Background
On a global scale, depression is one of the leading causes of premature mortality and disability-adjusted life years[
1]. Despite uncertainty about the scope of a potentially increased incidence in recent years[
2] depression makes a significant contribution to the global burden of disease and associated costs, in particular in rapidly aging populations[
3‐
5]. On a different level, work stress is now considered a growing threat to the health of employed people, especially so in association with aggravated competition, work intensification and job insecurity resulting from rapid spread of free market principles in a globalized economy[
6]. Whether there is an association between work stress and depression has been explored in a number of epidemiological studies (for reviews see[
7‐
10]. Although results are not fully consistent significantly increased odds ratios of incident depression were documented in a majority of cohort studies analysed in these reviews. Given the significance of exposure time for disease incidence one can assume that the association of stressful work with depression is particularly strong among older employees. Yet, this hypothesis has rarely been tested as most studies were conducted in middle-aged working populations[
7‐
10]. It is therefore important to analyse the association of work stress with depression in older employees. A further shortcoming of the current state of the art relates to the fact that most studies testing this hypothesis were restricted to samples of working men and women from a single country or from different countries, mostly within Europe[
11‐
13]. To our knowledge, no study has yet tested the hypothesis in working populations recruited from a variety of countries in different continents.
We set out to fill in these gaps of knowledge by analysing the association of work stress with depressive symptoms in a large sample of older employed men and women (aged 50 to 64) from several countries recruited from Europe, the USA and Asia. We test this association both cross-sectionally (17 countries) and longitudinally (13 countries) (see Methods). Despite their importance in times of growing economic globalization, studies that investigate cross-country comparisons face a considerable methodological problem as it is essential to ensure that the core constructs under study have an identical meaning across different cultures. Moreover, the measurement of these constructs should result in comparable information across the different data sets. In case of clinically relevant depressive symptoms this challenge has been met rather successfully, both at the level of professional judgments based on internationally valid criteria and assessment tools[
14‐
16], and at the level of self-reported standardized questionnaire data assessing depressive symptoms (see Methods).
Can we define stressful work in an identical way and provide a comparable assessment? To answer this question a theoretical model is needed. It is the function of a theoretical model to delineate particular stressful work conditions so that they can be identified at a level of generalization that allows for their use in a wide range of occupations in different socio-economic and socio-cultural contexts. While several such theoretical concepts have been developed[
17], two models received special attention in international research and were applied in epidemiological studies in a variety of working populations in different countries, the job strain (or demand-control) model and the effort-reward imbalance model. The job strain model defines work stress in terms of a distinct job task profile where jobs defined by high demands in combination with low decision latitude or low task control are stressful[
18]. A complementary model, effort-reward imbalance, focuses on the work contract and the principle of social reciprocity lying at its core. Rewards received in return to efforts spent at work include money, esteem, and career opportunities (promotion, job security). The model asserts that lack of reciprocity (high effort in combination with low reward) occurs frequently in a globalized economy and generates strong negative emotions and psychobiological stress responses with adverse long-term effects on health[
19]. Both models are measured by standardized, psychometrically validated questionnaires that are available in a variety of languages[
20,
21].
In summary, we analyse associations of job strain and effort-reward imbalance at work with depressive symptoms in older employed men and women recruited from countries in three continents, applying validated, cross-culturally comparable methods.
Results
The prevalence of current depressive symptoms varied considerably between countries, with a majority of countries exhibiting less than 14 percent. Yet, in a few countries a higher prevalence was observed (19,2 percent in Belgium, 23,0 percent in Italy, 26,1 percent in France, and 33,9 percent in Poland). Concerning the three world regions, the prevalence of current depression was lowest in Japan, medium in the USA, and highest in Europe.
To test our hypothesis, multivariate logistic regressions were calculated separately for the three regions of the world. Table
2 presents the results from cross-sectional data. Accordingly, odds ratios of depression were significantly elevated among men and women experiencing work stress in terms of high effort-reward imbalance in all three regions, compared to those who were free from stress at work. The strongest association was observed in the USA, followed by Europe and Japan. A similar trend was obvious for the complementary measure of work stress, low control. However, after adjusting for confounders the association was no longer statistically significant in Japan.
Table 2
Associations of work stress with depressive symptoms (multivariate logistic regression analyses) (N = sample with complete data)
Effort-reward imbalance | Yes | 2.28 (1.59 – 3.28) *** | 1.97 (1.75 – 2.23) *** | 1.64 (1.02 – 2.63) * |
| No (Ref.) | | | |
| N= | 1510 | 10047 | 1083 |
Low Control | Yes | 2.26 (1.57 – 3.25) *** | 1.66 (1.45 – 1.90) *** | 1.59 (0.88 – 2.86) |
| No (Ref.) | | | |
| N= | 1560 | 10342 | 1226 |
Findings from longitudinal analyses are displayed in Table
3, with the restricted sample of employed and self-employed men and women from 13 countries. Significantly elevated odds ratios of depression, assessed in 2006, are observed among men and women with a high level of work stress in terms of effort-reward imbalance and low control, assessed in 2004. Gender-specific analyses showed no statistically significant differences in effect sizes. Concerning the two statistical models, all odds ratios in the second model are attenuated once baseline depression in 2004 has been taken into account. In this case, elevated odds ratios for both work stress models remain statistically significant in the combined European samples (SHARE and ELSA). Here, an odds ratio of 1.51 (effort-reward imbalance) and of 1.42 (low control) respectively is observed.
Table 3
Associations of work stress with prospective depressive symptoms (multivariate logistic regression analyses)
Effort-reward imbalance (2004) | Model1 | 1.94 (1.21 – 3.13) ** | 1.79 (1.53 – 2.09) *** |
| N= | 683 | 6063 |
| Model2 + symptoms 2004 | 1.53 (0.91 – 2.57) | 1.51 (1.28 – 1.78) *** |
| N= | 673 | 6034 |
Low control (2004) | Model1 | 1.65 (1.03 – 2.63)* | 1.57 (1.34 – 1.84) *** |
| N= | 714 | 6246 |
| Model2 + symptoms 2004 | 1.46 (0.88 – 2.45) | 1.42 (1.20 – 1.68) *** |
| N= | 704 | 6213 |
The results of additional sensitivity analyses with self-rated health as outcome were similar to those obtained with depressive symptoms. Moreover, sensitivity analyses were conducted to confirm the longitudinal findings. First, we repeated analyses by excluding participants who retired between 2004 and 2006. Second, we restricted our mental health outcome to participants with incident depressive symptoms between 2004 and 2006. In either case, results remained basically unchanged (results not shown). The magnitude and statistical significance of respective odds ratios were well comparable in case of the European and American study samples, both in cross-sectional and longitudinal analyses. However, in Japan, low control was not associated with poor subjective health. Furthermore, instead of analysing the full sample and to adjust for baseline depressive symptoms, we repeated longitudinal analyses by excluding all participants with depressive symptoms at baseline, thus estimating associations with newly manifested depressive symptoms. Again, results were in line with the analyses displayed in Table
3. More specifically, odds ratios in case of effort-reward imbalance were 1.64 (0.87-3.08) for USA and 1.59 (1.29-1.96) for Europe, and in case of low control 1.69 (0.92-3.10) for USA and 1.52 (1.23-1.88) for Europe.
Discussion
Our results demonstrate that work stress, as measured by core components of two major theoretical models, job strain and effort-reward imbalance, is associated with significantly increased odds ratios of depressive symptoms in cross-sectional and, in part, in longitudinal analyses across different regions of the world. Based on cross-sectional data, the strength of associations is comparable between the three analyses conducted for countries from the three regions Europe, USA and Asia (Japan). Among employed men and women aged 50 to 64 exhibiting a high level of work stress there is an increased probability of depressive symptoms varying from about 60 percent to about 120 percent, compared to those with no or low level of work stress. Findings from longitudinal analyses are restricted to samples from USA and Europe. Although trends are similar to those reported from cross-sectional analyses, effects are attenuated after adjusting for baseline depression, where elevated odds ratios remain significant in Europe only – a finding that might be explained by the large sample size compared to the US sample.
Taken together, our results lend preliminary support to the notion that work stress is associated with an elevated probability of experiencing clinically relevant depressive symptoms in employed men and women of early old age recruited from a variety of countries in different regions of the world. Observed effects remain statistically significant after adjustment for gender, age, education, employment status, working hours and country, but adjustment for baseline depression systematically reduces the size of effects in longitudinal analyses. This latter observation is well known from other investigations and may in part be due to common method variance between two sets of simultaneously assessed self-report measures[
43]. Our findings are consistent with those reported from a single country or from studies comparing European countries[
11‐
13,
44].
To the best of our knowledge, this is the first report demonstrating an association of work stress with depression in early old age populations across a variety of countries from different regions of the world, using identical or highly comparable measurements of the constructs under study. However, there are several limitations to our analysis. First, the number of countries from which longitudinal data are available is smaller than is the case with cross-sectional data. Therefore, respective findings cannot be generalized to all samples under study. Moreover, results are restricted to the specific age group, and the Japanese sample was not intended to be representative of the respective national population group. Inclusion of more Asian countries would provide more robust estimates. In fact, two more aging studies with comparable study designs have been initiated meanwhile, in South Korea and in China, thus offering an extension of the current analysis in future research.
Second, we restricted the analysis of work stress to two core theoretically grounded notions of stress, lack of control and failed reciprocity between effort and reward[
45]. Although these theoretical notions are particularly relevant in delineating stressful working and employment conditions in the context of economic globalization future studies may broaden the range of work-related stressors, e.g. by including organisational injustice[
8] and precarious work[
46]. However, given the restrictions of applying short questionnaires in multi-disciplinary international investigations we were not in a position to measure all components of the two models with complete scales and to include additional theoretical notions. Yet, rather than overestimating the effects of work-related stress, this may have led to underestimating effect sizes.
A third limitation concerns the measurement of depression. Despite satisfactory criterion validity the two self-administered scales measuring depressive symptoms may not capture clinically relevant conditions to a sufficient extent, as is the case with an established standardized approach based on clinical interviews[
15]. Yet, correlations of the two scales, the CES-D scale and the Euro-D scale, were high, and both scales demonstrate convincing psychometric properties. Fourth, cultural variations in experiencing and expressing stressful work as well as depressive symptoms may not have been adequately captured by short standardized questionnaire methods[
15]. Therefore, in addition to applying item response theory to improve cross-cultural comparability of the scales[
29] qualitative studies are desirable to validate these preliminary findings[
47]. A more comprehensive understanding is expected from such studies, specifically in case of explaining the observed differences in effect sizes of work stress on depression between Japan and the other countries or between men and women (although, in our data, gender differences were not significant). In this context, socio-cultural differences in recall of depressive symptoms may also need a more detailed analysis[
48]. Finally, information on mental health in this analysis was restricted to depressive symptoms. The inclusion of additional health indicators would further support the hypothesis. In fact, in sensitivity analyses we repeated all calculations presented above with data on ‘poor self reported health’ as an outcome criterion. Results were highly comparable to the ones given in Tables
2 and
3. In further sensitivity analyses excluding participants who retired between 2004 and 2006 and focusing on incident depressive symptoms findings were almost identical to the ones presented in the Results section (results not shown).
These limitations are balanced by several strengths. All epidemiological studies on aging populations were designed in close coordination, and the data collection procedure met internationally established quality standards[
22‐
24]. The questionnaires applied to measure the core constructs, work stress and depression, were either fully identical or well comparable, thus offering opportunities to conduct intercultural comparisons. Furthermore, the four cohorts represent a relatively homogenous age group, 50 to 64 years, where exposure to work stress can be assumed to have occurred for a couple of years or even decades. In this age group, the prevalence of depressive symptoms is substantial, and the policy dimension of this fact is indicated, among others, in the prominent role of depression as one of the most frequent diagnoses justifying disability pensions in several countries included in this study[
3]. Finally, the two work stress models underlying this analysis were successfully applied to explain stress-related mental and physical disorders in a large number of prospective epidemiological studies[
10,
12,
44,
49], and the psychobiological pathways linking lack of control and frustration of rewards to disease were tested in a variety of experimental studies[
50,
51].
What are the policy implications of these results? At the international level, coordinated efforts are needed to strengthen measures that mitigate health-adverse effects of stressful work, e.g. by reducing the impact of neoliberal policies on labour market standards and quality of work. At the national and local levels, investments into worksite health promotion programmes and occupational health services targeting occupational high risk groups contribute to a reduction of the burden of stress-related disorders, given the high prevalence of low control and effort-reward imbalance amongst the workforce with low qualification[
52]. Preliminary favourable results from intervention studies that increase control and reward at work by implementing theory-based organizational changes support this notion[
53].
Acknowledgement
This paper uses data from SHARE release 2.3, as of November 2009. The SHARE data collection has been primarily funded by the European Commission through the 5th framework programme (project QLK6-CT-2001-00360 in the thematic programme Quality of Life), through the 6th framework programme (projects SHARE-I3, RII-CT- 2006-062193, COMPARE, CIT5-CT-2005-028857, and SHARELIFE, CIT4-CT-2006-028812) and through the 7th framework programme (SHARE-PREP, 211909 and SHARE-LEAP, 227822). Additional funding from the U.S. National Institute on Aging (U01 AG09740-13S2, P01 AG005842, P01 AG08291, P30 AG12815, Y1-AG-4553-01 and OGHA 04-064, IAG BSR06-11, R21 AG025169) as well as from various national sources is gratefully acknowledged (see
http://www.share-project.org for a full list of funding institutions).
Data from ELSA were made available through the UK Data Archive (UKDA). ELSA was developed by a team of researchers based at the National Centre for Social Research, University College London and the Institute for Fiscal Studies. The data were collected by the National Centre for Social Research. The funding is provided by the National Institute of Aging in the United States, and a consortium of UK government departments co-ordinated by the Office for National Statistics. The developers and funders of ELSA and the Archive do not bear any responsibility for the analyses or interpretations presented here.
The HRS (Health and Retirement Study) is sponsored by the National Institute on Aging (grant number NIA U01AG009740) and is conducted by the University of Michigan.
The Japanese Study of Aging and Retirement (J-STAR) was conducted by the Research Institute of Economy, Trade and Industry (RIETI) and Hitotsubashi University, Japan.
This research was conducted in the frame of research grant of the Hans-Boeckler-Foundation, Duesseldorf, Germany (project number: S-2009-311-4).
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
All authors read and approved the final manuscript. JS proposed the analysis and drafted the manuscript. TL and MW jointly conducted the empirical data analysis, helped with literature search and revised the draft. ND contributed to the study design, revised the manuscript and contributed to its final version.