Understanding the protective effect of social support on depression symptomatology from a longitudinal network perspective

Abstract

Background: Higher social support protects people from developing mental disorders. Limited evidence is available on the mechanism through which social support plays this protective role.

Objective: To investigate the stress-buffering process of social support on depressive symptoms using a novel longitudinal dynamic symptom network approach.

Methods: A total of 4242 adult participants who completed the first two waves (from May to October 2020) of the International Covid Mental Health Survey were included in the study. Cross-lagged panel network modelling was used to estimate a longitudinal network of self-reported social support, loneliness and depressive symptoms. Standardised regression coefficients from regularised cross-lagged regressions were estimated as edge weights of the network.

Findings: The results support a unidirectional protective effect of social support on key depressive symptoms, partly mediated through loneliness: A higher number of close confidants and accessible practical help was associated with decreased anhedonia (weight=-0.033) and negative self-appraisal symptoms (weight=-0.038). Support from others was also negatively associated with loneliness, which in turn associated with decreased depressed mood (weight=0.086) and negative self-appraisal (weight=0.077). We identified a greater number of direct relationships from social support to depressive symptoms among men compared with women. Also, the edge weights from social support to depression were generally stronger in the men's network.

Conclusions: Reductions in negative self-appraisal might function as a bridge between social support and other depressive symptoms, and, thus, it may have amplified the protective effect of social support. Men appear to benefit more from social support than women.

Clinical implications: Building community-based support networks to deliver practical support, and loneliness reduction components are critical for depression prevention interventions after stressful experiences.

Keywords: COVID-19; adult psychiatry; depression & mood disorders.

Figure 1

A longitudinal network of social support, loneliness and depressive symptoms (N=4242). Arrows represent longitudinal association between different nodes. Elements for different constructs are in different colours. Nodes with higher connections are automatically clustered together by the Fruchterman-Reingold algorithm. Autoregressive edges and weaker edges with weights lower than 0.03 were excluded to ease visual interpretation as suggested by previous studies. OSSS, Oslo Social Support Scale; PHQ, Patient Health Questionnaire.

Figure 2

Centrality (A) and bridge centrality (B) indices for elements of social support, loneliness and depressive symptoms in the network. Bridge EIs that focus on cross-construction edges. Autoregressive effects are removed when calculating the indices. EI, expected influence; OSSS, Oslo Social Support Scale; PHQ, Patient Health Questionnaire.

Figure 3

Comparison between the men’s and women’s networks. (A) Men’s network of social support, loneliness, and depressive symptoms; (B) women’s network of social support, loneliness, and depressive symptoms. To facilitate visual comparison, the same layout with the network in full sample was used to plot the networks and a higher threshold for weak edges (0.05) was used for men’s network. (C) Comparison of centrality indices for the nodes; (D) comparison of bridge centrality indices for the nodes. PHQ1: anhedonia; PHQ2: depressed mood; PHQ3: sleep disturbance; PHQ4: low energy; PHQ5: appetitive disturbance; PHQ6: negative self-appraisal; PHQ7: concentration difficulties; PHQ8: psychomotor disturbance; PHQ9: suicide/self-harm; OSSS1: number of close confidants; OSSS2: number of close confidants: concern from others; OSSS3: accessible practical help from neighbours. EI, expected influence; OSSS, Oslo Social Support Scale; PHQ, Patient Health Questionnaire.