Establishment of a predictive model for violent behavior among outpatient patients in psychiatric hospitals

Background To investigate the factors influencing the violent behavior of outpatients with mental illness and establish a nomogram prediction model with Rstudio statistical software. Methods Patients who visited the Outpatient Department of Wutaishan Hospital in Yangzhou, Jiangsu Province from February to July 2024 were selected as investigation objects. By collecting data, the independent influencing factors of patients’ violent behavior were screened. A multi-factor Logistic regression model was established with R software and a column graph was plotted. The model was evaluated by receiver operating characteristic (ROC), calibration and clinical decision curves (DCA), and validated by the validation group. Results Among 430 outpatient patients with mental illness, 114 had violent behavior, with an incidence rate of 26.5%. Statistical significance was shown in many aspects of outpatients with mental illness (all P  < 0.05): diagnosis, education, occupation, marriage, residence, family history, payment method, relationship with patients, psychological capital, adverse childhood experience, time of visit, smoking, drinking, age, number of visits, and psychotic symptoms (the Brief Psychiatric Rating Scale, BPRS). Educational background (odds ratio (OR) = 0.35), marriage (OR = 0.20), BPRS (OR = 1.10), visit time (OR = 4.83) and smoking (OR = 3.37) were independent risk factors for the violence of outpatients with mental illness. Among them, educational background and marital status were protective factors. Modeling and verification groups had an area under ROC curve (AUC) of 0.93 and 0.96, respectively. The calibration curve showed a good calibration degree of the model. The clinical decision curve demonstrated the good clinical effectiveness of the model. Conclusion The established risk prediction model for the violent behavior of outpatients with mental illness has good predictive efficacy, which is beneficial for clinical mental health personnel to early detect high-risk groups of violent behavior.