Article Text

Download PDFPDF

Socio-economic differences in the association between sickness absence and mortality: the prospective DREAM study of Danish private sector employees
  1. T Lund1,
  2. M Kivimäki2,3,
  3. K B Christensen4,
  4. M Labriola1
  1. 1
    Danish National Centre for Social Research, Copenhagen, Denmark
  2. 2
    Department of Epidemiology and Public Health, University College London, London, UK
  3. 3
    Finnish Institute of Occupational Health, Helsinki, Finland
  4. 4
    National Research Centre for the Working Environment, Copenhagen, Denmark
  1. Thomas Lund, Danish National Centre for Social Research, Herluf Trolles Gade 11, DK-1052 Copenhagen, Denmark; tlu{at}


Objectives: To examine duration of sickness absence as a risk marker for future mortality by socio-economic position among all private sector employees in Denmark in 1998–2004.

Methods: All residents in Denmark employed in the private sector receiving sickness absence compensation in 1998 were investigated in a prospective cohort study. 236 207 persons (38.2% women, 61.8% men, age range 18–65, mean age 37.8 years) alive on 1 January 2001 were included in the study. Mortality from 1 January 2001 to 31 December 2004 was assessed using national register data. Deaths in 1999 and 2000 were excluded to determine the status of sickness absence duration as an early risk marker. For analyses within occupational grades, data were available for a sub-population of 137 607 study participants.

Results: 3040 persons died during follow-up. The age-adjusted risk of future mortality increased by duration of sickness absence in a graded fashion among men and non-blue collar workers. Among women and blue collar workers, there was no association of mortality with duration of sickness absences below 6 weeks. However, employees with ⩾6 weeks of absence compared to those with 1-week absence had a substantial excess risk of death in all groups: adjusted hazard ratio 2.2 (95% CI 1.8 to 2.7) for women, 2.1 (95% CI 1.8 to 2.4) for men, 3.7 (95% CI 1.9 to 7.2) in white collar occupations, 3.3 (95% CI 2.2 to 5.0) in intermediate grade occupations and 2.0 (95% CI 1.7 to 2.3) in blue collar occupations.

Conclusion: Administratively collected data on sickness absence compensation for periods ⩾6 weeks identified “at risk” groups for future excess mortality in male and female private sector employees across occupational grade levels.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

There seems to be increasing recognition of the abilities of certain measures of sickness absence to measure physical, psychological and social functioning as well as to predict permanent labour market exclusion in terms of, for example, disability pension.16 In contrast, only a few studies have assessed the predictive abilities of sickness absence in terms of hard endpoints such as mortality in working populations. In the Finnish 10-Town study of 12 821 male and 28 915 female municipal employees followed for 4.5 years, having more than one sickness absence period longer than 3 days per year compared to no such absence periods was associated with 3–4-fold increased risk of mortality. A similar excess mortality was also found for employees with annual sickness absence exceeding 15 days.7 Among 6895 male and 3413 female civil servants participating in the British Whitehall II study, employees who had more than five medically certified absences spells exceeding 7 days per 10 years had a mortality 3–5 times greater than those with no such absence.8 In both studies, associations between sickness absence and mortality were slightly stronger among men than among women.

The predictive ability of other global measures of health, such as self-rated health, has been found to vary between occupational grades.9 No previous studies have addressed potential disparities in the sickness absence–mortality association according to occupational grade. As long sick leaves and incident mortality are rare events, the sample size should be substantially larger than in previous studies7 8 to allow a detailed analysis of the association between absence duration and mortality risk, especially if addressing the social patterning in sickness absence10 and mortality.11 Moreover, previous studies were based on selected samples, typically public sector personnel, and no data are available specifically for private sector employees.7 8 12

The purpose of this study was to examine the associations between sickness absence and future excess mortality risk, and whether these associations differed according to gender and occupational grade. Thus, we performed gender and occupational grade stratified analysis for a large cohort of private sector employees between 18 and 65 years of age at study entry.


Data and population

This study is based on the DREAM (a Danish acronym for The Register-based Evaluation of Marginalisation) register.13 DREAM contains weekly information on granted sickness absence compensation (SAC) in 1998 for all private sector employees and mortality for 1999–2004. In 1998, the private sector in Denmark consisted of 1 495 952 persons of whom 36.6% were women and 63.4% were men. SAC is given to the employer, who can apply for a refund from the state for employees after 2 weeks of sickness absence. The employer covers expenses for the first 2 weeks if the employee has been employed for 8 consecutive weeks prior to sickness absence and has worked for at least 74 h during this period. The employee is eligible for SAC from the municipality if he or she has been employed for at least 13 consecutive weeks prior to sickness absence and has worked for at least 120 h during this period. Exceptions to this rule can be made if the employee is a member of an unemployment insurance fund, has finished vocational training of at least 18 months’ duration within the last month or is a trainee. We assessed duration of sickness absence for all new cases occurring between 1 January and 31 December 1998 among those who had received no social transfer payments during the last week of 1997 and with no social transfer payment in the week preceding onset of SAC. Maternity related absence is excluded. This study assesses risk of future mortality among all employees who received SAC and worked in the private sector in Denmark.

Study design and outcome

A person can normally receive SAC for no longer than 12 months. After this, he or she will have to return to work, receive unemployment benefits or be disability pensioned. To eliminate confounding attributable to sickness absence periods immediately prior to death, the follow-up period for mortality started 24 months after the assessment period of sickness absence. Mortality was identified from 1 January 2001 to 31 December 2004.

Assessment of sickness absence

Duration of sickness absence is based on the weekly records of SAC in DREAM. The range of duration varies from 1 to 53 weeks. A total of 240 815 persons receiving SAC during 1998 were identified. They had not received SAC or any other social transfer payment in the final week of 1997. Those who died during the 2-year period from 1999 to 2000 were excluded from the study, as they were no longer under risk of death in the follow-up period from 2001 to 2004. Therefore, 236 207 persons (38.2% women, 61.8% men) between 18 and 65 years of age who received SAC in 1998 were included in the study.

Other variables

The study includes information about sex, age and occupational grade at baseline, and death during the study period. Occupational grade was assessed through information on membership of an unemployment fund, and divided into blue collar occupations, intermediate grade occupations and white collar occupations. This was only possible for the 137 607 participants enrolled in an unemployment fund at baseline.


Mortality rate according to employment grade was adjusted for age using direct standardisation with the total study population as the standard. The Cox proportional hazards model was used to study the associations between duration of sickness absence and incidence of mortality. Employees were censored at the time of old age retirement pension, death or loss to follow-up due to emigration. The time-dependent interaction term between duration of sickness absence and the logarithm of follow-up time was non-significant, confirming that the proportional hazards assumption was justified. Hazard ratios and 95% confidence intervals were adjusted for age. Gender interaction effects were tested using Wald tests. Stratified analyses were performed for employees according to occupational grade in three levels at study entry.


A total of 3040 persons (1.3%) died during follow-up. Of these, 876 (28.8%) were women and 2164 (71.2%) were men. The age-standardised mortality rates were higher among men (3.70 per 1000 person-years) than among women (2.53 per 1000 person-years). A significant interaction between gender and duration of sickness absence on mortality was found (Wald χ2 = 19.3, df = 4, p<0.001). Among men, each additional week of absence from the first sickness absence week onwards was associated with a higher mortality rate; among women mortality rate was almost the same for those with 1, 2 or 3 weeks of absence. However, among both men and women ⩾6 weeks of absence was associated with an approximate doubling of the mortality risk compared with 1 week of absence (table 1).

Table 1 Duration of sickness absence compensation in 1998 and mortality in 2001–2004 among 236 207 private sector employees in Denmark

Of the 236 207 employees included in this study, data for employment grades were available for 137 607 who were members of an unemployment fund at baseline. These employees were divided into three occupational grades. As there were no significant interactions between gender and absence duration within occupational grades (Wald χ2 for blue collar workers 7.50, df = 4, p = 0.1116; for intermediate level workers 0.52, df = 4, p = 0.9715; and for blue collar workers 4.35, df = 4, p = 0.3606), grade-specific analyses are shown for women and men combined. The overall age- and gender-standardised mortality rates were largely similar between the three occupational grades, ranging from 3.50 per 1000 person-years in intermediate level occupations to 3.92 per 1000 person-years in white collar occupations (table 2).

Table 2 Duration of sickness absence compensation in 1998 and mortality in 2001–2004 among 137 607 private sector employees in Denmark according to occupational grade

The difference in age- and gender-standardised mortality rates between employees with ⩾6 weeks of absence compared to those with 1 week of absence was 6.9 per 1000 person-years for white collar workers, 5.1 per 1000 person-years for those with intermediate grade occupations, and 5.0 per 1000 person-years for blue collar workers. The corresponding hazard ratios were 3.65, 3.27 and 1.95, respectively. For white collar and intermediate grade occupations, but not for blue collar occupations, there was a graded association between sickness absence duration and risk of mortality.


In the total population of employees working in the private sector in Denmark, 236 207 persons aged 18–65 years experienced periods with SAC in 1998 and were alive at the beginning of the follow-up period 2 years later. Of these, 3040 persons (1.3%) died during the 4-year follow-up, with more men (71.2%) than women (28.8%) dying. Sex-stratified analyses for the total cohort and occupational grade-stratified analyses for 137 607 study participants enrolled in an unemployment fund showed a 2- to >3-fold excess mortality for employees with 6 or more weeks of absence among men, women and in each occupational grade. However, excess mortality risk for shorter durations of absence varied between groups; a graded association between absence duration and risk of death was seen for men and employees in white collar and intermediate occupations. Among women and blue collar workers, absences shorter than 6 weeks did not differentiate mortality risk. To our knowledge, this is the largest investigation of associations between sickness absence and mortality, and also the first to address the question of socio-economic differences in this association and be performed on data covering all private sector employees within a specific country.

The present study was based on register data, which allowed us to avoid sample attrition. Furthermore, as the sex-specific analyses were based on the total population of private sector employees receiving SAC in Denmark, problems such as selection bias into the study are eliminated and the study is representative by definition. The design utilises a 24-month wash-out period in accordance with the maximum possible duration of SAC, thereby avoiding deaths that occurred immediately after the sickness absence assessment period. As a result, sickness absence does not in most cases lead directly into mortality, which could otherwise seriously inflate observed associations. However, this does not rule out that some cases were recurrences of sickness absence. For example, a cancer patient or a person with a psychiatric disorder can be treated in 1 year, experience a period of good health and be able to work, and then suffer a relapse. This would result in the disease exceeding the 24-month wash-out period in this study.

Main messages

  • A clear graded association between increasing length of absence and increasing risk of future mortality was found for men and all occupational grades except blue collar workers.

  • Experiencing more than 6 weeks of sickness absence identified a group at considerable excess risk of death among men and women and in all occupational grades including blue collar occupations.

Policy implications

The use of information on sick leaves may improve the effectiveness of secondary prevention interventions by policy makers, case managers, employers and physicians.

The DREAM register consists solely of recipients of social transfer payments. Therefore, it does not permit study of mortality risk for employees with no sickness absence, but merely allows comparison of risk associated with duration of received SAC. As SAC is registered in DREAM if a person has more than 1 day of SAC in a given week, this could lead to an overestimation with a single day on SAC counting as an entire week. In contrast, there could be an underestimation of sickness absence as a whole, because employers in some cases may not report all sickness absence, usually because of negligence. This could lead to underestimation of the true sickness absence level. Therefore, the study could potentially have fewer cases included than are actually sickness absent. However, this is unlikely to cause a major bias in the analyses of associations in this study, as the magnitude of misclassification is likely to be relatively small and there are no such concerns as regards the reporting of mortality; nor does it affect the implications of using the register information on granted SAC as an indicator of future mortality.

Only 58% of the study population were members of an unemployment fund and thus eligible to enter the analysis of sickness absence and mortality within occupational grades. This could potentially introduce selection bias in the analysis. However, non-membership of unemployment funds is known to be equally distributed across social strata in Denmark, suggesting that interpretation of the information supplied by the 137 607 study participants who were enrolled in an unemployment fund may not be biased.

This study is unique in having sufficient power to examine the association between sickness absence duration and mortality within occupational grades. The observed occupational grade disparities could reflect the fact that white collar workers are capable of working with a broader array of health conditions than blue collar workers: for example, various musculoskeletal symptoms would hinder blue collar workers from participating in physically strenuous work, which would not be the case for most white collar workers. Causes of sickness absence would thus be more severe for white collar than for blue collar workers. Another potential explanation could be differences in norms regarding sickness absence across occupations.

The lack of association between sickness absence and mortality among women for SAC of less than 6 weeks is in line with the findings from the Whitehall II study showing a weaker association between sickness absence and mortality in women than in men.8 Whether the stronger association among men is due to more severe underlying health conditions among male employees than among female employees is unknown. There is a strong body of literature showing that despite their lower mortality female employees have relatively higher risk of sickness absence than male employees.10 14 15 However, there is also evidence suggesting that female-dominated groups tend to develop more lenient norms and standards with regard to sickness absence than are found in more male-dominated contexts.16 These facts together suggest that the sickness absence of male employees is due to more severe health conditions than female employees’ absence of similar duration.

Conclusion and policy implications

These findings from Danish private sector employees indicate that administratively collected data on SAC for periods of 6 weeks or more can be used to identify groups at risk for future excess mortality in men and women and in specific occupational grade levels. Sickness absence duration determines smaller mortality risk differences for men and non-blue collar occupations. The use of information on sick leaves may improve the effectiveness of secondary prevention interventions by policy makers, case managers, employers and physicians. Replication in other countries would be important to determine whether the associations are generalisable across different SAC schemes and population characteristics in contemporary settings.



  • Funding: MK was supported by the Academy of Finland (grants 117604, 124322 and 124271).

  • Competing interests: None.

  • Ethics approval: The study has been notified to and registered by Datatilsynet (the Danish Data Protection Agency). According to Danish law, questionnaire and register based studies do not need approval by ethical and scientific committees or informed consent.