Association between Sleep duration and Incidence of type 2 Diabetes in China

Objective: We assess the relationship between night sleep duration and the incidence of type 2 diabetes in China. Methods: We used logistic regression models to estimate odds ratios (ORs) and 95% condence intervals (CIs) for the incidence of type 2 diabetes among 11,539 participants at baseline from the REACTION cohort. Results: Compared to people who sleep for 7-8 h/night, people with a longer sleep duration ( ≥ 9 hours/night) had a greater risk of type 2 diabetes (hazard ratio [OR] 1.27; 95% condence interval [CI]1.01–1.61), while shorter sleep (<6 h/night) had no signicant difference. When the dataset was stratied based on selected covariates, the association between type 2 diabetes and long sleep duration became more evident among individuals < 65 years of age, male, body mass index<24 or with hypertension or hyperlipidemia, no interaction effects were observed. Furthermore, compared to people persistently sleeping 7–9 h/night, those who persistently slept ≥ 9 h/night had a higher risk of type 2 diabetes. The optimal sleep duration was 6.3-7.5 h/night. Conclusion: Short or Long sleep duration were associated with a high risk of type 2 diabetes. Persistently long sleep duration increased the risk. The optimal sleep duration was 6.3-7.5 hours/night.


Introduction
Type 2 diabetes is a critical public health challenge worldwide. Patients with type 2 diabetes are at increased risk for premature mortality and hospitalization due to complications [1]. Given the global burden of type 2 diabetes, understanding the impacts of modi able risk factors is of great importance.
Sleep is essential to the health of patients with type 2 diabetes [2]. Although humans spend about a third of their time sleeping, they may not understand the importance of it. Adequate high-quality sleep is vital to maintain the normal physiological state of the body [3]. Insu cient sleep is a health problem [4], and long sleep duration is associated with increased body mass index (BMI) [5], impaired glucose tolerance [6], and increased probability of developing type 2 diabetes [7,8]. Although lifestyle changes, such as increasing physical activity and weight loss, are of great importance to the management of this disease, understanding the link between type 2 diabetes and sleep duration may help to reduce its incidence.
Short sleepers [9] and long sleepers [10] show increased incidence of type 2 diabetes. In addition, a study in Japan found a J-shaped relationship between sleep time and HbA 1 c level [11]. Therefore, this retrospective cohort study assessed the associations between both nighttime and daytime napping and risk for type 2 diabetes. We used data form the REACTION study cohort, which covered a 4-year period. To the best of our knowledge, no such study has yet explored this relationship in Chinese people with a risk for type 2 diabetes.

Study subjects
We used data from the REACTION cohort study, which investigated the association between type 2 diabetes and pre-type 2 diabetes and the risk of cancer in the Chinese population [12]. All subjects live in Laoshan, Jingding, and Gucheng communities of Beijing (China) or in Ningde and Wuyishan, Fujian Province (China). They were invited to complete baseline questionnaires and medical examinations in 2011. During the rst follow-up survey in 2015, the same community were investigated, and the study had a total size of 13,311 participants.
In all, 1,556 subjects were diagnosed as type 2 diabetes in 2011 and 216 subjects lacked self-reported sleep duration or reported sleep durations of <4h/night or > 12h/night. These subjects were excluded. The remaining 11,539 subjects (4,043 men, 7,496 women) were enrolled in the present study.
Standard protocol approvals, registrations, and patient consents

Statistical analysis
We performed all analyses with Stata statistical software, ver. 14.2 (Stata Corp, College Station, TX, USA). Data were compared using analysis of variance (ANOVA) for continuous variables and χ 2 analysis for categorical variables. Continuous variables are expressed as mean ± standard deviation (SD), and categorical variables aer expressed as number and proportion.
Binary logistic regression was constructed to assess the ORs and 95% con dence intervals (CIs) of sleep duration and type 2 diabetes using sleep duration of 7 to -8 h/night as the reference groups, as previous studies have suggested that sleeping for 7-8 h [15,16] is optimal. Potential covariates included in the multivariate-adjusted model were age, gender, body mass index, smoking and drinking status, hypertension and hyperlipidemia. Considering that type 2 diabetes risk might follow nonlinear trends, we used restricted cubic spline model with 3 knots at 25 th , 50 th and 75 th percentiles of sleep duration and 7 h/night as the reference group [8,17]. Subjects were divided into several groups by age (<65,≥65 years), sex (male, female), BMI (<24, ≥24 kg/m 2 ) [18], hypertension (yes, no), and hyperlipidemia (yes, no). In addition, potential interactions were tested using interaction terms of these covariates with sleep duration.
We evaluated the association between changes in sleep duration and incidence of type 2 diabetes. This association was examined using crude and multivariate-adjusted models, with subjects persistently sleeping 7-9 h/night in both surveys as the reference group.
We further evaluated the joint effects of sleep duration and midday napping and that of sleep duration and sleep quality on the risk of developing diabetes, using moderate sleep duration (7-8 h/night) with midday napping (1-29 min), and moderate sleep duration (7-8 h/night) with good sleep quality as the reference groups.

Results
Among 11,539 subjects, 13.11% (1,513)reported sleeping ≥ 9 h/night and 5.18% 598 reported midday napping > 90 minutes. Compared to subjects reporting 7-8 h/night of sleep, those reporting sleep duration ≥ 9 h/night were more likely to be female, current smokers, and current drinkers. Meanwhile, compared to subjects with 1-29 minutes of midday napping, those reporting midday napping ≥ 90 minutes were more likely to be male, hypertension and hyperlipidemia, current smokers, current drinkers (p < 0.05). In addition, compared to the reference groups, participants reporting ≥ 9 hours/night of sleep were less likely to have hypertension and hyperlipidemia (Table 1). hours/night(p = 0.04), respectively. After adjusting for age (continuous), sex, body mass index (continuous), smoking status (yes or no), drinking status (yes or no), hypertension (yes or no), hyperlipidemia (yes or no), similar associations were observed for those obtaining ≥ 9 h/night of sleep (p = 0.02) ( Restricted cubic spline regression analysis showed a J-shaped curve and con rmed that people who slept ≥ 9 h/night had a high risk of type 2 diabetes (Fig. 1). The optimal nighttime sleep duration was 7.2-7.5 h, and it was 6.3-7.5 h after adjusting for all variables. When strati ed by selected covariates, the association between type 2 diabetes and long sleep duration became more evident in individuals who were < 65 years of age, male, body mass index < 24 or with hypertension or hyperlipidemia, no interaction effects was observed (Fig. 2). Abbreviations: CI = con dence interval; OR = odds ratio.
*adjusted for age (continuous), sex, body mass index (continuous), smoking status (current, former, or never), drinking status (current, former, or never), hypertension (yes or no) and hyperlipidemia (yes or no).

Discussion
In this large retrospective cohort study, we found that subjects who slept more than 9 h per night had a high risk of type 2 diabetes. Moreover, optimal sleep duration at night was 6.3-7.5 h after adjusting for age, sex, body mass index, smoking status, drinking status, hypertension and hyperlipidemia. To avoid an in uence of region on our results, we selected people in both the northern and southern regions of China. To the best of our knowledge, this is the rst retrospective study to report that persistently sleeping ≥ 9 h/night is related to higher type 2 diabetes risk compared to persistently sleeping for a moderate duration (7-9 h/night) in the Chinese population.
Many studies have assessed the association between sleep duration and the incidence of type 2 diabetes [8,19] or blood glucose level [9]. In these studies, participants have usually divided participants into several groups according sleep duration. Therefore, such studies can only draw the conclusions that one group of people has the lowest incidence of type 2 diabetes, for example, people who sleep 7-8 hours at night [9] A meta-analysis suggested that for short duration of sleep (5-6 h/night), the RR was 1.28, while that for long duration of sleep (8 − 9 h/night) was 1.48 [19]. In the present study, we limited the optimal sleep duration into a narrow range according to the restricted cubic spline model.
The elevated risk for type 2 diabetes and long sleep duration appeared to be more pronounced in individuals who were < 65 years old, male, BMI < 24, and/or with hypertension or hyperlipidemia, no interaction noted. However, we did not nd signi cant differences between each subgroup based on the P-values of interaction terms. The reason for the lack of interactions may be that we observed individuals for a short period of time and type 2 diabetes is a chronic disease with a relatively low incidence. As a result, we observed fewer cases.
Afternoon napping is a common habit in many countries, including in China. The relationship of midday napping and the risk of type 2 diabetes has also been investigated in several previous cross sectional [20][21][22] and cohort studies [23,24]. These studies have suggested that the incidence of type 2 diabetes or elevated blood glucose levels are higher in individuals with longer sleep duration. However, midday napping can in uence the quantity and quality of nocturnal sleep [25]. Therefore, we considered the effects of midday napping on the incidence of type 2 diabetes alongside that of nocturnal sleep. To the best of our knowledge, this is the rst study to examine the impact of midday napping in combination with nocturnal sleep on the incidence of type 2 diabetes in the Chinese population. The OR for the incidence of type 2 diabetes was 2.35 or 2.28 in people who slept more than 9 h at night and more than 90 min at noon.
Only eight people in this group developed new type 2 diabetes within the follow-up period of 4 years, and therefore the difference was not signi cant. Observing the relationships between the incidence of type 2 diabetes and both sleep duration and midday napping demands a large number of subjects.
Previous research has suggested that type 2 diabetes patients prefer longer midday naps because they are more fatigued [26]. We excluded individuals with diabetics at the start of our study, and therefore we can conclude that longer midday napping lead to type 2 diabetes.
Previous studies have showed that the prevalence of poor sleep quality was signi cantly higher among people with diabetes than those without it [27]. In our study, only 16 participants reporting ≥ 9 h/night sleep and poor sleep quality developed diabetes during the follow-up period. As a result, the difference in the incidences of diabetes are not statistically signi cant between this group and the reference group. A larger sample size is needed to investigated the effects of sleep quality.
Numerous possible mechanisms could explain the relationship between long sleep duration and midday napping and the incidence of type 2 diabetes.
Research revealed that leptin and ghrelin are of great importance to the incidence of type 2 diabetes [28]. Short sleep can reduce leptin and elevate ghrelin in the blood, leading to increases in hunger and appetite, accompanied by a decrease in glucose tolerance [29]. In addition, short sleep may contribute to the secretion of adiponectin and insulin. Adiponectin, which is secreted by adipocytes, is associated with insulin sensitivity [30,31]. Sleep restriction can increase sympathetic nervous system activity, leading to decreased insulin sensitivity [32,33]. However, the mechanisms through which long sleep duration leads to increased risk for type 2 diabetes are not fully understood. Long sleep duration may re ect a more sedentary life style and, similar to short sleepers, long sleepers may engage in more snacking [34].
The study also had several limitations. First, the sleep duration at night and midday is de ned as the time from going to bed to waking in the questionnaire survey, which is slightly different from the actual time to sleep. However, it was impossible to botain objective measures of sleep duration and napping in large prospective population studies, and the self-administrated survey is the most commonly used method of evaluating sleep duration and napping in large population-based studies. Second, previous studies have suggested that poor sleep quality is associated with higher blood glucose levels in patients with type 2 diabetes [35]. However, some subgroups contained few cases of type 2 diabetes, causing their differences to be nonsigni cant. A larger sample size is needed in future studies on this subject.

Conclusions
We found a J-shaped relationship between sleep duration and the incidence of type 2 diabetes, with the lowest risk for type 2 diabetes in individuals sleeping 6.3-7.5 h after adjusting for covariates. Sleep duration that is too long or too short increase the risk of type 2 diabetes. Further studies are needed to reveal the mechanism driving the relationship between sleep time and the incidence of diabetes.