Ethical and Legal Foundations of PMHNP Care (discussion)

Vol.:(0123456789)

Child & Youth Care Forum (2021) 50:511–526 https://doi.org/10.1007/s10566-020-09585-y

1 3

ORIGINAL PAPER

Predictors of Injury to Youth Associated with Physical Restraint in Residential Mental Health Treatment Centers

Jonathan Bystrynski1 · Michael T. Braun2  · Catherine Corr1 · Deserai Miller1 · Courtney O’Grady1

Accepted: 14 October 2020 / Published online: 26 October 2020 © Springer Science+Business Media, LLC, part of Springer Nature 2020

Abstract Background The restraining of children is a pervasive but controversial practice that has resulted in the injury and death of children. Despite this threat of harm, little research has explored what factors contribute to the risk of a child becoming injured during a restraint event. Objective This study examined multiple child and restraint factors to better understand what predicts the injury of a child during a restraint event. Methods Longitudinal data (794 youth, 13,339 restraint events) from six residential treat- ment centers in the Midwest were utilized to predict injury associated with restraint using nested hierarchical logistic regression models. Results Injuries to the child are a frequent outcome following the use of restraints on a child. Findings show that a child  being male and older contributed to the likelihood of an injury, but the race of the child did not predict injury. In addition, three physical hold types—prone, supine, and settle—were associated with a greater likelihood of injury. Conclusions Results suggest those who use restraints on children should reevaluate their behavior management plans to reduce their risk of hurting children by restraints. Eliminat- ing or greatly reducing the use of prone, supine, and settle holds will likely lead to a drop in injuries of children.

Keywords Restraints · Injuries · Physical holds · Youth · Residential treatment · Mental health

Introduction

The use of restraints on children is a common but controversial tool used by staff in hospi- tals, schools, and therapeutic residential settings. Restraints are used to reduce the immedi- ate risk of self-harm or violence towards others (Chun et al. 2016a). However, restraints can negatively impact a child’s physical and psychological wellbeing and have resulted in

* Michael T. Braun [email protected]

1 University of Illinois at Urbana-Champaign, Champaign, IL, USA 2 Medical College of Wisconsin, Milwaukee, WI, USA

512 Child & Youth Care Forum (2021) 50:511–526

1 3

the deaths of hundreds of children (Mohr et  al. 2003; Nunno et  al. 2006). Researchers, families, and advocacy groups have sought practice reforms through changes in profes- sional standards and training (Murray and Sefchik 1992; Recupero et  al. 2011). Despite these attempts to decrease the use of child restraints, child restraint policies are often mis- understood or ignored by facilities. In these instances, the continued use of restraints can result in child injury and, in extreme cases, the death of a child (Chun, Mace, and Katz 2016a; Curie 2005; Mohr and Mohr 2003). Although some is known as to what contributes restraint use on children, less is understood about the factors that are associated with injury as part of a restraint event. In this paper, we endeavor to predict injuries to youth who are physically restrained in residential treatment centers (RTCs), using data from nearly 800 youth physically restrained over 13,000 times. In the following sections, we review research on who is restrained, the impact of being restrained, and restraint injuries, before outlining the current study.

Who is Restrained?

Youth in residential treatment centers (RTCs) may at times behave in ways that harm or threaten to harm themselves or others (Cordell and Snowden 2015); in these cases, staff may physically restrain the youth to reduce this risk. Yet not all youth are at equal risk of being restrained, even though restraint rates in RTCs are higher than rates in schools and other settings (Green-Hennessy and Hennessy 2015). To understand factors associated with the use of restraints and seclusion within RTCs, Roy and colleagues (Roy et al. 2019) completed a systematic synthesis of extant literature and identified that younger and male children were more likely to be restrained; further, there was some evidence that children of color were also more likely to be restrained than their White peers.

Individual traits beyond demographics also appear to contribute to the risk of being restrained with disability status appearing to be particularly salient. Children with a special education eligibility or who have mental health concerns are more likely to be restrained; some research has indicated that as many as two-thirds of children with disabilities have been restrained during their lifetime (Delaney and Fogg 2005; Westling et al. 2010). Addi- tionally, children with disabilities are more likely to receive care in a therapeutic residen- tial placement where rates of crisis events involving restraints are disproportionately high (Cordell and Snowden 2015; Powers et al. 2016).

There is also evidence that contextual factors contribute to the risk of being restrained. Trainings on restraint and seclusion interventions often result in spikes of restraint use (Grimes 2012). RTCs with younger, more aggressive staff members, and a higher propor- tion of male staff, also have higher use of restraints (Roy et al. 2019). Staff stress levels appear to be tied to the use of restraints (Hallman et al. 2014), and even the days of the week may contribute disparate risks for use of restraints suggesting the logistics and pro- gramming of a RTC may evoke changes in restraint use (Leidy et al. 2006).

The Impact of Being Restrained

Though restraints are intended to prevent youth from harming themselves or others, restraints are not without consequences for youth. Children perceive being restrained as hostile and traumatizing (Smith and Bowman 2009), and children report that restraints bring about intense feelings of shame, discomfort, and anger (Vishnivetsky et  al. 2013). These emotional reactions are detrimental to the therapeutic process and their relationship

513Child & Youth Care Forum (2021) 50:511–526

1 3

with staff (Chun et  al. 2016a). These negative consequences from being restrained may be particularly taxing for children who have previously experienced abuse. Many children in RTCs have experienced previous trauma (Berry et  al. 2013), and restraints may have lasting negative outcomes for children who have experienced trauma due to the risk of cumulative trauma (Hodges et al. 2013). Children with trauma-related symptoms may be at a higher risk of being restrained (Thomann 2009). Although underexplored, the stress of being restrained combined with the aggressive and self-injurious behaviors that are com- mon following a trauma may result in a cycle alternating between dangerous behaviors and being restrained.

Most importantly, restraints can be deadly. A review of 23 child restraint fatalities showed that the deaths occurred when restraints were used when the child had not reached a point of danger to themselves or others (Nunno et  al. 2006). Within a setting, each restraint of a child is not an independent event. Although understudied, there is evidence that when restraints are used on a child early on in a setting, the child is more likely to be restrained in the future; these harsh initial interactions may contribute to an antagonistic relationship between child and staff (Wu 2015; Goren, Singh, and Best 1993). Given that restraint use is accompanied by this potential loss of rapport and an increase of hostility between the child and staff member, a child who repeatedly engages in threatening behav- iors may evoke responders to restrain them prematurely, increasing the risk of injury.

Child Factors and Restraint Injuries

Not all children bear the same risk of experiencing an injury. Overall, children have a dis- proportionate risk to experience injuries and death while being restrained relative to adults (General Accounting Office 1999); adolescents are at a particular risk for injury or death (Mohr et al. 2003; Nunno et al. 2006). There is some evidence that male children are more likely to be restrained, and this difference may contribute to a higher overall risk of experi- encing an injury (Chun et al. 2016a; Roy et al. 2019). The role of race and ethnicity in the use of restraints is less clear and is understudied across fields (Barragán 2012; Schimmel- mann 2011). For adults in inpatient settings, the existence of racial bias in use of restraints is uncertain; the paucity of extant research has produced conflicting results (Flaherty and Meagher 1980; Gudjonsson, Rabe-Hesketh, and Wilson 2000; Price et  al. 2004; Dye, Brown, and Chhina 2009). Child studies on this potential disparity are even fewer, and they too have left the question unsettled (Donovan et al. 2003; Martin et al. 2008). Racial dis- parities in school discipline, including corporal punishment, have long been documented in American schools (Gershoff and Font 2016; Peguero, Shekarkhar, Popp, and Koo 2015; Shaw and Braden 1990). Although the use of restraints is explicitly not used for punish- ment, racial bias in interpreting the risk of violence may result in more restraint use on children of color (Wilson, Hugenberg, and Rule 2017). This higher number of restraint, if present, may result in a disproportionate rate of injury as a population. More research is needed to better understand race and restraints, but it appears children operate within dif- ferent contexts of risk based on their age, sex, and possibly their race.

Types of Restraints and Injuries

Some forms of restraints are more dangerous than others, and the decisions made by staff members when restraining a child may contribute the most risk of a child experiencing an injury. Prone positioning is particularly risky given its association with injury and

514 Child & Youth Care Forum (2021) 50:511–526

1 3

death, and its use is highly discouraged by medical officials (Chun et al. 2016a; General Accounting Office 1999); see Table 1 for information about different restraint positions. More broadly, any restraint that can result in airway obstruction is particularly dangerous, and any restriction to head rotation can also increase risk of injury; these holds are par- ticularly dangerous for children with obesity and respiratory issues (Johnson 2007; Mas- ters and Bellonci 2002). The length of a restraint is also associated with risk of injury; longer holds are more likely to result in serious injury of the child (Villani et al. 2012). The risk of serious injury during longer restraints is so substantial that it is recommended chil- dren receive health checks conducted by a licensed independent practitioner every 15 min during restraint to reduce risk of injury (Chun et al. 2016b). Taken together, the types of restraint and their duration play a role in the likelihood of experiencing an injury.

The timing of restraints is also important to understand the potential risk of injury. Staff at RTCs have to make complex decisions to balance the risk of violence occurring and the use of restraints. It is recommended that therapeutic residential facilities have individu- alized de-escalation and restraint plans for each child. However, having a plan in place does not guarantee safety. Premature use of restraints violates the goal of using the least restrictive method possible when interacting with children and may result in an escalation of aggression and in increase in risk of injury (Chun et al. 2016a).

Present Study

This current study explores the use of restraints on children at six therapeutic residential settings across the span of three years. We were interested in examining what factors, both restraint-specific and child-specific, predict the association between a restraint event and an injury to the child involved.

In particular, we expect that prone and other restraints that may restrict breathing will be more likely to result in the child receiving an injury. Also, we believe the number of restraints used on a child will increase the association between the restraint and injury. Due to the increase risk of injury when children are more aggravated, we expect periods of high restraint use for a child to be associated with injuries. Specifically, we anticipate the num- ber of restraint events a child has had in the past week will be associated with greater rate of injuries. Finally, we will look at demographic information (i.e., age, race, and gender)

Table 1 Description of physical holds used to restrain youth

Restraint Description

Standing hold The staff member stands behind the child while locking their arms behind their back so the child cannot move them; both are standing during the hold

Settle position From the standing hold, the staff member shifts into a kneeling position with the child sit- ting on the floor with their arms still locked

Seated hold From the settle position, the staff member shifts into a seated position with the child’s arms still locked behind them; typically done against a wall for support

Supine hold Using at least two staff members, the child is shifted onto the floor with their face up and their arms and legs held against the floor by staff

Prone hold Child is shifted onto the ground face down with their arms and legs held down by at least two staff members

Escort Child’s arms are restricted while staff members remove the child from the location of the incident

515Child & Youth Care Forum (2021) 50:511–526

1 3

to examine whether these play a role in predicting a child receiving an injury. Based on the current literature, we predict children who are male, older in age, non-White will have greater association between restraint and injury.

Method

A children’s mental health services company with residential treatment centers across sev- eral Midwestern states provided incident reports as part of a collaborative effort with the second author of this article to explore patterns of restraints in the organization’s facilities. The provider serves children ages 6 through 21 in their residential treatment centers which treats both internalizing and externalizing issues. In accordance with a memorandum of understanding between the provider and the second author and an exempt status granted by the authors’ university, all youth in the provided data were deidentified and assigned a unique numeric ID. Limited demographic information was provided for each youth, includ- ing birth year, sex, and race. Information on each incident included types of restraint used, whether an injury occurred, and reason for restraint.

Sample

Researchers were given access to data for all documented incidents over a three year period, and the types of incidents that were documented were extensive and ranged from threats of harm to self or others to medical issues to attempts to escape the facility. The data do not represent all children who were ever served in the organization’s RTCs; it is only the portion of children who were written up for disciplinary issues. Additionally, not all incident reports provided involved restraints, and these incidents were excluded from analysis. The final data included 794 youth restrained 13,339 times over three years. The majority (72.7%) of the youth were male, and nearly half (48.6%) of the youth were White. The average age at the time of the incident was 14.2 (median = 14, SD = 2.4, range = 7–21; 64 incidents were for youth without a birth year in the data).

Measures

Restraints Used

Eight different physical holds were used at the RTC, and workers reported on each type of restraint used on a child during an incident. These holds included standing hold, prone, seated, supine, escort, settle, small child, and tripod. Descriptions of each physical hold used in this study can be found in Table 1.

Injuries

The RTC also included documentation of any injuries associated with the incident. Reported injuries were primarily abrasions, cuts, joint injuries, and bruises; more severe injuries were also reported including strangulation and head injuries. Centers’ responses to injuries ranged from taking no action to visiting the emergency room; no child deaths were recorded during this three-year period.

516 Child & Youth Care Forum (2021) 50:511–526

1 3

Data Preparation

All data was loaded into R Studio running R version 3.5.3 (RStudio Team 2019), and this software was used for all data cleanup and analyses. The package lme4 was used for the multilevel analyses (Bates et al. 2015). All code is available from the second author.

Incident Characteristics

To prepare the data for analysis, string matching was used to identify the specific physi- cal holds used in each incident. These hold types were used to create dichotomous vari- ables indicating the presence or absence of the particular hold type for each incident. In addition, these holds were summed to create a count of the number of physical holds used in each incident. To increase the chance of model convergence, the number of holds was mean centered for each youth. The incident date was used to calculate the number of incidents for each youth in the prior 7 days, and this value was also mean centered. Injuries were indicated in the data as 0–No injury, 1–Injury. Of the 13,339 incidents, 200 (1.5%) did not include injury information. Table 2 includes the frequency of restraint types used in incidents involving physical restraints.

Youth Characteristics

Sex was coded as 0–Male, 1–Female, and race as 0–White, 1–Non-White. Age was calculated by subtracting the incident year from the youth’s birth year; thus, for some youth, age can vary by incident, if the youth has incidents that span a calendar year (e.g., one incident in December 2016 and another in January 2017 will produce two dif- ferent ages).

Table 2 Frequency of restraint types used in incidents involving physical restraints

n (%)

Mutually Inclusive Standing hold 6566 (49.2) Prone 5332 (40.6) Seated 2615 (19.9) Supine 2179 (16.6) Escort 2075 (15.8) Settle 1338 (10.2) Small child 507 (3.8) Tripod 368 (2.8) Mutually Exclusive 1 type of hold 9631 (72.2) 2 1809 (13.6) 3 989 (7.4) 4 491 (3.7) 5 + 419 (3.1)

517Child & Youth Care Forum (2021) 50:511–526

1 3

Results

Restraint incidents had an average of 1.6 physical holds per incident (median = 1, SD = 1.1, range = 1–12). Nearly 7% of incidents involved 4 or more holds. Standing holds were the most commonly used form of physical restraint, used in 49.2% of all incidents (n = 6,466).

The average number of restraint incidents for each youth in the data was 16.8 (median = 6.5, SD = 34.9, range = 1–497). Nearly a quarter of children (23.0%, n = 183) were reported as experiencing 20 or more restraint events. Overall, injuries associated with a restraint incident were not rare; out of the 13,339 incidents, 10.6% (n = 1,418) involved an injury to the child.

Group Differences in Injuries

There was not a significant difference between sexes and their number of restraint, (f  [1,786] < 1,  p = 0.35), nor in the rate of injury during a restraint event (f [1,786] = 3.17, p = 0.08). There was also no significant difference between races (White and Non-White) in number of restraint (f  [1,759] = 1.70,  p = 0.19) nor in rate of injury (f [1, 759] = 1.56, p = 0.21). There was also no significant difference by race (White, Afri- can American, and Mixed Race) in number of restraint (f [2, 689] = 2.93, p = 0.054] nor in injury rate (f [2, 689] = 2.83, p = 0.06). See Table 3 for more information regarding restraint injuries and different demographic groups.

Predicting Injuries

Analyses were conducted using multilevel models with three levels, incidents nested within youth nested within sites (Szmaragd and Leckie 2011). The lme4 package (Bates et al. 2015) in R (RStudio Team 2019) was used for all analyses. Predictors at the inci- dent level included number of physical holds used, types of hold, number of holds in

Table 3 Demographic differences with the average number of incidents and average number of injuries per incident

*6 youth had no sex recorded. 33 had no race recorded. 6 youth had no age recorded

n (%) Average number of incidents

F p Percent of Incidents Involving Injury

F p

Sex* Male 577 (72.7) 16.4 0.87 0.35 13.3 3.17 0.08 Female 211 (26.6) 18.8 10.4 Race White 386 (48.6) 14.9 1.70 0.19 13.7 1.56 0.21 Non-White 375 (47.2) 17.8 11.8 Age 7–10 39 40.6 37.5 < .001 6.9 2.17 0.14 11–13 163 27.1 12.0 14–17 527 13.3 13.0 18–21 149 9.8 13.2

518 Child & Youth Care Forum (2021) 50:511–526

1 3

previous 7 days, and youth age at time of incident. Two specific hold types in the data were excluded (small child and tripod) because they occurred infrequently and because their presence resulted in a model that would not converge. Predictors at the youth level included sex and race. There were no predictors at the site level, but site was included as part of the nested structure because of possible variation in use of restraints and injuries between sites.

Five nested models, including a null model, were tested, beginning with predictors at the incident level and ending with the child-level predictors. The models were compared using a likelihood ratio test to assess if each subsequent model explained significantly more variance than the previous model (see Table 4).

Model 1

The first model predicted injury using only the number of physical holds (mean centered) in each incident. This predictor was significant (γ = 0.33, SE = 0.03, z = 12.69, p < 0.001). A higher value indicates a number of physical holds in an incident that is greater than the average number of physical holds usually used with that youth. The positive value sug- gests that as number of physical holds increases above youths’ averages, the risk for injury significantly increases. The first model was compared to the null model using a likelihood ratio test to see if the model explained significantly more variance. The test was significant (χ2 [1] = 157.2, p < 0.001), suggesting the addition of number of holds used into the model explained significantly more variance than the null model.

Model 2

In the second model, we added in restraint types as dichotomous variables for standing, seated, prone, settle, supine, and escort restraints. Number of restraints remained sig- nificant (γ = 0.22, SE = 0.04, z = 6.10, p < 0.001). Additionally, three specific restraint types were significant. Prone restraints (γ = 0.94, SE = 0.08, z = 11.98, p < 0.001), set- tle restraints (γ = 0.31, SE = 0.10, z = 3.16, p = 0.002), and supine restraints (γ = 0.44, SE = 0.10, z = 4.42, p < 0.001) were all associated with significantly increased chance of injury. Standing, seated, and escort restraints were not significant. This model was com- pared to the first model to see if it explained significantly more variance. The likelihood ratio test was significant (χ2 [6] = 198.6, p < 0.001), suggesting the addition of hold types into the model explained significantly more variance than the model with number of holds alone.

Model 3

In the third model, we added in number of restraints in the past 7 days. All previous pre- dictors remained significant (number of physical holds, use of prone holds, settle holds, and supine holds). In addition, the number of restraints in the past 7 days was a signifi- cant predictor. As the number of restraints in the past 7 days increases above a youth’s average number of restraint events, the risk for injury decreases significantly. This model also explained significantly more variance than the previous model (χ2 [1] = 4.7, p = 0.03).

519Child & Youth Care Forum (2021) 50:511–526

1 3

Table 4 Nested hierarchical logistic regression models predicting likelihood of injury from physical restraint

Null Model Model 1 Model 2

Estimate SE Z Estimate SE Z Estimate SE Z

Incident Level Intercept −2.25 0.37 −6.11*** −2.26 0.36 −6.21*** −2.68 0.42 −6.45*** Number of

Holds 0.33 0.03 12.69*** 0.21 0.04 6.10***

Standing Hold

−0.14 0.08 −1.74

Seated Hold 0.06 0.09 0.69 Prone Hold 0.94 0.08 11.97*** Settle Hold 0.31 0.10 3.16** Supine Hold 0.44 0.10 4.42*** Escort Hold −0.14 0.10 −1.31 Model Fit Total DF 3 4 10 AIC 8026.9 7871.8 7685.2 BIC 8049.1 7901.4 7759.3 Log Likeli-

hood −4010.5 −3931.9 −3832.6

Deviance 8020.9 7863.8 7665.2 Chi Square

(DF) 157.2 (1)*** 198.6 (6)***

Model 3 Model 4

Estimate SE Z Estimate SE Z

Incident Level Intercept −2.69 0.42 −6.44*** −5.46 0.62 −8.87*** Number of Holds 0.22 0.04 6.14*** 0.20 0.04 5.66*** Standing Hold −0.14 0.08 −1.73 −0.12 0.08 −1.49 Seated Hold 0.06 0.09 0.71 0.10 0.09 1.18 Prone Hold 0.94 0.08 12.00*** 0.95 0.08 12.15*** Settle Hold 0.31 0.10 3.12** 0.33 0.10 3.36*** Supine Hold 0.44 0.10 4.46*** 0.43 0.10 4.29*** Escort Hold −0.14 0.10 −1.38 −0.11 0.10 −1.01 Holds in Past 7 Days −0.04 0.02 −2.20* −0.03 0.02 −1.95 Age 0.19 0.02 7.96*** Child Level Sex (1 = Female) −0.43 0.15 −2.87** Race (1 = Non-White) −0.09 0.11 −0.79 Model Fit Total DF 11 14 AIC 7682.5 7622.3 BIC 7764.1 7726.1 Log Likelihood −3830.3 −3797.1 Deviance 7660.5 7594.3 Chi Square (DF) 4.7 (1)* 66.2 (3)***

***p < .001, **p < .01,,*p < .05

520 Child & Youth Care Forum (2021) 50:511–526

1 3

Model 4

In the final model, we added in youth sex, race, and age. All previous predictors remained significant in the same direction except restraints in the past 7 days, which was not sig- nificant in this model. In addition, both age and sex were significant. Older youth were significantly more likely to be injured (γ = 0.19, SE = 0.02, z = 7.96, p < 0.001). Addition- ally, being female was associated with a significant decrease in the likelihood of injury (γ = −0.43, SE = 0.15, z = −2.87, p = 0.004). This model explained significantly more vari- ance than the previous model (χ2 [3] = 66.2, p < 0.001).

Discussion

The necessity of restraints in residential treatment facilities to prevent youth from harming themselves or others does not mean restraints themselves have no negative consequences. This study attempted to predict one of those consequences—injuries to youth—using mul- tilevel logistic regression with a sample of 794 youth in residential treatment facilities who were restrained in 13,339 incidents over the course of three years. In the following para- graphs, we review the findings of this study, discuss the implications for youth and facili- ties, and suggest directions for future research.

The Use of Restraints

The use of restraints on youth was a common phenomenon within the RTCs explored in this study, and these restraints were associated with physical injury to the restrained youth in over 10% of incidents. Our analysis of predictors of injury revealed several consistent predictors of injury associated with physical restraint. First, the number of physical holds used (mean centered per youth) was consistently associated with a greater risk of injury in every model explored. That is, when staff restrained a child in a single event that used more physical holds than was typical for that specific youth, the risk of youth injury associated with the incident increased significantly. Needing several types of restraints suggests the child may be highly agitated, and this elevated distress during a restraint may contribute to a child becoming injured (Paterson et al. 2003). Additionally, staff may also be changing their hold because they may have initially used a developmentally inappropriate hold on a child, which could increase their risk of injury. If a hold was performed inappropriately or inadequately, the child may become injured either from their effects to escape or by the hold itself. Multiple restraints in one event may also be a proxy for the length of time spent restraining a child, a factor associated with injury and death (Chun et al. 2016b).

Three specific hold types were also associated with greater risk of injury: prone holds, supine holds, and settle holds. As expected, prone holds, which involve a child being restrained facedown, were associated with a greater rate of injury, as were supine holds, the faceup counterpart of prone holds. This association between prone position and injury reflects the cautions shared by the American Academy of Pediatrics and the recommenda- tion that the use of these holds be limited (Chun et al. 2016a). The high frequency at which prone holds were used (40.6% of all restraint events) is alarming and suggests that this type of hold may be a preferred approach by staff. This is particularly chilling given that the majority of child deaths occurred in this position in a review of deaths by restraints

521Child & Youth Care Forum (2021) 50:511–526

1 3

(Nunno, Holden, and Tollar 2006). Retraining may be required to reduce or eliminate the use of this dangerous type of hold.

The age of the youth was significantly associated with risk of injury with older age youth having a greater risk of injury. Perhaps older youth are harder to restrain (e.g., more able to resist the hold due to their strength), or perhaps staff are more aggressive in their approach to physically restraining older youth because these youth are perceived as physi- cally stronger and able to resist the hold. Regardless of the explanations, this elevated risk for injury emphasizes the need for emotion regulation strategies and de-escalation efforts for this population, as restraining children does not contribute to their ability to self-regu- late (Cusack et al. 2003). Older children are more readily able to engage in emotion regula- tion strategies (Cole, Armstrong, and Pemberton 2010; Davis 2016), and RTCs should take advantage of this. These efforts to avoid restraining this population will have the dual bene- fit of increasing down-regulation skills while avoiding the potential trauma-related distress from being restrained.

Although number of restraints in the past 7 days was not a significant predictor in the final model, we note that this was a significant predictor in the penultimate model. Specifi- cally, an increased number of restraints in the past 7 days was associated with a decrease in risk of injury. This finding was unexpected, and we suggest future research explore this potential relationship as it may be influenced by the aforementioned factors such as the perception of the child by staff, previous injuries leading to less struggle when restrained again, or other contextual factors.

The sex of the youth was also significantly associated with injury; female youth had a significantly lower rate of injury associated with restraint than male youth. This difference does not give credence to the idea that restraining girls is safer. Rather, attention should be paid as to how workers are treating male and female youth differently. The literature suggests that boys are more likely to be restrained than girls, and this reflects the differ- ence in rates of aggression between genders (Zimmer-Gembeck, Geiger, and Crick 2005). Although this difference in likelihood of injury may be due to boys’ physical aggressive- ness (Lansford et al. 2012), there may be qualitative differences in the ways staff approach restraint of boys. For example, perhaps staff members judge boys as more aggressive, thus necessitating greater aggression from staff during restraint. More research is needed to understand the differing injury risk between boys and girls.

Implications

Practical experience and research confirm that physical restraints can result in injuries to those being restrained; there is likely no method of physically restricting the movement of someone else that can ever be consequence-free. The results of this study, however, sug- gest that understanding particular risks related to restraints may be particularly harmful to children, and this may offer a starting point for reducing the harm that restraints may cause.

First, selective attention should be paid to the use of restraints identified as uniquely dangerous: prone holds, supine holds, and settle holds. Facilities should explore how to eliminate or deeply limit the use of these holds; purveyors of restraint training programs can develop alternatives. Second, facilities could develop plans for the use of restraints that consider the unique dangers of physically restraining older youth and male youth. Although more research is required to fully understand the specific risks for older youth and males, RTCs may begin to focus on this population as being particularly vulnerable within their programs. Future research should include the experiences of staff because understanding

522 Child & Youth Care Forum (2021) 50:511–526

1 3

their roles and work culture may lead to better understanding of how restraint approaches may differ by youth characteristics like age and sex.

Lastly, we note that restraints are not a trauma-informed practice, and that the use of restraints may traumatize youth and hinder their progress in their treatment goals. The need to address trauma is increasingly viewed as an important component of effective behavio- ral health and educational service delivery (Substance Abuse and Mental Health Services Administration [SAMHSA] 2014). Perhaps training staff in RTCs on the traumatizing role that restraints may have on youth will help reduce restraints, as staff reinvest attention to other de-escalation strategies.

Limitations and Future Research

The limitations of this paper are myriad. First, the facilities at which these restraints occurred could not produce overall demographic data during the same period. Thus, we cannot make broader conclusions around gender and race as it relates to the larger popula- tion of children living at the RTCs. For example, we cannot conclude that a facility with a greater proportion of older youth will have a higher rate of injuries, because we cannot determine the overall age makeup of a facility. Future research should explore how the overall demographics of the residents of an RTC shape the risk of injury associated with restraint.

We also cannot determine the accuracy of injury reporting. Many injury reports were for less serious concerns (e.g., scratches, reported pain), but the subjective nature of non- obvious injuries may mean that some injuries occurred but were either unreported by chil- dren or undocumented by staff, It is also possible that the threshold for what was or was not considered an injury varied among staff, perhaps because youth felt more comforta- ble sharing injuries with some staff than with others. We have no evidence to suggest any systematic bias in injury reporting. Additionally, we have no means to determine when a restraint caused an injury or when an injury happened as part of an incident that also involved a restraint; this information would strengthen this paper’s conclusions.

We also have no information on staffing, including unique identifiers of staff performing restraints or a staff census at the time of restraint. For RTCs to act to reduce restraints and injuries, knowledge of staffing may be vital. For example, perhaps a small number of staff account for a disproportionate number of injuries; this knowledge would allow for targeted retraining. Perhaps restraints and injuries are more likely when staff to youth ratios are low or high; this knowledge would help RTCs adjust staffing to protect youth. Future research should consider staff members and staffing levels as possible predictors of injuries associ- ated with restraints.

Fourth, we have not considered the harm that restraints prevent because this information is not documented. According to policy, restraints are designed to prevent harm, and in 90% of restraint incidents, no associated injury was reported. The ability to predict injury associated with restraint does not negate the fact that most restraints occur without associ- ated injury. While policy goals may advise a reduction in restraint use, and while this paper may offer direction for which restraint types should be eliminated because of their danger, the use of restraints in a vast majority of cases are performed without associated physical injury. A minor injury associated with physical restraint may be an acceptable outcome when the youth restrained was engaged in suicidal behavior or threatening another youth with a weapon.

523Child & Youth Care Forum (2021) 50:511–526

1 3

Finally, we offer a general call for greater descriptive research on the use of restraints among vulnerable populations. We believe there is always value in research that describes and documents events that occur in facilities where little research has been done. We encourage more researchers to describe—quantitatively or qualitatively—restraints and their consequences for youth in RTCs. We are especially curious to learn more about these facilities through narrative accounts of restraints, from the perspectives of staff, youth who are restrained, and youth who witness restraints. A large dataset, like the one used in this paper, is valuable for some types of analysis, but we regret that it is only a shallow look at the complexity of restraints in treatment facilities.

Conclusion

Injuries to youth associated with physical restraint are not rare in residential treatment centers, and the risk of injury is higher when prone, supine, and settle holds are used, and when the restrained youth is older and male. Given the risk of physical injury and damage to therapeutic progress (Vishnivetsky et al. 2013) from restraints, treatment facilities and regulators should expand their efforts to limit or eliminate the use of restraints on children, particularly prone, supine, and settle holds. Future research is needed to understand more about how this process can occur safely for youth and staff.

Funding None.

Compliance with Ethical Standards

Conflict of interest The authors declare that they have no conflict of interest.

Ethical Approval The dataset provided was de-identified and thus analysis of the secondary data was judged exempt from IRB oversight by the authors’ academic institution.

References

Bates, D., Maechler, M., Bolker, B., & Walker, S. (2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software, 67(1), 1–48. https ://doi.org/10.18637 /jss.v067.i0

Barragán, A. (2012). Seclusion and mechanical restraints among ethnic minorities: Understudied and needed area of research. Mental Health Law and Policy Journal, 1(1), 99–125.

Berry, K., Ford, S., Jellicoe-Jones, L., & Haddock, G. (2013). PTSD symptoms associated with the experi- ences of psychosis and hospitalisation: A review of the literature. Clinical Psychology Review, 33, 526–538. https ://doi.org/10.1016/j.cpr.2013.01.011

Chun, T. H., Mace, S. E., & Katz, E. R. (2016a). Evaluation and management of children and adolescents with acute mental health or behavioral problems Part I: Common clinical challenges of patients with mental health and/or behavioral emergencies. Pediatrics, 138(3), e1–e22. https ://doi.org/10.1542/ peds.2016-1570

Chun, T. H., Mace, S. E., & Katz, E. R. (2016b). Evaluation and management of children with acute men- tal health or behavioral problems Part II: Recognition of clinically challenging mental health related conditions presenting with medical or uncertain symptoms. Pediatrics, 138(3), e1–e23. https ://doi. org/10.1542/peds.2016-1573

Cole, P. M., Armstrong, L. M., & Pemberton, C. K. (2010). The role of language in the development of emotion regulation. In S. D. Calkins & M. A. Bell (Eds.), Human brain development. Child

524 Child & Youth Care Forum (2021) 50:511–526

1 3

development at the intersection of emotion and cognition (p. 59–77). Washington DC: American Psychological Association.

Cordell, K., & Snowden, L. (2015). Emotional distress dispositions and crisis intervention for children treated for mental illness. Journal of Child and Family Studies, 24(9), 2699–2709. https ://doi. org/10.1007/s1082 6-014-0072-8

Curie, C. G. (2005). Special section on seclusion and restraint: Commentary: SAMHSA’s commitment to eliminating the use of seclusion and restraint. Psychiatric Services, 56(9), 1139–1140. https :// doi.org/10.1176/appi.ps.56.9.1139

Cusack, K. J., Frueh, B. C., Hiers, T., Suffoletta-Maierle, S., & Bennett, S. (2003). Trauma within the psychiatric setting: A preliminary empirical report. Administration and Policy in Mental Health and Mental Health Services Research, 30(5), 453–460. https ://doi.org/10.1023/a:10246 97715 409

Davis, E. L. (2016). An age-related mechanism of emotion regulation: Regulating sadness promotes children’s learning by broadening information processing. Child Development, 87(5), 1617–1626. https ://doi.org/10.1111/cdev.12552

Delaney, K. R., & Fogg, L. (2005). Patient characteristics and setting variables related to use of restraint on four inpatient psychiatric units for youths. Psychiatric Services, 56, 186–192. https ://doi. org/10.1176/appi.ps.56.2.186

Donovan, A., Plant, R., Peller, A., Siegel, L., & Martin, A. (2003). Two-year trends in the use of seclu- sion and restraint among psychiatrically hospitalized youths. Psychiatric Services, 54(7), 987–993. https ://doi.org/10.1176/appi.ps.54.7.987

Dye, S., Brown, S., & Chhina, N. (2009). Seclusion and restraint usage in seven English psychiat- ric intensive care units (PICUs). Journal of Psychiatric Intensive Care, 5(2), 69–79. https ://doi. org/10.1017/S1742 64640 90014 35

Flaherty, J. A., & Meagher, R. (1980). Measuring racial bias in inpatient treatment. The American Jour- nal of Psychiatry, 137, 679–682. https ://doi.org/10.1176/ajp.137.6.679

General Accounting Office. (1999). Improper restraint or seclusion use places people at risk. Washing- ton, D.C.: General Accounting Office.

Gershoff, E. T., & Font, S. A. (2016). Corporal punishment in US public schools: Prevalence, disparities in use, and status in state and federal policy. Social Policy Report, 30, 3–25.

Goren, S., Singh, N. N., & Best, A. M. (1993). The aggression-coercion cycle: Use of seclusion and restraint in a child psychiatric hospital. Journal of Child and Family Studies, 2, 61–73. https ://doi. org/10.1007/BF013 21301

Green-Hennessy, S., & Hennessy, K. D. (2015). Predictors of seclusion or restraint use within residential treatment centers for children and adolescents. Psychiatric Quarterly, 86(4), 545–554. https ://doi. org/10.1007/s1112 6-015-9352-8

Grimes, R. D. (2012). Predicting involvement in coercive interventions from individual and contextual risk factors and treatment context (Unpublished doctoral dissertation). Tuscaloosa, Alabama: Uni- versity of Alabama.

Gudjonsson, G. H., Rabe-Hesketh, S., & Wilson, C. (2000). Violent incidents on a medium secure unit: The target of assault and the management of incidents. The Journal of Forensic Psychiatry, 11(1), 105–118. https ://doi.org/10.1080/09585 18003 62391

Hallman, I. S., O’Connor, N., Hasenau, S., & Brady, S. (2014). Improving the culture of safety on a high-acuity inpatient child/adolescent psychiatric unit by mindfulness-based stress reduction train- ing of staff. Journal of Child and Adolescent Psychiatric Nursing, 27(4), 183–189. https ://doi. org/10.1111/jcap.12091

Hodges, M., Godbout, N., Briere, J., Lanktree, C., Gilbert, A., & Kletzka, N. T. (2013). Cumulative trauma and symptom complexity in children: A path analysis. Child Abuse & Neglect, 37, 891–898. https ://doi.org/10.1016/j.chiab u.2013.04.001

Johnson, T. D. (2007). Respiratory assessment in child and adolescent residential treatment settings: Reducing restraint-associated risks. Journal of Child and Adolescent Psychiatric Nursing, 20(3), 176–183. https ://doi.org/10.1111/j.1744-6171.2007.00108 .x

Lansford, J. E., Skinner, A. T., Sorbring, E., Giunta, L. D., Deater-Deckard, K., Dodge, K. A., & Uribe Tirado, L. M. (2012). Boys’ and girls’ relational and physical aggression in nine countries. Aggres- sive behavior, 38(4), 298–308.

Leidy, B. D., Haugaard, J. J., Nunno, M. A., & Kwartner, J. K. (2006). Review of restraint data in a resi- dential treatment center for adolescent females. Child & Youth Care Forum, 35, 339–352. https :// doi.org/10.1007/s1056 6-006-9021-z

Martin, A., Krieg, H., Esposito, F., Stubbe, D., & Cardona, L. (2008). Reduction of restraint and seclu- sion through collaborative problem solving: A five-year prospective inpatient study. Psychiatric Services, 59, 1406–1412. https ://doi.org/10.1176/appi.ps.59.12.1406

525Child & Youth Care Forum (2021) 50:511–526

1 3

Masters, K. L., & Bellonci, C. (2002). Practice parameter for the prevention and management of aggres- sive behavior in child and adolescent psychiatric institutions, with special reference to seclusion and restraint. Journal of the American Academy of Child and Adolescent Psychiatry, 41(2 Supple- ment), 4S-25S. https ://doi.org/10.1097/00004 583-20020 2001-00002

Mohr, W. K., & Mohr, B. D. (2000). Mechanisms of injury and death proximal to restraint use. Archives of Psychiatric Nursing, 14(6), 285–295. https ://doi.org/10.1053/apnu.2000.19091

Mohr, W. K., Petti, T. A., & Mohr, B. D. (2003). Adverse effects associated with physical restraint. Canadian Journal of Psychiatry, 48, 330–337. https ://doi.org/10.1177/07067 43703 04800 509

Murray, L., & Sefchik, G. (1992). Regulating behavior management practices in residential treat- ment facilities. Children and Youth Services Review, 14, 519–539. https ://doi.org/10.1016/0190- 7409(92)90004 -F

Nunno, M. A., Holden, M. J., & Tollar, A. (2006). Learning from tragedy: A survey of child and ado- lescent restraint fatalities. Child Abuse & Neglect, 30, 1333–1342. https ://doi.org/10.1016/j.chiab u.2006.02.015

Paterson, B., Bradley, P., Stark, C., Saddler, D., Leadbetter, D., & Allen, D. (2003). Deaths asso- ciated with restraint use in health and social care in the UK The results of a preliminary sur- vey. Journal of Psychiatric and Mental Health Nursing, 10(1), 3–15. https ://doi.org/10.104 6/j.1365-2850.2003.00523 .x

Peguero, A. A., Shekarkhar, Z., Popp, A. M., & Koo, D. J. (2015). Punishing the children of immigrants: Race, ethnicity, generational status, student misbehavior, and school discipline. Journal of Immi- grant & Refugee Studies, 13(2), 200–220. https ://doi.org/10.1080/15562 948.2014.95113 6

Powers, A., Fani, N., Cross, D., Ressler, K., & Bradley, B. (2016). Childhood trauma, PTSD, and psy- chosis: Findings from a highly traumatized, minority sample. Child Abuse & Neglect, 58, 111–118. https ://doi.org/10.1016/j.chiab u.2016.06.015

Price, T. B., David, B., & Otis, D. (2004). The use of restraint and seclusion in different racial groups in an inpatient forensic setting. Journal of the American Academy of Psychiatry Law, 32, 163–168.

Recupero, P. R., Price, M., Garvey, K. A., Daly, B., & Xavier, S. L. (2011). Restraint and seclusion in psychiatric treatment settings: regulation, case law, and risk management. Journal of the American Academy of Psychiatry and the Law Online, 39(4), 465–476.

Roy, C., Castonguay, A., Fortin, M., Drolet, C., Franche-Choquette, G., Dumais, A., et al. (2019). The use of restraint and seclusion in residential treatment care for youth: A systematic review of related factors and interventions. Trauma, Violence, & Abuse. https ://doi.org/10.1177/15248 38019 84319 6

RStudio Team. (2019). RStudio: Integrated development for R. Boston, MA: RStudio Inc. Schimmelmann, B. G. (2011). Seclusion and restraints: A neglected area of research in child and adoles-

cent psychiatry? European Child & Adolescent Psychiatry, 20(5), 219–220. https ://doi.org/10.1007/ s0078 7-011-0170-8

Shaw, S. R., & Braden, J. P. (1990). Race and gender bias in the administration of corporal punishment. School Psychology Review, 19(3), 378–383.

Smith, M. L., & Bowman, K. M. (2009). The restraint spiral: Emergent themes in the perceptions of the physical restraint of juveniles. Child Welfare, 88(3), 58–83.

Substance Abuse and Mental Health Services Administration. (2014). SAMHSA’s concept of trauma and guidance for a trauma-informed approach. HHS Publication No. (SMA) 14–4884.

Szmaragd, C., & Leckie, G. (2011). Module 7: Multilevel models for binary responses. Bristol, UK: Centre for Multilevel Modeling. Retrieved from https ://www.cmm.bris.ac.uk/lemma /cours e/view. php?id=13

Thomann, J. (2009). Factors in restraint reduction in residential treatment facilities for adolescents (Unpublished doctoral dissertation). Newton, Massachusetts: Massachusetts School of Professional Psychology.

Villani, V. S., Parsons, A. E., Church, R. P., & Beetar, J. T. (2012). A descriptive study of the use of restraint and seclusion in a special education school. Child & Youth Care Forum, 41, 295–309. https ://doi.org/10.1007/s1056 6-011-9165-3

Vishnivetsky, S., Shoval, G., Leibovich, V., Giner, L., Mitrany, M., Cohen, D., et al. (2013). Seclusion room vs physical restraint in an adolescent inpatient setting: Patients’ attitudes. The Israeli Journal of Psychiatry and Related Sciences, 50(1), 6–10.

Westling, D. L., Trader, B. R., Smith, C. A., & Marshall, D. S. (2010). Use of restraints, seclusion, and aversive procedures on students with disabilities. Research & Practice for Persons with Severe Dis- abilities, 35(3–4), 116–127. https ://doi.org/10.2511/rpsd.35.3-4.116

Wilson, J. P., Hugenberg, K., & Rule, N. O. (2017). Racial bias in judgments of physical size and formi- dability: From size to threat. Journal of Personality and Social Psychology, 113(1), 59–80. https :// doi.org/10.1037/pspi0 00009 2

526 Child & Youth Care Forum (2021) 50:511–526

1 3

Wu, W. W. K. (2015). Psychosocial correlates of patients being physically restrained within the first 7 days in an acute psychiatric admission ward: Retrospective case record review. East Asian Archives of Psychiatry, 25, 47–57.

Zimmer-Gembeck, M. J., Geiger, T. C., & Crick, N. R. (2005). Relational and physical aggression, prosocial behavior, and peer relations: Gender moderation and bidirectional associations. The Journal of Early Adolescence, 25(4), 421–452. https ://doi.org/10.1177/02724 31605 27984 1

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Child & Youth Care Forum is a copyright of Springer, 2021. All Rights Reserved.