Objective Unemployed individuals have poorer mental health than the employed, but the direction of this association is not clear. Our objective was to examine the links between major depression and unemployment in a European cohort of general practice attenders.

Design Prospective cohort.

Setting General practices in UK, Estonia, The Netherlands, Portugal, Slovenia, Spain and Chile.

Participants Consecutive general practice attendees aged 18 to 75 years recruited April 2003-September 2004 (n=9793), followed up at 6, 12 and, in a subset (n=3259), 24 months.

Main outcome measure The Depression Section of the Composite International Diagnostic Interview was completed at baseline, 6, 12 and 24 months. Employment status was self-reported at baseline, 6 and 24 months.

Methods Random effect regression models or robust standard errors were used to account for clustering at General Practice level. Multinomial logistic regression models investigated whether unemployment (at baseline or 6 months) predicted onset of depression at 12 months. Logistic regression models investigated whether major depression (at baseline or 6 months) predicted onset of unemployment at 6 or 24 months. All models were adjusted for age, gender, level of education and country of origin. Gender interactions were examined.

Results 3969 participants were employed or unemployed and seeking work at baseline and at 6 months. 6% (n=221) had major depression at 12 months. Participants who were employed baseline but unemployed at 6 months compared to those employed at both time points had adjusted RRR of 1.63 (95% CI 1.02 to 2.60) for presence of depressive symptoms and 1.50 (95% CI 0.83 to 2.72) for major depression at 12 months. Gender differences were not significant. 12% (n=465) were unemployed at 6 months. The adjusted OR of being unemployed at 6 months, for participants with major depression at baseline and 6 months was 1.63 (95% CI 1.01 to 2.64). The OR of unemployment at 24 months associated with major depression at baseline or 6 months in men was 3.52 (95% CI 0.97 to 12.75) adjusted for lifetime depression and was further attenuated on adjustment for prior employment. In women the equivalent OR was 0.98 (95% CI 0.54 to 1.79).

Discussion There is some evidence that both causation and health selection result in raised levels of depressive symptoms in the unemployed. General practitioners should carefully monitor recently unemployed adults for onset of depressive symptoms and possible subsequent major depression. Adults with depression may have raised risks of subsequent unemployment and so should be supported at work.