Organizational factors associated with work-related sleep problems in a nationally representative sample of Korean workers

Purpose The purpose of this study was to assess the association of organizational factors with work-related sleep problems (WRSP) among Korean workers. Methods The data were derived from the First Korean Working Conditions Survey conducted in 2006 with a representative sample of the Korean working population (n = 10,039). Results The overall prevalence of WRSP was 5.1 % (95 % confidence interval (CI) 4.7–5.5). Those who experienced sexual harassment at work (adjusted odds ratio (aOR) 3.47: 95 % CI 1.77–6.81), discrimination due to sex (aOR 2.44: 95 % CI 1.36–4.36) or age (aOR 2.22: 95 % CI 1.52–3.23), violence at work (aOR 1.98: 95 % CI 1.06–3.68), threat of violence (aOR 1.96: 95 % CI 1.05–3.66), poor work-life balance (aOR 1.78: 95 % CI 1.44–2.20), low job satisfaction (aOR 1.69: 95 % CI 1.37–2.09), high cognitive (OR 1.64: 95 % CI 1.32–2.03) and emotional (aOR 1.53: 95 % CI 1.22–1.91) demands, job insecurity (aOR 1.32: 95 % CI 1.07–1.63), and high work intensity (aOR 1.55: 95 % CI: 95 % CI 1.25–1.92) had an increased risk of WRSP compared to their respective counterparts (p < 0.01). Low social support was not significantly associated with WRSP (aOR 0.88: 95 % CI 0.67–1.15). Conclusion The results revealed that poor psychosocial working conditions may be related to a high prevalence of WRSP among representative Korean workers.


Introduction
Sleep problems have been one of the most commonly reported health complaints associated with a variety of physical and mental health outcomes (Ohayon 2002). According to a global estimate of sleep problems based on 10 different countries (n = 35,327), 31.6 % of individuals suffer from insomnia and 24.0 % report that they do not sleep well (Soldatos et al. 2005). In Korea, the prevalence of sleep problems ranges between 5.0 and 32.9 % (Cho et al. 2009;Kim et al. 2011;Nomura et al. 2010;Ohayon and Hong 2002), depending on the characteristics of the population sampled and the definition/case assessment. Sleep problems including insomnia, daytime sleepiness, and sleep apnea have been linked to a variety of occupational health issues; those employees with sleep problems reported an elevated risk of depression and anxiety disorders (Motohashi and Takano 1995;Nakata 2011b), alcohol abuse (Weissman et al. 1997), suicide (Goldstein et al. 2008) and chronic pain (Kuppermann et al. 1995) and are considered to be at high risk for hypertension (Murata et al. 2007;Yang et al. 2006) and coronary heart disease (Mallon et al. 2002). Sleep problems have a profound negative impact not only for individuals but also for the workplace and society as a whole. Consequences of sleep problems include reduced productivity (Nena et al. 2010;Rosekind et al. 2010), increased injuries at work (Kling et al. 2010;Lombardi et al. 2010;Nakata 2011a;Salminen et al. 2010;Vahtera et al. 2006), absenteeism (Akerstedt et al. 2010;Nakata et al. 2004b;Philip et al. 2001), and medical care expenditures (Leger and Bayon 2010;Metlaine et al. 2005). To date, a number of studies have examined the relationship between work organization factors and sleep problems; these studies have identified overtime work (Dahlgren et al. 2006), job dissatisfaction (Nakata et al. 2004a(Nakata et al. , 2007Scott and Judge 2006), overcommitment (Kudielka et al. 2004;Ota et al. 2005), effort-reward imbalance (Fahlen et al. 2006;Ota et al. 2005Ota et al. , 2009, low job control (Runeson et al. 2011), high job demands (Cahill and Landsbergis 1996;Kalimo et al. 2000;Knudsen et al. 2007;Nakata et al. 2007;Pelfrene et al. 2002;Runeson et al. 2011), role conflict (Knudsen et al. 2007), poor interpersonal relationships (Nakata et al. 2004a(Nakata et al. , 2007 job insecurity (Ferrie et al. 1998;Kim et al. 2011), workaholism (Kubota et al. 2010), and poor social support from colleagues/supervisors (Nakata et al. 2001(Nakata et al. , 2004aOta et al. 2009;Pelfrene et al. 2002;Runeson et al. 2011;Sinokki et al. 2010), as risk factors for sleep problems, although earlier studies have emphasized the negative impact of non-standard work schedules, that is, shift/night work, on sleep (Akerstedt et al. 2002;Estryn-Behar et al. 1990;Niedhammer et al. 1994). In addition, emerging workplace issues, that is, workplace bullying (Lallukka et al. 2011;Niedhammer et al. 2009;Takaki et al. 2010), violence at work (Eriksen et al. 2008), and occupational injustice (Elovainio et al. 2009;Kim et al. 2011), are found to be strongly related to sleep problems.
Although previous studies have suggested that work organization and the nature of work are associated with sleep problems, a few have drawn that conclusion based on representative samples of workers. The data from the National Employment Survey 2002-2003, a nationally representative random sample of 1,715 US full-time employees, indicated that work overload and repetitive work were associated with difficulty initiating and maintaining sleep while work overload and role conflict were related to non-restorative sleep (Knudsen et al. 2007). In addition, most previous studies used self-administered questionnaires rather than face-to-face interviews, which may introduce reporting/recall bias (Bowling 2005). Furthermore, little information is available about Korean workers; a study by Kim et al. (2011), which used a selfadministered questionnaire, reported that high job demands, insufficient job control, inadequate social support, job insecurity, organizational injustice, lack of reward, discomfort with the occupational climate, and overall job stress were related to a 13-45 % increased risk of insomnia (Kim et al. 2011).
Based on the above facts, continued effort is needed to explore the relationship between work organization factors and sleep problems. Therefore, this study was undertaken to investigate the relationship between work organization factors and sleep problems in a large nationally representative sample of Korean workers using data collected via face-to-face interviews.

Subjects and procedure
Data were derived from the First Korean Working Conditions Survey (KWCS), conducted in 2006 by the Korea Occupational Safety and Health Agency (KOSHA) (Park and Lee 2009). The survey population was a representative sample of the actively working population aged 15-65 years (in Korea, the legal work age is 15 years). 'Economically active' refers to subjects who were either employees or self-employed at the time of interview. Therefore, those who were retired, unemployed, housewives, or students were not included in the survey. The basic study design was a multistage random sampling of the enumeration districts used in the 2005 population and housing census (Park and Lee 2009). Data collection was performed by Gallup Korea during June 26 to September 26, 2006. A total of 46,498 households were visited, and 10,043 interviews were performed. A total of 36,515 households had dropped out of the interview. The number of households where a member of the household could not be interviewed after visiting 3 times was 14,680, while the number of households where a member of household was encountered but was not qualified to be a respondent was 2,671. The number of households without an employed person aged between 15 and 64 (non-qualified household) was 12,192, and the number of households that refused to take part was 6,972. We excluded workers who were under 18 (n = 4), which resulted in a final sample size of 10,039 respondents.
The survey weighting was carried out on the basis of the actively working population, which means that its distribution by age, sex, region, locality, size, economic activity, and occupation is identical to that of the active population distribution. Sociodemographic characteristics of the sample and total working population in Korea are shown in Table 1, suggesting that the distributions of the KWSC and the Korean total working population are comparable. The questionnaire contains questions about hours of work, physical risk factors, work organization, and the impact of work on health. The methodology and survey questionnaire of the first KWCS were based on that of the Fourth European Working Conditions Survey (EWCS) in 2005 (Parent-Thirion et al. 2006).

Sleep problems
Sleep problems in this study were assessed by the single item 'Do you currently suffer from work-related sleep problems (WRSP)?' which is identical to the question used in the EWCS. The response was either 'yes' or 'no.'

Work organization factors
Descriptions of work organization factors, response options, and response criteria are shown in Table 2. In all, 12 work organization variables were included in the questionnaire. The subjects were asked to answer 'yes' or 'no' about their experiences of discrimination regarding age and sex, sexual harassment, threat of violence, and violence at work during the past 12 months. Job insecurity, cognitive work demands, and emotional work demands were measured with a five-point scale. Job satisfaction and work-life balance were measured with a four-point scale. Social support at work and work intensity were measured by the sum of two items, both with five-point scales. The Cronbach's a for social support at work and for work intensity was 0.87 and 0.83, respectively.
According to the report provided by KOSHA (Park and Lee 2006), the test-retest reliability for the 1-month interval for the items 'working at very high speed,' 'working too tight deadlines,' and 'intellectually demanding work' had 60.1, 61.7, and 68.5 % consistency rates, respectively.

Statistical analyses
A series of univariate and multiple logistic regression analyses were conducted individually to examine the associations of organizational factors with sleep problems. All the work organization variables were dichotomized into two groups as suggested in Table 2. First, we tested the relationship between potential confounders and sleep problems with univariate analyses and then with forward stepwise multiple logistic regression analysis (p B 0.05 for inclusion and p C 0.10 for exclusion). Second, to examine the relationship between organizational factors and sleep problems, multiple logistic regression analyses were used and were incrementally adjusted for sex, age group, highest education level, and income (Model A); Model A variables ? smoking status, alcohol consumption, and presence of illnesses (Model B); Model B variables ? employment status, job type, employment contract, working hours, and work schedule (Model C). The significance level for all statistical analyses was p \ 0.05 (twotailed test).

Results
The characteristics of the study participants are shown in Table 3. There were 5,809 male and 4,230 female workers. The prevalence of sleep problems was 5.1 % (95 % CI: 4.7-5.5 %). Participants ranged in age from 18 to 65 (mean 42) years. More than one-third held a college degree or higher and 62 % earned a monthly income of 1-3 million Korean won. Overall, 32 % were current smokers, 13.9 % were former smokers, and more than 70 % were current alcohol drinkers. About a quarter of the workers reported one or more physical symptoms/ disorders, almost 30 % were self-employed or an employer, and 7.2 % of participants worked a shift/night schedule. The four dominant job types were professional/technical (19.1 %), clerical (14.0 %), service (12.4 %), and sales (11.4 %). More than half of the participants worked 45 h or more per week.
The covariates associated with sleep problems are shown in Table 4. The univariate logistic regression analyses revealed that male gender, older age (C55), current smoking, higher alcohol consumption, presence of illness, job type, long working hours (C45 h/week), and shift/night work were significant factors associated with sleep problems. In the stepwise multivariate logistic regression analyses, male gender, age group, presence of illness, job type, and shift/night work schedule remained significant.
The relationships between psychosocial work characteristics and sleep problems are shown in Table 5. Univariate logistic regression analyses showed that all 12 organizational variables were significantly associated with a 25-525 % increased prevalence of sleep problems. After controlling for covariates, social support at work did not remain significant, but the rest of the 11 variables remained significant.

Discussion
The purpose of this study was to investigate the relationship between work organization factors and WRSP in a Rarely, almost never large representative sample of Korean workers. There were three key findings from this study. First, organizational factors related to violence, discrimination, work-life imbalance, job dissatisfaction, high work demands and intensity, and job insecurity were associated with an increased prevalence of WRSP even after adjusting for an array of potential confounders. Second, male gender, age group, presence of illness, and shift/night work were background risk factors associated with high WRSP prevalence. Third, the overall prevalence of WRSP was 5.1 % in this population. Although the results must be interpreted with caution because of the cross-sectional nature of the study design, the analyses of this large population-based representative survey suggest that work organization factors are important risk factors for WRSP among Korean workers.
Those who experienced sexual harassment at work had a 3.5 times higher risk of WRSP compared to those who had not experienced sexual harassment at work. Although we could not locate studies specifically focused on a relationship between sexual harassment and workers' sleep problems, several studies have reported the relationship between sexual harassment and workers' physical and mental health. A study on female flight attendants showed that for those who experienced sexual harassment, the risk of poor self-rated health was 2.8 times higher than for those who had not had such an experience (Ballard et al. 2006). There are also reports that sexual harassment heightens the risk of depression, somatic symptoms, posttraumatic stress disorder (PTSD), and other medical conditions , which could relate to sleep problems. Sexual harassment also raises the risk of the victims' harmful alcohol use (Gradus et al. 2008). Given such evidence, workers who experienced sexual harassment may have an increased risk for suffering sleep problems.
This study found that the participants who perceived sex-and age-related discrimination had more than twice   the risk of WRSP than those workers who did not. Discrimination is a crucial social issue not only in multiethnic nations such as the United States but also in non-multiethnic nations as well. In the United States, the occurrence of perceived discrimination over one's lifetime is 33.5 %, but the prevalence differs greatly by racial/ethnic group; for non-Hispanic whites, it is 30.9 %, for non-Hispanic blacks, 48.9 %, and for other racial/ethnic groups, 50.2 % (Kessler et al. 1999). The results of the 1977-1989 US Longitudinal Survey of Mature Women (n = 1,778) indicated that perceived workplace discrimination ranged between 11.11 and 15.14 % in black women, while it ranged between 12.10 and 16.03 % in white women. Workplace discrimination was found to be one of the strongest predictors for emotional distress and functional limitation (Pavalko et al. 2003). In the current study, the occurrence of age and sex discrimination at the workplace was 3.4 and 1.4 %, respectively, which was lower than those of studies conducted in the United States (Kessler et al. 1999;Pavalko et al. 2003), but the impact on sleep seems substantial. Taking these findings into consideration, workplace discrimination may have a long-lasting negative influence on sleep behavior, possibly resulting in poor mental health of affected workers. Additionally, in this study, those who experienced violence at their work sites were twice as likely to suffer from sleep problems as those who did not. A study of Nurses' aides revealed that those who had been exposed to threats or violence at work had a 19 % increased risk of poor sleep compared to those without such exposures (Eriksen et al. 2008). With fear acting as a mediator, the experience of violence is known to adversely affect workers' health both mentally and physically (Rogers and Kelloway 1997). Even when an individual is not a direct victim of violence, being a witness to a threatening act has been reported to exert negative effects (anxiety, illness symptoms, and negative occupational outcomes) (Hall and Spector 1991). The result of this study corresponds with the notion and that workers who are exposed to threats of violence had an equivalent risk of sleep problems as those who actually had undergone violence at work.
Work-life imbalance has become an emerging issue in Korea because of an increase in working hours (Park et al. 2010). Work-family imbalance has been reported to be a risk factor for depression (Frone et al. 1996), reduced wellbeing (Grant-Vallone and Donaldson 2001), exhaustion (Demerouti et al. 2004), and alcohol abuse (Wang et al. 2010). The work-life interface has also been reported to be related to sleep. Those who had difficulties combining work and private life had increased odds for sleep disorders (men adjusted OR 1.54, 95 % CI 1.12-2.10 and women adjusted OR 1.81, 95 % CI 1.31-2.49) (Hammig and Bauer 2009). Another study in medical residents showed that work-family conflict was associated with sleep deprivation (Geurts et al. 1999). Our study found that work-life imbalance is related to increased sleep problems in Korean workers as well.
Job satisfaction has been consistently associated with sleep problems in earlier studies (Doi et al. 2003;Kuppermann et al. 1995;Nakata et al. 2004aNakata et al. , 2007Nakata et al. , 2008Scott and Judge 2006). The results of our study are in line with these findings. For example, Scott and Judge (2006) reported that insomnia is positively related to job dissatisfaction and this relationship is mediated by hostility, joviality, and attentiveness in US administrative employees (Scott and Judge 2006). Doi et al. (2003) found that job dissatisfaction is the second major factor for poor sleep quality, which resulted in a twofold increase in the prevalence of disturbed sleep among white-collar employees in Japan (Doi et al. 2003). Another study in Japan revealed that low job satisfaction created a significantly increased risk for insomnia including difficulty maintaining sleep (DMS) after adjusting for multiple confounding factors (Nakata et al. 2004a). Our study, together with those from other countries, indicates that job dissatisfaction is a risk factor associated with sleep problems. High cognitive and emotional demands, as well as high work intensity, increased the risk of sleep problems. A significant association between cognitive demands and difficulty initiating sleep (DIS) was found in male whitecollar daytime workers in Japan (Nakata et al. 2004a). Urponen et al. (1988) also reported that mental workload was one of the most important factors that interfered with falling asleep (Urponen et al. 1988). In terms of work intensity, there is consensus that high job demands are related to insomnia (Cahill and Landsbergis 1996;Kalimo et al. 2000;Pelfrene et al. 2002). Excessive mental/cognitive demands and working too hard may disturb the ability to fall asleep, which in turn may impair the quality of sleep.
In our study, social support at work was not associated with sleep problems after adjusting for confounding factors. Although the majority of published studies (Cahill and Landsbergis 1996;Eriksen et al. 2008;Jansson and Linton 2006;Kageyama et al. 1998;Kim et al. 2011;Nakata et al. 2001Nakata et al. , 2007Nordin et al. 2005;Pelfrene et al. 2002;Runeson et al. 2011) indicate that poor social support at work is related to sleep problems, some studies suggest that the statistical significance of this relationship is attenuated after controlling for confounders (Nakata et al. 2004a(Nakata et al. , 2006(Nakata et al. , 2008. This finding may be relevant to the fact that social support often exerts a buffering effect on health outcomes and that the significant relationship disappears if controlled for related variables. However, it is important to note that social support from one's workplace is often more protective than social support from family or friends, suggesting the importance of workplace social support (Nakata et al. 2001(Nakata et al. , 2004a. A significant association between job insecurity and sleep problems was found in this study. After the 1998 financial crisis in East Asia, Korea was no exception with regard to increased job insecurity. At the time of the crisis, a large number of workers lost their jobs and since then businesses have not been active in recruiting permanent employees (preferring temporary employees), and employers are facing organizational restructuring over time. Workers who feel their jobs are insecure may succumb to sleep disorders resulting in long-term mental stress. A study of civil servants in Britain reported that male workers who experienced organizational change tended to have increased sleep problems (Ferrie et al. 1998). Another Swedish study discovered that workers who expected that they would lose their jobs experienced sleep disturbances (Mattiasson et al. 1990). The results of this study support the notion that job insecurity is connected to sleep problems.
The overall prevalence of WRSP in this study was 5.1 %, which was comparable to that of 8.7 % in the fourth EWCS (Table 3). The sleep problems question used in both the KWCS and the EWCS was targeted specifically to work-related sleep problems. A study in Sweden (Swedish Work Environment Survey; SWES) used a similar method to define sleep disturbances as both the KWCS and EWCS and showed a strong predictability of medically certified sickness absence (Westerlund et al. 2008). In comparison with earlier studies that defined sleep problems in general, the definition used in the KWCS, EWCS, and SWES might have a stronger predictive validity than merely asking about general sleep problems because general sleep problems may also capture problems related to or caused by non-work-related issues. However, it is also true that the significant associations found in this study are subject to the 'triviality trap'; that is the measurement of the independent (WRSP) and dependent (organization factors) variables is conceptually overlapping and the observed associations may be spurious (Kristensen 1996). Thus, future studies should be undertaken to validate our finding by using objective sleep measures in a prospective study design.
The analyses of underlying factors associated with WRSP revealed that men had a 1.5 times higher odds of WRSP than women (Table 4). In studies investigating sex differences in sleep problems, the majority of studies discovered that sleep problems are more frequent in women than in men (Chen et al. 2005;Kim et al. 2011;Paparrigopoulos et al. 2010). However, in this study, as the definition of sleep problems was 'work-related,' it may be that working men in Korea have more sleep problems due to work than working women do. In the EWCS, the prevalence of sleep problems in men was 8.9 %, while it was 8.5 % in women. Thus, it is likely that the higher prevalence of sleep problems in men than in women may depend on how 'sleep problems' are defined. As suggested in Table 4, the higher prevalence of WRSP in workers with illness and working the shift/night schedule is in line with previous findings, indicating that the association was in the expected direction.

Strengths and limitations of the study
The specific strengths of this study are that: (a) the sample was both nationally representative of the Korean working population and was large in size, (b) the study measured a number of work organization factors, (c) the analyses controlled for a broad array of potential confounders related to work organization and sleep problems, and d) the survey measures were collected via face-to-face interviews resulting in very little missing data. A major criticism of the methodology of the present study is that we evaluated WRSP with a single question, which prevented us from judging the severity of sleep problems and did not allow us to compare our results with other studies that used more general questions. Moreover, the definition of WRSP may include not only those with general sleep problems, that is, insomnia, poor sleep quality, and sleep loss, but also those with more specific sleep disorders, that is, sleep apnea, excessive daytime sleepiness, severe bruxism, etc. We also acknowledge other potential limitations. First, the study is cross-sectional in nature; thus, no causal interpretations can be made. However, it may be speculated that sleep problems affect the rating of work conditions; workers with sleep problems may have issues with irritability with colleagues and supervisors, an inability to concentrate at work, difficulty accomplishing assigned tasks in a timely manner, and uncertainty that they will be able to continue their employment, leading to expressions of higher work stress (Nakata et al. 2007). Meanwhile, poor working conditions may influence sleep problems. A two-year prospective study of the effort-reward imbalance model, the job demand-control model, and insomnia revealed that those who were not insomniac at the baseline became insomniac when exposed to high overcommitment to work (OR 1.75, p \ 0.05) and high job strain (OR 1.72, p \ 0.05) (Ota et al. 2009). Second, most of the work organization measures consisted of single item that may raise questions as to the validity and reliability of the results. However, items such as 'job satisfaction' are known to hold as high a reliability as multi-item scales (Wanous et al. 1997). Third, even though we have statistically controlled for existing disorders, it is possible that those who are suffering from sleep problems may be affected by comorbid disorders.

Conclusions
This study found a significant relationship between a broad range of work organization characteristics and sleep problems, which has been understudied in representative samples of workers. Although a prospective study with objective sleep measures is warranted to prevent the 'triviality trap,' the present finding that work organization factors are related to sleep problems may be useful in developing strategies to prevent sleep problems in the Korean working population.