Longitudinal relationship between cognition and depressive symptoms in older adults from the Virginia Cognitive Aging Project: a latent dual change score analysis

Abstract

Background: the direction of the longitudinal association between depression and cognition remains a topic of intense debate. A unidirectional association where depression impacts the change in cognition (or vice-versa) and a bidirectional association where the trajectories of both dimensions affect each other lead to different clinical implications. Method: this study aimed to investigate the directionality of the depression-cognition association in a sample of 2,057 older adults aged between 60 to 99 years old from the Virginia Cognitive Aging Project (VCAP). We used the bivariate dual change score model to investigate the association between five cognitive domains (memory, speed, reasoning, space, and vocabulary) and three dimensions of depression (somatic, depressed affect, and positive affect) throughout five measurement points with an average interval of 2.5 years between the assessments. Three directions were tested for each pair of variables: two unidirectional models and one bidirectional model. Results: a unidirectional effect in which depression at time t predicts change in cognition presented the best data fit for most of the cognitive domains. One significant causal effect was found: depressed affect (DA) at time t significantly predicted the change in memory with a negative association, that is, higher levels of DA at baseline predicted a worse performance on memory over time (γDep = -0.32881; SE = 0.17; p < 0.05). Conclusions: our findings support a unidirectional association with depression predicting change for most of the cognitive domains tested. Higher levels of DA in older adults predicted an accelerated decline in memory.