Article Text

Download PDFPDF

Does dissatisfaction with psychosocial work climate predict depressive, anxiety and substance abuse disorders? A prospective study of Danish public service employees
  1. H K Jensen1,
  2. J Wieclaw1,
  3. T Munch-Hansen1,
  4. A M Thulstrup1,
  5. J P Bonde1,2
  1. 1Department of Occupational Medicine and Danish Ramazzini Center, Aarhus University Hospital, Aarhus, Denmark
  2. 2Department of Occupational and Environmental Medicine, Bispebjerg University Hospital, Copenhagen, Denmark
  1. Correspondence to Joanna Wieclaw, Department of Occupational Medicine, Aarhus University Hospital, Noerrebrogade 44, Bygn. 2C, DK–8000 Aarhus C, Denmark; joanna.wieclaw{at}stab.rm.dk

Abstract

Background The aim of this study was to examine if dissatisfaction with psychosocial work climate predicts psychiatrically diagnosed depressive, anxiety and substance abuse disorders.

Methods In Aarhus County, Denmark, 13 423 public service employees at 683 workplace units answered a questionnaire survey assessing psychosocial work environment. An average workplace unit score of overall satisfaction with psychosocial working conditions, rated on a scale from 0–10 with 10 being most satisfied, was computed and assigned to the individual employees at each work unit. Aggregated satisfaction scores were divided into three levels, according to the 25–75 percentiles. Data on hospitalisations and outpatient treatments for depressive, anxiety and substance abuse disorders was obtained from the Danish Central Psychiatric Research Register. HRs and 95% CIs were computed for first onset of studied disorders, starting from the baseline survey at 1 January 2002 through to 30 April 2008. Risk estimates were adjusted for sociodemographic variables.

Results A low satisfaction with psychosocial working conditions was associated with an increased risk of any mental health disorder, HRadj 1.71, 95% CI 1.04 to 2.82. The lower the satisfaction level, the higher was the risk of mental health disorders. Moreover, substance abuse disorders were more frequent among men dissatisfied with work climate, HRadj 3.53, 95% CI 1.55 to 8.03.

Conclusion Working in a dissatisfying psychosocial environment increases the risk of subsequent mental health disorders. Randomised, controlled intervention trials may help in resolving whether this association is causal.

  • Psychosocial
  • workplace
  • longitudinal studies
  • mental health disorders
  • depression
  • mental health DI

Statistics from Altmetric.com

Request Permissions

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.

Introduction

Depressive and anxiety disorders inflict substantial burdens on the Western world due to loss of work days, loss of quality of life and less effective work performance.1 Depressive and anxiety disorders are often comorbid2 and share a number of risk factors, such as middle-age,3 female gender4 5 and low socioeconomic position.6 Also, traumatic life events such as divorce, bereavement, unemployment,4 5 7 poor family environment,8 genetics and personality traits such as neuroticism and low self-esteem9 10 have been shown to be associated with these disorders. Prevalence of depressive and anxiety disorders varies across professions,11 12 and a number of studies have reported increased anxiety and depression especially in human service professions.12–14

Several studies point to psychosocial working conditions as an important risk factor,8 15–26 and reviews found evidence that psychosocial work environment may predict mental health symptoms and disorders.27–30 However, there is inconsistency in identifying risk factors for mental health disorders, due to heterogeneous measures of exposure and outcome and diverse study populations. Furthermore, most studies rely on self-reports for both job stressors and psychological distress symptoms.15–17 20–23 Generally, self-report studies show associations between psychosocial work environment and depressive and anxiety disorders more often than studies using objective outcome and/or exposure measures.25 31

Several attempts are made to use objectively obtained exposure and outcome measures in assessing risks of mental disorders. Stansfeld et al17 found no association between mental health disorder and psychosocial work environment when using work characteristics assessed by external observers compared to individual perceptions. Also, a recent study by Kouvonen et al19 showed that individual assessments of workplace social capital predicted depressive disorders, but no association was found for the aggregate workplace measure. In contrast, positive associations were found in the study by Waldenström et al32 that used assessment of working conditions by external observers and in the study by Wieclaw et al33 that used job exposure matrix and register data on depressive disorders.

With this background, a longitudinal study was conducted aiming to examine the relationship between the overall satisfaction with psychosocial work climate and subsequent risk of psychiatrically diagnosed mental health disorders. The study addresses some limitations of exposure and outcome ascertainment inherent in earlier studies, by using aggregated measures of perceived work environment in work units and independent psychiatric diagnoses retrieved from medical registers. Both hospitalisation and outpatient hospital-based treatments are considered. This restricts the outcome to individuals suffering from severe mental health disorders, but at the same time reduces bias inherent in self-reports. Only a few studies have looked at a number of mental disorders simultaneously. As there is a degree of comorbidity, depressive, anxiety and substance abuse disorders are all addressed in the present study.

Methods

Population and data collection

A cohort of 13 423 public service employees from 683 workplaces in Aarhus County participated in a workplace questionnaire survey that was carried out as part of human resource management activities from 1 January 2002 through to 31 December 2005. The County's main employment areas are healthcare and special care institutions. A more detailed description of the cohort and the survey has been provided elsewhere.34 The questionnaires were handed out and collected at workplaces. The employees not present at the time of the survey were contacted by mail. Participation was voluntary and anonymous, but through the county's salaries administrative system all employees were identified at the work unit level by means of the personal identification number (the CPR number), which consists of ten digits designating birth date and gender.34 35 Thus, the identity of the persons that filled in a particular questionnaire was unknown but the identity of all employees in the managerial work units was known. The questionnaire response rate was 78.8%. Data on occupation, departments and date of commencement and termination of employment were provided by administrative services in the county. Finally, data were obtained on hospital and outpatient treatments for depressive, anxiety and substance abuse disorders in Aarhus County from the Danish Central Psychiatric Research Register (DCPRR) during 1 January 1969 through to 30 April 2008, registered according to the WHO ICD classification: ICD8 for the period 1969–1994 and ICD10 for the period 1995–2008. The DCPRR36 contains information dating from the nineteenth century with systematic data collection since 1938. It was computerised in 1969 and includes data on all admissions to psychiatric hospitals and psychiatric wards in general hospitals in Denmark. From 1995 it also contains information about all psychiatric outpatient contacts. As there are no private psychiatric hospitals in Denmark, the DCPRR has complete records of hospital treatments whereas primary healthcare contacts are not included. Data linkage was achieved by using the Danish Civil Registration System.35 The CPR number is used in all national registers, and therefore it is possible to achieve a highly accurate linkage across registers. The register also contains data on the patients' sociodemographic background such as area of residence, marital status and number of children. Diagnoses in the clinical register have been validated, showing high concordance between clinical ICD-8 diagnoses and ICD-10 diagnoses based on research diagnostic criteria.37

Exposure measure

The surveys were based on a 40-item questionnaire designed to assess psychosocial work environment in the county's workplaces.34 In the present study only data regarding the overall satisfaction with work climate were analysed. The question regarding overall satisfaction was phrased: ‘How satisfied are you, all in all, with the psychosocial work conditions at your workplace?’ rated on a scale from 0 (unacceptable) to 10 (exceptional). Factor analyses of the responses to the questionnaire indicated the existence of an essential one-dimensional construct and the results of an earlier study showed that the overall satisfaction question had the strongest explanatory power.38

The individual responses were computed as aggregated, average workplace unit scores and subsequently assigned to each employee at the specific work unit independently of their individual response. A workplace unit was defined as the lowest organisational level up to the first level of management ranging in size from four to 120 workers. This approach was used, because it was assumed that an aggregated workplace unit score was a better reflection of the objective workplace environment than the individual scores.

Scores on satisfaction with working conditions were divided in three levels: a low satisfaction ‑ less than six points, an intermediate satisfaction ‑ six to eight points and a high satisfaction ‑ eight or more points, corresponding to the 25 and 75 percentiles. The satisfaction score was also entered into analyses as a continuous variable.

Outcome measure

The outcome measures were first-ever diagnoses of affective disorders (referred to as depressive disorders), neurotic, stress-related and somatoform disorders (referred to as anxiety disorders) and mental and behavioural disorders due to psychoactive substance use (referred to as substance abuse disorders), classified according to ICD-8 up to 1994 and from 1995 onwards according to ICD-10 (codes F10-19, F32-39, F40-48 respectively).39 The disorders were analysed separately and pooled, which provided four outcome categories namely depressive disorders (F32-39), anxiety disorders (F40-48), substance abuse disorders (F10-19) and all. Additionally, reaction to severe stress and adjustment disorders (F43) was analysed separately, as these disorders by definition are associated with psychosocial strain.40 Psychiatric records on all cohort members since 1969 up to baseline were obtained and participants with previous diagnoses of any of the studied disorders under the study were excluded to restrict analyses to first-ever diagnosis during the follow-up period.

Statistical analyses

Incidence rates, using HRs, of hospitalisation and outpatient treatment were compared for the studied mental health disorders among groups of employees working in units with high (reference), intermediate and low levels of work climate satisfaction. Follow-up started 1 January 2002 or when cohort members entered the work unit, whichever came last, and ended when they discontinued their job, died, emigrated, were hospitalised or were treated for mental health disorders or at 30 April 2008, whichever came first. Employees with a prior diagnosis of mental health disorders were excluded (n=35). HRs with 95% CIs were computed as a measurement of the association between work climate satisfaction and the mental health disorder of interest.

The levels of work climate satisfaction were entered into the models as dummies of the lower quartile and the two intermediate quartiles combined, using the upper quartile (most satisfied with work climate) as the reference. Models were adjusted for potentially confounding factors, which were introduced as dummy variables: gender (male yes/no), age at start of follow-up (41–50, 51–60, >60 with <40 as reference category), marital status (single, reference married/cohabitating) children less than 15 years at the residence (yes/no as reference) and occupational grade that was categorised as follows; low: nursing aids and unskilled service workers; middle-range: nurses, social care workers, office clerks, other health professionals, primary school teachers, technical assistants and craftsmen; and high: physicians, secondary school teachers and academics, with the latter as reference category (occupational grade is used as a proxy for educational level as they are closely correlated) and the size of work unit (under 30/30 and more).

Results

Sociodemographic characteristics of the study cohort according to perception of work climate are presented in table 1. The cohort comprised 13 423 participants, employed at 683 work units, with 62% of work units having less than 30 employees. The largest proportions of participants were women (79%) and worked in health services, mainly hospitals (70%).

Table 1

Sociodemographic characteristics of the study cohort according to overall satisfaction with psychosocial working conditions

During a total of 55.901 years of follow-up, there were 125 first-occurrences of psychiatrically diagnosed depressive disorders, anxiety disorders or substance abuse disorders.

Table 2 shows the association between work climate satisfaction and the risk of psychiatrically diagnosed mental health disorders. For all diagnoses combined, a low satisfaction level was significantly associated with a risk of any mental health disorder (HRadj 1.71, 95% CI 1.04 to 2.82). A dose‑response effect was observed, showing that the lower the satisfaction level, the higher the risk of mental health disorders (table 2). When using work climate satisfaction as a continuous variable, a one-point decrease in satisfaction level on the 11-point scale (0–10) was found to increase the risk of mental health disorders by 20% (HRadj 1.20, 95% CI 1.01 to 1.42).

Table 2

Hazard rates of mental disorders according to levels of satisfaction with psychosocial work climate

Regarding specific disorders, substance abuse disorders revealed the highest and significant association with low satisfaction with work climate, HRadj 9.18, 95% CI 1.16 to 72.59 (data not shown). However, due to a small number of cases CIs are wide. Being male was a significant risk factor of substance abuse, HRadj 3.53, 95% CI 1.55 to 8.03. A one-point decrease in satisfaction level increased the risk of substance abuse disorders by 48% (HRadj 1.48, 95% CI 1.04 to 2.10, data not shown).

The risks of remaining disorders are presented in table 2. Although none of the risks were significant, the risk pattern showed a dose‑response trend.

Regarding background variables, a low occupational grade was a significant risk factor for any mental health disorder throughout all analyses (HR 2.25, 95% CI 1.12 to 4.51) compared to a middle-range occupational grade (HR 0.99, 95% CI 0.51 to 1.93). Among the occupational groups, unskilled service workers had the highest risk of depressive disorders (HR 2.35, 95% CI 1.13 to 4.92), anxiety disorders, (HR 2.36, 95% CI 1.00 to 5.56), substance abuse disorders (HR 1.99, 95% CI 0.61 to 6.57) and severe stress and adjustment disorders (HR 3.41, 95% CI 1.34 to 8.68). No other occupational groups had a significant increased or decreased HR of developing mental disorders.

The risk of developing mental disorders was lower in older age groups (age 51–60 years: HR51‑60 0.46, 95% CI 0.26 to 0.80, >60 years: HR>60 0.12, 95% CI 0.02 to 0.90 respectively), than in younger age groups (41–50 years: HR41‑50 0.93, 95% CI 0.61 to 1.40.) Being single had a borderline significant HR of 1.36, 95% CI 0.92 to 1.99 of developing any mental disorder. No other background variables had a significant impact on HR.

Discussion

A low satisfaction with psychosocial working conditions was found to carry a significant risk of developing any of the studied psychiatrically diagnosed mental disorders. A one-point decrease in the level of satisfaction increased the risk of any mental disorder by 20%. However, the risks were significant only for the pooled diagnoses and substance abuse disorders. The highest, but not significant, risks were found for depressive and reaction to severe stress and adjustment disorders. Occupational grade and age were independent risk factors, but the risk estimates changed only marginally when adjusting for gender, age and occupational grade.

The present results are in line with a recent meta-analysis concluding that overall satisfaction with work climate was associated with mental health problems.41

In contrast to earlier studies,8 15 18 20 21 23 26 the present study found no gender differences in the risk of mental health disorders, except in the case of substance abuse disorders. As 79% of the study population were women, it is likely that only strong gender-related effects could be detected.

Methodological issues

The strengths of this study are its longitudinal design and the independency between exposure and outcome measures. The follow-up period has the benefit of providing sufficient time of exposure to create effects on the outcome variable; however, changes may still have occurred in both exposure and outcome variables within the follow-up period. Exposure data was collected independently of the psychiatrically diagnosed outcome variables, thus eliminating bias due to common methods variance. Furthermore, the cases are limited to patients suffering from severe mental health disorders and thus a sufficient contrast between the cases and the non-cases is provided and thereby a strengthening of the results.

The study addresses only more severe cases of mental disorders that require a hospital psychiatric treatment, and there is no information on subclinical symptoms of mental disorders that are likely to affect employees' vulnerability to job strain. Still, there might be a discrepancy between those who seek treatment for mental disorders and those who do not. Furthermore, individuals who are treated by their GP are not included in the present study. A study by Alonso et al42 showed that only one-third of people suffering from mood disorders seek treatment and the number is even lower for people suffering from anxiety disorders. One-third of all consultations are conducted by GPs only. Although the DCPRR ensures complete data on psychiatrically diagnosed mental disorders, it is clear that the present findings may not apply to mild cases of depression. On the other hand, it is unlikely that mental strain in the workplace is related only to severe depression.

The largest group in the public service sector is human service professionals, working primarily in the health sector and elderly care (see table 1). In Denmark these occupations are predominated by women. It is likely that these professionals have easier access to professional help and that suffering from mental disorders is more acceptable in this type of work environment. They may, therefore, be overrepresented in DCPRR.

The use of a single-item measure of satisfaction with work climate provides an easy and efficient measure of a global assessment of psychosocial factors at the workplace, which had in an earlier study proved to be stronger than any single scale predictor of sickness absence.38 It is also advantaged by lack of item-bias problems related to scales combining several questions.38 43

A single-item measure is useful as a screening instrument, but has its limitations with respect to recommendations regarding improvement of psychosocial working conditions as such rather crude measure does not make it possible to identify the specific aspects of psychosocial work environment factors that may constitute risks to mental health, if any.

The use of aggregated work unit satisfaction levels, which constitutes an independent exposure measure, lifts the findings out of the triviality trap but introduces a possible misclassification of exposure. It jeopardises assumptions of independent observations and psychosocial strain at the individual level and diminishes exposure contrast. Additionally, by using aggregated measures, cognitive appraisal processes are ignored. In contrast to the present study, a recent study by Kouvonen et al19 found association between individually assessed low workplace social capital and self-reports of psychiatrically diagnosed depression, whereas no association was found when using aggregated workplace social capital scores. Although items constituting social capital seem close to items in the work climate questionnaire, the discrepant findings can also be explained by different measures of exposure or outcome.

Data on potential confounding factors such as genetics, personality traits and life events were not available. This may have inflated the results, as these variables are known to be related to individual susceptibility to mental health disorders as well as proneness to report dissatisfaction with work climate. Difficulties in tackling these types of confounding problems are reflected by an ongoing debate of whether, for example, negative affectivity should be controlled for, following the controversy concerning it being a substantive or a biasing factor.44–46

All analyses were adjusted for several sociodemographic background variables, which should have minimised confounding by these factors. However, as measures used were rather crude there may still be a question of some residual confounding.

In order to eliminate the most susceptible individuals, employees with prior diagnosis of depressive, anxiety or substance abuse disorders were excluded, as there is a high recurrence of these mental health disorders.2 However, it is not possible to eliminate a vicious cycle, namely that suffering from depressive symptoms diminishes coping skills, which further impairs role functioning, and thus increases perceived job stress, which in turn constitutes a risk of depression.

Conclusion

The study showed that overall dissatisfaction with psychosocial working conditions is related to a significantly increased risk of developing psychiatrically diagnosed depressive, anxiety or substance abuse disorders. Actions to improve employees' satisfaction may be a promising path in preventing work related mental health problems. However, more preferably controlled randomised intervention studies are needed to disentangle causal pathways between specific psychosocial work environment exposures and the risk of specific mental disorders and their subclinical symptoms.

What is already known on this subject

  • Several psychosocial work environment factors have been shown to be related to symptoms of mental health problems, but whether associations are causal remains disputable.

  • Earlier studies using a general single item and aggregated measures of satisfaction with working conditions have failed to show association with psychiatrically diagnosed mental disorders.

What this study adds

This prospective study of a large cohort of public service employees that applied an aggregated, single-item measure of satisfaction with psychosocial working conditions has shown that general dissatisfaction with psychosocial work condition is a significant risk factor for psychiatrically diagnosed mental health disorders.

Policy implications

  • A global, single-item measure of satisfaction may be an effective screening instrument for identifying workplaces where psychosocial work environment constitutes a threat to employees' mental health.

  • Improving satisfaction of employees may decrease the risk of mental health disorders but requires identification of specific aspects of the psychosocial work environment that cause dissatisfaction. Further knowledge regarding prevention is of a great importance as mental health disorders carry high social and economic costs for employees and employers.

References

Footnotes

  • Funding The study was supported by a grant from the Danish Work Environment Research Fund (54-2003-08) and the Danish Research Council (2136-07-0049).

  • Competing interests None.

  • Ethical approval The data used in the study are register data and the use was approved by the Danish Data Protection Agency. There was no direct contact with human subjects and therefore no approval from an ethics committee was required.

  • Provenance and peer review Not commissioned; externally peer reviewed.