Establishment And Verification Of Clinical Prediction Model Of Non-Suicidal Self-Injury Behavior In Adolescents With Depressive Disorder

Objective The mechanism of Non-suicidal self-injury(NSSI)behavior in adolescents with depressive disorder is still unclear,and related risk factors are still in the research stage.This study aims to establish a clinical prediction model of NSSI behavior in adolescents with depressive disorder by analyzing basic clinical data of adolescents with NSSI behavior.Strengthen the initial assessment of the risk of NSSI behavior in hospitalized patients,prevent the occurrence of NSSI behavior during hospitalization,and further reduce the occurrence of adverse events such as suicide.Method A total of 366 inpatients with depressive disorder in the pediatric department of Shenzhen Mental Health Center from January 1,2021 to December 31,2021 were retrospectively collected.According to the DSM-5 diagnostic criteria of the Diagnostic and Statistical Manual of Mental Disorders(Fifth Edition),the patients were divided into a group with NSSI behavior(n=289)and a group without NSSI behavior(n=77).A total of 366 patients were randomly divided into a training set(n=258)and a validation set(n=108)in a ratio of 7:3 to collect clinical data.Logistic regression analysis was used to screen out the predictive variables of adolescent depressive disorder patients with NSSI behavior,and the prediction model was established.The area under receiver Operating characteristic(ROC)curve(AUC)was used to evaluate the differentiation of the training set and validation set models,the calibration curve was used to evaluate the calibration of the training set and validation set models,and the Homser-Lemeshow(HL)goodness of fit test was used to evaluate the fit of the models.Decision curve analysis(DCA)was used to evaluate the clinical benefit of the prediction model.Result The detection rate of NSSI behavior in adolescents with depressive disorder was 77.52%.Gender(β=1.734,OR=5.561,95%CI: 2.678-11.964)and grade(β=0.864,OR=2.737,95%CI: 1.174~4.795),history of attempted suicide(β=0.932,OR=2.539,95%CI: 1.253~5.144),only child(β=0.745,OR=2.106,95%CI: 1.029 ~4.311),depression severity(β=0.056,OR=1.058,95%CI: 1.025~1.092)were five independent predictive variables(all P < 0.05).ROC curve training set AUC was 0.808(95%CI: 0.746~0.870),cutoff value was 0.824,sensitivity was 0.828,specificity was 0.670.Hosmer-Lemesho’s goodness of fit test(HL)showed a good fit(P=0.832).The DCA curve suggests clinical net benefit.Validation set AUC was 0.722(95%CI: 0.581~0.864),cutoff value was 0.732,sensitivity was 0.632,and specificity was 0.820.HL test showed a good fit(P=0.561).Conclusion Gender,grade,depression severity,suicide attempt history,and single child with NSSI were independent predictive variables for adolescents with depressive disorder.A nomogram with clinical characteristics as the basic variables was constructed.The model has good sensitivity and specificity.