青少年抑郁障碍患者非自杀性自伤行为风险列线图预测模型的建立

Establishment of a nomogram prediction model for the risk of non-suicidal self-injury behavior in adolescents with depressive disorders

中文关键词:  青少年  抑郁障碍  非自杀性自伤  危险因素  列线图模型

英文关键词:Adolescents  Depressive disorder  NSSI  Risk factors  Nomogram model

中文摘要:

      【摘要】 背景 非自杀性自伤（NSSI）行为是亟需重视的公共卫生议题，抑郁青少年高发。既往研究多聚焦于青少年NSSI行为的相关因素分析，而面向临床快速评估、兼具直观量化特征的可视化模型构建相关研究较为有限。目的 调查青少年抑郁障碍患者NSSI行为的危险因素，并建立可视化的列线图风险预测模型，便于临床接诊时快速评估个体NSSI行为风险概率，高效筛查高危就诊青少年。方法 采用横断面研究，选取2025年2月—12月就诊于兰州大学第二医院心理卫生科的、符合《精神障碍诊断与统计手册（第5版）》（DSM-5）抑郁障碍诊断标准的青少年患者448例，按7∶3比例随机分为建模组（n=313例）和验证组（n=135）。采用抑郁自评量表（SDS）、焦虑自评量表（SAS）、Barratt冲动性量表第11版（BIS-11）、童年创伤问卷简版（CTQ-SF）及匹兹堡睡眠质量指数量表（PSQI）进行评定。采用二元Logistic回归分析筛选建模组NSSI行为的独立预测因子，进而构建列线图模型，采用受试者工作特征曲线、校准曲线和决策曲线依次评估该模型的区分度、准确度和临床应用价值。结果 二元Logistic回归分析显示，年龄越小、女性、SDS评分更高、BIS-11评分更高、CTQ-SF评分更高、PSQI评分更高是青少年抑郁障碍患者发生NSSI行为的危险因素（OR=0.816、2.671、1.131、1.030、1.064、1.102，P<0.05或0.01）。基于上述6个指标构建的列线图预测模型，Hosmer-Lemeshow拟合优度检验显示拟合度较好（χ2=15.144，P=0.056）。内部验证及外部验证结果显示，建模组AUC为0.895，验证组AUC为0.891；校准曲线显示，实际值与预测值间的平均绝对误差分别为0.043和0.023；决策曲线显示，当预测风险阈值大于0.11时，模型表现出显著临床净收益。结论 基于年龄、性别、抑郁症状严重程度、冲动性、童年创伤、睡眠质量这6个指标构建的青少年抑郁障碍患者NSSI行为风险列线图预测模型具有较高的区分度、准确度和临床应用价值。

英文摘要:

      【Abstract】 Background Nonsuicidal selfinjury (NSSI) constitutes a critical publichealth concern requiring urgent attention, given its high prevalence among adolescents suffering from depression. Prior research has largely centered on identifying correlates of NSSI behaviors in adolescent populations, whereas studies developing intuitive, quantifiable visual models to facilitate rapid clinical assessment remain comparatively limited. Objective To explore the risk factors for non-suicidal self-injury (NSSI) among adolescents with depressive disorders and construct a visual nomogram prediction model that can rapidly evaluate individual NSSI risk probability in clinical settings and enable efficient screening of high-risk adolescent patients. Methods This investigation constitutes a cross-sectional study. A total of 448 adolescent patients diagnosed with depressive disorders according to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) were enrolled from the Mental Health Department of Lanzhou University Second Hospital between February and December 2025. All participants were randomly divided into a modeling group (313 cases) and a validation group (135 cases) at a ratio of 7:3. Assessments were performed using the SelfRating Depression Scale (SDS), SelfRating Anxiety Scale (SAS), Barratt Impulsiveness Scale11 (BIS11), Childhood Trauma QuestionnaireShort Form (CTQSF), and Pittsburgh Sleep Quality Index (PSQI). Binary logistic regression analysis was applied to screen independent predictors of NSSI behaviors in the modeling group, based on which a nomogram model was constructed. The receiver operating characteristic curve, calibration curve and decision curve analysis were used to evaluate the discrimination, calibration and clinical utility of the model, respectively.. Results Binary logistic regression analysis revealed that younger age, female sex, higher SDS score, higher BIS11 score, higher CTQSF score, and higher PSQI score were risk factors for NSSI behaviors in adolescent patients with depressive disorder (OR?=?0.816, 2.671, 1.131, 1.030, 1.064, 1.102; P?<?0.05 or 0.01). The Homser-Lemeshow(HL) test of the nomogram prediction model constructed based on these six indicators showed a good fit(χ2=15.144, P=0.056). Internal and external validations indicated that the AUC of the modeling group was 0.895 and that of the validation group was 0.891. The calibration curves suggested that the mean absolute errors between the actual values and the predicted values were 0.043 and 0.023, respectively. The decision curve demonstrated that when the predicted risk threshold was>0.11, the model exhibited significant clinical net benefit. Conclusions The nomogram prediction model for NSSI risk in adolescent with depressive disorder, constructed based on six indicators including age, gender, severity of depression, impulsivity, childhood trauma, and sleep quality, has high discrimination, accuracy, and clinical application value.