Working hours and all-cause mortality in relation to the EU Working Time Directive: a Danish cohort study

Working hours and all-cause mortality in relation to the EU Working Time Directive: a Danish... Abstract Background In keeping with the need to protect the safety and health of workers, the EU Working Time Directive stipulates that a worker’s average working time for each 7-day period, including overtime, does not exceed 48 h. It has, however, not been settled whether or not the threshold at 48 working hours a week is low enough to protect against excess mortality from long work weeks. The aim of the present study was to examine all-cause mortality in relation to weekly working hours among employees in the general population of Denmark. A special attention was given to mortality rates among employees with moderately long work weeks, 41–48 h. Methods Interview data from cohorts of 20–64 year-old employees were drawn from the Danish Labour Force Survey. The participants (N = 159 933) were followed through national registers from the end of the calendar year of the interview (1999–2013) until the end of 2014. Rate ratios (RRs) for all-cause mortality were estimated as a function of weekly working hours while controlling for age, sex, social class, night-time work and calendar year. Results We found 3374 deaths during an average follow-up time of 7.7 years. With 32–40 working hours a week as reference, the RRs for all-cause mortality were 0.75 (95% CI: 0.66–0.85) for 41–48 and 0.92 (0.80–1.05) for >48 h. Conclusion Mortality rates in Denmark are significantly lower among employees with moderately long work weeks than they are among full-time employees without overtime work. Introduction It has often been suggested that long working hours may be associated with poor sleep, insufficient restitution between work shifts and a subsequent increased susceptibility to accidents as well as mental and physical health problems.1 It is therefore stipulated in the EU Working Time Directive (EUWTD) that Member States shall take the measures necessary to ensure that a worker’s average working time for each 7-day period, including overtime, does not exceed 48 h.2 The purpose of the directive, which was issued in 1993, with slight amendments in the years 2000 and 2003, is to protect the safety and health of workers. If the intent of the directive is fulfilled then the 48 h limit should be enough to protect against adverse health effects from long weekly working hours. It should, however, be noted that we are dealing with an arbitrary cut-point. The assumption that moderate overtime work, 41–48 h a week, does not constitute a public health problem, has not been sufficiently tested. A recent study by O'Reilly and Rosato3 examined the relationship between weekly working hours and premature death (the ultimate consequence of a poor safety and health situation) in a cohort of 20–64 year-old people from Northern Ireland (144 938 women and 270 011 men) who worked at least 35 h a week. During an 8.7 years follow-up period, they observed 1143 deaths among the women and 4447 deaths among the men. Among the women, the age, marital status and socioeconomic status (SES) adjusted all-cause mortality rate ratio (RR) was 0.99 (95% CI: 0.80–1.22) for employees working 41–48 h, 1.18 (0.88–1.58) for employees working 49–54 h and 0.86 (0.63–1.18) for employees working >54 h a week, when compared with employees working 35–40 h a week. Among the men, the corresponding RRs were 0.96 (95% CI: 0.88–1.04), 1.01 (0.91–1.12) and 0.96 (0.87–1.05). The concerned RRs were also estimated within four different SES groups and only one instance of a significantly elevated RR was found, namely among male employees who worked more than 54 h a week in routine occupations. Apart from the above, only a few small studies have dealt with all-cause mortality as a function of overtime work4–6 and none of these used exposure definitions that are compatible with the cut-point of EUWTD. There are, however, some large and interesting studies which suggest that long weekly working hours are associated with an increased risk of atrial fibrillation,7 coronary heart disease and stroke.8 It has moreover, been suggested or implied that various associations between long working hours and health or safety depends on gender,9 SES8 and night-time work.2 The results by O’Reilly and Rosato3 suggest that a threshold at 48 h a week would afford more than ample protection against excess mortality from long weekly working hours. However, O'Reilly and Rosato did not regard hours worked outside of the person’s main job. Long working hours due to one or more secondary jobs could thereby be misclassified as normal working hours (35–40 h a week), which would bias the estimated RRs toward unity. To rectify this shortcoming, the present study would examine the relationship between weekly working hours and all-cause mortality among employees in the general working population of Denmark, while taking hours worked in secondary jobs into account. Since long working hours also are associated with an increased income, which in turn has been associated with a decreased risk of stroke,10 ischaemic heart disease11 and mental health problems,12 there are theoretical arguments both for positive and negative associations between long working hours and all-cause mortality. Aims and hypotheses We wanted to know if all-cause mortality rates are independent of weekly working hours among full-time employees in Denmark, and we addressed this research question in a series of nested hypothesis tests. The following null-hypotheses were tested: All-cause mortality rates among full-time employees in Denmark are independent of weekly working hours as well as interaction between weekly working hours and SES, sex and night-time work, respectively There are no interaction effects between weekly working hours and SES. There are no interaction effects between weekly working hours and sex. There are no interaction effects between weekly working hours and night-time work. The all-cause mortality rates are independent of weekly working hours when we disregard interaction effects. The overall significance level was set at 0.05 and the multiple testing problem was solved by the following strategy: The first level null-hypothesis would be rejected if the P-value of its statistical test was less than or equal to 0.05. A null-hypothesis at the second level would be rejected if (i) the first-level null-hypothesis was rejected and (ii) the P-value of its statistical test was less than or equal to 0.05. Methods The statistical analysis was performed in accordance with a study protocol,13 which we published before we looked at any relations between the exposure and outcome data of the study. Methodological details of the study protocol will be repeated in the present method section. Data material The data of the present study were obtained through a person-based linkage between the Danish Labour Force Survey (LFS) 1999–2013, the central person register and the employment classification module. The Central Person Register contains, inter alia, information on sex and dates of birth, death, and migrations for every person who is or has been an inhabitant of Denmark sometime between 1968 and the present.14 The Employment Classification Module, which has existed since 1975 and covers all inhabitants of Denmark, contains person-based annual information on SES, occupation, and industry.15 LFS is based on quarterly random samples of 15- to 74-year-old people in the Danish population. Each participant is invited to be interviewed four times over a period of one and a half years. Structured telephone interviews are used to gather person-based information on, inter alia, usual weekly working hours, calculated by adding the hours worked in secondary jobs to the ones worked in a primary job, and night-time work (‘Yes, regularly’, ‘Yes, occasionally’, ‘No’).16 The response rate has decreased with time, from 70% in 2002–53% in 2013. The questions used to gather the information on working hours have changed slightly with time. Before 2001, the questionnaires did not specify whether or not meal breaks should be included in the number of working hours. In 2001–06, it was specified that all meal breaks should be excluded and from 2007, it was specified that meal breaks should be included if the person got paid while eating and excluded otherwise. Follow-up and inclusion criteria Participants who were between 20 and 64 years old at the start of the follow-up period and employed with 32–100 weekly working hours at the time of the interview were eligible for inclusion. The included participants were followed from the beginning of the calendar year that succeeds that of their baseline interview. The follow-up ended at the time they emigrated or died, or the study period ended (31 December 2014), whichever came first. A flow-chart for the inclusion and exclusion procedure is given in figure 1. In total, we included 159 933 participants (46% were women), and we found 3374 deaths in 1 237 999 person-years at risk (mean follow-up 7.7 years). Figure 1 View largeDownload slide Flow-chart for the inclusion and exclusion criteria. Figure 1 View largeDownload slide Flow-chart for the inclusion and exclusion criteria. Primary statistical analysis Poisson regression was used to analyse all-cause mortality rates as a function of weekly working hours (32–40, 41–48 or >48 h a week), night-time work (‘Yes, regularly’ or ‘Yes, occasionally’ vs. ‘No’), sex, age (10-year classes), calendar time (2000–04, 2005–09 or 2010–14), time passed since start of follow-up (0–4, 5–9 or ≥10 years) and SES (low, medium, high or unknown). SES was based on the participant’s main occupation during the calendar year of the interview and coded in accordance with the three-class version of the European Socioeconomic Classification (ESeC). The coding procedure is described by Hannerz et al.17 Age, calendar time and time passed since start of follow-up were treated as dynamic (time-varying) variables. The remaining variables were fixed at baseline (the calendar year of the interview). The logarithm of person-years at risk was used as offset and the maximum likelihood method was used to estimate the parameters. People who participated in more than one interview were classified in accordance with the responses given in their first interview. The full model contained the following covariates: calendar time, time passed since start of follow-up, age, sex, SES, night-time work, working hours, working hours × sex, working hours × SES and working hours × night-time work. Hypothesis 1 was tested by use of a likelihood ratio which compared the full model to a sub-model in which the main as well as the interaction effects of working hours were excluded. Hypotheses 1.1–1.3 was tested by use of likelihood ratios which compared the full model to sub-models in which the respective interaction terms were excluded. Hypothesis 1.4 was tested by use of a likelihood ratio which compared a model which contained all main effects of the full model to a sub-model in which the main effect of working hours was excluded. Parameter estimates were used to calculate RRs for all-cause mortality, with 95% confidence intervals, as a function of weekly working hours with and without stratification by sex, SES, and night-time work, respectively. In keeping with a set of previous studies,18–21 we defined fulltime work as 32 or more hours a week. The following contrasts were considered: 41–48 vs. 32–40 working hours a week and >48 vs. 32–40 working hours a week. Sensitivity analysis 1—only stable exposure To find out if the strength of the association between the examined work time arrangements and all-cause mortality increases when exposure is more stable over time, we conducted a sensitivity analysis, in which we only included people who (i) participated in more than one interview, (ii) were between 20 and 64 years old during their last interview, (iii) were employed according to their first as well as their last interview and (iv) did not move more than one step among the ordered working time categories (<32; 32–40; 41–48; >48 h a week) between the first and last interview. The included participants were thereafter re-categorized according to the mean of the reported usual working hours during their first and last interview. The follow-up of the sensitivity analysis started on January 1 of the calendar year which succeeded the calendar year of the participant’s last interview. Only main effects were regarded. In all other respects, the statistical models and covariates were the same as in the primary analysis. Sensitivity analysis 2—stratification by interview calendar period Since the questions used to obtain information about weekly working hours were slightly revised in 2001 and then again in 2007, we performed a sensitivity analysis with the results stratified by calendar period of interview (1999–2000, 2001–06 and 2007–13). Only main effects were regarded. The statistical model, covariates and inclusion criteria were otherwise the same as in the primary analysis. Results The main null hypothesis which stated that all-cause mortality rates among full-time employees in Denmark are independent of weekly working hours as well as interaction between weekly working hours and SES, sex and night-time work, respectively, was rejected by the likelihood ratio test (P = 0.0004). No significant interactions with working hours were found for SES (P = 0.86), sex (P = 0.10) or night-time work (P = 0.10). The sub-model, which only included main effects, yielded a P values for the effect of weekly working hours that was less than 0.0001, and the estimated RRs among employees with 41–48 and >48 weekly working hours were lower than those among the reference group (32–40 h). The estimated RRs from the primary statistical analysis are given in table 1, with and without stratification by sex, SES and night shift status, respectively. The RR for all-cause mortality among the employees with overtime work within the limits of EUWTD (41–48 h) was lower than unity in each of the examined sub-populations. They were also lower than unity in each of the sensitivity analyses (table 2). Table 1 RR with 95% confidence interval for all-cause mortality, as a function of weekly working hours among Danish employees 2000–14, with and without stratification by sex, SES and night shift status, respectively Population Weekly working hours Persons Person years Cases RR 95% CI All workersa >48 9497 78 543 228 0.92 0.80–1.05 41–48 15 643 133 667 275 0.75 0.66–0.85 32–40 134 793 1 025 789 2871 1.00 — Male workersb >48 7344 61 904 195 0.89 0.76–1.03 41–48 9707 83 932 191 0.69 0.59–0.80 32–40 68 961 517 105 1952 1.00 — Female workersb >48 2153 16 639 33 1.06 0.75–1.50 41–48 5936 49 735 84 0.93 0.75–1.17 32–40 65 832 508 684 919 1.00 — Workers with a high SESc >48 3587 29 357 83 0.87 0.69–1.10 41–48 5687 48 124 99 0.77 0.62–0.95 32–40 38 026 257 384 584 1.00 — Workers with a medium SESc >48 1250 11 180 27 0.87 0.59–1.29 41–48 2421 22 043 44 0.83 0.61–1.14 32–40 25 976 203 686 442 1.00 — Workers with a low SESc >48 2891 24 481 79 1.03 0.82–1.29 41–48 5780 49 255 107 0.76 0.62–0.92 32–40 57 155 464 401 1540 1.00 — Workers with unknown SESc >48 1769 13 525 39 0.84 0.60–1.17 41–48 1755 14 246 25 0.58 0.38–0.87 32–40 13 636 100 318 305 1.00 — Workers with night-time workd >48 2801 23 499 83 1.09 0.86–1.39 41–48 2216 18 933 32 0.58 0.41–0.84 32–40 15 060 117 152 349 1.00 — Workers without night-time workd >48 6696 55 045 145 0.84 0.71–1.00 41–48 13 427 114 734 243 0.78 0.68–0.89 32–40 119 733 908 638 2522 1.00 — Population Weekly working hours Persons Person years Cases RR 95% CI All workersa >48 9497 78 543 228 0.92 0.80–1.05 41–48 15 643 133 667 275 0.75 0.66–0.85 32–40 134 793 1 025 789 2871 1.00 — Male workersb >48 7344 61 904 195 0.89 0.76–1.03 41–48 9707 83 932 191 0.69 0.59–0.80 32–40 68 961 517 105 1952 1.00 — Female workersb >48 2153 16 639 33 1.06 0.75–1.50 41–48 5936 49 735 84 0.93 0.75–1.17 32–40 65 832 508 684 919 1.00 — Workers with a high SESc >48 3587 29 357 83 0.87 0.69–1.10 41–48 5687 48 124 99 0.77 0.62–0.95 32–40 38 026 257 384 584 1.00 — Workers with a medium SESc >48 1250 11 180 27 0.87 0.59–1.29 41–48 2421 22 043 44 0.83 0.61–1.14 32–40 25 976 203 686 442 1.00 — Workers with a low SESc >48 2891 24 481 79 1.03 0.82–1.29 41–48 5780 49 255 107 0.76 0.62–0.92 32–40 57 155 464 401 1540 1.00 — Workers with unknown SESc >48 1769 13 525 39 0.84 0.60–1.17 41–48 1755 14 246 25 0.58 0.38–0.87 32–40 13 636 100 318 305 1.00 — Workers with night-time workd >48 2801 23 499 83 1.09 0.86–1.39 41–48 2216 18 933 32 0.58 0.41–0.84 32–40 15 060 117 152 349 1.00 — Workers without night-time workd >48 6696 55 045 145 0.84 0.71–1.00 41–48 13 427 114 734 243 0.78 0.68–0.89 32–40 119 733 908 638 2522 1.00 — a The analysis was adjusted for calendar time, time passed since start of follow-up, age, night-time work, sex and SES. b The analysis was adjusted for calendar time, time passed since start of follow-up, age, night-time work, and SES. c The analysis was adjusted for calendar time, time passed since start of follow-up, age, night-time work, and sex. d The analysis was adjusted for calendar time, time passed since start of follow-up, age, sex and SES. Table 1 RR with 95% confidence interval for all-cause mortality, as a function of weekly working hours among Danish employees 2000–14, with and without stratification by sex, SES and night shift status, respectively Population Weekly working hours Persons Person years Cases RR 95% CI All workersa >48 9497 78 543 228 0.92 0.80–1.05 41–48 15 643 133 667 275 0.75 0.66–0.85 32–40 134 793 1 025 789 2871 1.00 — Male workersb >48 7344 61 904 195 0.89 0.76–1.03 41–48 9707 83 932 191 0.69 0.59–0.80 32–40 68 961 517 105 1952 1.00 — Female workersb >48 2153 16 639 33 1.06 0.75–1.50 41–48 5936 49 735 84 0.93 0.75–1.17 32–40 65 832 508 684 919 1.00 — Workers with a high SESc >48 3587 29 357 83 0.87 0.69–1.10 41–48 5687 48 124 99 0.77 0.62–0.95 32–40 38 026 257 384 584 1.00 — Workers with a medium SESc >48 1250 11 180 27 0.87 0.59–1.29 41–48 2421 22 043 44 0.83 0.61–1.14 32–40 25 976 203 686 442 1.00 — Workers with a low SESc >48 2891 24 481 79 1.03 0.82–1.29 41–48 5780 49 255 107 0.76 0.62–0.92 32–40 57 155 464 401 1540 1.00 — Workers with unknown SESc >48 1769 13 525 39 0.84 0.60–1.17 41–48 1755 14 246 25 0.58 0.38–0.87 32–40 13 636 100 318 305 1.00 — Workers with night-time workd >48 2801 23 499 83 1.09 0.86–1.39 41–48 2216 18 933 32 0.58 0.41–0.84 32–40 15 060 117 152 349 1.00 — Workers without night-time workd >48 6696 55 045 145 0.84 0.71–1.00 41–48 13 427 114 734 243 0.78 0.68–0.89 32–40 119 733 908 638 2522 1.00 — Population Weekly working hours Persons Person years Cases RR 95% CI All workersa >48 9497 78 543 228 0.92 0.80–1.05 41–48 15 643 133 667 275 0.75 0.66–0.85 32–40 134 793 1 025 789 2871 1.00 — Male workersb >48 7344 61 904 195 0.89 0.76–1.03 41–48 9707 83 932 191 0.69 0.59–0.80 32–40 68 961 517 105 1952 1.00 — Female workersb >48 2153 16 639 33 1.06 0.75–1.50 41–48 5936 49 735 84 0.93 0.75–1.17 32–40 65 832 508 684 919 1.00 — Workers with a high SESc >48 3587 29 357 83 0.87 0.69–1.10 41–48 5687 48 124 99 0.77 0.62–0.95 32–40 38 026 257 384 584 1.00 — Workers with a medium SESc >48 1250 11 180 27 0.87 0.59–1.29 41–48 2421 22 043 44 0.83 0.61–1.14 32–40 25 976 203 686 442 1.00 — Workers with a low SESc >48 2891 24 481 79 1.03 0.82–1.29 41–48 5780 49 255 107 0.76 0.62–0.92 32–40 57 155 464 401 1540 1.00 — Workers with unknown SESc >48 1769 13 525 39 0.84 0.60–1.17 41–48 1755 14 246 25 0.58 0.38–0.87 32–40 13 636 100 318 305 1.00 — Workers with night-time workd >48 2801 23 499 83 1.09 0.86–1.39 41–48 2216 18 933 32 0.58 0.41–0.84 32–40 15 060 117 152 349 1.00 — Workers without night-time workd >48 6696 55 045 145 0.84 0.71–1.00 41–48 13 427 114 734 243 0.78 0.68–0.89 32–40 119 733 908 638 2522 1.00 — a The analysis was adjusted for calendar time, time passed since start of follow-up, age, night-time work, sex and SES. b The analysis was adjusted for calendar time, time passed since start of follow-up, age, night-time work, and SES. c The analysis was adjusted for calendar time, time passed since start of follow-up, age, night-time work, and sex. d The analysis was adjusted for calendar time, time passed since start of follow-up, age, sex and SES. Table 2 R with 95% confidence interval for all-cause mortality, as a function of weekly working hours among the participants of the various sensitivity analyses Population Weekly working hours Persons Person years Cases RR 95% CI Workers interviewed more than once, with stable exposure to working hours >48 3935 29 856 85 0.91 0.73–1.14 41–48 12 511 96 355 246 0.94 0.82–1.07 32–40 93 896 592 859 1547 1.00 — Workers interviewed in 1999–2000 >48 1667 22 788 93 1.06 0.85–1.32 41–48 2716 37 844 83 0.68 0.54–0.85 32–40 21 733 307 190 1070 1.00 — Workers interviewed in 2001–06 >48 3701 37 197 98 0.82 0.67–1.02 41–48 6527 66 248 138 0.76 0.64–0.91 32–40 36 770 380 976 1058 1.00 — Workers interviewed in 2007–13 >48 4129 18 558 37 0.81 0.58–1.13 41–48 6400 29 576 54 0.83 0.63–1.09 32–40 76 290 337 623 743 1.00 — Population Weekly working hours Persons Person years Cases RR 95% CI Workers interviewed more than once, with stable exposure to working hours >48 3935 29 856 85 0.91 0.73–1.14 41–48 12 511 96 355 246 0.94 0.82–1.07 32–40 93 896 592 859 1547 1.00 — Workers interviewed in 1999–2000 >48 1667 22 788 93 1.06 0.85–1.32 41–48 2716 37 844 83 0.68 0.54–0.85 32–40 21 733 307 190 1070 1.00 — Workers interviewed in 2001–06 >48 3701 37 197 98 0.82 0.67–1.02 41–48 6527 66 248 138 0.76 0.64–0.91 32–40 36 770 380 976 1058 1.00 — Workers interviewed in 2007–13 >48 4129 18 558 37 0.81 0.58–1.13 41–48 6400 29 576 54 0.83 0.63–1.09 32–40 76 290 337 623 743 1.00 — Table 2 R with 95% confidence interval for all-cause mortality, as a function of weekly working hours among the participants of the various sensitivity analyses Population Weekly working hours Persons Person years Cases RR 95% CI Workers interviewed more than once, with stable exposure to working hours >48 3935 29 856 85 0.91 0.73–1.14 41–48 12 511 96 355 246 0.94 0.82–1.07 32–40 93 896 592 859 1547 1.00 — Workers interviewed in 1999–2000 >48 1667 22 788 93 1.06 0.85–1.32 41–48 2716 37 844 83 0.68 0.54–0.85 32–40 21 733 307 190 1070 1.00 — Workers interviewed in 2001–06 >48 3701 37 197 98 0.82 0.67–1.02 41–48 6527 66 248 138 0.76 0.64–0.91 32–40 36 770 380 976 1058 1.00 — Workers interviewed in 2007–13 >48 4129 18 558 37 0.81 0.58–1.13 41–48 6400 29 576 54 0.83 0.63–1.09 32–40 76 290 337 623 743 1.00 — Population Weekly working hours Persons Person years Cases RR 95% CI Workers interviewed more than once, with stable exposure to working hours >48 3935 29 856 85 0.91 0.73–1.14 41–48 12 511 96 355 246 0.94 0.82–1.07 32–40 93 896 592 859 1547 1.00 — Workers interviewed in 1999–2000 >48 1667 22 788 93 1.06 0.85–1.32 41–48 2716 37 844 83 0.68 0.54–0.85 32–40 21 733 307 190 1070 1.00 — Workers interviewed in 2001–06 >48 3701 37 197 98 0.82 0.67–1.02 41–48 6527 66 248 138 0.76 0.64–0.91 32–40 36 770 380 976 1058 1.00 — Workers interviewed in 2007–13 >48 4129 18 558 37 0.81 0.58–1.13 41–48 6400 29 576 54 0.83 0.63–1.09 32–40 76 290 337 623 743 1.00 — Discussion In the present study, we found that moderately long work weeks (41–48 h) were statistically significantly associated with decreased mortality rates among full-time employees in the general working population of Denmark. We did not find any significant effect of overtime work which exceeds the limit of EUWTD (>48 h a week), and we did not find any statistically significant interaction between weekly working hours and SES, sex or night-time work. Since participation in the LFS is voluntary, the results will be open to non-participation bias. We know that the response rates have declined with time. The results of our second sensitivity analysis thereby suggests that non-participation may have biased the main finding of the present study towards unity; the lower the response rate the weaker the estimated association between moderately long work weeks and all-cause mortality. The main advantages of the present study are (i) that the participants were randomly selected from the target population and (ii) that bias due to missing follow-up data were eliminated through the ascertainment of deaths in a national register which covers the entire target population. Another advantage is that information on working hours was available both for primary and secondary jobs. If we had disregarded hours worked in secondary jobs, then 25.0% of the workers with 49–100 h a week would have been misclassified as working less than 49 h and 24.6% of the workers with 41–48 h a week would have been misclassified as working less than 41 h. The RR among workers with >48 h would have changed from 0.92 to 1.00 while the RR for 41–48 h would have changed from 0.75 to 0.74. Due to these advantages, we can be quite certain that moderately long work weeks are associated with decreased mortality rates in our target population. We can however not know if we are looking at a healthy worker effect or a causal relationship. Apart from the Northern Irish study that we mentioned in the introduction,3 we found three studies in which all-cause mortality was examined in relation to long working hours or overtime work, one from Denmark,5 one from Great Britain6 and one from Sweden.4 The Danish study5 was based on a 30-year follow-up of a cohort of 5249 men, who were 40–59 years old at baseline (1971–72). The age-adjusted all-cause mortality RR was estimated at 1.07 (95% CI: 0.95–1.20) for 41–45 vs. <= 40 working hours per week and 0.91 (0.79–1.05) for >45 vs. <= 40 h week−1. The British study6 followed a cohort of 39–61 year old civil servants (n = 6014) who worked full time at baseline (1991–94), for on average 11 years. The age, sex, marital status and occupational grade adjusted all-cause mortality RR was 1.11 (95% CI: 0.75–1.63) for employees working 1 h overtime, 1.27 (0.83–1.94) for 2 h overtime and 1.35 (0.82–2.21) for 3–4 h overtime per day, when compared with employees not working overtime. The Swedish study4 used data from a twin registry to study mortality 1973–96 as a function of self-reported overtime and extra work at or prior to baseline among like-sexed twins (9500 women and 11 132 men) born in Sweden 1926–58. The subjects were treated as a sample from the general population regardless of twinning. Compared with other employees of the same sex, the age-adjusted all-cause mortality RRs for >5 h overtime work per week was estimated at 1.69 (95% CI: 1.06–2.69) among the women and 1.08 (0.89–1.31) among the men. The RRs for moderate overtime work of a maximum 5 h a week was estimated at 0.85 (0.52–1.38) among the women and 0.63 (0.49–0.81) among the men. A similar pattern was observed for extra work (work outside employment). In line with the studies from Northern Ireland3 and Denmark,5 we did not find any excess mortality among employees with long weekly working hours and in line with the Swedish study,4 we found significantly low mortality rates among employees with moderate overtime work. Our findings are, however, at odds with some recent meta-analyses, which suggest that long weekly working hours are associated with an increased risk of atrial fibrillation,7 coronary heart disease and stroke.8 This lack of agreement weakens the generalizability of the meta-analyses. It also indicates that caution should be exercised if we want to generalize our results to populations outside of Denmark. As previously mentioned, it is possible that the lowered risk among employees with moderately long hours was due to a healthy worker effect. It is also possible that the non-significant RR among employees with very long hours can be explained by a protective selection effect that has been cancelled out by malignant factors of long hours. Unfortunately, the LFS data did not include any health questions. We have therefore not been able to pursue this matter in the present study. We propose that the possible health selection into long working hours could be investigated in future research by use of health questionnaires and subsequent follow-up of employees with normal working hours at baseline. Conclusions Mortality rates in Denmark are significantly lower among employees with moderately long work weeks than they are among full-time employees without overtime work. The results of our study suggest that a threshold at 48 working hours a week generally affords more than ample protection against excess mortality from long work weeks. Funding The study was partially supported by a grant from the Danish Working Environment Research Fund (ref: 38-2013-09/20130069288). Key points The EU Working Time Directive stipulates that a worker’s average working time for each 7-day period, including overtime, does not exceed 48 h. Forty-eight working hours a week is an arbitrary threshold and it had not been settled whether or not it is low enough to protect against excess mortality from long work weeks. This study found that all-cause mortality rates in Denmark were significantly lower among employees with a moderately long work week (41–48 h) than they were among full-time employees without overtime. In spite of a sufficient statistical power, there were no significant effects of overtime work which exceeds the limit of the directive (> 48 working hours a week). The study indicates that a threshold at 48 working hours a week generally affords more than ample protection against excess mortality from long work weeks in Denmark. Acknowledgements The data of the project were supplied by Statistics Denmark. The study was partially supported by a grant from the Danish Working Environment Research Fund (ref: 38-2013-09/20130069288). Conflicts of interest: None declared. References 1 Spurgeon A , Harrington JM , Cooper CL . Health and safety problems associated with long working hours: a review of the current position . Occup Environ Med 1997 ; 54 : 367 – 75 . Google Scholar Crossref Search ADS PubMed 2 European Parliament , Council of the European Union. Directive 2003/88/EC of The European Parliament and of The Council of 4 November 2003 concerning certain aspects of the organisation of working time . Official J Eur Union 2003 ; 299 : 9 – 19 . 3 O'Reilly D , Rosato M . Worked to death? A census-based longitudinal study of the relationship between the numbers of hours spent working and mortality risk . Int J Epidemiol 2013 ; 42 ( 6 ): 1820 – 30 . Google Scholar Crossref Search ADS PubMed 4 Nylén L , Voss M , Floderus B . Mortality among women and men relative to unemployment, part time work, overtime work, and extra work: a study based on data from the Swedish twin registry . Occup Environ Med 2001 ; 58 ( 1 ): 52 – 7 . Google Scholar Crossref Search ADS PubMed 5 Holtermann A , Mortensen OS , Burr H , et al. Long work hours and physical fitness: 30-year risk of ischaemic heart disease and all-cause mortality among middle-aged Caucasian men . Heart 2010 ; 96 : 1638 – 44 . Google Scholar Crossref Search ADS PubMed 6 Virtanen M , Ferrie JE , Singh-Manoux A , et al. Overtime work and incident coronary heart disease: the Whitehall II prospective cohort study . Eur Heart J 2010 ; 31 : 1737 – 44 . Google Scholar Crossref Search ADS PubMed 7 Kivimäki M , Nyberg ST , Batty GD , et al. Long working hours as a risk factor for atrial fibrillation: a multi-cohort study . Eur Heart J 2017 ; 38 ( 34 ): 2621 – 28 . PubMed PMID: 28911189. Google Scholar Crossref Search ADS PubMed 8 Kivimäki M , Jokela M , Nyberg ST , et al. Long working hours and risk of coronary heart disease and stroke: a systematic review and meta-analysis of published and unpublished data for 603, 838 individuals . Lancet 2015 ; 386 ( 10005 ): 1739 – 46 . Epub 2015 Aug 19. Review. PubMed PMID: 26298822. Google Scholar Crossref Search ADS PubMed 9 Lee DW , Hong YC , Min KB , et al. The effect of long working hours on 10-year risk of coronary heart disease and stroke in the Korean population: the Korea National Health and Nutrition Examination Survey (KNHANES), 2007 to 2013 . Ann Occup Environ Med 2016 ; 28 : 64 . Google Scholar Crossref Search ADS PubMed 10 Toivanen S . Income differences in stroke mortality: a 12-year follow-up study of the Swedish working population . Scand J Public Health 2011 ; 39 : 797 – 804 . Google Scholar Crossref Search ADS PubMed 11 Andersen I , Osler M , Petersen L , et al. Income and risk of ischaemic heart disease in men and women in a Nordic welfare country . Int J Epidemiol 2003 ; 32 : 367 – 74 . Google Scholar Crossref Search ADS PubMed 12 Sareen J , Afifi TO , McMillan KA , Asmundson GJ . Relationship between household income and mental disorders: findings from a population-based longitudinal study . Arch Gen Psychiatr 2011 ; 68 : 419 – 27 . Google Scholar Crossref Search ADS PubMed 13 Hannerz H , Soll-Johanning H . ( 2017 ). General mortality in relation to the EU Working Time Directive: a Danish study protocol . figshare . https://doi.org/10.6084/m9.figshare.5297062.v1 (10 August 2017, date last accessed). 14 Pedersen CB . The Danish civil registration system . Scand J Public Health 2011 ; 39 : 22 – 5 . Google Scholar Crossref Search ADS PubMed 15 Petersson F , Baadsgaard M , Thygesen LC . Danish registers on personal labour market affiliation . Scand J Public Health 2011 ; 39 : 95 – 8 . Google Scholar Crossref Search ADS PubMed 16 Statistics Denmark . Arbejdskraftundersøgelsen. 2015 . Available at: http://www.dst.dk/da/statistik/dokumentation/statistikdokumentation/arbejdskraftundersoegelsen (16 June 2016, date last accessed) [WebCite Cache ID 6gXb0sUA8]. 17 Hannerz H , Larsen AD , Garde AH . Working time arrangements as potential risk factors for ischemic heart disease among workers in Denmark: a study protocol . JMIR Res Protoc 2016 ; 5 : e130 . Google Scholar Crossref Search ADS PubMed 18 Hannerz H , Albertsen K . Long working hours and subsequent use of psychotropic medicine: a study protocol . JMIR Res Protoc 2014 ; 3 : e51 . Google Scholar Crossref Search ADS PubMed 19 Kleppa E , Sanne B , Tell GS . Working overtime is associated with anxiety and depression: the Hordaland Health Study . J Occup Environ Med 2008 ; 50 : 658 – 66 . Google Scholar Crossref Search ADS PubMed 20 Larsen AD , Hannerz H , Møller SV , et al. Night work, long work weeks, and risk of accidental injuries. A register-based study . Scand J Work Environ Health 2017 ; 43 : 578 – 86 . doi: 10.5271/sjweh.3668. Epub 2017 Sep 15. PubMed PMID: 28914325. Google Scholar PubMed 21 van Amelsvoort LG , Schouten EG , Maan AC , et al. Changes in frequency of premature complexes and heart rate variability related to shift work . Occup Environ Med 2001 ; 58 : 678 – 81 . Google Scholar Crossref Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of the European Public Health Association. All rights reserved. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png The European Journal of Public Health Oxford University Press

Working hours and all-cause mortality in relation to the EU Working Time Directive: a Danish cohort study

Loading next page...
 
/lp/ou_press/working-hours-and-all-cause-mortality-in-relation-to-the-eu-working-eyTaI6q7ZJ
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the European Public Health Association. All rights reserved.
ISSN
1101-1262
eISSN
1464-360X
D.O.I.
10.1093/eurpub/cky027
Publisher site
See Article on Publisher Site

Abstract

Abstract Background In keeping with the need to protect the safety and health of workers, the EU Working Time Directive stipulates that a worker’s average working time for each 7-day period, including overtime, does not exceed 48 h. It has, however, not been settled whether or not the threshold at 48 working hours a week is low enough to protect against excess mortality from long work weeks. The aim of the present study was to examine all-cause mortality in relation to weekly working hours among employees in the general population of Denmark. A special attention was given to mortality rates among employees with moderately long work weeks, 41–48 h. Methods Interview data from cohorts of 20–64 year-old employees were drawn from the Danish Labour Force Survey. The participants (N = 159 933) were followed through national registers from the end of the calendar year of the interview (1999–2013) until the end of 2014. Rate ratios (RRs) for all-cause mortality were estimated as a function of weekly working hours while controlling for age, sex, social class, night-time work and calendar year. Results We found 3374 deaths during an average follow-up time of 7.7 years. With 32–40 working hours a week as reference, the RRs for all-cause mortality were 0.75 (95% CI: 0.66–0.85) for 41–48 and 0.92 (0.80–1.05) for >48 h. Conclusion Mortality rates in Denmark are significantly lower among employees with moderately long work weeks than they are among full-time employees without overtime work. Introduction It has often been suggested that long working hours may be associated with poor sleep, insufficient restitution between work shifts and a subsequent increased susceptibility to accidents as well as mental and physical health problems.1 It is therefore stipulated in the EU Working Time Directive (EUWTD) that Member States shall take the measures necessary to ensure that a worker’s average working time for each 7-day period, including overtime, does not exceed 48 h.2 The purpose of the directive, which was issued in 1993, with slight amendments in the years 2000 and 2003, is to protect the safety and health of workers. If the intent of the directive is fulfilled then the 48 h limit should be enough to protect against adverse health effects from long weekly working hours. It should, however, be noted that we are dealing with an arbitrary cut-point. The assumption that moderate overtime work, 41–48 h a week, does not constitute a public health problem, has not been sufficiently tested. A recent study by O'Reilly and Rosato3 examined the relationship between weekly working hours and premature death (the ultimate consequence of a poor safety and health situation) in a cohort of 20–64 year-old people from Northern Ireland (144 938 women and 270 011 men) who worked at least 35 h a week. During an 8.7 years follow-up period, they observed 1143 deaths among the women and 4447 deaths among the men. Among the women, the age, marital status and socioeconomic status (SES) adjusted all-cause mortality rate ratio (RR) was 0.99 (95% CI: 0.80–1.22) for employees working 41–48 h, 1.18 (0.88–1.58) for employees working 49–54 h and 0.86 (0.63–1.18) for employees working >54 h a week, when compared with employees working 35–40 h a week. Among the men, the corresponding RRs were 0.96 (95% CI: 0.88–1.04), 1.01 (0.91–1.12) and 0.96 (0.87–1.05). The concerned RRs were also estimated within four different SES groups and only one instance of a significantly elevated RR was found, namely among male employees who worked more than 54 h a week in routine occupations. Apart from the above, only a few small studies have dealt with all-cause mortality as a function of overtime work4–6 and none of these used exposure definitions that are compatible with the cut-point of EUWTD. There are, however, some large and interesting studies which suggest that long weekly working hours are associated with an increased risk of atrial fibrillation,7 coronary heart disease and stroke.8 It has moreover, been suggested or implied that various associations between long working hours and health or safety depends on gender,9 SES8 and night-time work.2 The results by O’Reilly and Rosato3 suggest that a threshold at 48 h a week would afford more than ample protection against excess mortality from long weekly working hours. However, O'Reilly and Rosato did not regard hours worked outside of the person’s main job. Long working hours due to one or more secondary jobs could thereby be misclassified as normal working hours (35–40 h a week), which would bias the estimated RRs toward unity. To rectify this shortcoming, the present study would examine the relationship between weekly working hours and all-cause mortality among employees in the general working population of Denmark, while taking hours worked in secondary jobs into account. Since long working hours also are associated with an increased income, which in turn has been associated with a decreased risk of stroke,10 ischaemic heart disease11 and mental health problems,12 there are theoretical arguments both for positive and negative associations between long working hours and all-cause mortality. Aims and hypotheses We wanted to know if all-cause mortality rates are independent of weekly working hours among full-time employees in Denmark, and we addressed this research question in a series of nested hypothesis tests. The following null-hypotheses were tested: All-cause mortality rates among full-time employees in Denmark are independent of weekly working hours as well as interaction between weekly working hours and SES, sex and night-time work, respectively There are no interaction effects between weekly working hours and SES. There are no interaction effects between weekly working hours and sex. There are no interaction effects between weekly working hours and night-time work. The all-cause mortality rates are independent of weekly working hours when we disregard interaction effects. The overall significance level was set at 0.05 and the multiple testing problem was solved by the following strategy: The first level null-hypothesis would be rejected if the P-value of its statistical test was less than or equal to 0.05. A null-hypothesis at the second level would be rejected if (i) the first-level null-hypothesis was rejected and (ii) the P-value of its statistical test was less than or equal to 0.05. Methods The statistical analysis was performed in accordance with a study protocol,13 which we published before we looked at any relations between the exposure and outcome data of the study. Methodological details of the study protocol will be repeated in the present method section. Data material The data of the present study were obtained through a person-based linkage between the Danish Labour Force Survey (LFS) 1999–2013, the central person register and the employment classification module. The Central Person Register contains, inter alia, information on sex and dates of birth, death, and migrations for every person who is or has been an inhabitant of Denmark sometime between 1968 and the present.14 The Employment Classification Module, which has existed since 1975 and covers all inhabitants of Denmark, contains person-based annual information on SES, occupation, and industry.15 LFS is based on quarterly random samples of 15- to 74-year-old people in the Danish population. Each participant is invited to be interviewed four times over a period of one and a half years. Structured telephone interviews are used to gather person-based information on, inter alia, usual weekly working hours, calculated by adding the hours worked in secondary jobs to the ones worked in a primary job, and night-time work (‘Yes, regularly’, ‘Yes, occasionally’, ‘No’).16 The response rate has decreased with time, from 70% in 2002–53% in 2013. The questions used to gather the information on working hours have changed slightly with time. Before 2001, the questionnaires did not specify whether or not meal breaks should be included in the number of working hours. In 2001–06, it was specified that all meal breaks should be excluded and from 2007, it was specified that meal breaks should be included if the person got paid while eating and excluded otherwise. Follow-up and inclusion criteria Participants who were between 20 and 64 years old at the start of the follow-up period and employed with 32–100 weekly working hours at the time of the interview were eligible for inclusion. The included participants were followed from the beginning of the calendar year that succeeds that of their baseline interview. The follow-up ended at the time they emigrated or died, or the study period ended (31 December 2014), whichever came first. A flow-chart for the inclusion and exclusion procedure is given in figure 1. In total, we included 159 933 participants (46% were women), and we found 3374 deaths in 1 237 999 person-years at risk (mean follow-up 7.7 years). Figure 1 View largeDownload slide Flow-chart for the inclusion and exclusion criteria. Figure 1 View largeDownload slide Flow-chart for the inclusion and exclusion criteria. Primary statistical analysis Poisson regression was used to analyse all-cause mortality rates as a function of weekly working hours (32–40, 41–48 or >48 h a week), night-time work (‘Yes, regularly’ or ‘Yes, occasionally’ vs. ‘No’), sex, age (10-year classes), calendar time (2000–04, 2005–09 or 2010–14), time passed since start of follow-up (0–4, 5–9 or ≥10 years) and SES (low, medium, high or unknown). SES was based on the participant’s main occupation during the calendar year of the interview and coded in accordance with the three-class version of the European Socioeconomic Classification (ESeC). The coding procedure is described by Hannerz et al.17 Age, calendar time and time passed since start of follow-up were treated as dynamic (time-varying) variables. The remaining variables were fixed at baseline (the calendar year of the interview). The logarithm of person-years at risk was used as offset and the maximum likelihood method was used to estimate the parameters. People who participated in more than one interview were classified in accordance with the responses given in their first interview. The full model contained the following covariates: calendar time, time passed since start of follow-up, age, sex, SES, night-time work, working hours, working hours × sex, working hours × SES and working hours × night-time work. Hypothesis 1 was tested by use of a likelihood ratio which compared the full model to a sub-model in which the main as well as the interaction effects of working hours were excluded. Hypotheses 1.1–1.3 was tested by use of likelihood ratios which compared the full model to sub-models in which the respective interaction terms were excluded. Hypothesis 1.4 was tested by use of a likelihood ratio which compared a model which contained all main effects of the full model to a sub-model in which the main effect of working hours was excluded. Parameter estimates were used to calculate RRs for all-cause mortality, with 95% confidence intervals, as a function of weekly working hours with and without stratification by sex, SES, and night-time work, respectively. In keeping with a set of previous studies,18–21 we defined fulltime work as 32 or more hours a week. The following contrasts were considered: 41–48 vs. 32–40 working hours a week and >48 vs. 32–40 working hours a week. Sensitivity analysis 1—only stable exposure To find out if the strength of the association between the examined work time arrangements and all-cause mortality increases when exposure is more stable over time, we conducted a sensitivity analysis, in which we only included people who (i) participated in more than one interview, (ii) were between 20 and 64 years old during their last interview, (iii) were employed according to their first as well as their last interview and (iv) did not move more than one step among the ordered working time categories (<32; 32–40; 41–48; >48 h a week) between the first and last interview. The included participants were thereafter re-categorized according to the mean of the reported usual working hours during their first and last interview. The follow-up of the sensitivity analysis started on January 1 of the calendar year which succeeded the calendar year of the participant’s last interview. Only main effects were regarded. In all other respects, the statistical models and covariates were the same as in the primary analysis. Sensitivity analysis 2—stratification by interview calendar period Since the questions used to obtain information about weekly working hours were slightly revised in 2001 and then again in 2007, we performed a sensitivity analysis with the results stratified by calendar period of interview (1999–2000, 2001–06 and 2007–13). Only main effects were regarded. The statistical model, covariates and inclusion criteria were otherwise the same as in the primary analysis. Results The main null hypothesis which stated that all-cause mortality rates among full-time employees in Denmark are independent of weekly working hours as well as interaction between weekly working hours and SES, sex and night-time work, respectively, was rejected by the likelihood ratio test (P = 0.0004). No significant interactions with working hours were found for SES (P = 0.86), sex (P = 0.10) or night-time work (P = 0.10). The sub-model, which only included main effects, yielded a P values for the effect of weekly working hours that was less than 0.0001, and the estimated RRs among employees with 41–48 and >48 weekly working hours were lower than those among the reference group (32–40 h). The estimated RRs from the primary statistical analysis are given in table 1, with and without stratification by sex, SES and night shift status, respectively. The RR for all-cause mortality among the employees with overtime work within the limits of EUWTD (41–48 h) was lower than unity in each of the examined sub-populations. They were also lower than unity in each of the sensitivity analyses (table 2). Table 1 RR with 95% confidence interval for all-cause mortality, as a function of weekly working hours among Danish employees 2000–14, with and without stratification by sex, SES and night shift status, respectively Population Weekly working hours Persons Person years Cases RR 95% CI All workersa >48 9497 78 543 228 0.92 0.80–1.05 41–48 15 643 133 667 275 0.75 0.66–0.85 32–40 134 793 1 025 789 2871 1.00 — Male workersb >48 7344 61 904 195 0.89 0.76–1.03 41–48 9707 83 932 191 0.69 0.59–0.80 32–40 68 961 517 105 1952 1.00 — Female workersb >48 2153 16 639 33 1.06 0.75–1.50 41–48 5936 49 735 84 0.93 0.75–1.17 32–40 65 832 508 684 919 1.00 — Workers with a high SESc >48 3587 29 357 83 0.87 0.69–1.10 41–48 5687 48 124 99 0.77 0.62–0.95 32–40 38 026 257 384 584 1.00 — Workers with a medium SESc >48 1250 11 180 27 0.87 0.59–1.29 41–48 2421 22 043 44 0.83 0.61–1.14 32–40 25 976 203 686 442 1.00 — Workers with a low SESc >48 2891 24 481 79 1.03 0.82–1.29 41–48 5780 49 255 107 0.76 0.62–0.92 32–40 57 155 464 401 1540 1.00 — Workers with unknown SESc >48 1769 13 525 39 0.84 0.60–1.17 41–48 1755 14 246 25 0.58 0.38–0.87 32–40 13 636 100 318 305 1.00 — Workers with night-time workd >48 2801 23 499 83 1.09 0.86–1.39 41–48 2216 18 933 32 0.58 0.41–0.84 32–40 15 060 117 152 349 1.00 — Workers without night-time workd >48 6696 55 045 145 0.84 0.71–1.00 41–48 13 427 114 734 243 0.78 0.68–0.89 32–40 119 733 908 638 2522 1.00 — Population Weekly working hours Persons Person years Cases RR 95% CI All workersa >48 9497 78 543 228 0.92 0.80–1.05 41–48 15 643 133 667 275 0.75 0.66–0.85 32–40 134 793 1 025 789 2871 1.00 — Male workersb >48 7344 61 904 195 0.89 0.76–1.03 41–48 9707 83 932 191 0.69 0.59–0.80 32–40 68 961 517 105 1952 1.00 — Female workersb >48 2153 16 639 33 1.06 0.75–1.50 41–48 5936 49 735 84 0.93 0.75–1.17 32–40 65 832 508 684 919 1.00 — Workers with a high SESc >48 3587 29 357 83 0.87 0.69–1.10 41–48 5687 48 124 99 0.77 0.62–0.95 32–40 38 026 257 384 584 1.00 — Workers with a medium SESc >48 1250 11 180 27 0.87 0.59–1.29 41–48 2421 22 043 44 0.83 0.61–1.14 32–40 25 976 203 686 442 1.00 — Workers with a low SESc >48 2891 24 481 79 1.03 0.82–1.29 41–48 5780 49 255 107 0.76 0.62–0.92 32–40 57 155 464 401 1540 1.00 — Workers with unknown SESc >48 1769 13 525 39 0.84 0.60–1.17 41–48 1755 14 246 25 0.58 0.38–0.87 32–40 13 636 100 318 305 1.00 — Workers with night-time workd >48 2801 23 499 83 1.09 0.86–1.39 41–48 2216 18 933 32 0.58 0.41–0.84 32–40 15 060 117 152 349 1.00 — Workers without night-time workd >48 6696 55 045 145 0.84 0.71–1.00 41–48 13 427 114 734 243 0.78 0.68–0.89 32–40 119 733 908 638 2522 1.00 — a The analysis was adjusted for calendar time, time passed since start of follow-up, age, night-time work, sex and SES. b The analysis was adjusted for calendar time, time passed since start of follow-up, age, night-time work, and SES. c The analysis was adjusted for calendar time, time passed since start of follow-up, age, night-time work, and sex. d The analysis was adjusted for calendar time, time passed since start of follow-up, age, sex and SES. Table 1 RR with 95% confidence interval for all-cause mortality, as a function of weekly working hours among Danish employees 2000–14, with and without stratification by sex, SES and night shift status, respectively Population Weekly working hours Persons Person years Cases RR 95% CI All workersa >48 9497 78 543 228 0.92 0.80–1.05 41–48 15 643 133 667 275 0.75 0.66–0.85 32–40 134 793 1 025 789 2871 1.00 — Male workersb >48 7344 61 904 195 0.89 0.76–1.03 41–48 9707 83 932 191 0.69 0.59–0.80 32–40 68 961 517 105 1952 1.00 — Female workersb >48 2153 16 639 33 1.06 0.75–1.50 41–48 5936 49 735 84 0.93 0.75–1.17 32–40 65 832 508 684 919 1.00 — Workers with a high SESc >48 3587 29 357 83 0.87 0.69–1.10 41–48 5687 48 124 99 0.77 0.62–0.95 32–40 38 026 257 384 584 1.00 — Workers with a medium SESc >48 1250 11 180 27 0.87 0.59–1.29 41–48 2421 22 043 44 0.83 0.61–1.14 32–40 25 976 203 686 442 1.00 — Workers with a low SESc >48 2891 24 481 79 1.03 0.82–1.29 41–48 5780 49 255 107 0.76 0.62–0.92 32–40 57 155 464 401 1540 1.00 — Workers with unknown SESc >48 1769 13 525 39 0.84 0.60–1.17 41–48 1755 14 246 25 0.58 0.38–0.87 32–40 13 636 100 318 305 1.00 — Workers with night-time workd >48 2801 23 499 83 1.09 0.86–1.39 41–48 2216 18 933 32 0.58 0.41–0.84 32–40 15 060 117 152 349 1.00 — Workers without night-time workd >48 6696 55 045 145 0.84 0.71–1.00 41–48 13 427 114 734 243 0.78 0.68–0.89 32–40 119 733 908 638 2522 1.00 — Population Weekly working hours Persons Person years Cases RR 95% CI All workersa >48 9497 78 543 228 0.92 0.80–1.05 41–48 15 643 133 667 275 0.75 0.66–0.85 32–40 134 793 1 025 789 2871 1.00 — Male workersb >48 7344 61 904 195 0.89 0.76–1.03 41–48 9707 83 932 191 0.69 0.59–0.80 32–40 68 961 517 105 1952 1.00 — Female workersb >48 2153 16 639 33 1.06 0.75–1.50 41–48 5936 49 735 84 0.93 0.75–1.17 32–40 65 832 508 684 919 1.00 — Workers with a high SESc >48 3587 29 357 83 0.87 0.69–1.10 41–48 5687 48 124 99 0.77 0.62–0.95 32–40 38 026 257 384 584 1.00 — Workers with a medium SESc >48 1250 11 180 27 0.87 0.59–1.29 41–48 2421 22 043 44 0.83 0.61–1.14 32–40 25 976 203 686 442 1.00 — Workers with a low SESc >48 2891 24 481 79 1.03 0.82–1.29 41–48 5780 49 255 107 0.76 0.62–0.92 32–40 57 155 464 401 1540 1.00 — Workers with unknown SESc >48 1769 13 525 39 0.84 0.60–1.17 41–48 1755 14 246 25 0.58 0.38–0.87 32–40 13 636 100 318 305 1.00 — Workers with night-time workd >48 2801 23 499 83 1.09 0.86–1.39 41–48 2216 18 933 32 0.58 0.41–0.84 32–40 15 060 117 152 349 1.00 — Workers without night-time workd >48 6696 55 045 145 0.84 0.71–1.00 41–48 13 427 114 734 243 0.78 0.68–0.89 32–40 119 733 908 638 2522 1.00 — a The analysis was adjusted for calendar time, time passed since start of follow-up, age, night-time work, sex and SES. b The analysis was adjusted for calendar time, time passed since start of follow-up, age, night-time work, and SES. c The analysis was adjusted for calendar time, time passed since start of follow-up, age, night-time work, and sex. d The analysis was adjusted for calendar time, time passed since start of follow-up, age, sex and SES. Table 2 R with 95% confidence interval for all-cause mortality, as a function of weekly working hours among the participants of the various sensitivity analyses Population Weekly working hours Persons Person years Cases RR 95% CI Workers interviewed more than once, with stable exposure to working hours >48 3935 29 856 85 0.91 0.73–1.14 41–48 12 511 96 355 246 0.94 0.82–1.07 32–40 93 896 592 859 1547 1.00 — Workers interviewed in 1999–2000 >48 1667 22 788 93 1.06 0.85–1.32 41–48 2716 37 844 83 0.68 0.54–0.85 32–40 21 733 307 190 1070 1.00 — Workers interviewed in 2001–06 >48 3701 37 197 98 0.82 0.67–1.02 41–48 6527 66 248 138 0.76 0.64–0.91 32–40 36 770 380 976 1058 1.00 — Workers interviewed in 2007–13 >48 4129 18 558 37 0.81 0.58–1.13 41–48 6400 29 576 54 0.83 0.63–1.09 32–40 76 290 337 623 743 1.00 — Population Weekly working hours Persons Person years Cases RR 95% CI Workers interviewed more than once, with stable exposure to working hours >48 3935 29 856 85 0.91 0.73–1.14 41–48 12 511 96 355 246 0.94 0.82–1.07 32–40 93 896 592 859 1547 1.00 — Workers interviewed in 1999–2000 >48 1667 22 788 93 1.06 0.85–1.32 41–48 2716 37 844 83 0.68 0.54–0.85 32–40 21 733 307 190 1070 1.00 — Workers interviewed in 2001–06 >48 3701 37 197 98 0.82 0.67–1.02 41–48 6527 66 248 138 0.76 0.64–0.91 32–40 36 770 380 976 1058 1.00 — Workers interviewed in 2007–13 >48 4129 18 558 37 0.81 0.58–1.13 41–48 6400 29 576 54 0.83 0.63–1.09 32–40 76 290 337 623 743 1.00 — Table 2 R with 95% confidence interval for all-cause mortality, as a function of weekly working hours among the participants of the various sensitivity analyses Population Weekly working hours Persons Person years Cases RR 95% CI Workers interviewed more than once, with stable exposure to working hours >48 3935 29 856 85 0.91 0.73–1.14 41–48 12 511 96 355 246 0.94 0.82–1.07 32–40 93 896 592 859 1547 1.00 — Workers interviewed in 1999–2000 >48 1667 22 788 93 1.06 0.85–1.32 41–48 2716 37 844 83 0.68 0.54–0.85 32–40 21 733 307 190 1070 1.00 — Workers interviewed in 2001–06 >48 3701 37 197 98 0.82 0.67–1.02 41–48 6527 66 248 138 0.76 0.64–0.91 32–40 36 770 380 976 1058 1.00 — Workers interviewed in 2007–13 >48 4129 18 558 37 0.81 0.58–1.13 41–48 6400 29 576 54 0.83 0.63–1.09 32–40 76 290 337 623 743 1.00 — Population Weekly working hours Persons Person years Cases RR 95% CI Workers interviewed more than once, with stable exposure to working hours >48 3935 29 856 85 0.91 0.73–1.14 41–48 12 511 96 355 246 0.94 0.82–1.07 32–40 93 896 592 859 1547 1.00 — Workers interviewed in 1999–2000 >48 1667 22 788 93 1.06 0.85–1.32 41–48 2716 37 844 83 0.68 0.54–0.85 32–40 21 733 307 190 1070 1.00 — Workers interviewed in 2001–06 >48 3701 37 197 98 0.82 0.67–1.02 41–48 6527 66 248 138 0.76 0.64–0.91 32–40 36 770 380 976 1058 1.00 — Workers interviewed in 2007–13 >48 4129 18 558 37 0.81 0.58–1.13 41–48 6400 29 576 54 0.83 0.63–1.09 32–40 76 290 337 623 743 1.00 — Discussion In the present study, we found that moderately long work weeks (41–48 h) were statistically significantly associated with decreased mortality rates among full-time employees in the general working population of Denmark. We did not find any significant effect of overtime work which exceeds the limit of EUWTD (>48 h a week), and we did not find any statistically significant interaction between weekly working hours and SES, sex or night-time work. Since participation in the LFS is voluntary, the results will be open to non-participation bias. We know that the response rates have declined with time. The results of our second sensitivity analysis thereby suggests that non-participation may have biased the main finding of the present study towards unity; the lower the response rate the weaker the estimated association between moderately long work weeks and all-cause mortality. The main advantages of the present study are (i) that the participants were randomly selected from the target population and (ii) that bias due to missing follow-up data were eliminated through the ascertainment of deaths in a national register which covers the entire target population. Another advantage is that information on working hours was available both for primary and secondary jobs. If we had disregarded hours worked in secondary jobs, then 25.0% of the workers with 49–100 h a week would have been misclassified as working less than 49 h and 24.6% of the workers with 41–48 h a week would have been misclassified as working less than 41 h. The RR among workers with >48 h would have changed from 0.92 to 1.00 while the RR for 41–48 h would have changed from 0.75 to 0.74. Due to these advantages, we can be quite certain that moderately long work weeks are associated with decreased mortality rates in our target population. We can however not know if we are looking at a healthy worker effect or a causal relationship. Apart from the Northern Irish study that we mentioned in the introduction,3 we found three studies in which all-cause mortality was examined in relation to long working hours or overtime work, one from Denmark,5 one from Great Britain6 and one from Sweden.4 The Danish study5 was based on a 30-year follow-up of a cohort of 5249 men, who were 40–59 years old at baseline (1971–72). The age-adjusted all-cause mortality RR was estimated at 1.07 (95% CI: 0.95–1.20) for 41–45 vs. <= 40 working hours per week and 0.91 (0.79–1.05) for >45 vs. <= 40 h week−1. The British study6 followed a cohort of 39–61 year old civil servants (n = 6014) who worked full time at baseline (1991–94), for on average 11 years. The age, sex, marital status and occupational grade adjusted all-cause mortality RR was 1.11 (95% CI: 0.75–1.63) for employees working 1 h overtime, 1.27 (0.83–1.94) for 2 h overtime and 1.35 (0.82–2.21) for 3–4 h overtime per day, when compared with employees not working overtime. The Swedish study4 used data from a twin registry to study mortality 1973–96 as a function of self-reported overtime and extra work at or prior to baseline among like-sexed twins (9500 women and 11 132 men) born in Sweden 1926–58. The subjects were treated as a sample from the general population regardless of twinning. Compared with other employees of the same sex, the age-adjusted all-cause mortality RRs for >5 h overtime work per week was estimated at 1.69 (95% CI: 1.06–2.69) among the women and 1.08 (0.89–1.31) among the men. The RRs for moderate overtime work of a maximum 5 h a week was estimated at 0.85 (0.52–1.38) among the women and 0.63 (0.49–0.81) among the men. A similar pattern was observed for extra work (work outside employment). In line with the studies from Northern Ireland3 and Denmark,5 we did not find any excess mortality among employees with long weekly working hours and in line with the Swedish study,4 we found significantly low mortality rates among employees with moderate overtime work. Our findings are, however, at odds with some recent meta-analyses, which suggest that long weekly working hours are associated with an increased risk of atrial fibrillation,7 coronary heart disease and stroke.8 This lack of agreement weakens the generalizability of the meta-analyses. It also indicates that caution should be exercised if we want to generalize our results to populations outside of Denmark. As previously mentioned, it is possible that the lowered risk among employees with moderately long hours was due to a healthy worker effect. It is also possible that the non-significant RR among employees with very long hours can be explained by a protective selection effect that has been cancelled out by malignant factors of long hours. Unfortunately, the LFS data did not include any health questions. We have therefore not been able to pursue this matter in the present study. We propose that the possible health selection into long working hours could be investigated in future research by use of health questionnaires and subsequent follow-up of employees with normal working hours at baseline. Conclusions Mortality rates in Denmark are significantly lower among employees with moderately long work weeks than they are among full-time employees without overtime work. The results of our study suggest that a threshold at 48 working hours a week generally affords more than ample protection against excess mortality from long work weeks. Funding The study was partially supported by a grant from the Danish Working Environment Research Fund (ref: 38-2013-09/20130069288). Key points The EU Working Time Directive stipulates that a worker’s average working time for each 7-day period, including overtime, does not exceed 48 h. Forty-eight working hours a week is an arbitrary threshold and it had not been settled whether or not it is low enough to protect against excess mortality from long work weeks. This study found that all-cause mortality rates in Denmark were significantly lower among employees with a moderately long work week (41–48 h) than they were among full-time employees without overtime. In spite of a sufficient statistical power, there were no significant effects of overtime work which exceeds the limit of the directive (> 48 working hours a week). The study indicates that a threshold at 48 working hours a week generally affords more than ample protection against excess mortality from long work weeks in Denmark. Acknowledgements The data of the project were supplied by Statistics Denmark. The study was partially supported by a grant from the Danish Working Environment Research Fund (ref: 38-2013-09/20130069288). Conflicts of interest: None declared. References 1 Spurgeon A , Harrington JM , Cooper CL . Health and safety problems associated with long working hours: a review of the current position . Occup Environ Med 1997 ; 54 : 367 – 75 . Google Scholar Crossref Search ADS PubMed 2 European Parliament , Council of the European Union. Directive 2003/88/EC of The European Parliament and of The Council of 4 November 2003 concerning certain aspects of the organisation of working time . Official J Eur Union 2003 ; 299 : 9 – 19 . 3 O'Reilly D , Rosato M . Worked to death? A census-based longitudinal study of the relationship between the numbers of hours spent working and mortality risk . Int J Epidemiol 2013 ; 42 ( 6 ): 1820 – 30 . Google Scholar Crossref Search ADS PubMed 4 Nylén L , Voss M , Floderus B . Mortality among women and men relative to unemployment, part time work, overtime work, and extra work: a study based on data from the Swedish twin registry . Occup Environ Med 2001 ; 58 ( 1 ): 52 – 7 . Google Scholar Crossref Search ADS PubMed 5 Holtermann A , Mortensen OS , Burr H , et al. Long work hours and physical fitness: 30-year risk of ischaemic heart disease and all-cause mortality among middle-aged Caucasian men . Heart 2010 ; 96 : 1638 – 44 . Google Scholar Crossref Search ADS PubMed 6 Virtanen M , Ferrie JE , Singh-Manoux A , et al. Overtime work and incident coronary heart disease: the Whitehall II prospective cohort study . Eur Heart J 2010 ; 31 : 1737 – 44 . Google Scholar Crossref Search ADS PubMed 7 Kivimäki M , Nyberg ST , Batty GD , et al. Long working hours as a risk factor for atrial fibrillation: a multi-cohort study . Eur Heart J 2017 ; 38 ( 34 ): 2621 – 28 . PubMed PMID: 28911189. Google Scholar Crossref Search ADS PubMed 8 Kivimäki M , Jokela M , Nyberg ST , et al. Long working hours and risk of coronary heart disease and stroke: a systematic review and meta-analysis of published and unpublished data for 603, 838 individuals . Lancet 2015 ; 386 ( 10005 ): 1739 – 46 . Epub 2015 Aug 19. Review. PubMed PMID: 26298822. Google Scholar Crossref Search ADS PubMed 9 Lee DW , Hong YC , Min KB , et al. The effect of long working hours on 10-year risk of coronary heart disease and stroke in the Korean population: the Korea National Health and Nutrition Examination Survey (KNHANES), 2007 to 2013 . Ann Occup Environ Med 2016 ; 28 : 64 . Google Scholar Crossref Search ADS PubMed 10 Toivanen S . Income differences in stroke mortality: a 12-year follow-up study of the Swedish working population . Scand J Public Health 2011 ; 39 : 797 – 804 . Google Scholar Crossref Search ADS PubMed 11 Andersen I , Osler M , Petersen L , et al. Income and risk of ischaemic heart disease in men and women in a Nordic welfare country . Int J Epidemiol 2003 ; 32 : 367 – 74 . Google Scholar Crossref Search ADS PubMed 12 Sareen J , Afifi TO , McMillan KA , Asmundson GJ . Relationship between household income and mental disorders: findings from a population-based longitudinal study . Arch Gen Psychiatr 2011 ; 68 : 419 – 27 . Google Scholar Crossref Search ADS PubMed 13 Hannerz H , Soll-Johanning H . ( 2017 ). General mortality in relation to the EU Working Time Directive: a Danish study protocol . figshare . https://doi.org/10.6084/m9.figshare.5297062.v1 (10 August 2017, date last accessed). 14 Pedersen CB . The Danish civil registration system . Scand J Public Health 2011 ; 39 : 22 – 5 . Google Scholar Crossref Search ADS PubMed 15 Petersson F , Baadsgaard M , Thygesen LC . Danish registers on personal labour market affiliation . Scand J Public Health 2011 ; 39 : 95 – 8 . Google Scholar Crossref Search ADS PubMed 16 Statistics Denmark . Arbejdskraftundersøgelsen. 2015 . Available at: http://www.dst.dk/da/statistik/dokumentation/statistikdokumentation/arbejdskraftundersoegelsen (16 June 2016, date last accessed) [WebCite Cache ID 6gXb0sUA8]. 17 Hannerz H , Larsen AD , Garde AH . Working time arrangements as potential risk factors for ischemic heart disease among workers in Denmark: a study protocol . JMIR Res Protoc 2016 ; 5 : e130 . Google Scholar Crossref Search ADS PubMed 18 Hannerz H , Albertsen K . Long working hours and subsequent use of psychotropic medicine: a study protocol . JMIR Res Protoc 2014 ; 3 : e51 . Google Scholar Crossref Search ADS PubMed 19 Kleppa E , Sanne B , Tell GS . Working overtime is associated with anxiety and depression: the Hordaland Health Study . J Occup Environ Med 2008 ; 50 : 658 – 66 . Google Scholar Crossref Search ADS PubMed 20 Larsen AD , Hannerz H , Møller SV , et al. Night work, long work weeks, and risk of accidental injuries. A register-based study . Scand J Work Environ Health 2017 ; 43 : 578 – 86 . doi: 10.5271/sjweh.3668. Epub 2017 Sep 15. PubMed PMID: 28914325. Google Scholar PubMed 21 van Amelsvoort LG , Schouten EG , Maan AC , et al. Changes in frequency of premature complexes and heart rate variability related to shift work . Occup Environ Med 2001 ; 58 : 678 – 81 . Google Scholar Crossref Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of the European Public Health Association. All rights reserved. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Journal

The European Journal of Public HealthOxford University Press

Published: Oct 1, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off