Participants and Setting
The PIU operated within an urban academic medical center providing specialized psychiatric care for adolescents experiencing psychiatric crises (e.g., following a suicide attempt). Clinical procedures took place on an 18-bed inpatient unit. The unit afforded each patient their own bedroom. Common programming areas included three classrooms (approximately 20 to 30 m2 in size housing two tables and at least 10 chairs per classroom) and a cafeteria. Unit staff divided the classrooms by the age of the patients (i.e., tweens between 12 and 14 years of age; adolescents between the ages of 15 and 17 years). Patients averaged 15.4 years of age. Please refer to Tables 1 and 2 for additional demographic information about the patient population served on this PIU. The PIU was staffed by university faculty and hospital staff. The university faculty consisted of three psychiatrists and a psychologist. The hospital staff consisted of 45 direct-care staff (holders of bachelor’s or master’s degree in psychology or related field), 10 nurses (holders of bachelor’s degree in nursing), and three licensed behavioral health clinicians (master’s degree in professional counseling or social work).
Dependent Variables
Patient Outcomes. Patients receiving treatment on the PIU during the study period were asked to complete the Anger, Depression, and Anxiety Scales from the Patient-Reported Outcomes Measurement Information System (PROMIS). Patients responded to questions structured according to a 5-point ordinal scale about whether the item was never, rarely, sometimes, often, or always true in the past 7 days28. The anger scale measured angry moods, verbal aggression, and physical aggression29. The depression scale measured symptoms of depression, such as feeling sad28. Finally, the anxiety scale measured fear, anxious misery, hyperarousal, and somatic symptoms related to arousal30. A previous study indicated alpha-coefficient values of 0.90, 0.95, and 0.93 for the anger, depression, and anxiety scales, respectively, which collectively reflects a high degree of assessment reliability31. An additional study by the same author32 indicated a strong convergent validity of the PROMIS anger, depression, and anxiety measures with other established measures of these domains of emotions31-32. These outcome measures were administered upon admission to the PIU and at discharge (either the night before or the morning of patients’ departure from the unit). Patients completed the Patient Satisfaction Inventory at the conclusion of their admission. The authors created this inventory to measure patient satisfaction with medication management, therapy services received as well as whether they felt that the skills learned during their admission would be helpful outside of the hospital. This inventory consisted of a 1 – 7 likert-type scale in which scores of 1 represented low satisfaction and scores of 7 indicated high satisfaction. Patients completed these questionnaires using a tablet computer as overseen by the unit psychologist or a psychology trainee. Assistance was provided if the patient struggled to read or reported being unable to understand the question.
Staff Member Injury. The research team extracted pertinent data from the daily records collected by the Occupational Health Services department of the study site and from the clinic registry. These data included the number of staff and patients working on the unit throughout each day, frequency of restraint or seclusion usage, the number of staff member injuries, and the time each staff member injury occurred. Additional demographic information about the patients who injured the staff were gathered from the electronic medical record at the study site. Considering the complexity of the data extraction process, data entry integrity was ensured by having a second, independent observer auditing the work of the primary data coder. This second observer input data on each of the variables described above for 30% of days across each month of the study period. We defined exact agreement as when the primary data coder reported the same value for the dependent variables described above (e.g., number of restraints used each day) as the secondary data coder. A disagreement was defined as when the primary data coder reported a different value than the secondary data coder for the variables described above. The research team calculated agreement by dividing the number of agreements by the number of agreements plus disagreements and multiplying this quotient by 100 to produce a percentage. Agreement between the two observers was 100%.
Procedures
The PIU delivered interdisciplinary treatment by a child and adolescent psychiatrist, social worker, and psychologist. The psychiatrist evaluated each patient’s medication regimen as well as assisted with diagnostic conceptualization. The social worker provided family therapy and assisted with identifying after-care services for the families. The psychologist organized the DBT treatment each patient received on the Unit. The social worker and psychologist closely collaborated to ensure that the DBT material contacted by the patients could be integrated into the family therapy services.
Prior to starting data collection on patient outcomes, the unit’s psychologist and social workers trained PIU staff to deliver the DBT groups. The training lasted 2 hrs and consisted of didactic presentation and role-play with coaching and feedback. The didactic presentation began by describing the history of DBT along with a description of the groups the unit would be conducting with patients (30 minutes). After this, the psychologist and social workers modeled one of the therapeutic groups (15 minutes). The trainers then provided all staff with a protocol of one of the DBT groups and divided them into groups of four. PIU staff took turns implementing the group according to the protocol, while either the psychologist or social workers provided feedback and support (60 minutes). At the end of the coaching period, the psychologist or social workers encouraged PIU staff to ask additional questions (15 minutes).
DBT can have a lengthy course of implementation, which is not feasible for short-term PIU admissions. Thus, DBT programming at the study site was modified to selectively focus on learning about mindfulness skills, distress tolerance, and emotional regulation. The unit schedule incorporated one DBT group, a group led by an occupational therapist, and snack/meals during the morning. After lunch, patients visited with their families and then participated in a DBT and social skills group. Free time and snacks/meals occurred during the afternoon programing period too. During the evening, patients ate meals, visited with their families, and engaged in a wrap-up group that reviewed programming topics for the day. The Tween and Adolescent groups participated in the same activities but with their own groups. That is, patients assigned to the Tween group did not program with the Adolescent group.
Patients began each group session by participating in a mindfulness activity (e.g., mindful eating, mindful exercise, and mindful coloring) as led by PIU staff. Afterwards, PIU staff conducted the assigned DBT group. The unit conducted DBT groups according to a weekly schedule. On Mondays, groups focused on introducing mindfulness and DBT to patients. Tuesdays and Fridays involved discussion of distress, identification of personally distressing situations, and formulation of coping strategies to manage distress using the six senses or other distraction strategies. On Wednesdays and Sundays, staff facilitated discussions on emotion regulation, including distinguishing negative emotions and antecedents to these emotions, recognizing positive attributes or experiences, and identifying and preparing to use various coping strategies. Finally, on Thursdays and Saturdays PIU staff reviewed dialectics to discern the existence of multiple truths to a situation as a way to reinforce positive behaviors and decrease maladaptive behaviors. Given the abbreviated nature of the PIUs DBT treatment, we collaborated with an outpatient team trained to deliver extended DBT programming to offer a step-down care approach to continue patient therapy following their discharge.
Data Analyses
Descriptive and inferential statistical approaches (e.g., count, percent frequency, mean ± standard deviation, paired t-test, MANOVA, Spearman correlation) were applied to summarize trends and determine significant differences in staffing, cases of restraint and seclusion, incidence of injury, and patient psychological scores, at admission to the PIU and at discharge from the PIU. All statistical analyses were completed using IBM SPSS Statistics for Windows (Version 26.0). Non-parametric approaches were applied when data did not show normal distribution. Significance level was set at p<0.05. All data analyses were performed on de-identified data and key findings were presented based on aggregate data.