7 - (IOP 23) Intersectionality and Youth Suicide: A Multilevel Analysis of Individual Heterogeneity and Discriminatory Accuracy and Lasso-regularized Logistic Regression

Suicide is a leading cause of death among U.S. youth, with disproportionately high rates among sexual and ethnoracial minority adolescents. Traditional approaches to studying suicide risk often fail to account for the complex interplay of intersecting social identities, limiting the ability to identify high-risk groups and key predictors of suicidal thoughts and behaviors (STBs). To address this gap, the present study applies the Multilevel Analysis of Individual Heterogeneity and Discriminatory Accuracy (MAIHDA) to examine intersectional risk for STBs among U.S. adolescents using data from the Youth Risk Behavior Surveillance System. We analyzed a nationally representative sample of 113,819 high school students in grades 9-12 across the 50 states and the District of Columbia. We modeled intersectional social identities—including sex, ethnoracial identity, grade level, and sexual orientation—as predictors of STBs, which included suicidal ideation, suicide plan, suicide attempt, and serious suicide attempt (requiring medical attention). Findings revealed a complex social patterning of STB prevalence, highlighting disproportionate suicide risk among multiply marginalized youth. The highest suicidal ideation prevalence was observed among bisexual White females in 10th grade (47.58%; 95% highest posterior density [HPD], 44.89% - 50.40%), while suicide plans were most common among bisexual Hispanic/Latino females in 9th grade (39.3%; 95% HPD, 35.66% - 43.08%). Similarly, suicide attempts were most prevalent among bisexual Black/African females in 9th grade (30.02%; 95% HPD, 26.39% - 34.12%), and serious suicide attempts were highest in lesbian/gay Black/African females in 9th grade (9.42%; 95% HPD, 5.73% - 13.59%). To identify key predictors of STBs, we applied a LASSO-regularized generalized linear model, which selects the most relevant risk factors by penalizing less informative variables. Across all models, mental health symptoms (e.g., feeling sad and hopeless) and prior suicidal ideation consistently emerged as key predictors of STBs. Sexual abuse was a top predictor of suicidal ideation, while frequent smoking was among the strongest predictors of suicide attempts and serious attempts. By integrating MAIHDA and LASSO-regularized modeling, this study advances intersectional suicide research by both identifying high-risk subgroups and prioritizing key risk factors for targeted interventions. Findings highlight the need for structurally informed suicide prevention efforts tailored to the distinct vulnerabilities of multiply marginalized youth, ensuring targeted and equitable intervention strategies.