Reported sleep duration reveals segmentation of the adult life-course into three phases

Sleep is crucial for well-being, impacting numerous biological and cognitive functions. While age is a significant factor influencing sleep duration, with younger adults reporting less sleep than older adults, and significant cross-national variations have been observed, large-scale epidemiological studies encompassing diverse populations remain scarce. Previous studies often relied on small sample sizes and were predominantly focused on high-income countries, limiting the generalizability of findings. This study aims to address these limitations by analyzing a massive dataset of self-reported sleep durations from a globally diverse population to identify patterns in sleep duration across the adult life course and investigate the relationship between sleep duration and cognitive function, while considering the influence of geographical and socioeconomic factors. The study utilizes data collected through Sea Hero Quest, a mobile game designed to assess navigation ability, providing a unique opportunity to gather sleep data from a large, geographically diverse participant pool. By analyzing this extensive dataset, the researchers hope to gain a more comprehensive understanding of sleep patterns across the adult lifespan and their associations with cognitive performance.

Existing literature demonstrates a strong link between sleep and various bodily functions, from gene expression and cellular processes to complex brain functions related to cognition and mental health. Genetic and environmental factors, including age and artificial lighting, modulate sleep patterns. Studies have shown variations in reported sleep duration across nations, with Asians often reporting shorter sleep durations than other populations. However, previous research often suffered from small sample sizes, limited geographical representation (mostly high-income countries), and methodological heterogeneity, hindering the ability to draw robust conclusions about global sleep patterns and their correlates. This study seeks to overcome these limitations by utilizing a substantially larger and more diverse dataset.

Data was collected from 3,881,449 participants who played the Sea Hero Quest mobile game. Of these, 730,187 participants from 63 countries provided their average sleep duration, after excluding participants over 70 (due to selection bias), those reporting less than 5 hours or more than 10 hours of sleep, and countries with fewer than 500 participants. Participants also provided demographic information (gender, education, home environment, commute duration). Spatial navigation ability was assessed using a subset of participants who completed specific wayfinding levels in the game. A wayfinding performance metric (WF) was calculated to quantify spatial ability, corrected for video-gaming skills. A training performance metric (TP) was also calculated for comparison. Linear-mixed models were used to analyze the relationships between sleep duration, age, demographics, and cognitive performance. Change points in the age-sleep duration curve were identified using a parametric global method. Analysis included comparisons between WEIRD (Western, Educated, Industrialized, Rich, and Democratic) and non-WEIRD countries, and exploration of geographic clustering of sleep duration using established supra-national and cultural clusters. Statistical analyses involved ANOVAs and linear regressions to assess the significance of various predictors. The change points in the age-sleep duration curve were estimated using a parametric global method that minimized the total residual error by systematically varying the location of division points. The method incorporated a penalty term to avoid overfitting by penalizing the addition of extra change points unless the reduction in error met a certain threshold. Linear-mixed models were used to predict reported sleep duration using various fixed effects (age, gender, etc.) and random intercepts clustered by country. Analyses of the spatial navigation data utilized ANOVAs to test the effects of sleep duration and other variables on wayfinding and training performance.

The world average self-reported sleep duration was 7.01 hours. Women reported sleeping 7.5 minutes more than men. Analysis revealed three distinct phases in adult life course regarding sleep duration: a sharp decrease from early adulthood (19-33 years) to a plateau in mid-adulthood (34-53 years), followed by an increase in late adulthood (54+ years). The proportion of short sleepers increased with age, while the proportion of long sleepers showed a U-shaped relationship with age. Age was strongly associated with sleep duration, as was gender. A strong interaction was observed between the effects of age and education, with tertiary education leading to shorter sleep durations among younger participants. Commute duration also showed an association, with longer commutes related to shorter sleep duration for those commuting more than one hour. There was a weak association between home environment and sleep duration. In late adulthood (54-70 years), spatial navigation performance showed an inverted U-shaped relationship with reported sleep duration, with optimal performance at 7 hours of sleep. This relationship was most significant in this age group and held when stratifying by gender and education. Country-level average sleep duration varied considerably, showing geographic clustering consistent with supra-national regions. This clustering was strongly associated with latitude and GDP per capita. No significant difference in sleep duration patterns was found between WEIRD and non-WEIRD countries.

This study provides the largest single analysis of self-reported sleep duration across the lifespan and its relationship with geographical location and socioeconomic factors. The identification of three distinct phases in sleep duration across the adult life course is a significant finding, suggesting that these periods may be characterized by different sleep needs and influences. The observed increase in reported sleep duration in late adulthood aligns with previous research suggesting a reduction in childcare and other demands that may have previously limited sleep. The inverted U-shaped relationship between sleep duration and spatial navigation performance in late adulthood extends prior findings from smaller, WEIRD-focused studies. The lack of a similar relationship in younger age groups might indicate that the impact of sleep duration on cognition varies across the lifespan, potentially related to different neural mechanisms involved in cognitive processes at different ages. The geographic clustering of sleep duration highlights the influence of cultural norms and environmental factors such as latitude. Further research is needed to explore the underlying mechanisms linking latitude to sleep duration and to determine the extent to which these factors can account for the observed geographical patterns. The relatively consistent patterns observed across WEIRD and non-WEIRD populations suggest that the identified relationships between sleep duration, age, and cognition may be broadly generalizable, although further research with objective sleep measures is warranted.

This study provides valuable insights into global sleep patterns and their relationship with cognitive function. The findings highlight the importance of considering age and geographical context when interpreting sleep data. Future research should focus on validating these findings using objective sleep measures and investigating the causal mechanisms underlying the observed relationships. Further research could also explore the role of additional demographic and environmental factors in influencing sleep duration and cognitive performance across different populations and cultures. The use of large-scale citizen science initiatives, such as Sea Hero Quest, offers a powerful approach for gathering globally representative sleep data and advancing our understanding of the global sleep landscape.

The study relies on self-reported sleep duration, which might differ from actual sleep patterns. The concordance between self-reported and objectively measured sleep can be influenced by various factors, such as age, sex, and the presence of sleep disorders. The study did not collect information on sleep quality, ethnicity, socioeconomic status, neurological or psychiatric disorders, inflammation levels, day of the week, time of day, and light levels during cognitive testing—factors known to affect sleep and cognitive performance. Future studies should incorporate these factors to gain a more comprehensive understanding. The cross-sectional nature of the study limits inferences about causal relationships between sleep duration and cognitive function.