Socioeconomic Position and Excessive Daytime Sleepiness: A Systematic Review of Social Epidemiological Studies

The objectives of this empirical study are to describe and discuss the current literature available on the relationship between excessive daytime sleepiness (EDS) and the socioeconomic position (SEP) as well as to provide recommendations for consideration of SEP in sleep medicine and biomedical research. Databases Medline/PubMed, Web of Science, Google scholar and Scopus were screened from January 1990 to December 2020 using PRISMA guidelines and 20 articles were included in the final synthesis. Nineteen studies were cross-sectional and one study was longitudinal. Among these studies, 25.00% (n = 5) are focused on children and adolescent and the remaining 75.00% (n = 15) focused on adults and seniors. Ages ranged from 8 to 18 years old for children/adolescent and ranged from 18 to 102 years old for adults. Main SEP measures presented in these studies were education, income, perceived socioeconomic status and employment. The sample size in these studies varied from N = 90 participants to N = 33,865 participants. Overall, a lower educational level, a lower income and full-time employment were associated with EDS. Symptoms of EDS are prevalent in women, especially those with a low income or no job; and children and adolescents with difficult living conditions or working part time reported more sleep disturbances. SEP is already considered as an important determinant for many health outcomes, but if SEP is embedded in the experimental design in psychosomatic research, biomedical research and clinical practice as a constant variable regardless of outcome; it will move forward future investigations.


Introduction
Sleep is an essential state to restore and revitalize physiological and mental processes, occupying up to a third of a person's lifetime. The prevalence of sleep disturbances has been increasing steadily over the past decades [1][2][3][4][5], alongside modern societal and technological changes that influence quality and quantity of sleep.
Alterations in sleep have been linked to a wide range of health issues including cardiovascular or metabolic disorders [6][7][8][9][10][11][12], as well as an increase in workplace and occupational health issues due to sleep disturbances such as excessive daytime sleepiness (EDS) [13][14][15]. According to the National Sleep Foundation 2000 Omnibus Sleep in America Poll: 43% of adults report that they are so sleepy during the day to the point that it interferes with their daily activities a few days per month or more [16].
Sleepiness is an imperative biological function defined by the likelihood of falling asleep, however, when this tendency is increased and the propensity to sleep becomes compulsive, it is referred to as excessive daytime sleepiness (EDS) [17,18]. The latter is defined variously as daytime somnolence or when wakefulness is difficult to maintain [17,18]. This burdensome disorder has been widely studied in order to understand its physiological and neurological substrates; however, few studies have looked into the association between EDS and socioeconomic indicators of social class or socioeconomic position (SEP) [7,13,[19][20][21][22][23].
It has been previously shown that a lower socioeconomic status, including employment, educational level or income, is an important stressor and impacts sleep negatively [24]. Despite the known adverse consequences of excessive sleepiness during the day, literature is scarce on the topic. Substantial evidence has shed some light on the graded association of SEP and health problems [8,19,[25][26][27][28][29][30]. However, despite the numerous findings linking SEP with health, there is an enormous lack of research conducted on the relationship between SEP and sleep disturbances. Due to the high implication of SEP in health status and the public health issue that sleep disturbances represent [31], a better understanding of this relationship is necessary.
A systematic review seemed more appropriate to investigate this question, because the indicators needed to assess SEP, such as income, education or employment, are used differently depending on the context. Therefore, the heterogeneity of the results prevents a statistical combination of the findings necessary for a meta-analysis. The present systematic review was performed for various purposes. It has as an objective to: (1) describe the current literature on the relationship between EDS and SEP, and (2) provide recommendations for consideration of SEP in sleep medicine and biomedical research.

Literature Review
Using Prisma [32], articles included in this review were identified by screening in PubMed, Web of Science, Google scholar and Scopus. Keywords/Mesh terms used for SEP were a mix of the following: "socioeconomic", "socioeconomic position", "socioeconomic status", "social position", "social rank", ''social class" AND the following keywords/Mesh terms used for EDS: were "sleep", "sleep disorders", "sleep disturbances", "sleep complaints", "sleep duration", "sleepiness", "daytime sleepiness", "somnolence" "sleep quality", "sleepiness", "somnolence", "sleep" (Figure 1). Screening ranged from January 1990 to December 2020. Only peer review articles in English with an available pdf were considered. Studies conducted before 1990 were not considered for this review, since they were mostly empirical studies conducted on animals. Clinical studies examining sleep health began after 1990. This systematic review was not pre-registered.

Inclusion and Exclusion Criteria
Cross-sectional, retrospective and longitudinal studies which include individual SEP indicators such as education, income, occupation, employment status as well as perceived SES were considered, as shown in Figure 1. Full text articles were evaluated for inclusion by two independent reviewers and disagreements were resolved by discussion to reach a consensus. Proxy measures of SEP, such as composite measures and aggregate measures

Inclusion and Exclusion Criteria
Cross-sectional, retrospective and longitudinal studies which include individual SEP indicators such as education, income, occupation, employment status as well as perceived SES were considered, as shown in Figure 1. Full text articles were evaluated for inclusion by two independent reviewers and disagreements were resolved by discussion to reach a consensus. Proxy measures of SEP, such as composite measures and aggregate measures of SES were also included. Family SEP measures, such as parental education and household income were used when study participants were children or adolescents. Studies were excluded when they: (1) were not original research, (2) were case report/series, reviews or meta-analysis, (3) were not written in English, (4) did not have a fully accessible text or (5) included participants that already presented specific conditions at baseline (for example chronic disease, high mental health medication, pregnant women, etc.).

Data Extraction and Quality Rating of Studies
Eligible studies were screened to extract interrater variability measurements using Cohen's Kappa scores to assess diagnostic agreement. Interrater variability demonstrated good agreement (κ = 0.79; 95 % CI = 0.68-0.92) as shown in Table 1. Data extracted from included studies were: Country (where research took place), year, type of population, sample size, age (mean or range depending on studies), percentage of each gender in the sample, study design, settings, SEP measures, sleep measures, findings and statistical strength of the association SEP-EDS (Odds Ratio, p value and r) ( Table 2). SEP was considered the exposure variable and EDS the outcome variable. The National Institute of Health's Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies was used to rate the quality of included studies (Table 3). This is a tool that focuses on key factors for a critical evaluation of the internal validity of a study. SEP was considered the independent variable and sleep disturbances the dependent variable. Rating of the independent and the dependent variables of each study was performed by the two co-authors. Studies employing subjective and community SEP indicators, self-reported or non-diagnostic symptoms of EDS were downgraded. Blinding of outcome assessors was non-applicable in self-reported outcomes. Age, gender and race were considered essential confounders in statistical analyses, while body mass index or obesity were also required in studies assessing sleep-disordered breathing. Regarding overall quality rating, we calculated the proportion of positive rating on the sum of applicable criteria. Studies with <50% positive rating were judged as poor quality, ≥65% as good quality and the rest as fair quality, respectively.

•
Education: Lower education was related to increased perceived sleepiness [22]. Having lower education was associated with more daytime sleepiness [21]. Higher educational attainment was associated with less excessive daytime sleepiness [37]. Men with a low educational level were more likely to suffer from poor sleep quality and short sleep duration [35]. One study reported that higher level of maternal education was negatively related to EDS [20] and another one reported that EDS was not associated with educational status [40]. • Income: Lower income-to-needs ratio INR was related to increased sleepiness [27]. Lower INR was associated with greater daytime sleepiness [33,41]. Lower family income was associated with EDS [23]. Lower income was associated with difficulty falling asleep and having no private insurance was associated with sleep latency >30 min and non-restorative sleep [28]. The prevalence of EDS was lower in adults with a higher family income [7]. • Employment: Individuals employed full-time had more sleepiness than those employed part time or not employed [18]. Men with a low occupational position were more likely to suffer from poor sleep quality and short sleep duration [35]. Men with a low occupational position were also more likely to have long sleep latency [35]. Women with a low occupational position were more likely to have long sleep latency, excessive daytime sleepiness and short sleep duration [35]. One study reported that office workers had lower EDS scores than line workers [13], and another one reported that there is no association between EDS and employment status [7]. • Perceived/subjective SES: In children, higher subjective SES predicted less daytime sleepiness in adolescents [41], higher subjective SES was associated with fewer sleep disturbances [19]. Being in the highest deprivation quintile was a significant independent predictor of EDS [41], compared to being in the lowest deprivation quintile [24]. Lower household food security was associated with poor sleep quality, short sleep duration and excessive daytime sleepiness [36]. Subjective SES better predicted daytime sleepiness than objective SES [39]. Lower SES was also associated with more excessive daytime sleepiness [36]. One study reported that higher SES was associated with less daytime sleepiness [39] and another one reported that there was no association between SES and EDS [38].

Discussion
The aim of the present systematic review of observational cohort longitudinal and cross-sectional studies is to provide the current status of knowledge on the link between socioeconomic position (SEP) and excessive daytime sleepiness (EDS). Overall, a lower educational level, a lower income and full-time employment were associated with more excessive daytime sleepiness (Table 2). EDS symptoms are prevalent in women, especially those with a low income or no job [21]; and finally, children and adolescents with difficult living conditions or working part time reported more sleep disturbances [19,23,27].

•
Association between EDS and SEP in children and minors EDS was evaluated among young children and minors, a population in which development and growth rely on many physiological processes, including sleep. However, their sleep needs are often unmet. The actual average number of hours slept for minors was 6.7 h on weekdays and 7.4 on weekends [17,22,38,42]. EDS was inversely related to the number of hours of sleep and was associated with lower grade average and elevated scores of anxiety and depression [13,17,43]. These results, along with other studies, show that sleep disturbances are linked to poor academic performance, cognitive impairment and mental issues [37,43]. Disruptive sleep environment, conditions non-conducive to sleep or great worries could explain why children living in disadvantageous conditions have more sleep disturbances [21]. Moreover, the omnipresence of technological devices, the commotion of urban environment as well remunerated work influences their sleep duration and quality [13,43].

• Association between EDS and SEP in adults
Low SEP was strongly associated with poor sleep quality in adults. Important factors such as workload, varying shifts, family commitments or financial stress could account for the observed disparity in the burden of sleep disturbances on people with low SEP [35]. This association persists even after the adjustment for other sociodemographic, psychological or behavioral factors [35].
Moreover, EDS symptoms seem to be more frequent in women, especially in those with a lower family income or no job [21]. Women consider psychological factors to be the most important causes of sleep disturbances [20,44,45], therefore a low SEP could be perceived as a bigger burden by females rather than males. Furthermore, anxiety is significantly more common in females. However, psychological stress could be a consequence rather than a cause [44,46]. Women included in sleep health studies are often of the childbearing age, therefore, the higher number of sleep disturbance complaints could be due to the presence of young children [27,44,45].

•
Association between EDS and SEP in people with mental, physical and functional disability Excessive daytime sleepiness was also significantly associated with mental disorders, chronic physical conditions and greater functional disability among vulnerable populations [7]. Sleep disturbance is commonly observed in a wide range of psychiatric disorders reported by disadvantaged communities [17,45] and sleep disturbances increase negative emotions and decrease positive emotional responses among these communities more than their pairs [47].
EDS also seems to be more prevalent in low SEP individuals with chronic conditions than high SEP people with the same chronic conditions [48]. These chronic conditions such as diabetes, arthritis and obesity have been linked to poor quality of sleep [7,36,49], which in turn partly explains the high prevalence of sleep disturbances that have been observed in the elderly population [22,38,46].

•
Association between EDS and SEP in the elderly The lack of a significant association between sleep disturbances and SEP in geriatric research may be due to the presence of comorbidities. Multiple chronic diseases can cause bodily pain and can cause sleep disturbance independently of SEP. Reports of daily bodily pain account for 35% of complaints of sleep disturbance, including EDS [48].
Other factors that contribute to sleep problems in old age independently of SEP are changes in social engagement and lifestyle [49]. As their daily demands diminish and lifestyle becomes more sedentary, alteration in their sleep might not be as noticeable [50]. Furthermore, there are normal physiological alterations in sleep that arise with older age [51]. These changes include shortened nocturnal sleep duration as well as a higher frequency of daytime napping [51]. Although more studies are required to understand the association between sleep disturbances and SEP, the findings of this review point out a potential relationship that deserves more investigations. As previously mentioned, the older population often suffers from multiple diseases, therefore, those with no health insurance might be more worried about their medical expenses, which in turn might impair their sleep quality [51].

•
Recommendation for future sleep research Given the wide range of negative impacts associated with EDS, further research needs to be conducted to raise awareness among healthcare workers and improve clinical guidelines. Social class is a strong predictor of health outcomes and yet, it is a variable that remains particularly underexamined in sleep science. It is vital to gather more information about the contribution of other SEP parameters on EDS, such as access to insurance, ethnicity or immigration status. The temporality of the relationship between EDS and SEP needs to be further assessed in longitudinal studies, rather than cross-sectional, which can discern the direction of the interaction and allow clear causality interpretations on how SEP influences EDS. In order to implement and design clinical and public health initiatives to improve the overall population health and reduce the gap in health disparities, inclusion of objective measures of sleep, obtained by actigraphy or polysomnography should be included in future research [52,53]. Considering that sleep is an intricate and individual process that is heavily influenced by complex and dynamic interrelations across a lifetime, the development of a multivariable model that allows to consider and describe the various modifiable factors that take play in EDS would lead to better novel clinical interventions.

Conclusions
Modern medical care revolves mainly on a biomedical model where quantitative measurements are very often considered the gold standard for clinical practice as well as The recent shift in the societal paradigm accompanied by technological advances render our lifestyles incompatible with good sleep. The incessant need for productivity and the increasing hours of work have impacted the quality and quantity of sleep that affects mostly developed countries. Furthermore, the abundant availability of technological devices has also diminished the duration of sleep that affects people of all ages. The proliferation of screens and electronics increases our exposure to blue wavelengths, which in turn suppresses the secretion of melatonin. This phenomenon has been linked to diabetes, obesity, as well as depression.
On a smaller scale, SEP reflects the conditions of an individual's environment, such as their community, workspace and family. Sleep is an intricate process that is heavily affected by environmental cues; thus, a lower SEP is linked to more sleep disturbances, more worries and anxiety that all lead to poor sleep hygiene. Socioeconomic disparity creates a greater gap between different individuals, based on their age, sex or ethnicity.
Although sleep is impacted by the larger spheres of society and community, it is a personal physiological function. Therefore, the lifestyle, behavior and psychology of the individual is of the utmost importance. Behavioral risk factors such as smoking, drinking, drug use or eating disorders are linked with a lower SEP as well as with poor health outcomes (including sleep health). Disadvantageous living conditions impact sleep health directly and indirectly via mental health. The bidirectional association between sleep and mental health is significant and could account for a large part of sleep disorders.

Conclusions
Modern medical care revolves mainly on a biomedical model where quantitative measurements are very often considered the gold standard for clinical practice as well as public health administrators. However, this approach has limitations in assessing longterm health outcomes and their relationship with socioeconomic disparities. SEP is already considered as an important determinant for many health outcomes, but if SEP is embedded in experimental design in psychosomatic research, biomedical research and clinical practice as a constant variable regardless of outcome, it will move forward future investigations. It is a challenge to capture the multidimensional nature of SEP, which highlights the need for better measures that would be comparable among various populations and across different types of studies. Future research is still required to understand how socioeconomic status, social class and socioeconomic position influence sleep health while analyzing possible mediators and moderators. Achieving a sustainable and wide impact on sleep disparities is a challenging feat that requires a coordination of efforts of various systems to instill a perceivable change.