Context: Depressive symptoms are common in older persons, and may predict mortality.
Objectives: To determine: (1) If depressive symptoms predict mortality; (2) If there is a gradient in this effect; and (3) Which depressive factors predict mortality.
Population: In 1991-1992, 1751 community-dwelling older persons, sampled from a population-based registry, were interviewed.
Measures: The Center for Epidemiologic Studies - Depression (CES-D), age, gender, the Modified Mini-Mental State Examination, self-rated health, and functional status.
Outcome measure: Time to death.
Analysis: Those scoring 16+ on the CES-D were considered depressed. To determine if a gradient was present, the CES-D was treated as a continuous variable. Four depressive factors from the CES-D (depressed affect, positive affect, somatic, and interpersonal) were analyzed. Cox regression models were constructed.
Results: The mortality in those with depressive symptoms was higher in those without depressive symptoms (Hazard Ratio of 1.71, p < 0.001, Log rank test). In multivariable models, this association was no longer significant after accounting for self-rated health and functional status. There was a gradient in risk of mortality across the range of the CES-D. Somatic factors, depressed affect, and positive affect were all associated with mortality in bivariate analyses, but not in multivariable models adjusting for functional status. Interpersonal factors were not associated with mortality.
Conclusions: Depressive symptoms predict mortality in older persons.