0257 Associations of individual level and job-group level estimates of psychosocial work factors with depressive symptoms

Background Job exposure matrix (JEM) methodology is useful in occupational psychosocial epidemiology for eliminating reporting bias and analysing low-prevalence outcomes in register based populations. This investigation aims to compare patterns of associations between psychosocial factors, assessed by JEM estimates and individual-level estimates, respectively, with depressive symptoms and to test the linearity of the associations. Methods In this cross-sectional analysis, we used data from the Danish Work Environment Cohort Study 2000 (n=8583) to generate JEM and individual-level estimates of quantitative demands, work pace, influence, opportunities for development, emotional demands, and role conflicts at work. JEM estimates were attained from regression models providing sex- and age specific estimates for each job-group. Depressive symptoms were measured with the MHI-5 scale of the Short Form 36 questionnaire. The shape of the association between psychosocial exposures and depressive symptoms were assessed by use of linear splines. Using F-tests we tested whether reducing model flexibility (i.e. number of splines) led to statistically significant changes in model fit. Results Preliminary results indicate that associations between individual-level estimates of psychosocial work factors with depressive symptoms were largely linear and statistically significant. The associations of JEM estimates of psychosocial job factors with depressive symptoms showed varied patterns of non-linearity and were generally not statistically significant, after adjustment for individual-level measures. Discussion Our study indicates that individual estimates of psychosocial work factors are consistently, strongly and linearly associated with depressive symptoms, whereas JEM estimates showed varied and non-linear patterns. JEM psychosocial work estimates may capture different phenomena than individual-level estimates.