Depressive Symptoms and Cognitive Function in Older Adults: A Cross-Lagged Network Analysis

Abstract

Background: Depressive symptoms commonly co-occur with cognitive decline in older adults. However, prospective interrelationships between different cognitive function domains and depressive symptoms are not well understood. This study evaluated prospective interrelationships between depressive symptoms and cognitive functioning components among individuals aged 50 years or older from a perspective of network analysis. Method: Longitudinal data from the English Longitudinal Study of Aging were analyzed. Depressive symptoms were measured with the eight-item Center for Epidemiologic Studies Short-Depression Scale. Cognitive functions assessed included memory, orientation, and executive function. Contemporaneous network analyses were conducted using mixed graphical model, while a temporal network model was assessed using cross-lagged panel network model. To identify important predictors and outcomes, centrality indices, including expected influence, out-expected influence, and in-expected influence, were calculated. Results: A total of 6,433 older adults were included in the network analysis. Baseline "Not enjoy life" (CESD-6) was negatively associated with executive function at the follow-up assessment. Moreover, improvements in "Everything was an effort" (CESD-2) and "Loneliness" (CESD-5) were related to less future decline of executive function and memory ability. Furthermore, analyses suggested targeting "Lack of happiness" (CESD-4) could be useful in reducing the co-occurrence of depression and cognitive decline among older adults. Conclusions: This network analysis study highlighted dynamic interrelationships between depressive symptoms and cognitive decline in older adults. Findings suggest that interventions targeting specific depressive symptoms may have the potential to alleviate declines in executive function and memory for this population.

Keywords: cognitive decline; cross-lagged network analysis; depression; executive function.

Figure 1

Contemporaneous networks of depression and cognitive components. Note: Contemporaneous networks for the baseline and follow-up datasets. The relationship of the components is indicated by the edge's color (blue = positive, red = negative), while the strength of the relationship is indicated by the edge's thickness (thicker = stronger).

Figure 2

Temporal network of depressive and cognitive components. Note: The cross-lagged panel networks for depressive and cognitive components. The relationship of the components is indicated by the arrow's color (blue = positive, red = negative), and the strength of the relationship is indicated by the arrow's thickness (thicker = stronger). In these networks, autoregressive effects are excluded, and sex, age, married status, education level, drink, and smoke confounding were controlled.

Figure 3

In-prediction and out-prediction of temporal network. Note: Symptom centrality estimates for the cross-lagged networks using z values. Greater values indicate greater centrality. Out-expected-influence is the degree to which individual symptom could predict other symptoms at the subsequent relevant point. In-expected-influence is the degree to which symptom is predicted by other symptoms at the subsequent relevant point.