The prevention of Physical Violent Behavior (VB) toward others during psychiatric hospitalization is a major concern of clinicians. These VBs can have a deleterious impact on the victims, inpatients or caregivers, as well as on the therapeutic milieu. Such violence can also have negative consequences for the assailant patients, such as repeatedly being hospitalized under restraint, stigmatization, and difficulties reintegrating into the community.
This study explored individual (age, gender, marital status, living status, diagnostic) and institutional (type of admission, length of stay, number of previous hospitalizations) risk factors, and how their interactions could increase the risk of VB during psychiatric hospitalizations.
The study was carried out over a period of four years in the psychiatry department of the Lausanne University Hospital, on the 15 wards (219 beds) specialized in acute psychiatric care for adults. All the patients admitted to one of these wards during this period (n=4518), aged between 18 and 65 years, were included in the study. The sample was divided in two groups: non-violent patients (NVPs) and violent patients (VPs). VBs, defined as physical aggressions against another person, were assessed by the Staff Observation Aggression Scale - Revised (SOAS - R). Only physical assaults, associated or not with other types of violence, involving hospitalized patients were analyzed. Personal and institutional factors were extracted from the hospital database. Chi ² independence tests were used to assess differences between groups. Logistic regression models were used to identify the links between each factor and the VB. Classification and regression trees were used to study the hierarchical effect of factors, and combinations of factors, on VBs.
During the study period, 414 VBs were reported involving 199 patients (4.40 % of all patients). VPs were significantly younger, male, more likely to be unmarried and living in sheltered housing before hospitalization. In this group, the proportion of patients with diagnoses of schizophrenia, and/or schizophrenia with comorbid substance abuse and cognitive impairment, were higher compared to NVPs. VPs were more frequently admitted involuntarily, had a longer average length of stay and a greater number of previous hospitalizations. The logistic regression model performed on individual factors have shown a significant link between age (OR=0.99; CI: 0.97-1.00; P-value=0.024), living in sheltered housing before admission (OR=2.46; CI: 1.61-3.75; P-value<0.000), schizophrenic disorders (OR=2.18; CI: 1.35-3.57; P-value=0.001), schizophrenic disorders with substance abuse comorbidity (OR=2.00; CI: 1.16-3.37; P-value=0.016), cognitive impairment (OR=3.41; CI: 1,21-8.25; P-value=0.010), and VBs. The logistic regression model on institutional factors have shown a significant link between involuntary hospitalization (OR=4.38; CI: 3.20-6.08; P-value<0.000), length of previous stay (OR=1.01; CI: 1.00-1.01; P-value<0.000), number of previous hospitalizations (OR=1.06; CI: 1.00-1.12; P-value=0.031), and VBs. The logistic regression model on individual and institutional factors have shown a significant link between age (OR=0.99; CI: 0.97-1.00; P-value=0.008), living in sheltered housing before admission (OR=2.46: CI: 1.61-3.75; P-value=0.034), cognitive impairment (OR=3.41; CI: 1.21-8.25; P-value=0.074), involuntary hospitalization (OR=3.46; CI: 2.48-4.87; P-value<0.000), length of previous stay (OR=1.01; CI: 1.00-1.01; P-value<0.000), and VBs. The classification and regression trees have shown that the relationship between long length of stay and repeated hospitalizations mainly potentiate the risk of violence.
The results of this study have shown the existence of a small group of vulnerable patients who accumulate constrained hospital stays during which violence occurs. Exploring the clinical profiles and institutional pathways of patients could help to better identify these patients and promote a more appropriate mode of support, such as intensive clinical case management. This model could facilitate the development of a clinical network and the links between the structures and partners caring for a patient. This would create a continuous support, avoiding or limiting the lack of continuity of care and care disruption.