Hostility may explain the association between depressive mood and mortality: Evidence from the French GAZEL cohort study Running title: Hostility, depression and mortality

Background Depressive mood is associated with mortality. Because personality has been found to be associated with depression and mortality as well, we aimed to test whether depressive mood could predict mortality when adjusting for several measures of personality. Methods 20,625 employees of the French national gas and electricity companies gave consent to enter in the GAZEL cohort in 1989. Questionnaires were mailed in 1993 to assess depressive mood, Type A behaviour pattern, hostility, and the six personality types proposed by GrossarthMaticek & Eysenck. Vital status and date of death were obtained annually for all participants. The association between psychological variables and mortality was measured by the Relative Index of Inequality (RII) computed through Cox regression. Results 14,356 members of the GAZEL cohort (10,916 men, mean age: 49 years, 3,965 women, mean age: 46 years) completed the depressive mood scale and at least one personality scale. During a mean follow-up of 14.8 years, 687 participants had died. Depressive mood predicted mortality, even after adjustment for age, sex, education level, body mass index, alcohol consumption, and smoking [RII (95% CI) = 1.56 (1.16-2.11)]. However, this association was dramatically reduced (RII reduction: 78.9%) after further adjustment for cognitive hostility (i.e. hostile thoughts) [RII (95% CI) = 1.12 (0.80-1.57)]. Cognitive hostility was the only personality measure remaining associated with mortality after adjustment for depressive mood [RII (95% CI) = 1.97 (1.39-2.77)]. Conclusions Cognitive hostility may either confound or mediate the association between depressive mood and mortality. in se rm -0 04 64 81 3, v er si on 1 18 M ar 2 01 0