Sleep occupies nearly one third length of an individual’s life span and sleep duration becomes shorter with increasing age [1]. Many older adults complain of sleep disturbances even they have the opportunity to sleep for longer periods [2]. Sleep disturbances are frequently manifested by difficulties with initiation or maintenance of sleep, multiple and prolonged awakenings during night, and early awakening in the morning [2, 3]. If the sleep difficulties become continual and chronic, the diagnostic criterion of insomnia might be made [4].

The incidence of insomnia is much higher in older adults, due to medical and social factors, as well as comorbid disorders in the elderly [5]. The lack of high-quality nocturnal sleep leads to day-time dysfunction, such as fatigue, tiredness, drowsiness, distress and impairment in social, occupational, behavioral, academic, or other functional domains [2, 3, 6, 7].

Accumulating evidence suggests that insomnia contributes to cognitive decline and incidence of dementia [8, 9, 10]. A meta-analysis found that insomnia was a risk factor for cognitive disorders, including Alzheimer’s disease (AD) [11]. Adults with insomnia exhibited impaired performance for several cognitive functions, including working memory [12], episodic memory [13], and some aspects of executive functioning [14], although these impairments were of mild to moderate magnitude [15]. Xu and his colleagues found that individuals with both elevated amyloid-beta protein (A$\beta$) and insomnia experienced faster cognitive decline than those with only elevated A$\beta$ or insomnia, suggesting a link between insomnia and the progression of AD [16].

Most older adults experience some types of sleep disturbance, but the condition does not necessarily meet the diagnostic criteria of insomnia. Self-reported sleep disturbance, such as sleep duration, has been reported to reflect poor overall quality of sleep [17, 18, 19, 20]. Although self-reported sleep duration is a habitual character with highly individual variation, it has been found to be closely associated with objective polysomnographic data [21] and is widely used in the epidemiological studies. The effect of sleep duration on cognitive function in general population has been widely reported [22]. A study in Finnic adults showed that short (7 h) periods of sleep duration were associated with cognitive decline, based on objective testing and self-reporting data [23]. Ohayon and Vecchierini found that self-reported lack of sleep ($\le$6 h) was associated with impaired self-reported cognitive to impairment in older adults [24]. These studies provide more evidence that insufficient sleep duration is related to impaired cognitive function.

Higher levels of education has been shown to be protective against the onset of dementia [25, 26, 27], while lower education level ($\le$8 years) have been reported to increase the risk of dementia or lower cognitive impairment by 80% [26]. Several studies observed that short sleep was seen among subjects with lower education levels as compared to college graduates [1, 28, 29, 30, 31, 32, 33]. However, whether self-reported sleep duration impacts cognitive function in the non-demented older adults with a higher education level ($\ge$9 years) is still unclear. Clarifying the impact of sleep duration on cognitive function in non-demented older adults may help to determine potential risk factors and provide better preventive strategies against cognitive disorders at an earlier stage. Therefore, we conducted a cross-sectional study to explore the relationship between sleep duration and cognitive function in non-demented older adults with high school or above education.

In this study, we explored the correlation between sleep duration and multiple domains of cognitive function in the non-demented older adults with high school or above education. The CFT scores measuring visuospatial ability in subjects with sleep disturbance were significantly lower than those in subjects with normal sleep. There were no significant differences in global cognition as measured by MMSE, or memory function as measured by AVLT total and AVLT N5, or language ability as measured by VFT, or attention and executive function as measured by TMT-A/B and SDMT between the two subgroups. However, we noticed that there were more females and participants with smaller BMI in sleep disturbance group, which contributed to cognition [45, 46]. We further analyzed the correlation between sleep duration and multiple domains of cognitive function using linear regression controlling these cofounders which may affect cognitive 214 function. After adjusting for age, gender, education and BMI (in model 1), both the nocturnal sleep duration and total sleep duration exhibited a significant positive correlation with global cognitive function as measured by MMSE scores. A similar result was observed after adjusting comorbidity with hypertension, diabetes, hyperlipidemia, coronary heart disease and household conditions (model 2). The visuospatial function as measured by CFT was also correlated with nocturnal and total sleep duration after adjusting age, gender, education and BMI in linear regression analysis of model 1, but not in model 2.

The effect of insomnia on cognitive function in the general population has been explored [3], but the results are still inconsistent. A few studies reported no significant differences in cognitive performance between subjects with insomnia and good sleepers [41, 47]. However, more observational studies reported that, as compared with good sleepers, insomniacs exhibited impaired function in episodic memory [12, 13], attention networks, vigilance and executive function [41, 47]. Yaffe and his colleagues reported that individuals with sleep disturbance had a 27% increased risk of dementia during the 8-year follow-up [48]. Another study using Taiwan’s National Health Insurance Research Database discovered that patients (aged 50 years and over) with long-term use of hypnotics have more than a 2-fold increased risk of dementia after 3 years of follow-up [49]. These 230 conflicting results may be attributed to the sample size, the definition criteria of insomnia [2], the sensitivity and specificity of cognitive assessment scales, and the statistical methodology used in these studies.

Since numerous observational studies were conducted in the general population which contains demented patients, we sought to explore the effect of insomnia on cognition in subjects without cognitive impairment to determine potential therapeutic interventions to prevent dementia. MMSE has been used to assess general cognitive functioning or screen dementia in elderly individuals [15, 50, 51, 52]. In the present study, global cognitive function as measured by MMSE, initially did not exhibit a significant difference between subjects with self-reported sleep disturbance and normal sleepers (Table 1). But after controlling for age, gender, education and BMI, comorbidity and household conditions, there was a significant positive correlation between sleep duration and MMSE scores. The CFT scores measuring visuospatial ability in subjects with sleep disturbance were also significantly lower than those in good sleepers. Our data further supports the premise that short sleep duration is harmful to cognition even in non-demented older adults.

Sleep duration direct reflects the quantity of sleep. The mean sleep duration of the elderly participants without sleep disturbance in our study (7.24 h) was similar to that reported in older individuals in France (6–8 h) [24] or the United States (7 h) [53]. Faubel et al. [22] reported that sleep duration over 11h per day is associated with poorer cognitive function in older adults.

Tsapanou et al. [54] reported that longer sleeper ($>$480 min per day) performed worse in memory assessment in adults without dementia and MCI. A meta-analysis by Xu et al. [11] recently found that insufficient (4 hours per night or total daily) or extensive ($>$10 hours per night and $>$12.5 h for total daily) sleep duration could elevate the risk of all-cause cognitive disorders or AD dementia. Several studies have demonstrated that both acute total and cumulative partial sleep loss lead to deteriorations in a wide range of cognitive performance, such as sustained attention, executive and memory functions [55, 56]. These studies suggest that short sleepers in the general population may show inferior cognitive functions when compared with 7–8 h sleepers [23]. In our study, the mean total sleep duration per day was 5.55 h in the sleep disturbance subgroup and 7.24 h in normal sleep subgroup respectively. Although the two values of sleep duration were distributed in the reported normal range (4–10 h) [11], our results suggest that there is a significant positive correlation between sleep duration and global cognition within this sleep duration range (5.55–7.24 h) in non-demented older adults.

Visuospatial ability refers to the ability to understand what we see around us and how we interpret spatial relationships, which is affected in multiple types of dementia, including the early stages of AD. We used the copy part of CFT to evaluate visuo-constructional ability and memory function. Worse CFT scores in sleep disturbance group were observed as compared with those in normal sleeper, even after controlling several covariates in the linear regression analysis. Van Dijk et al. [57] reported poorer performance on the tests of visuospatial function and executive of the working memory in subjects with sleep problems. Our data further indicated that visuospatial ability was impaired in subjects with sleep disturbance even in the very early non-demented stage.

An elevated risk of sleep disturbance was seen in women in the present study, which was consistent with a previous study [58]. Women have a 1.5 times higher risk of sleep disorders than men, which may be the result of a complicated interplay of biological, psychological, and social factors that play different roles throughout the life span [59]. These potential mechanism require further investigations.

Several limitations should be noted in the current study. First, the potential for selection bias exists 275 since the participants included in this cohort have a higher level of education that may not be found in 276 other study participants involved in this type of research. Second, the relatively small population size may limit our interpretation of the results, a larger sample size will be necessary in future studies. Third, other objective clinical techniques such as Pittsburgh sleep quality index and polysomnography, will need to be included in future insomnia studies. Fourth, the study could not 280 evaluate non-rapid eye movement (NREM) or rapid eye movement (REM) sleep disturbances, which have been associated with cognitive deficits in elderly adults [60].

In summary, the present results confirmed the effect of sleep duration on cognitive performance in non-demented elderly adults with high school or above education, and highlight the potential beneficial role of sleep duration on global cognitive performance in preclinical stages.

A$\beta$, amyloid-beta protein; AD, Alzheimer’s disease; AVLT, the Auditory Verbal Learning Test; AVLT N5, RAVLT-delayed recall; BMI, body mass index; CFT, Rey-Osterriech Complex Figure Test; MMSE, Mini-Mental State Examination; NREM, non-rapid eye movement; RBD, rapid eye movement sleep behavior disorder; REM, rapid eye movement; SDMT, Symbol Digit Modalities Test; TMT-A/B, the Trail Making Test-A and B; VFT, Verbal Fluent Test.

All authors contributed significantly to this research and preparation of the manuscript. Conceived and designed the experiments and wrote the manuscript—XP, XD, CZ, GF. Performed the experiments and analyzed the data—XD, XP, XC, JZ, LW, SS, CZ, GF. The corresponding authors—GF, CZ. All authors have been involved in the drafting, critical revision and final approval of the manuscript for publication. All authors agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

The study was approved by the Committee on Medical Ethics of Zhongshan Hospital, Fudan University, code 2009-013. Participants provided written informed consent before beginning the study.

We thank all of the participants and their families for their time and effort in the research. We would also like to thank Department of Neurology in Zhongshan Hospital Affiliated to Fudan University, and the State Key Laboratory of Medical Neurobiology, as well as Institutes of Brain Science & Collaborative Innovation Center for Brain Science in Fudan University.

This study was supported by grants from the National Natural Science Foundation of China (91332201, 82171411). Medical Innovation Program of Fujian Health Committee (2020CXB049), Medical and Health Technology Program of Science and Technology Bureau in Xiamen (3502Z20194028).

The authors declare no conflict of interest.

Supplementary material associated with this article can be found, in the online version, at https://doi.org/10.31083/j.jin2106168.