Sleep
13(5):403-409, Raven Press, Ltd., New York
© 1990 Association of Professional Sleep Societies
Sleep Patterns Related to Menstrual Cycle
Phase and Premenstrual Affective Symptoms
Kathryn A. Lee, Joan F. Shaver, Elizabeth C. Giblin, and Nancy F. Woods
Department of Physiological Nursing, University of Washington, Seattle, Washington, U.S.A.
Summary: An ovulatory menstrual cycle is characterized by fluctuating levels
of progesterone. Progesterone, a gonadal hormone known for its soporific and
thermogenic effects, is present in negligible levels prior to ovulation and in high
levels after ovulation. To describe and compare sleep patterns in relation to
ovulatory cycles and premenstrual mood state, sleep was monitored in healthy
women at two phases of the menstrual cycle. Results indicated that rapideye-movement (REM) latency was significantly shorter during the postovulatory (luteal) phase compared to the preovulatory (follicular) phase, but there
was no significant difference in latency to sleep onset or the percentage of
REM sleep. While there were no menstrual cycle phase differences in the
percentages of various sleep stages, the women with negative affect symptoms
during the premenstruum demonstrated significantly less delta sleep during
both menstrual cycle phases in comparison with the asymptomatic subjects.
Key Words: Women's sleep-Progesterone-Menstrual cycle-Premenstrual
symptoms.
In both males and females, progesterone is secreted by the adrenal cortex and accounts for the low plasma concentrations of -1 mg/day. For females, however, luteinization of the granulosa cells in the ovary results in additional synthesis of high levels
of progesterone during the ovulatory and postovulatory (luteal) phases of the menstrual
cycle. As a result, the production rate of progesterone reaches 25 mg/day at its peak
during the midluteal phase, and the level begins to decline at various rates during the
premenstrual week (1). It is the thermogenic effect of this hormone that allows women
to utilize basal body temperatures as an indicator of ovulation (2,3).
The soporific effect of progesterone has been documented under various experimental conditions in animal models (4-7) and in various clinical situations with both males
and females (8-10). Yet, little attention has been given to the effects of its monthly
fluctuation in sleep studies involving women of child-bearing age (11). One purpose of
this study was to compare sleep patterns during the follicular phase, when progesterone
levels are negligible, with sleep patterns during the luteal phase, when progesterone
Accepted for publication March 1990.
Address correspondence and reprint requests to Dr. K. A. Lee at Department of Family Health Care
Nursing, N4I1Y, University of California, San Francisco, San Francisco, CA 94143-0606, U.S.A.
403
404
K. A. LEE ET AL.
levels and core body temperatures are significantly elevated. The second purpose of the
study \-vas to describe hovi sleep patterns of women who experience premenstrual
negative affect symptoms might differ from those who do not experience symptoms.
METHODS
A convenience sample, consisting of 17 healthy and presumably ovulating women
between 25 and 35 years of age, volunteered to participate in a study of sleep and
temperature rhythm patterns and were paid $75 upon completion. They had a history of
regular, predictable menstrual cycles of between 25 and 35 days in length. With the
exception of two subjects taking daily thyroid replacement, all subjects were free of
medications during the study. Subjects had not been pregnant or lactating for a period
of3 months prior to the study. Three additional subjects taking oral contraceptives, and
thus anovulatory, were also monitored for the purpose of establishing the ovulatoryanovulatory menstrual cycle criteria using salivary progesterone (P sal) levels and mean
rectal temperature (Tr) values. Alterations in the circadian rhythm for core temperature
during the follicular and luteal phases of the menstrual cycle have been reported elsewhere for these subjects (12).
After giving informed consent and receiving an orientation to the sleep laboratory,
subjects began the study at either the follicular (days 3-10 after menses onset) or the
luteal (3-12 days before onset of menses) phase of their cycle, whichever occurred at
the first mutually available opportunity. Subjects were instructed to maintain their
normal daily routine during each phase of the study. Daily diary recordings of meals,
activity, and sleep were evaluated for compliance with instructions.
Arriving ~ 1 h prior to their desired bedtime and awakening at their desired times,
they spent 2 nights at each phase in the sleep laboratory. For each subject, lights-out
time for the second menstrual phase occurred within 15 min of lights-out time for the
first menstrual phase. Electroencephalogram, electrooculogram, and electromyogram
were recorded continuously and scored in 30-s epochs according to standardized criteria (13). The sleep scorer remained blind to subjects' menstrual cycle phase until
completion of the data analysis phase. A 10% random sampling for interrater reliability
was consistently above 0.90 with another sleep researcher and another laboratory.
Upon awakening each morning in the sleep laboratory, subjects were asked to produce a 3-ml sample of unstimulated saliva. Samples were frozen at - 20°C until completion of the study when one batch radioimmunoassay (RIA) was performed. A Coata-Count (Diagnostic Products) RIA kit, utilizing a solid-phase, no-extraction protocol
with 125I-Iabeled progesterone to compete for antibody sites, was used to assay for P sal '
In a previous pilot study (K. A. Lee, unpublished data), saliva and serum samples for
women during both follicular and luteal phases were significantly correlated (r = 0.92).
Samples were analyzed in duplicate and mean P sal values for the 2 days at each menstrual cycle phase were used in further analyses. In conjunction with mean P sal values,
the day of the menstrual cycle and Tr were also used to document follicular and ovulatory luteal phases.
To quantify menstrual cycle negative affect symptomatology without biasing the
results, subjects were asked to complete the Profile of Mood States (POMS) before
sleep onset and upon awakening. Subjects assumed that changes in these evening and
morning data would be used in a study of sleep's effect on daytime mood. Approximately 30 min prior to lights out, subjects completed the POMS, rating 65 adjectives
Sleep, Vol. 13, No.5, 1990
MENSTRUAL CYCLE SLEEP PATTERNS
405
according to how they felt "right now" on a scale of "not at all" (score of 0) to
"extremely" (score of 4). The POMS contains six subscales: Tension-Anxiety, Depression-Dejection, Anger-Hostility, Vigor-Activity, Fatigue-Inertia, and ConfusionBewilderment (14). A total evening POMS score was calculated for each subject at each
phase by subtracting the Vigor-Activity score from the sum of the other five subscales.
POMS subscale scores had internal consistency reliabilities ranging from 0.77 to 0.93,
with the total POMS scores reaching a Cronbach alpha coefficient of 0.90 during the
follicular phase and 0.81 during the luteal phase.
In addition to the POMS, subjects completed a 7-day diary that contained space to
rate both positive and negative symptoms as modified from the Woods Women's
Health Diary (15). This diary included 22 negative affect symptoms rated on a scale of
not present (score of 0) to extreme (score of 4). Like the POMS, the diary was completed within 30 min of bedtime and upon awakening. It was also completed on each of
7 days, during which time 2 nights included the sleep laboratory experience. A total
negative affect score for 7 entire evenings was calculated for each subject.
Based on the total 7-day diary scores during the follicular and luteal week or the
evening total POMS scores during the follicular and luteal sleep laboratory experience,
subjects were categorized as either premenstrually symptomatic or asymptomatic. The
symptomatic subjects were defined as those with at least a 30% increase in total POMS
or diary scores during their luteal phase in comparison with their own follicular phase.
Analysis
Subjects failing to meet ovulation criteria for significant menstrual cycle phase differences in P sal and Tr were eliminated from analyses related to changes in sleep
patterns between the two phases of the menstrual cycle. The three subjects taking oral
contraceptives, and thus anovulatory, had a cycle phase difference in P sal of < 100
pmol/L. Therefore, one criterion for ovulatory cycles was a P sal difference of > 150
pmol/L between cycle phases.
Two subjects were eliminated because subsequent menses onset occurred 2 days
earlier than predicted and they had <50 pmol/L difference in P sal . A third subject was
eliminated because of a substantial change in ambient temperature in the sleep laboratory between her two menstrual cycle phases, resulting in mean Tr levels at extreme
outlying levels. A fourth subject was eliminated after she reported an evening of dancing and alcohol consumption prior to coming into the sleep laboratory during one of the
two menstrual cycle phases. The mean P sal values and circadian rhythm-adjusted mean
(mesor) Tr values are presented in Table 1 for the 13 remaining subjects included in
subsequent analyses of cycle phase differences in sleep patterns.
Analyses for menstrual cycle phase comparisons consisted of Wilcoxon matchedpairs, signed-ranks tests for within-subject comparisons between follicular and luteal
cycle phases. Mann-Whitney U analyses were performed for asymptomatic and symptomatic group comparisons.
RESULTS
Sleep pattern changes related to menstrual cycle phase
The 13 subjects in the within-subject comparisons of sleep characteristics between
menstrual cycle phases had total sleep times ranging from 308 to 482 min. The only
significant finding (Wilcoxon Z = 1.99, p < 0.05) was a shorter REM latency during the
Sleep, Vol. 13, No.5, 1990
K. A. LEE ET AL.
406
TABLE 1. Salivary progesterone (P sa ') levels and
rhythm-adjusced mean rectal iemperaiUres (T,) at two
phases of the menstrual cycle (n = 13)
P saJ (pmoI/L)
Mean ± SD
Median
Range
Tr Cc)
Mean ± SD
Median
Range
Follicular
Luteal
216.0 ± 70.57
201
107-381
526.4 ± 183.78a
457
315-950
36.93 ± 0.072
36.93
36.80-37.04
37.26 ± 0.141a
37.29
37.00-37.46
a Significantly different from follicular phase: Wilcoxon Z = 3.18,
p < 0.001.
luteal phase (93.5 ± 33.8 min) when compared to the follicular phase (109.8 ± 38.5 min).
Rapid-eye-movement (REM) latency was defined as the number of minutes from stage 1
sleep onset to the first full minute of REM sleep. There were no significant menstrual
cycle phase differences in total sleep time, sleep onset latency to stage 1 or stage 2,
percentages of sleep period time (SPT) spent in various stages of REM and non-REM
(NREM) sleep, or sleep efficiency (see Table 2).
In the three women taking oral contraceptives and not experiencing ovulation or
changes in core body temperature, REM latencies were short, falling into a narrow
range of 54-66 min, and varied by <5 min between the follicular and luteal phase. There
were no other noticeable differences in sleep for these three women when compared to
the group of ovulating women.
Sleep patterns associated with premenstrual negative affect symptoms
Ofthe 13 women, 6 were categorized as asymptomatic and 7 met the POMS or diary
criteria for inclusion in the luteal phase symptomatic group. Table 3 contains the sleep
variables for asymptomatic and symptomatic women during both the follicular and the
luteal phases. When data were analyzed by group, there was a significant difference (p
< 0.001) in the percentage of delta sleep (stages 3 and 4) between the asymptomatic and
TABLE 2. Sleep pattern characteristics (mean ± SD) at two phases of
the menstrual cycle (n = 13)
Follicular
Latency to
Stage 1 (min)
Stage 2 (min)
REM (min)
Time spent (% SPT) in
Stage 0
Stage 1
Stage 2
Stage 3--4
REM
Total sleep time (min)
Sleep efficiency (% TST/TIB)
6.9 ± 3.9
14.4 ± 7.4
109.8 ± 38.5
5.3
12.2
52.6
10.0
19.9
417.8
94.6
± 3.5
±
±
±
±
±
±
5.3
7.9
6.9
4.0
44.3
0.5
Luteal
7.8 ± 4.2
13.2 ± 5.5
93.5 ± 33.8 a
5.7
11.0
51.8
10.0
20.6
417.7
94.0
± 4.9
± 4.1
± 7.9
± 6.8
± 3.8
± 49.4
± 0.5
SPT, sleep period time; TST, total sleep time; TIB, time in bed.
a Significantly different from follicular phase: Wilcoxon Z = 1.99, p < 0.05.
Sleep, Vol. 13, No.5, 1990
MENSTRUAL CYCLE SLEEP PATTERNS
407
TABLE 3. Sleep pattern characteristics (mean ± SD) during the two phases of the menstrual
cycle for the asymptomatic and symptomatic group at both the follicular (F) and the luteal (L)
menstrual cycle phases
Menstrual
cycle
phase
Sleep onset latency (min) to
Stage I
F
L
Stage 2
F
L
REM
F
L
Percentage of SPT in
Stage 0
F
L
Stage I
F
Stage 2
F
L
L
Stage 3-4
F
L
REM
F
L
Total sleep time (min)
F
L
Sleep efficiency index
(% TSTffIB)
F
L
Asymptomatic
(n = 6)
Symptomatic
(n = 7)
7.9
10.1
17.3
14.8
103.3
84.4
±
±
±
±
±
±
3.3
4.0
9.3
4.9
35.3
17.9
6.1
5.9
12.2
11.8
11504
101.2
±
±
±
±
±
±
4.3
3.5 a
4.8
6.1
43.0
43.2
504
4.3
12.7
9.9
46.8
48.3
15.6
15.5
19.4
22.0
420.3
434.5
4.6
1.9
7.0
3.7
7.1
4.6
4.7
5.1
2.8
± 3.9
± 44.3
± 30.9
5.2
7.0
11.7
11.9
57.6
54.7
5.1
404
2004
21.5
415.6
403.6
±
±
±
±
±
±
±
±
±
±
±
±
2.7
6.4
3.8
4.6
4.6 b
9.1
4.lc
2.5 c
4.9
4.1
56.8
51.5
±
±
±
±
±
±
±
±
±
94.5 ± 0046
95.6 ± 0.20
94.7 ± 0.28
92.7 ± 0.67
SPT, sleep period time; TST, total sleep time; TIB, time in bed.
Significantly different from asymptomatic group: aMann-Whitney U, p < 0.05; bMann-Whitney U, p <
0.01; cMann-Whitney U, p < 0.001.
symptomatic subjects at both the follicular and the luteal phases of their menstrual
cycles. The asymptomatic group had a mean of 15.5 ± 5.1% delta sleep, while the
symptomatic group had an average of only 4.4 ± 2.5% during the luteal phase. This
finding was similar in the follicular phase.
In addition to delta sleep differences, there was also a significant difference (p =
0.05) between the groups in sleep onset latency to the first full minute of stage 1. The
latency for the asymptomatic group was 10.1 ± 4.0 min and the latency for the symptomatic group was 5 min shorter (5.9 ± 3.5 min) during the luteal phase. The same
pattern was also evident during the follicular phase, although statistical significance
was not reached. The symptomatic group also experienced an average of 7-10% more
stage 2 sleep during both phases of the menstrual cycle. During the follicular phase, this
difference was statistically significant (p < 0.01), whereas only a trend was revealed
during the luteal phase (p = 0.18).
DISCUSSION
Little information exists in the literature with which to compare findings from this
study. Williams et al. (16, p. 43) acknowledged the menstrual cycle's potential influence
on sleep patterns by studying adult women only during their follicular phase. Follicular
phase data from this current study closely resemble those reported by Williams and
colleagues for sleep onset latency, REM latency, sleep efficiency, and the percentages
Sleep, Vol. 13, No.5, 1990
408
K. A. LEE ET AL.
ofNREM sleep. However, they reported 25% REM sleep, only 0.5% stage 0, and only
4% stage 1 sleep.
In a study of sleep-related breathing in 11 healthy women, Stahl and colleagues (17),
in contrast to this study, reported only 13% REM sleep, more awake time, and a low
sleep efficiency index without any demonstrable differences between the two menstrual
cycle phases. These discrepancies may be explained by the lack of a night's adaptation
to the sleep laboratory, the imposed limitation on their total sleep time due to awakening at 0530 h, or their extensive monitoring with equipment for assessing the respiratory parameters.
While it is difficult to generalize from this small sample, findings suggest that sleep
patterns, particularly altered REM latency, might indeed be affected by the phase of the
menstrual cycle. The significantly shorter luteal phase REM latency for all ovulating
subjects might be a function of progesterone's thermogenic effect. A relationship between REM latency and core body temperature has been established in both freerunning (18) and clinical (19) situations.
In a previous report of circadian temperature rhythms for some of the women in this
study (12), it was found that temperature rhythm amplitudes were dampened and mesors were elevated during the luteal phase when compared to the follicular phase. In
these entrained subjects, however, no difference in temperature rhythm acrophases
between the two menstrual cycle phases was apparent. Altered REM latency, in concert with the menstrual cycle change in temperature amplitUde and no change in the
temperature acrophase, adds further support for the circadian amplitude hypothesis
(20). In this hypothesis, a dampened amplitude of the circadian rhythm is thought to be
responsible for lowering the threshold for REM sleep at sleep onset.
The finding of no menstrual cycle phase difference in REM sleep in this study supports earlier work by Ho (21) and Cluydts and Visser (22) but conflicts with an earlier
report by Hartmann who concluded that "all seven women showed a tendency to have
higher D-times [REM sleep] late in the cycle" (23, p. 413). Yet in comparing data from
each menstrual cycle (23, p. 409), there was a great deal of variance in REM sleep. In
one subject, for example, REM for the luteal phase ranged from 15.2% in the first cycle,
to 25% in the third cycle and 32.5% in the fourth cycle. Furthermore, three of the seven
subjects in Hartmann's study were psychiatric inpatients and one was anovulatory due
to Enovid therapy.
Overall, the 13 women in this study had amounts of stage 3--4 sleep comparable with
that reported in the literature (16, p. 52). Yet when these women were categorized by
premenstrual symptomatology, the symptomatic negative affect group had only about
one-third as much stage 3--4 sleep as the asymptomatic group, regardless of menstrual cycle phase. Although the sample size is small, these findings support data from
Cluydts and Visser (22) and suggest that women who experience premenstrual negative
affect symptoms have significantly less deep sleep than would be expected for their age
cohort, regardless of menstrual cycle phase.
In comparison to the asymptomatic group, the symptomatic group also differed significantly in sleep onset latency during the luteal phase. While the mean difference was
only 4 min and may not be of clinical relevance, falling asleep within 5 or 6 min is an
indication that the symptomatic group may be more sleep deprived than the asymptomatic group, who took an average of 10 min to fall asleep.
Until this study is replicated, sleep clinicians and researchers should be cautious in
interpreting REM latency data from women when the phase of their menstrual cycle is
Sleep, Vol. 13, No.5, 1990
MENSTRUAL CYCLE SLEEP PATTERNS
409
unknown or uncontrolled. This study also begins to provide clues to possible mechanisms for premenstrual syndrome. Symptomatic women who exhibit less slow-wave
sleep might have an underlying physiologic deficit that is accentuated by monthly
gonadal hormone fluctuations, allowing symptoms of negative affect to emerge. Alternatively, women who experience a chronic deficit in slow-wave sleep may be more
sensitive to monthly hormonal fluctuations, with that increased sensitivity manifested
as negative affect symptoms.
Acknowledgment: This study was supported by a National Research Service A ward (no.
5F3INU0567) from the Department of Health and Human Services, the American Nurses' Foundation, the Psi Chapter of Sigma Theta Tau, and a bequest from Hester McLaws to the University of Washington School of Nursing.
REFERENCES
1. Baird DT, Corker CS, Davidson DW, Hunter WM, Michie EA, VanLook PFA. Pituitary-ovarian relationships in polycystic ovary syndrome. J Clin Endocrinol Metab 1977;45:798-809.
2. Little BC, Matta RJ, Zahn TP. Physiological and psychological effects of progesterone in man. J Nerv
Ment Dis 1974;159:256-62.
3. de Mouzon J, Testart J, Lefevre B, Pouly J, Frydman R. Time relationships between basal body temperature and ovulation or plasma progestins. Ferti! SterilI984;41:254-9.
4. Selye H. Acquired adaptation to the anesthetic effect of steroid hormones. J Immunol 1941 ;41 :259--68.
5. Heuser G, Ling GM, Kluver M. Sleep induction by progesterone in the pre-optic area in cats. Electroencephalogr Clin Neurophysiol 1967;22: 122-7.
6. Kawakami M, Sawyer CH. Effects of sex hormones and antifertility steroids on brain thresholds in the
rabbit. Endocrinology 1967;80:857-71.
7. Komisaruk BR, McDonald PG. Effects of progesterone and sensory stimulation on EEG and neuronal
activity in the rat. Exp Neurol 1967;19:494-507.
8. Merryman W, Boiman R, Barnes L, Rothchild I. Progesterone "anaesthesia" in human subjects. J Clin
EndocrinoI1954;14:1567-9.
9. Blumer D, Migeon C. Hormone and hormonal agents in the treatment of aggression. J Nerv Ment Dis
1975;160:127-37.
10. Herrmann WM, Beach RC. Experimental and clinical data indicating the psychotropic properties of
progestogens. Postgrad Med J 1978;54(suppl 2):82-7.
11. Lee KA, Shaver J. Women as subjects in sleep studies: methodological issues. Sleep Res 1985;14:271.
12. Lee KA. Circadian temperature rhythms in relation to menstrual cycle phase. J Bioi Rhythms
1988;3:255-63.
13. Rechtschaffen A, Kales A, eds. A manual of standardized terminology, techniques and scoring system
for sleep stages of human subjects. Brain Information Service/Brain Research Institute, University of
California, 1968.
14. McNair DM, Lorr M, Droppleman LF. Profile of Mood States manual. San Diego: Educational and
Industrial Testing Service, 1971.
15. Woods NF. Premenstrual symptoms: another look. Public Health Rep [Suppn 1987;July/August: 106-12.
16. Williams RL, Karacan I, Hursch DJ. The Electroencephalography (EEG) of human sleep: clinical applications. New York: John Wiley & Sons, 1974.
17. Stahl ML, Orr WC, Males JL. Progesterone levels and sleep-related breathing during menstrual cycles
of normal women. Sleep 1985;8:227-30.
18. Czeisler CA, Zimmerman JC, Ronda JM, Moore-Ede MC, Weitzman ED. Timing of REM sleep is
coupled to the circadian rhythm of body temperature in man. Sleep 1980;2:329-46.
19. Avery D, Wildschiodtz G, Rafaelsen O. REM latency and temperature in affective disorder before and
after treatment. Bioi Psychiatry 1982;17:463-70.
20. Schulz H, Lund R. On the origin of early REM episodes in the sleep of depressed patients: a comparison
of three hypotheses. Psychiatry Res 1985;16:65-77.
21. Ho A. Sex hormones and the sleep of women. Sleep Res 1972;1: 184.
22. Cluydts R, Visser P. Mood and sleep. 1. Effects of the menstrual cycle. Waking Sleeping 1980;4:193-7.
23. Hartmann E. Dreaming sleep (the D-state) and the menstrual cycle. J Nerv Ment Dis 1966;143:406-16.
Sleep, Vol. /3, No.5, 1990