Objectives To investigate the extent improvement or deterioration in employee job security, control or workload is associated with a change in mental health.

Design Self-report panel data (2000, 2004) on mental health (symptoms of depression and generalised anxiety) and job demands, control and insecurity. Changes in exposures and outcomes were calculated by subtracting wave 1 from wave 2 scores. Changes in mental health were regressed onto changes in work conditions, adjusting for confounders. Sensitivity analyses assessed reverse causation, floor and ceiling effects.

Setting Two adjoining cities in south-east Australia.

Participants 1975 employees aged 40–48 years, 50% (n=995) male.

Results Improvements and deterioration in each work condition were associated with corresponding improvements or deterioration in mental health. The association between changes in job insecurity and symptoms of depression was B=0.386 (95% CI 0.245 to 0.527) and with anxiety symptoms was B=0.434 (95% CI 0.267 to 0.601). Similarly, changes in job control were associated with changes in depressive (B=−0.548; 95% CI −0.791 to −0.304) and anxiety symptoms (B=−0.608; 95% CI −0.896 to −0.319) as were changes in job demands (B depression=0.386; 95% CI 0.245 to 0.527; B anxiety=0.434; 95% CI 0.267 to 0.601). Excluding people with severe symptoms at baseline did not alter the findings; however, path analyses indicated that depression may precede a worsening of work conditions.

Conclusion Among mid-aged employees, deteriorating work conditions may amplify population health burdens, especially anxiety. Furthermore, better quality jobs, combining an array of positive conditions, could alleviate major population health burdens.