Racial and Ethnic Disparities in Use of Chemical Restraint in the Emergency Department

Dear Academic Emergency Medicine editors,

Agitation is defined as excessive psychomotor activity leading to violent or aggressive behavior and is a prevalent symptom for patients presenting to the emergency department (ED) in behavioral crises.¹ Etiologies can be broad and are often multifactorial, including substance intoxication, decompensated mental illness, metabolic derangements, infection, and intracranial trauma or infection. Agitated individuals often carry conditions that are considered stigmatizing, with 65% having diagnoses of serious mental illnesses and/or substance use disorders.² In addition, the most historically marginalized populations are overrepresented in patients that present to the emergency setting with agitation, including >30% from underrepresented races and ethnicities, 10% with housing insecurity, and 72% from low socioeconomic status.³

In 2012, the American Association of Emergency Psychiatry published the Best Practices in the Evaluation and Treatment of Agitation to address the need for quality guidelines for the treatment of agitation.⁴ In these guidelines, experts recommended applying patient-centered techniques as first-line treatment to manage agitation, including verbal de-escalation and environmental modification.^5,6 In cases where these techniques fail or using them is not possible to prevent harm to the patient and/or staff, physical restraint or chemical restraint via the use of sedatives may then be necessary.

However, recent research has found that these recommendations may be differentially adopted based on the characteristics of the patient presenting with agitation. Specifically, cross-sectional analysis of ED visits revealed significantly elevated odds of physical restraint orders associated with Black race, likely due to both implicit and systemic biases.⁷ Although this association with race/ethnicity has been explored with regard to use of physical restraints, less research has examined if similar associations exist with the use of chemical restraints during agitation events. This is of potential concern, as chemical restraint is associated with increased likelihood of adverse patient outcomes such as cardiac arrhythmia, respiratory depression, and hypotension.⁸ To address this gap in research, this study aimed to explore if patients’ race and ethnicity were differentially associated with use of chemical restraint during an ED visit.

We hypothesized that Black patients would have higher odds of receiving chemical restraint as compared to white patients. We conducted a cross-sectional study of all ED visits occurring between January 2015 and August 2021 for patients greater than 16 years of age to three hospitals within a regional healthcare network in New England using data collected in the electronic health record at the time of encounter. Among the three study sites, one was situated at a large academic hospital in an urban setting and two were in community and suburban settings. We defined our primary outcome of chemical restraint as the presence of a completed order for an intramuscular sedative in a patient’s electronic health record during an ED visit. We included medications that were most commonly used for the purposes of chemical restraint in the emergency setting,⁸ consisting of antipsychotics (haloperidol, droperidol, olanzapine), benzodiazepines (lorazepam, midazolam), ketamine, and diphenhydramine. For the purposes of this model, we considered any treatment with an intramuscular sedative during a visit to be equivalent to receipt of a chemical restraint.

We performed descriptive analysis of the data and used a generalized linear multivariable mixed model with a binary logistic link to evaluate associations between chemical restraint and race/ethnicity, incorporating nesting to accommodate for multiple visits by the same patient. Additional covariates in the model included sex, age, type of primary chief complaint, and history of psychiatric and/or substance use diagnoses as documented in the chart for that visit. Primary chief complaints (chosen from a drop-down list by the triage team at the beginning of an ED visit) were grouped manually by the research team into five categories in accordance with prior work regarding agitation in the ED,³ consisting of medical/non-behavioral, primary or undifferentiated agitation, cognitive/neurologic, mental health, and substance use or intoxication. Analyses were conducted using R software (v. 4.1.3). We examined the interaction between race and study site within our model and did not find any statistical significance. We obtained approval from our institution’s Human Investigation Committee. Informed consent was waived.

A total of 1,163,148 unique ED visits occurred within the study period. Of these visits, 12,491 (1.1%) had associated chemical restraint orders. For those with a chemical restraint during their visit, 59.3% were male, 45.0% were White non-Hispanic, 34.1% were Black non-Hispanic, and 17.6% were Hispanic. In addition, 84.7% of chemically restrained patients had a history of psychiatric or substance use diagnoses.

Within our model, we found race to have a significant association with chemical restraint use (joint p<0.01). Black non-Hispanic individuals had higher odds of chemical restraint than White non-Hispanic individuals, with an adjusted odds ratio (AOR) of 1.43 (95% CI 1.34, 1.51). Those reporting race as “Other/Not Listed or Other” (AOR 1.49, 1.22–1.83) and “Unknown/Refused” (AOR 2.02, 1.57–2.58) also had significantly higher odds of chemical restraint compared to White non-Hispanic race. Black non-Hispanic individuals similarly had significantly higher odds of chemical restraint than White non-Hispanic individuals when the same model was performed but restricted to individual categories of chief complaints (AOR 1.25, 1.13–1.38 for medical/non-behavioral; AOR 1.26, 1.07–1.48 for agitation; AOR 4.50, 2.38–8.48 for cognitive/neuro; 1.83, 1.63–2.06 for mental health; and AOR 1.63, 1.45–1.84 for substance use). In addition, we found significantly higher odds of chemical restraint for Hispanic/Latinx individuals compared to White non-Hispanic individuals for cognitive/neuro (AOR 2.69, 1.35–5.34), mental health (AOR 1.30, 1.13–1.49) and substance use chief complaints (AOR 1.18, 1.03–1.36).

We also found significant associations with sex (p<0.01) and age (p<0.01). Males had higher odds of being chemically restrained than females (AOR 1.33, 1.26–1.40), and those with ages 26–35 (AOR 1.41, 1.29–1.53) and 36–45 (AOR 1.38, 1.26–1.51) had higher odds of chemical restraint than those age 16–25. Chief complaint type (p<0.01) and history of psychiatric and/or substance use diagnoses (p<0.01) were significantly associated with chemical restraint use as well. Those with the complaint of agitation had significantly higher odds of chemical restraint (AOR 22.23, 20.47–24.14) compared with those with medical or non-behavioral complaints. Those with a history of psychiatric and/or substance use diagnoses were at higher odds of chemical restraint (AOR 2.96, 2.78–3.15) compared with those who did not have such diagnoses.

Our study identified a significant association between the receipt of chemical restraint during ED visits and race and ethnicity. Black non-Hispanic individuals were identified as having higher odds of receiving chemical restraint as compared to White non-Hispanic individuals. Evidence suggests that decision-making based on biased heuristics versus a more systematic approach is more likely to occur under stressful situations.⁹ During agitation events, a combination of physical danger along with a need for quick decision-making may lead to significant bias and disparities in use of coercive measures like chemical restraints for disadvantaged populations. Indeed, recent studies have shown that emergency clinicians have frustration and negative attitudes towards individuals with substance use and mental illness,¹⁰ and those of minority race or ethnicity groups are particularly vulnerable to negative outcomes in the ED.¹¹ Those listed as either “Other” or “Unknown” race also had significantly higher odds of chemical restraint compared to White non-Hispanic race, potentially reflecting similar biases against multiracial individuals.

This study has several limitations. It is possible that the medications used to evaluate chemical restraint may have been used in a context other than for agitation events as a direct marker for presence of agitation symptoms did not exist in our health record data. The cross-sectional design of our study may also limit interpretation of causality. Because our analysis was limited to a single state and regional healthcare system, our results might not apply to other institutions or geopolitical areas. Further, although the interaction between study site and race was found to be non-significant, it is possible that differences among the sites may still have an impact on the result of the analysis.

Our findings have important implications, as differential odds of chemical restraint based on race and ethnicity may reflect implicit and systemic biases that are influencing decisions to chemically restrain patients.⁷ These findings also highlight that while best practices for managing patient agitation may have been established, they are not always being applied in the same manner for all patients. Future work can focus on better understanding of the influence of race and ethnicity on medical decision-making during agitation management, as well as potential differences between different chemical sedative agents or visits with multiple doses of agents and associations with race/ethnicity. In addition, the association between race and ethnicity and use of chemical restraints highlights the need to identify interpersonal, institutional, and structural factors which may mediate these disparities. Policymakers and administrators can then establish protocols and interventions targeting those specific factors to reduce bias and to allow decision-making to occur in a more systematic way. Ultimately, this can help to reduce unnecessary use of chemical restraints and the marginalization of disadvantaged individuals.

TABLE 1.

Demographic and Visit Characteristics of Emergency Department Patients by Presence of a Chemical Restraint Order and Adjusted Odds of a Chemical Restraint Order by Variable, January 2015 - August 2021

	Chemical Restraint Use, No. (%)a		Adjusted OR (95% CI)	P valueb
	No N=1,150,657 (98.9%)	Yes N=12,491 (1.1%)
Age			Joint Test p<0.01
16–25	155,566 (13.5)	1,525 (12.2)	Ref	--
26–35	199,345 (17.3)	2,848 (22.8)	1.41 (1.29, 1.53)	<0.01
36–45	164,776 (14.3)	2,412 (19.3)	1.38 (1.26, 1.51)	<0.01
46–55	186,831 (16.2)	2,383 (19.1)	1.06 (0.96, 1.16)	0.26
56–64	156,973 (13.6)	1,503 (12.0)	0.94 (0.85, 1.05)	0.27
65+	287,166 (25.0)	1,820 (14.6)	1.07 (0.97, 1.17)	0.18
Sex			Joint Test p<0.01
Female	620,972 (54.0)	5,083 (40.7)	Ref	--
Male	529,685 (46.0)	7,408 (59.3)	1.33 (1.26, 1.40)	<0.01
Race/Ethnicity			Joint Test p<0.01
White non-Hispanic	569,948 (49.5)	5,616 (45.0)	Ref	--
Hispanic/Latinx	212,455 (18.5)	2,197 (17.6)	1.06 (0.99, 1.14)	0.11
Black non-Hispanic	325,697 (28.3)	4,253 (34.1)	1.43 (1.34, 1.51)	<0.01
Asian non-Hispanic	15,546 (1.4)	94 (0.8)	0.98 (0.77, 1.25)	0.88
APIc non-Hispanic	3,086 (0.3)	38 (0.3)	1.52 (0.91, 2.51)	0.11
Otherd non-Hispanic	16,450 (1.4)	202 (1.6)	1.49 (1.22, 1.83)	<0.01
Unknown/Refused	7,475 (0.7)	91 (0.7)	2.02 (1.57, 2.58)	<0.01
Primary Chief Complaint e			Joint Test p<0.01
Medical or Non-Behavioral	1,019,242 (88.6)	3,481 (27.9)	Ref	--
Agitation	18,311 (1.6)	1,640 (13.1)	22.23 (20.47, 24.14)	<0.01
Cognitive or Neurologic	9,110 (0.8)	136 (1.1)	5.07 (4.19, 6.13)	<0.01
Mental Health	51,299 (4.5)	3,502 (28.0)	13.40 (12.49, 14.37)	<0.01
Substance Use	52,695 (4.6)	3,732 (29.9)	14.41 (13.41, 15.49)	<0.01
History of Psychiatric/Substance Use Diagnoses f			Joint Test p<0.01
No	616,532 (53.6)	1,913 (15.3)	Ref	--
Yes	534,125 (46.4)	10,578 (84.7)	2.96 (2.78, 3.15)	<0.01

^a.Percentages may not add to 100% due to rounding.

^b.P value of <.05 was considered statistically significant.

^c.The “API” group includes “American Indian or Alaska Native”, “Native Hawaiian”, “Native Hawaiian or Other Pacific Islander”, and “Other Pacific Islander”

^d.The “other” group includes “Other/Not Listed” and “Other”

^e.Primary chief complaints were grouped into five categories in accordance with prior work regarding agitation in the ED.³

^f.We mapped diagnoses to the AHRQ Clinical Classification System and manually grouped psychiatric and substance use diagnoses.

Appendix

TABLE 2.

Adjusted Odds of a Chemical Restraint Order by Variable, Stratified by Primary Chief Complaint, January 2015 - August 2021

	Medical/Non-Behavioral N=1,022,723	Agitation N=19,951	Cognitive/ Neuro N=9,219	Mental Health N=54,801	Substance Use N=56,434
	Adjusted OR (95% CI)	Adjusted OR (95% CI)	Adjusted OR (95% CI)	Adjusted OR (95% CI)	Adjusted OR (95% CI)
Age
16–25	Ref	Ref	Ref	Ref	Ref
26–35	1.35*a (1.15,1.58)	1.60* (1.19,2.14)	0.67 (0.27,1.68)	1.66* (1.45,1.90)	1.60* (1.36,1.88)
36–45	1.38* (1.17,1.63)	1.41* (1.04,1.90)	0.60 (0.25,1.45)	1.67* (1.44,1.94)	1.44* (1.22,1.71)
46–55	1.01 (0.93,1.29)	0.65* (0.48,0.89)	0.08* (0.02,0.27)	1.42* (1.21,1.66)	1.23* (1.03,1.47)
56–64	1.04 (0.88,1.23)	0.41* (0.30,0.56)	0.10* (0.03,0.31)	1.369*** (1.14,1.64)	1.14 (0.94,1.38)
65+	1.61* (1.38,1.87)	0.31* (0.23,0.40)	0.17* (0.07,0.44)	1.23 (0.99,1.53)	0.94 (0.70,1.24)
Sex
Female	Ref	Ref	Ref	Ref	Ref
Male	1.41* (1.30,1.53)	1.38* (1.20,1.59)	2.01* (1.30,3.40)	1.26* (1.14,1.39)	0.88* (0.79,0.98)
Race/Ethnicity	Joint Test p<0.01
White non-Hispanic	Ref	Ref	Ref	Ref	Ref
Hispanic/Latinx	0.92 (0.81,1.03)	1.13 (0.91,1.40)	2.69* (1.35,5.34)	1.30* (1.13,1.49)	1.18* (1.03,1.36)
Black non-Hispanic	1.25* (1.13,1.38)	1.26* (1.07,1.48)	4.50* (2.38,8.48)	1.83* (1.63,2.06)	1.63* (1.45,1.84)
Asian non-Hispanic	0.90 (0.59,1.39)	0.68 (0.27,1.75)	0.90 (0.09,9.27)	1.55* (1.05,2.29)	0.62 (0.32,1.20)
APIb non-Hispanic	1.52 (0.75,3.06)	0.99 (0.22,4.36)	-- (0.22,4.36)	3.09* (1.32,7.24)	1.68 (0.62,4.50)
Otherc non-Hispanic	1.34 (0.96,1.88)	1.38 (0.73,2.61)	4.27 (0.96,19.00)	1.82* (1.26,2.63)	1.78* (1.21,2.61)
Unknown/Refused	1.545 (0.98,2.44)	2.75* (1.27,5.96)	2.81 (0.21,37.10)	1.67 (0.97,2.86)	2.11* (1.34,3.33)
History of Psychiatric/Substance Use Diagnoses
No	Ref	Ref	Ref	Ref	Ref
Yes	4.77* (4.36,5.21)	2.55* (2.17,3.01)	5.27* (2.87,9.65)	1.66* (1.40,1.96)	1.54* (1.38,1.73)

^a.* indicated any ORs with P value <.05 and considered statistically significant.

^b.The “API” group includes “American Indian or Alaska Native”, “Native Hawaiian”, “Native Hawaiian or Other Pacific Islander”, and “Other Pacific Islander”

^c.The “other” group includes “Other/Not Listed” and “Other”