Video Abstract
Children hospitalized with a mental health crisis often receive pharmacologic restraint for management of acute agitation. We examined associations between pharmacologic restraint use and race and ethnicity among children admitted for mental health conditions to acute care nonpsychiatric children’s hospitals.
We performed a retrospective cohort study of children (aged 5–≤18 years) admitted for a primary mental health condition from 2018 to 2022 at 41 US children’s hospitals. Pharmacologic restraint use was defined as parenteral administration of medications for acute agitation. The association of race and ethnicity and pharmacologic restraint was assessed using generalized linear multivariable mixed models adjusted for clinical and demographic factors. Stratified analyses were performed based on significant interaction analyses between covariates and race and ethnicity.
The cohort included 61 503 hospitalizations. Compared with non-Hispanic Black children, children of non-Hispanic White (adjusted odds ratio [aOR], 0.81; 95% confidence interval [CI], 0.72–0.92), Asian (aOR, 0.82; 95% CI, 0.68–0.99), or other race and ethnicity (aOR, 0.68; 95% CI, 0.57–0.82) were less likely to receive pharmacologic restraint. There was no significant difference with Hispanic children. When stratified by sex, racial/ethnic differences were magnified in males (aORs, 0.49–0.68), except for Hispanic males, and not found in females (aORs, 0.83–0.93). Sensitivity analysis revealed amplified disparities for all racial/ethnic groups, including Hispanic youth (aOR, 0.65; 95% CI, 0.47–0.91).
Non-Hispanic Black children were significantly more likely to receive pharmacologic restraint. More research is needed to understand reasons for these disparities, which may be secondary to implicit bias and systemic and interpersonal racism.
Children in mental health crises are increasingly admitted to children’s hospitals to wait for inpatient psychiatric placement. Acute agitation and behavioral escalation may occur. Little is known about racial and ethnic disparities in the pharmacologic restraint of hospitalized patients.
Non-Hispanic Black children received pharmacologic restraint more often than other racial groups. Implicit bias and systemic racism may contribute to these findings. This work adds to a body of evidence highlighting a need to address disparities in mental health care.
The ongoing pediatric mental health crisis, further exacerbated by the COVID-19 pandemic, prompted the American Academy of Pediatrics, the American Academy of Child and Adolescent Psychiatry, and the Children’s Hospital Association to declare a national state of emergency for children’s mental health.1 Part of this declaration was in response to a 30% increase in children and adolescents presenting to the emergency department (ED) for suspected suicide attempts from 2019 to 2021.2 A significant proportion of children who present to the ED with a primary mental health condition meet criteria for inpatient psychiatric hospitalization. However, if a psychiatric bed is unavailable, children are increasingly admitted to nonpsychiatric acute care children’s hospitals while awaiting transfer to a primary psychiatric hospital (ie, boarding).3,4
While admitted for primary mental and behavioral conditions, acute agitation and behavioral escalation may occur.5 Acute safety concerns, particularly perceived risk of harm to patient or staff, prompt the use of verbal deescalation followed by pharmacologic restraint or physical restraint. Our recent work demonstrated pharmacological restraint was used in 12.6% of encounters of children admitted to a nonpsychiatric acute care hospital for a primary mental health reason.6 Multiple factors, including long boarding times, underlying psychiatric conditions, and a restrictive environment, likely contribute to agitation and behavioral dysregulation in hospitalized patients.5,7,8
Treatment of mental health conditions has a longstanding history of racial and ethnic disparities. Notably, in 2021 the American Psychiatric Association apologized for “enabling discriminatory and prejudicial actions…and racist practices in psychiatric treatment for Black, Indigenous, and People of Color.”9 There are numerous examples of ongoing structural, institutional, interpersonal, and internalized racism experienced by racial and ethnic minority groups in mental health care.10 Even after controlling for demographic and clinical variables, minoritized groups, especially Black patients, are more likely to be involuntarily admitted to a psychiatric facility.11 Non-Hispanic Black children are also more likely to be diagnosed with disruptive behavior disorders, whereas non-Hispanic white children are more likely to be diagnosed with attention-deficit/hyperactivity disorder (ADHD). Providers may be interpreting the same or similar behaviors in children as evidence of a disruptive behavior disorder in non-Hispanic Black children and ADHD in non-Hispanic white children because of unconscious bias.12 This leads to a cycle of negative response to behavioral dysregulation rather than providing needed mental health services.12–15 A label of disruptive disorder can be perpetuated in patient charts and influence provider behavior, including actions that may trigger previous traumatic experiences in these children (ie, physical or pharmacologic restraint).16 Racial and ethnic minority youth access mental health care services less frequently than non-Hispanic white children.14,17 Postulated reasons in the literature include cost,18 stigma,19 and cultural mistrust,20 which are results of negative lived and group experiences with the health care system and other systems of oppression. In the specific instance of acute agitation and behavioral escalation in the pediatric ED, non-Hispanic Black children experiencing mental health crises are more likely to be submitted to both physical and pharmacologic restraint compared with non-Hispanic white children.21,22
Given the known disparities in mental health diagnoses between racial/ethnic groups, we hypothesized similar associations may be evident in the inpatient pediatric setting and that non-Hispanic Black children would be more likely to experience pharmacologic restraint regardless of their diagnostic label. In this study, we examined the association between race and ethnicity and pharmacologic restraint use among children admitted for mental health conditions in acute care, nonpsychiatric children’s hospitals.
Methods
Study Design and Database
We performed a multicenter, retrospective cohort study using the Pediatric Health Information System (PHIS) database of children hospitalized for primary mental health conditions from March 1, 2018, to February 29, 2022, at 41 US children’s hospitals. The PHIS database includes demographic, billing, and resource use data from 49 tertiary-care children’s hospitals in the United States affiliated with the Children’s Hospital Association (Lenexa, Kansas). Eight hospitals were excluded because of missing or incomplete race and ethnicity data, a known reporting issue in administrative databases.23 The study was approved as nonhuman subjects research by our institutional review board.
Study Population
We used the Child and Adolescent Mental Health Disorders Classification System (CAMHD-CS) to identify primary mental health hospitalizations of children aged 5 to ≤18 years.24 This classification system organizes diagnostic codes into 30 groups that correspond to the Diagnostic and Statistical Manual of Mental Disorders, fifth edition (DSM-V), psychiatric diagnosis groups. The CAMHD-CS has been used in prior similar studies.6,25,26 We further subdivided mental health diagnoses by category: autism, anxiety, disruptive, mood, neurocognitive, other (mental health symptoms, sexuality and gender identity disorders, and dissociative disorders), psychotic, somatic, suicide or self-injury, and trauma and stressor (Supplemental Table 5). Patients who died or who were admitted directly to inpatient psychiatric units from the ED were excluded. Encounters with an operating room charge were excluded because it is not possible in PHIS to ascertain if a medication was used for procedural sedation or pharmacologic restraint. For similar reasons, calendar days spent in the ICU were excluded from the analysis, and non‐ICU days of those encounters were included. Electroconvulsive therapy, a potential procedure used in severe depression and other diagnoses and done during a psychiatric inpatient hospital stay when needed, was only performed in 3 of the acute care hospitalizations in this cohort.
Outcome Measures
The primary outcome measure was the use of pharmacologic restraint during an acute care hospitalization encounter, defined as parenteral (intramuscular or intravenous) administration of benzodiazepines, barbiturates, antipsychotics, antihistamines, and others (Table 1). Medications were chosen based on expert opinion, consensus emergency medicine guidelines, and previous studies of pharmacologic restraint given a lack of inpatient pediatric guidelines for pharmacologic restraint use for acute agitation and behavioral escalation.6,27–30 Indications for medications are not reported in PHIS. Therefore, oral medications were not included in our definition of pharmacologic restraint because they could represent a regularly scheduled psychiatric medication or have been given for reasons other than pharmacologic restraint.6 Because some included medications have other indications (ie, nausea, itching), the primary outcome measure was also assessed using a sensitivity analysis with a more specific definition (parenteral haloperidol, olanzapine, and ziprasidone) to reduce potential outcome misclassification.
Parenteral Medications Included in Pharmacologic Restraint by Drug Class and Frequency of Use Among Cohort (n = 7309)
Drug Class . | Drug Name . | N (Frequency of Use (%)) Individual Medication Doses/Total Number of Medication Doses . |
---|---|---|
Benzodiazepines | Lorazepam | 5906 (41.6) |
Diazepam | 156 (1.1) | |
Midazolam | 1759 (12.4) | |
Barbiturates | Phenobarbital | 25 (0.2) |
Methohexital | 7 (0.05) | |
Pentobarbital | 25 (0.2) | |
Antipsychotics | Ziprasidone | 185 (1.3) |
Aripiprazole | 7 (0.05) | |
Haloperidol | 1639 (11.6) | |
Olanzapine | 793 (5.6) | |
Paliperidone | 13 (0.1) | |
Prochlorperazine | 356 (2.5) | |
Chlorpromazine | 483 (3.4) | |
Droperidol | 16 (0.1) | |
Antihistamines | Diphenhydramine | 2190 (15.4) |
Hydroxyzine | 27 (0.2) | |
Promethazine | 173 (1.2) | |
Other | Ketamine | 326 (2.3) |
Benztropine | 107 (0.7) |
Drug Class . | Drug Name . | N (Frequency of Use (%)) Individual Medication Doses/Total Number of Medication Doses . |
---|---|---|
Benzodiazepines | Lorazepam | 5906 (41.6) |
Diazepam | 156 (1.1) | |
Midazolam | 1759 (12.4) | |
Barbiturates | Phenobarbital | 25 (0.2) |
Methohexital | 7 (0.05) | |
Pentobarbital | 25 (0.2) | |
Antipsychotics | Ziprasidone | 185 (1.3) |
Aripiprazole | 7 (0.05) | |
Haloperidol | 1639 (11.6) | |
Olanzapine | 793 (5.6) | |
Paliperidone | 13 (0.1) | |
Prochlorperazine | 356 (2.5) | |
Chlorpromazine | 483 (3.4) | |
Droperidol | 16 (0.1) | |
Antihistamines | Diphenhydramine | 2190 (15.4) |
Hydroxyzine | 27 (0.2) | |
Promethazine | 173 (1.2) | |
Other | Ketamine | 326 (2.3) |
Benztropine | 107 (0.7) |
Primary Exposure: Race and Ethnicity
The primary exposure was a 5-category variable of race and ethnicity: non-Hispanic Black, non-Hispanic white, Hispanic, Asian, and Other (Pacific Islander, Native American, and other unspecified). In this analysis, the variable race and ethnicity identifies a social construct and categorization of people based on shared physical traits that have no correlation with biology. As such, it was to act as a proxy for racism as a means to identify associations of race and ethnicity external to the study population.31–35 Race and ethnicity data are gathered by caregiver report at the time of hospital admission and/or hospital registration assignment then submitted to PHIS from hospitals.
Covariates
Additional demographic variables included age, sex, season, hospital disposition (home, psychiatric facility/other), and social determinants of health including median neighborhood household income quartile, payer (government, private, other), and rurality. Patient‐level variables included DSM‐V diagnosis categories grouped by the CAMHD-CS, complex chronic conditions, ICU stay, length of hospital stay, and Hospitalization Resource Intensity Scores for Kids (H‐RISK). Complex chronic conditions are medical conditions expected to last at least a year, have a high probability of hospitalization, and involve multiple organ systems.36 The H‐RISK is a relative resource-intensity weight given to each hospital encounter based on the level of severity and diagnosis‐related groups.37
Statistical Analysis
The association of race and ethnicity and pharmacologic restraint use was assessed using generalized estimating equations adjusting for age, sex, payer, DSM-V diagnosis, hospital disposition, and H-RISK severity index, while accounting for hospital clustering. Non-Hispanic Black children were used as the reference group because we a priori hypothesized based on literature review and expert opinion that non-Hispanic Black children would be at highest risk of pharmacological restraint with a goal to evaluate the odds of all other racial/ethnic groups compared with them. We also recognize the potential harm (whether intention or unintentional) of uniformly using dominant and/or advantaged groups as reference groups.31,32,35 We tested for interaction effects with race and ethnicity and several covariates (sex, age, payer, period, and DSM-V diagnosis). Additional stratified analyses were conducted for covariates with significant interaction effects (sex). A sensitivity analysis was performed using 3 antipsychotic medications (haloperidol, olanzapine, and ziprasidone), which when administered parenterally are almost exclusively used in instances of acute behavioral dysregulation requiring pharmacologic restraint.
Results
The cohort included 61 503 acute care hospitalizations for a mental health crisis (Table 2). Of those, 7309 (11.9%) included the use of pharmacologic restraint. Non-Hispanic white children represented 54.8% of the cohort compared with 19.8% non-Hispanic Black. Non-Hispanic Black children had the highest use of pharmacological restraint (14.8%) among all racial/ethnic groups. More non-Hispanic Black children had a primary diagnosis of disruptive disorder (9.2%) than other races and ethnicities (2.8% to 4.7%) during their study encounter. The average hospital length of stay for all children was 2 days and 68% were female.
Characteristics of Pediatric Mental Health Admissions to Children’s Hospitals in 2018–2022
. | Overall . | Non-Hispanic White . | Non-Hispanic Black . | Hispanic . | Asian . | Othera . |
---|---|---|---|---|---|---|
Hospitalizations | 61 503 | 33 680 (54.8) | 12 174 (19.8) | 9383 (15.3) | 1245 (2.0) | 5021 (8.1) |
Age, y | ||||||
a. 5–9 | 4197 (6.8) | 2232 (6.6) | 1083 (8.9) | 555 (5.9) | 41 (3.3) | 286 (5.7) |
b. 10–14 | 28 153 (45.8) | 15 269 (45.3) | 5607 (46.1) | 4418 (47.1) | 496 (39.8) | 2363 (47.1) |
c. 15–18 | 29 153 (47.4) | 16 179 (48) | 5484 (45) | 4410 (47) | 708 (56.9) | 2372 (47.2) |
Sex | ||||||
a. Female | 41 980 (68.3) | 23 105 (68.6) | 7802 (64.1) | 6633 (70.7) | 921 (74) | 3519 (70.2) |
b. Male | 19 490 (31.7) | 10 559 (31.4) | 4362 (35.9) | 2748 (29.3) | 324 (26) | 1497 (29.8) |
Payer | ||||||
a. Government | 30 777 (50) | 12 804 (38) | 8620 (70.8) | 6337 (67.5) | 426 (34.2) | 2590 (51.6) |
b. Private | 27 045 (44) | 18 843 (55.9) | 2918 (24) | 2438 (26) | 743 (59.7) | 2103 (41.9) |
c. Other | 3681 (6) | 2033 (6) | 636 (5.2) | 608 (6.5) | 76 (6.1) | 328 (6.5) |
Hospital disposition | ||||||
Home | 32 131 (52.2) | 17 413 (51.7) | 6879 (56.5) | 4573 (48.7) | 598 (48) | 2668 (53.1) |
Psychiatric facility | 29 372 (47.8) | 16 267 (48.3) | 5295 (43.5) | 4810 (51.3) | 647 (52) | 2353 (46.9) |
DSM V categoryb | ||||||
Autism | 1358 (2.2) | 680 (2) | 302 (2.5) | 274 (2.9) | 20 (1.6) | 82 (1.6) |
Anxiety | 2289 (3.7) | 1493 (4.4) | 290 (2.4) | 289 (3.1) | 49 (3.9) | 168 (3.3) |
Disruptive | 3046 (5) | 1382 (4.1) | 1118 (9.2) | 275 (2.9) | 35 (2.8) | 236 (4.7) |
Mood | 11105 (18.1) | 6567 (19.5) | 2137 (17.6) | 1370 (14.6) | 198 (15.9) | 833 (16.6) |
Neurocognitive | 1433 (2.3) | 757 (2.2) | 352 (2.9) | 209 (2.2) | 12 (1) | 103 (2.1) |
Otherc | 3509 (5.7) | 1877 (5.6) | 882 (7.2) | 469 (5) | 50 (4) | 231 (4.6) |
Psychotic | 1243 (2) | 411 (1.2) | 451 (3.7) | 243 (2.6) | 46 (3.7) | 92 (1.8) |
Somatic | 4008 (6.5) | 2290 (6.8) | 819 (6.7) | 542 (5.8) | 77 (6.2) | 280 (5.6) |
Suicide or self-injury | 31 219 (50.8) | 17 044 (50.6) | 5236 (43) | 5421 (57.8) | 724 (58.2) | 2794 (55.6) |
Trauma and stressor | 2293 (3.7) | 1179 (3.5) | 587 (4.8) | 291 (3.1) | 34 (2.7) | 202 (4) |
CCC | ||||||
No | 52 114 (84.7) | 28 333 (84.1) | 10 477 (86.1) | 7932 (84.5) | 1042 (83.7) | 4330 (86.2) |
Yes | 9389 (15.3) | 5347 (15.9) | 1697 (13.9) | 1451 (15.5) | 203 (16.3) | 691 (13.8) |
ICU | ||||||
No | 54 635 (88.8) | 30 033 (89.2) | 11 207 (92.1) | 8075 (86.1) | 1060 (85.1) | 4260 (84.8) |
Yes | 6868 (11.2) | 3647 (10.8) | 967 (7.9) | 1308 (13.9) | 185 (14.9) | 761 (15.2) |
Length of stay | ||||||
Days | 2 [1, 4] | 2 [1, 4] | 2 [1, 4] | 2 [1, 3] | 2 [1, 4] | 2 [1, 4] |
H-RISK | 0.86 (0.44) | 0.87 (0.43) | 0.87 (0.4) | 0.86 (0.49) | 0.83 (0.46) | 0.86 (0.47) |
. | Overall . | Non-Hispanic White . | Non-Hispanic Black . | Hispanic . | Asian . | Othera . |
---|---|---|---|---|---|---|
Hospitalizations | 61 503 | 33 680 (54.8) | 12 174 (19.8) | 9383 (15.3) | 1245 (2.0) | 5021 (8.1) |
Age, y | ||||||
a. 5–9 | 4197 (6.8) | 2232 (6.6) | 1083 (8.9) | 555 (5.9) | 41 (3.3) | 286 (5.7) |
b. 10–14 | 28 153 (45.8) | 15 269 (45.3) | 5607 (46.1) | 4418 (47.1) | 496 (39.8) | 2363 (47.1) |
c. 15–18 | 29 153 (47.4) | 16 179 (48) | 5484 (45) | 4410 (47) | 708 (56.9) | 2372 (47.2) |
Sex | ||||||
a. Female | 41 980 (68.3) | 23 105 (68.6) | 7802 (64.1) | 6633 (70.7) | 921 (74) | 3519 (70.2) |
b. Male | 19 490 (31.7) | 10 559 (31.4) | 4362 (35.9) | 2748 (29.3) | 324 (26) | 1497 (29.8) |
Payer | ||||||
a. Government | 30 777 (50) | 12 804 (38) | 8620 (70.8) | 6337 (67.5) | 426 (34.2) | 2590 (51.6) |
b. Private | 27 045 (44) | 18 843 (55.9) | 2918 (24) | 2438 (26) | 743 (59.7) | 2103 (41.9) |
c. Other | 3681 (6) | 2033 (6) | 636 (5.2) | 608 (6.5) | 76 (6.1) | 328 (6.5) |
Hospital disposition | ||||||
Home | 32 131 (52.2) | 17 413 (51.7) | 6879 (56.5) | 4573 (48.7) | 598 (48) | 2668 (53.1) |
Psychiatric facility | 29 372 (47.8) | 16 267 (48.3) | 5295 (43.5) | 4810 (51.3) | 647 (52) | 2353 (46.9) |
DSM V categoryb | ||||||
Autism | 1358 (2.2) | 680 (2) | 302 (2.5) | 274 (2.9) | 20 (1.6) | 82 (1.6) |
Anxiety | 2289 (3.7) | 1493 (4.4) | 290 (2.4) | 289 (3.1) | 49 (3.9) | 168 (3.3) |
Disruptive | 3046 (5) | 1382 (4.1) | 1118 (9.2) | 275 (2.9) | 35 (2.8) | 236 (4.7) |
Mood | 11105 (18.1) | 6567 (19.5) | 2137 (17.6) | 1370 (14.6) | 198 (15.9) | 833 (16.6) |
Neurocognitive | 1433 (2.3) | 757 (2.2) | 352 (2.9) | 209 (2.2) | 12 (1) | 103 (2.1) |
Otherc | 3509 (5.7) | 1877 (5.6) | 882 (7.2) | 469 (5) | 50 (4) | 231 (4.6) |
Psychotic | 1243 (2) | 411 (1.2) | 451 (3.7) | 243 (2.6) | 46 (3.7) | 92 (1.8) |
Somatic | 4008 (6.5) | 2290 (6.8) | 819 (6.7) | 542 (5.8) | 77 (6.2) | 280 (5.6) |
Suicide or self-injury | 31 219 (50.8) | 17 044 (50.6) | 5236 (43) | 5421 (57.8) | 724 (58.2) | 2794 (55.6) |
Trauma and stressor | 2293 (3.7) | 1179 (3.5) | 587 (4.8) | 291 (3.1) | 34 (2.7) | 202 (4) |
CCC | ||||||
No | 52 114 (84.7) | 28 333 (84.1) | 10 477 (86.1) | 7932 (84.5) | 1042 (83.7) | 4330 (86.2) |
Yes | 9389 (15.3) | 5347 (15.9) | 1697 (13.9) | 1451 (15.5) | 203 (16.3) | 691 (13.8) |
ICU | ||||||
No | 54 635 (88.8) | 30 033 (89.2) | 11 207 (92.1) | 8075 (86.1) | 1060 (85.1) | 4260 (84.8) |
Yes | 6868 (11.2) | 3647 (10.8) | 967 (7.9) | 1308 (13.9) | 185 (14.9) | 761 (15.2) |
Length of stay | ||||||
Days | 2 [1, 4] | 2 [1, 4] | 2 [1, 4] | 2 [1, 3] | 2 [1, 4] | 2 [1, 4] |
H-RISK | 0.86 (0.44) | 0.87 (0.43) | 0.87 (0.4) | 0.86 (0.49) | 0.83 (0.46) | 0.86 (0.47) |
Data presented as N (%) except length of stay is median (interquartile range). CCC, complex chronic conditions; DSM-V, Diagnostic and Statistical Manual of Mental Disorders, fifth edition; H-RISK, Hospitalization Resource Intensity Scores for Kids.
Other includes Pacific Islander, Native American, and other unspecified.
Mental health discharge diagnoses were identified using the Child and Adolescent Mental Health Disorders Classification System.
Other includes mental health symptoms, sexuality and gender identity disorders, and dissociative disorders.
Compared with non-Hispanic Black children, non-Hispanic white children (adjusted odds ratio [aOR], 0.81; 95% confidence interval [CI], 0.72–0.92), Asian children (aOR, 0.82; 95% CI, 0.68–0.99), and other race and ethnicity children (aOR, 0.68; 95% CI, 0.57–0.82) were significantly less likely to receive pharmacologic restraint, whereas there was no statistically significant difference observed for Hispanic children (aOR, 0.87; 95% CI, 0.73–1.03). When stratified by sex, compared with non-Hispanic Black males, children of non-Hispanic white (aOR, 0.68; 95% CI, 0.59–0.77), Asian (aOR, 0.68; 95% CI, 0.47–0.98), and other race and ethnicity (aOR, 0.49; 95% CI, 0.37–0.65) were significantly less likely to receive pharmacologic restraint. Findings were not significant for Hispanic males (aOR, 0.89; 95% CI, 0.72–1.11). Associations were attenuated and nonsignificant in females (Table 3). When limiting to a more specific pharmacological restraint definition, we observed a significantly amplified disparity in restraint use for all racial/ethnic groups, including Hispanic youth (aOR, 0.65; 95% CI, 0.47–0.91), compared with non-Hispanic Black youth (Table 4).
Pharmacologic Restraint Use by Race and Ethnicity
. | Stratified by Sex . | |||
---|---|---|---|---|
. | Overall Pharmacologic Restraint Use Frequencya . | Overall aORb (95% CI) . | Male aORc (95% CI) . | Female aORc (95% CI) . |
Non-Hispanic Black | 14.8% | Reference | Reference | Reference |
Non-Hispanic white | 11.1% | 0.81 (0.72–0.92) | 0.68 (0.59–0.77) | 0.93 (0.8–1.08) |
Hispanic | 12.3% | 0.87 (0.73–1.03) | 0.89 (0.72–1.11) | 0.88 (0.73–1.05) |
Asian | 11.6% | 0.82 (0.68–0.99) | 0.68 (0.47–0.98) | 0.93 (0.73–1.19) |
Other | 9.5% | 0.68 (0.57–0.82) | 0.49 (0.37–0.65) | 0.83 (0.68–1.01) |
. | Stratified by Sex . | |||
---|---|---|---|---|
. | Overall Pharmacologic Restraint Use Frequencya . | Overall aORb (95% CI) . | Male aORc (95% CI) . | Female aORc (95% CI) . |
Non-Hispanic Black | 14.8% | Reference | Reference | Reference |
Non-Hispanic white | 11.1% | 0.81 (0.72–0.92) | 0.68 (0.59–0.77) | 0.93 (0.8–1.08) |
Hispanic | 12.3% | 0.87 (0.73–1.03) | 0.89 (0.72–1.11) | 0.88 (0.73–1.05) |
Asian | 11.6% | 0.82 (0.68–0.99) | 0.68 (0.47–0.98) | 0.93 (0.73–1.19) |
Other | 9.5% | 0.68 (0.57–0.82) | 0.49 (0.37–0.65) | 0.83 (0.68–1.01) |
aOR, adjusted odds ratio; CCC, complex chronic conditions; CI, confidence interval; DSM-V, Diagnostic and Statistical Manual of Mental Disorders, fifth edition; H-RISK, Hospitalization Resource Intensity Scores for Kids.
Use of pharmacologic restraint is defined as the administration of any included parenteral medication during hospitalization. Medications included benzodiazepines (lorazepam, diazepam, midazolam), barbiturates (phenobarbital, pentobarbital), antipsychotics (ziprasidone, aripiprazole, haloperidol, olanzapine, paliperidone, prochlorperazine, chlorpromazine, risperidone), antihistamines (diphenhydramine, hydroxyzine, promethazine), and other (ketamine, benztropine).
Adjusted for age, sex, payer, DSM-V category, CCC, hospital disposition, and H-RISK.
Adjusted for age, payer, DSM-V category, CCC, hospital disposition, and H-RISK.
Pharmacologic Restraint Use by Race and Ethnicity Sensitivity Analysis
. | Sensitivity Analysisa aORb (95% CI) . |
---|---|
Non-Hispanic Black | Reference |
Non-Hispanic white | 0.64 (0.54–0.76) |
Hispanic | 0.65 (0.47–0.91) |
Asian | 0.53 (0.34–0.82) |
Other | 0.55 (0.42–0.71) |
. | Sensitivity Analysisa aORb (95% CI) . |
---|---|
Non-Hispanic Black | Reference |
Non-Hispanic white | 0.64 (0.54–0.76) |
Hispanic | 0.65 (0.47–0.91) |
Asian | 0.53 (0.34–0.82) |
Other | 0.55 (0.42–0.71) |
aOR, adjusted odds ratio; CCC, complex chronic conditions; CI, confidence interval; DSM-V, Diagnostic and Statistical Manual of Mental Disorders, fifth edition; H-RISK, Hospitalization Resource Intensity Scores for Kids.
Use of pharmacologic restraint only included haloperidol, olanzapine, and ziprasidone.
Adjusted for age, sex, payor, DSM-V category, CCC, hospital disposition, and H-RISK.
Discussion
In this study of pharmacologic restraint use in US nonpsychiatric children’s hospitals, non-Hispanic white, Asian, and children of other races and ethnicities (with the exception of Hispanic youth) had significantly decreased odds of pharmacologic restraint use compared with non-Hispanic Black children, especially males. In other words, non-Hispanic Black youth, particularly boys and young men, are more likely to receive pharmacologic restraint than other racial/ethnic groups when admitted for mental health diagnoses. Stronger associations were observed in a sensitivity analysis when using a more specific pharmacologic restraint definition and were significant across all examined races and ethnicities, including Hispanic youth.
Results from our study align with recent findings of increased physical restraint use (80% increased odds) and pharmacological restraint use (22% increased odds) among non-Hispanic Black children presenting to the pediatric ED.21,22 We adjusted for several important factors to avoid confounding of results, including DSM-V diagnosis grouping, hospital length of stay, and hospital disposition. When considered through a racial disparity lens, our work may be explained in the context of the 4 levels of racism: interpersonal (between people), internalized (within people), structural (among institutions and society), and institutional (within institutions and systems).10
Our findings suggest that Black children experiencing a mental health crisis, especially Black boys, are more often perceived to present an imminent risk of harm to self or staff and are, therefore, more likely to be administered pharmacologic restraint compared with children in crisis from other racial/ethnic groups. Studies have found that seeing the faces of Black people, particularly Black males, facilitate the identification of dangerous stimuli.38 This held true in Black boys as young as age 5 years.39 Adultification bias, another form of interpersonal racism in which Black children are perceived as older than their true age, may also be contributing.40 This may lead to health care worker expectations of Black children that are not developmentally appropriate given the perception that they are older or more mature than their actual age. We posit that implicit bias and interpersonal racism in which Black males are perceived as threatening and older than their actual age rather than experiencing fear or frustration is contributing to increased pharmacologic restraint use. The interplay of these well-documented examples of interpersonal racism with the effects of structural and institutional racism which contribute to Black children entering the health care system in a state of higher acuity are potential explanations for the disparities identified in this and similar studies. Although our cohort demonstrates Black children have higher rates of hospitalization because of disruptive disorders (eg, conduct disorder, impulse control disorder, intermittent explosive disorder), it should be noted that this differential diagnostic categorization by race has been widely described in the literature.13,14 Additionally, these discrepancies may be explained by provider bias.12,15,16 This may create a feedback loop in which initial provider biases (the diagnosis/label itself) lead to a higher likelihood of restraint use later in care as patients now carry a label such as conduct disorder or oppositional defiant disorder that would make future providers more likely to perceive their behavior as threatening. Although validated standardized behavior escalation scores exist, they are not yet widely used in children boarding on inpatient medical units.41 Thus, identification of an escalated behavioral health patient is particularly vulnerable to bias in health care workers.42
Structural and institutional racism may also contribute to disparities in pharmacologic restraint use. Structural racism is linked to poverty/social disadvantage and poverty is strongly linked to adverse childhood experiences.43,44 Racial and ethnic minority youth, despite experiencing a higher frequency of trauma/adverse childhood experiences, exhibit lower rates of utilization when it comes to medical and mental health services.45 In turn, adverse childhood experiences are known to have a direct association with psychologic distress, potentially a reason for crisis mental health care among Black children.46 Generations of oppression, abuse, and disparate treatment of Black Americans within the US health care system has also led to distrust of mental health services. Inequitable utilization of outpatient mental health services secondary to cost, stigma, cultural mistrust, and other factors may also contribute to Black children seeking crisis mental health care acutely in emergency departments at significantly higher rates than their peers of other races/ethnicities.47
Furthermore, institutional racism in health care and other systems mimics parts of society. An example of a system external to health care influencing disparate treatment and outcomes of Black Americans is the criminal justice system. People with severe mental illness are 12 times more likely to experience police use of force than those without severe mental illness, and non-Hispanic Black Americans are 3 times more likely to experience police use of force than non-Hispanic white Americans.48 Khatri et al hypothesize that these elements of institutional racism demonstrated by the police response to Black citizens experiencing a mental health crisis continue within many hospitals, even in the setting of different personnel, training, and goals of care.49 Thus, institutional racism outside the hospital may inform both the hospital staff and the patient expectations regarding hierarchy and authority and the resulting behaviors inside hospital walls that perpetuate these disparities.
Documenting racial disparities in mental health care is a first step to developing solutions. Given the increasing numbers of children in mental health crisis who are boarding on the inpatient wards, our findings are of particular importance to tailor solutions to the inpatient nonpsychiatric setting. Managing acute behavioral dysregulation in the hospital is a critical aspect of care for children experiencing a mental health crisis. At the interpersonal level, training in antiracism, implicit bias, cultural competency, and deescalation should be incorporated into provider on-boarding and continuing education. Evidence for the effectiveness of implicit bias training is mixed. However, mandated training strengthens and prioritizes the importance of equity in health care.50 At the institutional level, standardized behavior escalation scoring, standardized behavioral escalation pharmacologic management, and use of an experienced behavioral health emergency response team could help guide the equitable use of pharmacologic restraint. Standardized protocols implemented in the hospital setting have been found to be effective in mitigating disparities in other conditions.51 Acknowledging the trauma of lived and inherited experiences that disproportionately affect minority youth will also be important. For example, children’s hospitals should consider universal trauma-informed practices and explore safe alternatives to using police as a part of behavioral escalation response in children’s hospitals.16,45 Upstream solutions at the structural level should include policy changes to reduce poverty, increase housing and food security, and increase the affordability and availability of mental health care access. Downstream solutions such as increases in psychiatric bed availability would limit time spent boarding in hospitals ill-equipped to provide true psychiatric care for children in crisis.
Our study has limitations. First, the PHIS database is primarily composed of tertiary care children’s hospitals and may not be generalizable to community hospitals. Second, there may also be variations in race and ethnicity reporting practices among PHIS hospitals. Although we excluded hospitals with incomplete data, there is no evidence of differences in pharmacologic restraint use between included and excluded hospitals (similar range of 8%–29% vs 11%–26%, respectively). Third, although we adjusted for several covariates, additional confounding such as disease severity within DSM-V diagnosis groups may exist, especially when using an administrative database. Fourth, we were not able to examine use of physical restraint as a next measure or alternative to pharmacologic restraint, which could result in over- or underestimation of pharmacologic restraint. Finally, our definition of pharmacological restraint includes only parenteral medications and omits oral medications, which could underestimate pharmacologic restraint.
Conclusions
In this multicenter retrospective cohort study of 41 children’s hospitals, we demonstrated disparities in pharmacologic restraint use in children hospitalized for mental health conditions with highest odds of use among Black male children. It is imperative that we recognize how interpersonal and structural racism contribute to these findings and implement tangible solutions to combat mental health disparities among our youth, especially as the need for mental health services grows in the pediatric population.
Drs Wolf and Johnson led the overall conceptualization and design of the study, analyzed and interpreted the data, drafted the initial manuscript, and reviewed and revised the manuscript; Dr Hall led the acquisition and analysis of the data and contributed to the conceptualization and design of the study and critical review of the manuscript; Drs Williams, Antoon, Carroll, Gastineau, Ngo, Herndon, and Bell, and Ms Hart contributed to the overall conceptualization and design of the study, analysis and interpretation of data, and critical review of the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
FUNDING: Dr Antoon was supported by the National Institute for Allergy and Infectious Diseases Institutes of Health (K23 AI168496). Dr Carroll was supported by grant number T32HS026122 from the Agency for Healthcare Research and Quality. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health and Agency for Healthcare Research and Quality.
CONFLICT OF INTEREST DISCLOSURES: The authors have indicated they have no potential conflicts of interest to disclose.
- ADHD
attention-deficit/hyperactivity disorder
- aOR
adjusted odds ratio
- CAMHD-CS
Child and Adolescent Mental Health Disorders Classification System
- CI
confidence interval
- DSM-V
Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition
- ED
emergency department
- H-RISK
Hospitalization Resource Intensity Scores for Kids
- PHIS
Pediatric Health Information Systems
Comments