopausal status. Second, the results of the test must be coupled closely with the clinical examination for proper interpretation. We excluded reviews and articles that included men, hormone replacement therapy, cancer, or osteoporosis as major foci of the papers. We developed the search strategy with a medical librarian, and this is available on request by the author. Two authors (L.A.B. and C.M.S.) systematically reviewed and identified titles and abstracts for content and quality. Articles using a definition of perimenopause different from 3 to 11 months of amenorrhea or irregular periods, those lacking a control group (a remote premenopausal group), and studies whose data could not be classified into contingency tables were excluded. Articles using a young control group (ie, 20-year-old women) or an older postmenopausal group (ie, 60- to 70-year-old women) or including women receiving hormone replacement therapy also were excluded. Two authors (L.A.B. and C.M.S.) abstracted the articles using a standardized abstraction form. Each publication was given a grade of A, B, or C based on the study design and level of evidence.⁷³ Discrepancies about quality were resolved by a third author (K.N.).

The MEDLINE search identified 1221 articles, and from the references cited in these and other publications known to us, another 25 articles were added to the review pool. Sixteen articles^{10 - 11 ,20 - 22 ,26 ,28 ,39 ,42 ,54 ,68 ,74 - 78} met all the inclusion criteria described above and were included in the final analysis (Table 1).

We calculated values and CIs for sensitivity rate, specificity rate, and positive and negative likelihood ratios (LR + , LR − ) using statistical software (SAS version 8.0; SAS Institute Inc, Cary, NC). Perimenopause is the target condition and the reference standard is based on the definition by Brambilla et al.⁷ Sensitivity is the true positive rate and specificity is the true negative rate, and both numbers estimate the accuracy of the test's identification of those with and without perimenopause. The perfect test has a sensitivity and specificity equal to 1.00.

The LR + (sensitivity/[1−specificity]) is a measure of how well a positive test result rules in perimenopause whereas the LR − ([1−sensitivity]/specificity) is a measure of how well a negative test result rules out perimenopause. A likelihood ratio close to 1 does not appreciably predict the likelihood of perimenopause. A likelihood ratio greater than 1 increases the likelihood of perimenopause, whereas a likelihood ratio less than 1 decreases the likelihood of perimenopause. We tested the values for sensitivity rate, specificity rate, LR + , and LR − from different studies examining the same symptom or sign for homogeneity. When the χ² statistic suggested homogeneity (P>.05), we combined the data to produce a random-effects estimate.⁷⁹ For heterogeneous data, variables are given as ranges.

Findings that were similar across studies (Table 2); that is, those that had the greatest LR + and were therefore best at ruling in perimenopausal status were hot flashes (LR + range, 2.15-4.06), night sweats (LR + 1.90, 95% CI, 1.63-2.21), and vaginal dryness (LR + range, 1.48-3.79). The absence of findings was not efficient at ruling out perimenopausal status; self-rating (LR− range, 0.18-0.36) and hot flashes (LR− range, 0.54-0.87) had the smallest negative LR. Only 1 study each reported enough data to calculate sensitivity rate, specificity rate, and likelihood ratios for FSH and the inhibins^{71 ,77} ; no study reported enough data to calculate these values for estradiol. High FSH levels (≥24 IU/L) and low inhibin B levels (≤30 ng/L) provided weak evidence to rule in perimenopause (LR+ 3.06, 95% CI, 2.06-4.54 and LR+ 2.05, 95% CI, 0.96-4.39, respectively). However, neither normal FSH levels nor normal inhibin B levels could rule out perimenopause (LR− 0.45, 95% CI, 0.36-0.56 and LR− 0.70, 95% CI, 0.51-0.96, respectively).

Case 1 describes a 45-year-old woman with a moderately high pretest probability of being perimenopausal or postmenopausal (40%) based on her age (Figure 1) and probably even higher because she has had a hysterectomy and is experiencing climacteric symptoms. Because she has reported hot flashes (LR + range, 2.15-4.06) and irritability (LR + 1.23, 95% CI, 1.12-1.34), the calculated posttest probability of her being perimenopausal ranges from 40% to 100%. Our recommendation would be to not order FSH or other laboratory tests but to tell her that she is perimenopausal, to counsel her on increasing her calcium intake, and to increase exercise to prevent osteoporosis. If her climacteric symptoms are causing her discomfort, we would suggest a low-dose transdermal or oral estrogen medication. Although recent evidence has downplayed the role of hormone replacement therapy for long-term prevention of disease,⁸⁰ it is still the most widely prescribed therapy for climacteric symptoms. Other therapies shown to reduce hot flashes include clonidine,^{81 - 82} tibolone,^{83 - 84} and alternative treatments such as phytoestrogens and black cohosh.⁸⁵

In case 2, a 41-year-old woman might have a pretest probability of being perimenopausal or postmenopausal (estimate, 10%) based on her age. Because she is currently smoking cigarettes (LR + 1.33, 95% CI, 1.22-1.45), this aspect of her medical history raises her probability of being perimenopausal to a range of 12% to 14%. Because she thinks she is starting menopause (LR + range, 1.53-2.13), her calculated probability of being perimenopausal might be further increased to the range of 18% to 30%. We would inquire about her menstrual patterns and presence of climacteric symptoms, tell her she may be close to perimenopause, and discuss contraception and smoking cessation with her.

Finally, the 47-year-old woman in case 3 has a different type of question. Her pretest probability of being perimenopausal or postmenopausal is high (50%) based on her age (Figure 1). In this patient, we cannot assess for menstrual patterns or menopausal symptoms because she is using an oral contraceptive. Although she is likely perimenopausal, ovulation may still be possible.⁸⁶ If she desires to continue with her oral contraceptive, the American College of Obstetricians and Gynecologists recommends discontinuing this therapy between the ages of 50 and 55 years.⁸⁷ As suggested in case 1, we would counsel her on increasing calcium intake and increasing exercise for osteoporosis prevention.

No single element of the history or clinical examination is powerful enough to confirm the probability of being perimenopausal. Besides menstrual history, the most powerful predictor of menopausal status may be age. As by McKinlay et al,⁹ the median age at perimenopause is 47.5 years, and 87% of women are perimenopausal or postmenopausal by the age of 51 years. The clinical question of perimenopausal status is more challenging in patients in their early to middle 40s. Many clinicians rely on the measurement of hormone levels, such as FSH, to confirm the diagnosis. In the clinical scenarios we evaluated, FSH measurement did not help the clinician make a diagnosis. Further research needs to be conducted to document the additional benefit of these hormone level tests in making a diagnosis of perimenopause.