The Gutenberg health study: a five-year prospective analysis of psychosocial working conditions using COPSOQ (Copenhagen psychosocial Questoinnaire) and ERI (effort-reward imbalance)

Abstract Background Psychosocial working conditions were previously analyzed using the first recruitment wave of the Gutenberg Health Study (GHS) cohort (n = 5000). We aimed to confirm the initial analysis using the entire GHS population at baseline (N = 15,010) and at the five-year follow-up. We also aimed to determine the effects of psychosocial working conditions at baseline on self-rated outcomes measured at follow-up. Methods At baseline, working GHS participants were assessed with either the Effort-Reward-Imbalance questionnaire (ERI) (n = 4358) or with the Copenhagen Psychosocial Questionnaire (COPSOQ) (n = 4322); participants still working after five years received the same questionnaire again (ERI n = 3142; COPSOQ n = 3091). We analyzed the association between working conditions and the outcomes job satisfaction, general health, burnout, and satisfaction with life at baseline, at follow-up and also prospectively from baseline to follow-up using linear regression models. We examined the outcome variance explained by the models (R2) to estimate the predictive performance of the questionnaires. Results The models’ R2 was comparable to the original baseline analyses at both t0 and t1 (R2 range: ERI 0.10–0.43; COPSOQ 0.10–0.56). However, selected scales of the regression models sometimes changed between assessment times. The prospective analysis showed weaker associations between baseline working conditions and outcomes after five years (R2 range: ERI 0.07–0.19; COPSOQ 0.07–0.24). This was particularly true for job satisfaction. After adjusting for the baseline levels of the outcomes, fewer scales still explained some of the variance in the distribution of the outcome variables at follow-up. The models using only data from t0 or t1 confirmed the previous baseline analysis. We observed a loss of explained variance in the prospective analysis models. This loss was greatest for job satisfaction, suggesting that this outcome is most influenced by short-term working conditions. Conclusions Both the COPSOQ and ERI instruments show good criterion validity and adequately predict contemporaneously measured self-reported measurements of health and (occupational) well-being. However, the COPSOQ provides a more detailed picture of working conditions and might be preferable for improvment strategies in workplaces. Additional prospective research with shorter follow-up times would be beneficial for estimating dose-response relationships.