Gender differences at the workplace: sickness absence and productivity loss at work and their association with health and work-related factors

Introduction: Differences in sickness absence and productivity loss at work between men and women are recognized but need to be better understood.

Materials and methods: In a cross-sectional study, 10,407 employees from 37 companies in the Netherlands participated in a web-based health risk assessment, between 2010 and 2014. Self-reported short-term (<9 days) and long-term (10 or more days) sickness absences during the preceding 12 months were assessed. The questionnaire also asked about productivity loss at work, physical and psychosocial work-related factors, and health problems. Log-linear models were used to estimate prevalence ratios (PR), with 95% confidence intervals (CI).

Results: Women, more often than men, experienced short-term and long-term sickness absence (PR 1.06, 95% CI 1.01 – 1.11, and PR 1.33, 95% CI 1.21 – 1.46, respectively) but were less likely to have productivity loss at work (more than 30%), (PR 0.90, 95% CI 0.81 – 0.99). In short-term sickness absence, gender differences were reduced by 20%, after controlling for psychosocial work-related factors, and by 60%, after controlling for health problems, separately. None of the factors mentioned above could explain a large productivity loss at work among men.

Conclusions: Higher prevalence of sickness absence among women can partly be explained by psychosocial work-related factors and health problems. Further studies are needed to explore large productivity loss at work among men.

INTRODUCTION

Reduced productivity at work and sickness absence are considerable threats to sustainable employability. They are associated with substantial costs for employers and society, mainly due to social benefits [1],[2],[3]. It is widely recognized that both the prevalence and duration of sickness absence are higher among female workers than their male colleagues [4],[5],[6]. This difference may be attributed to the higher prevalence of chronic diseases among women, although cardiovascular diseases, as leading chronic non-communicable diseases, are more prevalent among men [4],[7],[8],[9]. However, the explanation of gender differences in sickness absence goes beyond biological differences. They can also be ascribed to different physical and psychosocial working conditions that the workers experience at the workplace, widely recognized as ‘gender differences’ or ‘gender inequality’ [10].

Gender differences are reflected in several work-related issues. A systematic review showed that employed women were more likely to have job insecurity, lower job control, and worse working conditions than men [10]. Unfavorable work-related factors are also associated with chronic health problems and, consequently, with longer and more frequent sickness absence among women [11]. While gender differences in sickness absence have been confirmed in many studies, less is known about productivity loss at work, which is considered another important factor for sustainable employability. Some studies report a higher prevalence of productivity loss at work among women than among men [12], while others have not found any gender differences in productivity loss at work [13]. As reported above, there is evidence that employed women report more health problems and unfavorable work characteristics than their male colleagues. These factors have also been found to be associated with productivity loss at work [14],[15] and may contribute to gender differences in productivity loss at work [14],[15],[16],[17],[18].

It is important to study both sickness absence and productivity loss at work, since both are part of the same decision-making process of workers in distress, whereas men and women might respond differently. We aim to get more insight into the interplay between work-related factors, health problems, and work outcomes among men and women. Therefore, this study investigates gender differences in sickness absence and productivity loss at work and estimates the contribution of health problems and work-related factors to these differences. We hypothesize a higher prevalence and duration of sickness absence and productivity loss at work among women than among men and that both health problems and work-related factors play a role in gender inequalities in these outcomes.

MATERIALS AND METHODS

Study design and population

The population of this cross-sectional study consisted of 34,264 employees from 37 companies in the Netherlands who expressed an interest in being involved in the web-based health risk assessment entitled Prevention Compass (described in the text below). Data were collected in the period between 2010 and 2014. The majority of the individuals invited to participate in the health risk assessment were employed at companies in the public (57.2%), information technology (16.6%), health care (9.5%), and other (16.7%) sectors. The number of men and women in the public sector was equally distributed, whereas in the information technology sector the participants were predominantly men (71.8%), and in the health care sector they were predominantly women (89.6%).

All employees were invited to participate in a webbased health risk assessment (HRA) entitled the Prevention Compass [21]. Invitations to participate were sent by the human resources departments of these companies or by NIPED, the research organization involved in the development and data management of the Prevention Compass. Employees were informed that participation was voluntary, free of charge, and confidential, meaning that no individual results would be shared with their employer or any other third party. In total, 11,962 (34.9%) of the 34,264 employees invited to participate in the health risk assessment completed the HRA. In the current study, we focus on the subset of questions concerning individual characteristics, health-related factors, work-related characteristics (independent variables), as well as sickness absence and productivity loss at work (dependent variables). A total of 1,555 individuals were excluded from the study, based on missing information on productivity loss at work or sickness absence (n = 1,224), psychosocial work-related factors (explained later in the text), (n = 9), physical work-related factors (n = 56), health problems (n = 2), and demographic data (age, gender, occupation, education), (n = 264). After excluding these 1,555 individuals, 10,407 employees (87.0% of the respondents) were included in the analysis, of whom 53.2% were male (n = 5,541) and 46.8% (n = 4,866) were female.

Digital informed consent was obtained from all study participants prior to the study, following requirements for identifiable data collection in the Dutch Code of Conduct for Observational Research.

Sickness absence

Sickness absence was defined as being off work due to the presence of health problems and was derived from the work ability index (WAI) [22]. Sickness absence was self-reported, i.e., the participants were asked to indicate how many days in the preceding 12 months they were unable to work due to health problems, with answers ranging from 0 days to more than 100 days (0 days; 1 – 9 days; 10 – 24 days; 25 – 99 days, and 100 – 365 days). The answers were further categorized into three categories: no sickness absence, 1 – 9 days of sickness absence (short-term sickness absence), and ten or more days of sickness absence (long-term sickness absence) [14].

Productivity loss at work

Self-reported productivity loss at work was measured using the quantity dimension of the Quantity and Quality (QQ) scale [14], which showed a reasonable correlation with objective work output [15],[16]. Respondents were asked to indicate, on a scale from 0 (nothing) to 10 (regular amount), how much work they actually performed during regular working hours, on their most recent regular workday, compared to the expected. Productivity loss at work was classified into three categories: no productivity loss at work (score = 10), 10 – 20% productivity loss at work (score = 8 or score = 9), and 30% or more productivity loss at work (score of 7 or lower) [14].

Physical work-related factors

The following physical work-related factors were assessed by using yes/no questions: 1) exposure to vibrations at the workplace for more than five years, 2) daily lifting, pushing, and/or pulling heavy loads, and 3) working in awkward body postures more than two hours daily [23].

Psychosocial work-related factors

The following five psychosocial work factors were measured with an abbreviated version of a validated Dutch questionnaire on the experience and assessment of work: work demands, autonomy, skills discretion, social support by colleagues, and social support by the supervisor [24],[25].

Work demands were measured using three questions (Cronbach’s alpha = 0.66) concerning too much work to do, the need to hurry, and problems with work pressure. Job autonomy was measured using three questions (Cronbach’s alpha = 0.83) regarding the freedom in carrying out work activities, the influence in the planning of work activities, and personally deciding how much time one needs for a specific activity. Skill discretion was assessed by two questions related to variations in the work and personal input (Cronbach’s alpha = 0.66). Social support by colleagues (three items, Cronbach’s alpha = 0.60) and social support by the supervisor (three items, Cronbach’s alpha = 0.76) were measured with questions about conflicts, the possibility of counting on colleagues/the supervisor when encountering difficulties at work, and the work atmosphere. For all questions, a four-point scale was used with the following ratings: 'never,' 'sometimes,' 'often,' and 'always', in such a way that, for every item, the highest score was assigned to low work demands, high autonomy, high skill discretion, and high social support. For each dimension separately, a standardized sum score ranged from 0 to 100, and workers with a score in the lowest quartile were regarded as exposed to adverse psychosocial risk factors.

Health problems

Health problems were defined to be present when respondents reported that they had been diagnosed with a certain disease, when they were currently taking medication for it and/or when they visited a physician during the preceding 12 months. Cancer was defined as being present when respondents reported they had been diagnosed with this disease and had not been declared cured. Health problems were classified into five groups of most common diseases: cardiovascular (angina pectoris, heart attack, transient ischemic attack, heart failure, cerebrovascular accident, hypercholesterolemia, hypertension, or other heart or cardiovascular disease), musculoskeletal disorders (back, upper and lower extremities), psychological disorders (anxiety disorder, burn out, depression, other psychological disorders), cancer (breast, colon, pharyngeal, urinary tract, or other), and other diseases (including respiratory, gastrointestinal, metabolic and endocrine diseases). Participants could report several health problems simultaneously.

Individual characteristics

In the questionnaire, the respondents were asked to report their gender, age, educational level, marital status, and number of working hours per week. Age was classified into the following three categories: 18 – 34 years, 35 – 49 years, and 50 years and above. Educational level was classified into the following two groups: low (primary school level, lower and intermediate secondary school level, or lower vocational training) and medium/high (higher secondary school level or intermediate vocational school level, higher vocational school level or university). The respondents’ marital status was marked as ‘married’ if they were cohabiting or married.

Statistical analysis

Descriptive statistics was used to report the characteristics of the study population. Spearman’s correlation was used to analyze the correlation between productivity loss at work and sickness absence, as well as between the dependent variables and the potential mediators.

In order to analyze the association between gender (independent variable), work-related factors and health problems, and sickness absence and productivity loss at work (dependent variables), a mediation analysis was performed. Poisson log-linear regression analysis was performed to obtain prevalence ratios (PR) for the associations between the prevalence of individual characteristics, work-related factors, health problems, and sickness absence and productivity loss at work. The association between gender and the dependent variables was investigated first. The association between gender, work-related factors, and health problems was analyzed second. Thirdly, in different models, the work-related factors and health problems were added to the regression model of sickness absence and productivity loss at work, for both genders. In the first model, the association between gender and the dependent variables was only adjusted for age, education, and marital status (Basic Model, Model 1). The second, third, and fourth models consisted of the Basic Model and the following: physical work-related factors (Model 2), psychosocial work-related factors (Model 3), and health problems (Model 4), respectively. The fifth model consisted of the Basic Model, psychosocial work-related factors, and health problems (Model 5). The sixth model contained the Basic Model, physical and psychosocial work-related factors, as well as health problems, together (Model 6). We calculated the percentage change in PRs for adjusted gender associations (Models 1 to 6), using the formula ((PR _{(Basic Model association)} – PR _{(Model 2-6)})/(PR _{(Basic Model association)} -1))100) [26]. All analyses were conducted in SPSS V22.

Table 1. Characteristics of 10,407 participants included in the health risk assessment: individual characteristics, health factors, work related factors, and productivity loss at work

In the text:

RESULTS

The results of the study show that women were more likely than men to report dealing with daily lifting, pushing, and/or pulling of heavy loads, lower autonomy, and lower skill discretion (Table 1). Women were also more likely to report all types of health problems, except cardiovascular problems (women 15.9% vs. men 21.1%). Men were more likely to report exposure to vibrations for longer than five years and lesser social support by their supervisor.

Both short-term (up to 9 days) and long-term (10 days and more) sickness absence was more prevalent among women (41.4% vs. 39.8%, and 14.9% vs. 11.8%), while productivity loss at work of 30% and more was more prevalent among men (13.5% vs. 12.0%). The correlation between sickness absence and productivity loss at work in the total population was low (r_s = 0.09), and even lower between the occurrence of any health problem and the presence of at least one strenuous physical or psychosocial work-related factor, calculated separately for women and men (r_s = 0.05), (not presented in a table).

Short-term and long-term sickness absence

Women more often reported short-term sickness absence (PR 1.08, 95% CI 1.03 – 1.13), as well as longterm sickness absence (PR 1.30, 95% CI 1.18 – 1.43), than men (not presented in a table). All psychosocial factors, except higher job demands, were associated with short-term sickness absence among both men and women, and more strongly with long-term sickness absence (Table 2). Physical factors were not associated with long-term sickness absence and showed contradictory associations for short-term sickness absence among women. While exposure to vibrations increased the likelihood of short-term sickness absence among women (not statistically significant), exposure to awkward back postures or lifting/pushing was associated with a lower likelihood of short-term sickness absence. Regarding health problems, patterns of associations with sickness absence were similar in women and men: the presence of health problems increased the likelihood of sickness absence.

Table 2. Adjusted associations between demographic characteristics, health problems, and work-related factors and sickness absence among women (n = 4,866) and men (n = 5,541)

Small and large productivity loss at work

Small productivity loss at work was not associated with gender. However, in contrast to sickness absence, productivity loss at work of ≥ 30% was reported by women less often than by men (PR 0.90, 95% CI 0.81 - 0.99), (not presented in a table). Psychosocial work factors and health problems, except cancer, were associated with large productivity loss at work, in both men and women (Table 3). The presence of one health problem (mostly a psychological disorder) and, in particular, two or more health problems, was statistically significantly associated with the likelihood of having ≥ 30% productivity loss at work. Cardiovascular health problems were only associated with productivity loss at work among men.

Table 3. Adjusted associations between demographic characteristics, health problems, and work-related factors and productivity loss at work among women (n = 4,866) and men (n = 5,541)

Analysis of gender differences in sickness absence and in productivity loss at work

Adding physical work-related factors to the Basic Model did not explain gender differences in sickness absence (PR 1.06, 95% CI 1.02 – 1.11, Model 2), (Table 4). In comparison, a 20% reduction, i.e., percentage change in PRs between the Basic Model and the additional models, was achieved after additional adjustment for psychosocial work-related factors (Model 3), and 60% after adjustment for health problems (Model 4). A total of 80% of the gender difference was explained by individual characteristics, psychosocial work-related factors, and health problems (Model 5, PR 1.01, 95% CI 0.97 – 1.06). Adding physical work-related factors to this model reduced the mediating influence of all factors together to 60% (Model 6). Similar results were obtained for gender differences in long-term sickness absence (Table 4).

Table 4. Multinomial regression analysis of associations between gender, sickness absence and productivity loss at work, adjusted for a number of factors

Compared to men, women were less likely to report higher productivity loss at work (PR 0.87, 95% CI 0.78 – 0.96), regardless of age, education, ethnicity, and marital status. Further adjustment for physical and psychosocial work-related factors and health problems did not explain these gender differences.

DISCUSSION

This study shows that women reported more sickness absence but less productivity loss at work than men. Almost one fifth (19%) of the difference in long-term sickness absence between women and men could be explained by psychosocial work-related factors, and almost 70% (69%) by health problems. The combination of psychosocial work factors and health problems could explain roughly 80% of the gender difference in sickness absence. In contrast, health problems or work-related factors did not explain gender differences in large productivity loss at work.

The major strength of this study is the size of our study population (n = 10,407). Another strength is that this study provides insight into sickness absence and productivity loss at work simultaneously. Since just sickness absences (but not productivity loss at work) are usually noticed and registered at the workplace, this could create a blind spot for employers.

The results regarding sickness absence are in line with a consistently reported higher prevalence of sickness absence among women in other studies. This is mainly mediated by higher morbidity among women, as compared to men [6],[27],[28],[29],[30]. However, it seems that factors other than morbidity or occupational environment influence differences in productivity loss at work between men and women. Higher productivity loss at work by men, as compared to women, was also reported by Van den Heuvel [31], but not in some other studies [32],[33],[34]. Altogether, differences in results regarding productivity loss at work seem to be related to whether measured productivity loss at work was focused on health-related productivity loss at work or productivity loss at work in general. This finding might confirm the theory that there are different mechanisms of responses to distress among women and men, which is described as ‘illness behavior’ [35]. In our study, women were more exposed to distress by the perceived occupational environment (psychosocial work-related factors). However, it seems that, with regard to productivity loss at work, these factors do not make a difference in women, as compared to men.

Increased sickness absence and productivity loss at work were found for all the studied health problems, as compared to individuals without a particular health problem. This is in line with other publications on various medical conditions, such as rheumatoid or inflammatory arthritis [36],[37], diabetes [38],[39], psychological disorders or depression [29],[40], and cancer [41]. However, some authors found no gender differences in sickness absence or health-related productivity loss at work [37], while others reported that women suffered more loss than men [29],[42]. There is also evidence that shift work affects women more negatively than men when productivity loss at work is concerned [43].

Psychosocial work-related factors and health problems could largely explain differences between men and women regarding longer sickness absence, but they did not explain gender differences regarding productivity loss at work. The results may suggest that women tend to choose to call in sick, in case of unwell-being as the result of work-related or health factors, while men more often stay at work, although their productivity goes down. In other studies, combining paid work with domestic and caregiver tasks and responsibilities was mentioned as potentially affecting women more than men in their choices regarding sickness absence [34]. Our study suggests that gender differences in sickness absence are mainly due to occupational factors and health. However, gender differences regarding large productivity loss at work could not be explained by these factors, so further research is needed in this field, especially because productivity loss at work is associated with significant costs and is known as a predictor of future long-term sickness absence [12]. To improve sustainable employability at the level of both individuals and organizations, the focus should be on the prevention and adequate treatment of health problems and on addressing psychosocial factors at work. Additionally, bearing in mind the fact that women, in general, work fewer hours per week than men, and have less productivity loss at work, the practical implication could be considering shortening the expected working week. A systematic review conducted by Voglino et al., in 2022, already proved the positive impact of reduced working hours on the quality of life at work, sleep, and stress [44]. However, work flexibility, which combines work from home and at the office, might have a different impact on women and men [45], and international, culture-specific research might inform national decision-makers about the optimal working schemes.

Many intervention studies taking place in the work environment are addressing these factors, and their results are promising, although there are many different ways of studying changes in psychosocial work-related factors, and a lot of gaps in knowledge still exist [46],[47],[48],[49]. In addition, workplace interventions to improve employees' health and reduce sickness absence should look at men and women separately, in order to realize and maximize their effectiveness and cost-effectiveness.

This study has some limitations as well. Due to the cross-sectional design, we cannot assume any causal, one-way relationship between exposure to adverse psychosocial work-related factors, health problems, and sickness absence, as they actually might be reciprocal to each other. Another concern is the fact that our data on sickness absence were based on self-reporting and might be subject to recall bias. Finally, women in the Netherlands work fewer hours per week than men. Since we did not take work hours per week into account, the weekly cumulative work exposure for women could have been lower than for men. Therefore, if women were equally exposed as men, the cumulative exposure would, in fact, become higher, and future studies should take this into account.

CONCLUSION

Our study suggests that gender differences in sickness absence are mainly due to occupational factors and health. However, gender differences in large productivity loss at work could not be explained by these factors, so further research is needed in this field. Gender-sensitive policies might be needed to recognize the differences in behavioral patterns between men and women.

LIST OF ACRONYMS

IT        Information technology
HRA    Health risk assessment
QQ      Quantity and Quality
WAI    Work ability index
PR       Prevalence ratio