Is the association between education and depressiveness mediated by work stress and social isolation and moderated by regional unemployment rate? – A confirmatory moderated lower level mediation model of the lidA (leben in der Arbeit)-study

Introduction/Background: As a consequence of the demographic change the German work force is ageing and shrinking. Therefore older employees are of rising value to hold up the German productivity, competitiveness and welfare state. Goal of the lidA-study is to identify factors which keep ageing German employees healthy and motivated at work. Depressiveness is one of the leading reasons for days absent from work, loss of productivity and early retirement. Social differences to the disadvantage of lower status groups in depressiveness are well known. Our question was by which pathways these differences could be explained. We hypothesized that part of the social differences in depressiveness could be explained by work-related stress and social isolation and there may be regional differences in this association which could be explained by regional unemployment rate. Data/Methods: This confirmatory analysis was conducted on the 6339 employees of the first wave of the lidA-cohort study. Work-related stress was parameterized by effort-reward imbalance (ERI-ratio), depressiveness by the short version of the Becks depression inventory (BDI-V). Education was measured by a combined score of school education and vocational training, social isolation by a simple item sum score. Regional unemployment rate came from the Federal Statistical Office. First level mediation was defined according to the criteria of Baron and Kenny (1986). Moderated lower level mediation was analyzed by the method of Bauer et al (2006). The multilevel-model was adjusted for gender, negative affectivity and overcommitment. Age-specific differences were analysed by age-stratification. Results: There was a stronger relationship between educational level and depressiveness in the younger age cohort. Whereas in the younger age cohort education was both, directly and indirectly (i.e. mediated by work stress), significantly inversely associated with depressiveness in the older age cohort only the indirect association was significant. In both cohorts social isolation was a significant non-work related mediator in the association between education and depressiveness after control for covariates. Considering the total effect there was a clear inverse association between education and depressiveness in the younger but not in the older cohort. Regional variation between work-related stress and depressiveness was observed for both age cohorts. Yet, the regional variation was not explained by regional unemployment. No regional variation was found for the association between education and depressiveness or education and work-related stress. Discussion/Conclusion: Our results indicate that younger employees with lower SES might have a higher probability of showing depressive symptoms. There was also a stronger indirect association between educational level and depressiveness mediated by work-related stress in the younger cohort. This observation could be of preventive value as a social gradient in depressiveness could be reduced by preventive measures lowering work-related stress. Regarding the stronger direct association between educational level and depressiveness in the younger age group improvements in vocational training in younger lower educated employees could have an additional preventive effect in the long run. Both preventive actions could be worthwhile with regard to preservation of work power and competitiveness of an ageing and shrinking German work force.