Psychosocial and organisational work factors as predictors of sickness absence among professionally active adults with common mental disorders

The study population was drawn from a subsample of those participating in any of the SWES 1993–2013 [25] and who had been diagnosed with a CMD up to five years before they answered the SWES. Data regarding CMDs were obtained from the National Patient Register [26] and were defined as a diagnosis for depression, anxiety- or stress-related disorders taken from inpatient or outpatient health care in Sweden. According to the International Statistical Classification of Diseases (ICD), they correspond to the codes in version 10 (ICD-10): F32 – F43; and version 9 (ICD 9): 296, 298, 300, 301, 308, 309 or 311 (n = 1237). The largest share of participants was diagnosed with a stress-related diagnosis (ICD-10: F43, ICD-9: 308 and 309, n = 556 (14.6%)) followed by anxiety disorder (ICD-10: F40-F42, ICD-9: 300, n = 370 (9.8%), and depression (ICD-10: F33-F39, ICD9: 296, 298, 301 and 311, n = 230 (6.1%)). Following the example of previous studies, also those who were prescribed antidepressants (ATC-code N06A) up to five years before the interview in the SWES were added [6]. This group made out the largest group of the sample, n = 2,639 (69.5%). The inclusion process was hierarchical where individuals with a record of a diagnosis of CMD were included first, and in a second step, those without a diagnosis but with a record of prescription of antidepressant medication were included. After the exclusion of those who received disability pension before participating in SWES, 3,795 persons remained in the study population.

In Sweden, all individuals from 16 years and onwards, with an income above a certain level, can receive benefits when having sickness absence. The employer is responsible for the payment of benefits during the first 14 days and this period is not covered by registers. Moreover, there is one qualifying day (3 or 30 days depending on chosen fee among self-employed) without benefits. Thus, sickness absence in this study covers sick-leave periods of 15 days or more. Moreover, two days of half-time sickness absence were counted as one net day of sickness absence. Information on sickness absence up to three years after participation in SWES was derived from the “Longitudinal integrated database for health insurance and labour market studies” (LISA) and linked to survey data by the unique national identification number.

Job strain was defined following the work demand-control model [14, 27] with several variables serving as proxies. Four variables were used as indicators of job demands and job control, respectively [28]. To follow the questions stated below, the SWES from 2015 is attached as a supplementary file (SWES 2015). Indicators of high job demands, with cut-offs given in parenthesis after the respective item, were a) “have to miss lunch, work late, or take work home’’ (≥ 1 day of 2), b) ‘‘do not have time to talk or even think about something other than work’’ (≥ ¾ of the time), c) ‘‘the work requires your full attention and concentration’’ (nearly all the time) or d) ‘‘have far too much to do at work’’ (too much to do, response 1). Indicators of low job control were e) “are you able to determine when various working duties are to be carried out’’ (no, not at all), f) ‘‘participate in decisions on the arrangement of your work’’ (≤ mostly not), g) ‘‘have the opportunity to determine your work pace’’ (≤ 1/10 of the time), or h) ‘‘have too little influence at work’’ (≤ too little influence, response 1 and 2). Item a) was answered on a five-point scale reaching from 1 = every day to 5 = not at all/rarely in the last three months. Items b), c), and g) were answered on a six-point scale reaching from 1 = Nearly all of the time to 6 = No, not at all. Item d) was answered on a five-point scale reaching from 1 = far too much to do to 5 = far too little to do. In similarity, item h) was answered from 1 = too little influence to 5 = too much influence. Items e) and f) were answered on a four-point scale reaching from 1 = always to 4 = no, not at all. The upper quartile was defined as having been exposed to high job demands and low job control, respectively. If a person answered positively to at least two of the indicators of work demands or control, this indicated high job demands or low job control, respectively [28].

Social support was indicated by two items “Can you receive support and encouragement from your superiors when your work becomes troublesome?” and “Can you receive support and encouragement from your fellow workers when your work becomes troublesome?”, answered on a 4-point Likert scale from 1 = always to 4 = never. Receiving social support (always or most of the time) from both colleagues and superiors was calculated as receiving social support, while all other options were coded as not receiving social support.

Further, one item measured the atmosphere at work (open or closed) “Are you reluctant to express critical views in the workplace regarding your working conditions?” answered on the same scale as described for items e and f (see above). Having this possibility for always or most of the time indicated an open atmosphere. Two other indicators of psychosocial work were: “Do you feel ill at ease and downhearted as a result of the difficulties you face at work?” and “At the end of your workday, do you feel that your work input is inadequate?” answered on the same scale as item a (see above). Feeling badly or experiencing despair at least one day of the week (1 day of 5) indicated a poor work environment.

The indicators of the organisational work environment factors were work flexibility, usually working at home, inconvenient working hours, and full-time or part-time work. Flexibility at work was indicated by two items. “In general, are you able to decide your working hours, within certain limits?” with response options “Yes, I have flextime (begin and end within certain fixed times, but not at an exact time point)” and “Yes, I have relatively free working hours in another way” indicating flexibility, while “No, in general, I cannot change my working hours” was coded as not having flexibility.

Another organisational workplace factor was “How much of your normal working time do you usually work at home”. Having this possibility at least some hours a week was indicating the possibility to work from home.

Inconvenient working hours were indicated by the need to work evenings, nights, or weekends (Saturday or Sunday). If any of this was the case, it was coded as inconvenient working hours.

Age, sex, educational level, and year of participation in SWES were considered potential confounding variables. Age, sex, and educational level were obtained from the LISA database [29]. Age was coded into five age groups (16 to 25 years, 26 to 35 years, 36 to 45 years, 46 to 55 years, and 56–64 years). Sex is binary, men and women. Educational level was based on the Swedish classification system of education (SUN2000) and coded into three groups (low (elementary school), medium (upper secondary school), and high (university)).

Relative risk (RR) regression models with a 95% confidence interval (CI) were calculated to analyse associations between psychosocial and organisational workplace factors and subsequent sickness absence three years after the interview in SWES using proc genmod in SAS (SAS Institute Inc. 2013. SAS® 9.4.) [30]. In the first step, crude models were calculated. In the next step, we controlled for the year of interview. In a third model, we additionally controlled for age, sex, and educational level. Further, relative risk regression models stratified by the three diagnoses (stress, anxiety, and depression) were used.