Objectives To examine the associations between socio-economic position (SEP) and the onset of psychiatric work disability, return to work and recurrence of disability.
Methods Prospective observational cohort study (1997–2005) including register data on 141 917 public-sector employees in Finland. Information on International Classification of Diseases, 10th Revision diagnosis-specific psychiatric work disability (≥90 days) was obtained from national registers.
Results During a mean follow-up of 6.3 years, 3938 (2.8%) participants experienced long-term psychiatric work disability. Of these, 2418 (61%) returned to work, and a further 743 (31%) experienced a recurrent episode. SEP was inversely associated with onset of disability owing to depressive disorders, anxiety disorders, personality disorders, schizophrenia and substance-use disorders. No association was found between SEP and disability owing to bipolar disorders or reaction to severe stress and adjustment disorders. High SEP was associated with a greater likelihood of a return to work following depressive disorders, personality disorders, schizophrenia and substance-use disorders, but not bipolar disorders, anxiety disorders or reaction to severe stress and adjustment disorders. Low SEP predicted recurrent episodes of work disability.
Conclusions High SEP is associated with lower onset of work disability owing to mental disorders, as well as return to work and lower rates of recurrence. However, the socio-economic advantage is diagnosis-specific. SEP predicted neither the onset and recovery from disability owing to bipolar disorders and reaction to severe stress and adjustment disorders, nor recovery from disability owing to anxiety disorders. SEP should be taken into account in the attempts to reduce long-term work disability owing to mental disorders.
- mental health
- return to work
- longitudinal studies
Statistics from Altmetric.com
- mental health
- return to work
- longitudinal studies
What this paper adds
An inverse socio-economic gradient in mental disorders is well documented but the association with the onset, recovery and recurrence of diagnosis-specific work disability is not known.
This study showed that the incidence of long-term psychiatric disability was 2.8%. Of the disabled, 61% returned to work, and a further 31% experienced a recurrent episode. A high variation between diagnoses was found in the onset and recovery rates.
High socio-economic position was associated with lower onset of work disability, faster return to work and lower rates of recurrence.
However, the socio-economic advantage was diagnosis-specific. Socio-economic position predicted neither the onset and recovery from disability owing to bipolar disorders and adjustment disorders, nor recovery from disability owing to anxiety disorders.
Socio-economic position should be taken into account in the attempts to reduce long-term work disability owing to mental disorders.
Psychiatric work disability has important implications both for worker productivity and for individual quality life. Based on benefits paid out, there is an increasing trend in psychiatric work disability throughout Europe and North America.1 Although it is well established that a higher socio-economic position (SEP) is protective against the onset of mental disorders,2 less is known about the extent to which this gradient translates into socio-economic differentials in work disability and, specifically, how the patterns vary by specific psychiatric diagnoses.3–5
Previous investigations have shown either an inverse association6 7 or no association8 9 between SEP and onset of psychiatric work disability, but the evidence has been limited to psychiatric disability in general7 8 or a single diagnostic group such as neurotic disorders6 7 or alcohol-use disorders.9 Evidence remains sparse on the socio-economic differences in return to work following an episode of psychiatric disability. One study reported substantial socio-economic disparities in the risk of permanent work disability following an initial episode of psychiatric sick leave,10 but information on specific diagnoses was not included. Another study did not find any significant association between managerial versus non-managerial status and return to work after a short-term disability episode owing to depression,11 and still another study suggested a slower return to work among employees with higher education.12 To our knowledge, the association of SEP with recurrent psychiatric work disability episode has not been reported previously.
We therefore sought to examine the socio-economic differences in the onset of psychiatric work disability, in recovery from each episode (ie, time until return to work), as well as risk of recurrence of work disability.
Study design and participants
This study was part of the Finnish Public Sector Study13 covering employees of 10 towns and 21 public hospitals in Finland. The study was approved by the Ethics Committee of the Finnish Institute of Occupational Health.
In Finland, all residents are covered by the National Health Insurance (NHI) scheme. The NHI comprises an earned income insurance component and a medical care insurance component. The former covers the sickness, parenthood and rehabilitation allowances as well as occupational health services, and the latter comprises such benefits as the reimbursements for treatment and examination charges, prescription-drug costs, doctors' fees and costs of rehabilitation services. The NHI is funded by employers, employees and government contributions towards medical care insurance. After a year of sickness absence, disability compensation is provided by insurance companies, where the funding is based on an obligatory pension insurance premium paid by the employer and the employee. Sickness absence and work disability benefits are earnings-based with a minimum compensation for those who have no earnings.
In addition, equal access to primary healthcare is provided for all citizens. Occupational health service is part of primary healthcare including, for example, medical treatment by an occupational physician and rehabilitation referral, if needed. In cases of long-term psychiatric work disability (more than 3 months), a certificate from a psychiatrist is usually required to receive compensation.
The eligible population comprised 151 347 employees who were employed for at least 6 months at any point between 1997 and 2005. Their employment and records were linked to national health registers through unique personal identification codes which are assigned to all citizens in Finland. Linkage to registers was 100% complete, and there was no sample attrition during the follow-up. We excluded those 9282 employees who were on long-term sick leave or disability pension at any time during the 6 months prior to the beginning of the follow-up. We further excluded 148 employees with missing data on SEP, resulting in a final analytical sample of 141 917 participants (94% of the original sample). Information on return to work during follow-up was obtained from the register kept by the Finnish Centre for Pensions. The date of exit from the labour market was coded as the date when disability compensation began to be paid out, while the date when disability compensation ended was taken as the date when a person returned to work life. Consecutive compensation periods were combined.
Work disability outcomes
We defined the onset of work disability as the first long-term sickness absence (≥90 days) or granted disability pension (irrespective of type) during the follow-up. National register data kept by the Social Insurance Institution and the Finnish Centre for Pensions were used to quantify the onset of work disability. Both registers require physician diagnoses of disability and include the beginning and end dates of all work-related disability (including long-term sickness absence, and temporary or permanent full-time or part-time work disability), all lasting at least 90 days. Diagnosis-specific sickness absence data were available from 1997. Primary diagnosis was available for both sickness absences and disability pensions, and they were coded according to the International Classification of Diseases, 10th Revision.14 We grouped the diagnoses in Mental and behavioural disorders (Chapter F) as follows: depressive disorders, mania and bipolar affective disorder, anxiety disorders (phobias, panic disorder, obsessive–compulsive disorder and generalised anxiety disorder (GAD)), reaction to severe stress and adjustment disorders, disorders of adult personality and behaviour, schizophrenia, schizotypal and delusional disorders, and mental and behavioural disorders owing to psychoactive substance use.
Socio-economic position (SEP)
SEP was derived from the employers' registers at the beginning of the follow-up and categorised according to the occupational-title classification of Statistics Finland15 as follows: upper-grade non-manual workers (eg, physicians, teachers, professionals), lower-grade non-manual workers (eg, technicians, registered nurses, kindergarten teachers) and manual workers (eg, cleaners, maintenance workers, kitchen workers).
Age, sex, geographic area (Southern, Central, Northern Finland) and employer (municipality, hospital) were obtained from the employers' records. The presence of any chronic physical disease during a 3-year period preceding the baseline (no, yes) was identified from the national health records. Data on prevalent hypertension, cardiac failure, ischaemic heart disease, diabetes, asthma or other chronic obstructive lung disease, and rheumatoid arthritis were obtained from the Drug Reimbursement Register kept by the Social Insurance Institution of Finland. Information about malignant tumours diagnosed during the preceding 5 years was obtained from the Finnish Cancer Register covering all diagnosed cancer cases in Finland. Previous history of psychiatric treatment was assessed through linkage to individual records in several health registers (reimbursed psychopharmacological treatment coded according to the Anatomical Therapeutic Chemical (ATC) classification,16 psychiatric hospitalisation or state-subsidised psychotherapy) prior to baseline. Data were obtained from the Drug Prescription Register and Rehabilitation registers of the Social Insurance Institution of Finland and hospital discharge register of National Institute for Health and Welfare. Similarly, psychiatric treatment received during episodes of work disability was determined when individuals were recorded as receiving psychotropic drug reimbursements, state-subsidised psychotherapy or hospitalisation within 90 days of the beginning of the first disability episode.
The earliest start date for follow-up was 1 January 1997 and for those who entered the cohort later, the latest beginning date was 1 January 2005. The follow-up ended upon reaching any of the following events: a long-term (≥90 days) sickness absence, disability pension, old-age pension, death or end of the study (31 December 2005), whichever came first. When a period of work disability was noted, the length of the work-disability period was determined from the beginning date of the disability to the end of such period owing to a return to work, old-age pension, death or end of the study (31 December 2005), whichever came first. Among those who returned to work, follow-up for the recurrent episode began from the day the employee returned to work. Cox proportional hazard models were used to study the associations between SEP and outcomes. For both risk of work disability and return to work, analyses were conducted separately for all-cause and cause-specific psychiatric work disability. For recurrent episodes, the small number of cases allowed us only to run an all-cause analysis.
Three main-effects models were performed by entering the covariates into the models in three steps: first, age, sex, geographic area and employer; then, the foregoing variables+psychiatric treatment history; and finally, the foregoing variables+chronic physical disease. The results were presented as HR and their 95% CIs. In the analysis of return to work, we adjusted for calendar year of the disability period and treatment received during the disability spell. In the analysis of recurrent work disability, additional adjustment was made for the length of the previous disability period. Finally, to evaluate whether treatment during the disability period varied between socio-economic groups, we used a χ2 test to compare different forms of treatment (psychotropic drug use, psychotherapy, hospitalisation) between different SEP groups. We used the SAS statistical software, version 9.2.
Table 1 shows that manual workers were slightly older, were more often male and had more chronic physical diseases than non-manual employees. Lower-grade non-manual workers were more often employed by hospitals, while higher-grade non-manual employees worked more often in Southern Finland. No difference was found in the history of psychiatric treatment between the socio-economic groups.
During a mean of 6.3 years of follow-up, the cumulative incidence of any psychiatric work disability episode was 2.8% (table 2). The highest incidence was found for depressive disorders (1.9%) followed by anxiety disorders (0.2%), reaction to severe stress and adjustment disorders (0.2%), schizophrenia and related disorders (0.2%), and bipolar disorders (0.1%). Work disability owing to personality or substance-use disorders was extremely rare, 0.04% of the population for both diagnoses. This may partly be due to those disorders being usually the secondary rather than primary cause of work disability.
We found a socio-economic gradient in several diagnostic categories of psychiatric work disability (table 2). Compared with upper-grade non-manual workers, manual workers had a 1.38-fold (95% CI 1.26 to 1.50) risk, while lower non-manual workers had a 1.27-fold (95% CI 1.17 to 1.37) risk of all-cause psychiatric work disability. Manual workers also had a 1.23-fold to 3.82-fold excess risk of disability stemming from depressive disorders, anxiety disorders, personality disorders, schizophrenia and related disorders, and disorders that were caused by psychoactive substance use. Lower non-manual employees had an excess risk of depressive disorders and schizophrenia when compared with higher non-manual employees. The associations changed little after sequential adjustment for covariates.
We examined the association between SEP and recurrent depressive episode (F33) and found that manual workers had an HR of 1.60 (95% CI 1.25 to 2.06), and lower non-manual workers had a HR of 1.36 (95% CI 1.08 to 1.71) of work disability owing to a recurrent depressive episode when compared with higher-grade non-manual employees (not shown in the table).
Of the participants with a disability episode, 61% returned to work, with an average follow-up time of 1.5 years (table 3). In order to examine the percentage of participants who returned to work within a year,17 we restricted the sample to those with at least a 1-year follow-up. Sixty-two per cent of this subpopulation returned within a year. Among participants with different psychiatric diagnoses, the most likely to return to work were those with a diagnosis of ‘reaction to severe stress and adjustment disorders’ (93%) followed by participants with anxiety disorders (76%), depressive disorders (63%) and bipolar disorders (54%). Participants with substance-use disorders (36%), schizophrenia and related disorders (41%), and personality disorders (45%) were the least likely to return to work. Higher non-manual occupational grade predicted return to work in depression, personality disorders, schizophrenia and related disorders, and substance-use disorders, though not in the case of bipolar disorder, anxiety disorders, or reaction to severe stress and adjustment disorders. Covariate adjustment did not materially affect these conclusions, with the exception of substance-use disorders, where the adjustment for treatment received during the disability period attenuated the association with SEP to statistical non-significance.
Recurrent episodes of long-term psychiatric work disability occurred for 31% of the subjects. The follow-up time for incident recurrent episode lasted 2.2 years on average (table 4). In the fully adjusted model, those with a manual occupational grade experienced a 1.25-fold (95% CI 1.02 to 1.53) higher risk of recurrent disability compared with those in higher non-manual occupations.
As shown in table 5, we found that during the first 90 days of the first disability period, 28% of the participants had no indication of recorded treatment, 70% had pharmaceutical treatment only, and a minority (2%) had state-subsidised psychotherapy or a combination of pharmaceutical treatment and therapy. Low SEP was associated with not receiving treatment, although the differences were small (28% in the highest SEP group vs 26% and 31% in the lower SEP groups).
This study of 141 917 public sector employees showed that during a mean of 6.3 years of follow-up, there were socio-economic differentials in onset and prognosis across the majority of diagnostic groups, with a couple of notable exceptions.
Strengths and limitations
The strengths of our study include its large sample size, prospective study design and reliance on register-based data which covers all employees. The national registers of work disability indicators reliably cover all disability periods of 90 days or more and include physician-determined International Classification of Diseases, 10th Revision diagnoses, as well as information on labour market participation. However, the study population comprised exclusively Finnish public sector employees, who were predominantly female (75%).
In this study, it was not possible to examine the association between SEP and work disability of less than 90 days' duration. Our measure of recurrent episode included only those with long-term disability episodes available in national registers, which may have resulted in misclassification of cases with multiple shorter-term episodes. If socially patterned, such misclassifications would introduce bias. Owing to the low number of events in some groups, it is also possible that our analysis did not have sufficient power to detect an effect in these cases. It is therefore important to replicate the study with outcomes of work disability including also shorter periods. We defined work disability as long-term sickness absence or the receipt of disability pension (fixed-term or permanent). Socio-economic disparities may vary depending on the severity of disability outcome. Therefore, in the future, it would be useful to examine disability outcomes (both short- and long-term absence, as well as fixed-term and permanent disability pension) separately.
Furthermore, although these data were prospective, we were not able to control for episodes of psychiatric disability during the life course or early selection to low-grade jobs owing to mental disorders. In addition, we had no measurement of actual incidence of mental disorders but only long-term disability owing to those disorders, and thus, no comparison could be made to evaluate the possible discrepancy between actual incidence and disability. Furthermore, although many covariates were adjusted in the models, these data did not include all potential confounding factors, such as severity of the psychiatric condition, education level, marital status, job tenure and psychosocial work characteristics. Our estimate of treatment was based on treatment in the public sector; thus, the true socio-economic differences could be even larger if we had data on private, self-covered treatment.
Finally, further research on population-based samples is needed to assess the generalisability of our findings because our sample comprised employed people in the public sector; thus, selection on the basis of being fit enough to work is likely to mean that our cohort was healthier at baseline than the general population. However, as the compensation system for long-term work disability in Finland is virtually universal, our findings are not likely to be biased, owing to differences in the national system. Furthermore, Finland is considered as a country with a relatively generous and accessible benefit system.18 Thus, our findings may not be generalisable to countries with a different disability policy orientation. Therefore, a comparative study using data from other countries would be informative.
Evidence of socio-economic gradient in diagnoses of psychiatric work disability
In line with earlier evidence on the inverse association between SEP and incidence of mental disorders,2 7 our findings suggest an inverse socio-economic gradient in the onset of a new episode of psychiatric work disability across a range of diagnoses including depressive disorders, personality disorders, schizophrenia and related psychotic disorders, psychoactive substance use and anxiety disorders.
The mechanisms that could explain the observed socio-economic gradient relate to differential access on the resources—such as income, prestige and job control—that are believed to be protective against the onset of psychiatric illness.2 Adverse life events, financial hardship and psychosocial work characteristics, such as job control and rewards at work, have also been demonstrated to have an important role in explaining the socio-economic gradient in mental health.19 Another explanation links personal resources, for example, coping style and self-esteem, as buffers against the impact of external stressors on mental disorders, and suggests that high-SEP individuals are better equipped with such resources.2
However, a part of the socio-economic differences in psychiatric work disability may be also due to the social consequences of early-onset mental disorders.20 Mental disorders in adolescence have been shown to be associated with a drop-out of high school and even higher education.20 Median age of onset can be as early as 11 years for specific anxiety disorders such as phobias and impulse-control disorders (corresponding to adult personality disorders). The median age at onset is also quite early for bipolar disorders (18 years), substance-use disorders (20 years) and schizophrenia (22–23 years), while for depression it tends to be later, 32 years.21–23 In depression, downward drift into low SEP may therefore be less likely than in other disorders, which may result in a weaker association between SEP and work disability in depression.
High SEP participants were 1.5–4.5 times more likely to return to work following a disability period compared with low SEP individuals. The greatest differences were found for personality disorders, followed by substance use, schizophrenia and related disorders, and depressive disorders. Our findings are in agreement with an earlier study reporting an inverse socio-economic gradient in the risk of permanent exit from labour market among employees on psychiatric sick leave.10 Our study extends existing evidence, since we were able to differentiate between specific psychiatric diagnoses.
To explain why high SEP is a protective factor for mental health, first, comorbid physical and mental disorders (which may complicate the process of return to work) are likely to be less common in high-SEP individuals.24 Second, high-SEP individuals are more likely to have access to material resources that they can deploy towards receiving various forms of treatment, including private sector treatment.25 Third, the availability of different forms of social support—an important factor in the recovery from depression—might be poorer for low-SEP individuals.19 Fourth, low-SEP individuals may have less discretion in adjusting their work schedules and arrangements to accommodate their condition.26 Among those who do not have a permanent job, re-employment prospects may be less favourable for low-SEP people.27 Fifth, success in treatment may be lower among low-SEP individuals owing to medical bias28 as well as poorer treatment compliance and greater treatment resistance among low-SEP patients.29
Psychiatric work disability diagnoses with no evidence of socio-economic differences
For two diagnostic groups—viz bipolar disorders and reaction to severe stress and adjustment disorders—we found no evidence of a socio-economic gradient in the onset of long-term psychiatric work disability. Additionally, even though we did find an association between low SEP and the onset of a work-disability episode owing to anxiety disorders, no association was observed in return to work following such an episode.
For these specific diagnostic categories, our findings are not entirely consistent with reports in previous studies. Regarding bipolar disorders, earlier population-based studies have shown higher prevalence rates among low SEP individuals.30 31 However, studies carried out among clinical samples32–34 suggest higher SEP and educational attainment in bipolar patients when compared with the general population. As our dataset comprised individuals who had contact with healthcare and had received a medical certification of work disability, the lower likelihood of low SEP patients to seek help may have diluted the socio-economic gradient in the onset of work disability. Furthermore, in severe mental disorders with an early onset, the socio-economic gradient may partly reflect the social consequences of the disorder (ie, downward social mobility owing to mental illness at a young age).20 The age of onset in bipolar disorder in general is later in women than in men, and later in type II disorder than in type I disorder.35 Bipolar II disorder is also more common in women than in men.36 Thus, a part of our subjects may have attained a higher SEP prior to the onset of disorder. A later age at onset is also associated with a less severe clinical prognosis,35 and in addition, hypomania can even enhance cognitive performance and thereby success in worklife.37 However, as these data did not allow us to differentiate between bipolar I and II cases, more research needs to be done in samples with more detailed information.
A recent follow-up study suggested a mean duration of bipolar I mood episode to be 13 weeks, and 75% of patients recovered from their present bipolar mood episodes within 1 year of onset.38 Of the bipolar disorder participants in our study, 54% returned to work within a year. It seems that recovery from symptoms does not necessarily lead to employment. A novel finding in our study is that returning to work from a bipolar-disorder disability period seems to be unrelated to the socio-economic position. However, as our sample of participants with that diagnosis was quite small (n=150), further studies are warranted.
In the group with anxiety disorders, including phobias, panic disorder, obsessive–compulsive disorder and GAD, we also found no socio-economic gradient in return to work. The distribution of specific diagnosis in this group was: 90% either panic disorder or GAD, 5% phobic anxiety disorders and another 5% obsessive–compulsive disorder. The literature on the socio-economic gradient in anxiety disorders and reaction to severe stress and adjustment disorders is scarce. The latter diagnostic group includes acute stress reaction, post-traumatic stress disorder (PTSD) and adjustment disorders. Low education39 and low maternal education40 have been related to a higher prevalence of PTSD among adults and adolescents. Research on the prevalence of work disability owing to anxiety disorders and adjustment disorders is also scarce. Because the duration of acute stress reaction is less than a month, our data did not include cases of acute stress reaction. With a three-digit specification of diagnoses, we were not able to distinguish between PTSD and adjustment disorder. An earlier report found a return-to-work rate of 84% after a 1-year follow-up, not including measures of SEP.17 Our study with the high return-to-work rate (93%) within a year is in accordance with that study and provides new evidence in terms of the lacking socio-economic differences both for the onset of work disability and for the return to work.
The lack of a socio-economic gradient in return to work in bipolar, anxiety and adjustment disorders may relate to several factors, such as clinical characteristics of the disorder, (eg, severity of symptoms) and work-related factors. However, there is at least one earlier study in which higher education predicted a longer duration of sickness absence for individuals with depression, anxiety or adjustment disorder.12 The disorder itself may be equally severe in each socio-economic group, making return to work equally difficult. For example, psychotic mania or severe panic disorder may make management of work tasks very complicated in a high-SEP job. However, for adjustment disorder, the picture may be different. Usually, the more severe the disorder, the greater the improvement of symptoms required for the resolution of work disability. Adjustment disorder is defined more in terms of social dysfunction than in terms of symptoms and does not necessarily reach the severity seen, for example, in depression.41 The less severe nature of adjustment disorder may explain the high return-to-work rates observed and the lack of socio-economic gradient in this study. These hypotheses, however, need further exploration with more detailed datasets.
We found that manual and lower non-manual workers had an increased risk of recurrent episode of psychiatric work disability compared with those with a high SEP. In general, little is known about SEP and recurrent mental disorders, and as far as we are aware, this is the first study to show the association with psychiatric work disability as an outcome. A recent systematic review on the predictors of recurrence of major depressive disorder did not suggest an association with SEP.42 However, other studies have shown that low SEP among depressed patients is associated with poorer treatment compliance25 and greater treatment resistance,29 which are both associated with recurrence.
Our results suggest that SEP is associated with the entire spectrum of disability associated with psychiatric diagnoses, that is, lower SEP workers are at a higher risk of onset, longer periods before return to work and a higher risk of recurrence of disability owing to mental illnesses. This study indicates that socio-economic position should be taken into account in the attempts to reduce long-term work disability owing to mental disorders. The exceptions to this patterns were bipolar and adjustment disorders where we did not find a socio-economic gradient. To explain the socio-economic differences in psychiatric work disability, future research should focus on the contribution of severity of the disease, help-seeking behaviour, employee motivation and compliance, coping resources and cognitive capacity, effectiveness of treatment and work characteristics.
Funding The study was supported by Academy of Finland (grant nos 124322, 124271, 123621, 133535 and 129262) and the BUPA Foundation, UK.
Competing interests None.
Ethics approval Ethics approval was provided by the Ethics Committee of the Finnish Institute of occupational health, Helsinki, Finland.
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.