A
A
A
A
A
A
A
Introduction
Sleep is essential for optimal health. Insufficient or poor-quality sleep is associated with increased risk of all-cause mortality as a result of higher cardiovascular (CV) and metabolic disease incidence, cognitive decline, and obesity1,2. Sleep quality declines during the menopausal transition, with over 50% of menopausal women experiencing sleep disorders like insomnia, obstructive sleep apnea, and early morning awakenings, which impact daily function and quality of life3,4. These changes in sleep quality contribute to the heightened disease risk with menopause, where CV disease, Type II diabetes, and cognitive decline become more prevalent in women vs. men5. By 2030, it is projected that over 1.2 billion women globally will be menopausal6, highlighting the importance of understanding menopause-related sleep changes. To date, there is minimal, yet conflicting research about how estrogen and progesterone, the primary hormones that decline during menopause7, modify sleep composition. In particular, it is unclear how menopause-related hormonal changes modify two main drivers of optimal sleep quality: circadian rhythm and body temperature regulation during sleep.
Despite worsening subjective sleep, there is currently no consensus on how sleep composition changes with menopause (i.e., time spent in light, deep, rapid-eye-movement (REM) sleep, and wake)8–10. Some studies report more deep sleep and higher sleep efficiency in postmenopausal women11, whereas others report less deep sleep and lower sleep efficiency despite similar total sleep time10. These inconsistencies likely reflect differences in reference groups (which often confound aging with hormone status) and varying sample sizes9,11–13. A major gap is that no studies have directly compared postmenopausal women to age-matched men when evaluating sleep composition, a comparison that could help separate menopause-related hormonal effects from the generally negative effects of aging on sleep composition14. Similarly, although postmenopausal women have shown reduced HRV and elevated HR during sleep, relative to premenopausal women15,16, it remains unclear whether these differences reflect menopause-related hormonal changes or declining CV recovery associated with aging more broadly17.
During healthy sleep, core body temperature (TC) follows a U-shape, where TC declines approximately 30 min before sleep onset18, and continues to drop throughout the first half of the night, reaching its lowest point around 4 AM19,20. The majority of deep sleep occurs during the first half of the night21, and cooler TC during this period has been linked to more deep sleep22,23. Additionally, a blunted decline in TC pre-sleep has been associated with worse CV recovery during sleep, as reflected by a higher sleeping heart rate (HR) and lower sleeping heart rate variability (HRV)24. During the second half of the night, REM sleep dominates21, as TC rises until waking21. Generally, the body is more sensitive to environmental temperature disturbances during this time, as thermoregulatory function is compromised25. Temperature regulation during sleep also involves how heat is distributed across the skin. TC is reduced leading up to sleep onset partially as a result of sending blood flow to the periphery (e.g. hands and feet)18. Once sleep is initiated, maintaining a more uniform skin temperature across the body helps with sleep continuity26. Overall, the coordination of these core and skin temperature rhythms support stable sleep, and disruption of these rhythms has been linked to poorer sleep quality18,26,27. With aging and menopause, there is some evidence that the rhythm of body temperatures are blunted28,29, which may contribute to poor quality sleep reported with menopause. However, studies and sample sizes are limited.
Very little is known about how the hormonal transition of menopause, independent of aging, impacts temperature regulation and sleep. We know that postmenopausal women not taking hormone replacement therapy (HRT) have a lower average TC during sleep and throughout a full 24-hour cycle30 vs. premenopausal women. Yet it is unclear whether the circadian rhythm of TC is modified with menopause. Evidence from one paper suggests that the circadian rhythm of skin temperature is blunted with menopause, showing a lower daily amplitude (smaller temperature range) and an earlier timing (approximately a one-hour phase advance; i.e. earlier nadir) in postmenopausal compared with premenopausal women28. Together, these findings suggest that menopause-related hormonal changes can reshape both core and skin temperature curves; however, none of the aforementioned research explored how these changes in circadian rhythm and body temperature impacted sleep. It remains unclear how estrogen and progesterone affect circadian rhythm and temperature regulation during sleep, and how these changes influence sleep composition and CV recovery. Thus, one aim of our study is to compare how circadian rhythm and body temperatures are altered in postmenopausal women taking vs. not taking HRT, and how these potential changes may impact sleep and CV recovery.
A
One approach taken to improve sleep quality in postmenopausal women is to prescribe HRT31,32, however, as a lot of women report significant side-effects33, the need for other, potentially non-pharmacological, interventions is pressing. Additionally, using hormone replacement therapy (HRT) increases TC34, potentially due to estrogen decreasing heat dissipation mechanisms and increasing thermogenesis35. Yet it is unclear how this altered temperature regulation may benefit or harm temperature regulation during sleep. Given the outlined importance of circadian rhythmicity and body temperature shifts for sleep onset and continuity, temperature-based interventions during sleep may provide a solution to improve sleep quality in menopausal women. As evidence of this concept, one study in postmenopausal not using HRT (nHRT) utilizing a high-heat capacity mattress (HHCM), which passively absorbs body heat and reduces TC, showed improved deep sleep and sleeping HR22. At the same time, HHCM approaches are typically limited in their ability to deliver individualized, time-varying temperature control across the full night (i.e. they absorb body heat during the first half of the night and do not adapt to sleep-stage needs or individual thermoneutral ranges). As a result, it may be that Active Temperature Regulation (ATR) throughout the night, via a smart mattress cover, may further enhance sleep and CV recovery in postmenopausal women.
Altogether, critical questions remain about how temperature regulation and circadian rhythm during sleep impact sleep and CV recovery with the menopausal transition, and whether ATR could be a key non-pharmacological solution to improve sleep and reduce disease risk during this critical time. Therefore, the goals of this study are to understand (1) how the menopausal transition (independent of aging) impacts body temperature and circadian rhythm during sleep, and therefore sleep composition and CV recovery, and (2) whether changing body temperatures and circadian rhythm via ATR throughout the night can improve sleep and CV recovery in these populations.
Results
Sleep Composition and Body Temperatures Differed by Cohort
Independent of ATR condition (i.e. across ATR ON and ATR OFF nights), we observed some differences in sleep composition, CV recovery, and body temperatures among cohorts (see Table 1). When exploring cohort-differences independent of ATR condition, women using HRT had slightly better sleep composition than age-matched men, observed through 2.1 ± 1.0% less WASO (p = 0.040), 4.8 ± 1.5% less light sleep (p = 0.006), and 2.8 ± 1.1% more REM sleep (p = 0.011). However, there were no differences in sleep composition between age-matched men and women not taking HRT, or women taking vs. not taking HRT (both p > 0.05). Independent of ATR conditions, there were no differences among cohorts for overnight CV recovery, deep sleep, total sleep time, sleep onset latency, or sleep efficiency (all p > 0.05).
Independent of ATR condition, women using HRT had slightly 0.10 ± 0.04°C higher nightly TC values than age-matched men (p = 0.028), but there were no statistical differences in TC between age-matched men vs. women not taking HRT, or between women taking vs. not taking HRT (both p > 0.05). However, no main effect of cohort was evident for weighted mean skin temperature (TSK) or body temperature (TB) (all p > 0.05).
All cosinor comparisons below for TC and HR are independent of ATR condition. Men had a significantly lower TC mesor compared to both women using HRT (-0.54 ± 0.03°C) and women not using HRT (-0.17 ± 0.03°C; both p < 0.001), indicating a lower 24-hour mean TC in age-matched men vs. women. Women not taking HRT had a significantly higher TC amplitude than both women using HRT (+ 0.11 ± 0.03°C; p = 0.002) and age-matched men (+ 0.15 ± 0.03°C; p < 0.001), suggesting a larger overall range of TC over a 24-hour period for women not using HRT. Age-matched men had a significantly lower HR mesor compared to both women taking HRT (-4.7 ± 1.5 bpm, p = 0.009) and not taking HRT (-3.9 ± 1.5 bpm, p = 0.031), suggesting that men had a lower mean 24-hour HR than either female cohort. There were no differences among cohorts for HR amplitude, TC acrophase or HR acrophase (all p > 0.05).
Table 1
Differences in sleep composition, cardiovascular (CV) recovery, core temperature (TC), body temperature (TB), and skin temperatures by cohort and active temperature regulation condition (ATR, ON or OFF)
Metric
Women using HRT
Women nHRT
Age-matched men
Sleep composition and cardiovascular (CV) recovery (mean ± SE)
Wake after sleep onset (WASO) (%) ‡
OFF = 10.2 ± 0.8
ON = 9.8 ± 0.8
OFF = 11.8 ± 0.8
ON = 11.4 ± 0.8
OFF = 12.6 ± 0.8
ON = 12.2 ± 0.8
Rapid eye movement (REM) sleep (%) ‡
OFF = 19.9 ± 0.7
ON = 20.0 ± 0.7
OFF = 18.9 ± 0.7
ON = 18.9 ± 0.7
OFF = 17.1 ± 0.7
ON = 17.2 ± 0.7
Light sleep (%) ‡
OFF = 52.9 ± 0.9
ON = 53.4 ± 0.9
OFF = 53.8 ± 0.9
ON = 54.3 ± 0.9
OFF = 55.4 ± 0.9
ON = 55.9 ± 0.9
Deep sleep (%)
OFF = 13.2 ± 0.6
ON = 13.2 ± 0.6
OFF = 12.0 ± 0.6
ON = 12.0 ± 0.6
OFF = 11.3 ± 0.6
ON = 11.2 ± 0.6
Sleeping HR (bpm) *
OFF = 63.9 ± 1.4
ON = 62.0 ± 1.4
OFF = 62.4 ± 1.4
ON = 60.6 ± 1.4
OFF = 59.8 ± 1.4
ON = 58.0 ± 1.4
Sleeping HRV (ms) *
OFF = 32.2 ± 2.4
ON = 35.9 ± 2.4
OFF = 33.4 ± 2.4
ON = 37.0 ± 2.4
OFF = 29.8 ± 2.4
ON = 33.4 ± 2.4
Total sleep time (hours)
OFF = 7.0 ± 0.1
ON = 7.0 ± 0.1
OFF = 7.1 ± 0.1
ON = 7.1 ± 0.1
OFF = 6.7 ± 0.1
ON = 6.8 ± 0.1
Sleep onset latency (min)
OFF = 20.9 ± 1.6
ON = 19.7 ± 1.6
OFF = 20.3 ± 1.6
ON = 19.2 ± 1.6
OFF = 20.3 ± 1.6
ON = 19.2 ± 1.5
Sleep efficiency (%)
OFF = 86.0 ± 0.8
ON = 86.5 ± 0.8
OFF = 84.7 ± 0.8
ON = 85.2 ± 0.8
OFF = 83.8 ± 0.8
ON = 84.3 ± 0.8
Overnight core temperature (TC) and skin temperature (mean ± SE)
TC (°C) * ‡
OFF = 36.70 ± 0.03
ON = 36.55 ± 0.03
OFF = 36.62 ± 0.03
ON = 36.46 ± 0.03
OFF = 36.59 ± 0.03
ON = 36.46 ± 0.03
Mean weighted skin temperature (TSK, °C)
OFF = 33.9 ± 0.1
ON = 34.0 ± 0.1
OFF = 33.7 ± 0.1
ON = 33.7 ± 0.1
OFF = 33.6 ± 0.1
ON = 33.8 ± 0.1
Body temperature (TB, °C) *
OFF = 36.42 ± 0.03
ON = 36.29 ± 0.03
OFF = 36.34 ± 0.03
ON = 36.21 ± 0.03
OFF = 36.32 ± 0.03
ON = 36.19 ± 0.03
Circadian rhythm measurements (i.e. cosinor analysis) of TCand heart rate (HR, mean ± SE)
TC mesor (°C) *‡ ⧻
OFF = 36.99 ± 0.03
ON = 36.94 ± 0.02
OFF = 37.01 ± 0.02
ON = 36.94 ± 0.02
OFF = 36.85 ± 0.03
ON = 36.77 ± 0.03
TC amplitude (°C) *✝ ⧻
OFF = 0.36 ± 0.03
ON = 0.44 ± 0.03
OFF = 0.47 ± 0.02
ON = 0.55 ± 0.02
OFF = 0.33 ± 0.03
ON = 0.39 ± 0.03
TC acrophase (hours after sleep onset)
OFF = 15.2 ± 0.4
ON = 15.1 ± 0.3
OFF = 15.3 ± 0.3
ON = 15.6 ± 0.3
OFF = 15.6 ± 0.4
ON = 15.8 ± 0.4
HR mesor (bpm) ‡ ⧻
OFF = 73.8 ± 1.0
ON = 73.9 ± 1.0
OFF = 73.2 ± 1.0
ON = 72.9 ± 1.0
OFF = 69.6 ± 1.2
ON = 68.7 ± 1.2
HR amplitude (bpm) *
OFF = 13.0 ± 1.0
ON = 14.9 ± 1.0
OFF = 14.3 ± 0.9
ON = 15.4 ± 1.0
OFF = 11.9 ± 1.2
ON = 13.6 ± 1.2
HR acrophase (hours after sleep onset)
OFF = 15.8 ± 0.3
ON = 15.7 ± 0.3
OFF = 15.6 ± 0.2
ON = 15.9 ± 0.3
OFF = 15.8 ± 0.3
ON = 16.2 ± 0.3
A
Note: Model estimated means ± standard error (SE) of core temperature (TC), skin temperature and sleep composition including heart rate (HR) and HR variability (HRV) by Active Temperature Regulation (ART, ON/OFF) and cohort (women using hormone replacement therapy (HRT), women not using HRT (nHRT), and age-matched men). For a complete overview of sample sizes,
A
see Supplementary Information table S1. The mesor indicates the mean value around which TCor HR oscillates, the amplitude reflects the half-width of 24-hour max-min, and acrophase represents the hours since sleep onset where amplitude peaks. The following symbols indicate statistical significance (p < 0.05): *= main effect of ATR, ‡= difference between women using HRT vs. age-matched men, independent of ATR condition, ✝= difference between women using HRT vs. women nHRT, independent of ATR condition, ⧻ = difference between age-matched men vs. women nHRT, independent of ATR condition, using linear mixed-effects modeling.
Active Temperature Regulation Improves Cardiovascular Recovery and Lowers Body Temperatures
Independent of cohort, ATR ON compared to OFF was associated with significantly better CV recovery, as observed through an average nightly decrease of 1.8 ± 0.3 bpm (-3%) in sleeping HR and an average nightly increase of 3.6 ± 0.6 ms (+ 11%) in sleeping HRV (both p < 0.001).
ATR ON was also significantly linked to cooler core and body temperatures, observed through an average reduction in TB of 0.13 ± 0.02°C and TC of 0.16 ± 0.01°C across nights and cohorts. The extent to which TC was reduced was significantly dependent on the participant’s ATR OFF TC values, with reductions estimated to occur by 0.55 ± 0.11°C for each 1°C in ATR OFF TC above 36.34°C, meaning that the warmer a participant’s mean overnight TC during ATR OFF, the greater the reduction in TC with ATR ON (all comparisons p < 0.001). For visualizations, see Supplementary Information, Fig. S1.
A
The larger the decrease in TC from ATR OFF to ON, the greater the reduction in HR with ATR ON (Fig. 1a, r = 0.34, p = 0.037). Moreover, colder ATR ON temperatures during the second half of the night were significantly correlated with larger improvements in nightly HRV from ATR OFF to ON (Fig. 1b, r = -0.31, p = 0.004). Overall, regardless of ATR condition, lower TB was linked to significantly greater improvements in CV recovery, such that each 1°C decrease in TB predicted a 8.8 ± 1.4 bpm decrease in HR (p < 0.001) and a 10.60 ± 3.82 ms increase in HRV (p = 0.010).
A
In addition to the overall decrease in TC and HR with ATR ON, changes were also observed in the circadian rhythm of TC and HR (via cosinor analysis). ATR ON (vs. OFF) was significantly associated with a 0.07 ± 0.01°C decrease in mesor and a 0.07 ± 0.02°C increase in TC amplitude. Together, these changes indicate a larger range of TC during the 24-hour period (both p < 0.001). Notably, these two changes indicate a similar maximal (peak) TC value between ATR conditions, but a reduced minima (i.e. nadir, Fig. 2a). We also observed a significant 11.6% increase in the HR amplitude (1.5 ± 0.4 bpm, p < 0.001) with ATR ON, which is largely a result of the lower sleeping HR (i.e. nadir) vs. a higher peak HR during wake (Fig. 2b). Lastly, ATR-related changes in TC and HR amplitude were significantly positively correlated, indicating that larger increases in TC amplitude with ATR ON were correlated with larger increases in HR amplitude (r = 0.46, p = 0.045). Visualizations of the TC and HR 24-hour changes with ATR are presented in Fig. 2.
Active Temperature Regulation Balances Sleep Composition
For those with below average sleep in a specific sleep stage with ATR OFF, sleeping with ATR ON helped rebalance sleep composition. In general, participants with lower percentages of deep or REM sleep during ATR OFF achieved higher percentages with ATR ON, while participants with higher percentages of light sleep or WASO during ATR OFF showed reductions with ATR ON.
Specifically, sleeping with ATR ON improved sleep for those with abnormal amounts of light sleep (i.e. >65% light sleep; r=-0.28; p = 0.007), deep sleep (< 13.7% deep sleep; r=-0.30; p = 0.004), REM sleep (< 21.3% REM sleep; r=-0.27; p = 0.012), or WASO sleep (i.e. >9.5% WASO; r=-0.26; p = 0.014) with ATR OFF. In total, when observing changes in sleep composition with ATR conditions across the cohorts, we observed that 26% of participants had a 2+% reduction in the percentage of light sleep (n = 23), 18% of participants had a 2+% increase in the percentage of deep and REM sleep (n = 16), and 19% of participants had a 2+% reduction in WASO % (n = 17) with ATR ON.
Mean Weighted Skin Temperature Differs Across Sleep Stages
A
TSK varied significantly by sleep stage, independent of cohort and ATR condition. During REM sleep, TSK was the warmest, being 0.08 ± 0.02°C warmer than in deep sleep (p = 0.014) and 0.11 ± 0.02°C warmer than in light sleep (Fig. 3a, p < 0.001). During WASO, TSK was significantly the coldest, with temperatures 0.51 ± 0.02°C lower than in REM sleep, 0.40 ± 0.02°C lower than light sleep, and 0.43 ± 0.02°C lower than deep sleep (Fig. 3a, all p < 0.001). No significant differences in TSK were observed between light and deep sleep (p = 0.500) or between ATR ON and OFF in any sleep stage (all p > 0.05).
When exploring the relationship between TSK and sleep stages from the Initial to Final phases, women using HRT exhibited a significant 0.38 ± 0.15°C higher TSK during WASO in the Final phase (Fig. 3b, p = 0.018), compared to the Initial phase. No other phase- or cohort-related TSK differences were found across sleep stages or cohorts (all p > 0.05).
A
TSK also predicted sleep stage probabilities, supporting the findings that a colder TSK linked to WASO and a warmer TSK linked to REM (Fig. 4a). Relative to deep sleep, each 1°C TSK increase was associated with a 59% decrease in the likelihood of being in WASO (OR = 0.41, 95% CI [0.31, 0.53]), with WASO being more likely below 32.9°C (CI’s non-overlapping at 32.5°C). Additionally, each 1°C TSK increase was associated with a 17% increase in the likelihood of being in REM sleep (OR = 1.17, 95% CI [0.99, 1.38]), with REM being more likely above 33.8°C (however, note overlapping CI’s, making this prediction more uncertain). No clear differences in cohorts were observed due to overlapping CI’s (Fig. 4b). However, the Tsk where deep sleep and WASO were equally likely was 33.0°C for women using HRT, 33.2°C for women nHRT, and 32.5°C for age-matched men, suggesting men may require ~ 0.5–0.7°C colder skin temperatures for wake to predominate.
Discussion
We examined how body temperatures and circadian rhythm during sleep are modified by the menopausal transition, and how these physiological changes impact sleep and CV recovery. In general, we found that the circadian rhythm of TC was blunted during sleep across all cohorts, including women taking HRT and age-matched men. Next, we explored whether sleeping on a mattress cover with ATR throughout the night would improve circadian rhythm of TC and therefore sleep composition and CV recovery in older adults. Across all cohorts we found that sleeping with ATR ON improved the circadian rhythm of TC and HR (i.e. reinstated the U-shape), which improved CV recovery (lower sleeping HR and higher sleeping HRV). Yet despite the improvement in circadian rhythmicity of TC and CV recovery, we did not see a uniform improvement in sleep composition with TC cooling as previously seen in younger individuals. Instead, we found that sleeping with ATR ON primarily rebalanced sleep composition in individuals with sub-optimal sleep composition with ATR OFF.
Our findings suggest that age and sex influence the circadian rhythm of core temperature, but not necessarily in the way that we expected or that is seen in younger adults. Notably, our data do not align with previous research indicating that more favorable TC dynamics (i.e. a U-shape during sleep, reflecting a larger pre-sleep drop in TC) leads to better sleep composition, especially increased deep sleep18,24. For example, non-HRT women had a larger TC drop with ATR ON, yet this did not result in more deep sleep or better CV recovery than the other cohorts (see Table 1 and Fig. 2). Conversely, women using HRT exhibited higher TC and an opposite circadian rhythm response (i.e. inverted U-shape) with ATR OFF, yet had a slightly more favorable sleep composition than age-matched men (Table 1 and Fig. 2). These patterns suggest that with age (independent of hormone status) there may be a decoupling of body temperature with sleep composition from that which is often observed in younger populations18,24. We expected that as TC decreased, we would see an increase in deep sleep; however, we did not see any significant relationship between TC and deep sleep in our cohorts. Furthermore, we found that in some cases sleep composition can remain relatively favorable even when TC is higher throughout the night and circadian rhythm does not mimic the expected U-shape (e.g. HRT women; Fig. 2). As very little is known about how body temperature and sleep composition are mechanistically related (e.g. the neuronal pathways), more research is needed to understand why this decoupling might occur from a cellular/molecular perspective with age.
This apparent dissociation between body temperature and sleep with age begs the question: If TC and sleep composition are not tightly coupled in older adults, can an external thermal intervention still drive improvements through a lower TC and restoration of circadian rhythm? More specifically, when ATR restores the 24-hour TC rhythm, does this translate into meaningful improvements in sleep composition, CV recovery, or both? We found that ATR produced consistent improvements across cohorts in reinstating the U-shape of TC circadian rhythm during sleep. ATR was linked to a nightly decrease in TC across cohorts, along with lower TC mesor and higher TC amplitude across 24 hours, indicative of enhanced TC circadian waveform. These findings align with previous work showing that nocturnal cooling lowers TC22, though to our knowledge we are the first to show that ATR can restore the U-shaped circadian rhythm of TC during sleep in older adults (Fig. 2a). Although the improvements in circadian rhythm of TC did not lead to uniform improvements in sleep composition (as seen in data on younger adults), it may be that enhancing the U-shape of the TC curve during sleep resulted in metabolic, cognitive or emotional processing improvements that we did not measure. Previous research shows there are strong links between the TC rhythm during sleep and metabolic flexibility, cognitive flexibility and emotional processing36–38.
We observed that ATR-related reductions in TC were positively correlated with reductions in sleeping HR across the night, indicating that a larger decrease in TC from ATR OFF to ON was associated with a greater reduction in sleeping HR (Fig. 1a). While decreases in HR have been observed with TC reductions before22,24, to our knowledge we are the first to demonstrate a direct relationship between the two. We also observed that cooler ATR temperatures in the second half of the night were related to larger increases in HRV, suggesting that the magnitude of cooling may play an important role in parasympathetic activity during sleep39. Finally, ATR-related increases in TC amplitude were accompanied by increases in HR amplitude, largely driven by the nadir of TC and HR decreasing in tandem with ATR ON. Together, these findings support the conclusion that TC decreases via ATR can meaningfully improve CV recovery during sleep in older adults.
For sleep composition, ATR's effects were selective rather than universal. ATR did not produce uniform sleep stage increases; rather, it rebalanced sleep composition by improving time spent in sleep stages where individuals were initially below- or above-average. This pattern is particularly relevant for older adults, given that sleep composition commonly shifts toward more light sleep and less deep/REM sleep with age21. Although we did not observe a direct link between changes in TC with sleep stages from ATR OFF to ON, we did observe clear differences in TSK based on sleep stage, similar to previous research40,41. REM sleep was associated with the warmest TSK, whereas WASO was associated with the coldest TSK. Moreover, lower TSK values were associated with higher likelihood of wake relative to deep sleep. These findings partly support the notion that aging does not fundamentally alter the relationship between TSK and sleep-wake states, as lower TSK during WASO than during light sleep has also been reported in young men and women40. However, Zhang et al.’s study in younger women40 did not observe higher TSK during REM relative to light or deep sleep stages, and their absolute TSK values were approximately 0.2–0.6°C higher than those observed in our study. It is unclear whether the discrepancies reflect changes to skin temperature during sleep with aging or methodological differences in how TSK was computed (7-site estimate versus our 3-site estimate). Nevertheless, studies comparing younger and older populations’ skin temperatures during wake at rest similarly found TSK was lower by 0.3°C for the older group42. Although the lower TSK was not statistically significant, this supports the theory that skin temperatures may change with age during sleep. Further work using similar TSK estimates across age groups during sleep is needed to confirm these differences.
A
Limitations. Although our data suggest that the blunted circadian rhythm and the decline in sleep quality and cardiovascular recovery observed during the menopausal transition are largely attributable to natural aging, confirming that these changes are driven solely by aging would require inclusion of additional reference groups, such as premenopausal women and younger age-matched men. Additionally, HRT formulation or dose was not standardized across the HRT women in the study, which resulted in a wide range of progesterone values in the HRT group (SD = 8.0 µg/mL, see Supplementary Information, Table S2). As progesterone impacts temperature regulation and sleep43,44, it may be that our HRT results would have been modified had we ensured everyone was on similar HRT dosing and formulations. However, doing so would have severely limited our sample sizes and applicability of results to all women taking HRT. Moreover, the generalizability of these results is somewhat limited by (1) the exclusion of individuals with chronic conditions, thus not adequately reflecting the disease burden present in the overall population at this age45, and (2) only including participants that already owned an Eight Sleep Pod. Pod-owners can be presumed to be more invested in their sleep and general health, thus reflecting a selection bias in our data. However, participants’ average sleep metrics with ATR OFF were below-average, as expected with this age group (Table 1) indicating that sleep composition in these individuals matched that of the general older adult population. Lastly, because TC was scheduled for one night per ATR condition mid-week, we considered whether the TC differences between ATR ON vs. OFF could reflect short-term adaptation to ATR rather than an acute thermal effect. Several observations argue against this. First, physiological cold acclimation typically requires repeated exposures over ~ 1 week or longer46, making it unlikely that one to two nights of mild surface cooling would produce meaningful acclimation. Second, the GI pill often remained in the system for more than 24 hours, allowing us to capture ≥ 2 nights of TC data from 70.4% of participants. When we evaluated the TC cosinor metrics from night 1 to night 2, there was no difference between the first night and subsequent nights, suggesting stable within-person rhythms across repeated measurements. Finally, with ATR ON, TC declined progressively across the sleep period and reached its nadir ~ 4–5 hours after sleep onset (Fig. 2a), consistent with an ongoing, real-time cooling influence rather than a fixed acclimation shift.
In conclusion, we found that aging blunts (i.e. flattens) the circadian rhythm of TC during sleep, independent of menopausal status. However, sleeping on a mattress cover with ATR reinstated the circadian rhythm of TC and significantly improved CV recovery during sleep. Although the improvements in circadian rhythm of TC did not lead to uniform improvements in sleep composition (as seen in data on younger adults), enhancing the U-shape of the TC curve during sleep has been shown to improve metabolic flexibility, cognitive flexibility, and emotional processing36–38. Overall, these findings support ATR as a promising non-pharmacological strategy to restore circadian rhythm of TC, improve CV recovery during sleep, and to selectively improve sleep composition in older adults with sub-optimal sleep quality, including postmenopausal women experiencing disturbed sleep. Future research should examine whether the ATR-induced improvements in TC and CV recovery lead to long-term reductions in CV, metabolic, and cognitive disease risk.
Methods
Participant Characteristics
A
A
A
In total, 90 participants completed the experimental protocol, which comprised three cohorts of n = 30 each: 1) postmenopausal women using combination HRT (i.e. estrogen and progesterone), 2) postmenopausal nHRT, and 3) age-matched men not taking any supplement hormones (like testosterone). On average, participants weighed 161 ± 29 lb, were 5′7″ ± 3″ tall, and were 56 ± 6 years old. Additional participant characteristics are provided in Supplementary Information, Table S2. Urinary hormone levels of pregnanediol glucuronide, luteinizing hormone, follicle-stimulating hormone, and estrone-3-glucuronide were assessed the morning after Night 3 on each ATR condition (Mira Clarity Kit, Miracare, Pleasanton, California, US; see Supplementary Information, Table S2 for values).
Experimental Design
A
A
All participants slept on their Eight Sleep Pod for 14 consecutive nights: one week with ATR disabled (ATR OFF, i.e. temperatures turned off) and one week with ATR enabled at their preferred temperature settings (ATR ON, i.e. temperatures turned on). The order of these conditions was randomized across participants, with half of participants starting with ATR ON and half of participants with ATR OFF. Temperature selections during the ATR ON week were recorded for each participant in Eight Sleep’s database (see Supplementary Information, Table S3, for mean Pod temperature values). Throughout the study, participants were instructed to maintain their normal routines regarding exercise, diet, alcohol consumption, and sleep habits to maximize ecological validity.
A
During the 14-day study and during the 2 nights before, participants continuously wore wearable rings (Generation 4, Oura Health, Oulu, Finland) to track their HR (both during sleep and across 24 hours), sleeping HRV, and sleep composition, including the duration of time spent in deep, light, and REM sleep stages, time awake, sleep onset latency, sleep efficiency, and total sleep time. Both weeks followed a similar 7-night structure, differing only in whether ATR was ON or OFF (Fig. 5). On Nights 2–4 of each week, participants recorded skin temperatures during sleep at three sites (forehead, left chest, and left foot) using iButton temperature loggers (Thermochron iButtons, Model DS1922L; Maxim Integrated, San Jose, USA). On Night 3 of each week, participants recorded their Tc during sleep via gastrointestinal pills (e-Celsius Performance, BodyCAP, Hérouville Saint-Clair, France) (see the ‘Physiological data’ section in Methods for details).
Participant Exclusion Criteria
In total, 90 participants were enrolled and completed the study.
A
A
All participants provided written informed consent in accordance with the Declaration of Helsinki. The study protocol was approved by the Sterling Institutional Review Board in June 2025 (IRB: 14003).
Participants were excluded if they were under 45 years of age, transgender, did not own a Pod, unable to sleep on their Pod for 14 consecutive days, unwilling to have ATR OFF for 7 nights, could not comply with the testing protocol, or reported regularly sleeping less than 4 hours per night for more than half of the week. Participants were also excluded if they had CV disease, sleep disorders (e.g., insomnia, sleep apnea, narcolepsy), and/or one or more chronic diseases known to affect estrogen, progesterone, testosterone, glucose regulation, thermoregulation, and/or sleep quality and regularity. These exclusions minimized individual variability that could confound results.
A
Postmenopausal women confirmed through self-report that they had not had a menstrual period for at least 12 consecutive months47. Postmenopausal women were excluded if they had changed their HRT status (i.e. started or stopped taking HRT) in the 3 months prior to the study. Women in the HRT cohort were required to use combination estrogen and progesterone HRT, but there was no restriction on the dose of these hormones or in the application (e.g. oral pill, topical gel, etc). Women were also excluded if they had a current diagnosis of polycystic ovarian syndrome (PCOS) or endometriosis, and/or had an oophorectomy at any time in their life. These exclusion parameters were implemented to ensure a similar hormonal profile within each cohort.
To minimize risks associated with taking the gastrointestinal pill, participants were excluded according the manufacturers' instructions: body weight below 88 lbs (40 kg) or a BMI exceeding 44.6, current or previous gastrointestinal disorders (e.g., gastroparesis, diverticulitis, Crohn’s disease, or recent gastrointestinal surgery), those unable to swallow a capsule-sized pill, those with implanted pacemakers or other electromagnetic medical devices, and/or a scheduled MRI during or within one week after the study period.
Physiological Data
The Eight Sleep Pod’s Active Temperature Regulation
The Eight Sleep Pod is a temperature-regulated mattress cover that modulates bed surface temperature in real-time based on the biometric signals it captures throughout the night (e.g. HR, HRV, sleep stages). The system consists of a Hub, positioned beside the bed, which has a water reservoir that also heats and cools the water temperature flowing through the ATR cover. The temperature is controlled by the participant via the Eight Sleep app. Importantly, the Pod Cover allows independent temperature control for each side of the bed.
Participants can set discrete temperatures in the Eight Sleep App (allowable range from 13–43°C) at three time points: (1) Bedtime phase, from bed entry until ~ 15 minutes after sleep onset; (2) Initial phase, covering the subsequent four hours; and (3) Final phase, from four hours post-sleep onset to wake. Within the Initial and Final phases, the Pod automatically adjusts temperatures based on the individual’s current sleep stage, their biological sex, and age. Generally, the Pod will cool slightly during deep and light sleep stages, and warm during REM sleep, to help maximize time in each sleep stage22,26. The mean ± SD temperature settings the cohorts chose are provided in Supplementary Information, Table S3. Ambient room temperature was slightly higher with ATR ON vs. OFF by 0.3 ± 0.1°C (p < 0.001); however, this small change in room temperature has not been shown to impact temperature regulation48.
Two functional modes were implemented: ATR ON and ATR OFF. During ATR ON, the Pod actively regulated bed surface temperature according to each participant’s programmed thermal profile with automatic adjustments based on the current sleep stage, while collecting continuous biometric data. During ATR OFF, temperature regulation was disabled, allowing only biometric data collection. Note that none of the CV recovery or sleep stage metrics reported in this paper are from the Pod – they are all from the wearable ring. Pod-derived sleep staging was only used to align TSK to sleep stages, as the Pod was able to detect when participants left the bed.
Sleep and Cardiovascular Recovery Metrics
The Oura Ring (Generation 4, Oura Health, Oulu, Finland) is a validated wearable device that continuously records physiological and behavioral parameters through photoplethysmography (PPG), skin temperature sensing, and a tri-axial accelerometer49. For this study, Oura data were used to characterize both nightly sleep composition and associated CV recovery patterns. Four data summaries were downloaded from the Oura API: (1) night-level summary data, providing a single observation per night, (2) epoch-level (i.e. 30 s) sleep stage data, detailing the timing and duration of light, deep, and REM sleep, as well as awake periods throughout each sleep episode, (3) continuous HR and HRV estimates sampled every 5 minutes during sleep, and (4) continuous 24-hour HR data, sampled irregularly every 5–60 seconds depending on signal stability and participant motion50. The nightly summaries included mean or total estimates of the following: average HR (bpm), average HRV (ms), sleep efficiency (%), time spent awake (min), as well as stage-specific durations for light, deep, and REM sleep. Additional variables included total sleep time, sleep onset latency (min), bedtime start, and bedtime end. The sleep window was defined as the period from sleep onset (calculated as the time elapsed after bedtime start to sleep onset latency) until bedtime end, reflecting total sleep time. Oura’s HRV values are derived from interbeat interval (IBI) data collected during sleep, calculated as the root mean square of successive differences (RMSSD) between consecutive heartbeats.
Skin Temperature Measurements
Skin temperature was wirelessly recorded using iButtons (Thermochron iButtons, Model DS1922L; Maxim Integrated, San Jose, USA), a validated methodology for acquiring skin temperature51, every 5 minutes at three sites during sleep on Nights 2–4. Participants wore iButtons on their forehead, left chest, and top of left foot. These locations allowed us to calculate the most accurate mean-weighted skin temperature that has been shown to reflect thermal sensation during sleep52. Devices were attached using 3M Transpore™ medical tape at least 30 minutes prior to sleep onset and were removed immediately upon waking. This protocol timing ensured thermal equilibration before sleep and minimized post-awakening artifacts. Skin temperature data were stored internally on the iButtons and then downloaded after participants shipped back the units.
Core Body Temperature Monitoring
TC was continuously monitored every minute during sleep via a gastrointestinal pill (e-Celsius Performance, BodyCAP, Hérouville Saint-Clair, France). This pill is a wireless, single-use device approximately the size of a standard vitamin capsule, designed to measure internal temperature from the intestines. The gastrointestinal pill has been validated against both esophageal and rectal temperatures and deemed reliable and comparable to these other TC locations53. When taking the pill each week, participants were instructed to ingest the pill at least 4 hours before their bedtime on Night 3 to ensure that the pill traveled to the small intestines and temperature readings were normal before sleep onset. Once swallowed, the pill recorded TC and transmitted data in real time via bluetooth to a lightweight external monitor kept within a one-meter range of the participant. The monitor stored the temperature data locally and after participants shipped back the unit, temperatures were downloaded to a secure application for processing and analysis.
Data Preprocessing and Filtering
Data preprocessing and filtration were performed using Python (version 3.13.5) in Visual Studio Code (version 1.102.3). All data types were aligned to each participant’s local timezone.
All data types were summarized not only as nightly means, but also separately for the first (Initial) and second (Final) portions of the night to assess time-of-night effects. For sleep, CV, and body temperature metrics, these phases were defined using each participant’s average sleep midpoint from the prior three months: Initial spanned sleep onset to the average midpoint, and Final spanned the average midpoint to sleep offset. Unless otherwise specified, “Initial/Final” refers to this midpoint-based segmentation. In contrast, ATR temperature phases followed the Pod’s built-in schedule: Initial corresponds to the first ~ 4 hours of sleep and Final corresponds to the period from ~ 4 hours post–sleep onset until waking.
Because data were collected in free-living conditions across multiple sensors, filtration was required to remove artifacts, ensuring that analyses reflected typical sleep and thermophysiology. For all datatypes, nights/periods with extreme sleep duration (< 4 or > 14 h) were excluded. Sleep and CV outcomes were filtered by participant-relative outliers. This retained 1,408 nights from 90 participants for cohort comparisons of sleep composition and CV recovery (including 1,230 nights within the ATR ON/OFF experimental weeks). Skin temperature data were filtered to remove nights indicative of sensor detachment, poor-quality recordings, or partial detachment episodes identified by rapid non-physiological shifts (with minor trimming where appropriate), resulting in 552 nights from 89 participants. TC data were filtered to remove non-physiological periods (shipping and pre-ingestion), retain only stable physiological readings, remove beverage-related artifacts, and exclude nights with < 4 h of valid data.
A
This yielded 408 nights from 87 participants. For circadian analyses, TC required continuous 24-h windows with exclusions for > 4 h missing data (with limited manual inclusions when gaps still allowed cosinor fitting), resulting in 279 24-h TC periods from 78 participants. Similarly, 24-h HR windows were excluded for gaps > 4 h, or insufficient data density for cosinor fitting, resulting in 669 24-h HR periods from 54 participants. Full filtration criteria and justification are provided in ‘Data Filtration’ in Supplementary Information. For a visual overview of the filtration pipeline, see Supplementary Information, Fig. S2.
Sleep Metrics and Cardiovascular Recovery
The proportions of deep, light, and REM sleep were derived by dividing the duration of each stage (in min) by the total sleep time (in min) and multiplying by 100 to obtain a percentage of time in each sleep stage. WASO percentage was computed as the time spent awake after falling asleep until waking in the morning, divided by total time in bed minus sleep onset latency, and expressed as a percentage. All other metrics (HR, HRV, sleep efficiency, sleep onset latency, total sleep time) were pulled directly from the Oura API.
A
To calculate sleep composition in the Initial phase, the total time in light, deep and REM sleep from sleep onset to the night’s midpoint was summed both within and across each sleep stage to calculate the percent time in each sleep stage (time in each sleep stage divided by total sleep time in the Initial phase, multiplied by 100). Percent WASO in the Initial phase was calculated as the time awake divided by the time from sleep onset to the average midpoint multiplied by 100. HR and HRV mean estimates for the Initial phase were similarly determined by taking a mean of the continuously sampled (every 5 min) HR and HRV from sleep onset to the average midpoint. The same approach was used to obtain sleep composition, WASO, and CV mean estimates in the Final phase by using the same computations from the average sleep midpoint to sleep offset. For overview of the calculations for each sleep phase, see Fig. 6.
To evaluate how ATR ON influenced sleep relative to their ATR OFF sleep composition, we calculated participant-level means for each sleep stage during ATR OFF and ON, and then computed the mean difference from ATR OFF to ON. Similarly, we computed ATR ON-induced changes in HRV and HR relative to ATR OFF. Percent change in HRV and HR was defined as (ON − OFF)/OFF×100, using each participant’s mean nightly values in each of the two ATR conditions.
Skin Temperatures
Mean weighted skin temperature (TSK) was calculated at the data-point level (sampled every 5 minutes) using the following equation to reflect thermal sensation52:
The nightly means for TSK were calculated by taking the average TSK from the Pod-derived sleep onset to the Pod-derived offset (after filtration). To get a mean TSK for each sleep stage, across and within cohorts, each TSK datapoint was aligned with the sleep stage its timestamp aligned with according to the Pod’s sleep staging algorithm, then averaged within each sleep stage across the night. WASO episodes when participants left the bed during the night were excluded, as TSK values might have been altered due to movement and/or changing environments. Sleep-stage specific TSK was also calculated separately for the Initial and Final phases by taking the average TSK values within each sleep stage across the time-windows of the two phases.
Core Temperature
TC was sampled every minute and averaged nightly, using Pod-estimated sleep onset/offset timestamps after filtration. Body temperature (TB) was calculated as 0.9×TC + 0.1×TSK54 using the mean nightly TC and TSK values for each participant during each condition.
TC changes from ATR OFF to ATR ON were calculated together with sleeping HR changes on a participant level. Mean nightly TC changes between conditions were calculated for nights with HR data available. Similarly, HR percent change was calculated as (ON − OFF)/OFF×100 using mean nightly HR values where TC data were available.
Circadian Rhythm Data for Core Temperature and Heart Rate
TC data for circadian rhythm analysis required continuous 24-hour periods for cosinor analysis (see ‘Data Analysis’). For each participant, the 24-hour windows started at the individual’s mean sleep onset time, calculated as the circular mean to account for the wrap-around from 24:00 to 00:00, across all valid nights (i.e. nights not removed during TC filtration) in the study. For visualization, a secondary 24-hour window starting one hour prior to sleep onset was created to better show the TC drop at sleep onset18. Because the sampling times varied slightly (by < 1 minute) across and within TC pills, we rounded each datapoint timestamp to the nearest minute to standardize timing across pills and enable minute-by-minute averaging for visualization.
The 24-hour HR data also underwent cosinor analysis, with 24-hour periods starting at each participant’s mean sleep onset across the study. HR data were binned in 5-minute increments to prevent high frequency sampling during physical activity from biasing the cosinor fit. For visualization, 24-hour HR was plotted alongside TC, starting one hour before sleep onset, utilizing the same HR data as for the cosinor analysis.
Statistical Analysis
Analyses included only nights with complete data across all required variables. Therefore, statistical models comparing Tsk and sleep metrics included 527 nights from 89 participants. Models including TC and Tsk to predict TB included 273 nights from 85 participants, and the model predicting HR from TB included 258 nights from 85 participants. For ATR-related changes, differences from ATR OFF to ATR ON were calculated at the participant level. Comparisons of ATR-related changes in sleep composition based on ATR OFF estimates included 90 participants. Comparisons of ATR-related changes in TC and HR included 70 participants, while comparisons of ATR temperatures to HRV percent change included 90 participants.
An a priori power analysis, based on parameter estimates from published literature52,55,56, indicated that ~ 30 participants per cohort would provide 80% power to detect group-level differences in HRV and TC.
Statistical analyses were performed using R in R Studio (version 2025.05.1 + 513). Given that most dependent variables were measured on multiple nights within both ATR conditions (within-participant predictors), the data was primarily analyzed on a nightly basis utilizing linear mixed effects models (LMMs) with participant ID as a random intercept. The model uses the individual observations (nights) to make the overall estimates, with each participant having their own intercept for expected values in the dependent variable. LMMs were used for all comparisons except those involving participant-level mean estimates for ATR OFF and/or ATR-related changes. For cohort and ATR status comparisons, estimated marginal means contrasts were created for pairwise comparisons of the LMMs using the emmeans R-package. These contrasts had Tukey adjustments for multiple comparisons for between-cohort comparisons and Holm adjustments for within-cohort comparisons. As LMMs are robust to normality violations57, heteroscedasticity and influential clusters were the primary focus of assumption testing. All LMMs and pairwise comparisons were run with CR2-adjusted estimates to account for potential heteroscedasticity and influential participants biasing the results.
For participant-level ATR-related changes (e.g. comparing changes in sleep composition based on ATR OFF mean values), linear models with HC3-adjusted estimates were utilized to similarly prevent issues induced by heteroscedasticity and influential clusters. For correlation plots based on participant mean values, Pearson correlation coefficient (r) and associated p-values were reported.
Cosinor analysis with a single harmonic was used to estimate 24-hour TC and HR dynamics. Cosinor fitting was conducted as a linear model using ordinary least squares (OLS) in base R to estimate
,
, and
following the equation below58,59. The mesor (24-hour average), amplitude (half-width of 24-hour max − min) and acrophase (hours since sleep onset where amplitude peaks) were extracted from the cosinor function59 and analyzed for each day and subject using LMMs.
To examine how mean TSK predicted sleep stage and WASO probability, a Bayesian multinomial logistic mixed-effects model was fitted using a categorical logit link with deep sleep as the reference category. The model was estimated across four Markov chains with 4,000 iterations per chain (first 2,000 for warm-up). Default brms priors were applied. The model outputs consisted of the log odds and credible intervals (CIs), which both converted to odds ratio (OR) and the CIs of OR for interpretability. These estimates represent the expected change in the odds of being in a given sleep stage (REM sleep, light sleep, or WASO) relative to deep sleep (model reference category), per 1°C increase in TSK.
We wanted to identify the Tsk at which the model predicts people are more likely to be in one sleep stage versus another. To do this, posterior predicted probabilities were computed across a continuous temperature range to visualize how sleep stage likelihoods varied with TSK. From these predictions, we derived equal-likelihood temperatures (where two sleep stages were equally probable) and CI-overlap temperatures (where 95% CIs of adjacent sleep stages intersected). These crossing points reflect nonlinear transformations of the model estimates calculated from predicted probabilities rather than raw log-odds coefficients. Consequently, the reported crossing Tsk represents values on the posterior prediction surface, where population-level probabilities of two sleep stages converge, rather than direct translations of the underlying logit-scale effects.
A
Author contributions
Conceptualization: S.S.J., M.L.H., D.D.H., N.E.M.; Data collection: M.L.H., E.R.C., S.S.J.; Data storage: E.R.C.; Data analysis and visualization: S.S.J., N.E.M; Data interpretation: S.S.J., N.E.M., M.L.H., T.M.; Writing (original draft): S.S.J., M.L.H., E.R.C., B.C.W., D.D.H., T.M., N.E.M.
Acknowledgements
We thank the study participants for their time and participation. We also thank the Eight Sleep team, particularly Natasha G. Ragland, for technical assistance and equipment preparation, and Jonathan Taso for firmware development that allowed us to turn ATR off.
A
Competing interests
& funding
This study was funded by Eight Sleep, Inc. Authors S.S.J., M.L.H., E.R.C., B.C.W., D.D.H., and N.E.M. are employees of Eight Sleep, Inc. and hold, or may be eligible to hold, equity in the company. They declare no other competing interests. Author T.M. declares support from the Canada Research Chairs Program (CRC-2022-00245); otherwise no other competing interests.
Materials & Correspondence
All correspondence and requests for materials should be addressed to corresponding author Nicole E. Moyen.
A
Data availability
The datasets generated for this study are not publicly available due to proprietary restrictions.
Code availability
The underlying code for this study is not publicly available for proprietary reasons but may be made available to qualified researchers on reasonable request from the corresponding author.
Bibliography
1.
Li J et al (2022) Sleep duration and health outcomes: an umbrella review. Sleep Breath 26:1479–1501
2.
Shah AS et al (2025) Effects of Sleep Deprivation on Physical and Mental Health Outcomes: An Umbrella Review. Am J Lifestyle Med 15598276251346752. 10.1177/15598276251346752
3.
Salari N et al (2023) Global prevalence of sleep disorders during menopause: a meta-analysis. Sleep Breath Schlaf Atm 1–15. 10.1007/s11325-023-02793-5
4.
Skibiak K, Dębski J, Przybyłowski J, Walędziak M, Różańska-Walędziak A (2025) The influence of menopausal status on sleep quality in different populations – a narrative review. Przegla̜d Menopauzalny Menopause Rev 24:53–65
5.
Lobo RA (2014) What the future holds for women after menopause: where we have been, where we are, and where we want to go. Climacteric 17:12–17
6.
Delanerolle G et al (2025) Menopause: a global health and wellbeing issue that needs urgent attention. Lancet Glob Health 13:e196–e198
7.
Burger HG (2002) Hormonal Changes in the Menopause Transition. Recent Prog Horm Res 57:257–275
8.
Baker FC, Lampio L, Saaresranta T, Polo-Kantola P (2018) Sleep and sleep disorders in the menopausal transition. Sleep Med Clin 13:443–456
9.
Hachul H, Bittencourt LRA, Soares JM, Tufik S, Baracat EC (2009) Sleep in post-menopausal women: Differences between early and late post-menopause. Eur J Obstet Gynecol Reprod Biol 145:81–84
10.
Kalleinen N et al (2008) Sleep and the menopause – do postmenopausal women experience worse sleep than premenopausal women? Menopause Int 14:97–104
11.
Young T, Rabago D, Zgierska A, Austin D, Finn L (2003) Objective and Subjective Sleep Quality in Premenopausal, Perimenopausal, and Postmenopausal Women in the Wisconsin Sleep Cohort Study. Sleep 26:667–672
12.
Kravitz HM et al (2003) Sleep difficulty in women at midlife: a community survey of sleep and the menopausal transition *. Menopause 10
13.
Moreno-Frías C, Figueroa-Vega N, Malacara JM (2014) Relationship of sleep alterations with perimenopausal and postmenopausal symptoms. Menopause 21:1017–1022
14.
Ehlers CL, Kupfer DJ (1989) Effects of age on delta and REM sleep parameters. Electroencephalogr Clin Neurophysiol 72:118–125
15.
Baker FC et al (2019) Changes in heart rate and blood pressure during nocturnal hot flashes associated with and without awakenings. Sleep 42
16.
Subhashri S et al (2019) Assessment of Heart Rate Variability in Early Post-menopausal Women. Int J Clin Exp Physiol 6:11–14
17.
Jandackova VK, Scholes S, Britton A, Steptoe A (2016) Are Changes in Heart Rate Variability in Middle-Aged and Older People Normative or Caused by Pathological Conditions? Findings From a Large Population‐Based Longitudinal Cohort Study. J Am Heart Assoc Cardiovasc Cerebrovasc Dis 5:e002365
18.
Kräuchi K, Cajochen C, Werth E, Wirz-Justice A (2000) Functional link between distal vasodilation and sleep-onset latency? Am J Physiol -Regul Integr Comp Physiol 278:R741–R748
A
19.
Del Bene VE, Temperature (1990) In: Walker HK, Hall WD, Hurst JW (eds) Clinical Methods: The History, Physical, and Laboratory Examinations. Butterworths, Boston
A
20.
Monk TH, Buysse DJ, Reynolds CF, Kupfer DJ, Houck PR (1995) Circadian temperature rhythms of older people. Exp Gerontol 30:455–474
21.
Carskadon M, Dement W (1989) Normal Human Sleep: An Overview. Principles and Practice of Sleep Medicine. M.H. Kryger (Ed.). WB Saunders Phila. 3–13
22.
Reid KJ et al (2021) Effects of manipulating body temperature on sleep in postmenopausal women. Sleep Med 81:109–115
23.
Moyen NE et al (2024) Sleeping for One Week on a Temperature-Controlled Mattress Cover Improves Sleep and Cardiovascular Recovery. Bioengineering 11:352
24.
Bigalke JA, Cleveland EL, Barkstrom E, Gonzalez JE, Carter JR (2023) Core body temperature changes before sleep are associated with nocturnal heart rate variability. J Appl Physiol 135:136–145
25.
Okamoto-Mizuno K, Mizuno K (2012) Effects of thermal environment on sleep and circadian rhythm. J Physiol Anthropol 31:14
26.
Raymann RJEM, Swaab DF, Someren EJ (2008) W. V. Skin deep: enhanced sleep depth by cutaneous temperature manipulation. Brain 131:500
27.
Lack LC, Gradisar M, Van Someren EJW, Wright HR, Lushington K (2008) The relationship between insomnia and body temperatures. Sleep Med Rev 12:307–317
28.
Gómez-Santos C et al (2016) Menopause status is associated with circadian- and sleep-related alterations. Menopause 23:682–690
29.
Van Someren EJW, Raymann RJEM, Scherder EJA, Daanen HAM, Swaab DF (2002) Circadian and age-related modulation of thermoreception and temperature regulation: mechanisms and functional implications. Ageing Res Rev 1:721–778
30.
Neff LM et al (2016) Core body temperature is lower in postmenopausal women than premenopausal women: potential implications for energy metabolism and midlife weight gain. Cardiovasc Endocrinol 5:151–154
31.
Cintron D et al (2018) Effects of oral versus transdermal menopausal hormone treatments on self-reported sleep domains and their association with vasomotor symptoms in recently menopausal women enrolled in the Kronos Early Estrogen Prevention Study (KEEPS). Menopause 25:145–153
32.
Tranah GJ et al (2010) Postmenopausal hormones and sleep quality in the elderly: a population based study. BMC Womens Health 10:15
33.
Bakken K, Eggen AE, Lund E (2004) Side-effects of hormone replacement therapy and influence on pattern of use among women aged 45–64 years. The Norwegian Women and Cancer (NOWAC) study 1997. Acta Obstet Gynecol Scand 83:850–856
34.
Gupta N et al (2000) Thermoregulation and hormone replacement in postmenopausal women. J Therm Biol 25:165–169
35.
Zhang Z, DiVittorio JR, Joseph AM, Correa SM (2021) The Effects of Estrogens on Neural Circuits That Control Temperature. Endocrinology 162:bqab087
36.
Windred DP et al (2024) Higher central circadian temperature amplitude is associated with greater metabolite rhythmicity in humans. Sci Rep 14:16796
37.
Zhang S et al (2021) Metabolic flexibility during sleep. Sci Rep 11:17849
38.
Karatsoreos IN, Bhagat S, Bloss EB, Morrison JH, McEwen BS (2011) Disruption of circadian clocks has ramifications for metabolism, brain, and behavior. Proc. Natl. Acad. Sci. 108, 1657–1662
39.
Okamoto-Mizuno K, Tsuzuki K, Mizuno K, Ohshiro Y (2009) Effects of low ambient temperature on heart rate variability during sleep in humans. Eur J Appl Physiol 105:191–197
40.
Zhang Y, Xiang S, Cheng Z, Lu Y, Xu J (2024) Dynamic thermal comfort skin temperatures of young adults during sleep and the effects of gender and sleep stage. J Build Eng 98:111236
41.
Body Temperature, Saunders (2011) doi:10.1016/B978-1-4160-6645-3.00028-1
42.
Coull NA, West AM, Hodder SG, Wheeler P, Havenith G (2021) Body mapping of regional sweat distribution in young and older males. Eur J Appl Physiol 121:109–125
43.
Stachenfeld NS, Silva C, Keefe DL (2000) Estrogen modifies the temperature effects of progesterone. J Appl Physiol 88:1643–1649
44.
Caufriez A, Leproult R, L’Hermite-Balériaux M, Kerkhofs M, Copinschi G (2011) Progesterone Prevents Sleep Disturbances and Modulates GH, TSH, and Melatonin Secretion in Postmenopausal Women. J Clin Endocrinol Metab 96:E614–E623
45.
Roser M, Ritchie H, Spooner F (2021) Burden of Disease. Our World Data https://ourworldindata.org/burden-of-disease
46.
Gordon K et al (2019) Seven days of cold acclimation substantially reduces shivering intensity and increases nonshivering thermogenesis in adult humans. J Appl Physiol Bethesda Md 1985 126:1598–1606
47.
Harlow SD et al (2012) Executive Summary of the Stages of Reproductive Aging Workshop + 10: Addressing the Unfinished Agenda of Staging Reproductive Aging. J Clin Endocrinol Metab 97:1159–1168
48.
Haskell EH, Palca JW, Walker JM, Berger RJ, Heller HC (1981) The effects of high and low ambient temperatures on human sleep stages. Electroencephalogr Clin Neurophysiol 51:494–501
49.
Svensson T, Madhawa K, Chung NTH, U., Svensson AK (2024) Validity and reliability of the Oura Ring Generation 3 (Gen3) with Oura sleep staging algorithm 2.0 (OSSA 2.0) when compared to multi-night ambulatory polysomnography: A validation study of 96 participants and 421,045 epochs. Sleep Med 115:251–263
50.
Cao R et al (2022) Accuracy Assessment of Oura Ring Nocturnal Heart Rate and Heart Rate Variability in Comparison With Electrocardiography in Time and Frequency Domains: Comprehensive Analysis. J Med Internet Res 24:e27487
51.
Hasselberg MJ, McMahon J, Parker K (2013) The validity, reliability, and utility of the iButton® for measurement of body temperature circadian rhythms in sleep/wake research. Sleep Med 14:5–11
52.
Lan L, Xia L, Tang J, Wyon DP, Liu H (2019) Mean skin temperature estimated from 3 measuring points can predict sleeping thermal sensation. Build Environ 162:106292
53.
Koumar OC, Beaufils R, Chesneau C, Normand H, Bessot N (2023) Validation of e-Celsius gastrointestinal telemetry system as measure of core temperature. J Therm Biol 112:103471
54.
Wilson TE, Cui J, Crandall CG (2005) Mean body temperature does not modulate eccrine sweat rate during upright tilt. J Appl Physiol Bethesda Md 1985 98:1207–1212
55.
Pérez-Medina-Carballo R et al (2023) The circadian variation of sleep and alertness of postmenopausal women. Sleep 46
56.
Behbahani S, Jafarnia Dabanloo N, Motie Nasrabadi A, Dourado A (2018) Gender-Related Differences in Heart Rate Variability of Epileptic Patients. Am J Mens Health 12:117–125
57.
Schielzeth H et al (2020) Robustness of linear mixed-effects models to violations of distributional assumptions. Methods Ecol Evol 11:1141–1152
58.
Doyle MM et al (2022) Enhancing Cosinor Analysis of Circadian Phase Markers Using the Gamma Distribution. Sleep Med 92:1–3
59.
Cornelissen G (2014) Cosinor-based rhythmometry. Theor Biol Med Model 11:16
A
60.
Byrne C, Lim CL (2007) The ingestible telemetric body core temperature sensor: a review of validity and exercise applications. Br J Sports Med 41:126–133
A
61.
Sun WM, Houghton LA, Read NW, Grundy DG, Johnson AG (1988) Effect of meal temperature on gastric emptying of liquids in man. Gut 29:302–305
Yes
Total words in MS: 7708
Total words in Title: 0
Total words in Abstract: 0
Total Keyword count: 0
Total Images in MS: 0
Total Tables in MS: 1
Total Reference count: 61