Pharmacologic Restraint Use During Mental Health Admissions to Children’s Hospitals

This was a retrospective, cross-sectional study of the Pediatric Health Information System (PHIS) database, which contains clinical and billing data from 49 tertiary care children’s hospitals. Of the 49 participating hospitals, 43 had consistent data reporting sufficient for inclusion in this study. All submitted data are deidentified. Data quality and reliability are assured through a joint effort between the Children’s Hospital Association (Lenexa, Kansas) and the respective hospitals. This study was approved for exemption by the institutional review board of the University of Texas Southwestern.

The study population included patients aged 5 to 17 years between January 1, 2016, and December 31, 2021, with a primary discharge diagnosis of a mental health condition. The Childhood and Adolescent Mental Health Disorders Classification System was used to create categories of mental health diagnoses.³¹  The categories included: Anxiety disorders, attention-deficit/hyperactivity disorder, autism, bipolar and related disorders, depressive disorders, disruptive disorders, eating disorders, neurocognitive disorders, psychotic disorders, somatic disorders, substance-related and addictive disorders, suicide or self-injury, and trauma and stressor-related disorders (Supplemental Table 4). To focus the study on medical inpatient units, hospitalizations involving a charge for an inpatient psychiatric unit or discharge from an ICU were excluded. Patient days involving an ICU charge were also excluded, whereas patient days after transfer to the acute-care floor from the ICU were included (Supplemental Fig 3).

Data elements included sociodemographic characteristics of age, sex, race/ethnicity, insurance type, median household income, and Child Opportunity Index (COI), and hospitalization characteristics including disposition and length of stay. The COI measures the quality of resources and conditions important for childhood development at the zip code level, and it provides a score from 1 to 100, with 100 representing the highest opportunity level.³² 

Pharmacologic restraint use for acute agitation was defined as parenteral (intravenous [IV] or intramuscular [IM]) administration of short-acting formulations of antipsychotic medications including aripiprazole, droperidol, haloperidol, olanzapine, and ziprasidone (Supplemental Table 5). These medications were included on the basis of previous studies.^27,33–36  This narrow definition of pharmacologic restraint was used because these medications, when used in the short-acting, IV/IM form, have a specific association with the management of acute agitation. Thus, oral formulations were excluded because regularly scheduled oral medications could not be distinguished from as-needed usage. Similarly, benzodiazepines (often used for nausea, seizures, drug withdrawal, and anxiolysis), antihistamines (often used for nausea, allergic reaction, pruritis, and insomnia), and other previously cited medications with broad nonrestraint indications were also excluded to avoid outcome misclassification.^11,28–30  Antipsychotic medications (chlorpromazine and prochlorperazine) with other common indications were also excluded. Finally, admissions involving an operating room charge were excluded because these may represent a nonrestraint indication for parenteral administration.

Descriptive statistics were used to characterize the sociodemographic (age, sex, race/ethnicity, insurance status, median household income, COI) and clinical (disposition, length of stay) characteristics of the study population. χ² tests were used to compare hospitalizations with versus without pharmacologic restraint use for these characteristics. The rate of pharmacologic restraint use by primary mental health discharge diagnosis was represented by days involving pharmacologic restraint use per 1000 patient days. Using Poisson regression, usage rates were adjusted for age within each diagnosis category and adjusted for age and mental health condition for the overall rate with accounting for hospital clustering. The Poisson regression model was also used to assess trends in pharmacologic use for each diagnosis and to calculate adjusted rate ratios (RRs) comparing pharmacologic restraint use days per 1000 patient days for 2016 and 2021. All statistical analyses were performed with SAS v.9.4 (SAS Institute, Cary, North Carolina), and P values < .05 were considered statistically significant.

During the study period, 91 898 mental health hospitalizations were identified representing 358 711 patient days. Pharmacologic restraint was used in 2754 (3%) of admissions and 4679 (1.3%) of patient days.

Most patients admitted for a primary mental health diagnosis were non-Hispanic white (57.9%), female (69.7%), and aged 9 to 12 (44.4%) or 13 to 17 (49.4%) years. There was a significant difference in nearly all sociodemographic characteristics when comparing hospitalizations with versus without pharmacologic restraint use. Disproportionately higher rates of pharmacologic restraint were observed for: Ages 5 to 8 years, males, non-Hispanic Black patients, publicly insured, lowest median household income, lowest COI, and stays ≥7 days. (Table 1).

The mental health diagnosis categories with the highest pharmacologic restraint use per 1000 patient days included autism (79.4; 95% confidence interval [CI], 56.2–112.3), substance-related and addictive disorders (45.0; 95% CI, 35.9–56.4), and disruptive disorders (44.8; 95% CI, 25.1–79.8) (Table 2). The overall rate of pharmacologic restraint per 1000 patient days did not significantly change during the study period, with the highest yearly rates occurring in 2016 (16.2 per 1000 patient days) and 2021 (19.7 per 1000 patient days) (Fig 1). The number of pharmacologic restraint days increased by 141% between 2016 and 2021, whereas the number of mental health patient days increased by 138% (Fig 1). Additionally, an increase in mental health patient days was observed for every mental health diagnosis except bipolar and related disorders (Supplemental Table 6).

Trends in the rate of pharmacologic restraint use significantly increased for children admitted with disruptive disorders (RR, 1.4; 95% CI, 1.1–1.6), bipolar and related disorders (RR, 2.0; 95% CI, 1.4–3.0), eating disorders (RR, 2.4; 95% CI, 1.5–3.9), and somatic disorders (RR, 4.2; 95% CI, 1.9–9.1) between 2016 and 2021 (Table 3, Fig 2). During the same time, there was a statistically significant decrease in the rate of pharmacologic restraint use among patients admitted with autism (RR, 0.8; 95% CI, 0.6–1.0) and anxiety disorders (RR, 0.3; 95% CI, 0.2–0.6) (Table 3, Fig 2). Trends for all other diagnoses are shown in Supplemental Fig 4.

This is a national, multicenter study of children hospitalized with primary mental health conditions in medical inpatient units, which characterized the rates and trends in pharmacologic restraint use by diagnosis between 2016 and 2021. In this study, 3% of hospitalizations and 1.3% of patient days involved pharmacologic restraint. The diagnoses with the highest rates of pharmacologic restraint use were autism (7.9% of patient days), substance-related and addictive disorders (4.5%), and disruptive disorders (4.5%). There was a 141% increase in the incidence of pharmacologic restraint use and a 138% increase in mental health patient days during the study period; however, the overall adjusted rate of pharmacologic restraint use remained stable. Trends in rates of pharmacological restraint use varied by diagnosis, with multiple diagnoses showing increasing rates of restraint over time.

A clear definition of pharmacologic restraint is important for studying usage rates and temporal trends. In this study, a narrow definition of pharmacologic restraint was used by including only short-acting, parenteral (IV/IM), antipsychotic medications without other common, nonrestraint indications.^27,33–36  Previous studies using multicenter, billing data of pediatric EDs and inpatient units have used a broader definition of pharmacologic restraint including IV/IM antipsychotics, benzodiazepines, antihistamines, ketamine, and barbiturates with variable inclusion of other medications previously cited as being used for pharmacologic restraint.^11,27,37  The narrow definition used in this study was designed to be as specific as possible for the treatment of acute agitation while minimizing the inclusion of nonrestraint-related medication use as a confounding factor. With this narrow definition, pharmacologic restraint was found in 3.0% of admissions in this study versus 12.6% of hospitalizations, and 3.5% to 4.1% of pediatric ED visits using a broader definition.^11,27,37 

The adjusted rate of total pharmacologic restraint use did not demonstrate a statistically significant change between 2016 and 2021; however, the number of pharmacologic restraint days increased by 141% in the setting of a 138% increase in patient days. This rising trend in mental health patient days was previously recorded in a large national study from 2005 to 2014 showing an increase of 106%.⁴  The escalating rate of rise in mental health patient days has been attributed to an increased severity and frequency of mental health conditions in the setting of the coronavirus disease 2019 pandemic.^9,10  Another major contributing factor is increased duration of inpatient boarding, with boarding times nearly doubling during the pandemic.⁶  Together, these factors may represent the sequelae of a worsening shortage of inpatient and outpatient pediatric psychiatric resources, leading to an increase in patients receiving psychiatric care within acute-care pediatric hospitals.^5,8,38 

The sociodemographic and clinical data examined in this study demonstrated statistically significant differences in hospitalizations with versus without pharmacologic restraint for nearly all characteristics. The characteristics observed to have disproportionately higher rates of pharmacologic restraint use included: 5 to 8 years of age, males, non-Hispanic Black patients, public insurance, lowest median household income ($0–$25 000), lowest COI (1–20), and length of stay ≥7 days. This disproportionate rate of pharmacologic restraint observed in prolonged hospitalizations (≥7 days) is noteworthy, especially in the setting of psychiatric bed shortages and rising boarding durations.^6,8  This may point to increased boarding duration as a contributing factor to the rising incidence of pharmacologic restraint. Regarding sociodemographic characteristics, recent studies of pediatric ED visits have found similar patterns of pharmacologic restraint use with higher rates in male and Black patients.^11,37  The association with public insurance was also found in 1 study¹¹  but absent in the other.³⁷  Additional studies are needed to understand associations between sociodemographic factors and pharmacologic restraint use among hospitalized children. The observations noted in this data set, as well as the evidence from these recent national studies, are suggestive of racial and socioeconomic disparities within pharmacologic restraint use in the United States.^11,37 

Not enough is known about the factors that are associated with increased agitation among children hospitalized with primary mental health diagnoses, although the hospital environment has been associated with agitation in some mental health diagnoses.^17,39  In general, externalizing conditions with impairments to impulse control such as autism, attention-deficit/hyperactivity disorder, disruptive disorders, bipolar disorder, and psychotic disorder have been associated with higher rates of acute agitation, whereas internalizing conditions such as depression, anxiety, and stressor-related disorders have been associated with lower rates.^{11,14,19,27,28}  These findings were consistent with the results of this study. Children with autism had the highest rate of pharmacologic restraint use (7.9% of patient days) despite a significant decrease in the rates of restraint between 2016 to 2021. The elevated rate of restraint could be related to a combination of condition characteristics and sensitivity to the hospital environment.³⁹  Agitation history and documented sensory sensitivities have been shown to be significant risk factors for agitation among children with autism on pediatric medical inpatient units.³⁹  Also among the externalizing conditions, it is important to note the significantly increased rate of pharmacologic restraint use in disruptive disorder (RR 1.4) and bipolar and related disorders (RR 2.0). Although not classically associated with pharmacologic restraint, eating disorders (RR 2.4) and somatic disorders (RR 4.2) also showed significant increases in the rate of pharmacologic restraint use. Because of a lack of literature regarding pharmacologic restraint use within these conditions, it is difficult to infer the reason for increased restraint use in this population. Increased incidence and severity of cooccurring psychiatric conditions could be possible contributors toward these trends; however, both conditions are primarily associated with anxiety and depressive disorders, which have low rates of restraint use.⁴⁰  Overall, a rising trend in the rate of pharmacologic restraint use was noted in 4 conditions within this study. Possible causes are likely multifactorial and may include increased prevalence and severity of mental health conditions, decreased access to outpatient psychiatry care, and increased boarding time in acute-care hospital environments, which may not be designed to manage these conditions and might even exacerbate acute agitation.^1,3,9,10,15 

Literature is limited on interventions to decrease avoidable restraint use within the pediatric medical inpatient setting.^35,41  A decrease in physical restraint use has been demonstrated with interventions such as: Behavioral deescalation protocols, deescalation training for staff, postevent debriefs, electronic medical record order sets, and hospitalist–psychiatrist–nursing collaboration.^42–44  In the adult setting, behavioral emergency response teams have shown evidence of decreased physical restraint use and increased staff satisfaction, but similar studies are lacking in pediatric hospitals.⁴⁵  Behavioral modification programs and increased child psychiatrist staffing have been associated with decreased pharmacologic restraint use in the pediatric inpatient psychiatric setting.¹⁴  How these would translate to the medical inpatient setting is not known, and further research and quality improvement efforts are needed to assess if these or similar interventions could feasibly reduce pharmacologic restraint in the pediatric medical inpatient setting.

This study had several limitations. The study population was identified and categorized on the basis of the primary discharge diagnosis, which is contingent on appropriate coding. The focus on the primary diagnosis also excluded secondary diagnoses, which may be psychiatric in nature and could have affected the rates of pharmacologic restraint use. The PHIS database does not show the indication for the use of a medication, nor confirms if the administration was successful. The definition of pharmacologic restraint used in this study omitted oral medications and parenteral medications previously cited for pharmacologic restraint with broader indications. Thus, it is likely that the rates of pharmacologic restraint use are underestimated. Furthermore, results were based on days with pharmacologic restraint use, and it is unclear how many times a pharmacologic restraint was used on a given day. Physical restraint was not included in this analysis because it is not accurately captured in the PHIS database, but it also represents a significant portion of restraint use among pediatric patients.^17,29,46  Finally, this study focused solely on large, tertiary care children’s hospitals, which may not be generalizable to the community hospital setting.

Pharmacologic restraints are used in 3% of hospitalization and 1.3% of patient days in children admitted with a primary mental health diagnosis. Certain diagnoses had higher rates of restraint usage including autism (7.9% of patient days), substance-related disorders (4.5%), and disruptive disorders (4.5%). Although the overall rate of pharmacological restraint use was not significantly higher between 2016 and 2021, the increase in the number of pharmacologic restraint days by 141% is noteworthy. Trends in the rate of pharmacologic restraint use varied by diagnosis, with a significant increase in the rate of restraint use for multiple mental health conditions. This study excluded oral medications and physical restraints, likely resulting in an underestimation of total restraint usage. These findings can assist practitioners in identifying at-risk patients, guiding targeted interventions, and developing quality improvement initiatives aimed at identifying and reducing factors that impact avoidable pharmacologic restraint use in the pediatric inpatient environment.

CI

confidence interval

COI

Child Opportunity Index

ED

emergency department

IM

intramuscular

IV

intravenous

PHIS

Pediatric Health Information System

RR

rate ratio