Background: The objectives of this study are to identify family and job characteristics associated with long work hours, to analyse the relationship between long work hours and several health indicators, and to examine whether gender differences for both objectives exist.
Methods: The sample was composed of all salaried workers aged 16–64 years (3950 men and 3153 women) interviewed in the 2006 Catalonian Health Survey. Weekly work hours were categorised as less than 30 h (part-time), 30–40 (reference category), 41–50 and 51–60 h. Multiple logistic regression models separated by sex were fitted.
Results: Factors associated with long working hours differed by gender. Among men, extended work hours were related with being married or cohabiting and with being separated or divorced. In men, working 51–60 h a week was consistently associated with poor mental health status (aOR 2.06, 95% CI 1.31 to 3.24), self-reported hypertension (aOR 1.60, 95% CI 1.12 to 2.29), job dissatisfaction (aOR 2.05, 95% CI 1.49 to 2.82), smoking (aOR 1.33, 95% CI 1.03 to 1.72), shortage of sleep (aOR 1.42, 95% CI 1.09 to 1.85) and no leisure-time physical activity (aOR 2.43, 95% CI 1.64 to 3.60). Moreover, a gradient from standard working hours to 51–60 h a week was found for these six outcomes. Among women it was only related to smoking and to shortage of sleep.
Conclusion: The association of overtime with different health indicators among men could be explained by their role as the family breadwinner: in situations of family financial stress men work overtime in order to increase the income and/or accept poor working conditions for fear of job loss, one of them being long working hours.
Statistics from Altmetric.com
Although in recent years interest in health problems related to long working hours has intensified, the number of studies concerning the different areas of health is still low and the results are often contradictory.1–4 The inconsistent findings can be related to several methodological shortcomings of previous research, including small sample sizes or unique industry-specific circumstances5–9 that limit generalisability of the findings.
Most studies have analysed samples composed of self-employed and salaried workers, but the reasons and the compensations associated with working long hours may differ between these two groups, and therefore the impact of long working hours on health status. It is noteworthy that most epidemiological studies on long work hours and health have been conducted in Japan and have focused on very long working hours but little is known about moderately long work hours.10 Some studies that have failed to find an association between poor health and long working hours have considered working hours as a continuous variable and assumed a linear relationship.11 12 However, individuals working few hours are more likely to do so due to a reduction of their usual work hours as a consequence of serious health conditions. Therefore, these studies could have underestimated the association between poor health status and long working hours. It has been pointed out that further studies are needed that take into account some potential confounding or interacting variables such as gender, work characteristics or domestic work.1 3
Most studies on the health impact of long work hours have been carried out among men or have considered sex as an adjusting variable. Nevertheless, there are several reasons why the impact of long work hours can differ by gender.1 On one hand, physiological changes related to long work hours may not be the same for men and women.13 Moreover, long work hours can contribute to the double burden of full-time and domestic work. Several studies have reported the damaging effects of work conflict and work overload among women who have to balance the responsibilities of job and family.14–16 In addition, more noticeable effects of long work hours have been observed among women, something often attributed to the additional domestic stressors experienced by women as a result of gender inequalities in the distribution of domestic tasks.17–19 Working long hours depends to a great extent on family responsibilities and constraints imposed by the labour market. Owing to both gender segregation within the labour market and family responsibilities, men and women probably work long hours for different reasons; hence the work conditions associated with their schedules also differ. Consistent with the gender division in the domestic sphere, with women responsible for housework and caring tasks and men usually assuming the breadwinner role, living with children is related to part-time work among women, while, among men, it is likely to be associated with long hours.20
Regulation on working hours differs across countries. The European Union (EU) work hours directive has included a maximum of 48 h of work per week.17 Yet, there are differences among European countries. For example, Spanish law is more restrictive and establishes a maximum of 40 contractual work hours and a maximum of 80 h of overtime per person and year.21 However, it has been pointed out that overtime regulation, not only in Spain but in the EU, is undermined by illegal practices, as well as the widespread use of special rules and exemptions for certain sectors, types of company, and occupational categories.22 The Japanese government has implemented administrative guidance on overtime, which suggests that all employees should not work 45 h of overtime per month and that there will be government intervention if overtime exceeds 100 h per month.23 In the USA there is no regulation limiting working time for adults.
The objectives of this study were (a) to identify family and job characteristics associated with long work hours; (b) to examine the relationship between long work hours and a variety of health outcomes and health-related behaviours; and (c) to examine whether gender differences exist for both objectives.
The data have been taken from the 2006 Catalonian Health Survey, a cross-sectional survey based on a representative sample of the non-institutionalised population of Catalonia, a region in the north-east of Spain that has about 7 million inhabitants. The survey included self-reported information on morbidity, health status, health-related behaviours and use of healthcare services, as well as sociodemographic data. A sample of 18 126 subjects was selected using a multiple-stage random sampling strategy with a maximum global error of ±0.7%. Trained interviewers administered the questionnaires at home in a face-to-face interview. Only 27% of respondents were replaced as a result of refusal or absence.24 For the purposes of this study a subsample of all salaried workers with a contract aged 16–64 was selected. In order to avoid the effect of extremely long work hours, people working more than 60 h a week were excluded (2.1% of men and 0.7% of women). The final sample under analysis was composed of 3950 men and 3153 women.
Working hours were determined using the question: “In the last week (or the last whole week you worked) how many hours did you work?” The response alternatives were <30 h (part-time job), 30–40 h (reference category), 41–50 h and 51–60 h a week.
Health and well-being outcomes
Self-perceived health status was elicited by asking the respondents to describe their general health as “excellent”, “very good”, “good”, “fair” or “poor”. This is a broad indicator of health-related well-being25 and has also proved to be a good predictor of mortality, even better than the medical diagnostic.26 The variable was dichotomised by combining the categories “fair” and “poor” to indicate perceived health as poor health status and the categories “excellent”, “very good” and “good” to indicate good health status. Mental health status was measured with the 12-item version of the General Health Questionnaire. We used a two-point scoring method, rating a problem as absent (0) or present (1). The responses were summed, and the participants scoring 3 or more were classified as cases.27 Hypertension was determined by asking people whether they suffered from hypertension or had been diagnosed by a healthcare professional as having hypertension. Self-reported hypertension has been found to agree with reports in medical records28 29 and with personal blood pressure measurements.30 Job satisfaction was measured with the question “All in all, how satisfied would you say you are with your job?”. There were four categories, from “very satisfied” to “very dissatisfied”. The responses were dichotomised by combining the categories “very satisfied” and “satisfied” to indicate satisfaction and the categories “dissatisfied” and “very dissatisfied” to indicate job dissatisfaction.
Smoking behaviour was categorised as “no” (current non-smokers and smokers of fewer than one cigarette a day) or “yes”. For leisure-time activities, a dichotomous variable was created, comparing no physical activity with slight, moderate or high intensity of sport activities. The daily number of hours slept was categorised as ⩽6 h and >6 h.
Occupational social class, assigned according to the respondent’s current occupation, was measured with a widely used Spanish adaptation of the British classification.31 Because of the small number of people in some categories, the six original social classes were collapsed into the following three broad classes: I and II (more privileged classes), III and IVA–IVB–V (manual workers). Type of contract had three categories: Permanent, fixed-term temporary contract and non-fixed term temporary contract. Shift type was coded as day, night, shifts and variable depending on the day.
Family and domestic characteristics
Marital status was classified into four categories: married or cohabiting, single, separated or divorced, and widowed. The number of children living at home was coded into three categories: none, one and two or more. Weekly hours of domestic work were recorded as <10 h, 10–20 h and >20 h.
First, gender differences for all of the dependent and independent variables were tested at the bivariate level using the χ2 test for categorical variables and the t test for age. Second, factors associated with working 41–50 or 51–60 h a week compared with working 30–40 h were identified by fitting multiple logistic regression models for the men and women separately and including age, and job and family characteristics. Third, multiple logistic regression models separated by gender and adjusted for age, and job and family characteristics were fitted in order to test the association between long work hours and all of the health indicators.
General description of the population
Table 1 shows a general description of the population. Regarding health outcomes, no gender differences were found for job satisfaction, women were more likely to report poor self-perceived health status and poor mental health status, whereas self-reported hypertension was more frequent among men. Smoking prevalence was higher among men but no gender differences were observed for leisure time physical activity or daily number of hours slept.
Women were more likely to work part-time, whereas working 41–50 h or 51–60 h was more frequent among men: 30.1% of men reported working 41–50 h a week and 7.6% 51–60 h. The corresponding proportions for women were 14.5% and 2.3% respectively. Conversely, the number of hours of domestic work was higher among women. Whereas 27.5% of women worked more than 20 h a week, the proportion among men was 7.0%.
Factors related to long work hours
Table 2 shows factors associated with working 41–50 h compared with working 30–40 h by gender. In both sexes, long work hours were more frequent among those with variable shifts depending on the day and was negatively associated with age. Additionally, among women, it was positively associated with night shifts and negatively related to non-fixed-term temporary contracts and to more than 20 h of domestic work a week. Although among women no association was found between long work hours and family roles, men married or cohabiting (aOR 1.42, 95% CI 1.13 to 1.77), separated or divorced (aOR 2.19, 95% CI 1.38 to 3.46) or having two or more children (aOR 1.25, 95% CI 1.02 to 1.55) were more likely to work 41–50 h.
Table 3 shows factors associated with working 51–60 h compared with working 30–40 h, by gender. In both sexes, an association with variable shifts depending on the day was found. Additionally, among men, working 51–60 h a week was also related to being married or cohabiting (aOR 1.55, 95% CI 1.05 to 2.29) and being separated or divorced (aOR 2.33, 95% CI 1.09 to 4.97). No association with number of children living at home was found in either sex.
Association between long work hours and health
Table 4 shows the association between health outcomes and working hours. The relationship of long work hours with health and health-related behaviours was much more consistent among men. In males, working 51–60 h a week was associated with six out of the seven outcomes analysed. Moreover, a gradient from standard working hours to 51–60 h a week was found for the six outcomes.
In both sexes, working 41–50 h a week was associated with job dissatisfaction (aOR 1.52, 95% CI 1.23 to 1.87 for men and aOR 1.40, 95% CI 1.06 to 1.85 for women). Additionally, among men it was related to being sedentary during leisure time (aOR 1.73, 95% CI 1.31 to 2.27) and to sleeping 6 h or less a day (aOR 1.30, 95% CI 1.11 to 1.52).
In both sexes, people working 51–60 h a week were more likely to smoke (aOR 1.33, 95% CI 1.03 to 1.72 for men and aOR 2.27, 95% CI 1.39 to 3.70 for women) and to sleep 6 h or less a day (aOR 1.42, 95% CI 1.09 to 1.85 for men and aOR 2.21, 95% CI 1.34 to 3.63) for women. Additionally, men working 51–60 h a week were more likely to report poor mental health status (aOR 2.06, 95% CI 1.31 to 3.24), self-reported hypertension (aOR 1.60, 95% CI 1.12 to 2.29), job dissatisfaction (aOR 2.05, 95% CI 1.49 to 2.82) and no leisure-time physical activity (aOR 2.43, 95% CI 1.64 to 3.60).
This study of representative data from a Spanish region adds to the growing body of evidence indicating that long working hours can substantially increase the risk of several poor health outcomes. Unlike other studies, our investigation has the advantage of covering a large variety of jobs, and overcoming many limitations of earlier research, for example by using a large sample and controlling for the potential confounding effect of age and a variety of job and family characteristics as well as by adopting a gender perspective.
The present study produced three main findings. First, factors associated with long work hours differed by gender. Second, among men there was a consistent association of long working hours with six of the seven health indicators analysed. Third, a notable result of our analysis was the detection of a dose–response relationship among men, from a standard number of hours to 51–60 h a week for these six health indicators.
As has been mentioned before, results of studies about long working hours and health status are contradictory. This is the case for the seven indicators analysed in this study. For example, whereas some authors have found a positive relationship between long working hours and poor general health status,32 others have not.5 33 Some studies have reported a positive association with depression,5 but others have not observed any relationship.34 Findings about hypertension are also contradictory. Whereas some studies have reported a positive relationship,33 35 others have found no association11 12 and, even, a negative relationship has also been observed.36 As in the present study, earlier research has found an association between long working hours and job dissatisfaction.33 37 Although some studies have observed a positive relationship between long working hours and smoking,33 38 this has not been found in other studies.7 39 40 Although a positive association has been reported between extended hours and lack of physical activity during leisure time,33 other studies have not observed any relationship.7 39 41 Finally, shortage of sleep is the most consistent finding in the relationship between long working hours and health status.4 33 36 42 Consistently, in our study, working 51–60 h was associated with sleeping 6 h or less a day in both sexes.
Therefore, the consistent association of extended work hours with a variety of health indicators among men is the main finding of this study. Gender differences in the association between long work hours and health outcomes is an interesting observation that can provide some clues for understanding the mechanisms that underlie the relationship between long working hours and health.
Whereas among women, working 41–50 h was only associated with one health outcome, among men it was associated with three. Moreover, working 51–60 h was positively related to six out of the seven health outcomes analysed among men but with only two among women. Additionally, a gradient was found among men. It can be argued that the more consistent association of working 51–60 h a week with health outcomes among men could be explained by an insufficient statistical power among women due to the low number of women working this schedule. However, when new logistic regression models were fitted with a single category of long work hours, working 41–60 h a week was associated with all the outcomes except self-perceived health status among men but only with job dissatisfaction among women (results not shown).
It has been suggested that the relationship between work hours and ill health could be mediated by stress in that long hours act both directly as a stressor increasing the demands on those who attempt to maintain performance levels in the face of increasing fatigue and indirectly by increasing the time that a worker is exposed to other sources of workplace stress.2 43 It has also been suggested that long working hours can be one indicator of a global poor psychosocial work environment among vulnerable workers.33 Finally, the importance of choice in determining a person’s response to long hours has also been pointed out.33 37 44 45
Therefore, there are several factors that can mediate in the relationship between long working hours and health status, some of them having a positive effect on health and others being negative. The results of our study among men suggest that there should be a strong and consistent factor that explains such a consistent association between long work hours and health status, which could be marital status. In this study, compared with men working 30–40 h a week, those working 41–50 h and 51–60 h were consistently more likely to be married or cohabiting and separated or divorced, whereas no association with marital status was found among women. This finding suggests a possible relationship between long working hours and family financial stress among men, since they traditionally assume the role of family breadwinner. The obligatory nature of working long hours because of family financial stress among people who have the breadwinner role could explain the relationship between long working hours and poor health outcomes found in the subsample of men. Pressure to work long hours in order to increase income and/or acceptance of poor working conditions, one of them being long working hours, due to fear of job loss in a situation of economic vulnerability, could explain the consistent pattern of association of long working hours with several health outcomes among men (fig. 1).
Our interpretation about overtime related to the need of increasing the income among men is supported by a study based on the 5th Spanish Working Conditions Survey, which reported that receiving economic compensation for overtime was significantly more frequent among men (60.1% vs 34.4%).46 This hypothesis is also supported by the consistent association between working long hours and job dissatisfaction and the gradient found among men, whereas among women the association existed only for those working 41–50 h. This interpretation about the importance of the mandatory nature of long working hours is also consistent with the study of Van der Hulst et al47 that concluded that even a limited number of hours of involuntary overtime is associated with adverse mental health in low reward situations. Nakanishi et al36 reported a negative association between long working hours and hypertension among male white collar workers. In that study, architects or research workers did more overtime than clerks, and being an architect or a research worker (vs being a clerk) was negatively associated with blood pressure. The authors related this finding to the non-mandatory nature of long working hours in this highly qualified group. As architects or research workers who work overtime might be especially competitive, or may particularly enjoy their work, they may not feel working long hours as job strain or stress.40
Our findings are consistent with those of Artazcoz et al,33 who observed a positive relationship between long working hours and a variety of health outcomes only among women, among whom working long hours was associated with being separated or divorced, while among men there was no relationship with marital status. They are also consistent with a study showing a relationship between financial stress and risk of acute myocardial infarction among men but not among women.48 It should be taken into account that between 2002 and 2005, in Spain, financial stress among families has significantly increased. For example, whereas in 2002 the percentage of household units with a ratio between debt and income higher than 3 was 3.4%, in 2005 it was 9.5%.49 In Spain financial stress is closely related to the high increase in housing prices over the last decade. This situation is likely to be common in other developed countries. According to a study of 17 industrialised economies between 1970 and 2003,50 most of the countries experienced a house price boom after the mid-1990s. During this period, housing price growth was particularly strong in Ireland, the Netherlands and the UK followed closely by Australia, Spain and a number of Nordic countries, where the pace of growth has accelerated in more recent years. Additionally, it should be taken into account that Spain has the highest home ownership rate in Europe. Unfortunately, studies about financial stress and health are still scarce.
Self-perceived health status was the only health indicator not related to long working hours either in men or women. It seems that long work hours would be primarily related to psychosocial health conditions rather than with pain or physical limitations which would be likely to prevent them from working long hours. Actually, whereas no differences in the prevalence of long-standing limiting illness by working hours were found among women, among men the prevalence of long-standing limiting illness was 14.8% among those working part-time, 8.1% among those working 30–40 h a week and 7.4% among those working 41–50 or 51–60 h a week.
This study may be limited by its cross-sectional design. However, reverse causation, whereby long working hours would not lead to poor health status and health-related behaviours, but rather the opposite—in other words, that people with poor health status and health-related behaviours would be more likely to work more than 40 a week—does not seem plausible. Data about long-standing limiting illness support this hypothesis, whereas no differences in the prevalence of sickness leave longer than 3 months by working hours were found among women (n = 99); among men (n = 53) the prevalence decreased with increasing working hours (4.5% among men working part-time, 1.4% among those working 30–40 h, 0.9% among those working 41–50 h and 0.3% among those working 51–60 h). Therefore, health problems associated with long working hours in this study are primarily health conditions and behaviour that have often been related to stress or to lack of time for oneself.
It can be argued that there should be differences in the association between long working hours and health status by social class. Yet, multiple logistic regressions models additionally separated for manual and non-manual workers were fitted and the patterns of associations were similar (results not shown). It should be taken into account that between 2001 and 2005, the proportion of Spanish families with a housing debt increased from 21.6% to 26.1%. However, there was no such increase in households in the lowest two deciles of income. Moreover, the percentage of families with this kind of debt was higher among families with higher income.49 These facts could explain the lack of interaction between long working hours and social class.
Although we have hypothesised that gender differences in the relationship between long working hours and poor health outcomes can be related to the mandatory nature of working long hours related to family financial stress among men, and have provided many arguments supporting this hypothesis, no specific questions about these issues were collected in the survey. Further research should be carried out in order to confirm this hypothesis.
As far as we know, this is the first study to report such a consistent association between moderately long work hours and a variety of health indicators and health-related behaviours. Competition from countries with low wages and low social costs seems at present to influence working hours to start to increase again.17 However, our findings suggest that even moderately long work hours may act as a risk factor for health, mainly when it is involuntary; for example, because of family financial stress among breadwinners as seems to be the case in this study. Accordingly, policies most fruitful for improving workers’ health should focus on reducing long working hours when it is mandatory and harmful.
What is already known on this subject
In the last years several studies have found an association between long working hours and different health indicators.
However, the number of studies concerning the different areas of health is still low and the results are often contradictory.
The inconsistent findings can be related to several methodological shortcomings of previous research, including small sample sizes or unique industry specific circumstances.
Moreover, most studies have focused on very long working hours but little is known about moderately long work hours.
What this study adds
As far as we know, this is the first study based on a large representative sample of the working population that shows such a consistent association between moderately long working hours and a variety of health indicators.
A notable result of our analysis is the detection of a systematic dose–response relationship between long working hours and health and health-related behaviours among men from an standard number of hours to 51–60 h a week.
The consistent association of long working hours with six out of seven health indicators among men could be explained by their role as family breadwinners that in situations of family financial stress can be related to pressure to work long hours in order to increase the income and/or acceptance of poor working conditions, one of them being long working hours, due to fear of job loss.
Competing interests: None declared.
Funding: This work was partially supported by the CIBER en Epidemiología y Salud Pública (CIBERESP), Spain