Association between quality and duration of sleep and subjective cognitive decline: a cross-sectional study in South Korea

Sleep is being emphasized as a factor that improves mental health and quality of life. Here, we aimed to investigate the association between the quality and duration of sleep and subjective cognitive decline in the Korean population. We used the 2018 Korean Community Health Survey data that are nationwide representative data collected by the Korea Centers for Disease Control and Prevention. Sleep quality was measured using the Korean version of Pittsburgh Sleep Quality Index. The study population comprised 206,719 individuals aged 19 years and over. We used multiple logistic regression for the analysis. Individuals of both sexes with poor sleep quality were more likely to experience subjective cognitive decline compared with the reference group (good sleep quality) (men, odds ratio (OR) = 1.97 [95% confidence interval (CI) 1.86–2.09]; women, OR = 1.75 [95% CI 1.67–1.84]). U-shape associations were found between sleep duration and subjective cognitive decline. Additionally, the presence of depressive symptom or stress and health-related behaviors, including smoking, drinking, and not walking, were high-risk factors for subjective cognitive decline. Our results indicate that poor sleep quality might contribute to subjective cognitive decline in the Korean population. We suggest the implementation of intervention measures for poor sleep behaviors to prevent cognitive decline.


Materials and methods
Data collection and study population. Data were derived from the 2018 Korea Community Health Survey (KCHS), a constitutes community-based, nationwide representative data. KCHS was designed by the Korea Centers for Disease Control and Prevention (KCDC) to assess the efficacy of community-level health promotion and disease prevention programs since 2008 21 . KCHS has been conducted annually at national public health centers. The surveyed subjects are aged 19 years or older and are selected by the sampling method. KCHS is annually reviewed and approved by the institutional review board of the KCDC, and written informed consent is obtained from all the participants. In 2018, KCHS conducted a nationwide sleep quality survey using the Pittsburgh Sleep Quality Index for the first time in Korea. The initial study population comprised 228,340 individuals. Among them, we excluded the following participants who did not respond to the questions or those whose data contained missing values for variables: no record of subjective cognitive decline (n = 167); no record of sleep quality (n = 11,428); no record of PHQ-9 score (n = 411); no record of health-related behavior (n = 186); no record of stress (n = 87); no record of marriage status (n = 276); no record of educational level (n = 356); no record of occupation (n = 964); no record of income (n = 7,746). Finally, 206,719 participants (91,805 male and 114,914 female) were used as a sample in the study.
Subjective cognitive decline. The main objective of this study was to analyze subjective cognitive decline.
SCD was self-observed impairment of more frequent or worsening of memory loss or confusion within the prior 12 months. The questionnaire on cognitive function includes questions regarding subjective cognitive decline. For example, the survey question for SCD is as follows: "During the last year, have you experienced memory loss or confusion getting worse or happening more often?" The response categories are classified as "yes" or "no. " Sleep quality and duration. The main exposures of interest were sleep quality and duration. Sleep quality was measured using the Korean version of Pittsburgh Sleep Quality Index (PSQI-K) 22 . PSQI-K is a self-reported questionnaire that measures the quality and patterns of sleep over a period of a month. It contains 19 items and seven sleep components: subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbance, use of sleep medication, and daytime dysfunction. Each component is scored on a scale ranging from 0 to 3. The global PSQI score is calculated by adding the scores of all the components together and it ranges from 0 to 21. Scores of greater than 5 are generally used to indicate poor sleep quality. Thus, in this study, we classified the survey participants into two groups as follows: the good sleep quality (scores of 0-5) and poor sleep quality (scores of 6-21) groups. As a reference, we also validated that previous studies on the Korean population also used a score of 5 as the cutoff point 17,23 . Sleep duration was measured by evaluating the responses to the following question: "How much sleep do you have per day?" Further, the questionnaire was designed for the participants to respond in hours and minutes. For the analysis, we classified the participants based on the hours of sleep as follows: < 5, 5 ≤ to < 6, 6 ≤ to < 7, 7 ≤ to < 8, 8 ≤ to < 9, and 9 ≤ h. According to World Sleep Society, the optimal sleep duration is about 7 to 8 hours 24 . Therefore, we set the individuals with the optimal sleep duration, ≤ 7 to < 8 h, were assigned to the reference group.
Covariates. In this study, demographic, socioeconomic, and health-related characteristics were included as covariates in the fully adjusted models. The demographic and socioeconomic covariates included age, marital status (married, divorced, separated, or widowed, unmarried), educational level (high school or below, college, graduate school or above), region (metropolitan, urban, rural), household income (low, medium-low, medium-high, high), and job type (specialized job, office worker, sales and service, agriculture and fishery, manual worker, others). The household income groups were divided into quartiles using the monthly average household income. The health-related covariates included depression, stress, health-related behaviors, and body mass index (underweight, normal, overweight, obese). Depressive symptom was measured using the Patient Health Questionnaire-9 (PHQ-9) and was classified as "yes" (PHQ-9 score ≥ 10) and "no" (PHQ-9 score < 10). The level of stress was determined using the following question "How much stress do you feel in your daily life?" www.nature.com/scientificreports/ Response categories were classified as "yes" (very much or much) and "no" (a little or rarely). The health-related behaviors that were analyzed included the current smoking status, high-risk alcohol use, and engaging in walking/physical activity for more than 30 min per day for more than 5 days in the last week. Individuals who did not currently smoke, who did not engage in high-risk alcohol consumption, and who had walked more than 30 min per day for more than 5 days in the last week were classified as "yes, " indicating that they had a healthy lifestyle.
Ethical standards. Korea Community Health Survey data provides anonymous, secondary data that is publicly available for scientific use. A more detailed description of the KCHS can be found on the website (https:// chs. kdca. go. kr/).

Statistical analysis.
Due to gender differences in sleep quality, all analyzes were stratified by sex 25 . All analyses included the use of sampling weighted variables constructed by the KCHS.weighted variables. A descriptive analysis was performed to examine the distribution of the general characteristics of the study population. We calculated the frequency and percentages of each variable and performed chi-square tests to examine significant differences in cognitive function depending on sleep quality and duration. The statistical significance level was defined as a p-value of < 0.05. A multiple logistic regression analysis was performed to determine odds ratios (ORs) and 95% confidence intervals (CIs) to identify the association between sleep quality, duration, and cognitive function after adjusting for sociodemographic and health-related covariates. ORs and 95% CIs were calculated to compare parameters between those who had good sleep behavior and those who had poor sleep behavior. A subgroup analysis of depression, stress, and health-related behavior was performed based on sleep quality and cognitive function. All the statistical analyses were performed using the SAS 9.4 software (SAS Institute, Cary, NC, USA).

Ethical approval. Korea Community Health Survey (KCHS) data is publicly accessible and written
informed consent is obtained from all the participants before participating in the survey. Instruments and study processes used for the study were approved by the Korea Centers for Disease and Control and Prevention Institutional Review Board. Respondents' information was completely anonymized for use for research purposes.
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2000.

Results
The general characteristics of the sex-stratified study population are presented in .1% in the < 5, 5 ≤ to < 6, 6 ≤ to < 7, 7 ≤ to < 8, 8 ≤ to < 9, and 9 ≤ h groups, respectively. Table 2 presents the logistic regression analysis results of both men and women adjusted for all the covariates. Individuals with poor sleep quality were more likely to experience SCD in the case of both men and women. Compared with those of the reference group (good sleep quality), the ORs (95% CIs) for cognitive decline in the group with poor sleep quality were as follows: OR = 1.89 [95% CI 1.75-2.04] in men and OR = 1.74 [95% CI 1.63-1.85] in women. Additionally, our study identified that sleep duration was significantly associated with cognitive decline. Our analysis revealed a U-shaped pattern between sleep duration and cognitive decline. The ORs were significant in all the groups for both the sexes except for in the group of men with a sleep duration of 6 ≤ to < 7 h. In both sexes, the group that overslept for 9 h or longer was the most likely one to experience SCD.
In order to analyze the relationship between each component of the PSQI and SCD, all the independent variables were adjusted for, and a multiple logistic regression analysis was performed (Table 3). Among the responses to each component item, the best response was considered as 0, and individuals with this response were set as the reference group in the analysis. We observed a dose-response relationship between the subjective sleep quality, sleep disturbance, and daytime dysfunction components and SCD in both men and women. In men, there was also a dose-response relationship between the component for the use of sleep medication and SCD. The ORs were significant in all the groups. Figure 1

Discussion
Sleep, an essential element of everyday life, is closely related to health. It is widely believed that poor sleep behaviors have a negative effect on an individual's general health. This study was conducted to identify the association between sleep quality and duration and SCD in the Korean population using representative data recorded via the KCHS. We also conducted a subgroup analysis based on depression, perceived stress, and health-related behaviors, which are factors related to sleep patterns. We observed that individuals who had poor sleep quality and short or long sleep durations were more likely to exhibit SCD compared to those who had good sleep quality and an adequate sleep duration. Odds ratios for men and women with poor sleep quality were 1.97 and 1.75 more likely to experience SCD, respectively. Additionally, in both sexes, the likelihood of developing cognitive impairment increased in individuals with insufficient or excessive sleep and not in those with an adequate sleep duration. The findings of the present study were consistent with those of earlier studies, which suggests that poor sleep is associated with cognitive function 17,26,27 . Another study conducted on Koreans also revealed that poor sleep quality can cause cognitive and functional decline 17 .
Several studies have demonstrated the effect of sleep quality on cognitive decline. A previous study has shown that poor sleep quality is associated with amyloid deposition 26 . Amyloid can affect the Alzheimer disease pathogenesis 28 . Also, sleep problems can cause more of the stress hormone cortisol, which can lead to brain inflammation and tissue damage 29 . In several cross-sectional studies, insufficient or decreased sleep quality was Table 3. The results of the analysis on the association between each component of the PSQI and subjective cognitive decline. PSQI Pittsburgh Sleep Quality Index, CI confidence interval, OR odds ratio. *OR adjusted for all covariates considered in the study. www.nature.com/scientificreports/ found to be associated with poor cognitive function [30][31][32] . In addition, prior cross-sectional study of older adults has shown that a nighttime sleep duration of 6 h or less and daytime sleep duration of 1 h or more could be associated with poor cognitive impairment with daytime sleepiness 33 .
Prior studies have also reported on a link between sleep duration and cognitive decline 34 . In a study that examined the association between sleep related factors and memory impairment in older Chinese individuals, an inverted U-shaped association was found between sleep duration and scores on the word delayed test, with the peak occurring at 7 to 8 h 31 . However, another study reported that sleep quality is more relevant to cognitive decline than sleep duration. The results of research conducted on elderly women suggest that the disturbance of sleep rather than quantity could be likelihood of poorer cognition. Sleep disturbances, such as waking up an increasing number of times after sleep onset, lower sleep efficiency, and lower sleep latency, were found to be consistently related to poorer cognition levels whereas total sleep time was not 30 . Another prospective study demonstrated that obstructive sleep apnea, a common sleep disorder that causes sleep disruption and hypoxia, increased the prospective risk of dementia in a cohort of elderly women 35 .
In addition, there are some studies that have demonstrated that sleep is associated with health or mental health outcomes, such as stress 8,9,[36][37][38] . An epidemiological study that included over one million American subjects reported that under 5 h or 10 h or more of sleep were predictors of mortality that were almost as strong as "ever had" diabetes, heart disease, stroke, or high blood pressure. The increased mortality rates were seen in both sexes and across all age groups 10 .
Previous study has shown that there is a greater association between poor health-related behaviors, such as smoking and high-risk drinking, and poor sleep quality in women 18 . In terms of psychiatry, subjective health status, stress, depressive symptoms, and subjective cognitive decline were strongly associated with poor sleep quality in both men and women. Some studies found that sleep disturbances were more relevant in subjects with poor mental health 39,40 . The result of this study was consistent with previous studies indicating that depressive symptoms were associated with cognitive decline 41,42 .
This study has several limitations that should be considered when interpreting the results. First, as this study used cross-sectional data, a clear causal relationship between the quality and duration of sleep and cognitive function could not be inferred. Second, because sleep quality and average sleep duration were evaluated based on the respondent's memory, there is a likelihood of recall bias. Third, cognitive decline was evaluated using subjective measurement in a self-report format and not a clinical diagnosis or experimental test. Therefore, this study has a limitation in that it did not objectively measure cognitive performance with appropriate tests. Fourth, it was not possible to consider unmeasured confounders, such as clinical sleep disorders or caffeine use, in this study. www.nature.com/scientificreports/ Despite these limitations, the present study has some strengths. First, we used nationwide representative data that are suitable for conducting Korean studies. The study subjects were randomly sampled from communitybased populations. Second, the study was based on the first nationwide survey that used the PSQI to measure sleep quality in Korea. The PSQI is a highly validated and reliable index of sleep quality, which is used worldwide. Third, mental health-related confounders, including depression and stress, which are strongly associated with sleep were included in the analysis. Furthermore, this study is meaningful in that its results support and add to the findings of previous studies by demonstrating that good sleep quality and an adequate sleep duration can help in preventing cognitive impairment. Thus, we could suggest that public policy and education developed for the promotion of mental health in Koreans should be focused on good sleep behaviors.

Conclusion
This study demonstrated that poor sleep behaviors are negatively associated with cognitive function in the Korean population. In addition, depression, stress, and bad health-related behaviors tend to increase the risk of cognitive impairment. These findings indicate that the potential benefits of good sleep quality and an optimal sleep duration could effectively prevent cognitive decline. Therefore, our research could help health policymakers and professionals in recognizing sleep behaviors as risk factors for a decline in cognitive function and in implementing prevention and intervention strategies for managing public mental health by promoting the benefits of good sleep quality and an optimal sleep duration.