Abstract
OBJECTIVES: To estimate the population-level incidence of emergency department (ED) visits for violent behaviour, identify associated factors and quantify the contribution of people presenting to ED for violent behaviour on the total number of ED visits for mental health problems. DESIGN: Retrospective cohort study. SETTING: Regional healthcare authority serving a population of 1.2 million in Romagna, Italy, January 2022 to December 2023. PARTICIPANTS: 871 119 residents (70% of the regional population; 55.6% female) alive on 1 January 2022 with data on comorbidities. Participants were followed until 31 December 2023 with censoring at death. PRIMARY AND SECONDARY OUTCOME MEASURES: Primary outcome: incidence of ED visits for violent/homicidal. Factors associated with violent ED visits were examined. SECONDARY OUTCOME: quantification of the contribution of people presenting to ED for violent behaviour on the total number of ED visits for any mental health problem. RESULTS: 286 individuals (76.9% male) had 573 ED visits for violent behaviour, representing 4.2% of all psychiatric ED visits with an annual incidence rate of 3.48 per 10 000 person-years (95% CI 3.21 to 3.78). Male sex was associated with violent behaviour (OR 3.85, 95% CI 2.60 to 5.70; OR 4.64, 95% CI 3.12 to 6.92 among mental health service users). High comorbidity and prior mental health service use increased the risk. Having an ED visit for violent behaviour was associated with higher odds (OR 10.9, 95% CI 8.02 to 14.96) and higher incidence rate (incidence rate ratio 1.51, 95% CI 1.35 to 1.69) of psychiatric ED visits for other mental health problems. CONCLUSIONS: ED visits for violent behaviour occur at relatively low population rates but predominantly affect high-risk groups: males with prior mental health service contact and either minimal or very high comorbidity. These findings supported the need for tailored interventions that address both the immediate risk of violence and underlying mental health issues, especially for high-risk groups.