Abstract
BACKGROUND: Suicide-related concerns are a clinical priority in general-hospital psychiatry, yet evidence on suicidal ideation (SI) at first visit and its links with sociodemographic factors, personality traits, and depressive symptom patterns remains limited. METHODS: We analyzed electronic intake assessments from first-visit psychiatric outpatients at Shanghai Tongji Hospital (May 2023 to December 2024). SI was defined as PHQ-9 item 9 > 0. Personality traits were assessed using the Eysenck Personality Questionnaire (EPQ). Multivariable logistic regression estimated adjusted correlates of SI. We examined the latent structure of depressive symptoms using factor-analytic approaches, and then applied structural equation modeling (SEM) to compare one-factor versus two-factor depressive symptom structures and to evaluate their associations with SI while accounting for demographic and personality characteristics. RESULTS: Of 4,389 patients, 2,430 (55.4%) reported SI in the prior 2 weeks. In logistic regression, SI was associated with younger age, female sex, lower education, higher neuroticism and psychoticism, and higher depressed mood, guilty, appetite disturbance, and psychomotor change scores; higher extraversion and fatigue were inversely associated. In SEM, overall depression was positively associated with SI (R² = 0.50). The two-factor model fit well (CFI = 0.977, TLI = 0.987, RMSEA = 0.055, SRMR = 0.045) and explained more variance in SI (R² = 0.93). The cognitive–affective factor was positively associated with SI (b = 3.357, p < 0.001), whereas the somatic–functional factor showed an inverse conditional association (b = − 2.893, p = 0.001). CONCLUSIONS: SI was common at first psychiatric presentation and demonstrated heterogeneity across demographic characteristics, personality traits, and depressive symptom patterns. Differentiating cognitive–affective from somatic–functional symptoms provided additional perspective beyond overall depression severity. SI was more closely aligned with cognitive–affective distress once shared variance was considered, suggesting that attention to symptom configuration in the context of demographic and personality characteristics may be relevant when evaluating SI at intake. TRIAL REGISTRATION: Not applicable. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s12888-026-07960-y.