Institutes for long-term psychiatric inpatient care
The study was conducted in closed psychiatric long-stay wards belonging to three regional mental healthcare centres in urbanized areas of the Netherlands: Rivierduinen Psychiatric Center in Oegstgeest, Parnassia Psychiatric Institute in The Hague, and Inforsa in Amsterdam. All three institutes provide long-term inpatient care. In all three centres, patients had been diagnosed predominantly with psychosis spectrum disorders or severe personality disorders, with approximately 80% of patients meeting diagnostic criteria for either type of disorder. Patients were mainly admitted involuntarily. The Medical Ethical Committee of Leiden University Medical Center granted permission for this study.
Definition of aggression incidents
We categorised aggression incidents into four categories, based on the Overt Aggression Scale (OAS) (18). First, we defined ‘verbal aggression’ as yelling, shouting, using obscenities or swearwords, sexual remarks, and threatening others (with or without a threatening posture). Second, we defined ‘physical aggression towards an object’ as kicking, hitting, throwing objects (e.g. chairs, dishes, or cups), and slamming doors. Third, we defined ‘self-harm’ as any act of physical aggression towards the self, such as hitting, cutting, burning, strangulation, overdosing on medication, and jumping from heights (with or without suicidal intent). Fourth, we defined ‘physical aggression towards others’ as a physical assault on another person by means of hitting, pushing, pulling, holding, scratching, kicking, biting, spitting, touching inappropriately, strangulating, and/or attacking someone with an object (e.g. a chair or a knife). Case vignettes illustrating each of the categories are presented in the Supplementary Material.
To estimate the incidence of aggression across the four categories, we applied a random sampling procedure, which included 21 nursing shifts over a 6-month period (February–July 2014) at all three facilities. All weekdays (Mon–Sun) and shift types (day, evening, and night) were covered. Psychiatric nurses were interviewed via telephone at the end of each of their shifts; we asked them to recall all incidents during the preceding 8 hours. If one patient caused several incidents, or if an incident escalated to a graver category, only the most severe incident was recorded for that shift. Using a random number generator, participating staff members were randomly selected from the total available ward staff per shift. The number of patients observed per shift was used to calculate the total number of observed patient years. This total, along with the observed incidents, was then used to calculate incidence rates per patient year. The ratio of the occurrence of the four types of aggression was used as a weight factor to calculate the time spent and the total cost of all aggression incidents.
Time spent on aggression
The time spent on each type of incident by staff members was monitored in real-time by a researcher who was present during day shifts and evening shifts over a period of one week at each of the three facilities. The researcher followed all incidents from start to finish in order to record all the activities by all staff members who handled the incident. These activities were collapsed into two categories: direct time (i.e. de-escalating conversation, restraining patients, checking up on isolated patients, administering medication, and tending to medical needs) and indirect time (i.e. administrative activities, discussing and evaluating the incident, information transfer, consultation, and transport).
Costs of aggression incidence
In order to estimate costs for each type of aggression incident from an institutional perspective, staff wages and material costs of both damaged property and immediate medical care were taken into account. We defined ‘nursing staff’ in a psychiatric ward as licensed nurses and psychiatric social workers whose responsibilities include the day-to-day care for patients. We labelled other assistive personnel as ‘paraprofessionals’, which included patient supporters, activity supervisors, and specially trained security personnel who support nursing staff during aggression incidents. Average yearly wages were derived from the Collective Labour Agreement for the Mental Health Sector (2017–2019) (19). Consistent with the guidelines of the Dutch Healthcare Authority, these wages were adjusted incrementally with the mandatory insurances provided by the employer. The same Collective Labour Agreement also provides the yearly number of available work hours per staff member, taking into account average days of sick leave and annual leave. Yearly salary costs were divided by the total available work hours, resulting in hourly wages for each staff category. To account for the incidents occurring outside the standard 40-hour working week, hourly wages were increased with a percentage surcharge for irregular shifts in proportion to the number of incidents taking place during evening and night shifts and during the weekend. Finally, staff costs were calculated by multiplying appropriate hourly wages with the average amount of time spent by each type of professional per incident type.
Material-related costs consisted of property damages caused by the patient (e.g. broken furniture or windows) and immediate medical expenses at the ward (e.g. costs of administering medication or necessary medical treatment of injured patients or members of staff as a direct consequence of the incident). The costs of non-immediate medical treatment or sick leave taken by staff members as a result of the incident could not be included, as it was impossible to link these to individual aggression incidents.
Adding the average personnel and material costs for each type of incident allowed us to calculate the average costs per incident for each of the four incident categories. Using the previously estimated number of incidents per patient year, we inferred the annual costs (i) per patient and (ii) per ward (assuming 20 patients per ward). The psychiatric clinical capacity for long stay (defined as a clinical stay of more than one year) in the Netherlands in 2017 was 6250 beds, of which 1438 (or about 23%) are in a closed setting (20). Although patients can be admitted or discharged during the year, the total number of admitted patients at any time stays close to 1438 — a rather high occupancy rate for long-stay clinical psychiatry.