Objectives Low socioeconomic position is consistently associated with higher rates of sickness absence. We aimed to examine whether working conditions, health-related behaviours and family-related factors explain occupational class differences in medically certified sickness absence.
Methods The study included 5470 women and 1464 men employees of the City of Helsinki, surveyed in 2000–2002. These data were prospectively linked to sickness absence records until the end of 2005, providing a mean follow-up time of 3.9 years. Poisson regression was used to examine the occurrence of medically certified sickness absence episodes lasting 4 days or more.
Results Medically certified sickness absence was roughly three times more common among manual workers than among managers and professionals in both women and men. Physical working conditions were the strongest explanatory factors for occupational class differences in sickness absence, followed by smoking and relative weight. Work arrangements and family-related factors had very small effects only. The effects of psychosocial working conditions were heterogeneous: job control narrowed occupational class differences in sickness absence while mental strain and job demands tended to widened them. Overall, the findings were quite similar in women and men.
Conclusions Physical working conditions provided strongest explanations for occupational class differences in sickness absence. Smoking and relative weight, which are well-known determinants of health, also explained part of the excess sickness absence in lower occupational classes. Applying tailored work arrangements to employees on sick leave, reducing physically heavy working conditions and promoting healthy behaviours provide potential routes to narrow occupational class differences in sickness absence.
- Socioeconomic factors
- sickness absence
- working conditions
- health behaviours
Statistics from Altmetric.com
Sickness absence has been increasingly used as an outcome in health research. One of the most consistent findings concerning sickness absence is that those in lower occupational classes have more sickness absence than those in higher classes.1 For instance, in Finland municipal employees in manual occupations have two to three times more medically certified sickness absence episodes than managers and professionals.2 A difference of this size is well in line with the evidence from other countries.3 Understanding the reasons for such large a difference is important, since the findings imply that the lower occupational classes are likely to have the largest potential for reducing sickness absence.
There are three previous longitudinal cohort studies that have examined whether various factors explain occupational class differences in sickness absence. In the UK, a study on civil servants found that health-related behaviours, psychosocial working conditions, job satisfaction and adverse social circumstances outside work accounted for a third of occupational class differences in both self-certified and medically certified sickness absence.4 All explanatory factors were adjusted for simultaneously, and it is not possible to determine which ones were most important. Physical working conditions were not included in the study. A French study on national gas and electricity company employees examined whether physical and psychosocial working conditions explain occupational class differences in medically certified sickness absence of any length.5 After adjusting for age, demographic characteristics and health-related behaviours, manual workers and clerical employees had three times higher sickness absence rates than managers. Adjusting for all working conditions simultaneously reduced the occupational class gradient in all-cause sickness absence by 16% in men and 25% in women. A nationally representative Danish study examined whether occupational class differences in sickness absence could be explained by health-related behaviours, physical working conditions and psychosocial working conditions.6 A strong occupational class gradient in sickness absence exceeding eight consecutive weeks was found, adjusting for age and family status. Adjusting for health-related behaviours attenuated the association by 5–18%. Additional adjustment for physical working conditions attenuated the original association by up to 57% among unskilled and semi-skilled male workers. Further adjustment for psychosocial working conditions had a minor effect in women only. Furthermore, two cross-sectional studies, one in Norway7 and one in France,8 have examined different working conditions as explanations for occupational class differences in sickness absence.
The aim of this study was to explain occupational class differences in medically certified sickness absence using a large prospective dataset, with register-based sickness absence records on middle-aged women and men employed by the City of Helsinki. Previous studies suggest that working conditions are likely to partly explain higher sickness absence rates in the lower occupational classes, but the contribution of different types of working conditions and their relative importance as compared to other types of explanatory factors remains unclear. We studied various kinds of working conditions, health-related behaviours and family-related factors as potential explanations for occupational class differences in sickness absence.
Employing nearly 40 000 persons, the City of Helsinki is the largest employer in Finland.9 In 2000, the personnel register of the City was used to identify all employees who, during that year, reached the age of 40, 45, 50, 55 or 60, and a self-administered questionnaire was mailed to them. Similar procedure was repeated in 2001 and 2002. Altogether, these three independent cross-sectional surveys included 13 346 persons of whom 67% responded. According to non-response analysis, the data are generally representative of the target population, although men, younger people and manual workers were slightly under-represented among the respondents.10
The survey data were prospectively linked with the employer's sickness absence records using the unique personal identification number assigned to each Finnish citizen. However, the linkage was not possible for 22% of respondents who did not provide a written consent for register linkages when returning the questionnaire. Our non-response analysis showed only small differences in consenting to register linkages by age, occupational class, income, type of employment contract and employment sector, and associations of these characteristics with sickness absence by responding and consenting.10
Among Finnish municipal employees, a medical certificate is required for sickness absence lasting 4 days or more. In this study, the number of medically certified sickness absence episodes during the follow-up period was used as the outcome variable. Supplementary analyses were made for self-certified absence episodes of 1–3 days as well as medically certified absence episodes >2 weeks. Results of these supplementary analyses are not shown in tables but are commented in the text. The follow-up time was started from the day the respondent returned the questionnaire and continued until the end of 2005 or until the work contract had terminated. All interruptions in working due to reasons other than own illness, such as maternal or parental leaves, were subtracted from the follow-up time. The overall number of person-years in the analyses was 28 198 with a mean follow-up time of 3.9 years.
Measurement of occupational class and the explanatory variables
Occupational class was divided into four hierarchical groups: managers and professionals, semi-professionals, routine non-manual employees and manual workers. Manual workers and non-manual employees were distinguished using the socioeconomic classification of Statistics Finland,11 and the occupational classification of the City of Helsinki was used to further divide non-manual employees into three groups on the basis of proficiency requirements and supervisory status. Managers have subordinates and do managerial or administrative work. Professionals include employees with a university degree, such as teachers and medical doctors. Semi-professionals include occupations like nurses and technicians. Routine non-manual employees include clerical employees and non-professional occupations within the social and health care, such as child-minders and home assistants. Typical manual occupations include cleaning, kitchen work and public transport.
Among the potential explanatory factors, a special emphasis was put on working conditions. Three kinds of working conditions were included: work arrangements, physical working conditions and psychosocial working conditions. Furthermore, health-related behaviours and family-related factors were examined.
Work arrangements included three measures: Those who worked >40 h per week were classified as working overtime. Shift workers included all those who did not do regular daytime work. Permanent and temporary employees were separated using the type of work contract.
Physical working conditions included four measures: from an inventory consisting of 18 work characteristics and working environment factors, three measures were constructed using factor analysis.12 The first measure indicates physical work load—for example, repetitive trunk rotation, repetitive movements and lifting and carrying. The second measure indicates exposure to hazardous substances—for example, dirt and dust, solvents, dampness and wetness. The third measure indicates sedentary work and computer use. Furthermore, a single-item question asking how physically strainful the respondents considered their work was included.
Psychosocial working conditions included six measures: mental strain was asked with a question parallel to that of physical strain. Job demands and job control were measured using the Framingham version of Karasek's job strain questionnaire.13 Social support at work was measured by an index constructed from four workplace-related questions of the Sarason inventory.14 Job dissatisfaction was measured with a single-item question with seven response alternatives. A question asking whether bullying existed at the workplace was used as a measure of workplace climate.
Health-related behaviours included four measures: Smoking was divided into heavy smokers, moderate smokers, quitters and never smokers. Alcohol use was divided into non-drinking, moderate drinking and high drinking using the cut-point of 14 weekly drinks for men and seven weekly drinks for women. Physical activity was assessed using the MET index and divided into quartiles.15 Relative weight was measured by body mass index and classified as underweight, normal weight, overweight and obesity.
Family-related factors included six measures: marital status was divided into three categories: married or cohabiting, single and previously married. The number of children aged <18 years was categorised as 0, 1 and 2 or more. Work–home conflict was measured with an item asking satisfaction with combining paid work and family life. Social networks were measured by asking how often the respondents met friends or relatives outside the nuclear family. Social support was measured by the full Sarason inventory.14 A sum of negative life events in the family during the preceding year was calculated from six items such as divorce or separation and serious illness of a family member.
Sickness absence rates for medically certified sickness absence episodes were first calculated in the four occupational classes separately for women and men. The rates are reported per 100 person-years.
The effects of working conditions, health-related behaviours and family-related factors on occupational class differences in sickness absence were examined using Poisson regression. Differences in the individual follow-up times were taken into account using the logarithm of the time until censoring as the offset. Overdispersion was corrected by scaling.16 Age-adjusted rate ratios with 95% CIs were first calculated comparing the other occupational classes to managers and professionals. The effects of individual working conditions, health-related behaviours and family-related factors on the association between occupational class and sickness absence were then analysed by adjusting for these factors one at a time. Because the measures within each group of explanatory factors are correlated and may overlap, we also simultaneously adjusted for all factors within the group regardless of whether they individually affected the association or not. Finally, all explanatory factors were adjusted for simultaneously. All analyses were conducted separately for women and men using SAS V. 8.02 for Windows (SAS Institute Inc).
The study population included 5461 women and 1463 men, reflecting the excess of women among the employees of Finnish municipalities, with the average age of 49 years (SD 6.6 years; table 1). Women were most likely to work in routine non-manual occupations, while men were most likely to work as managers and professionals or manual workers. There was a strong association between occupational class and occurrence of sickness absence in both women and men: in women, the number of sickness absence episodes for 100 person-years ranged from 46 among managers and professionals to 118 among manual workers, while in men, the corresponding range was from 31 to 100.
Table 2 shows relative differences in medically certified sickness absence in other occupational classes compared to managers and professionals among women. A strong graded association between occupational class and sickness absence was found. Manual workers had nearly three times more sickness absence than managers and professionals. Adjusting for work arrangements had practically no effect on this association. Adjusting for physical working conditions attenuated the risk of sickness absence in all occupational groups compared to managers and professionals by about 40%. Physical work load and physical strain had the largest effects. The total effect of psychosocial working conditions on occupational class differences in sickness absence was negligible. However, adjusting for job control attenuated the difference, while job demands and mental strain increased it. Adjusting for smoking and relative weight attenuated occupational class differences by one fifth. Family-related factors had negligible effects in women.
Among men as well, occupational class was strongly associated with medically certified sickness absence (table 3). Adjusting for shift work slightly attenuated the association. Physical working conditions accounted for two thirds of the excess risk for sickness absence among manual workers and one third among routine non-manual employees and semi-professionals. Hazardous exposures had the strongest effect. Among psychosocial working conditions, adjusting for job demands and mental strain widened occupational class differences in sickness absence. Adjusting for the other psychosocial working conditions slightly narrowed the differences, but in general, psychosocial working conditions rather widened than narrowed the differences. Smoking and relative weight attenuated occupational class differences in sickness absence by up to one fourth. Among family-related factors, social support and having children in the family slightly narrowed occupational class differences in sickness absence among men.
Supplementary analyses were conducted for self-certified sickness absence episodes of 1–3 days and medically certified sickness absence episodes >2 weeks (results available from the authors). There were occupational class differences in sickness absence episodes of all lengths. The same factors explained occupational class differences in sickness absence episodes of all lengths. Physical working conditions were most important, followed by smoking and relative weight. The effects, especially those of physical working conditions, were slightly weaker for self-certified absence episodes. Among women, also work arrangements, psychosocial working conditions and family-related factors slightly contributed to the explanation of occupational class differences in self-certified sickness absence episodes.
There was a strong association between occupational class and occurrence of medically certified sickness absence in both women and men. Sickness absence was roughly three times more common among manual workers than among managers and professionals. The association was slightly stronger in men than in women. Physical working conditions were the strongest explanatory factors for occupational class differences in sickness absence, followed by health-related behaviours, especially smoking and relative weight. The effects of psychosocial working conditions were heterogeneous: adjusting for job control narrowed occupational class differences in sickness absence, while mental strain and job demands tended to widen them. The effects of work arrangements and family-related factors were very small.
While in both women and men, physical working conditions were the strongest explanatory factors for occupational class differences in sickness absence, there was, however, heterogeneity in the explanatory factors within this group. Among women, physical work load and physical strain were the strongest explanatory factors, and their adjustment reduced sickness absence in all occupational classes compared to managers and professionals. Among men, hazardous exposure was the most important explanatory factor, and its effect was clearly strongest among manual workers. There were no differences in sickness absence rates between routine non-manual employees and manual workers after adjusting for physical working conditions among men. Previous cross-sectional studies have also shown that physical working conditions are important explanatory factors for occupational class differences in sickness absence.7 8
Except for manual workers, physical working conditions explained somewhat more of occupational class differences in women than in men. Because of the occupational sex segregation, the distribution of occupational class and the allocation of occupations within these classes differ between women and men. In our data, the routine non-manual class including occupations in nursing and child care is much larger in women than in men. Many of these occupations include physically heavy work tasks. Also, the manual class includes physically heavy women-dominated occupations like kitchen work. A previous Danish study6 found higher sickness absence rates in lower non-manual employees than among skilled manual workers among women. The lower non-manual class included large occupational groups of nursing and teaching. A corresponding observation was made in the supplementary analyses from our data (not shown) for self-certified sickness absence episodes of 1–3 days, with the routine non-manual class having the highest sickness absence rates in both women and men. Because of its large size the routine non-manual class is of great importance among women.
Of the other work-related factors, work arrangements had only very small effects on occupational class differences in sickness absence. This agrees with previous cross-sectional studies using similar kinds of measures.7 8 The total effect of psychosocial working conditions was also small, but the effects of individual measures were heterogeneous: adjusting for job control narrowed occupational class differences in sickness absence while mental strain and job demands tended to widen them. In a Danish study,6 adjusting for psychosocial working conditions on top of health-related behaviours and physical working conditions slightly narrowed occupational class differences in sickness absence among women. Other previous longitudinal studies have included psychosocial working conditions simultaneously with other kinds of factors, and their individual contributions cannot be assessed.4 5 However, apart from job control, their effects appear to be quite small,5 7 8 although larger effects may be found for special diagnostic groups like mental disorders.5
Of the non-work-related factors, smoking and relative weight contributed to occupational class differences in sickness absence. A Danish study6 showed somewhat smaller effects for health-related behaviours than our study: adjusting simultaneously for the same health-related behaviours than our study reduced the occupational class differences in sickness absence by 5–18%. In a British study,4 smoking and frequent alcohol consumption partially explained occupational class differences in sickness absence, but their individual effects were not identified. Our supplementary analyses for self-certified sickness absence episodes and medically certified sickness absence episodes >2 weeks showed roughly similar effects for health-related behaviours as the analyses for medically certified sickness absence of any length.
With a large variety of potential explanatory factors, a total of 40–60% of occupational class differences in medically certified sickness absence could be explained. The unexplained proportion might be due to inadequate measurement of the studied explanatory factors which would be likely to reduce their explanatory power. Another possibility is that there are other factors that explain the remaining differences. Health status differences between occupational classes were not considered in this study concentrating on social and structural explanations. Furthermore, attitudes and values concerning sickness absence may differ between occupational classes. In lower occupational classes, norms might be more tolerant and less severe illnesses considered as legitimate reasons for sickness absence. Those in upper occupational classes may have higher internal motivation and commitment to work which may make working while ill more common than in lower classes.17 In this study population, the requirements for sickness certification are the same in all occupational classes. However, it is possible that those in higher classes may be more often absent from work without being sick-listed, although this is likely to affect more short absences not requiring a medical certificate.
Sickness absence has been increasingly used as an outcome in health research. Long-term sickness absence is a generic health measure that indicates chronic morbidity and predicts future disability pension as well as mortality.18 19 An advantage of this measure is that absence episodes requiring a medical certificate from a physician may be more objective than many other commonly used measures of general health that are often based on self-reports, although peoples' own estimation of their condition is likely to play a major role also when a medical certificate is issued. However, sickness absence as a health measure has the disadvantage that it may be more than many other outcomes affected by contextual and environmental factors. In particular, working conditions may contribute more to occupational class differences in sickness absence than in other health outcomes since similar impairment may cause inability to carry out one's work tasks in some occupations but not in others. Working conditions20 21 and health behaviours22 23 have been found to explain occupational class differences in other health outcomes as well. However, exact comparison for explanatory factors in sickness absence and other health outcomes is difficult to make.
Our study was made among municipal employees in an affluent economy with a tradition of employee rights. The data included a large number of blue-collar and white-collar occupations, but generalisability of these results to the private sector is not warranted since the nature of work, occupational structures and social security schemes differ from those of the public sector. Furthermore, information on working conditions, health behaviours and family-related factors were based on self-reports. Some of the explanatory factors were measured by single-item questions which may have reduced their measurement validity. The response rate to the surveys was 67%, and the number of eligible respondents was further reduced because all respondents did not provide a consent for the register linkage. However, the non-response analysis suggests that this may not seriously affect the results.10
This study included a large sample, register-based data on sickness absence and a prospective study design. A broad variety of potential explanatory factors was included. These factors explained a large part of occupational class differences in medically certified sickness absence. The explanation was largely attributable to physical working conditions as well as smoking and relative weight. Thus, directly work-related factors and broader, culturally determined factors contribute to occupational class differences in sickness absence. Applying tailored work arrangements to the employees on sick leave, reducing physically heavy working conditions and promoting healthy behaviours are likely to be most effective ways to narrow occupational class differences in sickness absence and to reduce sickness absence rates in general.
What is already known on this subject
Low socioeconomic position is consistently associated with higher rates of sickness absence.
The contribution of different types of working conditions on social class differences in sickness absence and their relative importance as compared to other types of explanatory factors remains unclear.
What this study adds
Physical working conditions were strongest explanatory factors for occupational class differences in medically certified sickness absence, followed by smoking and relative weight.
Psychosocial working conditions had small effects and they were in opposite directions. Work arrangements and family-related factors had very small effects only.
Funding Academy of Finland.
Competing interests None.
Patient consent Obtained.
Ethics approval This study was conducted with the approval of the University of Helsinki and the City of Helsinki Health Authorities.
Provenance and peer review Not commissioned; externally peer reviewed.
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.