WED-119 - Distal, Developmental, and Proximal Risk Factors for Suicide Attempts in Youth
Wednesday, April 16, 2025
6:00 PM – 7:00 PM PST
Location: Pacific I/II, 2nd Floor
Subcompetencies: 1.3 Analyze the data to determine the health of the priority population(s) and the factors that influence health., 1.3.3 Identify the social, cultural, economic, political, and environmental factors that impact the health and/or learning processes of the priority population(s). 1.4.4 Develop recommendations based on findings. Research or Practice: Research
Graduate Student Indiana University Bloomington Bloomington, Indiana, United States
Learning Objectives:
At the end of this session, participants will be able to:
Understand and critically evaluate the application of machine learning techniques in identifying and predicting risk factors for suicide, including the strengths, limitations, and ethical considerations involved in utilizing these methods.
Examine the various risk and protective factors associated with suicide attempts in youth, with a focus on psychological, social, and environmental influences.
Understand how the complex interplay of distal, developmental, and proximal factors contribute to the prediction of suicidal behavior in youth, and explore the multifaceted risk and protective mechanisms involved.
Suicide is the second leading cause of death among youth aged 10-24 in the United States (Curtain & Garnett, 2023). Suicide-related outcomes result from the accumulation of interacting distal, developmental, and proximal risk factors throughout development (Carter et al., 2017). Distal risk factors are biological or environmental determinants shaping suicide vulnerability. Developmental risk factors are phenotypic diathesis expressions. Proximal risk factors are recent clinical conditions or negative life events that may precipitate suicide-related outcomes. Despite research linking these risk factors to suicide attempts, more research is needed to differentiate unique markers of risk. This study’s goal was to determine the most predictive subset of distal, developmental, and proximal risk factors for suicide attempts. Data was drawn from the Emergency Department Screening for Teens at Risk for Suicide (ED-STARS) Study, a large-scale investigation of suicide attempts during the 3- and 6-months following adolescents’ Emergency Department (ED) visits (King et al., 2019). Adolescents (n=6,448) were recruited from thirteen EDs between 2015-2016. Youth completed baseline assessments of suicidal ideation and behavior, depression, and connectedness, among other risk factors. A subset of youth (N=2,104; 63.1% female; Mage=15.1) were randomized to follow-up and completed additional assessments at 3- and 6-months. Machine learning was used; specifically, Least Absolute Shrinkage and Selection Operator (LASSO; Tibshirani, 1996) regression was utilized because it shrinks large regression coefficients to prevent overfitting while simultaneously producing robust and parsimonious models (Foubister et al., 2021). The LASSO regression identified which a priori selected distal, developmental, and proximal risk factors predicted future suicide attempts. In a complete case analysis (N=1,870) of 22 predictors, nine were retained via the LASSO regression: past suicide attempts, depression, number of non-suicidal self-injury (NSSI) methods used in the past year, recent suicidal ideation, negative life events, NSSI in the past year (yes/no), mood and affective states, school connectedness, and impulsivity. Previous suicide attempt was the strongest predictor (β=.67). Depression (β=.15), past year NSSI (β=.05), number of NSSI methods (β=.14) used in the past year, and impulsivity (β= -.14) also predicted higher risk for a future suicide attempt. Higher school connectedness (β= -.11) and mood (β= -.11) predicted lower risk for a future suicide attempt. As predicted, the strongest future suicide attempt predictor was a previous attempt. Developmental risk factors appear the most important subsequent risk predictor; other predictors such as number of NSSI methods and school connectedness should be explored in future research
