Survey on Non-Suicidal Self-Injury Behavior in Adolescents with Bipolar Disorder and Construction of a Risk Early Warning Model

doi:10.12104/j.issn.1674-4748.2025.24.007

Abstract：Objective: To investigate the current situation of non-suicidal self-injury behavior in adolescents with bipolar disorder, analyze its risk factors, construct a risk early warning model, and validate it. Methods: A total of 140 adolescents with bipolar disorder treated at the Third People's Hospital of Shangrao City from January 2022 to December 2023 were selected as the study subjects. According to whether they experienced non-suicidal self-injury behavior, they were divided into the occurrence group and the non-occurrence group. Relevant data were collected for risk factor analysis and model construction. Another 60 adolescents with bipolar disorder treated from January to December 2024 were selected as the validation group, and their data were collected to verify the risk early warning model. Results: There were 61 cases (43.57%) in the occurrence group and 79 cases (56.43%) in the non-occurrence group. Multivariate analysis showed that suicidal ideation, the Bech-Rafaelsen Mania Scale (BRMS) score, the Hamilton Depression Rating Scale (HAMD) score, and the National Suicide Screening Tool for Adults (NGASR) score were independent risk factors for non-suicidal self-injury behavior in adolescents with bipolar disorder (P < 0.05). Based on these factors, a risk early warning model was constructed: Logit(P) = 1.748 × suicidal ideation + 0.164 × BRMS + 0.237 × HAMD + 0.393 × NGASR - 16.380. The area under the receiver operating characteristic (ROC) curve was 0.923 [95% CI (0.883, 0.964)], with a sensitivity of 95.1%, specificity of 72.2%, and Youden index of 0.673. The Hosmer-Lemeshow test showed χ² = 14.468, P > 0.05. Conclusion: The constructed risk early warning model for non-suicidal self-injury behavior in adolescents with bipolar disorder can be used to screen high-risk patients early, enabling timely targeted interventions, and has significant application value.