Effects of hormone therapy with estrogen and/or progesterone on sleep pattern in postmenopausal women

1. Introduction 

Sleep disturbances increase in frequency as women approach and pass through menopause. These include insomnia [1], poor sleep efficiency [2], difficulty in maintaining sleep [3], breathing irregularity [4], [5], and hot flashes. The close relationship between sleep problems and decreased levels of reproductive hormones in menopausal women has suggested hormone therapy (HT) for their relief, with good results, especially because hot flashes are reduced [6]. The decreased production of progesterone after menopause has also been associated with breathing irregularities during sleep [7], likely the result of a decreased respiratory drive [8], and micronized progesterone has been shown to induce a better quality of sleep than medroxyprogesterone acetate when associated with estrogens [9].

Few studies have objectively assessed sleep quality in postmenopausal women undergoing HT. Yet, studies that address the relationship between sleep quality and reproductive hormone levels in postmenopausal women are not only important for understanding the neuroendocrinology of this period of women's lives, but may also shed light on the distinct role each hormone plays in the major changes associated with menopause.

This single-center, prospective, placebo-controlled study was designed to determine the effects of synthetic conjugated estrogens and medroxyprogesterone acetate on subjective and objective sleep quality in postmenopausal women. In addition, we assessed clinical and psychological complaints as well as reproductive hormone levels before and after treatment with estrogen, progesterone, or a combination of the two.

3. Results 

All participants completed the study protocol. Baseline characteristics and reproductive hormone values are shown in Table 1. The 2 groups consisted of participants matched for age, BMI, FSH level, ESS score, and sleep complaints, and were not significantly different for anthropometric characteristics, years since menopause, clinical complaints, KI score, or PSG results.

In both groups, the total KI score was markedly lower at testings 2 and 3 than at testing 1 (Table 2). Although the score was slightly higher at testing 3 than at testing 2 for group 1 (the estrogen, then estrogen plus progesterone group), the difference in total KI score between testing 1 and testing 2 or testing 3 was statistically greater for group 1 than for group 2 (the placebo, then placebo plus progesterone group).

Table 2. Analysis of Kupperman index scoresa

	Group	Testing 1(baseline)	Testing 2(after 12 wk)	Testing 3(after another 12 wk)
	Group 1	17.4 (8.6)	2.9 (2.7)b	3.1 (2.9)b
	Group 2	17.1 (9.5)	7.3 (5.2)b,c	6.7 (5.1)b,c

a Values are expressed as mean (SD). Group 1 received estrogen for 12 weeks, then estrogen plus progesterone for another 12 weeks; group 2 received placebo for 12 weeks, then placebo plus progesterone for another 12 weeks. b P<0.001 compared with the value at testing 1. c There was a statistical difference between the 2 groups.

The persons administering the tests and recording the data were blind to the participants' complaints and treatment. Although an analysis of the polysomnograms did not reveal any statistically significant differences between the 2 groups for sleep latency during the study period, fewer sleep complaints were reported in both groups after the progesterone treatment (Table 3). The main complaints were frequent arousal, snoring, restless legs, excessive somnolence during the day, and, especially, difficulty in falling asleep (of which 42.8% of the participants in group 1 and 52.6% in group 2 complained). In group 1, the prevalence of hot flash and apnea complaints was decreased at testing 2 (after estrogen treatment) (P<0.01), and the prevalence of both somnolence and restless legs complaints was decreased at testing 3 (after treatment with both estrogen and progesterone). In group 2, compared with baseline values, progesterone treatment reduced the prevalence of complaints of both hot flashes and bruxism (teeth clenching and grinding) (P<0.01) at testing 3. There were no significant differences between baseline values and values at testing 2 or testing 3 for the other sleep characteristics studied, but a smaller percentage of women reported snoring, apnea, or arousal from sleep after taking progesterone. Comparisons between the groups showed a significantly lower prevalence of hot flashes, and snoring at testing 2, and apnea at testing 3 in group 1 than group 2 (P<0.01).

Table 3. Subjective responses to a questionnaire regarding menopausal symptomsa

	Complaint	Group 1				Group 2				Pairwise comparisons between the 2 groups(P value)
		Testing 1(baseline)	Testing 2(after 12 wk)	Testing 3(after another 12 wk)	P value	Testing 1(baseline)	Testing 2(after 12 wk)	Testing 3(after another 12 wk)	P value	Testing 1(baseline)	Testing 2(after 12 wk)	Testing 3(after another 12 wk)
	Hot flashes	78.5	14.2b,d	0b,c,d	0.01	68.4	57.8d	42.1d	NS	NS	0.01	0.01
	Difficulty falling asleep	42.8	40.0	38.4	NS	52.6	37.5	47.0	NS	NS	NS	NS
	Frequent arousal	50.0	60.0	50.0	NS	52.6	62.5	41.1c	0.01	NS	NS	NS
	Restless legs	50.0	62.5	38.4 b	0.01	42.1	50.0	58.8	NS	NS	NS	NS
	Bruxism	21.4	11.1	0b,c,d	0.01	10.5	18.7	13.3d	NS	NS	NS	0.01
	Snoring	42.8	40.0d	58.3	NS	73.6	71.4d	58.8b,c	0.05	NS	0.01	NS
	Apnea	14.2	0b,d	18.1c	0.01	26.3	25.0d	7.7b,c	0.01	NS	0.01	NS
	ESS score >9	50.0	50.0d	30.7b,c	NS	31.5	26.6d	35.2	NS	NS	0.01	NS
	Anxiety	64.2	60.0	61.7d	NS	52.6	68.7	47.0 c.d	0.01	NS	NS	0.04
	Memory impairment	64.2	70.0	77.0d	NS	63.1	75.0	53.0 c.d	0.01	NS	NS	0.01
	Depression	28.5	22.2	18.1	NS	31.5	37.5	23.5	NS	NS	NS	NS
	Attention difficulties	35.7	33.3d	16.6 b,c,d	0.01	42.1	50.0d	41.1d	NS	NS	0.03	0.01

Abbreviations: ESS, Epworth sleepiness scale; NS, not significant.

a Values are given as prevalence of the complaint. Group 1 received estrogen for 12 weeks, then estrogen plus progesterone for another 12 weeks; group 2 received placebo for 12 weeks, then placebo plus progesterone for another 12 weeks. b Statistical difference with the value at testing 1. c Statistical difference with the value at testing 2. d Statistical difference between the 2 groups.

Table 3 also reports the prevalence of complaints regarding mood, memory, and attention. There was a decrease in the prevalence of complaints of attention difficulties at the end of the study, but the decrease was greater in group 1 (P<0.03) than in group 2 (P<0.01). Moreover, treatment with progesterone alone resulted in a lower percentage of women complaining of anxiety or memory impairment (P<0.01).

Although both groups had normal baseline sleep efficiency (Table 4), 12 of the 33 participants, 5 from group 1 and 7 from group 2, had a PLM score higher than 5. The only statistically significant differences at the end of the study were lower PLM scores in group 1 and fewer arousals in both groups. The PLM score was decreased for 8 women after estrogen plus progesterone treatment. At baseline, 14 of the 33 participants had an AHI score higher than 5, 4 in group 1 and 10 in group 2. The AHI score improved for 3 of the 4 participants in group 1 following treatment with estrogen (P>0.05) and in 8 of the remaining 11 following treatment with progesterone (P<0.06). After being treated with progesterone, the women experienced fewer arousals during sleep than at baseline (P<0.03). No statistical differences in other sleep characteristics were noted, whether across treatments or between the groups.

Table 4. Sleep characteristics obtained from polysomnogramsa

	Characteristic	Group 1				Group 2			P value	Pairwise comparisons Groups 1 and 2 (P value)
		Testing 1(baseline)	Testing 2(after 12 wk)	Testing 3(after another 12 wk)	P value	Testing 1(baseline)	Testing 2(after 12 wk)	Testing 3(after another 12 wk)		T1	T2	T3
	Sleep latency, min	5.5 (21.6)	12.0 (1.9)	10.7 (7.2)	NS	14.9 (14.9)	12.1 (15.3)	10.7 (16.9)	NS	NS	NS	NS
	REM sleep latency, min	80.4 (36.6)	93.3 (33.1)	97.4 (59.3)	NS	97.7 (43.8)	97.7 (38.3)	103.1 (63.2)	NS	NS	NS	NS
	Early awake, min	8.1 (10.8)	2.8 (3.8)	6.9 (11.6)	NS	4.7 (8.2)	6.4 (11.0)	3.9 (6.7)	NS	NS	NS	NS
	Sleep efficiency, % of total time	83.2 (7.6)	82.4 (8.1)	83.1 (6.9)	NS	83.2 (7.7)	84.8 (8.2)	84.1 (7.4)	NS	NS	NS	NS
	Sleep stages 3 and 4, % of total time	18.4 (8.9)	21.9 (7.8)	20.8 (7.9)	NS	19.3 (5.6)	21.2 (13.0)	22.6 (7.1)	NS	NS	NS	NS
	REM, % of total time	21.7 (4.0)	19.8 (5.7)	18.1 (3.0)	NS	20.3 (5.5)	18.6 (6.3)	18.6 (4.6)	NS	NS	NS	NS
	Arousals, No. her hour	8.5 (4.6)	4.4 (4.3)	2.8 (2.5)b	0.01	16.9 (15.1)d	7.3 (11.1)b	4.9 (5.3)b	0.03	0.04	NS	NS
	AHI score	4.2 (3.9)	4.1 (4.1)	3.6 (2.9)	NS	6.5 (4.2)	5.0 (5.8)	5.4 (6.3)	NS	NS	NS	NS
	SaO2min, %	81.0 (6.2)	85.5 (6.7)	84.9 (7.7)	NS	83.4 (6.3)	83.4 (6.6)	84.2 (5.2)	NS	NS	NS	NS
	PLM score	4.3 (4.6)	8.1 (10.8)	2.8 (3.7)c	0.04	6.2 (7.7)	3.7 (6.0)	2.8 (3.9)	NS	NS	NS	NS
	No. of sleep stage shifts	87.2 (23.3)	77.8 (23.1)	83.0 (15.1)	NS	91.3 (49.9)	3.5 (30.0)	97.9 (4.4)	NS	NS	NS	NS

Abbreviations: AHI, apnea-hipopnea index; NS, not significant; PLM, periodic leg movement; SaO2 minimum arterial oxygen saturation per minute.

a Values are given as prevalence of the complaint. Group 1 received estrogen for 12 weeks, then estrogen plus progesterone for another 12 weeks; group 2 received placebo for 12 weeks, then placebo plus progesterone for another 12 weeks. b Statistical difference with the value at testing 1. c Statistical difference with the value at testing 2. d Statistical difference between the 2 groups.

4. Discussion 

This study showed that HT-mediated improvement of subjectively reported sleep disturbances related to menopause can be independent of objective effects on sleep characteristics. Its main findings are that treatment with estrogen plus progesterone was effective in reducing menopausal symptoms, mostly suppressing hot flashes and bruxism events. Progesterone treatment alone resulted in fewer objective sleep arousals per hour, fewer complaints of anxiety and memory impairment, and fewer complaints of snoring or apnea.

In light of the large number of women experiencing menopausal symptoms, and the wide endocrinologic research aiming at alleviating these symptoms, which are known to include sleep-related respiratory disorders [16], [17], there is a surprising paucity of literature addressing sleep difficulties in menopausal women.

Polysomnographic studies have found that menopausal women with nocturnal hot flashes had lower sleep efficiency than postmenopausal women without these symptoms [4], and other studies have found strong relationships between self-reported hot flashes and sleep complaints [20]. However, self-reports of sleep disturbance may reflect subjective distress rather than objective sleep disruption. A 2003 study by Young et al. [5] and a 2004 study by Freedman and Roehrs [18] found, like ours, that sleep quality was not decreased in these women, even those with menopausal symptoms. The latter authors specifically reported that hot flashes did not result in alterations of sleep architecture, sleepiness, fatigue, or psychomotor performance in postmenopausal women. They excluded more than one-third of the prospective subjects because they had sleep disorders unrelated to menopause. By rigorously screening the study participants they removed potential confounding factors.

In the present study, KI scores were improved in both groups at testings 2 and 3. Hormone therapy may result in physical and psychological improvement regardless of any sleep improvement [19]; however, by improving menopausal symptoms, it is a valuable predictive factor for a beneficial effect on sleep [21]. In particular, the prevalence of complaints regarding hot flashes was significantly reduced by estrogen therapy, and estrogen plus progesterone led to a complete cessation of complaints. This remarkable effect may be related to improvement in subjective sleep quality. The women in group 2 reported fewer arousals at night although they saw no change in their diurnal somnolence.

We speculate that women affected with sleep problems during menopause may be more likely to seek HT. Estrogen seems to improve mood disturbances, although the subjective finding is not always supported by objective results [22]. Young et al. [5] found no association between HT use and better sleep quality whereas Sarti et al. [6] did. Since estrogens seem to reduce the complaint of insomnia, we wondered whether it was a placebo effect, and whether such effect could last through the study. At any rate, we observed a reduction in the number of apnea complaints after 12 weeks in the women taking estrogen.

The ventilatory effects of short-term medroxyprogesterone acetate suggest that periodic administration might be sufficient to improve ventilation, and that periodic administration may even be superior to continuous therapy [4]. Pickett et al. [23] observed that postmenopausal women taking conjugated estrogen plus medroxyprogesterone had fewer respiratory disorders during sleep. In our study, complaints of apnea were significantly reduced following treatment with progesterone. Despite this subjective improvement of apnea, and a decrease in snoring, no significant improvement in AHI score was noted, however. The reduced prevalence of arousals at testing 3 could be due to the well-documented hypnotic effect of progesterone. The effects of progesterone on sleep-related breathing disorders [7] are still promising, although additional studies should be conducted.

Objective sleep was not significantly improved after estrogen and/or progesterone treatment, although there was a trend toward improvement in objective sleep quality. We did not detect any major alterations in polysomnographic recordings in either group, although studies have found that estrogen treatment led to significant relief of menopausal symptoms [19]. This discrepancy may be due to differences in study populations (eg, whether the women were perimenopausal or postmenopausal, had a natural or a surgically induced menopause, or were younger or older) and methods (eg, different dosage, form, and duration of the hormonal treatment).

Our confidence in our findings is strengthened by our use of a well-characterized, randomized, double-blind, placebo-controlled design. The screening interview included a complete medical history, an evaluation of the women's hormonal levels, and a detailed description of their menopausal status. Furthermore, all participants underwent 3 polysomnograms. We were thus able to demonstrate the distinct roles of estrogen, progesterone, and their combination on sleep and menopause-related symptoms.

Interestingly, although baseline PLM scores were within normal ranges for both groups, the combined HT decreased the number of complaints of restless legs and PLM events with arousal. Few studies have reported the effects of estrogen plus progesterone on PLM. Polo-Kantola et al. [24] observed that estrogen improved subjective sleep quality regardless of PLM or related arousal. Recently, we reported that estrogen decreased PLM and increased REM sleep, in addition to improving overall menopause symptomatology in postmenopausal women with high PLM scores [25].

Despite the small number of participants and the possibility of a type 2 error, the present results indicate that HT combining estrogen and progesterone improves subjective sleep in postmenopausal women. Since sleep complaints are common during menopause, future studies should include complete investigations of both subjective and objective measures of sleep prior to receiving HT. Additionally, it is important to select study participants carefully for the actual assessment of the effects of HT on sleep disturbances, and to dissociate these effects from those normally produced by aging.