抑郁症自杀行为简易预测模型构建及其预测价值研究

背景　抑郁症作为最常见的心境障碍之一，具有患病率高、复发率高、致残率高和致死率高等特点，给患者造成巨大的疾病负担，甚至出现自杀行为。但是，目前快速筛查抑郁症患者自杀行为的手段相对有限。目的　调查影响抑郁症患者出现自杀行为的心理社会因素，建立抑郁症患者自杀行为简易预测模型，为抑郁症患者自杀防治工作提供参考依据。方法　采用整群抽样方法选择2018年1-12月在南昌大学第二附属医院和南昌大学第一附属医院就诊的抑郁症患者为调查对象，采用一般情况问卷、抑郁自评量表（SDS）、焦虑自评量表（SAS）和Landeiman社会支持量表进行调查，采用多因素Logistic回归分析探讨抑郁症患者出现自杀行为的影响因素，Risk score法构建抑郁症患者自杀行为简易预测模型，并检测其预测效果。结果　共发放问卷2 233份，回收有效问卷2 090份，问卷有效回收率为93.60%。2 090例抑郁症患者中，142例（6.79%）出现自杀行为。经常吸烟、重度饮酒、既往抑郁发作次数≥1次、既往因抑郁症住院次数≥1次、伴焦虑症状、伴精神病性症状、伴自杀意念、有精神障碍家族史、正在用抗抑郁药物、有其他内外科疾病的抑郁症患者自杀行为比例高（P<0.05）。多因素Logistic回归分析结果显示，既往抑郁发作次数≥1次〔OR=4.308，95%CI（3.547，5.232）〕、伴焦虑症状〔OR=2.329，95%CI（1.201，4.518）〕、伴精神病性症状〔OR=2.492，95%CI（1.448，4.287）〕、伴自杀意念〔OR=4.044，95%CI（2.305，7.096）〕、SAS标准分高〔OR=1.036，95%CI（1.003，1.071）〕均是抑郁症患者自杀行为的危险因素（P<0.05），正在用抗抑郁药物〔OR=0.110，95%CI（0.057，0.212）〕是抑郁症患者自杀行为的保护因素（P<0.05）。基于Logistic回归建立的Risk score预测模型为：Risk score=40.56×既往抑郁发作次数+23.50×伴焦虑症状+25.36×伴精神病性症+38.81×伴自杀意念-61.25×正在用抗抑郁药物+1.00×SAS标准分。按照Risk score预测模型绘制的受试者工作特征曲线（ROC）下面积（AUC）为0.920〔95%CI（0.907，0.931）〕，Youden指数最大时为0.7，截断值为193.23分，灵敏度为76.8%，特异度为94.2%。结论　抑郁症患者自杀行为发生率较高，既往抑郁发作次数≥1次、伴焦虑症状、伴精神病性症状、伴自杀意念、SAS标准分高均为抑郁症患者自杀行为的危险因素。基于Logistic回归建立的Risk score预测模型预测抑郁症患者自杀行为的灵敏度为76.8%，特异度为94.2%。

Development and Value Evaluation of a Simple Prediction Model of Suicidal Behavior in Depressive Disorder

Background　The depression disorder is one of the most common mood disorders，with characteristics of high morbidity，recurrence rate，disability rate and mortality，imposing a huge burden of disease on the patients，even leading to suicide.However，means for rapidly screening of suicidal behavior in patients with depressive disorder are relatively limited.Objective　To develop a simple prediction model for suicidal behavior in patients with depressive disorder using identified associated psychosocial factors of suicidal behavior，offering a reference for the prevention and treatment of suicide behavior in such patients.Methods　By use of cluster sampling，depressive disorder patients treated in the Second Affiliated Hospital of Nanchang University and the First Affiliated Hospital of Nanchang University of Nanchang from January to December 2018 were recruited.The General Information Questionnaire developed by our research group，Zung Self-rating Depression Scale（SDS），Zung Self-Rating Anxiety Scale（SAS）and Landeiman Social Support Scale were used to gather sociodemographic data，prevalence of depression and anxiety and social support information，respectively.Multivariate logistic regression was used to screen the risk factors for suicidal behaviors.Risk scoring measuring the effect of factors on suicidal behaviors was used to develop the prediction model.Results　The survey obtained a response rate of 93.60%（2 090/2 233）.The prevalence of suicide attempt was 6.79%（142/2 090）.Frequent smoking，heavy drinking，having one or more depressive episodes，having one or more hospitalization due to depressive disorder，presence of coexisting anxiety symptoms，psychotic symptoms，and suicidal ideas，family history of mental disorders，antidepressant use，and prevalence of other internal and surgical diseases were associated with a high risk of suicide attempts（P<0.05）.Multivariate Logistic regression analysis showed that the number of previous depressive episodes≥1〔OR=4.308，95%CI（3.547，5.232）〕，coexisting anxiety symptoms〔OR=2.329，95%CI（1.201，4.518）〕，coexisting psychotic symptoms〔OR=2.492，95%CI（1.448，4.287）〕，coexisting suicidal ideas 〔OR=4.044，95%CI（2.305，7.096）〕，and high SAS standard score〔OR=1.036，95%CI（1.003，1.071）〕 were risk factors for suicide attempts（P<0.05），while antidepressant use〔OR=0.110，95%CI（0.057，0.212）〕was associated with decreased risk of suicidal attempts（P<0.05）.The risk score model established based on Logistic regression results is：risk score=40.56×previous depressive episodes+23.50×coexisting anxiety+25.36×coexisting psychotic symptoms+38.81×coexisting suicidal idea-61.25×antidepressant+1.00×SAS standard score.The AUC of the model in predicting suicide was 0.920〔95%CI（0.907，0.931）〕，and when the cut-off value was determined as 193.23，the Youden's index was 0.7，reaching the maximum value，with a sensitivity of 76.8%，and a specificity of 94.2%.Conclusion　The prevalence of suicide attempts in patients with depressive disorder is high，the number of previous depressive episodes≥1，coexisting anxiety symptoms，coexisting psychotic symptoms，coexisting suicidal ideas，and high SAS standard score were risk factors for suicide attempts. The risk score prediction model based on Logistic regression was 76.8% in sensitivity and 94.2% in specificity.