The burden of sleep disturbances on quality of life and mental well-being in nearly 50,000 perimenopausal and postmenopausal women with and without concurrent vasomotor symptoms from the United States and Europe.

This was a secondary analysis of data collected from perimenopausal and postmenopausal women aged 40 to 65 years who participated in the cross-sectional NHWS—an annual self-administered online questionnaire designed to capture the health status of the general population. The survey operates through a stratified random sampling procedure based on age, sex, and ethnicity (United States only) among participating countries to provide samples representative of the demographic of the adult population of the respective country. Data were taken from the 2019 and 2021 surveys for US respondents (N=27,621) and from the 2017 and 2020 surveys for respondents from Europe (France, Germany, Italy, Spain, and the United Kingdom; N=20,220). Data from two survey years were used to ensure sufficient sample sizes and these were the most recent years with available data on selected sleep-related patient-reported outcomes. Only data from the most recent survey was used for women who completed surveys in both years. Data from earlier NHWS surveys (from 2010 and 2005) have been used previously in smaller studies of women with menopausal symptoms from either the US 6 , 7 or Europe. 12 This study was exempt from full or expedited review by the Pearl Institutional Review. Women were categorized as perimenopausal if they reported that their menstrual bleeding ceased ≤12 months ago or was irregular (in frequency, duration, or flow heaviness), and postmenopausal if they reported that their menstrual bleeding ceased >12 months ago. Both groups were subsequently categorized into four subgroups according to the presence/absence of self-reported sleep disturbances and VMS and: (i) absence of both symptoms (neither symptom), (ii) no sleep disturbances but presence of VMS (VMS only), (iii) sleep disturbances in the absence of VMS (sleep disturbances only), and (iv) presence of both sleep disturbances and VMS (both symptoms). The presence of VMS was defined as self-reported hot flashes/night sweats in the past 12 months (based on responding ‘yes’ to this specific question). Women were also classed as having VMS if they responded ‘yes’ to a question asking whether they had experienced menopausal symptoms (eg, hot flashes) in the past 12 months even if they had responded ‘no’ to the specific question on VMS; the reason for this was to minimize the chance of women with VMS being placed in the ‘neither symptom’ group. Sleep disturbance was defined as self-reported hot flashes/night sweats in the past 12 months or sleep symptoms experienced at least weekly, which included difficulty falling asleep, night-time awakenings, daytime sleepiness, or poor sleep quality (see Supplementary Methods in Supplemental Digital Content 1, http://links.lww.com/MENO/B428 )—all of which were considered as potentially related to the menopause transition or postmenopause. To evaluate HRQoL, data captured from four validated patient-reported outcome measures were used: Short-Form Health Survey (SF-36 for United States and SF-12vs2 for Europe) physical component summary (PCS) score and mental component summary (MCS) score, 13 - 15 EuroQol Visual Analogue Scale (EQ-VAS), and EQ-5D-5L. 16 , 17 The first three are measured using a scale of 0 to 100 and the latter is measured using a scale of 0 to 1; higher scores equate to better HRQoL. The EQ-5D-5L questionnaires were based on country-specific five-level value sets for each of the countries, that is, United States, France, Germany, Italy, Spain, and the United Kingdom. For the United Kingdom, the EQ-5D-3L was mapped to a five-level value set. 18 To evaluate mental well-being, we assessed depressive symptoms based on responses to the Patient Health Questionnaire (PHQ)-9 on a scale of 0 to 27, 19 and anxiety using the Generalized Anxiety Disorder assessment (GAD)-7 on a scale of 0 to 21. 20 For both measures, scores of 0 to 4 indicate no depression/anxiety risk, 5 to 9 mild risk, and 10 to 14 moderate risk. For PHQ-9, scores of 15 to 19 indicate moderately severe depression risk, and 20+ indicate severe depression risk. For GAD-7, scores of 15+ indicate severe anxiety risk. Further details can be found in Supplemental Digital Content 1, Supplementary Methods, http://links.lww.com/MENO/B428 . Data on the following self-reported sociodemographic, lifestyle, and physical and mental health variables were obtained: age, race/ethnicity (US respondents only), level of education, employment status, current insurance, body mass index, marital status, level of exercise, income, alcohol use, and smoking status. In Addition, data were obtained on Charlson Comorbidity Index, and diagnosis of clinical conditions in the previous 12 months including: endometriosis/fibroids, depression, generalized anxiety disorder, other mood conditions, and sleep conditions/sleep symptoms (other than sleep disturbance) experienced at least once per week. All covariates described were considered potential confounders for statistical modeling of associations between sleep disturbance, VMS, QoL, and mental well-being outcomes. Characteristics of perimenopausal and postmenopausal women and the four respective subgroups were summarized using frequency counts and percentages for categorical variables and means with SD or median with minimum and maximum for continuous variables. Bivariate analyses were used to assess patient characteristics, HRQoL, and mental well-being measures across all four subgroups. For categorical variables, χ 2 tests were used to determine potential differences between all groups, whereas ANOVA was used for continuous variables. Multivariable generalized linear regression models with appropriate distribution were conducted to evaluate associations between sleep disturbances/VMS (both absent, occurring independently, or co-occurring) and each HRQoL and mental well-being outcome measure. From these models, marginal means with 95% CIs were then calculated to assess differences between subgroups; one model was used for each outcome with adjustment made for all covariates described above. It should be noted that all reported p -values should be interpreted in a hypothesis-generating (rather than confirmatory) context, as no adjustment for multiplicity was carried out in this analysis. SPSS 29.0 was used for all analyses.

Data were analyzed from a total of 27,621 participants aged 40 to 65 years in the United States of whom 6,005 (21.7%) were perimenopausal and 21,616 (78.3%) were postmenopausal (Supplemental Fig S1, Supplemental Digital Content 2, http://links.lww.com/MENO/B429 ), and from 20,220 participants aged 40 to 65 years in Europe of whom 5,312 (26.3%) were perimenopausal and 14,908 (73.3%) were postmenopausal (Supplemental Fig S2, Supplemental Digital Content 2, http://links.lww.com/MENO/B429 ). Characteristics of these women are shown in Table 1 , as well as Supplemental Tables S1 and S2, Supplemental Digital Content 2, http://links.lww.com/MENO/B429 . Mean age of perimenopausal women was 48.5 years (SD±5.1; United States) and 48.6 (SD±4.6; Europe), and mean age of postmenopausal women was 57.6 years (SD±5.7; United States) and 57.1 (SD±5.8; Europe). Ethnicity/race distribution (United States only) was as follows: 64.3% White, 15.5% Hispanic, 11.0% Black/African American, 5.4% Asian, and 3.8% other (The specific categories included in the ‘other’ section are unclear due to the survey’s design. For the question regarding race and ethnicity, participants were instructed to select the ‘other’ category if they did not identify as White, African American/Black, Asian, or Hispanic/Latino, with no additional information requested.). Characteristics of the study population—United States and Europe BMI, body mass index; NA, not applicable. Data on race/ethnicity were only available for US women. The specific categories included in the ‘other’ section are unclear due to the survey’s design. For the question regarding race and ethnicity, participants were instructed to select the ‘other’ category if they did not identify as White, African American/Black, Asian, or Hispanic/Latino, with no additional information requested. Data are n (%) unless otherwise specified. All variables shown were included as potential confounders in the regression models, in addition to endometriosis/fibroids, other mood conditions, and sleep conditions/sleep symptoms (other than sleep disturbance) experienced at least once per week. Further characteristics of the study population are found in Tables S1 and S2, Supplemental Digital Content 2, http://links.lww.com/MENO/B429 . Vasomotor symptoms were reported by 60.1% (United States) and 60.3% (Europe) of perimenopausal women (Supplemental Fig S1, Supplemental Digital Content 2, http://links.lww.com/MENO/B429 ) and 43.4% (United States) and 44.0% (Europe) of postmenopausal women (Supplemental Fig S2, Supplemental Digital Content 2, http://links.lww.com/MENO/B429 ). Sleep disturbances were reported by 53.8% (14,851/27,621) women from the United States and 51.3% (10,375/20,220) women from Europe; 52.3% (United States) and 52.7% (Europe) for perimenopausal women (Supplemental Fig S1, Supplemental Digital Content 2, http://links.lww.com/MENO/B429 ) and 54.2% (United States) and 50.8% (Europe) of postmenopausal women (Supplemental Fig S2, Supplemental Digital Content 2, http://links.lww.com/MENO/B429 ). Among perimenopausal women, sleep disturbances were reported by 61.7% (United States) and 60.6% (Europe) with VMS, and 38.0% (United States) and 40.8% (Europe) without VMS. Among postmenopausal women, sleep disturbances were reported by 66.7% (United States) and 63.4% (Europe) with VMS, and 44.5% (United States) and 40.9% (Europe) without VMS. Bivariate analyses suggested potential differences across subgroups for patient characteristics and each HRQoL and mental well-being outcome for both perimenopausal and postmenopausal women (both United States and Europe; Supplemental Tables S3 to S6, Supplemental Digital Content 2, http://links.lww.com/MENO/B429 ). Estimated marginal means (EMMs) with 95% CIs for HRQoL outcomes among each subgroup of perimenopausal and postmenopausal women are shown in Figure 1 for SF-36/SF-12v2 MCS and PCS scores, and in Figure 2 for EQ-5D-5L and EQ-VAS scores (EMMs and pairwise p -values for all subgroups are included in Supplemental Tables S7-S12, Supplemental Digital Content 2, http://links.lww.com/MENO/B429 ). Among European women, the lowest scores (worst HRQoL) were seen in women with both sleep disturbances and VMS (EMMs for postmenopausal women were 44.7 for SF-12v2 MCS (overall mental well-being), 47.7 for SF-12v2 PCS (overall physical health status), 0.77 for EQ-5D-5L (HRQoL measured across five dimensions), and 68.5 for EQ-VAS (overall self-reported health); EMMs for perimenopausal women were 42.3 for SF-12v2 MCS, 48.9 for SF-12v2 PCS, 0.76 for EQ-5D-5L, and 67.5 for EQ-VAS. The second lowest scores (worse HRQoL) were seen in women with sleep disturbances only (EMMs for postmenopausal women were 45.4 for SF-12v2 MCS, 47.9 for SF-12v2 PCS, 0.78 for EQ-5D-5L, and 68.5 for EQ-VAS; EMMs for perimenopausal women were 43.0 for SF-12v2 MCS, 49.7 for SF-12v2 PCS, 0.79 for EQ-5D-5L, and 69.3 for EQ-VAS). Although this pattern was not as evident among US women, among postmenopausal women from the US scores were lower (worse HRQoL) among the two subgroups of women with sleep disturbances than the VMS only and neither symptom subgroups; among perimenopausal women this was only seen for the EQ-VAS score. Adjusted marginal mean (95% CI) SF-36/SF-12v2 (A) mental component summary and (B) physical component summary scores. Lower scores indicate worse HRQoL. P -values shown are from the linear regression analysis comparing symptomatic subgroups to the subgroup with neither symptom. Nonoverlapping CIs in this figure indicate differences ( P <0.05) for each respective perimenopausal and postmenopausal subgroup. Pairwise p -values for specific subgroup comparisons are available in Supplementary Tables 9a and 9b, Supplemental Digital Content 2, http://links.lww.com/MENO/B429 . All variables shown in Table 1 were included as potential confounders in the regression models, in addition to endometriosis/fibroids, other mood conditions, and sleep conditions/sleep symptoms (other than sleep disturbance) experienced at least once per week. HRQoL, health-related quality of life; MCS, mental component summary; PCS, physical component summary; SF, Short-Form; VMS, vasomotor symptoms. Adjusted marginal mean (95% CI) (A) EQ-VAS and (B) EQ-5D-5L scores. Lower scores indicate worse HRQoL. P -values shown are from the linear regression analysis comparing symptomatic subgroups to the subgroup with neither symptom. Nonoverlapping CIs in this figure indicate differences ( P <0.05) for each respective perimenopausal and postmenopausal subgroup. Pairwise p -values for specific subgroup comparisons are available in Supplementary Tables 12a and 12b, Supplemental Digital Content 2, http://links.lww.com/MENO/B429 . All variables shown in Table 1 were included as potential confounders in the regression models, in addition to endometriosis/fibroids, other mood conditions, and sleep conditions/sleep symptoms (other than sleep disturbance) experienced at least once per week. EQ-VAS, EuroQol Visual Analog Scale; HRQoL, health-related quality of life; VAS, Visual Analog Scale; VMS, vasomotor symptoms. After adjusting for confounders, postmenopausal women with sleep disturbances had lower mean outcome scores (worse HRQoL) than women with neither sleep disturbances nor VMS irrespective of whether they also had VMS ( P <0.001); this was seen for all four patient-reported outcomes in women from both the United States and Europe. Relatively lower HRQoL among perimenopausal women with sleep disturbances with/without VMS was apparent for some, but not all, HRQoL outcome measures; for instance, P ≤0.01 for EQ-VAS and P <0.05 for SF-36/SF-12v2 MCS and PCS (among European women). In addition, among postmenopausal women, those with sleep disturbances alone had lower HRQoL scores than those with VMS alone ( P <0.001). Among postmenopausal women, those with VMS had lower HRQoL scores than those with neither sleep disturbances nor VMS, irrespective of whether they also had sleep disturbances ( P <0.001 for all subgroups from the United States and Europe apart for EQ-VAS among US women where P =0.175). Among perimenopausal women, relatively lower HRQoL was seen for those with VMS with/without sleep disturbances for some outcome measures among European women; for instance, P =0.004 for SF-12v2 MCS, P =0.015 for PCS, P =0.096 for EQ-VAS, and P =0.006 for EQ-5D-5L. The subgroup with both sleep disturbances and VMS had the highest proportion of women with moderate-to-severe anxiety and depression (Supplemental Fig S3 and S4, respectively, Supplemental Digital Content 2, http://links.lww.com/MENO/B429 ). Estimated marginal means with 95% CIs for PHQ-9 (depression) and GAD-7 (anxiety) scores for each subgroup among peri- and postmenopausal women are shown in Figure 3 (pairwise p -values for all subgroups are shown in Supplemental Tables S13 and S15, Supplemental Digital Content 2, http://links.lww.com/MENO/B429 ). Apart from PHQ-9 scores in US perimenopausal women, the highest PHQ-9 and GAD-7 scores were seen in women with both sleep disturbances and VMS (EMMs for postmenopausal women were 5.3 [United States] and 5.1 [Europe] for PHQ-9, and 3.8 [United States] and 5.1 [Europe] for GAD-7; EMMs for perimenopausal women were 6.9 [United States] and 8.0 [Europe] for PHQ-9, and 5.1 [United States] and 6.2 [Europe] for GAD-7). The second highest PHQ-9 and GAD-7 scores were seen in women with sleep disturbances only (EMMs for postmenopausal women were 5.0 [United States] and 4.5 [Europe] for PHQ-9, and 3.4 [United States] and 4.5 [Europe] for GAD-7; EMMs for perimenopausal women were 7.3 [United States] and 7.5 [Europe] for PHQ-9, and 5.0 [United States] and 6.7 [Europe] for GAD-7). Notably, for both the US and Europe, after adjusting for confounders, both postmenopausal and perimenopausal women with sleep disturbances had higher PHQ-9 and GAD-7 scores than women with neither symptom irrespective of whether they also had VMS ( P <0.001 for postmenopausal and P <0.05 for perimenopausal). In addition, among both perimenopausal and postmenopausal women, those with sleep disturbances only had lower PHQ-9 and GAD-7 scores than those with VMS only (for PHQ-9, P <0.001 for both perimenopausal and postmenopausal; for GAD-7, P <0.001 for postmenopausal and P <0.05 for perimenopausal). Adjusted marginal mean (A) PHQ-9 (depression) and (B) GAD-7 (anxiety) scores. Note higher scores indicate worse depression/anxiety. P -values shown are from the linear regression analysis comparing symptomatic subgroups to the subgroup with neither symptom. Nonoverlapping CIs in this figure indicate differences ( P <0.05) for each respective perimenopausal and postmenopausal subgroup. Pairwise p -values for specific subgroup comparisons are available in Supplementary Tables 15a and 15b, Supplemental Digital Content 2, http://links.lww.com/MENO/B429 . All variables shown in Table 1 were included as potential confounders in the regression models, in addition to endometriosis/fibroids, other mood conditions, and sleep conditions/sleep symptoms (other than sleep disturbance) experienced at least once per week. GAD, Generalized Anxiety Disorder; PHQ, Patient Health Questionnaire; VMS, vasomotor symptoms. Among postmenopausal women from Europe, those with VMS had higher PHQ-9 and GAD-7 scores than those with neither symptom, irrespective of whether they also had sleep disturbances; however, among US postmenopausal women, PHQ-9 and GAD-7 scores among women with VMS only were not noticeably different to those with neither symptom. Relatively higher PHQ-9 and GAD-7 scores were seen among European perimenopausal women with VMS with/without sleep disturbances ( P =0.001 for PHQ-9 and p ≤0.017 for GAD-7), but not among perimenopausal US women.

In this large-scale study of nearly 50,000 perimenopausal/postmenopausal women, sleep disturbances were associated with worse HRQoL, depression, and anxiety than VMS alone or absence of both symptoms. These associations were noted irrespective of co-existing VMS, and after adjusting for sociodemographic, lifestyle, and other confounding factors. Overall, these associations were generally stronger among postmenopausal women. Our results show the broad burden of sleep disturbances in the presence or absence of concomitant VMS and highlight a need for proper symptom recognition and prompt management to improve outcomes and well-being among these women. They build upon the existing knowledge on the negative impact of sleep disturbances and VMS on HRQoL in menopausal women from smaller studies using less contemporaneous data and different outcomes measures. 2 - 8 , 12 , 21 , 22 Furthermore, they support previous research demonstrating a clear link between sleep disturbances and declining mental health during the menopausal years, 23 and the recognition that the perimenopausal and early postmenopausal years are a ‘window of vulnerability’ for mood and anxiety disorders. 23 - 25 Associations between VMS and depressive symptoms/anxiety have also been reported previously. 26 - 28 In our present study, depression and anxiety scores were worse in perimenopausal than postmenopausal women across subgroups. It should be noted, however, that no formal comparisons were made between these two groups. Health-related quality of life measured by the SF-36/SF-12v2 MCS, EQ-5D-5L, and EQ-VAS, was also worse in perimenopausal, whereas scores on the physical component of the SF-36/SF-12v2 were worse in postmenopausal women. In an analysis from SWAN, however, menopause stage itself was not found to be associated with HRQoL after adjusting for a wide range of factors including symptoms, sociodemographics, medical conditions, and psychological factors. 3 , 29 Our findings indicate a clear unmet need among menopausal women experiencing sleep disturbances because there are no pharmacological treatments specifically indicated for likely menopausal-related sleep disturbances. This is important considering that sleep disturbances were reported by over 50% of perimenopausal/postmenopausal women in this study. In addition, smaller studies have demonstrated that women can be grouped into phenotypic clusters, including those where sleep problems predominate over VMS. 10 , 11 Furthermore, the occurrence of these two symptoms independently supports the current view that they are not always intrinsically linked and that different pathophysiological pathways could be involved. 23 It is noteworthy that among postmenopausal women from both continents, the combined impact of sleep disturbances and VMS on depression and anxiety was mostly greater than when either symptom occurred individually. Similarly, for HRQoL, the co-occurrence of both sleep disturbances and VMS was greater than the occurrence of just one of these symptoms, although only in European women (both post- and perimenopausal). This observation, along with the fact that postmenopausal women with VMS reported lower HRQoL scores compared with those without these symptoms—regardless of whether they also faced sleep disturbances—highlights the importance of promptly recognizing symptoms. Whether these symptoms occur independently or together, it is essential to tailor management strategies to meet individual needs. A key strength of this study is the use of contemporary data from a population-based large sample of US and European women deemed representative of the demographic of the wider menopausal population. Women included were those at any stage of their menopausal journey (ie, perimenopausal/postmenopausal) irrespective of whether they were symptomatic and/or seeking treatment, and overall, our results can be considered to have good external validity. Outcomes were measured using several validated patient-reported measures and the large sample size meant analyses were sufficiently powered to identify associations of interest. Further, the EMMs and 95% CIs enabled the interpretation of between-individual subgroups, which pointed towards overall trends across the various outcome measures. There were no missing data as the NHWS requires responses for all survey questions, and we adjusted for a wide range of confounders including sociodemographics, lifestyle factors, and medical conditions. Limitations include the cross-sectional study design meaning results could not infer potential directions of effect of the observed associations. The self-reported data could have led to some misclassification of study variables due to entry inaccuracies, and recall bias from reliance on memory. Because of the definition used for experiencing VMS, it is possible that some women without VMS were misclassified as having VMS and placed in one of the VMS subgroups. Also, some selection bias is possible due to women with less internet access, time availability, and/or interest in the topic being unrepresented. Lastly, although analyses were adjusted for several confounders, residual confounding remains possible.

In conclusion, sleep disturbances were common among menopausal women irrespective of VMS, and were independently associated with negative effects on HRQoL, and mental well-being. Effective management of sleep disturbances and VMS in menopausal women—through proper and timely recognition and prompt treatment—is needed to mitigate the associated burden and improve overall well-being.