The association between mental health, chronic disease and sleep duration in Koreans: a cross-sectional study

Background Sleep duration holds considerable importance as an indicator of mental/physical health. The objective of this study was to investigate the association between sleep duration, mental health, and chronic disease prevalence in Koreans. Methods Of 31,596 subjects eligible for the Korean National Health and Nutrition Examination Survey V (2010–2012), 17,638 participants who answered items on sleep duration (aged ≥19 yrs) were analyzed in a cross-sectional study. Association between sleep duration, mental health, and chronic disease prevalence was assessed using logistic regression, and adjusted for various socioeconomic and lifestyle characteristics. Results Short or long sleep duration showed correlations with mental health, and items of significance showed gender-specific patterns. Women displayed significant associations with stress and depressive symptoms, and men with stress, thoughts of suicide, and psychiatric counseling. While stress was related with short sleep duration in both genders, depressive symptoms showed a relationship with long duration in men, and short duration in women. Prevalence of any chronic disease was associated with ≤6 h sleep when adjusted for factors including mental health, and among chronic diseases, cancer and osteoarthritis showed associations with short sleep duration, while diabetes and dyslipidemia were associated with normal sleep duration. Conclusions Mental health problems were associated with sleep duration with gender-specific patterns. Associations with osteoarthritis, cancer, diabetes, dyslipidemia and abnormal sleep duration persisted after adjustment for mental health. Electronic supplementary material The online version of this article (doi:10.1186/s12889-015-2542-3) contains supplementary material, which is available to authorized users.


Background
Sleep is a daily process of restoration and recovery. Lack of sleep and sleep-related disorders are related with mental and physiological conditions, and may modulate the endocrine and immune systems and metabolism, consequently affecting quality of life, and emotional state. They may also increase mortality and risk of disease [1][2][3]. Despite individual differences, 7-8 h of sleep per day is generally considered appropriate to sustain physiological and mental health [4,5].
Prior studies report that sleep duration is associated with various mental health aspects such as stress, depression, thoughts of suicide, mental disorders, and with chronic disease [6][7][8][9][10][11][12]. However, few studies have assessed the influence of different subcategories of mental health in men and women. This study investigated gender-specific patterns in associations between sleep duration and mental health.
Most previous studies failed to consider mental health in assessing the relationship between sleep duration and chronic disease, and it is uncertain whether sleep duration is affected by mental health factors consequent of chronic disease or by pathological expression of chronic disease itself. Despite results from a Brazilian population-based cross-sectional study on chronic disease and sleep that show depression and anxiety are related with sleep duration, it is limited in that it did not take into account whether depression and anxiety affected chronic diseases in the association between chronic disease and sleep duration [13]. A large-scale cohort by Ayas et al. reported that short and long sleep duration were both related to cardiovascular disease, but only depression was included as a covariate [14].
This study investigated whether sleep duration, mental health disorder and chronic disease prevalence were associated, and whether sleep duration displays gender-specific patterns in these associations using cross-sectional data from the Korean National Health and Nutrition Examination Survey (KNHANES).

Study population and sampling
This study was conducted using data obtained from the KNHANES V, which employed a rolling sampling method involving stratified, multistage, probability-clusters to yield a representative sample of South Koreans. The survey was administered by the Korean Ministry of Health and Welfare and was made up of 3 parts: health survey, nutrition survey, and health examination. Further details are listed in "The 5 th KNHANES Sample Design" and reports, made accessible on the KNHANES website https://knhanes.cdc.go.kr/ knhanes/index.do. The KNHANES annual reports, user manuals and instructions, and raw data are available on request. The KNHANES V health examination and survey were completed by 24,173 participants (76.5 % of the target population, n = 31,596). Analysis was performed in 17,638 subjects aged ≥19 yrs who answered to survey items on sleep duration, and subjects with missing values in mental health, chronic disease, and covariates were excluded from each corresponding analysis.

Sleep duration
Sleep duration was categorized as average ≤6 h, 7-8 h, and ≥9 h of sleep per day as classified in the health survey.

Mental health
Stress perception was classified into high (very high or high degree of stress), low (low degree of stress), and almost none (almost no stress). Assessment of depressive symptom morbidity, psychiatric counseling, and thoughts of suicide was binary in the health survey: those with and those without experience of continuous depressive symptoms for ≥2 weeks; those who had and those who had not sought counseling for psychiatric issues during the past year; and those with and those without thoughts of suicide over the past year, respectively. Those who had contemplated suicide were further divided into those who had attempted suicide and those who had not.

Statistical analysis
We employed a complex sampling design analysis approach using stratified, cluster, and weighted variables, and all analyses were performed using statistical package SAS ver. 9.3 (SAS Institute Inc, Cary, NC, USA) with p < 0.05 considered to be statistically significant. Continuous variables were expressed as mean and standard deviation, and categorical variables as frequency count and percentage (%), and differences in sleep duration by characteristics were calculated using Rao-Scott chi-square test or ANOVA. The impact of mental health factors on sleep duration was assessed with a complex sampling design multinomial logistic regression analysis adjusting for covariates to obtain odds ratios (OR) and 95 % confidence intervals (CI).

Ethics, consent and permissions
The interviewer was not given information about a participant before conducting the interview, and all participants gave written informed consent to participate. The protocol was approved by the Institutional Review Board of Jaseng Hospital of Korean Medicine in Seoul, Korea.

Results
The In assessing the association between mental health and sleep duration in KNHANES participants, age, sex, socioeconomic (income, education, occupation), lifestyle-related (marital status, drinking, smoking), and health-related factors (subjective health, current chronic disease morbidity, BMI, use of medical care, level of daily activity, bed rest for the past month, absence from work for the past month) were included as covariates. All factors except smoking were shown to be associated with sleep duration ( Table 1).
The results showed that mental health was related to sleep duration. Univariate analysis showed that subjects who reported mental health issues such as stress, depressive symptoms, thoughts of suicide, and mental health counseling were more likely to sleep ≤6 h or ≥9 h than the recommended 7-8 h. Related mental health items showed gender-specific patterns in multivariate analysis (fully adjusted model

Discussion
The results show that stress, experience of depressive symptoms, thoughts of suicide, and psychiatric counseling were all associated with short or long sleep duration. Mental health factors and sleep duration associations revealed gender-specific patterns, with thoughts of suicide and psychiatric counseling more related in men, and stress and depressive symptoms in women. It has been generally acknowledged that one of the reasons women are more susceptible to depression than men is the direct influence of follicular hormones [15,16]. Other explanations include that the HPA axis which regulates stress is more dysfunctional in women [17], and that follicular hormones affect HPA regulation [18]. Most women experience premenstrual symptoms over their lifetime, and 1/5 report serious symptoms including depression [19]. Also, erratic cyclic variation in estrogen and progestogen levels occurs in perimenopausal and postmenopausal periods, rendering menopausal women susceptible to major depression, and this tendency is more pronounced in women with a history of depressive episodes [20]. Women also differ in response and adaptation to stress, and adolescent girls have been shown to focus more internally on stressful emotions and mental distress [21]. Meanwhile, men tend to be more impulsive and prone to high-risk behavior than women [22]. Additionally, men are more prone to antisocial personality disorders, behavioral disorders, ADHD, and intermittent explosive disorders [23][24][25][26][27].
Associations were found between sleeping ≤6 h and higher levels of stress in women, and in ≥9 h sleep and lower stress levels. Although the causal relationship regarding short sleep and stress has not been established, recent studies point toward a bi-directional association [6,28,29].
Magee et al. purported that short sleep duration could induce more fatigue or mood and cognition disturbances, resulting in poor self-rated health (SRH) or quality of life (QOL) [29]. Subjects with poorer health notably had more sleep disturbance and short sleep duration, and Kessler et al. reported that significantly more subjects sleeping 7-8 h a day had low scores (<16) in the Kessler Psychological Distress Scale than those with short or long sleep duration [26].
Thoughts of suicide were associated with both short and long sleep duration in this study, whereas suicidal intentions and increased risk of suicide attempt have been shown to be related with short sleep in previous literature [7][8][9][10], and a 2001 review concluded that sleep disturbance was more frequent in psychiatric patients displaying suicidal behavior [30].
Although the underlying mechanism linking short sleep duration and suicide attempts is not yet known, insufficient sleep is known to have negative effects on cognitive function, resulting in impaired judgment and increased compulsion, exhaustion, and despair [8,31], placing patients under greater risk of impulsive behavior [11].
This study found that while women sleeping ≤6 h experienced more depressive symptoms than those sleeping 7-8 h, men did not show association between depressive symptoms and sleep duration., Gangwisch et al. purported that depression was associated with both long and short sleep duration [32]. Martin-Merino et al. suggested that sleep-related problems were a strong indicator of depression, and deprivation of sleep may alter mood and individual wellbeing [33]. Patel et al. reported that longer sleep was associated with higher risk of depression, use of antidepressants and benzodiazepine [34], which may be due, in part, to the sedative effect of psychotropic medication [35].
Psychiatric counseling showed significant correlation with short or long sleep duration in men. Men typically show weaker restraint over impulse, and higher prevalence of psychological disorder presenting overt or impulsive behavior [23][24][25][26][27], which may be why psychiatric counseling was also more significant in men. However, the total prevalence of psychiatric counseling was much lower than that of depressive symptoms, thoughts of suicide, and stress, indicating that many patients with mental health problems were not receiving appropriate counseling.
Many studies have addressed the relationship between chronic disease and mental health: it has been reported   [36], and type II diabetes mellitus patients have a twofold higher risk of depression than non-diabetics [37]. Also, chronic disease has been shown to be a major factor for low HRQoL through EQ-5D studies in chronic disease patients [38]. While some report an inverse association between chronic disease and mental health, some state that mental health remains stable and unaffected by chronic disease [39]. Some previous studies have reported relationships between stress-prone personalities and cardiovascular risk factors [40,41]. Stress is associated with total cholesterol, low-density lipoprotein and triglycerides [42], and modern stress-ridden lifestyles may seriously impact metabolism and may bring on pathological changes [43]. Prolonged changes in lipid metabolism from chronic stress may lead to cardiovascular disease [44].
We adjusted for mental health to adjust for the potential relationship between chronic disease and mental health. Participants with any chronic disease, and participants with osteoarthritis, diabetes, dyslipidemia, or cancer out of the various chronic diseases were shown to have significant relations with sleep duration, and osteoarthritis and cancer patients showed higher prevalence in short sleep duration. Diabetes was associated with normal sleep (compared to short sleep duration), and dyslipidia with normal sleep duration (compared to long sleep duration).
Numerous studies have been conducted on the association between chronic disease and sleep duration. A largescale cohort on 71,617 women aged 45 to 65 conducted by Ayas et al. concluded that both long and short sleep duration was associated with cardiovascular disease [14]. Given the sex and age group, it is highly probable that many, if not most, participants were menopausal, while this study included both male and female adults, and did not find a significant relationship between cardiovascular disease and sleep duration.
The results of a cross-sectional study investigating chronic disease and sleep duration in the Brazilian population are consistent with this study in reporting the association between depression and anxiety with sleep duration, but it failed to adjust for other mental health variables [13].
Dyslipidemia and diabetes were shown to be associated with normal sleep duration, indicating less association with abnormal sleep duration. This is contrary to previous studies which purport positive correlations between short sleep duration and dyslipidemia [45] and diabetes [46], while Kaneita reported in a 2008 study that subjects with short or long sleep duration had higher serum triglyceride and lower high-density lipoprotein cholesterol levels than those with appropriate sleep [47]. Other studies have put forth that longer sleep duration may increase diabetes incidence, or that longer sleep duration itself may be an early symptom of diabetes [48].
Osteoarthritis was shown to be related to short sleep duration as in a previous Brazilian study where prevalence of shorter or longer sleep duration was higher in osteoarthritic, rheumatic, osteoporotic, and arthritic subjects, and the reason was suggested to be disease-related pain and consequent disruption of sleep patterns [13].
Our study also revealed that cancer was associated with short sleep, contrary to reports from previous studies. One meta-analysis concluded that short sleep duration was unrelated to cancer, and though longer sleep duration displayed a positive relationship with colorectal cancer occurrence, it was found to reduce risk in hormoneassociated cancers [49]. The association between long sleep duration and cancer incidence may vary by course, and more cancer patients of <5 yrs duration presented with depressive symptoms compared to those of ≥5 yrs [50]. Mental health is highly relevant in cancer progress as it entails high mortality risks, and should be considered in future studies.
Some major strengths of this study are that the population is a nationally representative sample of Koreans, and the health examinations and surveys were conducted by prior trained surveyors. Also, many factors that may confound sleep duration, mental health, and chronic disease (age, sex, income, education, employment, marital state, smoking, drinking habits, subjective health, BMI, use of medical care, level of daily activity, bed rest for the past month, absence from work for the past month) were included for adjustment. We would also like to draw attention to the fact that we were able to determine genderspecific patterns in relation to sleep duration by considering various aspects of mental health, and that we gave consideration to chronic disease in assessing the relationship between mental health with sleep duration and vice versa, including mental health factors as confounding variables in associations between sleep and chronic disease.
Meanwhile, the biggest limitation of our study is due to its cross-sectional design. Though causal relationships cannot be deduced, these results may be used as basic data offering clues to how chronic disease, mental health and sleep duration is associated and underlying physiological and biochemical mechanisms. Other limitations include that while national-level surveys hold considerable power, the data entails certain limitations as questions and diagnoses have to be kept relatively simple, the validity and reliability of data could be questionable as data collection relied on self-report, and quality of sleep or history of sleep-related disorders were not investigated, thus restricting full evaluation of sleep disturbance problems. For example, only 4 items were related to mental health, and criteria for stress perception is somewhat ambiguous as it was assessed through subjective self-evaluation of high, low, or almost no stress as opposed to objective criteria or professional opinion. In addition, there was overlapping between the constructs of depressive symptoms, suicidal thoughts, psychiatric counseling, and stress. For instance, suicidal thoughts and sleep disturbance may be symptoms of depression or stress.
Still, the authors performed secondary analyses by group (only depressive symptoms, depressive symptoms with psychiatric counseling, depressive symptoms with suicidal thoughts, and depressive symptoms with psychiatric counseling and suicidal thoughts) hypothesizing that such additional analyses would reflect severity of depressive symptoms and stronger association with sleep duration. However, statistically significant difference shown in crude analysis was not maintained in adjusted analyses. This may be due to decreased sample size or as the possible pathways between such constructs as counseling and sleep disturbance are likely to be bi-directional (i.e. disturbed sleep is likely to be associated with increased chances of seeking psychiatric counseling, but psychiatric counseling should be associated with a subsequently decreased level of disturbed sleep) (see Additional file 1).
While the present study interpreted gender-specific patterns to be due to biological differences such as hormones, Multinomial logistic regression analysis a Adjusted for age, sex, household income, education, marital status, employment, BMI, smoking, alcohol consumption, level of daily activity, subjective health, bed rest for the past month, absence from work for the past month, stress, depressive symptoms, thoughts of suicide future studies may give more consideration to social gender roles and psychosocial differences between men and women in study design and interpretation of mental health and sleep pattern findings. For example, men may be more reluctant to seek psychiatric counseling and thus men seeking treatment may be more adversely affected than women. Further multidimensional studies should also be conducted with thorough consideration of construct and measures and evaluate the mediation effect of mental health in the association between chronic disease and sleep.

Conclusions
In conclusion, poor mental health was associated with short or long sleep duration, showing significance in contemplation of suicide, psychiatric counseling and stress in men, and stress and depressive symptoms in women. After adjusting for covariates including mental health factors, chronic disease morbidity, osteoarthritis, and cancer showed relationships with short or long sleep duration. Diabetes was associated with normal sleep (compared to short sleep duration), and dyslipidia with normal sleep duration (compared to long sleep duration). Further prospective studies are required to identify the pathways and causal relations by which chronic disease is associated with sleep duration.