Work ability and psychological distress in a working population: results from the Stockholm Public Health Cohort

Aims: Psychological distress is a global public health concern with individual and societal implications causing work-related disability and loss of productivity. It is less known how much work ability contributes to the development of psychological distress. This study aimed to assess the association between self-perceived physical and mental work ability in relation to job demands, and the incidence of psychological distress in a Swedish working population. Methods: Data were obtained from three subsamples of the Stockholm Public Health Cohort with baseline in 2010 and follow-up in 2014, based on a working population in Stockholm County aged 18–60 years, with no or mild psychological distress at baseline (n=29,882). Self-perceived physical and mental work ability in relation to job demands were assessed at baseline with a subscale from the Work Ability Index. Study participants scoring 4 or more on the General Health Questionnaire 12 at follow-up were classified as having developed psychological distress during the study period. Poisson log linear regression was used to calculate crude and adjusted rate ratios with 95% confidence intervals. Results: At follow-up, 2543 participants (12%) had developed psychological distress. Reporting poor physical and/or poor mental work ability in relation to job demands at baseline was associated with an almost doubled rate ratio of psychological distress at follow-up, compared to reporting good work ability (rate ratio 1.8; 95% confidence interval 1.6–2.0). Conclusions: Poor work ability is associated with a higher incidence of future psychological distress compared to good work ability.


Background
Mental health problems are leading causes of years lived with disability in high and middle-income countries. Depression and anxiety are most common within this wide scope of diagnoses, with an estimated total global prevalence of 7%, according to the Global Burden of Disease Study [1]. In Sweden, mental health problems are the primary reason for long-term sick leave, which, apart from personal suffering, is associated with a great societal burden and loss of productivity [2]. Psychological distress (PD) is characterised by a continuum of depressive and anxiety symptoms, which may develop into mental illness with greater severity if unrecognised. Notably, a systematic review of prospective studies and a longitudinal study report that subthreshold depression is associated with future major depressive disorder as well as other psychiatric conditions [3,4].
Working conditions promoting physical and mental health are of importance for a sustainable and long-lasting working career, including prioritising primary prevention of occupational health hazards such as PD. Previous studies have examined the role of psychosocial factors in work environments for Work ability and psychological distress in a working population: results from the Stockholm Public Health Cohort future mental health. In a meta-review by Harvey et al. (2017), low job control, high job demands, low relational justice, high effort-reward imbalance, role stress, bullying, low procedural justice and low social support were identified as risk factors for depression and anxiety symptomatology [5]. Similar findings were presented in a systematic review by Theorell et al. (2015) presenting low decision latitude, job strain and bullying as risk factors for depressive symptoms [6]. Furthermore, Madsen et al. (2017) concluded in a systematic review that job strain was associated with not only depressive symptoms but also an increased risk of clinically diagnosed depression [7]. evidently, work-related factors are important determinants for mental health.
Work ability is defined as having occupational competence and health to manage reasonable tasks in an acceptable working environment [8]. Assessing work ability in occupational settings is valuable to recognise ill health and prevent workers from discontinuing due to work-related disability [9]. Although measures of work ability ideally include an assessment of both objective and subjective estimation of workers' own resources in relation to job demands [10], self-reporting instruments such as the Work Ability Index (WAI) are measures with high validity often used in occupational research. The WAI consists of seven subscales measuring current work ability compared to the lifetime best, work ability in relation to job demands, current diagnosed diseases, work impairment due to disease, sick leave during the past year, own prognosis of work ability and mental resources [11]. The WAI is widely used in occupational health research with results that have been proved to be a good predictor of frequent sickness absence [12] as well as future work-related disability and sick leave [13][14][15]. Moreover, ebener and Hasselhorn (2019) concluded that measuring work ability with only the subscale about work ability in relation to job demands of the WAI is feasible in occupational health research [16].
In a systematic review by van den Berg et al. (2009), high physical workload, poor physical work environment, high mental work demands, lack of autonomy, obesity, older age, lack of leisure-time vigorous physical activity and poor musculoskeletal capacity were associated with poor work ability [17]. leijon et al. (2017) found that PD was associated with decreased work ability in a longitudinal study of Swedish workers [18]. To date, little is known about whether there is also a reversed association; that is, if poor work ability is of importance for the incidence of PD as other psychosocial work-related factors have shown relevance for mental health problems [5][6][7].

Aims
The aim of this study was to assess the role of selfperceived physical and mental work ability in relation to job demands for the incidence of PD in a Swedish working population.

Design and study sample
This study is based on data obtained from the Stockholm Public Health Cohort (SPHC), a prospective study set within the framework of Stockholm County Council's public health surveys aiming to assess health, lifestyle and socioeconomic factors. Further description and details of the SPHC is found in the cohort profile article [19]. Men and women between 18 and 60 years of age who participated in any of the three subsamples of the SPHC were included in the study if they reported no or mild PD at baseline; that is, scoring 2 or less on the General Health Questionnaire 12 (GHQ-12, described below) [20]. Another criterion for inclusion was responding to any of the two items on the subscale about physical and mental work ability in relation to job demands from the WAI (described below); that is, indicating that the participants had an active working life. Those with sickness absence for more than 90 days during the past 12 months were excluded from participation.

Exposure
At baseline in 2010, physical and mental work ability was measured with the subscale about work ability in relation to job demands included in the WAI. Participants were asked to answer the questions: (a) How do you rate your current work ability with respect to the physical demands of your work? and (b) How do you rate your current work ability with respect to the mental or psychological demands of your work? Physical and mental work ability was merged in which 'poor work ability' was defined as answering 'moderate', 'rather poor' or 'poor' in one or both of the items, and 'good work ability' was defined as answering 'rather good' or 'very good' on both items. Previous studies show that the psychometric properties of this subscale from the WAI are considered stable, predictive and internally coherent on group level [21,22].

Outcome
The incidence of PD at follow-up in 2014 was assessed with the GHQ-12. Participants were asked to respond to six positive and six negative statements in relation to their psychological health during the past weeks. The positive statements (i.e. ability to concentrate, playing a useful part, capability of making decisions, ability to enjoy day-today activities, ability to face problems and feeling reasonably happy) had multiple answer alternatives ranging from 'more/better than usual' to 'much worse/less than usual'. For the negative statements (i.e. loss of sleep over worrying, feeling constantly under strain, incapability to overcome difficulties, feeling unhappy and depressed, losing confidence and thinking of self as worthless), answer alternatives ranged from 'not at all' to 'much more than usual'. A score of 4 or more on the GHQ-12 was defined as having PD at follow-up, in which the chosen cut-off corresponding to moderate and severe PD was based on a previous validation study and is the recommended standard scoring method for PD assessed with this instrument [23]. Several studies have concluded that the GHQ-12 is a valid and reliable instrument in general population samples [20,[23][24][25].

Confounding
Potential confounders were considered in accordance with previous research on risk factors for mild mental illness, clinical considerations and availability, after careful discussion about whether they instead could be intermediators or colliders. Potential confounders are summarised in Table I.

Statistical analysis
Generalised linear models with Poisson log linear regression were used to estimate the associations between work ability and PD. results are presented as rate ratios (rrs) with 95% confidence intervals (CIs). Factors considered as confounders in the association between the exposure work ability and the outcome PD were age, sleep disturbance and neck/back pain in the past 6 months, which were added to the final model. Also, a variable including the origin of the three subsamples (i.e. years 2002, 2006 or 2010) were added to the model because the baseline in this study (2010) was not the baseline for two of the subsamples. To assess potential selection bias, an attrition analysis was conducted by comparing the prevalence of poor work ability among those lost to follow-up and those with missing data on any of the outcome variables, with the prevalence of this exposure among those successfully followed. Statistical analyses were performed using IBM SPSS Statistics version 25.

Ethical considerations
Participants gave their written informed consent prior to data collection. The study was approved by the ethical review board of Stockholm (Dnr 2010/1185-31/1 and Dnr 2016/987-32).

results
The total study population at baseline was 29,882, as presented in Figure 1. The follow-up rate was 69%, hence 20,539 participants were successfully followed    Table II displays a detailed description of baseline characteristics stratified by good and poor work ability. The mean age was 45.6 years and 55% were women. Most participants were intermediate non-manual workers and employed/self-employed professionals and 22% had previous long-term illness. Out of the included participants, 92% reported good work ability at baseline.
Crude and adjusted associations between work ability and PD are presented in Table III. At followup, 2543 participants (12%) had developed PD according to the GHQ-12. Of those with poor work ability, 14% (n=360) developed PD during the study period. The adjusted association between poor work ability and PD yielded a rr of 1.8 (95% CI 1.6-2.0). Thus, reporting poor work ability at baseline was associated with an almost doubled rr of PD at follow-up after 4 years, compared to reporting good work ability at baseline. The attrition analysis showed that of those participants lost to follow-up, 10% (n=934) reported poor work ability according to the WAI. The corresponding prevalence for those successfully followed was 8% (n=1575).

Discussion
This prospective study assessed the role of self-perceived physical and mental work ability for the incidence of PD in a Swedish working population. The results indicate that self-perceived poor physical and/ or mental work ability in relation to job demands is associated with an almost doubled rr of developing PD over time, compared to a good work ability.
To the best of our knowledge, this is the first study to investigate poor work ability as a potential risk factor for PD. Previous studies in occupational health research have identified other psychosocial workrelated risk factors, including high job demands, low support, bullying [5,6] and job strain [6,7] for depression and anxiety symptomatology [5,6] as well as for clinically diagnosed conditions [7]. leijon et al.
(2017) found PD to be a risk factor for future poor work ability in a longitudinal study of Swedish workers [18]. In this study, however, PD was instead the outcome of interest in relation to work ability and not the exposure.
The results of this study are of societal as well as individual importance. A Swedish report states that sick leave resulting from mental health problems on average is longer compared to sick leave from other diagnoses [2]. Furthermore, although individuals with poor work ability may be able to perform work which is less physically demanding, it is reasonable to assume that poor work ability related to mental demands hampers the ability to perform work which is also less demanding. Also, the possibility to start working after sick leave due to mental health problems is poorer than for other diagnoses, adding to the present study's result and stressing the importance of acknowledging mental health in occupational settings, maybe even to a greater extent than previously. High comorbidity between PD and conditions such as spinal pain [26] and cardiovascular disease, among others, is another incentive for increasing the awareness of mental health, which is also of high priority in the united Nations' Millennium Development Goals striving towards sustainable health for everyone [27]. The results of this study emphasise the importance of future studies assessing the role of work ability among workers in occupational settings for preventing future PD.
This population-based study has a substantial sample size, covering residents in the largest county in Sweden. extensive questionnaires enabled a careful consideration of potential confounders even though there is a risk of unmeasured and residual confounding (e.g. job loss, job dissatisfaction, etc.) that may entail an under-or overestimation of associations. Moreover, selection bias may have an impact on the results if loss to follow-up was associated both with the prevalence of poor work ability (exposure) and the incidence of PD (outcome). In this case, however, the prevalence of the exposure was similar among successfully followed participants and drop-outs (10% and 8%, respectively), indicating that selection bias is not a big threat.
The exposure and outcome were assessed with stable instruments of high validity previously used in occupational research. Work ability was assessed with only one subscale of the WAI, and the exposure may hence be prone to non-differential misclassification. Nonetheless, this subscale from the WAI has been validated both against the full WAI (i.e. all seven subscales) as well as for sickness absence [28]. If misclassification biases the results, this would have led to diluted associations. Furthermore, changing jobs during the study period may also result in changed perceived work ability. Measuring job changes during the study period would possibly result in a more accurate measure of the longitudinal association with PD; however, this was not assessed in this study and cannot be taken into consideration.
To avoid morbidity influencing participants' judgement of work ability for illness not related to PD, those with sickness absence for more than 90 days during the past 12 months were excluded from the cohort. The GHQ-12 is a non-specific screening measure of PD often used in primary care, and is not an instrument used for diagnosing mental disease [29]. In this study, only participants scoring 2 or less on the GHQ-12 at baseline were included to be able to study the risk of PD rather than the prognosis.
The outcome PD was defined as scoring 4 or more on the GHQ-12, which previously has been suggested to be the best cut-off for identifying PD with this instrument [23]. Scoring 4 or more on the GHQ-12 corresponds to moderate/severe PD, which may reflect a condition that is more than subclinical, hence motivating a more thorough mental health examination.
The aim of the study was to investigate the incidence of PD in a healthy cohort. rai et al. (2012) concluded that also milder forms of PD (scoring 1-2 on the GHQ-12), corresponding to 'good mental wellbeing' in Swedish surveys that form the basis of public mental health policies, were associated with receiving a disability pension after 5 years [30]. The findings by rai et al. (2012) may indicate that this study has examined the prognosis rather than the incidence of PD and is a rationale for acknowledging also mild PD in future research in occupational settings [30]. Moreover, it is reasonable to assume that the two different dimensions of work ability measured in this study (i.e. physical and mental, respectively) differ in their contribution to PD, which motivate future research studying these separately rather than collectively. By doing this, it is possible to identify the demands of highest importance for developing PD and implement future preventive strategies in occupational settings accordingly.
In conclusion, this study adds new knowledge to the field of occupational research by indicating that poor work ability is associated with a higher incidence of future PD compared to good work ability.

Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/ or publication of this article.

Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by AFA Insurance (grant no. 170095).