3 - (SYM 7) Idiographic Network Models Reveal Highly Individualized Proximal Risk Processes for Suicidal Ideation in Adults After Hospitalization for a Suicide Attempt

Despite a >50-year concerted research effort to enhance suicide prevention, suicide rates have continued to rise in the United States and researchers’ ability to predict suicide remains poor. While many common risk factors have been identified, robust research indicates the contextual processes that drive suicide risk are highly individualized and dynamic over time. State-of-the-science idiographic (i.e., N = 1) analytic approaches enable modeling person-specific risk processes in time-series data, which may identify when and under what conditions different people are at highest risk for suicide. Although there has been recent enthusiasm for idiographic methods in suicide research, most prior idiographic studies have utilized predictive models that examined only one or a few risk factors at a time. Understanding how dynamic interrelations between multiple risk factors influence an individual’s suicidal ideation (SI) could provide more nuanced context for how risk unfolds in daily life. The present study demonstrates the use of idiographic network models to examine person-specific pathways of risk for SI in a small sample of adults following hospitalization for suicidality, a period and patient population with extraordinarily high suicide risk. Psychiatric inpatients (N = 23) completed an ecological momentary assessment (EMA) protocol for 65 days following hospital discharge (Mean # EMA responses per participant = 89, M_age = 35.87 years, 70% White, 74% Female sex, 77% heterosexual). Graphical vector autoregressive models estimated idiographic lagged and contemporaneous pairwise correlations between momentary SI, positive and negative affect, rumination, distress intolerance, emotion dysregulation, and negative life events. Results identified unique network structures and centrality patterns for all participants, suggesting substantial heterogeneity in participants’ risk processes. For example, one participant evidenced only one direct effect on SI, with positive affect serving as a protective factor, while another participant’s SI appeared to be driven by lagged associations between negative events, negative affect, and rumination. Results demonstrate the feasibility of using idiographic networks to identify complex person-specific patterns of risk for SI, and support the need for precision approaches to detect and intervene on suicide risk. Practical considerations for implementing idiographic network methods in patients with high suicide risk will be discussed.