Gender differences in long term sickness absence

In The Netherlands, it is a requirement for all employers to ensure that their employees have access to occupational health care, which is typically provided by an occupational health service (OHS). An OHS is responsible for registering sickness absences, and for providing guidance to sick-listed employees by medical consultations and advice for returning to work (RTW). When an employee reports sick, the OHS registers this in the sickness absence register. Sickness absence can be due to any (i.e., work-related and non-work-related) physical or mental illness or injury. In The Netherlands, the employer financially compensates sickness absence for a period of 104 weeks. Most employers cover 100% of the worker’s salary in the first year of sickness absence and 70% in the second year. The OHS follows employees during 104 weeks of sickness absence, after which the employee may apply for a disability pension provided by the Employee Insurance Agency (UWV) and the employer may end the job contract.

For our study we retrieved data from a sickness absence register of a large Dutch national OHS, registering sickness absence data of approximately 1.24 million Dutch employees from about 11.6 thousand companies of various economic sectors throughout the country. The dataset included all reported employee sickness cases from January 2010 to December 2020. When an employee experienced multiple periods of sickness absence, we included each separate period in the dataset as a single case. For each case, we calculated the duration of sickness, defined as the interval from the first to the last registered day of sickness absence. For our study we included cases aged between 16 and 70 with a sickness duration between 1 day and 104 weeks. We excluded cases diagnosed as pregnancy and pregnancy-related diseases.

Most of the sickness absence cases we observed were short-term, typically lasting less than two weeks. These short-term absences were commonly due to medical conditions like upper respiratory infections or gastrointestinal disturbances. For longer sickness absence periods, employees consult an occupational health physician (OHP) for return-to-work (RTW) advice. In the Netherlands, it is mandatory for employees on sick leave to consult an OHP within 42 days of their absence. The OHP then documents the diagnosis in the Occupational Health Service (OHS) register. To classify the employees’ diagnoses, the OHS uses the Dutch classification system for Occupational and Social insurance physicians (CAS) [20]. This system is based on the ‘International Statistical Classification of Diseases and Related Health Problems’ (ICD-10) and contains similar main categories. An important distinction between the CAS system and the ICD10 is the classification of neoplasms. In the CAS system, neoplasms are categorized under the relevant organ system, whereas in the ICD10, they are considered a distinct class [21].

For our detailed analysis, we utilized cases that had been OHP diagnosed, excluding cases with unknown diagnoses.

Statistical analyses were done using the lifelines library for survival analysis and Python 3.10 [22]. Employees were right-censored when the job contract ended during the sickness absence. Data about sickness duration were analyzed descriptively using the mean, median, and standard deviation.

Descriptive statistics were computed for each diagnostic category and gender. For each diagnostic category, the difference in means between genders was calculated. We focused on differences in means including confidence intervals instead of applying tests of significance. This is because the latter are heavily influenced by sample size and will almost always demonstrate a significant difference, even for small differences that may not have practical significance [23].

We have analyzed the most important causes for sickness absence, in terms of both frequency and duration, in more detail. For these diagnostic categories, hazard rates have been determined using the Nelson-Aalen estimator [24]. The estimated hazard function gives the recovery rate at each point in time t, and is defined as the probability that an employee will recover in the next moment \(t + 1\), conditional on the fact that the employee was still sick at time t. For instance, if 100 employees are absent at the start of a day, and 2 have recovered by the start of the next day, the daily recovery rate is 0.02.

Considering that the onset and recovery of sickness absence are not evenly distributed across weekdays, with a higher percentage of employees reporting recovery on Mondays, we used a one-week moving average filter to smooth hazard the rate \(h_t\) at each point in time t.

In the period between January 2010 and December 2020 there were 4,998,455 sickness cases that fulfilled our inclusion criteria, cf., Fig. 1. Of these cases 52% were male, and 48% were female. This closely reflects the male/female ratio of the Dutch working population during the same period (about 53% male and 47% female) [25]. The mean absence duration was 23 ± 42 days. Among these approximately 5 million cases, approximately 11% (n = 562,395) were consulted and diagnosed by an occupational health physician (OHP). For further analysis, we excluded cases where employees were diagnosed with an unknown code (n = 6).

Tables 1 and 2 present descriptive statistics regarding sickness absence duration for men and women across various diagnostic categories. The most prevalent causes of sickness absence were diseases of the musculoskeletal system, mental disorders, diseases of the respiratory system, and diseases of the digestive system. For women, the most prevalent diagnoses were mental disorders, whereas for men diseases of the musculoskeletal system were most prevalent. Across all OHP diagnosed cases, the mean sickness absence duration was 158 days for women and 117 days for men, with an average gender difference of 41 days (95% confidence interval: 40.3-42.0). Among the most prevalent causes, mental disorders had the longest sickness absence duration (186 ± 162 days).

Except for diagnoses of diseases of the blood and blood forming organs, women had a longer average sickness absence duration than men. For each specific diagnosis, the 95% confidence interval indicates that the results are not only statistically significant but also practically relevant. The largest gender differences in sickness absence duration were found for cases diagnosed with diseases of the musculoskeletal system and cases diagnosed with diseases of the genitourinary system.

The most important causes for sickness absence, in terms of both frequency and duration, were diseases of the musculoskeletal system, mental disorders, diseases of the nervous system and diseases of the circulatory system. We have explored these diagnostic categories in more detail.

Figure 2 displays the recovery rates for these diagnostic categories. For male employees with mental disorders, the recovery rate remains rather constant during the initial 1.5 years. This suggests that the conditional probability of recovery does not change during this period and is independent of the actual sickness absence duration. In contrast, for female employees with mental disorders, we observe considerably lower recovery rates in the first few months. This indicates that a smaller proportion of women recover during this early period relative to men.

For diseases of the musculoskeletal system, the gender difference in recovery rate is even more pronounced, particularly in the initial few months, where the percentage of women reporting recovery is relatively low. However, after approximately three months, the recovery rates become comparable between women and men. For diseases of the nervous system we observe a small difference in recovery rate during in the initial stage of sickness absence. The recovery rates for diseases of the circulatory system are similar for men and women.

Table 3 shows the proportion of cases with either mental or musculoskeletal disorders that have recovered at three different points in time (13, 26, and 52 weeks). We see that the proportion of recovered cases differs most among genders within the first three months. For instance, after 3 months, 67% of men with musculoskeletal disorders have recovered, versus 51% of women. For mental disorders, the proportions are 44% for men and 37% for women.

Besides gender differences the recovery curves show another interesting pattern around one year of sickness absence. Around that period, we observe a sudden increase in recovery rate for both men and women. This might be an effect of a decrease in salary in the second year of sickness absence, as most employers decrease the salary to 70% after the first year of sickness absence. We will investigate this topic in more detail in a future paper.

Various gender-related factors may contribute to differences observed in absence duration and recovery rate, including medical, biological, personal, family and work-related factors [12]. We will investigate some of these factors and explore how these could account for the observed delay in recovery for women compared to men.

Person-related factors Person-related factors, such as coping style and work attitude, can also play a role in sickness absence. Tamres et al. (2002) conducted a meta-analysis of 50 studies on gender differences in coping and identified 17 coping strategies which they classified as problem-focused or emotion-focused behaviors [28]. Problem-focused coping strategies include active behaviors (such as changing the situation, removing the stressor), planning (review possible solutions), seeking instrumental social support that is directed towards solving problems, and general problem-focused behavior. Emotion-focused behaviors aim to alter the response to the stressor and include seeking emotional support, avoidance, denial, positive reappraisal, isolation, venting, rumination, wishful thinking, self-blame, positive self-talk, and exercise. According to the study, women tend to make more use of coping strategies compared to men, particularly more emotion-focused strategies. Van Rhenen et al. (2008) investigated the role of different coping styles on both the duration and frequency of sickness absence [29]. They found that both the use of an active problem-solving coping style and seeking emotional support decreased the mean sickness absence duration, with a stronger effect of the problem-solving coping style. Other emotion-based strategies had either no effect (expression of emotions) or a negative effect (avoidance) on sickness absence duration. Conversely, Loset et al. (2018) conducted a survey study to explore differences in attitudes and norms regarding sickness absence and found no significant differences between genders [30].

Further research is necessary to investigate variations in coping styles throughout the entire period of sickness absence, with particular emphasis on potential differences in coping styles during the initial stage of sickness absence and on changes in coping styles after these initial period.

Daily life characteristics Several differences in daily life and occupational characteristics may also influence sickness absence frequency and duration [13, 15‐17]. Women generally tend to spend more time to household tasks and childcare compared to men. The double burden hypothesis proposes that the combination of different roles, such as being an employee and a parent, can increase stress and consequently increase the risk for sickness absence. The strain associated with having multiple roles can be reflected by perceived work-family conflicts, where the demands of one’s professional role interfere with their family role, or vice versa. In a systematic review by Nilsen et al. (2017), they found moderate evidence for a positive correlation between work-family conflict and subsequent sickness absence, indicating that the strain from balancing work and family roles can indeed lead to higher levels of sickness absence [17]. However, the evidence was insufficient to draw conclusions about the role of gender in the prospective association between work-family conflict and subsequent sickness absence.

In our study we found that gender differences in particular influence the recovery rate during the first months of sickness absence. The work-family conflict provides a possible explanation for this observed differential recovery. The strain associated with managing different roles may affect the recovery process differently for men and women. For instance, during the initial recovery phase, women might prioritize resuming their family and social roles, while men may focus more on returning to their professional roles. This potential variance in prioritization could account for the lower recovery rates among women during the early stages of sickness absence. Further research is necessary to investigate the relationship between work-family conflict and recovery rates during the complete sickness absence period.

Occupational characteristics In addition to daily life characteristics, sector-specific gender representation patterns may also play a role. Women are predominantly employed in the healthcare, social services, and education sectors, whereas men are more commonly found in industries such as construction, manufacturing, information technology, and transportation [31]. Some studies suggest that higher rates of sickness absence are associated with occupations dominated by women [32, 33]. The emotionally demanding nature of jobs in healthcare and social services often entails working directly with patients or clients. Such roles may require a more complete recovery from mental disorders before work can be resumed. In contrast, physically demanding work might offer a distraction from mental health issues, which could partly explain why men might return to work sooner.

However, studies about the association between job occupation and sickness absence are not conclusive. In a recent study Østby et al. (2018) found no evidence that the type of occupation is related to gender differences in sickness absence [15]. Mastekaasa (2014) even found an increase in gender differences when adjusting for the type of occupation [11].