Published online Feb 19, 2026. doi: 10.5498/wjp.v16.i2.112575
Revised: September 20, 2025
Accepted: November 10, 2025
Published online: February 19, 2026
Processing time: 183 Days and 17.8 Hours
To understand the current situation of violent behavior among hospitalized patients with severe mental disorders (SMDs), analyze its influencing factors, establish a predictive model and draw a nomogram, providing screening tools for medical staff to accurately identify SMDs who have violent behavior and the direction of early intervention.
To investigate the determinants of violent actions in hospitalized patients with SMDs.
This research included 440 inpatients with SMDs who were admitted to the Wutaishan Hospital from January 2025 to June 2025. Data collection and analysis aimed to pinpoint independent contributors linked to aggression in this patient group. An advanced logistic regression analysis with multiple variables was performed using R, followed by the creation of a line chart to display the forecast outcomes of the model.
Of 120 patients exhibited violent behavior (incidence rate = 27.30%). Education level, cigarette smoking, length of hospitalization, age, psychotic symptoms based on the Brief Psychiatric Rating Scale, and C-reactive protein were independent risk factors for violent behavior. Education level and age served as protective elements among the factors analyzed. The receiver operating characteristic curve area for the training and test sets was calculated to be 0.94 and 0.93, respectively. The calibration graph demonstrated that the model was accurately adjusted. The clinical decision curve demonstrated that the model provided significant practical benefits.
The predictive mode provided a valuable theoretical basis for ward staff to identify inpatients with SMDs at elevated risk of aggression in the early phase.
Core Tip: This study constructed a predictive model by identifying the independent influencing factors of violent behavior in hospitalized patients with severe mental disorders, internally validated the model, and visually presented the model through a nomogram. The result of research shows that: Education level, cigarette smoking, length of hospitalization, age, Brief Psychiatric Rating Scale, and C-reactive protein were independent risk factors. The predictive model provided a valuable theoretical basis for ward staff to identify inpatients with severe mental disorders at elevated risk of aggression in the early phase.