Pediatric behavioral health admissions to children’s hospitals for disposition planning are steadily increasing. These children may exhibit violent behaviors, which can escalate to application of physical limb restraints for safety. Using quality improvement methodology, we sought to decrease physical restraint use on children admitted to our children’s hospital for behavioral health conditions from a baseline mean of 2.6% of behavioral health patient days to <1%.
We included all children ≥3 years of age admitted to our hospital medicine service with a primary behavioral health diagnosis from July 1, 2016, to February 1, 2020. A multidisciplinary team, formed in July 2018, tested interventions based on key drivers targeted toward our aim. The primary outcome measure was the percent of behavioral health patient days on which physical restraints were ordered. The balancing measure was the percent of patient days with a staff injury event. Statistical process control charts were used to view and analyze data.
Our cohort included 3962 consecutive behavioral health patient encounters, encompassing a total of 9758 patient days. A 2-year baseline revealed physical restraint orders placed on 2.6% of behavioral health patient days, which was decreased to 0.9% after interventions and has been sustained over 19 months without any change in staff injuries.
Team-based quality improvement methodology was associated with a sustained reduction in physical restraint use on children admitted for behavioral health conditions to our children’s hospital. These results indicate that physical restraint use can be safely reduced in children’s hospitals.
Children with primary behavioral health conditions are presenting to children’s hospitals with increasing frequency. Encounters for suicidal ideation and suicide attempts in United States children’s hospitals have more than doubled from 2008 to 2015.1 Because of a shortage in pediatric behavioral health services, children requiring admission for behavioral health conditions are frequently admitted to children’s hospitals for safety while pending inpatient psychiatric hospital placement.2 Medical units that temporarily house these children are often not adequately equipped to treat behavioral health challenges, namely agitated and violent behaviors.3 Children’s hospitals work to balance the safety of the children and the hospital staff during de-escalation events. As a last resort, children at high risk of self-harm or injury to staff members may be placed in physical restraints to reduce the risk of violence to themselves and others.4 A physical restraint refers to a physical or mechanical device that is used to limit a patient’s movement.5 In many children’s hospitals, including ours, the physical restraints ordered for violent behavior are typically 4-point polyurethane limb restraints that mechanically restrict patients to their bedframe.
Scarce data exist regarding the use of restraints on children admitted to children’s hospitals. Previous studies from pediatric emergency departments (EDs) indicate ∼ 7% of children who present with primary behavioral health conditions are restrained.6 Although restraints are used for safety purposes, their application is not without the risk of harm. Restrained patients report feeling loneliness and humiliation,7 akin to being imprisoned.8 The trauma of restraint use may exacerbate underlying psychopathology9 because children with behavioral health conditions often have trauma histories.10 The use of physical restraints increases the risk of physical injury to both staff and patients.11 Furthermore, staff members report feeling pity, frustration, and helplessness themselves during restraint events.4
Restraint use has been successfully reduced in pediatric and adult inpatient psychiatric settings.9,12 Previous effective interventions include debriefings,9 staff training on the management of challenging behaviors,13 and the implementation of trauma-informed care.9 Restraint reduction has been shown to decrease both patient12 and staff14 injuries. No previous studies have extrapolated these interventions to reducing restraint use in the setting of a children’s hospital without dedicated psychiatric facilities.
Our study objective was to decrease physical restraint use on children admitted to Monroe Carell Jr. Children’s Hospital at Vanderbilt for a primary behavioral health condition from a 2-year baseline average of 2.6% of behavioral health patient days to <1% by July 2019.
Methods
Setting and Context
Monroe Carell Jr. Children’s Hospital at Vanderbilt is a university-affiliated tertiary care children’s hospital with an associated pediatric ED. Our children’s hospital has 47 000 visits annually, with >16 000 annual discharges from 293 inpatient beds. There is no inpatient psychiatric unit within our children’s hospital, but there is a 28-bed child and adolescent unit at the neighboring Vanderbilt Psychiatric Hospital, which requires a separate acceptance process.
Children presenting for behavioral health concerns are initially seen in our ED. They are evaluated for concurrent medical conditions by multiple providers, including an attending physician. When a primary medical condition is not identified, disposition is then discussed with an ED psychiatry consult service composed of trained pediatric behavioral health assessment specialists.
When admission for a primary behavioral health condition is recommended, the availability of pediatric inpatient psychiatric beds in the surrounding area is evaluated by the assessment specialist. If an appropriate bed is not available, the child is admitted to the pediatric hospital medicine service behavioral health team, and the child and adolescent psychiatry service is consulted to address the child’s acute behavioral health needs. Alternatively, if a child with a primary behavioral health condition is admitted to the hospital for a concurrent medical diagnosis (such as an ingestion), they are transferred to the behavioral health team once medical clearance is confirmed by the attending physician.
Children admitted to the behavioral health team are physically located on 1 of 3 medical and surgical floors, each comprising 3 distinct units. During times of high patient volume, children may remain in 1 of the 33 core beds in the ED until an inpatient bed becomes available. Bedside staff includes nurses in a 3:1 or 4:1 nurse to patient ratio and a 1:1 staff member with varying levels of training, from home health sitters to behavioral health specialists trained by our institution in trauma-informed care.
After project inception in July 2018, the behavioral health team was staffed by pediatric hospital medicine attending physicians, residents, and a case manager. A pediatric psychologist joined the team in October 2018, followed by a pediatric acute care nurse practitioner in November 2018.
The behavioral health team cares for patients on weekdays during business hours. The pediatric hospital medicine teams, staffed by pediatric residents and attending physicians, care for behavioral health patients on nights and weekends. Behavioral health patients are evaluated during their admission by a pediatric psychologist or child and adolescent psychiatrist. The vast majority are placed at an inpatient psychiatry facility; others are discharged from the hospital, with varying degrees of outpatient support, if subsequent evaluations determine that inpatient psychiatric hospitalization is no longer appropriate.
Interventions
Pediatric hospital medicine and child and adolescent psychiatry physicians met with nursing leaders in June 2018. To improve the system of care for children admitted to our hospital for behavioral health conditions, a multidisciplinary team was composed of nurses, physicians, police officers, social workers, and administrative staff. This group began meeting monthly in July 2018 under the context of a staff safety workgroup.
A key driver diagram (Fig 1) was created to display the key drivers and proposed interventions for our aim of decreasing restraint use. This diagram was regularly adjusted on the basis of the results of our plan-do-study-act (PDSA) cycles.
The first key driver was centered on the ability of providers and staff to de-escalate behavioral health patients through a revised de-escalation protocol that based treatments on the patient’s level of agitation. With the second key driver, we addressed timing and type of as-needed medications administered for agitation and violent behaviors by linking the de-escalation protocol to an order set in the electronic medical record (EMR). The third key driver was focused on the efficacy of communication between providers, staff, and families, via a multidisciplinary behavioral health team with restructured team meetings. With our fourth key driver, we worked to improve the child and family’s tolerance of their hospitalization by individualizing treatment plans and revamping the available toys and activities.
Study of the Interventions
Data were collected from the EMR of children admitted for primary behavioral health conditions between July 2016 and February 2020. An automated restraint order report was developed in August 2018 and was compiled weekly, triggering a manual chart review for each physical restraint event. This refreshable report allowed us to obtain retrospective baseline data and follow data prospectively through the project duration. Chart reviews determined if the restrained child was admitted to the behavioral health team at the time of the restraint order. Children admitted for primary eating or substance use disorders, children <3 years of age, and children undergoing concurrent medical workups were excluded.
Staff injury events involving behavioral health patients that resulted in Days Away, Restricted, Transfer15 (DART) time, as defined by the Occupational Safety and Health Administration, were tracked monthly as a balancing measure. Monthly updates of restraint use and staff injury events were reviewed at staff safety workgroup meetings. Multiple rounds of feedback from bedside staff were integrated into our PDSA cycles.
Measures
Outcomes
The primary outcome was the percentage of patient days in which 4-point polyurethane limb restraints were ordered for a behavioral health patient (Table 1). Restraints were considered to be placed only once per patient per day. Foam restraints, mittens, and net beds were excluded from the analysis because our institution does not typically use these devices as restraints for aggressive patients.
Project Measures
| Measurement Group | Measure | Description |
|---|---|---|
| Primary outcome | Percentage of behavioral health patient days with a physical restraint order for violent behavior (4-point polyurethane limb restraints) | A measure of the percentage of patient days in which admitted pediatric behavioral health patients were placed in physical restraints; this measure was calculated weekly as follows: No. patient days in which behavioral health patients had a restraint order placed divided by the weekly total behavioral health population LOS multiplied by 100. |
| This excludes foam restraints, mittens, and net beds. | ||
| Secondary outcome | Days between restraint orders | A measure of duration, in calendar days, between orders for restraints on behavioral health patients. |
| Balancing measure | Percentage of behavioral health patient days with staff injury events which resulted in DART | A measure of the percentage of days on which admitted pediatric behavioral health patients were involved in a staff injury event resulting in DART; this measure was calculated monthly as follows: No. days with DART events involving behavioral health patients divided by the monthly total behavioral health patient LOS multiplied by 100. |
| Measurement Group | Measure | Description |
|---|---|---|
| Primary outcome | Percentage of behavioral health patient days with a physical restraint order for violent behavior (4-point polyurethane limb restraints) | A measure of the percentage of patient days in which admitted pediatric behavioral health patients were placed in physical restraints; this measure was calculated weekly as follows: No. patient days in which behavioral health patients had a restraint order placed divided by the weekly total behavioral health population LOS multiplied by 100. |
| This excludes foam restraints, mittens, and net beds. | ||
| Secondary outcome | Days between restraint orders | A measure of duration, in calendar days, between orders for restraints on behavioral health patients. |
| Balancing measure | Percentage of behavioral health patient days with staff injury events which resulted in DART | A measure of the percentage of days on which admitted pediatric behavioral health patients were involved in a staff injury event resulting in DART; this measure was calculated monthly as follows: No. days with DART events involving behavioral health patients divided by the monthly total behavioral health patient LOS multiplied by 100. |
The secondary outcome was the number of days between restraint events involving behavioral health patients.
Balancing Measure
The balancing measure was the percentage of days in which behavioral health patients were involved in staff injury events that resulted in DART time.
Analysis
The primary outcome was tracked weekly on a p-chart, which measured the percentage of behavioral health patient days with physical restraint orders. Our interventions were monitored over time at monthly staff safety workgroup meetings to determine their impact on the theory behind improvement.16 Special cause variation in our data was identified as a single point outside of the control limits or 8 consecutive points above or below the mean line.16
A rare event t-chart measured the secondary outcome of days between restraint events. A p-chart was used to track our balancing measure of staff safety events. Because of the potential for autocorrelation in our primary outcome, a p-chart examining the monthly percentage of behavioral health patient admissions with physical restraint orders was created after project completion.
Vanderbilt University Medical Center’s Institutional Review Board determined that our project did not qualify as human subjects research.
Results
Our study included 3962 consecutive behavioral health patient encounters in the analysis for the primary outcome. We reviewed data for 2092 patient encounters to establish a 2-year baseline from July 1, 2016, through June 30, 2018. Our intervention period encompassed 1870 patient encounters from July 1, 2018, through February 1, 2020.
PDSA Cycles
A standardized de-escalation protocol based on the Brøset Violence Checklist (Brøset)17 was created in July 2018, with input from pediatric psychiatrists, emergency medicine physicians, hospital medicine physicians, and nursing staff. The Brøset is a 6-item checklist used to assess the presence of 6 independent behaviors to help predict the risk of imminent violent behavior over the following 24 hours. Patients are scored on the Brøset by an observer, receiving 1 point for each of the 6 behaviors that are both present and increased from the baseline. Although validated in an adult population, our improvement team felt the extrapolation to children would be appropriate. Pharmacologic and nonpharmacologic interventions were assigned to each Brøset score as part of our de-escalation protocol.
This Brøset protocol was introduced to a small group of ED nursing staff in September 2018. After multiple PDSA cycles with frontline staff feedback, the Brøset protocol was transitioned from a paper document to an element of the EMR. This was implemented throughout the ED in October 2018. Pediatric residents were introduced to the Brøset protocol initially in July 2018 and again in October 2018. Badge cards were distributed listing the Brøset pharmacologic interventions. After its success in the ED, the Brøset protocol was initiated on our 3 acute care floors in March 2019. Feedback from inpatient nursing staff revealed difficulties obtaining as needed medications in a timely fashion. Thus, an EMR admission order set was created and implemented in May 2019. This order set linked as needed medication orders to Brøset scores, both standardizing medication choices and increasing medication availability. The implementation of this Brøset protocol was monitored by the multidisciplinary team formed through the staff safety workgroup.
In October 2018, the behavioral health team was restructured with the addition of a pediatric psychologist and shortly thereafter with the addition of an acute care pediatric nurse practitioner. A morning meeting of the behavioral health team and the inpatient pediatric psychiatry consultation team was created, helping triage the most acute behavioral health patients among available providers. The existing behavioral health team weekday afternoon meeting was restructured to focus on discharge planning and individualized mitigation plans to assist teams should de-escalation be needed, especially during nights and weekends when the behavioral health team was not present.
Our final intervention was focused on modifying nonpharmacologic care plans for behavioral health patients in accord with the Six Core Strategies of restraint reduction.18 Our strategies targeted trauma-informed care, inclusive treatment planning, and staff training on verbal de-escalation using our Brøset protocol. As an acute care children’s hospital that routinely manages behavioral health patients awaiting disposition, attention to a safe environment is important. Before our study, we addressed the safety of these patients through standardized behavioral health admission protocols. For instance, all behavioral health patients were expected to wear blue paper scrubs, and their daily activities were limited to a preapproved list. Children, especially those with autism spectrum disorder, often became triggered by the sensory changes imposed during this admission process.
In December 2018, a multidisciplinary team including physicians, occupational therapists, child life specialists, nursing staff, and our psychologist drafted a Stoplight Safe Activity System to individualize behavioral health patients’ care plans. This system specified toys and activities that were safe for all behavioral health patients (green light), those deemed safe when approved by a physician or psychologist on the basis of patient’s behavior (yellow light), and those never safe for behavioral health patients (red light). Once this system was tested and improved through PDSA cycles, it was released to all behavioral health patients in March 2019. Feedback from nursing staff regarding ambiguities in the approved patient activity level prompted the addition of a nursing communication to the admission order set in May 2019.
Primary Outcome
The percentage of behavioral health patient days in which physical restraints were ordered for violent behavior improved from our 2-year baseline of 2.6% (148 patient days with restraint orders over 5787 behavioral health patient days) to 0.9% (35 patient days with restraint orders over 3971 behavioral health patient days) during our intervention period from July 1, 2018, through February 1, 2020 (Fig 2). Special cause variation resulting in a shift occurred in July 2018, temporally related to the Brøset protocol introduction, and has been sustained for 19 months through February 1, 2020, with the additional interventions annotated on Fig 2.
The percentage of behavioral health patient days in which a physical restraint order was placed for violent behavior (p-chart). UCL, upper control limit.
The percentage of behavioral health patient days in which a physical restraint order was placed for violent behavior (p-chart). UCL, upper control limit.
Secondary Outcome
After our interventions began in July 2018, the reduction in our primary measure to 0 over multiple consecutive weeks rendered the p-chart less helpful for improvement and led to the use of a rare events t-chart to depict days between restraint events. This both reinforced the improvement depicted on our primary p-chart and provided us with timely and actionable data to inform our PDSA cycles, by allowing the team to recognize multiple spans of time above the upper control limit of 21 days between restraint events (Fig 3).
The number of days between restraint events involving behavioral health patients (t-chart). “Ongoing” is used to indicate the days between the last event and the end of data collection on February 1, 2020. Because of the large number of restraint events, each event is not specifically labeled with a date on the x-axis. UCL, upper control limit.
The number of days between restraint events involving behavioral health patients (t-chart). “Ongoing” is used to indicate the days between the last event and the end of data collection on February 1, 2020. Because of the large number of restraint events, each event is not specifically labeled with a date on the x-axis. UCL, upper control limit.
Balancing Measure
The monthly percentage of staff injury events involving behavioral health patients who resulted in DART time was not significantly changed during our intervention period (Fig 4).
The monthly percentage of behavioral health patient days with staff injury events resulting in DART (p-chart). UCL, upper control limit.
The monthly percentage of behavioral health patient days with staff injury events resulting in DART (p-chart). UCL, upper control limit.
The percentage of behavioral health patient admissions with physical restraint orders also improved from 3.0% at the baseline to 1.0% postintervention (Fig 5).
The percentage of behavioral health patient admissions during which a physical restraint order was placed for violent behavior (p-chart). UCL, upper control limit.
The percentage of behavioral health patient admissions during which a physical restraint order was placed for violent behavior (p-chart). UCL, upper control limit.
Discussion
Through quality improvement methodology,16 we achieved a decrease in the percentage of patient days on which physical restraints were ordered for children admitted for behavioral health conditions from 2.6% to 0.9%. This decrease was sustained for 19 months without increasing the rate of staff safety events attributed to these children. The increased number of days between the use of restraints was displayed as episodes of special cause variation that followed the inception of our interventions in July 2018. To our knowledge, this is the first set of interventions with demonstrable, sustained effects in reducing physical restraint use for violent behaviors on behavioral health patients admitted to an inpatient children’s hospital setting.
Similar to previous studies, we found that primary prevention principles consistent with the Six Core Strategies of restraint reduction played an important role in decreasing restraint use.9,18,19 Our study was unique in the creation of a de-escalation protocol based on the quantification of agitation. We found that the use of Brøset scores in our de-escalation protocol, when implemented through the EMR, allowed for the early identification and treatment of escalating agitated behavior. Additionally, the use of Brøset scores assisted in standardizing communication surrounding agitation between providers and staff and allowed for rapid de-escalation via nonpharmacologic and pharmacologic treatments defined by the admission order set.
Despite the overall success of our interventions, an investigation into the special cause variation, indicated by the points above our upper control limits in our primary measure, revealed that specific children continued to struggle with violent behaviors during their hospitalizations, leading to continued physical restraint use. This subset of children commonly had medical comorbidities, such as autism spectrum disorder or seizure disorders, making them more difficult to place at inpatient psychiatric facilities, and thereby increasing their length of stay (LOS) at our children’s hospital. Children with increased LOS have been found to have increased restraint use,18 thereby creating a cycle of restriction and episodic irritation, agitation, and violent behavior. Future interventions for continued improvement should be focused specifically on this population.
This study has several limitations. It was performed at a university-affiliated, tertiary care children’s hospital, which may limit the generalizability to other pediatric institutions, considering the behavioral health team model used in this study may not be reproducible. Restraint events were tracked by orders placed in the EMR by physicians and may not reflect restraint events that occurred without proper documentation. However, the impact of this limitation was mollified by the attention of hospital staff to physicians’ orders of restraints in compliance with Joint Commission requirements. Alternatively, there may have been situations in which restraint orders were placed but ultimately not used, such as a child that rapidly de-escalated before the restraints were used. Finally, the events in our primary outcome were not completely independent, which is an assumption of p-charts and other attribute data charts, because restrained patients may be more likely to be restrained on subsequent days. However, evidence of robust improvement is clearly demonstrated across all 3 of our control charts.
Conclusions
Implementation of a quality improvement initiative through a multidisciplinary team approach was associated with a sustained reduction in physical restraint use for violent behavior in children admitted for behavioral health conditions at Monroe Carell Jr. Children’s Hospital at Vanderbilt. Multiple interventions contributed to the success of this initiative, including the creation of a multidisciplinary behavioral health team, collaboration with the child and adolescent psychiatry consultation team through restructured team meetings, a Brøset de-escalation protocol, and an admission order set that incorporated our interventions into the EMR. Further efforts should be pursued to determine if the use of as needed medications for behavioral health patients can also be safely decreased.
Drs Herndon, Cundiff, Fuchs, Kreth, and Williams and Mrs Hart, Hughie, Morgan, and Ried conceptualized and designed the study as well as critically reviewed and revised the manuscript for important intellectual content; Drs Dalton and Johnson conceptualized and designed the study as well as critically reviewed and revised the manuscript for important intellectual content and drafted the initial manuscript, designed the data collection instruments, collected data, and conducted the initial analyses; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
A team-based quality improvement initiative successfully reduced the use of physical restraints on children admitted to our children’s hospital for behavioral health conditions.
Comments