Sleep and the menopause – do postmenopausal women experience worse sleep than premenopausal women?

Abstract

Objective

To examine the sleep characteristics in three cross-sectional populations: young, premenopausal and postmenopausal women, and the associations between sleep, menopause, mood and cognitive performance.

Study design

Twenty-one premenopausal (45–51 years), 29 postmenopausal (59–71 years) and 11 young (20–26 years, using oral contraceptives) women were recruited. Polysomnography was used to measure objective sleep quality. Subjective sleep quality, sleepiness and mood were assessed using questionnaires. Cognitive performance was investigated by means of three attentional tests.

Results

Total sleep time in pre- and postmenopausal women was similar (404.9 and 384.7 minutes), but shorter than in young women (448.2 minutes, P = 0.030 and <0.003, respectively). Sleep efficiency followed the same pattern, being 84.3% in premenopausal (P = 0.027), 80.2% in postmenopausal (P < 0.003) and 93.4% in young women. Pre- and postmenopausal women had less slow wave sleep (duration or activity) and more wake time after sleep onset (duration or frequency). Insomnia complaints were more frequent after the menopause (P = 0.023). Sleepiness and mood scores were similar in all groups. Reaction speeds slowed with increasing age. After the menopause, better cognitive performance was associated with more rapid eye movement sleep.

Conclusion

Objective sleep measures differed significantly between the young and postmenopausal groups. These differences may be more because of the physiology of ageing than the rapid changes across the menopause, since similar sleep characteristics were already present in the premenopausal women. The increase in sleep complaints after menopause was not associated with sleepiness or disturbances in objective sleep quality, mood or cognitive performance.

Keywords

Cognition estrogen menopause mood sleep

Introduction

Sleep complaints during or after menopause are a common medical problem. Leger et al. ¹ reported in their questionnaire study with over 12,000 participants of both sexes, that 15% of women over 50 years, but only 5% of women aged 18–24, suffered from severe insomnia. In men, the corresponding figures were 8% and 2%.¹ In the multicentre Survey of Women's Health Across the Nation, the odds ratios for troubled sleep were 1.6 for postmenopausal and 1.3 for perimenopausal when compared with premenopausal women.² In an earlier study by Kuh et al. ³ the ratios were even higher (3.4 and 1.5, respectively).³ Subjective sleep disturbances at menopause have been attributed to nocturnal hot flushes or sweats,^2–4 which may not, however, be directly involved in arousals.⁵ Vasomotor symptoms may associate with anxiety, depression, stress or tension, which may cause or at least contribute to sleep problems.^4,6

Several polygraphic sleep studies have been performed to elucidate the determinants and pathophysiological mechanisms or correlates of menopausal sleep complaints. These studies have failed to demonstrate the consistent sleep changes behind the symptoms. In general, studies comparing sleep architecture in pre- and postmenopausal women have shown either no difference⁷ or, contrary to initial predictions, better sleep after the menopause.^8,9 Different populations or methodology could explain part of the discrepancies. Moreover, the methods used may not have been sensitive enough to disclose subtle differences.

The clinical importance of decreased sleep quality is largely determined by associated sleepiness or impaired cognitive performance. Concentration, attention and reaction speed are some of the cognitive performances that deteriorate as a function of decreased sleep quality or increasing age.^10–14 Some of the brain areas involved in sleep regulation or cognitive functions, such as attention or memory, are also targets for sex hormone action.^15–17 Inclusion of cognitive tests was expected to add to the sensitivity of the study, consisting of conventional sleep parameters, for detecting menopause-induced physiologically important alterations in sleep.

Most of the studies examining the interactions between menopause and sleep compare pre-, peri- or postmenopausal groups in settings where it is difficult to differentiate between the menopausal and age effects. We added an age axis to assess sleep differences in three age groups of healthy women. The impact of possible objective or subjective menopausal sleep changes was investigated using attentional tests and questionnaires assessing sleepiness and mood. Impaired sleep was hypothesized to be associated with increased sleepiness or disturbances in attention or mood.

Methods

Subjects

This study was a part of a larger survey (European Union [EU]: QLK6-CT-2000-00499) investigating the effects of ageing and hormone therapy (HT) on sleep and cognitive functions in women. It was performed by a collaboration between Turku and Helsinki Universities in Finland. The study comprised 21 premenopausal women (aged 45–51 years), 29 postmenopausal women (aged 59–71) and 11 young women (aged 20–26), all of whom were enrolled through newspaper advertisements.

Premenopausal status was defined by serum follicle stimulating hormone (FSH) measurement (<23 IU/L) and ongoing menstrual cycle. The pre- and postmenopausal women were not using any HT, but the young women were taking oral contraception (OC = ethinyl estradiol 20 µg desogestrel 0.15 mg [Mercilon]). In the premenopausal group, two women had previously used the HT and the average time of use was 3.5 months (range 3–4 months). In the postmenopausal group, 21 women had previously used the HT, with an average time of 68 months (range 2–156 months). The washout period for the HT was at least 12 months in all women, except in one premenopausal woman, in whom it was five months. The characteristics of the study groups are summarized in Table 1.

Table 1

Basic characteristics, expressed as group means (SD), for the three study groups of the subjects with complete data

	Young n = 11	Premenopausal n = 20	Postmenopausal n = 28	P value
Age (years)	23.1 (1.6)^a	47.7 (2.3)^a	63.3 (3.6)^a	<0.001
BMI (kg/m²)	23.1 (3.1)^b	23.9 (2.4)^b	27.2 (4.2)	0.002
E₂ (pmol/L)	95.5 (83.6)^c	327.2 (330.2)^d	31.5 (10.5)	<0.001
FSH (IU/L)	3.8 (2.8)^a	10.6 (4.8)^a	75.2 (29.1)^a	<0.001
EQ-5D	7.4 (0.8)	7.1 (0.6)	7.4 (0.8)	NS
Vasomotor symptoms	2.5 (0.8)	2.9 (0.9)	4.0 (2.3)	NS
Education (years)	16.8 (1.5)^d	16.3 (4.1)^e	12.3 (3.6)	<0.001

BMI, body mass index; E₂, oestradiol; FSH, follicle stimulating hormone; EQ-5D, EuroQol quality-of-life questionnaire; SD, standard deviation

Bonferroni corrected P: ^a<0.003 between all groups, ^b0.009, ^c0.024, ^d<0.003 and ^e0.003 compared with postmenopausal women

Only healthy women were included. Exclusion criteria are described in Box 1. The washout time for any use of medicaments with central nervous system (CNS) effects or antioxidants was a minimum of three months. Regular sleep–wake schedules (22:00–23:00 hours to 06:00–07:00 hours) were required. All volunteers were first screened and informed about the study in a 10–30-minute telephone interview (NK, PP-K, ASU). Women who met the study requirements were invited to the sleep laboratory for a more detailed interview (1–1.5 hours; PP-K, ASU) where inclusion/exclusion criteria were confirmed, a physical examination carried out, the study protocol carefully explained and the laboratory environment introduced. The participants were required not to start using any antioxidants, hormones or CNS medication during the study. A caffeine-free diet was also required during the study and one week before it. Coffee drinkers were provided with caffeine-free coffee. Alcohol and travelling abroad were prohibited for one week before the study. The subjects kept a sleep diary for three weeks before and one week after the sleep studies to ensure a regular sleep–wake schedule. After oral and written information, all subjects signed an informed consent form. The study had the approval of the Ethical Committees of Turku University Hospital and the University of Helsinki.

Box 1

Exclusion criteria for subject participation

Ongoing malignancy

Cardiovascular disease (apart from treated hypertension)

Neurological disease

Significant loss of consciousness previously

Severe migraine

Previously diagnosed and treated nocturnal breathing disorder

Narcolepsia

Restless legs syndrome

Thyroid disease (4.5 mU/L <TSH <0.4 mU/L)

Fibromyalgia

Anaemia (Hb <118 g/L)

Depression (Beck depression inventory >13)

Current use of any medication with CNS effects

Shift work

Irregular sleep–wake rhythm

Smoking

Use of narcotics

Excessive consumption of caffeine (>5 cups/day)

TSH, thyroid stimulating hormone; Hb, haemoglobin; CNS, central nervous system

Study design

All pre- and postmenopausal women were studied in Turku and the young women in Helsinki. A uniform timetable and performance of procedures were carefully followed at both sleep centres. The supervisor of the study (PP-K) was in charge of the similarity of the procedures at both centres. The sleep recordings were carried out during two consecutive nights, the first night serving as the adaptation night. The menstruating women (premenopausal and young women) were studied during the first days of their cycle. On the first evening, the subjects arrived at the laboratory at 19:30 hours and went to bed at 23:00 hours (lights off). The subjects were woken up at 07:00 hours (lights on), cognitive tests were performed and blood samples taken for serum FSH and estradiol (E₂) (Table 1). The same schedule was repeated the next day, but without the blood samples. The subjects spent the time allowed for sleep (23:00 hours to 07:00 hours) in bed in a dark room without windows where only red light was allowed for illumination if needed.

Sleep recording and scoring

During both the adaptation and the study night, polysomnographic recordings were made. Polysomnography consisted of two electroencephalograms (EEG, C3/A2, C4/A1), two electro-oculograms, a mandibular electromyogram (EMG) and an electrocardiogram. For the pre- and postmenopausal women, two additional EEG channels were recorded (O1/A2, O2/A1) (Embla, Medcare Flaga hf. Medical devices, Reykjavik, Iceland). The staging of all sleep recordings (Turku and Helsinki) was carried out by the same scorer (NK). For quality control, all recordings were re-scored by a senior scorer (PP-K). Conventional criteria¹⁸ were used to classify the non-rapid eye movement (NREM) sleep, REM sleep and movement time (MT) as well as the wake time after sleep onset. NREM sleep includes stage 1 (S1), stage 2 (S2), slow wave sleep [SWS; stage 3 (S3) and stage 4 (S4) combined]. S1 sleep is considered to be a transition phase between wakefulness and sleep. This drowsiness-like state usually leads to S2 sleep, which is light sleep with a low arousal threshold. SWS represents the deepest sleep with the highest arousal threshold. REM sleep is characterized by desynchronized EEG frequencies, loss of EMG activity and the presence of REMs. NREM sleep, particularly its deeper stages, predominate early in the night and the REM sleep appears at around 90-minute intervals. As the night progresses, the REM episodes become longer, and the NREM sleep shorter as well as lighter.¹⁹

The sleep stages (1, 2, SWS, REM, MT, wake) were expressed as percentages of time in bed (from lights off to lights on). Sleep efficiency was calculated by dividing the total sleep time by the time in bed. The latency to sleep onset was defined as the time from lights off to first 90 seconds of stage 1 or to first 30 seconds of any other sleep stage. The latencies to SWS and to REM sleep were defined as the times from sleep onset to the first 30 seconds of the respective sleep stage. An awakening was classified as entering wake stage from sleep. The criteria of the American Sleep Disorders Association were used to score arousals²⁰ (Somnologica^®, Medcare Flaga hf. Medical Devices, Reykjavik Iceland). These arousals are a normal feature of sleep. They are brief and there is no awareness or recall of these events.¹⁹

Spectral analyses

Spectral analysis was used to quantify the slow wave activity (SWA, 0.75–4 Hz) of NREM sleep episodes. The sleep cycles were defined according to the rules of Feinberg and Floyd,²¹ excluding the requirement for the minimum duration of the first REM episode. Only full cycles were included. The first NREM episode was the sleep period from sleep onset to the beginning of the first REM episode. The following NREM episodes began from the end of each REM episode and lasted to the beginning of the next. The time course of SWA was normalized by calculating the SWA power per 30 seconds in each NREM sleep episode.

The power spectrum was calculated from the C3-A2/C4-A1 derivation in four-second epochs using the 256-point fast Fourier transform with 50% overlapping. EEG artefacts (movements, eye movements) and events triggering slow waves with increased muscle tone were visually identified and omitted (S-LH).²² To preserve sleep continuity, the power spectrum samples that overlapped manually marked artefacts were considered as missing data. To ensure that the spectral results were comparable between the study groups, the 100 Hz recordings of the young women were re-sampled into 200 Hz before the analysis.

Comparisons were carried out on the first four NREM sleep episodes in most subjects. Four women in the premenopausal group, eight in the postmenopausal group and three in the young group, had only three episodes. In addition, one woman in the premenopausal group and one woman in the postmenopausal group had only two episodes.

Questionnaires

Subjective sleep quality, sleepiness and mood were evaluated in the morning immediately after awakening. A questionnaire on sleep quality, sleep efficiency, sleep latency, number of awakenings, too early morning awakening and morning tiredness was used to assess the subjective sleep quality of the preceding night (the questions are shown in more detail in results, Table 3). The variables were categorical, with a low number referring to good sleep or to a low level of sleeping problems. Three questions (sleep efficiency, sleep latency and the number of awakenings) were used to investigate the correlations between subjective and objective sleep quality. Sleepiness and mood were evaluated with five visual-analogue scales (VAS) (Appendix 1). For statistical analyses, the answers concerning the levels of depression, sleepiness and irritability were reversed, resulting in a lower number referring to better mood in all questions. The Stanford sleepiness scale was also applied.²³ In addition, insomnia (scale 5–25) and sleepiness (scale 5–25) during the past three months were evaluated using the Basic Nordic Sleep Questionnaire (BNSQ)²⁴ (Appendix 2) and quality of life was assessed using the EuroQol quality-of-life questionnaire (EQ-5D).²⁵ Climacteric vasomotor symptoms were scored with two questions on the past six months (night sweats and hot flashes). The frequency of the symptoms was determined on the following four-point scale: one (seldom or never), two (approx. once a month), three (approx. once a week), four (almost every day). Vasomotor symptom score was a sum of the two answers.

Cognition

Cognitive tests were performed using the CogniSpeed program (AboaTech LTD, Turku, Finland).²⁶ Three attentional tests were used: simple reaction time (SRT), two-choice reaction time (2-CRT) and vigilance tests. The tests, apart from the test of vigilance, were carried out on both evenings before bedtime and on both mornings immediately after awakening. Only the tests performed in the morning following the study night were used for the final analysis. In the SRT test, the subject had to press zero (0) on the keyboard as soon as the target number zero (0) appeared on the computer screen. In the (2-CRT) test, there were two target numbers: one (1) and two (2). The numbers appeared on the screen in a random order and, as in the SRT, with random delays varying between one and four seconds. The subject was instructed to press the corresponding key as soon and accurately as possible whenever a target appeared. A practice session was performed before the first test. Reaction times (RTs) of correct responses were measured in milliseconds (ms). In the 2-CRT, the number of errors was counted as well.

Sustained attention was measured with the vigilance test, a visual test of letter cancellation. The test was carried out on the morning following the study night. Immediately before the test, there was a practice session. This monotonous task lasted 15 minutes. The three target letters mingled with non-targets occurred in the middle of the screen randomly, one at a time, with a probability of 15%. The subject pressed the space bar immediately after noticing a target letter. In the statistical analysis mean, individual RTs were used as a measure of speed and the omission rate as well as the number of errors (false positives) as measures of accuracy.

Statistics

Due to skewed data distribution, non-parametric tests of Kruskal-Wallis were first used to evaluate all the objective sleep variables, the answers from BNSQ, EQ-5D and Stanford questionnaires and cognitive variables. If the results were significant (P < 0.05), the Mann-Whitney U-test was used to analyse multiple comparisons among all the three study groups. These results were adjusted according to the Bonferroni procedure. The subjective sleep quality variables from the morning questionnaire were evaluated with Fisher's exact test. Spearman's correlation coefficient was used to investigate the correlations between subjective and objective sleep variables and between objective sleep variables and cognitive variables. In order to increase the power of the analyses to detect the differences between the pre- and postmenopausal groups, post hoc analysis using the Kruskal-Wallis test was used to analyse the objective and subjective sleep variables, the answers from BNSQ, EQ5D and Stanford questionnaires and cognitive variables. The power calculations included key sleep and cognitive variables (total sleep time, sleep latency, SWS%, REM%, SWA, 2-CRT and vigilance RT). The power of our population to show significant differences between the pre- and postmenopausal women was greater than 80%.

Results

All women completed the study. One subject from the premenopausal group and one from the postmenopausal group were excluded because of missing data. In addition, spectral analysis could not be performed for one young subject.

Objective

Objective measurements of sleep did not differ between the pre- and postmenopausal groups. All differences found were between the older groups (pre- and postmenopausal) and young women (Table 2). The older groups had less total sleep time and lower sleep efficiency. The REM latency in postmenopausal women was shorter than that in young women. No other differences in latencies were found. Pre- and postmenopausal women spent less time in SWS and were more awake, and they had less MT compared with young women. The SWA in the first four NREM sleep episodes was lower in the pre- and postmenopausal groups compared with young women (Figure 1). The most marked difference was seen in the first NREM sleep episode. Also, the total NREM SWA was lower in the pre- (mean 113.2 [standard deviation, SD 98.9], P < 0.003) and postmenopausal (mean 79.1 [SD 31.4], P < 0.003) groups than in young women (mean 284.6 [SD 89.9]).

Figure 1

Slow wave activity (SWA) in the three study groups. There were no differences between the pre- and postmenopausal groups. Mean SWA ± 1 SD are demonstrated for each group. Pre- and postmenopausal women had lower SWA compared with young controls ^a(P < 0.003), ^b(P = 0.012), ^c(P = 0.009) and ^d(P = 0.006)

Table 2

Sleep variables, expressed as group means (SD), for the three study groups

	Young	Premenopausal	Postmenopausal	P value
Total sleep time	448.2 (27.2)	404.9 (44.8)^a	384.7 (54.6)^b	<0.001
Sleep efficiency	93.4 (5.6)	84.3 (9.3)^a	80.2 (11.4)^b	<0.001
Sleep latency	14.1 (9.2)	15.2 (15.8)	16.8 (14.4)	NS
SWS latency	12.8 (6.4)	20.3 (17.3)	15.9 (16.3)	NS
REM latency	107.9 (36.3)	83.5 (31.1)	74.3 (25.5)^c	0.014
S1%	6.4 (2.8)	7.8 (3.2)	8.1 (3.3)	NS
S2%	43.7 (4.1)	43.2 (8.8)	39.7 (7.9)	NS
SWS%	22.1 (3.2)	12.6 (5.8)^b	14.2 (6.5)^d	<0.001
REM%	21.0 (4.2)	20.6 (6.1)	18.1 (5.3)	NS
Wake%	3.7 (5.3)	12.6 (8.4)^d	16.3 (9.4)^b	<0.001
MT%	0.2 (0.1)	0.05 (0.1)^d	0.06 (0.09)^d	<0.001
Sleep stage transitions	143.6 (31.5)	161.6 (35.3)	167.2 (42.0)	NS
No. of awakenings	6.1 (5.2)	17.1 (6.4)^b	19.9 (8.0)^b	<0.001
From S1	2.5 (2.4)	4.4 (3.2)	6.8 (5.3)^e	0.004
From S2	2.7 (2.0)	6.9 (3.9)^d	7.3 (2.9)^b	<0.001
From SWS	0.2 (0.4)	1.35 (1.5)^f	2.4 (1.7)^b	<0.001
From REM	0.7 (1.1)	4.5 (2.4)^b	3.4 (3.1)^g	0.001
No. of arousals	75.8 (23.0)	97.2 (45.7)	123.8 (70.8)	0.062

All significant differences were between the young controls and the pre- or postmenopausal women

Bonferroni corrected P: ^a0.030, ^b0.003, ^c0.012, ^d0.003, ^e0.006, ^f0.042, ^g0.015 compared with young controls

No differences between the pre- and postmenopausal women

Pre- and postmenopausal women had more awakenings than young women. They had more awakenings from S2, SWS and from REM sleep. In the postmenopausal group, there were also more awakenings from S1 than in the young group. In the post hoc analysis where only pre- and postmenopausal groups were compared, the groups were similar apart from fewer awakenings from S1 sleep (P = 0.046) and from SWS (P = 0.018) in the premenopausal group than in the postmenopausal group.

Subjective

According to the BNSQ, which enquired into various aspects of insomnia and sleepiness during the past three months, both pre- and postmenopausal women scored higher in insomnia than young women (premenopausal: mean 13.9 [SD 3.8], P = 0.036; postmenopausal: mean 16.2 [SD 3.6], P < 0.003 and young: mean 10.4 [SD 2.3]). The sleepiness scores did not differ (premenopausal: mean 10.5 [SD 3.9], postmenopausal: mean 10.8 [SD 3.3] and young: mean 10.0 [SD 2.4]). In the post hoc analysis, postmenopausal women scored higher in insomnia than premenopausal women (P = 0.023). Again there was no difference in the sleepiness scores.

No difference between the three groups was found on the VAS of sleepiness and mood or on the Stanford sleepiness scale (data not shown). In addition, the groups were similar according to the subjective sleep variables of the morning questionnaire (Table 3). Three questions about subjective sleep were used for comparisons with objective variables (sleep latency, efficiency and awakenings). Only two associations were found. The subjective and objective sleep latency correlated in the premenopausal group (r = 0.495, P = 0.026) and subjective and objective sleep efficiency in the postmenopausal group (r = −0.534, P = 0.003).

Table 3

Self-reported sleep characteristics of the study night, expressed as percentages of answers

Characteristic	Percentage
	Young	Premenopausal	Postmenopausal
Sleep latency
Fell asleep soon	9.1	30.0	17.9
Quite soon	90.9	50.0	60.7
Only after a long time	0.0	20.0	21.4
Sleep efficiency
Sleft excellently	18.2	15.0	17.9
Well	54.5	40.0	32.1
Rather well	27.3	25.0	32.1
Poorly	0.0	15.0	17.9
Very poorly	0.0	5.0	0.0
Awakenings
None	9.1	15.0	7.1
Once	54.5	40.0	28.6
Several times	36.4	45.0	64.3
Sleep quality
Better	36.4	15.0	32.1
Same	54.5	55.0	57.1
Worse than usual	9.1	30.0	10.7
Morning awakening
Normal	100.0	80.0	85.7
Earlier than usual	0.0	20.0	14.3
Morning tiredness
Fresh	9.1	10.0	14.3
Quite fresh	72.7	55.0	60.7
Quite tired	18.2	30.0	21.4
Tired	0.0	5.0	3.6

No significant differences in the answers between the study groups

Cognition

All groups performed well in the cognitive tests. In the SRT test, postmenopausal women had longer reaction times (P = 0.009) compared with young women (premenopausal mean 297.0 ms [SD 34.6], postmenopausal mean 312.1 ms [SD 42.5], young mean 271.1 ms [SD 30.7]). In the 2-CRT test, premenopausal women made fewer errors than postmenopausal or young women (premenopausal mean 0.1 [SD 0.3]; postmenopausal mean 0.5 [SD 0.6], P = 0.042; young mean 0.6 [SD 0.7], P = 0.021). In the same test, both pre- and postmenopausal women had longer reaction times compared with young women (premenopausal mean 489.3 ms [SD 67.4], P < 0.003; postmenopausal mean 521.4 ms (SD 79.4), P < 0.003; young mean 387.8 ms [SD 29.9]). In the test of vigilance, all women had similar rates of omissions (premenopausal mean 1.0% [SD 1.8]; postmenopausal mean 1.0 % (SD 1.8); young mean 0.9% [SD 1.1]) but premenopausal women made fewer errors (P = 0.048) and postmenopausal women had longer reaction times (P = 0.015) than young women (errors: premenopausal mean 1.2 [SD 1.0], postmenopausal mean 3.0 [SD 3.5], young mean 2.5 [SD 1.5]; reaction times: premenopausal mean 517.1 ms [SD 59.6], postmenopausal mean 527.8 ms [SD 48.2], young mean 484.5 ms [SD 33.2]).

The associations between selected objective sleep variables (sleep efficiency, sleep latency, REM latency, S2%, SWS%, REM%, SWA, number of awakenings) and cognitive variables (SRT, 2-CRT, 2-CRT errors, vigilance RT, vigilance errors, vigilance omissions) were studied in the three groups. The correlations were inconsistent and did not systematically support the initial hypotheses. In the premenopausal group, the error rate in 2-CRT correlated to the number of awakenings (r = −0.507, P = 0.022) and in vigilance, the number of errors correlated to the percentage of REM (r = −0.518, P = 0.019) and RT to REM latency (r = −0.469, P = 0.037). In the postmenopausal group, the errors in 2-CRT correlated to REM latency (r = 0.391, P = 0.040). The rate of omissions in vigilance correlated to the percentage of SWS (r = 0.413, P = 0.029) and to the percentage of REM (r = −0.483, P = 0.009). Also, the number of errors in vigilance correlated to the percentage of REM (r = −0.423, P = 0.025) and to REM latency (r = 0.383, P = 0.044). RT in vigilance correlated to the SWA in the second NREM episode (r = −0.412, P = 0.029) and in the fourth NREM episode (r = −0.559, P = 0.010). In the young group, three correlations were found: SRT correlated to the percentage of SWS (r = −0.774, P = 0.005) as well as to the SWA of the second NREM episode (r = −0.644, P = 0.044) and errors in vigilance correlated to sleep latency (r = 0.941, P < 0.001).

Discussion

Our results confirm earlier observations that postmenopausal women are less satisfied with their sleep than their premenopausal counterparts. However, neither direct (polygraphic sleep recordings) nor indirect measures of sleep disturbances (cognitive function tests, assessment of sleepiness and mood) revealed any clue of possible mechanisms or consequences of the poor sleep perceived by postmenopausal women. Major impairments in objective sleep quality were observed only when comparing pre- and postmenopausal women with young women. Thus, it seems that age rather than the menopausal state per se contributes to these changes.

The vast literature of questionnaire studies indicates that postmenopausal women suffer more from disturbed sleep than premenopausal women.^1–3,9 There are only a few exceptions. In the study by Sharkey et al. ⁸ subjective sleep ratings were independent of menopausal state,⁸ and further, in Owens and Matthews' study⁶ trouble with sleeping was associated with higher levels of anxiety, depression, stress, tension and public self-consciousness rather than menopausal status. We used two different sleep questionnaires and gained partly conflicting results. When evaluating the sleep quality of the one study night using the morning questionnaire, the three study groups were similar, but based on the BNSQ that assessed sleep quality during the past three months, postmenopausal women had worse sleep than premenopausal women. No difference in the scores of depression, mood or the perceived quality of life was noticed and thus these could not explain the variance in insomnia scores. No doubt our postmenopausal women experienced some disturbance in their sleep but the lack of increased daytime sleepiness suggests low clinical impact.

In the literature, few polysomnographic sleep studies have focused on the menopausal state per se: the majority of the studies conducted in menopausal women have addressed the effect of HT on sleep. Our results were in line with those of Shaver et al. ⁷ who evaluated sleep parameters in 76 women of whom 20 were pre-, 32 peri- and 24 postmenopausal and found no differences based on the menopausal state. In two more recent studies, the results showed that overall sleep architecture was worse in premenopausal women compared with postmenopausal women. Sharkey et al. ⁸ studied 13 premenopausal and 12 postmenopausal women and found that premenopausal women had more S1 sleep and longer SWS latency than postmenopausal women.⁸ In the largest study up to now by Young et al. ⁹ with 1024 sleep recordings (493 in the premenopausal group and 226 in the postmenopausal group), premenopausal women had lower sleep efficiency, less SWS and more S2 sleep than postmenopausal women. These latter studies, rather than interpreting that menopause improves sleep, disagree with the hypothesis drawn from subjective studies that postmenopausal women have worse sleep than premenopausal women. The fact that the findings were in different sleep variables further complicates the interpretation of the influence of the menopause on sleep. Variation in sample sizes could partly explain these different findings. Although our study groups are small, possibly causing bias, they are comparable with most previous studies.

Although premenopausal and postmenopausal women objectively had worse sleep than young women, the subjective sleep qualities of the corresponding night appeared to be similar. As a result, the correlations between subjective and objective sleep variables were scarce. This sleep misperception is a common finding.^9,27 It is possible that cortical events detected with polysomnography, the state-of-the-art method for studying sleep, may not be sensitive enough to detect alterations in deeper brain structures, such as autonomic nervous system centres.²⁸ In addition, the subjective variables we used were categorical and, therefore, may not be the most appropriate ones for correlations with objective sleep variables. It is also possible that with more symptomatic subjects, objective correlates of subjective symptoms would have become evident. The sleep complaints of menopausal women should also be viewed in relation to climacteric symptoms. Subjective sleep appears to be worsened in women with hot flushes^2,4,29,30 – the most common climacteric symptom – but polysomnographic sleep changes are more heterogeneous and less well-studied.^5,7,8,29,30 In our study, the subjective vasomotor symptom scores did not differ across study groups, confirming that the differences in objective sleep quality between the young controls and the two older groups were not dependent on the vasomotor sensations.

The idea of adding a young women's group into the study was to put the differences between pre- and postmenopausal women into a perspective of natural evolution. We were able to demonstrate that the major changes in sleep architecture occur before, not after, menopause. The objective and subjective sleep quality of premenopausal women was closer to that of postmenopausal women than that of young women. Studies exactly like ours are not available but our results are in line with the results of Lukacs et al. ³¹ whose study was conducted under sleep challenge.³¹ We could also reproduce the well-documented age-related sleep changes, such as the worsening of the overall sleep architecture and decline in slow wave sleep, between young women and the older groups.^32–35 The young women in our study were using OC. Earlier studies have reported less SWS in women using OC compared with naturally cycling women, but only in the luteal phase.³⁶ We evaluated the women in the follicular phase during which no differences have been found between the OC users and naturally cycling women. Furthermore, the age difference compared with the other study groups was so notable that the possible effect of OC use was expected to be minimal. It has previously also been reported that OC use has no effect on the reaction speed, sustained attention or alertness, which were precisely the variables we were interested in.³⁷

Good sleep is often considered as one of the basic sources of better cognitive performance. However, the effect of sleep architecture on cognitive performance is not consistent .^13,14,38 Some authors associate better attention with a higher representation of S2 sleep,³⁸ whereas others regard SWS and REM sleep as important.^39,40 In our study, most of the observed correlations between sleep and cognition appeared to be random. Only in postmenopausal women, more REM sleep or earlier REM onset was more consistently associated with better cognitive performance. The REM latency tends to shorten with increasing age, possibly contributing to this association. The discrepancy in study outcomes results partly from the fact that cognition is not a single issue, but refers to a variety of functions. We focused on attention since it is perceived as an essential cognitive function and deficits often cause concern to a person suffering from sleep problems. We confirmed the earlier reported slowing of reaction speeds with increasing age.^11,39,41 However, premenopausal women made fewer errors than either postmenopausal or young women, although the total number of errors was minimal in all groups. It is possible that young women emphasized speed over accuracy and premenopausal women concentrated on accuracy.

In conclusion, insomnia complaints increase at menopause. However, the objectively measured sleep of premenopausal women seems to depend on age more than on female sex hormone status. The major impairment in objective sleep quality seems to occur even before the middle age, and the menopausal state per se does not appear to essentially contribute to this. However, the longitudinal studies across menopause could address this better. In a healthy population, this deterioration in sleep quality does not seem to affect the general wellbeing or daytime sleepiness. The decrease in cognitive performance seems to be age-related rather than sleep-dependent. We suggest that not all sleep problems in the menopausal transition seem to be of menopausal origin and thus women in this age group presenting with insomnia complaints are a challenge for the physician. A careful evaluation of other contributing factors is essential.

Footnotes

Acknowledgements

This study was financially supported by the European Commission Grant (QLK6-CT-2000-00499) and the Finnish Sleep Research Society. We thank Tero Vahlberg, MSc, for statistical assistance and Yvonne Bearne, MA, for revising the language. The study had the approval of the Ethical Committees of Turku University Hospital and the University of Helsinki.

Competing interests

None declared.

Appendix 1

Appendix 2

References

Leger

, Quilleminault

, Dreyfus

, Delahaye

, Paillard

. Prevalence of insomnia in a survey of 12,778 adults in France. J Sleep Res 2000;9:35–42

Kravitz

, Ganz

, Bromberger

, Powell

, Sutton-Tyrrell

, Meyer

. Sleep difficulty in women at midlife: a community survey of sleep and the menopausal transition. Menopause 2003;10:19–28

Kuh

, Wadsworth

, Hardy

. Women's health in midlife: the influence of the menopause, social factors and health in earlier life. Br J Obstet Gynaecol 1997;104:923–93

Shaver

JLF

, Paulsen

. Sleep, psychological distress, and somatic symptoms in perimenopausal women. Fam Pract Res J 1993;13:373–84

Freedman

, Roehrs

. Lack of sleep disturbance from menopausal hot flashes. Fertil Steril 2004;82:138–44

Owens

, Matthews

. Sleep disturbance in healthy middle-aged women. Maturitas 1998;30:41–50

Shaver

, Giblin

, Lentz

, Lee

. Sleep patterns and stability in perimenopausal women. Sleep 1988;11:556–61

Sharkey

, Bearpark

, Acebo

, Millman

, Cavallo

, Carskadon

. Effects of menopausal status on sleep in midlife women. Behav Sleep Med 2003;1:69–80

Young

, Rabago

, Zgierska

, Austin

, Finn

. Objective and subjective sleep quality in premenopausal, perimenopausal, and postmenopausal women in the wisconsin sleep cohort study. Sleep 2003;26:667–72

10.

Lezak

. Neuropsychological Assessment. New York: Oxford University Press, Inc, 1995

11.

Wilkinson

, Allison

. Age and simple reaction time: Decade differences for 5,325 subjects. J Gerontol 1989; 44:29–35

12.

Samkoff

, Jacques

. A review of studies concerning effects of sleep deprivation and fatigue on residents' performance. Acad Med 1991;66:687–93

13.

Smith

, McEvoy

, Gevins

. The impact of moderate sleep loss on neurophysiological signals during working-memory task performance. Sleep 2002;25:56–66

14.

Bastien

, Fortier-Brochu

, Rioux

, LeBlanc

, Daley

, Morin

. Cognitive performance and sleep quality in the elderly suffering from chronic insomnia relationship between objective and subjective measures. J Psychosom Res 2003;54:39–49

15.

McEwen

, Alves

. Estrogen action in the central nervous system. Endocr Rev 1999;20:279–307

16.

Jones

, Kryger

, Roth

, Dement

. Basic mechanisms of sleep-wake states. In: Principles and Practice of Sleep Medicine. Philadelphia: W.B. Saunders Company, 2000:134–154

17.

Sherwin

. Estrogen and cognitive functioning in women. Endocr Rev 2003;24:133–51

18.

Rechtschaffen

, Kales

. A Manual of Standardized Terminology, Techniques and Scoring Systems for Sleep Stages of Human Subjects. Los Angeles, CA: Brain Information Service/Brain Research Institute, UCLA, 1968

19.

Shneerson

. Handbook of Sleep Medicine. Cambridge: Blackwell Science Ltd, 2000

20.

The Atlas Task Force: Bonnet

, Carley

, Carskadon

, ASDA report. EEG arousals: Scoring rules and examples. Sleep 1992;15:173–84

21.

Feinberg

, Floyd

. Systematic trends across the night in human sleep cycles. Psychophysiology 1979;16:283–91

22.

Heinzer

, Gaudreau

, DÃ©cary

, Sforza

. Slow-wave activity in sleep apnea patients before and after continuous positive airway pressure treatment. Chest 2001;119:1807–13

23.

Hoddes

, Zarcone

, Smythe

, Phillips

, Dement

. Quantification of sleepiness: a new approach. Psychophysiology 1973;10:431–6

24.

Partinen

, Gislason

. Basic nordic sleep questionnaire (BNSQ): a quantitated measure of subjective sleep complaints. J Sleep Res 1995;4:150–5

25.

Ohinmaa

, Sintonen

, Badia

, Herdman

, Segura

. Quality of life of Finnish population as measured by EuroQol. In: EuroQol, Plenary Meeting: Barcelona, Spain, 3–5 October 1995: Discussion Papers. Barcelona: Catalan Institute of Public Health, 1996:161–72

26.

Revonsuo

, Portin

. CogniSpeed: The Computer Based Measurement of Cognitive Processing. Turku, Finland: Aboatech LTD, University of Turku, 1995

27.

Baker

, Maloney

, Driver

. A comparison of subjective estimates of sleep with objective polysomnographic data in healthy men and women. J Psychosom Res 1999;47:335–41

28.

Polo

. Sleep in postmenopausal women: better sleep for less satisfaction. Sleep 2003;26:652–3

29.

Polo-Kantola

, Erkkola

, Irjala

, Helenius

, Pullinen

, Polo

. Climacteric symptoms and sleep quality. Obstet Gynecol 1999;94:219–24

30.

Freedman

, Roehrs

. Effects of REM sleep and ambient temperature on hot flash-induced sleep disturbances. Menopause 2006;13:576–83

31.

Lukacs

, Chilimigras

, Cannon

, Dormire

, Reame

. Midlife women's responses to a hospital sleep challenge: ageing and menopause effects on sleep architecture. J Women's Health 2004;13:333–40

32.

Bliwise

. Sleep in normal ageing and dementia. Sleep 1993;16:40–81

33.

Gaudreau

, Carrier

, Montplaisir

. Age-related modifications of NREM sleep EEG: from childhood to middle age. J Sleep Res 2001;10:165–72

34.

Miles

, Dement

. Sleep and ageing. Sleep 1980;3:119–220

35.

Reynolds

, Monk

, Hoch

, Electroencephalographic sleep in the healthy ‘old old’: a comparison with the ‘young old’ in visually scored and automated measures. J Gerontol 1991;46:M39–46

36.

Baker

, Waner

, Vieira

, Taylor

, Driver

, Mitchell

. Sleep and 24 hour body temperatures: a comparison in young men, naturally cycling women and women taking hormonal contraceptives. J Physiol (Lond) 2001;530:565–74

37.

Wright

, Badia

. Effects of menstrual cycle phase and oral contraceptives on alertness, cognitive performance, and circadian rhythms during sleep deprivation. Behav Brain Res 1999;103:185–94

38.

Walker

, Liston

, Hobson

, Stickgold

. Cognitive flexibility across the sleep-wake cycle: REM-sleep enhancement of anagram problem-solving. Cognitive Brain Res 2002;14:317–24

39.

Gottsdanker

. Age and simple reaction time. J Gerontol 1982;37:342–8

40.

Stickgold

, Scott

, Rittenhouse

, Hobson

. Sleep-induced changes in associative memory. J Cogn Neurosci 1999;11:182–93

41.

Fontani

, Lodi

, Felici

, Corradeschi

, Lupo

. Attentional, emotional and hormonal data in subjects of different ages. Eur J Appl Physiol 2004;92:452–61