Is endometriosis associated with poor sleep quality? A meta-analysis

Endometriosis is a hormone-driven chronic systemic inflammatory condition, characterized by the growth of endometrial-like lesions throughout the body, that affects up to 15% of women worldwide. 1 – 3 Symptomatology varies greatly among patients and is not correlated to the severity and number of endometriosis lesions. 4 Common symptoms include severe abdominal pain, chronic musculoskeletal pain, persistent fatigue, gastrointestinal issues, and infertility. 5 – 7 Additionally, endometriosis has been associated with a significant increase in depression and anxiety, with a 2023 study identifying a genetic link between endometriosis and these mental health conditions. 8 – 10 Many endometriosis patients report severe reductions in quality of life due to both the mental and physical symptoms they experience. 11 , 12 This reduction in quality of life associated with endometriosis has substantial negative impacts on both individuals and society as a whole. A 2018 study estimated that absenteeism and increased medical costs attributable to endometriosis across the United States lead to an annual financial impact of 69.4 billion dollars. 13 In line with these results, a study of 6,925 endometriosis patients found that over 45% of patients reported that endometriosis pain impacted activities such as socializing and walking, and over 50% reported their symptoms impacted getting out of bed or working. 14 Additionally, 42.6% of those patients reported endometriosis impacted their sleep quality. Despite these findings medical definitions and physicians rarely consider low sleep quality as a symptom of endometriosis. Sleep quality is a well-established component of quality of life and studies show that poor sleep quality harms both mental and physical health. 15 – 17 Overall sleep quality is generally evaluated on four aspects (sleep efficiency, sleep latency, sleep duration, and wake after sleep onset) and encompasses an individual’s satisfaction with these aspects as well. 18 A disruption in any of these four areas can impact an individual’s daily life and is associated with an increased risk of short and long-term health conditions, including psychosocial issues and cardiovascular disease. 19 Similarly, a 2024 study showed that after just two days of low sleep duration (4 hours in bed per night), individuals felt 4.44 years older compared to those who had 9 hours in bed per night. 20 Poor sleep quality and related sleep disorders are present globally with prevalence varying significantly throughout populations; however, women consistently experience lower sleep quality than men, and low sleep quality is a primary complaint of over 65% of individuals with chronic pain conditions. 21 – 26 The literature shows this association between chronic pain and low sleep quality is potentially bidirectional, but sleep quality tends to be a stronger predictor of chronic pain than the reverse. 27 , 28 In 2024, Sumbodomet et al. conducted a qualitative review of the current literature on the potential association of endometriosis and sleep quality. 29 The authors reported on nine papers that evaluated the difference in sleep quality between individuals with endometriosis and those without endometriosis. They reported evidence supporting an association between symptomatic endometriosis and poor sleep quality; however, no quantitative aggregation of sleep quality scores across studies was performed. Therefore, to draw more robust conclusions about the relationship and potential effect size across populations, and to better inform researchers and clinicians about sleep disturbance as an underrecognized symptom of endometriosis, we conducted a meta-analysis of studies evaluating quantitative differences in overall sleep quality between endometriosis patients and healthy individuals.

We performed a systematic literature search and review for papers related to sleep quality and endometriosis following PRISMA 2020 guidelines. 30 The review and meta-analysis were registered with PROSPERO (CRD42024592272). Predefined objectives, eligibility criteria, the search strategy (PubMed Boolean logic and sleep-quality terms are provided in the manuscript), and data extraction procedures were determined before data extraction; a formal, standalone protocol or statistical analysis plan was not prepared. Generative AI was used to confirm the following Boolean search logic and related sleep quality terms were used to search PubMed 31 : Endometriosis AND (“Sleep Disorders”[MeSH Terms] OR “Sleep Wake Disorders”[MeSH Terms] OR “Dyssomnias”[MeSH Terms] OR “Parasomnias”[MeSH Terms] OR “Insomnia”[MeSH Terms] OR “Sleep Apnea”[MeSH Terms] OR “Narcolepsy”[MeSH Terms] OR “Restless Legs Syndrome”[MeSH Terms] OR “Circadian Rhythm Sleep Disorders”[MeSH Terms] OR “Hypersomnia”[MeSH Terms] OR “Parasomnias”[MeSH Terms] OR “Sleep Bruxism”[MeSH Terms] OR “Sleep Paralysis”[MeSH Terms] OR “Sleep Disorders”[Text Word] OR “Sleep Disturbances”[Text Word] OR “Sleep Issues”[Text Word] OR “Sleep Problems”[Text Word] OR “Sleep Apnea Syndromes”[Text Word] OR “Obstructive Sleep Apnea”[Text Word] OR “Central Sleep Apnea”[Text Word] OR “Primary Insomnia”[Text Word] OR “Secondary Insomnia”[Text Word] OR “Narcolepsy with Cataplexy”[Text Word] OR “Narcolepsy without Cataplexy”[Text Word] OR “Willis-Ekbom Disease”[Text Word] OR “Delayed Sleep Phase Disorder”[Text Word] OR “Advanced Sleep Phase Disorder”[Text Word] OR “Non-24-Hour Sleep-Wake Disorder”[Text Word] OR “Shift Work Sleep Disorder”[Text Word] OR “Idiopathic Hypersomnia”[Text Word] OR “Sleepwalking”[Text Word] OR “Somnambulism”[Text Word] OR “Night Terrors”[Text Word] OR “Sleep Terrors”[Text Word] OR “REM Sleep Behavior Disorder”[Text Word] OR “Sleep Talking”[Text Word] OR “Somniloquy”[Text Word] OR “Sleep Quality”[Text Word] OR “Sleep Hygiene”[Text Word] OR “Sleep Deprivation”[Text Word] OR “Excessive Daytime Sleepiness”[Text Word] OR “Nightmares”[Text Word]). Endometriosis AND (“Sleep Disorders”[MeSH Terms] OR “Sleep Wake Disorders”[MeSH Terms] OR “Dyssomnias”[MeSH Terms] OR “Parasomnias”[MeSH Terms] OR “Insomnia”[MeSH Terms] OR “Sleep Apnea”[MeSH Terms] OR “Narcolepsy”[MeSH Terms] OR “Restless Legs Syndrome”[MeSH Terms] OR “Circadian Rhythm Sleep Disorders”[MeSH Terms] OR “Hypersomnia”[MeSH Terms] OR “Parasomnias”[MeSH Terms] OR “Sleep Bruxism”[MeSH Terms] OR “Sleep Paralysis”[MeSH Terms] OR “Sleep Disorders”[Text Word] OR “Sleep Disturbances”[Text Word] OR “Sleep Issues”[Text Word] OR “Sleep Problems”[Text Word] OR “Sleep Apnea Syndromes”[Text Word] OR “Obstructive Sleep Apnea”[Text Word] OR “Central Sleep Apnea”[Text Word] OR “Primary Insomnia”[Text Word] OR “Secondary Insomnia”[Text Word] OR “Narcolepsy with Cataplexy”[Text Word] OR “Narcolepsy without Cataplexy”[Text Word] OR “Willis-Ekbom Disease”[Text Word] OR “Delayed Sleep Phase Disorder”[Text Word] OR “Advanced Sleep Phase Disorder”[Text Word] OR “Non-24-Hour Sleep-Wake Disorder”[Text Word] OR “Shift Work Sleep Disorder”[Text Word] OR “Idiopathic Hypersomnia”[Text Word] OR “Sleepwalking”[Text Word] OR “Somnambulism”[Text Word] OR “Night Terrors”[Text Word] OR “Sleep Terrors”[Text Word] OR “REM Sleep Behavior Disorder”[Text Word] OR “Sleep Talking”[Text Word] OR “Somniloquy”[Text Word] OR “Sleep Quality”[Text Word] OR “Sleep Hygiene”[Text Word] OR “Sleep Deprivation”[Text Word] OR “Excessive Daytime Sleepiness”[Text Word] OR “Nightmares”[Text Word]). Next, we conducted an Embase search based upon PICO search criteria with the population set as “endometriosis/exp,” intervention and comparison left blank, outcome set to “sleep quality/exp” with all synonyms included, and study design as “case control study/exp” to filter to studies which compared health controls with endometriosis patients. The last update to these searches was October 2025. First author (JP) reviewed titles and abstracts for inclusion, eliminating papers that were not peer-reviewed, not published in English, or unrelated to endometriosis and sleep quality. We also reviewed the citations of all papers marked for full-text review for further papers to include. During the full-text review, evaluated the remaining papers for inclusion based on the following criteria: non-ecologic study design, original data was reported, overall sleep quality was compared between endometriosis patients and healthy individuals on a quantitative scale, and the mean and standard deviation (SD) or median and interquartile range (IQR) of the sleep quality scores were reported for endometriosis patients and controls. A second author (KH) then independently review the inclusion and exclusion decisions as validation and no discrepancies were found. Authors who reported sleep quality on a quantitative scale but did not include the mean and standard deviation were contacted for further data (3 studies). A traditional risk-of-bias assessment was not performed because these assessments have been shown to require highly subjective judgments and can be inconsistent and difficult to reproduce; instead authors jointly identified specific areas that could introduce bias into the aggregated results based on existing literature and the research. 32 – 35 Author JP then assessed how each study’s potential risk of bias might influence the results of the meta-analysis in terms of direction (toward the null, no effect, away from the null, or potentially either) and magnitude (small, medium, large) and confirming these assertions with author KH through discussion. Disagreements were settled through use of existing literature and directed acrylic graphs as needed ( Supplemental Table 1 ). Threats to potential selection bias were identified as whether controls were matched to cases, whether controls were screened for endometriosis, and whether there was a likelihood of Berkson’s bias due to use of hospital-based recruitment. Threats to potential information bias were determined to be greater when a scale other than the Pittsburgh Sleep Quality Index (PSQI), the predominant tool in most studies, was used, and if median/IQR was reported rather than mean/SD. The most important potential confounders were identified as age, body mass index, and hormonal therapy. Confounding was considered a potential source of bias if researchers either did not assess these variables or found a significant difference between cases and controls but did not account for it through adjustment, exclusion, or matching. We planned to evaluate publication bias using funnel plot techniques, Begg’s rank test, and Egger’s regression test; however all methods were underpowered due to limited sample size of included papers. For each study that met inclusion criteria, the following were extracted by author JP and reviewed by author KH: first author, year of publication, year(s) of data collection, the country where data was collected, the sample size for endometriosis patients and controls, method of endometriosis diagnosis, how controls were chosen and matched, which scale was used to measure sleep quality, covariates and potential confounders examined in the study or adjusted for in models, mean and SD or median and IQR of sleep quality scores for endometriosis patients and controls, whether power calculations were stated in the methodology, and any other quality-of-life measures examined ( Supplemental Tables 2 ). The abstraction was conducted a second time, blinded to the first abstraction, for quality control purposes. Data were standardized in preparation of the meta-analysis. To enable comparison between sleep scales, scales in which higher scores reflected better sleep quality were reversed by negating the means (mean = -mean), while standard deviations remained unchanged (SD = SD) so that higher scores consistently indicated poorer sleep quality. 36 If a study reported median and interquartile range (IQR) for sleep scores, the median and IQR/1.35 were substituted for the mean and standard deviation. 36 , 37 We performed a mixed effect meta-analysis pooling between-group standardized mean differences (SMD) from all studies. We used Knapp-Hartung and Hedge’s g adjustments to account for anticipated high heterogeneity between studies. 38 , 39 Results were expressed in a forest plot, and SMD was re-expressed for interpretability in terms of the most commonly used overall sleep quality scale, the PSQI. SMDs were multiplied by the average SD (3.534) in PSQI studies. We performed sensitivity analyses to examine how study variations may impact the results. First, we performed a leave-one-out analysis, excluding studies that did not use the PSQI scale to evaluate sleep quality. Next, we excluded studies that reported median and interquartile ranges instead of mean and SD. To assess how study differences may impact results and heterogeneity, we performed a univariate meta-regression on the year data collection started and ended, the case and control sample sizes in each study, and the location of data collection. Statistical analyses were carried out using R Version 4.4.2, and the metaphor package. 40 , 41

The literature search produced fifty-two potential papers ( Figure 1 ), of which six met the inclusion criteria for the meta-analysis ( Table 1 ). 42 – 47 Three additional papers included an evaluation of endometriosis patients and healthy individuals but did not report mean/SD or median/IQR. 48 – 50 The authors did not respond to requests for data and the papers were not included in the final 6. Research was performed in Brazil, Italy, Iran, and Spain. Five of the studies were conducted between 2016-2022; Nunes et al. 2015 was conducted earlier but did not report their collection years. All but one study used the PSQI, which represents better sleep quality with lower values, and the mean sleep scores ranged from 6.47 to 11.00 for endometriosis patients and 4.45 to 7.10 for healthy controls. 51 Nunes et al. 2015 used the Post Sleep Inventory (PSI) scale, which represents better sleep quality with a higher number, and reported a mean sleep score of 5.68 in endometriosis patients and a mean of 6.04 in healthy controls. 52 Sample size varied considerably between studies. Álvarez-Salvago et al., 2020 reported on the smallest study of 25 cases and 25 controls, while Iannuzzo et al., 2024 reported on the largest sample of 430 cases and 417 controls. All six studies reported statistically significantly poorer average sleep quality among endometriosis patients compared to healthy controls. Figure 1. Literature search and review flowchart for selection of studies. Table 1. Studies included in the meta-analysis of endometriosis and sleep quality and their characteristics. Reference Location Study years Sample size Sleep quality scale Sleep quality scores (mean, SD) Endo Control Endo Control Álvarez-Salvago, 2020 42 Spain 2016-2017 25 25 PSQI 8.32 (3.73) 6 (3.19) Chaichian, 2024 43 Iran 2019-2022 46 202 PSQI 10.60 (3.70) 7.1 (2.7) Facchin, 2021 44 Italy 2019-2020 123 123 PSQI 6.68 (3.59) 5.45 (3.03) Iannuzzo, 2024 ** , 45 Italy 2021 430 417 PSQI 11.00 (6.67) 5.00 (2.20) Nunes, 2015 46 Brazil NR 257 253 PSI * 5.68 (1.55) 6.04 (1.62) Youseflu, 2020 47 Iran 2016-2017 78 78 PSQI 6.47 (3.34) 4.45 (3.26) *Score on original scale PSI was reverse coded for analysis. **Reported median and interquartile range. Endo=Endometriosis, NR=Not Reported, PSQI=Pittsburgh Sleep Quality Index, PSI=Post-Sleep Inventory. Literature search and review flowchart for selection of studies. Studies included in the meta-analysis of endometriosis and sleep quality and their characteristics. *Score on original scale PSI was reverse coded for analysis. **Reported median and interquartile range. Endo=Endometriosis, NR=Not Reported, PSQI=Pittsburgh Sleep Quality Index, PSI=Post-Sleep Inventory. Potential threats to bias varied across the included studies ( Table 2 ). The most common area for potential bias was selection bias, with four of the six studies having one or more areas of potential selection bias identified. Two studies (Iannuzzo et al., 2024 and Nunes et al., 2015) were identified as having potential information bias threats. Four studies had potential confounding bias, primarily due to a lack of assessing or controlling for the use of hormonal therapies in endometriosis patients. One study (Facchin et al., 2021) met all requirements to avoid potential bias. The potential sources of bias identified in Nunes et al., 2015 were assessed as having a moderate impact on the results, potentially shifting them in either direction from the null depending on the strength of the biases. Chaichian et al., 2024 had the largest risk of bias toward the null due to high chance of selection bias and lack of screening for endometriosis in controls. Two studies (Iannuzo et al., 2024 and Youseflu et al., 2020) were scored to have a moderate risk of impact toward the null due to selection and information bias, and Berkson’s bias respectively. Alvarez-Salvago et al., 2020 was evaluated to have a small risk of bias toward the null due to potential confounding through HRT. Table 2. Assessment of threats to validity across studies included in meta-analysis. ​ Potential selection bias Potential information bias Potential confounding bias * Impact on study Reference Matching of controls Berkson’s bias Controls screened for endo PSQI sleep scale used Reported mean and SD Age assessed/controlled for BMI assessed/controlled for Use of HT assessed/controlled for Magnitude Direction Álvarez-Salvago, 2020 42 ✓ ✓ ✓ ✓ ✓ ✓ ✓ X small toward null Chaichian, 2024 43 X X X ✓ ✓ ✓ ✓ X large toward null Facchin, 2021 44 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ none none Iannuzzo, 2024 45 X ✓ X ✓ X ✓ ✓ ✓ moderate toward null Nunes, 2015 46 X ✓ X X ✓ X X ✓ moderate Potentially either Youseflu, 2020 47 ✓ X ✓ ✓ ✓ ✓ ✓ X moderate toward null ✓= avoided potential bias. X = did not avoid potential bias. Endo=endometriosis, PSQI= Pittsburgh Sleep Quality Index, SD= standard deviation, BMI= Body Mass Index, HT = hormonal therapy. *Confounding was considered as at risk for potential bias if researchers did not compare the distributions of the variables between cases and controls, or they did compare the distributions, found a significant difference between cases and controls, but the variable was not adjusted for, excluded on, or matched upon. Assessment of threats to validity across studies included in meta-analysis. ✓= avoided potential bias. X = did not avoid potential bias. Endo=endometriosis, PSQI= Pittsburgh Sleep Quality Index, SD= standard deviation, BMI= Body Mass Index, HT = hormonal therapy. *Confounding was considered as at risk for potential bias if researchers did not compare the distributions of the variables between cases and controls, or they did compare the distributions, found a significant difference between cases and controls, but the variable was not adjusted for, excluded on, or matched upon. The pooling of all six studies produced a standardized mean difference of 0.69 (95% CI: 0.28,1.09) with high heterogeneity (I 2 = 94% p<0.01), which equates to a 2.44 (95%CI: 0.99, 3.85) point increase on the PSQI scale for endometriosis patients compared to healthy controls ( Figure 2 ). When the single study that did not use the PSQI scale was removed from the meta-analysis, the pooled SMD was 0.79 (95% CI:0.36,1.23), which represents a 2.79 (95% CI: 1.27, 4.35) point increase on the PSQI scale for endometriosis patients compared to healthy controls with high heterogeneity between studies (I 2 : 89% p<0.01) ( Figure 3 ). When the study that reported median and IQR was removed, the pooled SMD was 0.57 (95% CI: 0.17, 0.97) with high heterogeneity (I 2 = 91%, p<0.01), representing a 2.01 (95% CI: 0.60,3.43) point increase on the PSQI scale for endometriosis patients compared to healthy controls ( Figure 4 ). Figure 2. Forest plot of the standardized mean difference meta-analysis of studies comparing sleep quality of endometriosis patients to healthy controls. Figure 3. Forest Plot of the standardized mean difference meta-analysis of studies comparing sleep quality of endometriosis patients to healthy controls, excluding study that did not use the pittsburgh sleep quality index. Figure 4. Forest plot of the standardized mean difference meta-analysis of studies comparing sleep quality of endometriosis patients to healthy controls, excluding study that reported median and interquartile range. Forest plot of the standardized mean difference meta-analysis of studies comparing sleep quality of endometriosis patients to healthy controls. Forest Plot of the standardized mean difference meta-analysis of studies comparing sleep quality of endometriosis patients to healthy controls, excluding study that did not use the pittsburgh sleep quality index. Forest plot of the standardized mean difference meta-analysis of studies comparing sleep quality of endometriosis patients to healthy controls, excluding study that reported median and interquartile range. Heterogeneity between studies remained high in univariate meta-regression analyses (I 2 ranging from 79.4%-90.59%) ( Table 3 ). Meta-analysis results indicated much of the differences in true effect sizes were explained by the location of data collection (53.04%), the sample size of endometriosis patients (35.96%), and the year of study (year started: 17.51%, year ended:14.32%). Table 3. Results of meta-regression examining how study characteristics impacted heterogeneity (I 2 ) of studies included in meta-analysis. Study characteristic Difference in effect size (R 2 ) Residual heterogeneity (I 2 ) Reduction in heterogeneity (94% - I 2 ) Location of Data Collection 53.04% 79.41% 14.59% Year Study Began 17.51% 85.8% 8.92% Year Study ended 14.32% 87.56% 6.44% Endometriosis patient sample size 35.96% 88.16% 5.84% Control sample size 8.13% 90.59% 3.41% R 2 = Measure of difference in effect size between studies included in meta-analysis, I 2 = Measure of heterogeneity between studies included in meta-analysis. Results of meta-regression examining how study characteristics impacted heterogeneity (I 2 ) of studies included in meta-analysis. R 2 = Measure of difference in effect size between studies included in meta-analysis, I 2 = Measure of heterogeneity between studies included in meta-analysis. Due to a small number of studies that met inclusion criteria, publication bias was not evaluated as funnel plots and regression-based methods were all underpowered.

This study highlights the potential for significant and clinically meaningful associations between endometriosis and reductions in sleep quality, which should be further investigated in ongoing high quality longitudinal research. Since sleep quality predicts chronic pain patterns, a better understanding of overall sleep quality in endometriosis patients may lead to improved patient care and treatments, helping to reduce pain, improve individuals’ quality of life, and decrease the overall burden of endometriosis on society.

This, to the best of the authors’ knowledge, represents the first quantitative evaluation of the association between endometriosis and poor sleep quality across populations through a meta-analysis. The 2.44 point increase in the average PSQI score, which ranges from 0-21 with good sleep quality scoring between 0 and 5 of endometriosis patients compared to individuals without the condition indicates a potential association between endometriosis and reduced sleep quality. This measure of association further supports and elucidates the association examined by the current literature. Sumbodo et al. 2024 found evidence of an association between endometriosis and sleep disturbances in their narrative synthesis. Meanwhile, two studies have found that surgical interventions significantly improved the sleep quality of endometriosis patients compared to before surgery. 53 , 54 In several survey-based studies, endometriosis patients have included low sleep quality as a personal burden experienced in connection to their symptoms. 14 , 55 , 56 Other studies have examined how endometriosis symptoms, including poor sleep quality, have impacted daily activities such as the ability to work. 57 , 58 This body of research indicates that endometriosis is likely associated with poor sleep quality, highlighting and documenting a burden of disease that should be further explored as a clinical symptom and target in treatment plans. The mechanisms by which this association may occur are unclear. However, research into other chronic pain conditions, such as fibromyalgia, arthritis, and back pain, suggests that chronic pain activates the hypothalamic-pituitary-adrenal (HPA) axis, increases sleep disturbances and inflammation in the body, and creates a feedback loop between pain and lowered sleep quality. 59 , 60 It is reasonable to assume a similar effect occurs in endometriosis patients, potentially further amplified by the inflammatory nature of endometriosis. Furthermore, research has begun to show that endometriosis’s immune and hormonal dysregulations may be directly interrupting sleep cycles. Pro-inflammatory cytokines, which play an important role in the progression of endometriosis, have been identified as a potential link between chronic pain and sleep disturbances. 61 Fluctuations in female-reproductive hormones that influence sleep and circadian rhythms may be negatively amplified by the estrogen-dependent imbalances of endometriosis. 1 , 3 , 62 Anecdotally, endometriosis patients report higher sleep disturbances during times of their menstrual cycle when progesterone and estrogen are rapidly changing. Studies interested in this potential mechanism have begun to examine the impact of melatonin, a hormone related to both sleep and the female reproductive system, has on endometriosis patients, with a phase II double-blind clinical trial finding both an improvement in sleep quality and a reduction in pain in patients treated with melatonin. 63 – 65 Additionally, comorbidities, particularly depression and anxiety, which are both highly prevalent in people with endometriosis, may modify the relationship between endometriosis and overall sleep quality. These conditions have been shown to independently worsen sleep and could confound or mediate the associations observed. Finally, a bidirectional relationship is plausible: poor sleep can amplify pain sensitivity, inflammatory signaling, and mood symptoms, which in turn may worsen endometriosis-related symptoms and further degrade sleep. This ‘feedback loop’ has been observed in other chronic pain conditions and highlights the need for longitudinal studies and mediation analyses to disentangle confounding, mediation, and directional causality.

Several limitations to this study should be considered when evaluating the results. The potential for selection bias was high among the included studies. Most studies were of a cross-sectional nature limiting the ability to interpret causally. Many studies did not screen for endometriosis in controls, potentially including undiagnosed individuals and biasing results toward the null. Adjustments for confounding variables varied greatly between included studies. In particular, half of the included studies did not assess or control for the use of hormonal therapies despite evidence that these treatments improve sleep quality by regulating hormones in menopausal women. 59 , 60 , 66 , 67 Given that hormonal therapy is a common treatment for endometriosis symptoms and is likely to be more prevalent among cases than controls, its omission may bias the results toward the null. The small sample size of included studies and high heterogeneity between studies make the effect estimate less interpretable than initially anticipated. Additionally, despite meta-regressions which elucidating potential factors leading to this high heterogeneity small sample size of included studies limited the ability to further subgroup by these areas. There is also a high potential for publication bias in the literature, which we were unable to evaluate based on sample size and power. Future research on endometriosis and overall sleep quality should aim to strengthen the ability to evaluate the causal relationship between endometriosis and poor sleep quality. Particularly longitudinal studies that properly adjust for variations in individual symptom burdens, potential confounders, including the use of hormonal therapy, comorbidities know to impact overall sleep quality including depression and anxiety, and the ability to assess the potential for a bi-directional relationship between overall sleep quality and endometriosis. In addition, controls should be properly screened for endometriosis or other chronic pain comorbidities to reduce non-differential misclassification. Despite these limitations, the study has numerous strengths that should be noted. The systematic and comprehensive nature of the review and analysis allows for examining associations across multiple populations and increases the generalizability of findings. The sensitivity analysis and meta-regressions indicate the effect estimate is potentially robust despite limitations related to sample size and heterogeneity. The included studies enrolled endometriosis patients who were diagnosed months or years before their sleep evaluations, minimizing concerns that poor sleep quality could exacerbate endometriosis symptoms through reverse causality. Lastly, this study has quantitatively evaluated an association that has only been summarized qualitatively and supports patterns emerging in the literature.

Supplemental material for Is endometriosis associated with poor sleep quality? A meta-analysis by Jillian Paul and Kim G. Harley in Women’s Health.