Longitudinal changes in the extended networks of depressive symptoms and cognitive functions following bereavement: comparison between progressively depressed and continuously non-depressed older adults

Abstract

Background Late-life depression and cognitive decline share a reciprocal relationship, but the impact of other contributing risk factors, such as bereavement, on this relationship remains unexplored. Temporal network modelling helps infer potential causal relationships between specific risk factors and depression onset. Objectives To compare the extended temporal networks of depressive symptoms and cognitive functions between older adults who developed depression after recent bereavement and those who remained resilient. Methods Unrelated adults over 70 were selected from the Longitudinal Study of Aging in Danish Twins based on three criteria: no prior diagnosis of depression, recent bereavement and complete data on depressive symptoms and cognitive function at two timepoints, set 2 years apart. At follow-up, 1770 remained non-depressed, while 222 developed depression. Cross-lagged panel network (CLPN) analyses were used to examine changes in 15 depressive symptoms and five cognitive functions over time. Results CLPN analyses revealed few temporal predictions between depressive symptoms and cognitive functions in the ‘resilient’ network. In contrast, the ‘depression-onset’ network displayed complex longitudinal interactions. While only one follow-up cognitive function (verbal fluency) was predicted by baseline depressive symptoms, all studied baseline cognitive functions (e.g. memory, attention) predicted variations in symptoms at follow-up, particularly thoughts of worthlessness, life not worth living and anhedonia. The ‘resilient’ group showed significant change in network structure over time, whereas ‘depression-onset’ group did not. Conclusions Following bereavement, resilient older adults showed a flexible extended network of depressive symptoms and cognitive functions. Lower baseline cognitive function within a structurally unchanging network appeared to trigger depression-onset.