This study provided new evidence between midpoint of sleep and T2DM in rural population, which had implications for better understanding of chronotype with T2DM. The current study found a U-shaped relationship between midpoint of sleep and T2DM, which means that early and late midpoint of sleep may increase the prevalence of T2DM. Besides, late midpoint of sleep and longer night sleep duration might be jointly associated with a higher likelihood of T2DM. Moreover, gender and age might influence the health effects of sleep which should be taken into consideration when making strategies to improve public health in rural areas.
On the basis of precious studies, we found late midpoint of sleep was related to increased odds of T2DM. For instance, former studies showed that late chronotype assessed by chronotype questionnaire was associated with increased risk of T2DM . Moreover, chronotype acquired by free-day mid-sleep time also showed similar results. A survey of large sample of diverse US Hispanics/Latino adults found that later sleep timing based on sleep midpoint was significantly associated with higher estimated insulin resistance (HOMA-IR) in individuals with and without diabetes combined independent of sleep duration . Furthermore, late chronotype was associated with poorer glycemic performance in individuals who were prediabetes and T2DM . In addition, consistent with other findings , long night sleep duration was associated higher risk of T2DM.
However, associations between midpoint of sleep and T2DM remained inconsistent with previous studies. This study suggested that early midpoint of sleep also increased the T2DM risk. A cross-sectional study involving 821 participants reported that M-type was associated with diabetes in middle-aged and elderly people . One possibility explanation for differences observed between these studies is that there are large differences in circadian patterns between rural and urban populations . The rural population presented a more predominantly early sleep pattern, as determined by the mid-sleep phase. In present study, people might have early midpoint of sleep due to work schedule, suggesting that mismatch between internal circadian rhythms and social factors could play a significant role for development of diabetes . Though there were no objective parameters of chronotype in the present study, the association between chronotype and rural population should be addressed in future studies.
Two studies previously explored the combined effect of sleep behaviors and T2DM. Liu et al. showed that longer napping and night sleep duration were jointly associated with T2DM . and Lou et al. presented that combined effect of sleep duration and sleep quality on the risk T2DM, which highlighted that both shorter sleep duration and poor sleep quality were jointly associated with T2DM . The present study found that midpoint of sleep and long night sleep duration might be jointly associated with a higher prevalence of T2DM. Investigating combined associations of midpoint of sleep and night sleep duration simultaneously with T2DM might have important implications for understanding the impact of sleep on T2DM.
Gender and age can modify these detrimental effects of midpoint of sleep on T2DM. Consistent with our results, Fabbian et al. reported that Eveningness may impact general health especially in women . However, one analysis of 1620 middle-aged Korean men and women (age range 47–59 years), stronger associations of Evening Chronotype with T2DM were presented among men . Consistent with Liu et al. [35 ] that women were more susceptive to the adverse effects of night sleep duration, we also found that short and long night sleep duration had more effects on women. Different biological and lifestyle characteristics may attribute to the gender differences. Besides, participants less than 60 years with early midpoint of sleep had statistically stronger associations than the others in present study which consistent with former findings . Individuals with younger age tend to under a mismatch between circadian and social activities such as working and studying and have adverse health outcome [32, 37].
All these above findings supported that disruption of circadian clock was associated to adverse changes on T2DM. There are several potential explanations of the biological mechanisms underlying the association between midpoint of sleep and night sleep duration with T2DM. One possibility is that circadian misalignment leads to fluctuation in the metabolic and endocrine function, including lower glucose tolerance and thyrotrophic concentration, increased evening cortisol concentrations and elevated activity of the sympathetic nervous system, which might accelerate the development of diabetes . Another is that melatonin secretion at night might have been disrupted by short sleep duration, circadian misalignment and/or light exposure, which may one of the endocrine mediators of circadian regulation of insulin sensitivity . Additionally, circadian rhythm was associated dietary intake and dietary behavior. An animal experiment showed that the Clock gene mutant mice have an attenuated diurnal rhythm of feeding behavior, and develop obesity, hyperphagia, reduced energy expenditure and visceral adiposity, as well as dysregulation of glucose and lipid metabolism .
There are several limitations needed to address in this study. First, the midpoint of sleep was calculated using sleep time and wake-up time of previous one month, but the midpoint of sleep was different between free days and work days and it is reported that free day midpoint of sleep was correlated better with chronotype scores assessed by chronotype questionnaire. However, because of the hot weather and any other reason (i.e. farming), most of the participants were on vacation during baseline assessment. Therefore, the participants probably had free time during the past month. Second, there were no objective parameters of chronotype in the present study which should be addressed in future studies. Third, the midpoint of sleep was classified into three categories by the 25th and 75th percentile based on previous study  and might lead to misclassification. Fourth, this study was a cross-sectional which was unlikely to yield a clear cause-effect conclusion and the information of sleep was self-reported which might suffer recall bias.