Three latent subtypes of suicide risk and network characteristics of scale item associations among adolescents with major depressive disorder: a cross-sectional study

This study aimed to identify latent suicide risk subtypes among first-episode, treatment-naive adolescents with major depressive disorder (MDD), and to explore the item-level association patterns across subtypes, so as to provide evidence for optimizing stratified suicide risk screening and supporting future longitudinal intervention research. Latent profile analysis (LPA) and regularized partial correlation network analysis were adopted to analyze data from 930 first-episode, treatment-naive adolescents with MDD aged 12–18 years. Suicide risk subtypes were identified based on the Suicide Behaviors Questionnaire-Revised (SBQ-R). Scale item association networks were constructed using items from the Patient Health Questionnaire-9 (PHQ-9), SBQ-R, and dimensions of Adolescent Self-Rating Life Events Checklist (ASLEC). Differences in high-centrality items and network structural characteristics were compared across subtypes. Three suicide risk subtypes were identified among adolescents with MDD: a low suicide risk group ( n = 248, 26.7%), a low-disclosure, high suicide vulnerability group ( n = 329, 35.4%), and an overt, high suicide risk group ( n = 353, 38.0%). Significant differences were observed between the groups in terms of total SBQ-R and ASLEC scores. PHQ-9 total scores were significantly lower in the low suicide risk group than in the other two groups, which did not differ from each other. The low-disclosure, high-suicide-vulnerability group exhibited a depressive symptom burden and partial suicide risk indicators comparable to those of the overt, high-suicide-risk group, yet demonstrated lower active disclosure of suicide-related information. Among the low suicide risk group, items with the highest strength centrality were PHQ-9 Item 1 (loss of interest), ASLEC dimension I (punishment) and PHQ-9 Item 2 (depressed mood). In the low-disclosure, high-suicide-vulnerability group, PHQ-9 Item 2, ASLEC dimension III (interpersonal relationships) and PHQ-9 Item 1 showed elevated node strength. In the overt high suicide risk group, ASLEC dimensions I and III, and PHQ-9 Item 4 (fatigue/loss of energy) had the strongest node connections. Network comparison test detected no significant differences in global strength for all pairwise subtype contrasts; significant structural divergence emerged only between the low suicide risk group and the overt high suicide risk group. SBQ-R Item 4 (future likelihood of suicide) was the only differentially weighted node. Given that this item contributed to latent profile derivation, this node-level difference is subject to partial mechanical circularity and must be interpreted cautiously. Marked heterogeneity exists in suicide risk among adolescents with MDD. The low-disclosure high suicide vulnerability subtype presents depressive burden and partial suicide risk indicators similar to the overt high-risk group, accompanied by insufficient disclosure of suicide-related information. Across subtypes, network structures showed considerable overlap with only limited item-level correlational discrepancies. Centrality metrics merely reflect correlational patterns among scale items, rather than causal relationships or definitive intervention targets. These findings inform stratified risk screening and warrant validation via longitudinal research. This study was prospectively registered in the Chinese Clinical Trial Registry (Registration No.: ChiCTR2400088411; Registration Date: August 19, 2024). Registration website: https://www.chictr.org.cn/showproj.html?proj=235322 .