Objectives Poor psychosocial working conditions are associated with increased risk of mental health disorders. Some studies suggest that work factors may also increase the risk of suicidal behaviour. However, these studies mostly measured suicidal ideation, rarely completed suicides, and hardly ever used longitudinal data. We aimed to examine the relationship between selected psychosocial work factors and death by suicide, as well as hospital treatment for intentional self-harm and similar diagnoses.

Methods Data on all persons born in Norway 1967–1976, employed in 2007 (n=420,895), were retrieved from national registers. Binary psychosocial work exposures were obtained from a job-exposure matrix for job demands, job control and leader support. Based on these, we constructed job strain (high demands, low control), and isostrain (high demands, low control, low support) exposure variables. These were linked to occupation in 2007 (ISCO88 four-digit codes). We estimated associations (Risk Ratios, RRs) between the psychosocial work factors and four-year risk of intentional self-harm and suicide in binomial regression models, adjusted for year of birth, education level, marital history and current family pattern.

Results During 2008–2011, 884 had diagnoses of intentional self-harm (annual rate 53 per 100,000; women 56, men 49), while 164 suicided (annual rate 9.7; women 4.5, men 14.5). Intentional self-harm was associated with all work factors, except job demands and leader support among men. Associations were strongest for high isostrain, adjusted RR 1.5 (95% confidence interval 1.2–1.7) and high job strain RR 1.3 (1.2–1.5), similar in women and men. Associations with suicide were positive only among men, with strongest associations for high job strain, RR 1.5 (1.0–2.1).

Conclusions Psychosocial work factors were associated with intentional self-harm in both sexes, but only in men for suicide. High isostrain and high job strain showed the strongest associations. Reasons other than a causal relationship cannot be excluded, particularly health selection.