Use of Modified Positive Behavioral Interventions and Supports in a Psychiatric Inpatient Unit for High-Risk Youths

This naturalistic, prospective study was based on data collected in a youth psychiatric inpatient service. The inpatient service has a 12-bed capacity. Data were collected from January 2010 through June 2014. Of note, similar to other efforts to reduce seclusion and restraint (12), the year of program implementation was not included in the analyses because of the fluctuating nature of fidelity to the program during the initial year. During the four-year interval of the study, the unit had an average of approximately 420 admissions per year, a bed occupancy of approximately 82%, and an average length of stay of 8.57±1.48 days. Approval was obtained from the institutional review board.

Demographic and clinical variables were abstracted from electronic medical records and included age, sex, race-ethnicity, insurance status (Medicaid versus other forms), standing neuroleptic medication, primary discharge diagnosis (collapsed into diagnostic categories), length of stay, and percentage occupancy. Outcomes of interest included use of seclusion and restraint (percentage of patients placed in seclusion or restraint and its mean duration and rate, expressed per 1,000 patient-hours) and use of PRN medications (percentage of patients who received PRN medications during their admission). Seclusion and restraint practices included open-door seclusion, locked-door seclusion, and any type of restraint. PRN medications included those given “as needed” for acute agitation or aggression. The various PRN medications were collapsed across medication type (diphenhydramine, haloperidol, lorazepam, risperidone, olanzapine, and aripiprazole), dose, and route of administration.

The M-PBIS intervention is a three-tiered continuum of support; tier 1 provides universal locationwide prevention practices, tier 2 consists of targeted interventions, and tier 3 provides intensive individualized interventions. On the inpatient unit, the tier 1 strategies included establishing commitment from the staff (of note, 100% of the unit staff voted to implement the program after their day of training); a defined set of positively worded behavioral expectations (that is, “be safe,” “be responsible,” and “be respectful”); a behavioral matrix identifying how to meet expectations; strategies for teaching the patient and the patient’s family the behavioral expectations during an initial orientation to the unit and throughout the course of the patient’s stay; a reward system in which positive and adaptive behaviors receive positive reinforcement (specifically, staff gave stamps to patients who showed appropriate behavior, with opportunities for stamps to be exchanged for various rewards); the extent to which a 5:1 positive to negative adult-to-child interaction ratio was met (for example, verbally labeling praise to describe positive behavior); a data monitoring system for ongoing decision making, including feedback to staff regarding rates of seclusion and restraint; staff recognition through weekly peer nomination; regular involvement of administrators; an action committee comprising various unit staff members to monitor program fidelity and to resolve unit behavior management issues; and weekly 20-minute in-service training sessions on topics relevant to applying the M-PBIS (for example, understanding how factors such as temperament affect a patient’s behavior).

Tier 2 included targeted problem-solving conversations with only the patients who demonstrated problem behavior on the unit. In most instances these problem-solving conversations yielded behavioral expectations for staff and patients that were incorporated into the patients’ treatment protocol and folded into the M-PBIS reward structure. Tier 3 interventions included functional behavior assessments and individualized behavior plans for a small minority of patients who continued to have problematic behavior after problem-solving conversations. M-PBIS was implemented at all times and throughout the unit.

Staffwide training included an initial daylong course (eight hours) provided to train unit staff across disciplines (including nurses, psychiatric assistants, social workers, and physicians), yearly booster sessions, and ongoing consultation with an on-site psychologist. Training sessions, which included education and role-playing, were scheduled in a way that maximized exposure to the model for all staff members. The program cost to the unit included compensation for 50% of a full-time–equivalent (FTE) licensed psychologist who served as the primary consultant for the M-PBIS intervention. In turn, the consultant’s time was devoted to training new staff, data management, and ongoing consultation. Additional yearly costs also included approximately $1,534 for rewards for patients (rewards included, for example, personal grooming supplies and stuffed animals).

After study implementation, trained rater observations were conducted on the unit to examine the ratio of positive to negative adult-to-child interactions and use of the reinforcement system—that is, providing positive reinforcement in the form of stamps within two seconds of adaptive behavior in tandem with labeled praise and contingent on a specified behavior. The ten-minute observations were conducted (N=335) at select times during the day (for example, at meals and community meetings) by doctoral-level psychologists.

A total of 1,485 admissions occurred during the course of the study, with 726 from January 2010 to October 2011 (preintervention) and 759 from November 2012 to June 2014 (postintervention). Patients’ mean±SD age was 13.18±2.93; 55.2% (N= 793) were female; 31.8% (N=456) were white, 59.2% (N=850) were black, 1.8% (N=26) were Hispanic, 1.3% (N=19) were Asian, and 5.9% (N = 86) were of other races or ethnicities. Demographic and clinical characteristics of the children are summarized in Table 1. The demographic characteristics of the patients changed significantly during the course of the study; the average age increased, whereas decreases were observed in the proportion of boys, black patients, patients covered by Medicaid, and those on a standing neuroleptic medication. The primary diagnoses were relatively stable; however, the proportion of patients with a primary discharge diagnosis of attention-deficit hyperactivity disorder or disruptive behavior disorder decreased, and the proportion of those with adjustment disorders as their primary discharge diagnosis increased. The average length of stay remained stable. The percentage occupancy of the unit significantly increased.

A total of 796 seclusion or restraint events occurred during the study period (some youths had more than one event), comprising 747 locked-door seclusions, 36 open-door seclusions, and 13 restraints. The seclusion-and-restraint rate, expressed per 1,000 patient-hours, was not related to personal characteristics (specifically, age, gender, race, and insurance type) or percentage of unit occupancy; as such, these variables were not included as covariates. After the complete implementation of M-PBIS, there was a marked reduction in the use of seclusion and restraint (from 543 events to 253). The mean seclusion rate significantly decreased from 1.49 to .73. The percentage of patients who were placed in seclusion or restraint significantly decreased, from 19.6% to 13.4%. The mean duration of seclusion and restraint incidents decreased from 20.43 minutes to 8.18 minutes per episode. In order to determine the potential moderating role of diagnosis (that is, whether the intervention had a differential effect by diagnosis), two linear regressions were conducted with number of seclusions and total minutes in seclusion as the outcome and primary diagnosis, intervention status (pre- or postintervention), and their interaction as predictors. In neither case was the interaction significant.

PRN medications were given 2,719 times during the study period. After the complete implementation of M-PBIS, there was a reduction in the use of PRN medications (from 1,705 events to 1,014). The percentage of patients who received a PRN medication significantly decreased, from 41.6% to 29.4%. Use of PRN medication was not related to percentage occupancy; however, it was related to patients’ gender, race, and insurance type. Controlling for gender, race, and insurance type, we found that a significant main effect of intervention group remained (F=5.96, df=1 and 37, p=.02; partial η²=.14). In order to determine the potential moderating role of diagnosis, we ran a linear regression with number of PRN medications given as the outcome and primary diagnosis, intervention status (pre- or postintervention), and their interaction as predictors. The interaction was not significant.

Postintervention, staff members exceeded the goal of at least five positive interactions to one negative interaction with patients. During observations, staff provided on average 8.13±6.52 praising comments, .54±.97 corrective comments, and .22±.67 critical comments. Furthermore, staff members largely awarded stamps appropriately (on average 8.00±6.78 stamps given during the observation)—that is, contingent on behavior (6.62±5.97), with labeled praise (7.02±6.04), and within two seconds of identified behaviors (7.12±6.16).

This study showed reductions in the use of seclusion and restraint and PRN medications in a psychiatric inpatient setting for high-risk youths after the implementation of M-PBIS. There were meaningful reductions in the number of seclusion and restraint events, mean duration of events, percentage of patients receiving this level of restrictive intervention, and the overall seclusion rate. A critical point to emphasize is the accompanying reduction in the use of PRN medications for agitation. Another finding to highlight is that the effect of the intervention was not contingent on a specific primary discharge diagnostic category or increased use of standing neuroleptics. These findings speak to the utility of M-PBIS in reducing multiple forms of restrictive interventions. Such reductions accordingly may serve to reduce the potential for emotional and physical harm to patients and possibly improve the atmosphere in the unit and among staff.

The strengths of the approach need to be considered along with the study’s limitations. First, our results could not identify the intervention components that were most instrumental in achieving outcomes. The study could be strengthened by the inclusion of data for patients receiving tier 2 and tier 3 interventions, assessment of fidelity across all aspects of M-PBIS implementation, perceptions concerning the model of care, and staff stress level and morale. The study covered a relatively short duration of intervention, which raises questions about sustainability over time. Finally, these findings highlight the need for a future randomized clinical trial to more rigorously test the efficacy and effectiveness of M-PBIS in other inpatient settings.