Towards A Socioeconomic Model of Sleep Health among the Canadian Population: A Systematic Review of the Relationship between Age, Income, Employment, Education, Social Class, Socioeconomic Status and Sleep Disparities

A better understanding of the contribution of the socioeconomic status (SES) in sleep health could guide the development of population-based interventions aiming to reduce “the silent public health issue” that are sleep disturbances. PRISMA was employed to identify relevant studies having examined the association between social class, social capital, education, income/assets, occupation/employment status, neighborhood deprivation and sleep health. Sixteen cross-sectional and three longitudinal studies were selected, having sampled 226,029 participants aged from 2 months to 85 years old. Findings showed that: (1) sleep health disparities among children and adolescent are strongly correlated to parental socioeconomic indicators; (2) poor parental income, poor family SES and poor parental education are associated with higher sleep disturbances among children and adolescents; (3) lower education is a predictor of increased sleep disturbances for adults; (4) low SES is associated with high sleep disturbances in adults and old people and; (5) low income and full-time employment was significantly associated with short sleep among adults and old people. In conclusion, sleep health should be an important public health target. Such intervention would be beneficial for populational health, for all taxpayers and public administrations, which would see a reduction in absenteeism and productivity losses attributable to sleep-related health problems in the global economy.


Introduction
Recently, disparities in development and wealth have continued to grow between G20 countries and other countries across hemispheres and continents. Over past decades, the literature has documented associations between social inequalities and several cardiovascular diseases, some chronic diseases, and even neurological diseases such as sleep disorders [1]. Researchers from different backgrounds report the existence of a social gradient involved in health and variations in socioeconomic indicators, such as income, education, employment and location [2]. In short, the higher one rises in the social hierarchy, the better the health status and the lower the mortality risk. However, the association between populational sleep health and socioeconomic status (SES) remains poorly documented, despite the recent increasing importance that sleep disturbances represent for public health authorities. A Canadian study published in December 2018, conducted between 2007 and 2015 among 21,826 respondents, found a 42% increase in insomnia symptoms among Canadians aged 18 years and older [3]. occupation * OR employment OR *employed OR asset * AND sleep * OR insomnia * OR circadian OR parasomnia * OR "restless leg *" OR "periodic leg movement *" AND Canada * OR Canadian * OR Quebec * OR Ontario * OR British Columbia * OR Saskatchewan * OR Alberta * OR Manitoba * OR Nova Scotia * OR New Foundland * OR New Brunswick *. The search period included studies published from January 1990 to December 2020.

Inclusion Criteria
Observational studies of any design (cross-sectional, prospective or retrospective cohort) that evaluated human subjects of any age, gender or race/ethnicity from the general population were kept. Multiple objective SES measures such as education, income, occupation, employment status, assets, composite scores, subjective SES, neighborhood or residence area, deprivation indexes and self-reported perceived SES were considered. For studies examining children or adolescents, parental SES indicators such as parental education, parental occupation and household income were used instead. Objective sleep parameters, such as wrist actigraphy, accelerometry or polysomnography (PSG), and subjective reports about sleep duration, sleep quality and any symptoms of sleep complaints were all considered.

Exclusion Criteria
Studies were excluded for the following reasons: (1) studies were interventional trials, reviews or meta-analyses, case series or case reports and did not present original research; (2) studies were not in English or French; (3) the full text was not accessible; (4) researchers recruited participants that already presented serious health issues at baseline (i.e., cancer, diabetes, neurodegenerative diseases, mental disorders or other chronic diseases); (5) results reported univariate associations and unadjusted estimates of the variables of interest and; (6) studies were not performed in Canada, or with Canadian participants.

Inclusion Criteria
Observational studies of any design (cross-sectional, prospective or retrospective cohort) that evaluated human subjects of any age, gender or race/ethnicity from the general population were kept. Multiple objective SES measures such as education, income, occupation, employment status, assets, composite scores, subjective SES, neighborhood or residence area, deprivation indexes and self-reported perceived SES were considered. For studies examining children or adolescents, parental SES indicators such as parental education, parental occupation and household income were used instead. Objective sleep parameters, such as wrist actigraphy, accelerometry or polysomnography (PSG), and subjective reports about sleep duration, sleep quality and any symptoms of sleep complaints were all considered.

Exclusion Criteria
Studies were excluded for the following reasons: (1) studies were interventional trials, reviews or meta-analyses, case series or case reports and did not present original research; (2) studies were not in English or French; (3) the full text was not accessible; (4) researchers recruited participants that already presented serious health issues at baseline (i.e., cancer, diabetes, neurodegenerative diseases, mental disorders or other chronic diseases); (5) results reported univariate associations and unadjusted estimates of the variables of interest and; (6) studies were not performed in Canada, or with Canadian participants.

Data Extraction, Quality Assessment and Synthesis of Results
The following information was extracted from included studies: epidemiological study type, type of population, percentage of each gender, age (mean age or age range), sample size, independent variables (IV) and their measurement, sleep variables and their measurement, results and conclusions. 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 [25]. SES was considered as the exposure, and sleep measures as the outcome variable. Studies employing subjective area SES measures, or self-reported sleep disturbances, were downgraded. Blinding of outcome assessors was non-applicable in self-reported outcomes. Studies with <50% positive rating were judged as "poor quality", those with ≥65% as "good quality" and the rest as "fair quality". The overall quality rating score was calculated with the proportion of positive rating on the sum of the 14 questions composing the tool and results, and are displayed in Table 1. Eligible studies were screened to extract interrater variability measurements using Cohen's Kappa scores to assess diagnostic agreement. Interrater variability demonstrated good agreement as shown in Table S1 (Supplementary Materials).

Children and Adolescents Sleep Health Disparities
Education: Higher subjective SES predicted less daytime sleepiness, longer self-reported sleep duration, better sleep quality and shorter parent-reported sleep duration [42]. Sufficient sleep was more likely among students attending schools in areas classified in the highest SES group [36]. Lower SES predicted sleep rhythmic movements in children [38].
Income: Higher household income predicted longer parent-reported sleep duration and fewer sleep disturbances [42]. Toddlers who came from a household with a higher annual income were less likely to sleep <11 h per night [35]. Compared with children without obstructive sleep apnea (OSA), those with OSA were more likely to reside in disadvantaged neighborhoods [26]. Attending schools in low-income areas predicted short and low-normal sleep duration trajectories over time [36]. Among pre-schoolers, low household income was significantly associated with short sleep [37].
Subjective SES and social class: Newborns of mothers without a university degree had significantly reduced sleep duration, compared to their peers who had mothers with a university degree [43].

Adults Sleep Health Disparities
Education: Overall, lower education is a predictor of increased sleep disturbances. One longitudinal investigation showed that, over a period of 10 years, prevalence of insomnia symptoms in Canada increased from 15.6% to 17.1% between 2002 and 2012, representing an absolute increase of 1.5% [28]. This increase in insomnia symptoms was significantly influenced by education [28]. Lower education was associated with more frequent insomnia symptoms [39]. People with less education were more likely to report insomnia [30]. Having some secondary (high school) education, or less, was associated with the presence of insomnia [29]. Like insomnia, a lower education level was also a risk factor of possible REM sleep behavior disorder (RBD) [40]. Finally, there is one study reporting that, among adults and older adults, less than secondary (high school) school education was significantly associated with both long sleep and short sleep [37], while another study reported that education was not significantly associated with sleep duration [44].
Income and employment: Overall, income and employment are strongly associated with sleep disturbances. A lower income was associated with the presence of insomnia [29]. People with lower income were more likely to report insomnia [30]. Higher annual income was associated with less sleep disturbances [33], while not enough money left over at the end of the month increased the risk of excessive daytime sleepiness (EDS) [32]. One study found that income was not significantly associated with sleep duration [44]. Not being in the work force was associated with the presence of insomnia [29]. Among older adults, full-time employment was significantly associated with short sleep, while unemployed older adults were more likely to sleep longer [37].
Subjective SES and social class: Overall, a low SES is associated with high sleep disturbances. A low socioeconomic status was associated with the presence of insomnia [29]. Higher SES was associated with better sleep quality, shorter sleep latency, longer sleep duration, shorter weekend oversleeps and less daytime sleepiness [41]. Subjective SES better predicted sleep duration, weekend oversleep and daytime sleepiness than objective SES, while objective SES better predicted sleep quality and latency [41]. Network capital increased the likelihood of restless sleep in men, but not women [31]. High generalized trust decreased the odds of restless sleep in women, while neighborhood disadvantages increased the odds of restless sleep in women, but not men [31]. Regardless of social class or SES, association between restless sleep, social capital and neighborhood environmental factors differed between males and females [31]. Parents with higher social capital tended to have children with fewer sleep disturbances, than parents with lower social capital [34]. One longitudinal study showed an association between incidence of subjective EDS and having a house in need of repairs [27].

Summary of Findings
A summarized overview of the literature shows that sleep health disparities among children and adolescents are strongly associated to parental socioeconomic indicators. Overall, poor parental income, poor family SES and poor parental education are associated with higher sleep disturbances among children and adolescents. Similar trends exist in adults' sleep health where: (1) lower education is a predictor of increased sleep disturbances; (2) low SES is associated with high sleep disturbances and; (3) low income and full-time employment was significantly associated with short sleep. The majority of included studies [29][30][31][32][33][34][35][36][37][38][39][40] were rated as "fair quality", while two longitudinal studies [27,38] and the single study using PSG to assess sleep health [26] were rated as "good quality".
An overview of the current findings confirms an underlying relationship between SES and sleep health disparities. Regardless of the sleep disturbances investigated, socioeconomic inequities appear in the sleep health of the general population, similar to what has been published in health disparities related to cardiovascular diseases, mental disorders and several chronic diseases [6,7,45,46]. While epidemiological data available on sleep medicine focused on a variety of parameters, such as sleep quality, sleep duration or sleep efficiency; the concept of ''Sleep health" for public health is more mitigated because it is more multidimensional [31,44,[47][48][49][50].

The Socioeconomical Model of Sleep Health
A detailed analysis of literature confirms a relationship between individual socioeconomic status and sleep health disparities. Regardless of which sleep disturbances are studied, socioeconomic disparities appear in the sleep health of the general Canadian population similar to what has been published in health disparities related to cardiovascular diseases, mental disorders and several chronic diseases; both in Canada and in other countries. While epidemiological data available on sleep medicine focused on a variety of parameters such as sleep quality, sleep duration or sleep efficiency; the concept of ''Sleep health'' for public health is more mitigated because it is more multidimensional and relatively new. Figure 2 presents a socioeconomical model of sleep health with multiple interactions at both individual, community and governmental levels. It will allow a mapping of socioeconomic and biobehavioral determinants, global patterns and public health

The Socioeconomical Model of Sleep Health
A detailed analysis of literature confirms a relationship between individual socioeconomic status and sleep health disparities. Regardless of which sleep disturbances are studied, socioeconomic disparities appear in the sleep health of the general Canadian population similar to what has been published in health disparities related to cardiovascular diseases, mental disorders and several chronic diseases; both in Canada and in other countries. While epidemiological data available on sleep medicine focused on a variety of parameters such as sleep quality, sleep duration or sleep efficiency; the concept of "Sleep health" for public health is more mitigated because it is more multidimensional and rel-atively new. Figure 2 presents a socioeconomical model of sleep health with multiple interactions at both individual, community and governmental levels. It will allow a mapping of socioeconomic and biobehavioral determinants, global patterns and public health trends related to sleep health in Canada, as well as other countries.

The Economic Policy
Worldwide, the shift between our lifestyles and the solar cycle is reaching epidemic proportions. It would seem that modern life is increasingly incompatible with adequate/sufficient sleep [51]. It is, therefore, no coincidence that the countries most affected by the "silent epidemic of sleep disorders" are either G20 countries (e.g., India and South Africa) or fast-growing countries (e.g., Indonesia) [48]. The economic costs caused by insufficient sleep are significant, especially in G7 countries: the United States suffers by far the biggest economic losses (up to US$411 billion per year) due to the size of its economy, followed by Japan (up to US$138 billion per year) [51]. On an annual basis, the United States loses the equivalent of about 1.23 million working days due to insufficient sleep; followed by Japan, with an average of 604,000 working days per year; the United Kingdom and Germany with 207,000 and 209,000 days, respectively; and finally, Canada, with about 78,000 working days lost [51]. Sleep thus seems incompatible with the endless capitalist imperatives of productivity and consumption that impose longer working days, at the expense of mental health and sleep. The mechanisms for this forced interference of the western lifestyle in health status are numerous: night work, longer commute times due to the urban sprawl, new wireless technologies and more recently, teleworking. All these environmental factors have a direct or indirect association with "sleep disorders", knowing that a sleep of less than 6 h per night affects all the functions of an organism (cellular rest, protein synthesis, mood regulation, memory capacity) [52][53][54][55].
On the other hand, political authorities, in coordination with public health managers, employers' associations and trade unions should redouble their efforts to align workers' conditions with their mental health [5]. Not only would such an intervention be beneficial for the sleep of individuals, and thus for the overall health of the population, but it could also be beneficial to companies and public administrations, which would see a reduction in absenteeism and productivity losses attributable to sleep disturbances in their employees [56]. Indirectly, all taxpayers would save money due to the reduction of sleep-related health problems.

The Individual, Family and Community SES
SES acts as a marker of the conditions of various environmental factors, such as the neighborhood [8], the social environment [6,57], work [46,58] and access to material resources [8,21,59,60]. Associations between socioenvironmental factors and multiple health outcomes are well established, suggesting a positive association between the SES and exposure to health-promoting socioenvironmental factors [7,[61][62][63][64][65].
Recent meta-analyses [66][67][68] and systematic reviews [69,70] show that the relation between SES indicators (such as education, income and employment) and sleep health may be modified by age, gender and ethnicity; but it can also be mediated by factors related to living conditions, behaviour and stress [8,18]. Age, sex/gender and race/ethnicity are the main moderators of the relation between SES and sleep health, with higher rates of self-reported sleep disturbances among non-Caucasian relative to Caucasian individuals [6,22,23,71] and an increase in sleep disturbances among old people compared to young adults and children [15,23,[72][73][74]. Black young adults reported shorter sleep duration as well as Hispanic young adults, and both ethnic group members slept less than their young Caucasian peers. Thus, these three moderators could be involved in a triple interaction [22,57,74,75].

The Living Conditions and Behavioral Risk Factors
The relation between living conditions and behavioral risk factors contributes to sleep health outcomes and influences chronic stress [38,72,76]. An irregular bedtime schedule due to working conditions (like rotative shift work and night shift work) and lack of public transportation, affect the circadian rhythm and sleep duration, leading to sleep disturbances like poor sleep quality and sleepiness [77][78][79][80]. Lifestyle factors such as smoking, drug use or unhealthy eating (the prevalence of which is associated with SES), are also associated with higher odds of being affected by sleep disturbances and an increase in both physical and chronic ailments [80,81]. Certain characteristics of a neighborhood, such as accessibility to green spaces, shops, presence of sufficient lighting or proximity to conveniences stores (restaurant, pharmacy, delivery services) and healthcare services are determinants of important healthy behaviors, including regular sleeping hygiene [41,[82][83][84]. Inadequate living conditions were found to be associated with mental disorders and chronic stress in several studies [4,23,61,64,74]. The chronic stress itself is associated with several health outcomes such as cardiovascular disease [16,63,85], mental health disorders [58,61,86,87] and sleep health [73,86].

Chronic Stress and Social Stress
Regarding chronic stress, its consequences are numerous and the links to social stress through living conditions, influence of SES and its moderators such as ethnicity and sex/gender, are manifold [6,8]. Neighborhood characteristics have been associated with high levels of stress among individuals living in noisy, unsafe neighborhoods, or ones with poor access to public transportation, parks or grocery stores [6,8,15,73]. A work environment with changing or irregular schedules contributes to the social isolation of the individual, while stressing the body with a disruption of the biological clock [5]. Psychological stress from work impacts the individual's mental well-being and exposes them to mental health problems [5,46]. A family's material deprivation can lead to chronic stress (i.e., late bills, poor quality housing), leading to further health problems [6,8,61]. Exposure to this social stress over time leads to chronic stress and subsequently to an increase in the allostatic load (AL) [88,89]. AL represents the physiological burden the body experiences when repeated responses to stressors are activated constantly, or for an unusually long period [85]. This chronic stress is associated with neuropsychiatric disorders such as affective and mood disorders, which are both associated with sleep health [90,91]. Chronic stress can adversely affect sleep quality and sleep duration, while insufficient sleep can increase stress levels [61,86]. In addition, both stress (chronic and social) and lack of sleep can lead to lasting physical and mental health problems [86]. Considering previous studies and reviews reporting on associations between SES and chronic stress [8,87], and other studies linking chronic stress and sleep health [23,61,92,93], the present conceptual socioeconomic model assumed that SES influences stress through living conditions, and this stress in turn influences sleep health as well as other physical and mental health disorders [15,57]. While acutely adaptive, brain reactivity and body resilience associated with chronic stress are sharply influenced by individual differences in behavioral risk factors [61] and living conditions [4,65,[94][95][96].

Sleep Disturbances
This term includes: (1) subjective sleep complaints reported by participants themselves, or measured with validated or customized questionnaires/items such as the Pittsburgh Sleep Quality Index or Insomnia Severity Index and; (2) objective "sleep disorders" confirmed with a quantitative measure of sleep complaints, such as actigraphy and PSG, or confirmed by a physician's diagnosis or a certified sleep specialist. Sleep disturbances are associated with anxiety, depression and stress [86]; most of these diseases (including sleep disturbances) are prevalent among low SES individuals [18,42,61,86,87,95,[97][98][99]. Knowing that chronic stress, social stress and AL are well known determinants of mental disorders [6,23] and difficult living conditions induced chronic stress [6,18,23], there is no doubt that sleep disturbances are consequences of these interactions [18,42,61,86,87,95,[97][98][99].

Implications for Public Health Policy
Living conditions, behavioral risk factors and health status can all be found in the pathway from SES to sleep health inequalities ( Figure 2). As such, clinical guidelines should consider them in the prevention and management of "sleep disorders", due to their multidimensional impacts on sleep. Since adopting and maintaining good sleep hygiene is a key non-pharmacological strategy for improving the mental health and general well-being of individuals (along with modification of other behavioral risk factors such as quitting smoking or engaging in physical activity), clinicians should, therefore, tailor their interventions to the specific needs of their patients. Clinicians should consider the SES of individuals as a marker of increased risk for sleep health disparities, but should also be aware that the negative impact of SES on risk of sleep disturbances varies according to individual characteristics (like age, gender/sex, ethnicity/race). Many studies reported lower rates of neuropsychiatric disorders and other chronic comorbidities among good sleepers, compared to people suffering from sleep disturbances [16,63,92,93,100], suggesting that good sleep quality and appropriate sleep duration may protect against some mental disorders [16,63,92,93,100]. At the same time, several studies showed that a majority of good sleepers are people with high SES under good living conditions [1,21,23,59,74,98,[101][102][103][104] and mental disorders (in addition to sleep disturbances) are very often reported among vulnerable populations with low SES [16,23,105]. This evidence together reveals that the protective factor that sleep can have on health is unequally distributed among different social classes and contributes to social inequalities in sleep health [23,59,73,105]. Thus, consideration of SES in sleep health management also applies to public health decision-makers and governments, who can take action through targeted interventions that support the development of healthy sleep environments.

Implications for Future Research
Future research should be careful in the use of the terms SES and social position. The vast majority of studies in sleep medicine unfortunately seem to equate social position with living conditions, leading to possible confusion as to what is really impacting health in general, and sleep health in particular. The present conceptual socioeconomic model clarifies the distinction between these two concepts and provides tips on which is more suitable for the research question. Researchers from biomedical fields, and applied sciences in particular, should think deeply about how to measure and integrate social inequalities in health into their experimental designs. Sleep health seems to be a common factor in many health outcomes [67,68,106] and it would be important to design new instruments or to update current questionnaires of self-reported sleep outcomes to include, as standard basic items, questions or variables related to SES, social position and living conditions.

Current Limitations and Challenges
The first challenge is the diversity of SES measures in the literature. The variety of definitions and conceptualizations of SES leads to a substantial heterogeneity in the measurements of SES. Nevertheless, considering that SES is an emerging concept, such diversity contributes to validating the role of SES in sleep health disparities, as multiple studies reported results pointing in the same direction. The second important challenge is how sleep health is measured. Several studies assessed sleep health with a single question, an approach that cannot capture the multi-dimensionality and day-to-day variability of sleep, and may be unable to detect the accurate effect SES has on sleep health. Furthermore, indicators were obtained for most of the studies through participants' self-report, which could mean that recall bias might have altered the accuracy of the findings. Effects of SES variations on objective sleep measures, like actigraphy and PSG, are still unknown. Investigating the association SES-sleep health has with objective measurement will certainly bring more knowledge to the entire scientific community and decision-makers, and it will also help to update this socioeconomic model of sleep health. Finally, another important challenge is the experimental setting of the study. The cross-sectional design employed by most studies limits and restricts conclusions about causality. It seems redundant to say, however, a reminder is of utmost importance. Longitudinal studies have the advantage to follow changes through time, which is a game changer for advancement of knowledge, especially with a complex biological function like sleep, in which physiological and behavioral parameters change a lot from youth to adulthood.

Conclusions
Sleep has been studied intensively in the past two decades and remains a subject of interest, overlapping several disciplines such as psychology, neurology, pneumology and psychiatry [107]. Incidence and prevalence of sleep disturbances have increased globally over the past decade [30,[108][109][110][111][112][113], while their determinants, as well as associated psychopathological mechanisms, remain not well understood. More than any other external factors, the interaction between SES and sleep health must be better understood. Future work should identify how factors related to living conditions and lifestyle/habits act on the incidence and progression of sleep health disparities, not only in the Canadian population, but worldwide. This socioeconomic model of sleep makes it possible to identify potentially vulnerable populations (i.e., low-income people, children living in disadvantaged families, individuals with limited education, people living in impoverished neighborhoods) for whom specific interventions could be developed. Public health interventions aimed at improving living conditions and reducing social inequalities are likely to contribute to the improvement of sleep health, helping at the same time in reducing chronic and social stress of low SES populations. Some examples of efficient interventions would include the reduction of rotative shift work, an increase in availability and accessibility to neighborhood green spaces and a massive promotion of good sleep health.