Depression predicts mortality in the young old, but not in the oldest old: results from the Berlin Aging Study

There is evidence that depression in old age is associated with an increased mortality risk, but studies have also yielded inconclusive results. Possible moderators of the depression-mortality association in old age discussed in the literature are differences in cardiovascular morbidity, effects of multimorbidity, and increasing effects of subthreshold depression symptoms, such as minor depression and loneliness, on mortality. This study is concerned with the depression-mortality association in old and very old age.

Information about mortality status and dates of death for 497 participants of the Berlin, Germany, Aging Study (mean age: 85.16 years; range: 70-103 years), a population based, age-stratified, longitudinal study, were obtained from the Berlin, Germany, City Registry for up to 15 years. The authors calculated proportional hazard regression models to examine associations between clinical diagnosis of depression at baseline assessment and subsequent mortality for young-old (70-84 years; N = 243; 68% deceased) and oldest-old participants (85+ years; N = 254; 98% deceased). In an additional step, the authors examined whether depression-mortality associations remained after statistically controlling for the effects of other mortality predictors including age, gender, education, dementia, cardiovascular risk factors, and other somatic diseases.

Our analyses revealed strong predictive effects of depression diagnoses for mortality among the young old (Relative Risk = 1.60, 95% Confidence Interval = 1.13-2.26) that were not due to the effects of other mortality predictors (Relative Risk = 1.56, 95% Confidence Interval = 1.09-2.22). Among the oldest old, no depression-mortality associations were found.

Depression is a significant risk factor for all-cause mortality in old age, yet the risk conveyed by depression does not hold in very old age. Possible underlying mechanisms in the very old are discussed.