Abstract

Importance: Suicide is a leading cause of death worldwide, yet risk prediction remains imprecise. While psychiatric disorders are strongly associated with suicide-related outcomes, most individuals with these conditions never exhibit suicidal behaviors. Polygenic risk scores (PRSs) may help identify additional vulnerability factors beyond clinical diagnoses. Objective: To evaluate the independent and interactive effects of polygenic risk for psychiatric disorders and clinical diagnoses on suicidal ideation (SI) and suicide attempts (SA) in a large, ancestrally diverse cohort. Design: Cross-sectional analysis of genetic and survey data from the All of Us Research Program. Setting: Population-based cohort study leveraging a diverse U.S. sample. Participants: 41,379 adults with genetic data and self-reported psychiatric diagnoses, SI, and SA. Main Outcomes and Measures: Lifetime SI and SA, assessed via self-reported surveys. Predictors included lifetime psychiatric diagnoses on 13 categories and PRSs for depression, bipolar disorder, and PTSD, derived from multi-ancestry genome-wide association studies. Ancestry-stratified multinomial logistic regression analyses were performed for African, Admixed Hispanic/Latino, and European American groups, followed by fixed-effects meta-analysis, adjusting for age, sex at birth, and socioeconomic factors. Results: Among 41,379 participants, 28.5% reported SI, and 12.6% reported SA. All psychiatric disorders were significantly associated with both outcomes, with depression, bipolar disorder, and PTSD showing the strongest independent effects (ORs=2.81-7.73 for SA, 1.62-3.32 for SI, all FDR < 0.05). Each additional psychiatric diagnosis more than doubled the odds of SA (OR=2.16 95% CI: 2.10-2.21). PRSs for depression, bipolar disorder, and PTSD remained significantly associated with SI and SA after adjusting for clinical diagnoses and sociodemographic covariates. For SA, depression PRS showed the strongest association (OR=1.36 [1.30-1.41], p=1.42x10-55), followed by PTSD (OR=1.33 [1.28-1.39], p=6.91x10-45) and bipolar disorder (OR=1.18 [1.13-1.23], p=1.41x10-16). Effect sizes were comparable among individuals with and without clinical diagnoses, suggesting transdiagnostic relevance. Conclusions: Polygenic risk for psychiatric disorders showed modest but significant associations with SI and SA, independent of clinical diagnoses and sociodemographic factors. These findings highlight the value of genetic information in identifying vulnerability not fully captured by diagnostic categories and underscore the importance of multi-dimensional approaches to suicide risk assessment across diverse populations.