Network analysis of the link between depressive symptoms and risk of suicidal behavior at three years: A national...

Depressive symptoms predict suicidal behavior, but the specific pathways linking individual symptoms to suicide attempts remain incompletely characterized. Network analysis combined with cross-lagged panel models (CLPM) offers a framework to identify bridge symptoms and disentangle their temporal directionality.

Using Waves 1 (2001-2002) and 2 (2004-2005) of the National Epidemiologic Survey on Alcohol and Related Conditions, we examined longitudinal associations between depressive symptoms during a past-12-month major depressive episode (MDE) at baseline (N = 3485) and suicide attempts three years later (n = 102). We estimated an Ising network, computed bridge centrality, and fitted five a priori CLPMs using the WLSMV estimator. Two sensitivity analyses excluded participants with a lifetime manic episode or with past-12-month alcohol or non-nicotine substance use disorders.

Suicidal ideation at follow-up emerged as the strongest bridge symptom to suicide attempt (bridge expected influence = 0.282), followed by baseline suicidal ideation and thoughts of death. In CLPMs, worthlessness at follow-up predicted concurrent suicidal ideation (β = 0.509, 95% CI [0.477; 0.541]) and thoughts of death (β = 0.508, [0.477; 0.540]), whereas baseline guilt predicted the same outcomes with substantially smaller coefficients (β = 0.067; β = 0.066) - a 7- to 8-fold differential. Follow-up suicidal ideation predicted suicide attempt (β = 0.515, p < 0.001). The worthlessness-guilt hierarchy replicated in both sensitivity analyses.

Suicide attempts occurred in a small sub-sample (n = 102), and binary symptom measures limit granularity.

Interventions targeting the schematic self-devaluation captured by worthlessness, rather than guilt-focused strategies, may offer greater clinical leverage in preventing the transition from depressive symptomatology to suicidal cognitions.