Abstract
IMPORTANCE: Although suicide attempt is the most robust estimator of suicide death, few individuals who attempt it go on to die by suicide (<10%), and approximately 50% of suicide deaths occur in the absence of evidence of prior attempts. The risks are particularly poorly understood in this group. OBJECTIVE: To study underlying polygenic liabilities among suicide deaths without evidence of prior nonfatal suicidality (SD-N) compared with suicide deaths with prior suicidality (SD-S), testing prior results showing significantly lower clinical risks of neuropsychiatric traits in SD-N vs SD-S. DESIGN, SETTING, AND PARTICIPANTS: In this cohort study, polygenic scores (PGS) were computed using summary statistics from 12 published source studies, then compared across SD-N and SD-S groups taken from the Utah Suicide Mortality Research Study (cases accrued between December 1998 and October 2022). PGS from the suicide death cohorts were also compared to unselected population controls. Evidence of prior suicidality was determined from diagnoses and clinical notes. MAIN OUTCOMES AND MEASURES: Cohort differences in PGS reflecting neuropsychiatric conditions were tested using analysis of covariance, adjusting for sex, age, and genetic ancestry, followed by additional analyses within sex and within subgroups defined by age at death (50 years or younger vs older than 50 years). PGS spanned 12 neuropsychiatric conditions. Data were analyzed between July 2024 and July 2025. RESULTS: The SD-N cohort (n = 1337) had significantly more male suicide deaths (1105 [82.65%] vs 974 [67.95%]), with an older mean (SD) age at death (47.5 [18.9] vs 41.4 [15.6] years) than the SD-S cohort (n = 1432). The control cohort (n = 19 499) had significantly fewer males (8597 [44.09%]) than both suicide death subsets. Genetic ancestry was similar across the SD-N and SD-S groups (96.77% and 96.81% European ancestry), and control (97.38% European ancestry) groups. Socioeconomic status was not significantly different across suicide cohorts adjusted for age and sex (occupation ranking SD-N mean [SD], 57.16 [24.54]; SD-S mean [SD], 54.72 [25.29]; t = 1.30; P = .70; maximum education SD-N mean [SD], 2.70 [1.12]; SD-S mean [SD], 2.67 [1.13]; Fisher exact test P = .38). Comparing SD-N to SD-S revealed significantly lower (false discovery rate P < .05) PGS in the SD-N group for major depressive disorder (adjusted mean difference, 0.085 [95% CI, 0.018-0.152]; P = .01), depressed affect (adjusted mean difference, 0.081 [95% CI, 0.012-0.149]; P = .02), anxiety (adjusted mean difference, 0.091 [95% CI, 0.021-0.161]; P = .01), neuroticism (adjusted mean difference, 0.102 [95% CI, 0.033-0.171]; P = .004), and Alzheimer disease (adjusted mean difference, 0.090 [95% CI, 0.021-0.1658]; P = .01), and lower (false discovery rate P < .10) PGS in SD-N for posttraumatic stress disorder (adjusted mean difference, 0.070 [95% CI, 0.001-0.139]; P = .04). Of note, SD-N PGS were not significantly different from controls for depressed affect (adjusted mean difference, 0.037 [95% CI, -0.019 to 0.093]), neuroticism (adjusted mean difference, -0.001 [95% CI, -0.057 to 0.055]), or Alzheimer disease (adjusted mean difference, -0.027 [95% CI, -0.083 to 0.029]). CONCLUSIONS AND RELEVANCE: In this cohort study, SD-N showed significantly different genetic liabilities to neuropsychiatric conditions from SD-S. Results have implications for future suicide research and prevention for persons at risk of mortality.