The World Health Organization projects the total number of people with dementia worldwide to reach 82 million in 2030, nearly doubling to 152 million by 2050 (Jia, Reference Jia2020). The estimated 2015 total global cost of dementia was a staggering 1.1% of global gross domestic product (World Health Organization, 2019). Dementia substantially impacts patients and their individual support systems as well as the broader community and societal structures. With approximately 25% of the world’s patients with dementia living in China, dementia poses a significant challenge for Chinese policymakers, public health and healthcare professionals, and communities (GBD 2016 Dementia Collaborators, 2019). The incidence of dementia and mild cognitive impairment in older Chinese adults (65+ years old) ranges from 17.7 to 24.0 per 1000 person-years (Jia, Reference Jia2020). The 2010 estimated cost of dementia in China was $47.2 billion (USD) and is projected to reach $114.2 billion in 2030 (Xu, Reference Xu2017). Without curative treatments for cognitive decline, better prevention strategies and early detection methods are key for reducing the impact of dementia. With recent evidence supporting a possible causal link with dementia, sleep-related factors are increasingly popular treatment and research targets.

Zhang and colleagues’ (Reference Zhang2021) article in the current issue examined links between sleep and cognitive impairment among 3,692 community-dwelling older Chinese participants from a 3-year longitudinal study. The authors reported that excessive sleep (≥10 hours/night) increased the risk of cognitive impairment by 30% (OR = 1.309, 95% CI: 1.019–1.683) relative to sleeping between 5-10 hours/night adjusting for sociodemographic, type of residence, lifestyle factors, and comorbidity covariates. Among those with long sleep duration, factors including being male, being 75+ years old, having at least primary education, and living in a rural area were associated with cognitive impairment at follow-up. Short sleep duration (≤5 hours/day) was not associated with cognitive impairment.

These findings contribute to a body of inconsistent evidence on the relationship between sleep duration and cognitive decline. The disparate findings may be linked to the heterogeneity of study samples related to sleep and cognitive functioning. For example, a 2007 US study found that black adults were more likely to have short and long sleep duration and Hispanic adults were more likely to have short sleep duration, relative to white respondents (Hale and Do, Reference Hale and Do2007). A 2016 US study found that black, Hispanic, and Asian adults were more likely to have short sleep compared to white adults (Cunningham, Reference Cunningham2016). Thus, diverse populations of older adults are warranted for greater generalizability.

One particular challenge of studying older adults in China is geographic and socioeconomic heterogeneity. The Zhang et al. (Reference Zhang2021) study included participants from the majority of China’s province and considered the diversity of environment and education in their analyses. Zhang and colleagues found education and rural living to be associated with cognitive impairment for long sleepers in this study. Given the differences across provinces, the study participants likely spoke different languages or dialects. Individuals with Alzheimer’s Disease often have worse cognitive performance in their nondominant language, compared to their dominant language (Ivanova et al., Reference Ivanova, Salmon and Gollan2014). Thus, whether the language used for the cognitive testing is the participant’s dominant language should be considered in the data analysis.

Level of education may also affect the validity of the cognitive measures used. The original Chinese version of the Mini-Mental State Exam (MMSE) was evaluated in a cohort of community-dwelling elderly in Shanghai (Katzman, Reference Katzman1988). Within the sample, 26.7% of the participants had not attended school and had a higher proportion of low scores as well as a different error pattern (Katzman, Reference Katzman1988). Other Chinese studies have established differential cut-offs based on years of education, rural residence, age, and sex – with dementia cut-offs of 16/17 for illiterate individuals, 19/20 for individuals with 1–6 years of education, and 23/24 for individuals with 7 or more years of education (Li, Reference Li2015).

The current study noted that among individuals with primary and above education, excessive sleep duration was linked to cognitive impairment (Zhang et al., Reference Zhang2021). Education may contribute to ‘cognitive reserve’ or the capacity of the brain to buffer against age-related decline and pathologic damage – thus delaying the onset of noticeable clinical symptoms (Stern, Reference Stern2012). For example, while brain pathology (such as decreased glucose metabolism and cortical thinning) may be worse among a group of individuals with higher education – performance on cognitive testing would be similar to individuals with less formal education (Jung, Reference Jung2018; Perneczky, Reference Perneczky2006). A cross-country comparison found formal education to have a large effect on cognitive function in most countries (Rodriguez et al., Reference Rodriguez, Hofbauer and Röhr2021). A meta-analysis by Meng et al. showed that older adults with education had significant reduction both in prevalence and incidence dementia, but they had accelerated cognitive decline (Meng and D’Arcy, Reference Meng and D'Arcy2012). This supports the concept of a ‘threshold effect,’ wherin higher education may aid in masking clinical symptoms initially, but as the brain pathology reaches a clinical level, performance declines more rapidly (Meng and D’Arcy, Reference Meng and D'Arcy2012). A 2014 study by Amieva and colleagues found that the first subtle signs of cognitive decline occurred 15–16 years (before the dementia diagnosis) in highly educated people, compared to 7 years in adults with less education (Amieva, Reference Amieva2014). The higher cognitive reserve and threshold effects could contribute to excessive sleep duration as a key modifiable factor for cognitive impairment among highly educated people, but not among less-educated older adults.

Future studies of cognitive trajectories in diverse populations could consider unique test items as well as assessing rate of decline rather than a threshold. A 2003 study adapted the Chinese version of the MMSE by substituting literacy-dependent items with equivalent non literacy-dependent items for illiterate older adults and established differential cut-offs for each group (Xu, Reference Xu2003). A 2019 study examined rates of cognitive decline and reported that those with rapid cognitive decline (greater than average reduction of MMSE scores) had a higher risk of death than those with stable cognitive function (Lv, Reference Lv2019).

Similar to tracking cognition, future studies may benefit from tracking sleep over a longer period of time or over individual’s lifespan. Like most large cohort studies, the Zhang et al. (Reference Zhang2021) study used self-report measures of sleep duration at baseline, and did not capture sleep duration across the lifespan – thus it is unclear if sleep duration during childhood, young adulthood, or middle age would be linked to future cognitive decline as well. Alterations in sleep may also be linked to aging milestones – such as having young children or retirement. Thus, both physiological and lifestyle changes linked to aging contribute to sleep habits.

In conclusion, the Zhang et al. (Reference Zhang2021) study and the diverse cohort of older Chinese study participants provide an important contribution to the growing knowledge regarding sleep duration and cognitive impairment. However, further exploration of education, language, and lifespan changes are warranted to establish these causal factors and identify which individuals would be most likely to benefit from potential preventative sleep-related interventions or use of sleep as an early risk factor for future cognitive impairment.