Does personality predict mortality? Results from the GAZEL French prospective cohort study Running head: Personality and mortality

Background: Majority of studies on personality and physical health have focused on one or two isolated personality traits. We aim to test the independent association of 10 personality traits, from three major conceptual models, with all-cause and cause-specific mortality in the French GAZEL cohort. Methods: A total of 14,445 participants, aged 39-54 in 1993, completed the personality questionnaires composed of the Bortner Type-A scale, the Buss-Durkee-Hostility-Inventory (for total, neurotic and reactive hostility), and the Grossarth-Maticek-Eysenck-PersonalityStress-Inventory that assesses six personality types (cancer-prone, coronary heart disease (CHD)-prone, ambivalent, healthy, rational, anti-social). The association between personality traits and mortality, during a mean follow-up of 12.7 years, was assessed using the Relative Index of Inequality (RII) in Cox regression. Results: In models adjusted for age, sex, marital status and education, all-cause and causespecific mortality were predicted by “total hostility”, its “neurotic hostility” component as well as by “CHD-prone”, “ambivalent” “antisocial”, and “healthy” personality types. After mutually adjusting personality traits for each other, only high “neurotic hostility” remained a robust predictor of excess mortality from all causes (RII=2.62; 95% CI=1.68-4.09) and external causes (RII=3.24; 95% CI=1.03-10.18). “CHD-prone” (RII=2.23; 95% CI=0.726.95) and “anti-social” (RII=2.13; 95% CI 0.61-6.58) personality types were associated with cardiovascular mortality and with mortality from external causes, respectively, but confidence intervals were wider. Adjustment for potential behavioural mediators had only a modest effect on these associations. Conclusions: Neurotic hostility, CHD-prone personality and antisocial personality were all predictive of mortality outcomes. Further research is required to determine the precise mechanisms that contribute to these associations. 1 H A L athor m anscript inerm -007304, version 1 in se rm -0 02 57 30 4, v er si on 1 18 F eb 2 00 9