Joint Modeling of Multivariate Longitudinal Depressive Symptoms and Survival with Application to an Aging Study

Background: The primary aim of this study was to use a joint analysis approach in order to examine the association between longitudinal depressive symptoms and survival in Mexican Americans. Methods: The separate Cox regression and joint modeling were applied to data from the Hispanic Established Population for Epidemiological Study of the Elderly (HEPESE). Depressive symptoms were measured by the Center of Epidemiological Studies Depression Scale (CES-D). The trajectories of CES-D, modeled by random effect, were used as independent variables to fit the mortality curve adjusted by other variables including demographics and physical functioning. Results: The separate Cox regression couldn’t identify association between depressive symptoms and survival. The joint analysis indicated that the slope of CES-D score was not associated with mortality in older Mexican-Americans, however the intercept had negative effects on mortality. Conclusion: There is significant association between baseline depression symptoms and mortality, whereas there is no association with slope in older Mexican Americans