Effects of Clinical Characteristics on Sleep Quality in Patients with Chronic Temporomandibular Disorders

Objectives: This study aimed to investigate and compare sleep quality between patients with chronic temporomandibular disorder and healthy controls, and to analyze the association of sleep quality with disease characteristics, obstructive sleep apnea risk factors, and excessive daytime sleepiness. Methods: Chronic temporomandibular disorder patients (n=503) and 180 age- and sex-matched healthy controls were included, who completed well-organized clinical report and answered questions on sleep quality (Pittsburgh Sleep Quality Index), sleep apnea risk factors (STOP-Bang questionnaire), and excessive daytime sleepiness (Epworth sleepiness scale). Results: Mean global Pittsburgh Sleep Quality Index scores were signicantly higher in the patients (6.25±2.77) than in healthy controls (6.25±2.77) (p<0.001). Poor sleep was signicantly more prevalent in the patient group (56.9%) than in healthy controls (22.2%) (p<0.001). Compared with healthy controls, chronic temporomandibular disorder patients had a higher likelihood of obstructive sleep apnea. (STOP-Bang total score ≥ 3; 7.2% vs. 16.1%; p<0.01) and higher excessive daytime sleepiness (Epworth sleepiness scale score ≥ 10; 12.8% vs. 19.7%; p<0.05). Age (odds ratio=2.551; p<0.001), female sex (odds ratio=1.885; p=0.007), total Epworth sleepiness scale score (odds ratio=1.839; p=0.014), and headache attributed to temporomandibular disorder (odds ratio=1.519; p=0.049) were the most powerful predictors of poor sleep (global Pittsburgh Sleep Quality Index score ≥ 5) in chronic temporomandibular disorder patients. Conclusion: including


Introduction
Sleep maintains homeostasis and optimizes various functions across multiple physiologic systems.
Good sleep quality is a key factor for good physical health, emotional well-being, brain functioning, daytime performance, and pain control. Humans require both sleep and pain for good health and survival.
However, 'chronic' pain can lower and deteriorate the quality of life 1-3 . Patients with chronic pain have poorer sleep than healthy controls in terms of sleep latency, sleep e ciency, and awakenings after sleep onset. Reciprocal, bidirectional interactions exist between chronic pain and sleep disorders, deterioration in either of them can ultimately become comorbid conditions 2 . In clinical settings, sleep problems have been found to impact 88% of patients with chronic pain 4 . Contrarily, more than 40% of patients who have sleep-related problems report chronic pain 5 . The prevalence of chronic pain ranges from 10-40% 6 , similar to that of sleep disorders, ranging from 10-36% 7 . Temporomandibular disorders (TMDs) are quite common chronic orofacial pain conditions. TMDs are highly prevalent, affecting up to 25% of the population, with a peak incidence at 20-40 years of age, and 1.5-2 times more prevalent in female than in male 8 . TMD may promote widespread idiopathic pain, and the etiology of chronic TMD is considered multifactorial. Pain is the most common symptom of TMDs, which can affect areas such as the ears, eyes and/or throat, frequently causing neck pain and headache, and involve musculoskeletal pain, disturbances in mandibular movement, and functional impairment 9 .
As the international diagnostic and classi cation systems for TMD, the research diagnostic criteria for TMD (RDC / TMD) proposed in 1992 is the most popular, and it has been updated to the diagnostic criteria for TMD (DC / TMD) in 2014, and has been subdivided into joint-and muscle-derived TMD pain as well as headache attributed to TMD 10,11 . Sleep quality in patients with TMD decreases as the number of diagnoses of painful TMD increases based on the International RDC / TMD Axis I 12 . Although several studies have been conducted on sleep quality in TMD patients [13][14][15] , few researchers use international RDC / TMD and DC / TMD. Deterioration of sleep quality and impairment of sleep structure occur in a signi cant proportion in TMD patients and are thought to be a risk factor for maintaining and worsening symptoms, but their impact is not clearly known due to several methodological limitations.
Chronic TMD can present persistent, recurrent, or chronic pain associated with TMJ and/or muscles involved in the masticatory system, which leads to highly disabling. Chronic TMD is also typically related to joint dysfunctions, such as disc displacement with or without reduction 16 . Approximately 90% of TMD patients reported poor sleep quality 15 . Although the etiology underlying the connection of chronic pain and poor sleep remains largely unknown, poor sleep might contribute to TMDs aggravation. In addition, TMD patient ratings of poor sleep are associated with increased clinical pain severity and psychological distress 15 . Chronic TMD is also accompanied by pain ampli cation characterized by hyperalgesia, allodynia, hyperalgesia, and upregulation of the serum tonicity pathway, and sleep problems and psychological disorders are common comorbidities 17 . Sleep bruxism is considered an aggravating factor for TMD pain, but the relationship remains controversial 18,19 . However, there is little evidence to indicate that chronic TMD patients have poor sleep quality, and it is not known which factors clearly affect the sleep quality of chronic TMD patients.
The Pittsburgh Sleep Quality Index (PSQI) is a valid, reliable, and internationally known instrument for assessing self-perceived sleep quality 20 . Compared to objective measurements, such as polysomnography or oximetry, PSQI addresses a longer time frame and is suitable for large scale research. In addition, a comparison of validity with respiratory indices from polysomnographic recordings has been made for PSQI with some degree of success. The primary concern for many patients presenting with lowered sleep quality is excessive daytime sleepiness (EDS). There may not be an obvious cause for EDS, lowered sleep quality and insu cient sleep duration are responsible for this symptom 21 . EDS is a common symptom in sleep disorders, chronic pain, headaches, and TMD 22,23 . The EDS status can be easily evaluated with the Epworth sleepiness scale (ESS), and 28.6% of TMD patients presented with EDS based on this questionnaire 23 . Obstructive sleep apnea (OSA) is a sleep breathing disorder characterized by repeated agitation of nocturnal breathing disruption caused by upper airway collapse, and is considered as a putative risk factor for chronic TMD 24 . For screening for OSA, the snoring, tiredness, observed apnea, high blood pressure (STOP)-BMI, age, neck circumference, and male gender (Bang) questionnaire with high reliability has been used in the clinical eld 24,25 . Therefore, it may be necessary to evaluate sleep-related factors using PSQI, STOP-Bang, and ESS simultaneously in chronic TMD patients.
This cohort study aimed to investigate the clinical characteristics and evaluate sleep-related factors to correlate these parameters with poor sleep quality in chronic TMD patients. We described sleep quality differences between chronic TMD patients and healthy controls, investigated subgroup of TMD in uences, and studied sleep quality in the full set of diagnostic DC/TMD and RDC/TMD subgroups of Axis I. To complementarily evaluate the factors related to EDS and OSA along with sleep quality obtained from PSQI, ESS and STOP-Bang questionnaire were additionally used. Chronic TMD has an idiopathic basis in which the pathophysiology mechanism is not well understood. Our ndings in this study will support the usefulness of an integrated model of demographics and disease characteristics in explaining sleep quality deterioration in chronic TMD patients.

Participants
To investigate the research purpose, the authors designed and implemented a retrospective cohort study conducted at the Department of Orofacial Pain and Oral Medicine at Kyung Hee University Dental Hospital of Seoul. The research protocol was reviewed in compliance with the Helsinki Declaration and approved by the Institutional Review Board of the Kyung Hee University Dental Hospital (KHD IRB no. 1804-2). Written informed consent was obtained from all individual participants.
Of the patients who had visited the Department of Orofacial Pain and Oral Medicine between June 1, 2018, and November 30, 2019, the study sample was composed of chronic TMD patients according to inclusion and exclusion criteria. Inclusion criteria of chronic TMD patients were as follows: completed a set of routine TMJ assessments, as well as constructive questionnaires, no treatment of the current episode other than medication, and no history of TMD prior to the present symptoms. Patients who were pregnant or had a history of systemic osteoarthritis, rheumatoid arthritis, other connective tissue diseases, general infection, neurological/neuropathic diseases, and those under 18 years of age were excluded from the study. Healthy controls were recruited through advertisements in the hospital. Inclusion criteria of healthy controls were as follows: systemic good health, no history of macrotrauma or surgery on TMJ and neck area, no previous diagnosis of TMD, no TMD signs and symptoms, and an absence of TMD on brief clinical examination. Healthy individuals who satis es the following inclusion criteria and exclusion, and who are sex-and age-matched with chronic TMD patients were nally selected. All individuals included in the study completed three questionnaires including PSQI, ESS, and STOP-Bang.

TMD Classi cation and clinical Evaluation
Clinical evaluation procedures included an oral examination, interview, panoramic radiography, and a comprehensive questionnaire in RDC/TMD criteria 26 and DC/TMD Axis I diagnostic algorithms 27 for TMD diagnoses. A subtype of diagnostics included myofascial pain, disc displacement, arthralgia using RDC/TMD, and headache attributed to TMD based on DC/TMD Axis I diagnostic algorithms.
We diagnosed patients who experienced TMJ pain for more than six months after onset as chronic TMD.
When the pain persisted more than three to six months, individuals were usually considered in a chronic state 28 . The intensity of TMD pain was measured using a visual analog scale (VAS) (0-10, 10 being the worst possible pain), and symptom duration in the masticatory muscles, TMJ, and adjacent structures were reported in days. Information concerning patient demographics included age, sex, height, body weight for body mass index (BMI = weight/height 2 ), and neck circumference were collected by a research assistant.

Sleep Quality Evaluation
Using PSQI PSQI was employed to measure sleep quality. Habitual sleep quality and sleep disturbance in the past month were assessed using the 19-item PSQI, a well-validated self-report questionnaire. The PSQI has seven components that concern subjective sleep quality, sleep latency, sleep duration, sleep e ciency, sleep disturbances, use of sleep medication, and daytime dysfunction. Each subscale is weighted equally, scored from 0 (good sleep/no problems) to 3 (poor sleep/severe problems), summing to a global PSQI score (range, 0-21). Higher scores denote worse sleep quality, and a global score > 5 has a diagnostic value in distinguishing poor from good sleep 29 . In this study, the authors investigated sleep quality using PSQI in multidimensional construction.

The risk evaluation of obstructive sleep apnea (OSA) with STOP-Bang
The STOP-Bang questionnaire is validated screening tool for identifying high likelihood for OSA. The STOP-Bang questionnaire includes eight dichotomous (yes/no) questions related to these clinical features of sleep apnea. For each question, answering "yes" scores 1, a "no" response scores 0, and the total score ranges from 0-8. The exposure of interest was a binary-low or high likelihood for OSA; the low likelihood of OSA: Yes to < 3 questions, high likelihood of OSA: Yes to ≥ 3 questions 30 . All patients were asked to complete the STOP-Bang questionnaire.

Excessive daytime sleepiness measured by ESS
The ESS is a validated clinical tool for the evaluation of excessive daytime sleepiness (EDS) 31 . Unlike other scales, which measure sleepiness at a single time point, the ESS is designed to evaluate the general level of sleepiness. The ESS is an eight-item, self-administered questionnaire designed to provide a measure of the subject's propensity to fall asleep in a variety of situations. The subject is instructed to answer how likely it is that he/she would fall asleep in those different situations, by giving a score on a 4point scale (0-3). Thus, the total score (the sum of scores of the eight items) of the ESS ranges from 0-24. The higher the score, the greater the possibility the individual will fall asleep during the daytime. The ESS total scores were dichotomized into scores ≤ 10 and > 10; the latter is considered to be clinically signi cant EDS 32 . We used a score of 10 or higher on the ESS to measure excessive sleepiness.

Statistical Analysis
The data were analyzed using SPSS Statistics for Windows, Version 20.0 (IBM Corp., Armonk, NY, USA). Continuous variables are presented as means and standard deviations (SD), and categorical variables are presented as frequencies and percentages. Differences between groups were examined by using the chisquare test for categorical variables and t-test and one-way analysis of variance (ANOVA) with Tukey post-hoc test for numeric variables. Three questionnaires included in this study were for sleep quality, risk for OSA, and daytime sleepiness. To understand the risk for poor sleep quality, we performed a multiple logistic regression analysis to determine the relative risk for poor sleep (PSQI global score of ≥ 5). In addition, dichotomous variables based on cut-off points of each questionnaire's global scores were used as dependent or independent variables of logistic regression analysis. The high likelihood of OSA (STOP-Bang total score ≥ 3), and EDS (ESS score of > 10) (independent variables), as well as age, sex, and symptom duration, were taken into consideration simultaneously to predict a value of a dependent variable (poor sleep) for chronic TMD patients. For all analyses, a two-tailed level of statistical signi cance of a p-value was set at less than 0.05.

General description
In this period, 525 patients who were diagnosed with chronic TMD were included, and 22 patients were excluded because they lacked medical documentations. Finally, 503 patients (mean age: 33.10 ± 13.26 years, 333 females) were designated as the chronic TMD group. One hundred and eighty age-(mean age: 32.77 ± 12.95 years, 116 females) and sex-matched TMD-free volunteers were designated to the healthy control group. Table 1 presents the demographic characteristics of chronic TMD patients compared to healthy controls.
Chronic TMD was more prevalent in females (66.2%) than in males (43.8%), and the female:male was 1.51:1. BMI scores were signi cantly higher in chronic TMD than in healthy controls (22.32 ± 3.57 vs. 21.45 ± 2.76, p = 0.001), and the mean values were in the normal range. In chronic TMD patients, the mean VAS score of chronic TMD patients was 4.89 ± 2.45, and the mean symptom duration was 589.74 ± 1315.01 days. The results were obtained via t-test, b : The mean difference among age subgroups was obtained by one-way analysis of variance and post-hoc analysis. c : The results were obtained from χ2 test. A pvalue < 0.05 was considered signi cant. **: p-value < 0.01. ***: p-value < 0.001. TMD: temporomandibular disorder, BMI: body mass index, SD: standard deviation, n.a.: not available.    : The p-value for the test of the mean difference of the PSQI global score between subgroups. ¶ : The p-value for the comparison results of the number (%) between subgroups via χ2 test and Boneferroni correction. The mean difference among age subgroups was obtained by one-way analysis of variance and post-hoc analysis. † : The PSQI global scores of subgroups over 20 years of age were signi cantly higher than that of 10s, and the scores of 20s, 30s, and over 40 were not signi cantly different. ‡ : The PSQI global scores over 30 years of age were signi cantly higher than that of 20s, and the scores of 30s and over 40 were not signi cantly different. p-Value signi cance was set at < 0.05. *: p-value < 0.05, **: p-value < 0.01, ***: p-value < 0.001. TMD: temporomandibular disorder, PSQI: Pittsburgh Sleep Quality Index, SD: standard deviation.

Differences in PSQI Global Scores and Poor sleep Between TMD group and Controls
As for age, an increase in age was associated with an increase in PSQI global scores and the occurrence of poor sleep. When we divided TMD patients into subgroups based on their age, the PSQI global scores were signi cantly higher and the ratio of patients with poor sleep quality was higher in patients in their 30s, 40s, or older as compared with those in their 20s and younger. The PSQI global score of chronic In addition, multiple TMD sub-diagnoses were allowed in one patient, and the PSQI global score was higher when patients were diagnosed with myofascial pain (6.31 ± 2.82 vs. 5.50 ± 1.98, p = 0.023) or headache attributed to TMD (6.58 ± 2.85 vs. 5.82 ± 2.60, p = 0.002) than in patients without the diagnoses. The majority of chronic TMD patients simultaneously suffer from pain of myofascial origin (n = 465, 92.4%), arthralgia (n = 355, 70.6%), headache attributed to TMD (n = 285, 56.7%), and disc displacement (n = 278, 55.3%). Therefore, we concluded that it is not appropriate to compare the PSQI global and component scores and presence of poor sleep by subgroup.

Differences in STOP-Bang and ESS Between TMD group and Controls
STOP-Bang total scores were signi cantly higher in TMD patients than in healthy controls (1.77 ± 1.86 vs. 1.53 ± 0.77, p = 0.017) (   (Fig. 1). STOP-Bang total score was not signi cantly different according to the number of sub-diagnoses. However, in ESS, the ESS total score was signi cantly higher when chronic TMD patients had four multiple diagnoses (7.40 ± 3.79) than when they had only one subdiagnosis (5.73 ± 3.83) (p < 0.001) ( Table 5).

Discussion
This study comprehensively investigated the clinical characteristics and sleep quality of chronic TMD patients who were diagnosed based on RDC/TMD and DC/TMD Axis I. The main ndings in the present study imply that chronic state TMD patients had poorer sleep than healthy controls, and the magnitude of impaired sleep was associated with increased age, female sex, certain subtypes of TMD diagnosis, including myofascial pain and headache attributed to TMD, and the number of TMD diagnoses in a person. Regarding summary scores and cut-off values of each questionnaire, the presence of EDS was a signi cant predictor for the poor sleep quality in chronic TMD patients.
The cause of chronic TMD is varied, and its localization, and clinical characteristics are vaguer than with acute pain. Chronic TMD is typically associated with joint dysfunctions such as disc displacement with or without reduction 16 , and psychological distress 17 . It is di cult to infer a causal relation between sleep and chronic pain; patients with chronic pain commonly suffer from poor sleep quality 33   Female sex was also a major contributor to poor sleep quality in chronic TMD patients. There is limited recent evidence of interactions among sex, TMD chronicity, and sleep. However, it has been found that females show higher clinical and experimental pain sensitivity, and worse sleep impairments than males 2 . Few probable causes for poor sleep quality in the female sex may be explained based on sex differences concerning mechanisms of pain of the craniofacial system 47 . Furthermore, contribution of female sex may re ect changes of systems beyond the physical axis of the orofacial area and in line with the biopsychosocial model, blending centrally mediated factors. Especially in postmenopausal female, an increase in sleep problems may be associated with the presence of noticeable hormonal changes, ageassociated changes in sleep and psychosocial distress 2 . In clinical research, females reported TMD symptoms, headache, and had muscle tenderness and joint sounds more often than males 48 . It will be crucial to determine whether the effect of sleep on chronic TMD pain, and vice versa, is moderated by key demographic variables, such as age or sex.
Considering the EDS, EDS was a signi cant predictor for poor sleep in chronic TMD patients. EDS was more prevalent in chronic TMD patients than in healthy controls (19.7% vs. 12.8%, p < 0.05), and its OR value for poor sleep quality was 1.069. The EDS prevalence in the present study was higher than the prevalence among the general population (12-16%) 49 , and lower than 28.57% of TMD patients, the rate previously reported 23 . The discrepancy may have occurred due to differences in age distribution, race, and method of study. A signi cant proportion of the general population, as well as in patients with sleep problems or chronic pain, may suffer from EDS for a variety of causes, not pathological mechanisms. In older adults, they are prone to have daytime napping and EDS and the presence of comorbid conditions such as chronic pain, sleep disorders, and frequent nighttime urination breaks 50 . Diminished melatonin secretion and a reduced circadian modulation of rapid-eye-movement sleep and less pronounced daynight differences in the lower alpha activity occurs in the older group 51 . Furthermore, females are more likely than males to have more trouble sleeping at night and experience EDS 52 . Thus, EDS in chronic TMD may have different underlying mechanisms of a homeostatic drive for sleep and pain control systems according to age and sex. Hence, it is necessary to consider age-and sex-related differences in chronic TMD patients to obtain accurate results translation.
The high likelihood of OSA was not the signi cant predictor for poor sleep quality in chronic TMD patients. The STOP-Bang questionnaire is a great tool to easily identify patients with suspected OSA 25 . However, traditionally, polysomnography (PSG) in a sleep laboratory has been considered as a standard tool for the diagnosis of OSA 53 . Once OSA is identi ed, these patients need a de nite diagnosis using the PSG. There is a limit to the extended interpretation of the results of the sleep quality and high risk of OSA in patients with chronic TMD, as the relationship between OSA and TMD. Of course, a bidirectional association between OSA and TMD have been suggested 24,54 . Although the clear mechanism has not been identi ed, oral appliances used in the treatment of OSA may be the cause of TMD with continuous mandibular advancement while sleeping 55 . Sleep bruxism, a major risk factor for TMD, may be linked to OSA through the sleep-related arousal reactions 56 . The link between sleep bruxism and OSA has been studied at various ages [57][58][59] , which could potentially suggest a link between TMD and OSA. Sleep bruxism and PSG were not covered in this study, but will be fully addressed in subsequent studies.
Headache attributed to TMD was associated with an increase of PSQI global scores, and a signi cant predictor for poor sleep. As headaches are a common symptom of accompanying TMD, few researches have been done on the nature of headache attributed to chronic TMD. TMD patients with headache reported signi cantly higher levels of pain and mandibular dysfunction than patients with only TMD 60 .
Headache can promote sleep disturbances, and sleep disturbances can also precede or trigger a headache attack 61 . Moreover, sleep deterioration has been associated with an increased risk for headaches, and in individuals with chronic headaches, shorter sleep duration has been associated with more severe pain 62 . Clearly, the underlying pathophysiology contributing to the close association and complex relationship among headache attributed to TMD, headache disorders, and sleep disorders are not fully explained. There may be complex bidirectional relationships, and can be explained by peripheral and central sensitization, malfunctions of neuroendocrine, immune, and vascular system, and even gene polymorphism.
Myofascial pain was also associated with poor sleep quality in chronic TMD patients. Similar to the present study, the substantial in uence of myofascial pain on poor sleep quality in patients with TMD was documented 63 . Furthermore, according to the TMD diagnostic subgroups, and the impact on quality of sleep and the symptom severity can vary. TMD patients with muscle-derived or myofascial pain exhibit more advanced stages of depression and somatization than patients diagnosed with TMJ disc displacement, a joint-derived problem 64 . Other researchers also observed a higher impact in patients with myogenous complaints than those with disc disorders 65,66 . While joint pain is characterized by a wellde ned in ammatory process, chronic muscle pain presents with enigmatic pathophysiologic mechanisms, of which central sensitization is the common factor unifying these conditions 67 . In addition, females have more pain and widespread pain in more body areas than males, which may be related to their worse quality of sleep 68 . Therefore, depending on the subgroup of TMD, the mechanisms by which TMD signs and symptoms occur are different, and further investigation is needed on the effects on sleep quality.
Finally, the quality of sleep was lower in chronic TMD patients with multiple diagnoses than in patients with a single diagnosis. According to Gil-Martínez et al., patients with mixed pain, having arthrogenous and myogenous origin simultaneously, showed greater craniomandibular and neck disability than patients diagnosed with chronic joint pain or muscle pain only 69 . Patients with headache and TMDs reported signi cantly higher levels of pain and disability compared to patients with only TMDs 60 . These ndings can be interpreted as increasing the number of TMD subgroup diagnoses can increase the severity of TMD symptoms. Overall, chronic TMD symptoms and multiple diagnoses may have a bidirectional cross-correlation, which can impair sleep quality. Chronic TMD, especially myofascial pain, headaches attributed to TMD, and sleep disturbance factors, may share the mechanism of occurrence and exacerbation.
Limitations of this study include the case-control study design, which cannot address a causal direction of effects and suggests only associations/correlations between the variables. A study design with repeated PSG may be more powerful in detecting phase-related differences. Instead of using PSG, we used self-assessment measures of sleep due to feasibility and convenience, especially because of large sample size. PSG is an objective measure of biophysiologic sleep parameters, so we are planning further studies to expand these ndings with PSG. In addition, to get a deeper understanding of the relation between chronic TMD and poor sleep quality, we need to investigate their biopsychosocial aspects; however, this study did not evaluate psychological distress in patients with chronic TMD. Further studies on the psychological aspects of chronic TMD patients are ongoing.

Conclusions
The strength of this study is a comprehensive analysis of how various issues affect poor sleep quality among many factors. Chronic TMD patients suffered more from impaired sleep than healthy control subjects, and poor sleep was associated with multiple comorbid symptoms. Thus, assessing sleep quality should be a routine part of the diagnostic work-up of chronic TMD patients. Furthermore, a multidisciplinary management approach is needed to address all the factors in addition to sleep that modulate pain experience. In the diagnosis and treatment of chronic TMD, a fragmented eld of view is not suitable, and a multidisciplinary approach involving experts in neurology, endocrinology, gerontology, and psychology, in addition to orofacial pain experts, is required. The results from this study will help to establish strategies for individual treatment and management of chronic TMD patients. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Declarations
Data Availability Statement: Since these are patient data, if there is a request for data disclosure, KHU-IRB will discuss it before disclosure.