Restfulness from sleep and subsequent cardiovascular disease in the general population

We aimed to clarify the association between restfulness from sleep and subsequent risk of cardiovascular disease (CVD). Medical records of 1,980,476 individuals with neither prior history of CVD nor sleep disorders were extracted from the Japan Medical Data Center. Restfulness from sleep was subjectively assessed using information from the questionnaires at initial health check-ups. The mean age was 45 ± 11 years and 1,184,937 individuals were men. Overall, 1,197,720 individuals (60.5%) reported having good restfulness from sleep. The mean follow-up period was 1122 ± 827 days. Myocardial infarction, angina pectoris, stroke, heart failure, and atrial fibrillation occurred in 3673 (0.2%), 30,241 (1.5%), 13,546 (0.7%), 28,296 (1.4%), and 8116 (0.4%) individuals, respectively. Multivariable Cox regression analyses including age, sex, and other CVD risk factors after multiple imputation for missing values showed that good restfulness from sleep was associated with lower incidence of myocardial infarction (hazard ratio [HR] 0.89, 95% confidence interval [CI] 0.83–0.95), angina pectoris (HR 0.85, 95% CI 0.83–0.87), stroke (HR 0.85, 95% CI 0.82–0.88), heart failure (HR 0.86, 95% CI 0.84–0.88), and atrial fibrillation (HR 0.93, 95% CI 0.89–0.97). The association of restfulness from sleep with CVD events was pronounced in subjects with younger age and female sex. In conclusion, good restfulness from sleep may be associated with the lower risk of myocardial infarction, angina pectoris, stroke, heart failure, and atrial fibrillation. Further studies are required to clarify the underlying mechanism and to develop a novel preventive approach for CVD from the perspective of sleep.

The problem with research on this topic might be the reliability of self-reported sleep duration. Self-reported sleep durations generally indicate total time spent in bed, which is different from the duration of actual physiological sleep. Lauderdale et al. reported that objective sleep duration evaluated using actigraphy was shorter than the self-reported sleep duration 19 . Furthermore, the optimal sleep time for an individual may depend on genetic factors 20 , and appropriate sleep duration varies depending on the individual. Therefore, risk stratification using sleep duration alone has various limitations.
Sleep quality is another important factor in the assessment of sleep. Various studies demonstrated that sleep quality could be also associated with cardiovascular health [21][22][23] . Several studies have demonstrated that poor subjective sleep quality was independently associated with higher all-cause mortality in older men 23,24 . It has recently been proposed that sleep quality might be more important than sleep duration 25 . Therefore, the importance of comprehensive assessment for restfulness from sleep using both sleep duration and quality is currently recognized. In Japan, the information on restfulness from sleep is routinely obtained using the standard questionnaire in Specific Health Checkups which is the annual health screening and promotion service organized by the Japanese Ministry of Health, Labour and Welfare (MHLW) 26 . In this study, we sought to clarify the association between restfulness from sleep and incident CVD among the general population without prevalent history of CVD using a nationwide epidemiological database.
Ethics. We conducted this study in accordance with the principles of the Declaration of Helsinki. The requirement for informed consent was waived because all data in the JMDC database were anonymized and de-identified. All data were compliant with the International Conference on Harmonization guidelines 34 . Institutional Review Board of the University of Tokyo approved this study (2018-10862). Institutional Review Board of the University of Tokyo also provided waiver for informed consent. We performed this study in accordance with relevant guidelines/regulations.

Restfulness from sleep.
We obtained information regarding restfulness from sleep from questionnaires in the individuals' initial health check-up (specific health check-up) records. The questionnaire presented at these health check-ups was almost uniform because a regular health check-up is mandatorily performed under the jurisdiction of the Ministry of Health, Labor, and Welfare for most Japanese employees using a standardized format and protocol. If a study participant answered "YES" to the question, "Do you have a good rest with sleep?" then this study participant was categorized as having good restfulness from sleep. If a study participant answered "NO" to this question, then this study participant was categorized as having poor restfulness from sleep.
Definition. Obesity was defined as body mass index of ≥ 25 kg/m 2 . High waist circumference was defined as ≥ 85 cm for men and ≥ 90 cm for women 35 . Hypertension was defined as systolic blood pressure of ≥ 140 mmHg, diastolic blood pressure of ≥ 90 mmHg, or use of anti-hypertensive medications. Diabetes mellitus was defined as a blood glucose level of ≥ 126 mg/dL, or use of antidiabetic medications including insulin. Dyslipidemia was defined as low-density lipoprotein cholesterol level of ≥ 140 mg/dL, high-density lipoprotein cholesterol level of < 40 mg/dL, triglyceride level of ≥ 150 mg/dL, or use of lipid-lowering medications.
Statistical procedures. Categorical and continuous data of the baseline characteristics are presented as numbers (%) and means (standard deviation). The chi-square test was used to compare the categorical variables. The unpaired t-test was used to compare continuous variables. We performed the multivariable Cox regression analysis (complete case analysis) to identify prognostic determinants for subsequent CVD events. To account for the missing data values shown in Table 1, we also used multiple imputation as a statistical procedure as previously described 36 . Multiple imputation is a statistical procedure to avert bias caused by missing data by creating multiple filling-in patterns to replace missing values with other plausible values. It is also recognized as an alternative approach for analyzing incomplete data 37 . In this study, we replaced each missing value with a set of substituted plausible values by creating 20 filled-in complete data sets using a multiple imputation with the chained equation method 38 . Covariates included in the multivariable Cox regression analysis were used in the multiple imputation process. Hazard ratio (HR) and standard errors were obtained using Rubin's rules. As subgroup analyses, we divided our study population by age (≥ 50 years, < 50 years), sex (men, women), and obesity (obese, non-obese), and conducted the multivariable Cox regression analysis in each subgroup. The P values for interactions between groups were calculated. A probability value of < 0.05 was considered statistically

Results
Study population. We analyzed the data of 2,218,206 individuals who were enrolled in the JMDC database between January 2005 and August 2018 and whose baseline health check-up data (including data on sleep) were available. Exclusion criteria were as follows: (1) age < 20 years (n = 19,948); (2) prior history of myocardial infarction, angina pectoris, coronary revascularization, heart failure, stroke, atrial fibrillation, or hemodialysis (n = 103,011); and (3) existence of sleep disorders defined as ICD-10 code of G47 (e.g., sleep apnea syndrome and insomnia) (n = 114,771). Ultimately, data from 1,980,476 individuals were analyzed in this study. The mean follow-up period was 1122 ± 827 days.

Baseline clinical characteristics.
Characteristics of the study population are shown in Table 1. Overall, the mean age was 45 ± 11 years, and 1,184,937 individuals (59.8%) were men. Among the total cohort, 1,197,720 individuals (60.5%) reported having good restfulness from sleep. Individuals with good restfulness from sleep were older. Obesity and high waist circumference were more common in individuals with poor restfulness from sleep. Cigarette smoking was less common, whereas alcohol drinking was more common, in individuals with good restfulness from sleep.  Table 3. Table 4. The association between restfulness from sleep and the incidence of myocardial infarction, angina pectoris, stroke, heart failure, and atrial fibrillation which was seen in the overall population, was observed in each subgroup except for the association of restfulness from sleep with the incidence of myocardial infarction, stroke, and atrial fibrillation among individuals with older age, and the association of restfulness from sleep with the incidence of atrial fibrillation among those with obesity. The association of restfulness from sleep with CVD events was significantly modified by age and sex, and was pronounced in subjects with younger age and female sex. The association of restfulness from sleep with CVD events was not modified by the presence of obesity.

Discussion
This comprehensive analysis of a nationwide epidemiological database included approximately two million individuals with neither prior history of CVD nor sleep disorders. The results demonstrated that approximately 60% of the general population reported having good restfulness from sleep and that good restfulness from sleep was independently associated with a lower risk of subsequent incidence of CVD, including myocardial infarction, angina pectoris, stroke, heart failure, and atrial fibrillation. Our results suggest the potential importance of having good restfulness from sleep in the primary prevention of CVD among the general population.
Although previous research has focused more on sleep duration and its effects on the incidence of clinical outcomes rather than on sleep quality 7,[9][10][11][12][13][14][15][16][17][18] , several studies have indicated that sleep quality could also affect clinical outcomes including subsequent CVD. For example, a population-based cohort study in Japan showed that poor self-reported sleep quality might be associated with an elevated risk of all-cause mortality independent of sleep duration 39 . Sleep quality assessed by the Jenkins Sleep Questionnaire showed that sleep disturbance was positively correlated with coronary heart disease 21 . A recent meta-analysis showed that sleep quality was associated with coronary heart disease but made no difference in mortality and other outcomes 7 . A systematic review with meta-analysis by Sofi et al. showed that insomnia was associated with the risk of CVD and CVD-related death 22 . Sleep duration and quality are both important, and therefore, we comprehensively assessed "sleep" using the answer of study subjects to the questionnaire regarding restfulness from sleep in this study.
Most studies regarding sleep and CVD have focused on its association with the incidence of coronary heart disease. Our study gives the first epidemiological data analyzing the relationship between restfulness from sleep and the incidence of various CVDs; this includes not only coronary heart disease but also other major CVDs such as stroke, heart failure, and atrial fibrillation. Furthermore, because sleep disorders are known to increase the risk of CVD 40,41 , we excluded individuals with established diagnoses of sleep disorders (e.g., sleep apnea syndrome and insomnia) from the analysis of this study. Still, having good restfulness from sleep was independently associated with the lower incidence of myocardial infarction, angina pectoris, stroke, heart failure, and atrial fibrillation.
Our study has clinical implications. Approximately 40% of the individuals reported poor restfulness from sleep, and that poor restfulness from sleep might elevate the risk of various CVDs. From this point, intervention for sleep may have great potential in preventive cardiology. It should also be noted that we could stratify the risk of CVD by simply asking individuals a short question about sleep. We believe that asking this simple question could be possible even in busy real-world clinical or public health settings and could provide physicians with important information on future CVD risk.
Results of subgroup analyses are important as well. The association of restfulness from sleep with CVD events was pronounced in subjects with younger age and female sex. Additionally, individuals not having restfulness from sleep were more common in those with younger age. The study population in this study are working for relatively large Japanese companies. Younger subjects serving for such companies in Japan, may be prone to be quite busy, have much mental or physical stress, and irregular lifestyles, which could be related to "poor sleep". Unfortunately, we did not have enough data assessing this point concretely. Further investigations are required to clarify the mechanism of these results. Considering that the importance of CVD prevention among young adults is increasingly recognized currently, restfulness from sleep could have a potential as a key element for CVD prevention in young generations, particularly younger women.
The next essential issues to be addressed are to clarify if interventions that improve restfulness from sleep is effective in CVD prevention and how to intervene in poor restfulness from sleep. Various psychological, physiological, social, and environmental factors could lead to poor restfulness from sleep, and there may be individuals with latent CVD among those with poor restfulness from sleep. Therefore, root cause investigation for poor restfulness from sleep is indispensable as the first step. The beneficial impact of good restfulness from sleep on subsequent incidence of CVD is seemingly similar regardless of the targeted disease. Therefore, we speculate that good restfulness from sleep might influence primitive development of CVD.
Several possible mechanisms can be suggested to explain the association between restfulness from sleep and the subsequent incidence of CVD. Poor sleep is known to induce leptin/ghrelin imbalance and to impair glycemic control, and these could lead to obese and metabolic disorders [42][43][44][45][46] . Sleep is also an important regulator of circadian rhythm, which is associated with the development of CVD 47 , and individuals with poor sleep quality may have disturbed circadian rhythms. Kondo et al. reported that the cyclic alternating pattern during sleep was associated with heart rate variability, blood pressure, and autonomic activity 48 , which could also affect the development of CVD. Meier-Ewert et al. demonstrated that sleep disorders were associated with elevated serum C-reactive protein levels, suggesting that inadequate sleep may cause chronic inflammation 49 , which plays a pivotal role in the pathology of various CVDs 50,51 . Sleep disturbances were also reported to accelerate subclinical CVD burden including coronary artery calcium, carotid intima media thickness, endothelial dysfunction,

Scientific Reports
| (2020) 10:19674 | https://doi.org/10.1038/s41598-020-76669-z www.nature.com/scientificreports/ Table 3. Multivariable Cox regression analysis for cardiovascular events after multiple imputation. Multivariable Cox regression analysis for cardiovascular events after multiple imputation showed that good restfulness from sleep was independently associated with the lower incidence of cardiovascular events including myocardial infarction, angina pectoris, stroke, heart failure, and atrial fibrillation.  The association between restfulness from sleep and the incidence of cardiovascular disease which was seen in the overall population, was observed in each subgroup except for the association of restfulness from sleep with the incidence of myocardial infarction, stroke, and atrial fibrillation among individuals with older age, and the association of restfulness from sleep with the incidence of atrial fibrillation among those with obesity. The association of restfulness from sleep with CVD events was significantly modified by age and sex, whereas not by the presence of obesity.  52 . Further studies are warranted to elucidate the underlying pathophysiological mechanisms of our results. There are several limitations in this study that need to be addressed. Although we performed multivariable Cox regression analyses, we could not eliminate the possibility of unmeasured confounders and residual bias. Data on sleep duration were unavailable in our database. Because the population in the JMDC database comprised employed working-age individuals, healthy worker bias should be acknowledged. Further investigations are required to confirm our results and to generalize them with populations of different ethnicities, races, education levels, and incomes. Comparing with other nationwide epidemiological data in Japan 53,54 , the incidence of CVD in our study is reasonable, and therefore, we believe that our data could have reflected real-world clinical practice. However, recorded diagnoses are generally considered less well validated due to the nature of retrospective design and administrative database. Although the mood is closely related to sleep, our database lacked the information on the mood.
In conclusion, this comprehensive analysis of a nationwide epidemiological database suggested that good restfulness from sleep was associated with a lower subsequent incidence of myocardial infarction, angina pectoris, stroke, heart failure, and atrial fibrillation among the general population without prior history of relevant CVD. This exemplifies the potential essential role of sleep in the primary prevention of CVD.