a b s t r a c t

Study objective: To investigate the holistic characteristics of patients administered or prescribed opioids to treat pain in the emergency department (ED).

Methods: We used National Hospital Ambulatory Medical Care Survey data for 2018 to examine the administration and prescribing of opioids for pain-related ED visits. Weighted logistic regression models were developed to evaluate the association between opioid administration and prescribing (OAP) in the ED and pa- tients’ pain/severity of conditions, demographic/socioeconomic factors, behavioral factors, contextual factors, and organizational factors. Then, subgroup analyses were conducted by type of pain.

Results: Nearly 55% of the ED visits in 2018 involved pain as a main reason for visiting the ED. The odds of receiv- ing opioids were 45% less in black patients than in white patients when other covariates were adjusted (OR: 0.55; CI: 0.430-0.703). Compared to patients with private insurance, Medicaid beneficiaries and uninsured/self-pay patients had a 45% (OR: 0.55; CI: 0.423-0.706) and 44% (OR: 0.56; CI: 0.386-0.813) lower chance of receiving or being prescribed opioids for a pain-related ED visit when all covariates were adjusted. Other significant predic- tors of OAP for pain in EDs included older age, higher pain level, ED arrival by ambulance, admission to hospital, ED arrival during a night shift, geographic region of the ED. Behavioral factors, such as ED return within 72 h and whether a patient has substance/alcohol abuse or dependence, were not significantly associated with OAP. The subgroup analysis indicated that black patients had lower odds of OAP than their white counterparts only for cer- tain pain categories.

Conclusion: Despite increasing awareness of potential implicit bias in managing pain in the ED, racial disparities in OAP still existed. More education and training on implicit bias would help with reduce the disparities. Also, our study result indicated that non-clinical factors may play a role in emergency physicians’ decision making in OAP. Increased recognition of the variation and systemic efforts to address factors affecting the variability are needed.

(C) 2022

1. Introduction

Acute or chronic pain is one of the most frequent reason patients visit the emergency department (ED) [1]. When used appropriately, opioid analgesics are important for pain treatment. In 2016, opioids were administered in the ED for 53.4 visits per 1000 adults, prescribed at discharge for 38.4 visits per 1000 adults, and both administered and prescribed for 35.4 visits per 1000 adults [2].

Studies have indicated that patients are often first exposed to opi- oids during an ED visit. Being prescribed opioids at ED discharge is re- lated to an increased risk for Opioid abuse and addiction [3,4]. To regulate analgesic opioid administration and prescribing (OAP) in EDs

* Corresponding author at: College of Applied Health Sciences, 2005 Huff Hall, Champaign, IL 61820, USA.

E-mail address: hyokang@illinois.edu (H. Kang).

and prevent opioid abuse and addiction, ED-specific opioid prescribing guidelines have been implemented [5,6]. As a result, opioid prescribing rates at ED discharge have decreased in recent years [7].

While controlling OAP is an important issue, it is also critical to pre- vent the undertreatment of pain or disparities in pain management. Ra- cial disparities in analgesic use, including opioids, in the ED have been reported [8-12]. However, these studies tend to focus on visits with par- ticular types of diagnoses (e.g., long bone fracture [13], abdominal pain [12]) or pain level (severe pain). The evidence for opioid-related racial disparities is mixed for certain types of pain.

Racial disparities in Opioid prescriptions may be attributed to unin- tended biases held by emergency medicine providers. Given the impact of racial/ethnic concordance between patients and providers on their communication and self-reported pain [14,15], physician race may also influence racial disparities in OAP. In addition to demographic

https://doi.org/10.1016/j.ajem.2022.02.043

0735-6757/(C) 2022

and socioeconomic factors [16-18], other patient behavioral characteris- tics, such as history of substance abuse, may affect physicians’ percep- tions and lead to healthcare disparities [19-21]. Also, contextual factors like physician workload may affect emergency physicians’ deci- sions on opioid prescribing [22]. However, the relationship between pa- tient behavioral characteristics and contextual factors with OAP for pain-related ED visits has been understudied.

The objective of this study is to investigate the holistic characteristics of patients administered or prescribed opioids to treat pain in the ED. We hypothesized that race/ethnicity, patient behavioral factors (e.g., substance use disorder, frequent ED visit), and contextual factors (e.g., arrival day and time) are associated with OAP in the ED. We also aim to determine the existence of racial disparities by pain type through subgroup analyses.

1. Methods
   1. Study setting

This study used National Hospital Ambulatory Medical Care Survey data for 2018 to examine the administration and prescribing of opioids for pain-related ED visits. The NHACMS is an annual survey of ambulatory care services at US hospitals, including visits to outpatient departments, EDs, and ambulatory surgery centers, collected by the Na- tional Center for Health Statistics (NCHS) [23]. The survey uses a multi- stage probability sample design to provide a nationally Representative sample of ED visits.

For this study, we focused on adults who were at least 18 years of age and visited the ED for pain. In the dataset, there were three data fields that represent patient’s complaints, named reasons for visit (RFV1 – RFV3). We identified pain-related ED visits if one or more of the RFV fields included the following terms: pain, aches, soreness, tenderness, burns, cramps, and spasms [24]. Then, to conduct subgroup analysis by type of pain, we classified pain-related reasons for visits into eight categories, including abdominal, back, chest, face, head, musculoskele- tal, neck, and other pain that includes genitourinary, pelvic, wound, skin, and unspecified pain.

• 1. Main outcomes

For each pain-related visit, we examined whether a patient received opioids, including administration and prescribing. A visit was classified as OAP if the answer was yes to the question, “Were medications or immu- nizations given at this visit or prescribed at ED discharge?”, and if one of the 30 medication entries (MED 1 – MED 30) included the following nine opioids: hydrocodone, oxycodone, oxymorphone, hydromorphone, morphine, codeine, fentanyl, tramadol, and meperidine [4,25].

Opioid administration/ prescribing

in the ED

Organizational factors

• ED arrival time
• ED arrival day

Contextual factors

• Self-reported pain scale
• Arrival by ambulance
• Admission to hospital

Pain/severity of conditions

• Seen in this ED within 72 hours
• Alcohol/substance abuse/dependence

Behavioral factors

• Age
• Gender
• Race/ethnicity
• Insurance

demographic factors

• 1. Conceptual model

Fig. 1 represents a conceptual model of the factors that can potentially affect patients’ OAP. The model includes pain/severity of conditions, demographic/socioeconomic factors, behavioral factors, contextual factors, and organizational factors. We conceptualize that pain and severity of conditions are main determinants of OAP, and other covariates can modify their effects. In particular, we hypothesize that OAP varies by racial/ethnicity, behavioral, or contextual factors when adjusting for other factors.

The self-reported pain score was recorded on a 0-10 scale. To reflect non-linearity of the pain scale we categorized the score into none (0), mild (1-3), moderate (4-7), severe (8-10), and unknown. Arrival by ambulance and admission to hospital were used as a proxy for severity of conditions, given the assumption that a patient who arrived at the ED by ambulance or were admitted to a hospital after ED discharge tend to have more acute and/or severe conditions that require ongoing inter- vention or continued monitoring. Both factors were coded as a binary variable.

Demographic and socioeconomic factors included age, gender, race/ ethnicity, and insurance. Age was classified into young (18-34 years old), middle-aged (35-54 years old), old (55-74 years old), and oldest (75+ years old) adult groups. Race/ethnicity was grouped into non- Hispanic (NH) white, NH black, Hispanic, and NH other racial minorities. Payment type was classified into private, Medicare, Medicaid including CHIP and other state-based programs, self-pay, and other insurance.

Behavioral factors included if a patient was seen in this ED within the last 72 h and if a patient had alcohol/substance abuse or dependence. mental and behavioral disorders were identified using the following icd10 codes in primary diagnosis: alcohol related disorders (F10); can- nabis related disorders; sedative, hypnotic; or anxiolytic related disor- ders; cocaine related disorders (F15); hallucinogen related disorders (F16); inhalant related disorders (F18); and other psychoactive sub- stance related disorders (F19).

Contextual factors included ED arrival time and day. Arrival time was grouped into morning (7 am – 3 pm), afternoon (3 pm – 11 pm), and night (11 pm – 7 am). Arrival day was represented as day of the week. Organizational factors included the geographic region of EDs (e.g., Northeast, Midwest, South, West) and metropolitan statistical

area (MSA) status (e.g., MSA, non-MSA).

• 1. Statistical analysis

To account for the complex sampling design in the NHAMCS, we conducted all statistical analysis using the Survey suite of programs in Stata version 15.1(StataCorp, College Station, TX). We used sampling

• Geographic region
• metropolitan statistical area status

Fig. 1. Conceptual model.

weights in the analysis when calculating population estimates. More details on the survey methodology can be found from the NCHS [23].

Before performing statistical analyses, we handled missing data in two ways: if a variable included missing data >5% of the entire dataset, we created a new category “blank”, and if missing data was <=5%, we treated them as missing at random and used a listwise deletion ap- proach. We first conducted univariate analysis to describe the charac- teristics of pain-related ED visits. Then, weighted logistic regressions were used to estimate odds ratios associated with individual factors in- cluded in the conceptual model. In the initial model, only factors related to pain/severity of conditions were included. Then, demographic/socio- economic factors, behavioral factors, contextual factors, and organiza- tional factors were incrementally added to the model in order to evaluate changes in the odds ratio for race/ethnicity. Using the full model, the significance of individual covariates associated with OAP was examined. Further subgroup analyses were performed for each pain category using the full model.

This study was exempt from IRB review because the NHAMC data is publicly available.

1. Results

There were 15,827 ED visits by adults in 2018 in the raw sample, which extrapolated to 100,680,659 visits in the weighted sample. Of these visits, nearly 55% involved pain as a main reason for visiting the ED. After applying the exclusion criteria, 8353 observations that extrap- olated to 54,673,807.9 ED visits were used in our study. Table 1 summa- rizes individual and Organizational characteristics of pain-related ED visits. During the study period, 14.17% of the adult patients who visited the ED for any kind of pain received opioids in the ED or at discharge. 6.8% patients reported a pain score of 0 (no pain), 5.7% reported a pain score between 1 and 3 (mild pain), 16.8% reported a pain score between 4 and 6 (moderate pain), 39.9% reported a pain score 7 and above (se- vere pain), and 30.7% did not report a pain score. About 12.4% patients arrived at the ED by ambulance, and 10.8% were admitted to a hospital after being seen in the ED.

The average age of patients who visited the ED for pain was 46 years old, in which 23.6% and 9.3% were in the older adults (55-74 years) and oldest adults (75 + years) groups. Among patients, about 55% were white, 27.6% were black, and 14.1% were Hispanic patients. For insur- ance, the proportion of Medicaid beneficiaries was the highest at 26.8%, followed by Commercial insurance holders (26.5%) and Medicare beneficiaries (20.7%). About 10.3% patients had no insurance.

About 4% of patients had return visits within 72 h of discharge, and

8.6% of patients had substance dependence or abuse including alcohol dependence or abuse. Pain-related patient arrivals tended to be evenly distributed across day of the week, but they varied significantly by time of day, being highest during the morning shift (43.9%) and lowest during the night shift (15.6%).

In terms of geographic location of the EDs, 42.4% included in the study were in the South and 15.4% in the Northeast, with 86.4% of the EDs in metropolitan statistical areas.

Of the 10 types of pain, abdominal pain (24.4%) and Musculoskeletal pain (24.3%) were reported most frequently, followed by chest pain (17.8%), back pain (16.7%), and head pain (10.3%).

Table 2 summarizes the results of the unadjusted and adjusted logis- tic models. As expected, the odds of OAP increased if patients presented more severe conditions. In an unadjusted model, compared to patients who reported no pain, the odds of being given or prescribed for opioids were higher in those who reported moderate pain (OR: 2.1, 95% CI: 1.242-3.677) or severe pain (OR: 3.78; 95% CI: 2.215-6.453). Also, pa-

tients who arrived at the ED by ambulance (OR: 1.75; CI: 1.383-2.206) or who were admitted to a hospital after being discharged from the ED (OR: 3.23; CI: 2.577-4.048) had higer odds of receiving opioids for

Table 1

Characteristics of adult patients with pain-related visits.

Variables N^{^} (%)

Opioids administered or prescribed 7,744,859 (14.17%) Pain scale

No pain (0) 3,726,280 (6.82%)

Mild pain (1-3) 3,106,951 (5.68%)

Moderate pain (4-6) 9,207,584 (16.84%)

Severe pain (7⁺) 21,833,941 (39.93%)

Other 16,799,052 (30.73%)

Hospital admission from the ED

Yes 5,894,753 (10.78%)

No 48,779,055 (89.22%)

Arrival mode

By ambulance 6,766,944 (12.38%)

Other mode 47,906,864 (87.62%)

Age, mean (standard deviation) 46.05 (18.66)

18-34 18,402,448 (33.66%)

35-54 18,253,686 (33.39%)

55-74 12,919,446 (23.63%)

75⁺ 5,098,228 (9.33%)

Sex

Female 32,944,822 (60.26%)

Male 21,728,986 (39.74%)

Race/Ethnicity

NH* White 30,067,344 (54.99%)

NH Black 15,089,829 (27.60%)

Hispanic 7,731,978 (14.14%)

Other 1,784,658 (3.26%)

Insurance

Commercial 14,505,875 (26.53%)

Medicare 11,317,256 (20.70%)

Medicaid 14,641,142 (26.78%)

Self-pay 5,634,630 (10.31%)

Other 8,574,906 (15.68%)

Seen in the ED within 72 h

Yes 2,177,020 (3.98%)

No 47,964,371 (87.73%)

Unknown 4,532,417 (8.29%)

Substance dependence or abuse

Yes 4,703,936 (8.60%)

No 49,969,872 (91.40%)

Arrival day

Sunday 7,055,002 (12.90%)

Monday 9,075,275 (16.60%)

Tuesday 8,407,271 (15.38%)

Wednesday 8,166,082 (14.94%)

Thursday 7,653,111 (14.00%)

Friday 7,365,371 (13.47%)

Saturday 6,951,695 (12.71%)

Arrival time

Morning 23,975,058 (43.85%)

Afternoon 22,165,815 (40.54%)

Night 8,532,936 (15.61%)

Geographic region

Northeast 8,419,330 (15.40%)

Midwest 11,437,493 (20.92%)

South 23,180,775 (42.40%)

West 11,636,210 (21.28%)

Location of the ED

MSA 47,043,188 (86.04%)

Non-MSA 7,630,620 (13.96%)

Pain type

Abdominal pain	13,331,444 (24.38%)
Musculoskeletal pain	13,280,389 (24.29%)
Chest pain	9,718,502 (17.78%)
Back pain	9,115,181 (16.67%)
Head pain	5,637,698 (10.31%)
Neck pain	4,669,822 (8.54%)
Face pain	3,724,901 (6.81%)
Other+ pain	5,930,581(10.85%)

NH: Non-Hispanic.

MSA: Metropolitan Statistical Area.

^{^} Weighted number of observations.

⁺ Other pain includes genitourinary, pelvic, wound, skin, and un- specified pain.

Table 2

Summary of logistic regression model results.

Variable	Odds ratio (95% CI)	AOR+ severity (95% CI)	AOR demographic (95% CI)	AOR full model (95% CI)
Pain scale
No Pain	1.0	1.0	1.0	1.0
Mild	1.64 [0.894, 3.001]	1.72 [0.933, 3.170]	1.62 [0.875, 3.013]	1.67 [0.888, 3.143]
Moderate	2.14? [1.242, 3.677]	2.26? [1.290, 3.952]	2.24? [1.260, 3.993]	2.25? [1.259, 4.022]
Severe	3.78?? [2.215, 6.453]	4.20?? [2.438, 7.235]	4,39?? [2.500, 7.719]	4.40?? [2.497, 7.748]
Other	2.42? [1.284, 4.562]	2.52?? [1.340, 4.727]	2.54?? [1.377, 4.679]	2.52?? [1.375, 4.620]
Admission
Not admitted	1.0	1.0	1.0	1.0
Admitted	3.23?? [2.577, 4.048]	3.16?? [2.466, 4.062]	2.80?? [2.202, 3.564]	2.73?? [2.177, 3.436]
Arrival Info
Not by ambulance	1.0	1.0	1.0	1.0
By ambulance	1.75?? [1.383, 2.206]	1.45? [1.119, 1.891]	1.45? [1.119, 1.879]	1.41? [1.092, 1.824]
Age
18-34	1.0	1.0		1.0
35-54	1.64?? [1.288, 2.084]	1.37? [1.087, 1.732]		1.38? [1.094, 1.742]
55-74	2.21?? [1.771, 2.755]	1.67?? [1.282, 2.177]		1.67?? [1.289, 2.172]
75+	1.71?? [1.260, 2.316]	1.21 [0.786, 1.858]		1.21 [0.790, 1.868]
Sex
Female	1.0	1.0		1.0
Male	1.03 [0.876, 1.217]	0.94 [0.795, 1.108]		0.92 [0.774, 1.086]
Race & ethnicity
NH-white	1.0	1.0		1.0
NH-black	0.52?? [0.407, 0.663]	0.56?? [0.438, 0.705]		0.55?? [0.430, 0.703]
Hispanic	0.77? [0.601, 0.982]	0.86 [0.659, 1.126]		0.85 [0.668,1.082]
Other	0.68 [0.434, 1.053]	0.65 [0.411, 1.025]		0.63? [0.404, 0.986]
Insurance
Commercial	1.0	1.0		1.0
Medicare	0.91 [0.707, 1.179]	0.75 [0.539, 1.043]		0.75 [0.543, 1.032]
Medicaid	0.52?? [0.402, 0.670]	0.56?? [0.434, 0.721]		0.55?? [0.423, 0.706]
Self-pay	0.50?? [0.343, 0.727]	0.57?? [0.398, 0.831]		0.56?? [0.386, 0.813]
Other	0.81 [0.468, 1.388]	0.85 [0.523, 1.391]		0.85 [0.551, 1.297]
Seen in 72 h
No	1.0			1.0
Yes	1.25 [0.906, 1.735]			1.26 [0.897, 1.782]
Other	0.84 [0.517, 1.365]			0.91 [0.606, 1.381]
Alcohol/substance abuse No	1.0	1.0
Yes	1.29 [0.776, 2.146]	1.05 [0.659, 1.676]
Visiting day
Sunday	1.0	1.0
Monday	1.23 [0.933, 1.625]	1.23 [0.908, 1.669]
Tuesday	1.00 [0.711, 1.407]	0.99 [0.690, 1.411]
Wednesday	1.16 [0.851, 1.568]	1.18 [0.837, 1.658]
Thursday	1.08 [0.803, 1.447]	1.07 [0.768, 1.478]
Friday	1.23 [0.901, 1.682]	1.22 [0.871, 1.695]
Saturday	1.12 [0.850, 1.470]	1.11 [0.843, 1.473]
Arrival time
Morning	1.0	1.0
Afternoon	0.95 [0.769, 1.173]	0.99 [0.795, 1.233]
Night	1.40?? [1.150, 1.696]	1.39?? [1.126, 1.727]
Region
Northeast	1.0	1.0
Midwest	1.34 [0.906, 1.977]	1.41 [0.951, 2.076]
South	1.33 [0.897, 1.983]	1.43 [0.989, 2.061]
West	1.58? [1.034, 2.425]	1.52? [1.032, 2.246]
MSA Urban	1.0	1.0
Rural	0.83 [0.547, 1.265]	0.81 [0.529, 1.231]

AOR: Adjusted Odds Ratio; MSA: Metropolitan Statistical Area.

* <0.05.

?? <0.005.

pain than those who arrived at the ED by other modes and those who were not admitted to a hospital after an ED visit. When all covariates were adjusted, these factors were still significantly associated with OAP in the ED.

In an unadjusted model, NH black patients (OR: 0.52; CI: 0.407-0.663) and Hispanic patients (OR: 0.77; CI: 0.601-0.982) were less likely than white patients to receive opioids in the ED or at dis- charge when their main reason for visit was pain. Compared to white patients, other minority group (OR: 0.68; CI: 0.434-1.053) patients were also less likely to receive opioids, but the difference was not statis- tically significant. After adjusting for pain scale and severity of

conditions, the disparities in receiving opioids for pain still existed for black patients (OR: 0.56; CI: 0.438-0.705). In a full model in which pain/severity levels, demographic, behavioral, contextual, and organiza- tional factors were adjusted, the odds of receiving opioids were 45% less in black patients than in white patients (OR: 0.55; CI: 0.430-0.703).

In a full model, in addition to race/ethnicity, age and insurance were also significant predictors of OAP for pain in EDs. Middle-aged (OR: 1.38; CI: 1.094-1.742) and older adults (OR: 1.67; CI: 1.289-2.172)

had higher odds of receiving opioids than the youngest adults, while oldest adults (OR: 1.21; CI: 0.790-1.868) did not have a significant dif- ference in odds than the youngest adults. Compared to patients with

private insurance, Medicaid beneficiaries (OR: 0.55; CI: 0.423-0.706) and uninsured/self-pay patients (OR: 0.56; CI: 0.386-0.813) had a 45% and 44% lower chance of receiving or being prescribed opioids for a pain-related ED visit when all covariates were adjusted.

In contrast to our initial hypothesis, behavioral factors, such as ED re- turn within 72 h and whether a patient has substance/alcohol abuse or dependence, were not significantly associated with OAP in any of the models. We also evaluated the moderating effect of a history of sub- stance abuse/dependence on racial disparities in the receipt of opioids for pain while controlling for other factors, but no significant modera- tion effect was detected. The moderating effect of repetitive ED visit behavior was not analyzed because of a small sample size.

The day of ED arrival did not influence the odds of OAP when adjusting for other covariates. However, arrival time was significantly associated with the odds of OAP in unadjusted and adjusted models. When all covariates were adjusted, the odds of being given or pre- scribed for opioids were 1.39 higher in patients who visited the ED dur- ing a night shift between 11 pm and 7 am (OR: 1.39; CI: 1.126-1.727) than those who visited the ED during a morning shift between 7 am and 3 pm.

The odds of OAP varied by geographic region of the ED. When other factors were held constant, patients who visited the ED in the West (OR: 1.52; CI: 1.032-2.246) had a 52% higher chance of being administered or prescribed opioids for pain than those in the Northeast. However, MSA status of EDs did not affect the odds of OAP.

For subgroup analysis by pain type, we focused on disparities in OAP between white and black patients only when considering a small sam- ple size for other race groups. The analysis indicated that black patients had lower odds of OAP than their white counterparts only for certain pain categories after adjusting for other covariates, which included musculoskeletal pain (OR: 0.61; CI: 0.415-0.882), back pain (OR: 0.32; CI: 0.190-0.556), chest pain (OR: 0.55; CI: 0.329-0.921), and neck

pain (OR: 0.25; CI: 0.090-0.694). The difference in OAP was not signifi- cant between the two race groups for abdominal pain (OR: 0.69; CI: 0.445-1.082), head pain (OR: 0.45; CI: 0.194-1.028), face pain (OR:

1.13; CI: 0.522-2.426), and other pain (OR: 0.83; CI: 0.427-1.610).

1. Discussion

Using the most recent NHAMCS data, this study determined inde- pendent predictors of OAP in the ED or at discharge for adult patients who visited the ED for pain-related issues. We found black patients were less likely to be administered or prescribed for opioids for equiva- lent levels of pain and severity of conditions compared to white pa- tients. These racial/Ethnic disparities did not disappear when other demographic factors and covariates were adjusted. However, the result changed when focusing on a specific type of pain. Among the top four most frequent pain types that accounted for more than 74% of pain- related visits, racial/ethnic disparities still existed for musculoskeletal, back, and chest pain, while the difference was not statistically signifi- cant for abdominal pain. To answer why there are disparities only in cer- tain pain types, further analysis on a patient’s medical history and other non-opioid treatments such as Non-steroidal anti-inflammatory drugs should be performed.

Previous studies using NHAMCS data showed similar findings on ra-

cial disparities in OAP, while the scope of the studies varied in terms of patient age and type of pain. Pletcher et al. (2008) [9] focused on pain- related ED visits using 1993-2005 data, finding black, Hispanic, Asian/ other patients were less likely to receive an opioid than white patients over the study period. In our study, racial disparities in OAP for Hispanic, Asian, and other minority groups were not significant. The different re- sult may be because our study only included adults patients while Pletcher et al. included all age groups. Also, it is possible disparities have been reduced for some racial/ethnic minority groups over time, and therefore the difference is no longer statistically significant. Joynt et al. (2013) [11] that used 2006-2009 data to focus on patients who

presented moderate to severe pain. They indicated that black patients were prescribed opioids less frequently than white patients, indepen- dent of their socioeconomic status. Using 2006-2010 data, Shah et al. (2015) [12] analyzed Opioid prescriptions for primary diagnosis of acute abdominal pain and concluded that NH black patients were less likely to receive opioids than their white counterparts.

Some studies did not detect any racial disparities in OAP in the ED. Using 2010-2014 NHAMCS data, Rosenbloom et al.(2019) [10] investi- gated opioids administration for appendicitis or Gallbladder disease for 12-55 years of age and did not detect differences in opioid administra- tion between white and non-white groups. Other studies [26,27] that used single hospital data also focused on specific types of pain such as back pain, abdominal pain, and musculoskeletal pain and did not find a difference in OAP between white and non-white patients in the ED. In our study, differences in OAP between black and white patients were detected for back and musculoskeletal pain. The difference be- tween the single hospital studies and our study may be due to different patient population and size.

A survey study that investigated important factors affecting deci- sions of emergency medicine providers on prescribing opioids identi- fied patient’s history of substance abuse or dependence as one of the highest rated factors [28]. Emergency medicine providers are recom- mended to use multimodal opioid-sparing strategies for patients with a prior Substance abuse history or behavioral health disorders to pre- vent adverse outcomes because they are risk factors for opioid overdose and Opioid use disorder [29-31]. Explicit and implicit prejudice of health care workers toward people with Substance use disorders and discrim- ination in pain management have been also reported [19-21,32,33]. Fre- quent ED users tend to present pain-related complaints and often have substance abuse disorders [34-36]. Like a patient’s history of substance dependence/use disorders, frequent ED visits may play an influencing role in emergency medicine providers’ unconscious bias and decision to administer or prescribe opioids for pain control. We examined an in- dependent association of these behavioral factors with OAP as well as the moderating effect of substance use dependence/disorder with race/ethnicity on OAP, but they were not a significant determinant of OAP. The results may be attributed to increasing awareness of potential implicit bias in managing pain in the ED. The insignificance of the be- havioral factors might be also due to the relatively small number of pa- tients with those conditions in the dataset. Qualitative research that involves patient interviews and surveys may help better evaluate the existence of disparities related to patients’ behavioral factors. Also, more education and training on implicit bias particularly in the region of pain tolerance and treating patients complaining pain with certain medications (e.g., opioids vs. non-opioids) would help with changing OAP in the ED.

Some studies found variations in opioid prescribing patterns by ED care providers. In a single hospital study, the number of opioid prescrip- tions and the mean number of pills per prescription significantly varied by individual providers even when adjusting for the number of patients seen [37]. A study that focused on ED visits by Medicare beneficiaries also indicated substantial variation in the rate of opioid prescribing among emergency physicians within the same ED [38]. We could not evaluate the variation in OAP by emergency medicine providers at the national level because of the lack of information in the data about the prescribing patterns of individual emergency medicine providers. In- stead, we considered arrival date and time as one of the contextual fac- tors that may influence provider workload [39,40]. Our study indicated that patients who arrived at the ED during the morning shift time (7 am to 3 pm) were about 39% less likely to receive opioids compared to those who arrived during the night shift (11 pm and 7 am) when all covariates were adjusted. There are more complex cases such as traumatic injuries during the night time than the day time, which would result in a higher need for pain control [41,42]. This finding may be also associated with ED census, patient population, and provider workload. Night shift phy- sicians may have higher workload than those on the day shift because

ED census still remain high until the night time [43]. A study showed that patients seen by providers whose workload was higher were more likely to receive opioid prescriptions [44]. Similarly, our study im- plies that non-clinical factors such as working environment and time pressure may play a role in emergency medicine providers’ decision making in OAP. Increased recognition of the variation and systemic ef- forts to address factors affecting the variability may help reduce the im- pact of non-clinical factors on OAP in the ED.

In this study, we only included ED visits that pertained to a certain type of pain as the chief complaint. As expected, the proportion of OAP in a severe pain group was highest at 18.6%, followed by moderate and mild pain groups at 11.4% and 9.0%, respectively. However, the rate of OAP in the no pain group (i.e., pain score of 0) was higher than ex- pected at 5.7%. These patients might not have had pain initially but re- ported pain over the course of their stay in the ED, which led to OAP at some point during their visit. OAP in the no pain group may also be due to errors in the pain scale by patients who reported a score or by staff who entered the score in the system [13]. It is also possible opi- oids were administered or prescribed unnecessarily to lower pain pa- tients because patients complained about a certain pain. To avoid overprescribing and mitigating any risk of developing opioid depen- dency or misuse, it may be worth examining cases that involve OAP for patients who reported mild or no pain.

Our study contributed to confirming findings from previous studies in OAP in the ED and identified additional predictors of OAP using the most recent NHAMCS data. However, it also has potential limitations. First, several shortcomings of the NHAMCS database, such as possible errors in data collection and coding, have been reported [45,46]. Cases that had pain for the main reason for visit but a pain scale of 0 may be an example of coding errors. The NHAMCS also does not provide infor- mation about the exact number of opioids given or prescribed per visit. To better examine the variability in OAP by patient characteristics or providers, more detailed information is needed. Furthermore, the data- base does not include the socioeconomic characteristics of patients, which may be significantly associated with patients’ health, ED visit pat- terns, and access to other health systems. Socioeconomic information could be linked to the NHAMCS if more detailed information about the ED location like county is provided. However, it only includes the aggre- gated geographic location of the ED a patient visited (e.g., Northeast, Midwest, South, West). In addition, our study was unable to distinguish acute pain from chronic pain because the data did not have sufficient in- formation to determine it (e.g., the time from onset of pain). Patients with chronic non-cancer pain may be already on opioid therapy. Given that racial disparities in prescribing opioids for chronic pain exist [47,48], it is important to consider the duration of pain. Future re- search that uses electronic health records with the duration of pain (acute vs. chronic) or longitudinal data with pain treatment histories for patients would determine the degree of racial disparities in OAP in the ED more accurately. Finally, it is possible that our analysis missed cases if RFVs did not include one of the pain-related terms we consid- ered. We assumed that a patient in acute pain (e.g., traumatic injuries) would complain about pain, but it is possible that only specific sites of pain or injuries were recorded for RFVs. Given that traumatic injuries are more likely in minority groups [49], future research that consider both patient complaint (i.e.,RFV) and diagnoses will lead to a stronger evidence in racial/ethnic disparities in OAP. Despite some limitations of the NHAMCS, the database has been widely used to study the provi- sion and use of emergency care because it provides a national sample of visits to the ED, which helps generalize study findings and understand overall trends.

In conclusion, racial/ethnic disparities in OAP persist in EDs for cer-

tain types of pain. Closer investigations are required to identify the causes of these disparities. Also, systemic interventions that can reduce variation in OAP should be implemented and ensure equity in pain management independent of patient and organizational characteristics as well as other situational factors.

Funding

This research did not receive any specific grant from funding agen- cies in the public, commercial, or not-for-profit sectors.

Source of support

N/A.

Credit authorship contribution statement

Hyojung Kang: Formal analysis, Methodology, Project administra- tion, Supervision, Writing – original draft, Writing – review & editing. Peng Zhang: Writing – original draft, Methodology, Formal analysis. Seokgi Lee: Writing – review & editing, Conceptualization. Sa Shen: Writing – review & editing, Software, Methodology. Eleanor Dunham: Writing – original draft, Supervision, Conceptualization.

Declaration of Competing Interest

The authors have no conflicts of interest to report.

References

1. Weiss AJ, Wier LM, Stocks C, Blanchard J. Overview of emergency department visits in the United States, 2011: statistical brief #174. Agency for Healthcare Research and Quality (US); 2006. http://www.ncbi.nlm.nih.gov/pubmed/25144109 Accessed May 7, 2019.
2. Rui P, Schappert S. Opioids prescribed at discharge or given during emergency department visits among adults in the United States, 2016 Hyattsville, MD; 2019.
3. Butler MM, Ancona RM, Beauchamp GA, et al. Emergency department Prescription opioids as an initial exposure preceding addiction. Ann Emerg Med. 2016;68(2): 202-8. https://doi.org/10.1016/J.ANNEMERGMED.2015.11.033.
4. Hoppe JA, Nelson LS, Perrone J, et al. Opioid prescribing in a cross section of US emer- gency departments. Ann Emerg Med. 2015;66(3):253-259.e1. https://doi.org/10. 1016/j.annemergmed.2015.03.026.
5. Broida RI, Gronowski T, Kalnow AF, Little AG, Lloyd CM. State emergency depart- ment opioid guidelines: current status. West J Emerg Med. 2017;18(3):340-4. https://doi.org/10.5811/westjem.2016.12.30854.
6. del Portal DA, Healy ME, Satz WA, McNamara RM. Impact of an opioid prescribing guideline in the acute care setting. J Emerg Med. 2016;50(1):21-7. https://doi.org/ 10.1016/J.JEMERMED.2015.06.014.
7. Rui P, Santo L, Ashman JJ. Trends in opioids prescribed at discharge from emergency departments among adults: United States, 2006-2017. Hyattsville, MD. https:// www.cdc.gov/nchs/data/nhsr/nhsr135-508.pdf; 2020.
8. Quazi S, Eberhart M, Jacoby J, Heller M. Are racial disparities in ED analgesia improv- ing? Evidence from a national database. Am J Emerg Med. 2008;26(4):462-4. https://doi.org/10.1016/j.ajem.2007.05.007.
9. Pletcher MJ, Kertesz SG, Kohn MA, Gonzales R. Trends in opioid prescribing by race/ ethnicity for patients seeking care in US emergency departments. JAMA. 2008;299

(1):70-8. https://doi.org/10.1001/jama.2007.64.

1. Rosenbloom JM, Burns SM, Kim E, August DA, Ortiz VE, Houle TT. Race/ethnicity and sex and opioid Administration in the Emergency Room. Anesth Analg. 2019;128(5): 1005-12. https://doi.org/10.1213/ANE.0000000000003517.
2. Joynt M, Train MK, Robbins BW, Halterman JS, Caiola E, Fortuna RJ. The impact of neighborhood socioeconomic status and race on the prescribing of opioids in emer- gency departments throughout the United States. J Gen Intern Med. 2013;28(12): 1604-10. https://doi.org/10.1007/s11606-013-2516-z.
3. Shah AA, Zogg CK, Zafar SN, et al. Analgesic access for acute abdominal pain in the emergency department among racial/ethnic minority patients: A Nationwide examina- tion. Med Care. 2015;53(12):1000-9. https://doi.org/10.1097/MLR.0000000000000444.
4. Ritsema TS, Kelen GD, Pronovost PJ, Pham JC. The national trend in quality of emer- gency department pain management for long bone fractures. Acad Emerg Med. 2007;14(2):163-9. https://doi.org/10.1197/j.aem.2006.08.015.
5. Takeshita J, Wang S, Loren AW, et al. Association of racial/ethnic and gender concor- dance between patients and physicians with patient experience ratings. JAMA Netw Open. 2020;3(11):e2024583. https://doi.org/10.1001/JAMANETWORKOPEN.2020. 24583.
6. Anderson SR, Gianola M, Perry JM, Reynolds Losin EA. Clinician-patient racial/ethnic concordance influences racial/ethnic minority pain: evidence from simulated clinical interactions. Pain Med Off J Am Acad Pain Med. 2020;21(11):3109. https://doi.org/ 10.1093/PM/PNAA258.
7. Centers for Disease Control & Prevention (CDC). CDC health disparities and inequal- ities report (CHDIR). Atlanta, CA. https://www.cdc.gov/minorityhealth/CHDIReport. html; 2013.
8. National healthcare quality and disparities reports. Rockville, MD. https://www. ahrq.gov/research/findings/nhqrdr/index.html; 2019.
9. Haider AH, Schneider EB, Sriram N, et al. Unconscious race and social class Bias among acute care surgical clinicians and clinical Treatment decisions. JAMA Surg. 2015;150(5):457. https://doi.org/10.1001/jamasurg.2014.4038.
10. von Hippel W, Brener L, von Hippel C. Implicit prejudice toward injecting Drug users predicts intentions to change jobs among drug and alcohol nurses. Psychol Sci. 2008; 19(1):7-11. https://doi.org/10.1111/j.1467-9280.2008.02037.x.
11. Brener L, von Hippel W, Kippax S. Prejudice among health care workers toward injecting drug users with Hepatitis C: does greater contact lead to less prejudice? Int J Drug Policy. 2007;18(5):381-7. https://doi.org/10.1016/j.drugpo.2007.01.006.
12. Paulson MR, Dekker AH, Aguilar-Gaxiola S. Eliminating disparities in pain manage- ment. J Am Osteopath Assoc. 2007;107:ES17-20.
13. Burgess DJ, Phelan S, Workman M, et al. The effect of cognitive load and Patient race on physicians’ decisions to prescribe opioids for Chronic low back pain: a random- ized trial. Pain Med. 2014;15(6):965-74. https://doi.org/10.1111/PME.12378.
14. Centers for Disease Control and Prevention. Ambulatory health care data: datasets and documentation; 2022.
15. Platts-Mills TF, Esserman DA, Brown DL, Bortsov AV, Sloane PD, McLean SA. Older US emergency department patients are less likely to receive pain medication than younger patients: results from a national survey. Ann Emerg Med. 2012;60(2): 199-206. https://doi.org/10.1016/j.annemergmed.2011.09.014.
16. Osborn SR, Yu J, Williams B, Vasilyadis M, Blackmore CC. Changes in provider pre- scribing patterns after implementation of an emergency department prescription opioid policy. J Emerg Med. 2017;52(4):538-46. https://doi.org/10.1016/j. jemermed.2016.07.120.
17. Heins JK, Heins A, Grammas M, Costello M, Huang K, Mishra S. Disparities in analge- sia and opioid prescribing practices for patients with musculoskeletal pain in the emergency department. J Emerg Nurs. 2006;32(3):219-24. https://doi.org/10. 1016/J.JEN.2006.01.010.
18. Mills AM, Shofer FS, Boulis AK, Holena DN, Abbuhl SB. Racial disparity in analgesic treatment for ED patients with Abdominal or back pain. Am J Emerg Med. 2011;29

(7):752-6. https://doi.org/10.1016/J.AJEM.2010.02.023.

1. Pomerleau AC, Schrager JD, Morgan BW. Pilot study of the importance of factors af- fecting emergency department opioid Analgesic prescribing decisions. J Med Toxicol. 2016;12(3):282-8. https://doi.org/10.1007/S13181-016-0553-9.
2. U.S. Department of Health and Human Services. Pain management best practices inter-agency task force report: updates, gaps, inconsistencies, and recommenda- tions. https://www.hhs.gov/sites/default/files/pmtf-final-report-2019-05-23.pdf; 2019.
3. St. Marie B, Broglio K.. Managing pain in the setting of opioid use disorder. Pain Manag Nurs. 2020;21(1):26-34. https://doi.org/10.1016/J.PMN.2019.08.003.
4. Webster LR. Risk factors for opioid-use disorder and overdose. Anesth Analg. 2017; 125(5):1741-8. https://doi.org/10.1213/ANE.0000000000002496.
5. Simon R, Snow R, Wakeman S. Understanding why patients with substance use dis- orders leave the hospital against medical advice: A qualitative study. Subst Abus. 2020;41(4):519-25. https://doi.org/10.1080/08897077.2019.1671942.
6. Dassieu L, Kabore JL, Choiniere M, Arruda N, Roy E. chronic pain management among people who use drugs: A health policy challenge in the context of the opioid crisis. Int J Drug Policy. 2019;71:150-6. https://doi.org/10.1016/J.DRUGPO.2019. 03.023.
7. Milbrett P, Halm M. Characteristics and predictors of frequent utilization of emer- gency services. J Emerg Nurs. 2009;35(3):191-8. https://doi.org/10.1016/J.JEN. 2008.04.032.
8. Blank FSJ, Li H, Henneman PL, et al. A descriptive study of heavy emergency depart- ment users at an academic emergency department reveals heavy ED users have bet- ter access to care than average users. J Emerg Nurs. 2005;31(2):139-44. https://doi. org/10.1016/j.jen.2005.02.008.
9. Birmingham LE, Cochran T, Frey JA, Stiffler KA, Wilber ST. Emergency department use and barriers to wellness: a survey of emergency department frequent users. BMC Emerg Med. 2016;17(1):16. https://doi.org/10.1186/s12873-017-0126-5.
10. Smulowitz PB, Cary C, Boyle KL, Novack V, Jagminas L. Variation in opioid prescribing patterns between ED providers. Intern Emerg Med. 2016;11(8):1121-4. https://doi. org/10.1007/s11739-016-1505-8.
11. Barnett ML, Olenski AR, Jena AB. Opioid-prescribing patterns of emergency physi- cians and risk of long-term use. N Engl J Med. 2017;376(7):663-73. https://doi. org/10.1056/NEJMSA1610524.
12. Clopton EL, Hyrkas EK. Modeling emergency department nursing workload in real time: an exploratory study. Int Emerg Nurs. 2020;48:100793. https://doi.org/10. 1016/J.IENJ.2019.100793.
13. Ganguly S, Lawrence S, Prather M. Emergency department staff planning to improve patient care and reduce costs. Decis Sci. 2014;45(1):115-45. https://doi.org/10. 1111/DECI.12060.
14. Welch SJ, Jones SS, Allen T. Mapping the 24-hour emergency department cycle to improve patient flow. Jt Comm J Qual Patient Saf. 2007;33(5):247-55. https://doi. org/10.1016/S1553-7250(07)33029-8.
15. Fitschen-Oestern S, Lippross S, Lefering R, et al. Does the time of the day affect mul- tiple trauma care in hospitals? A retrospective analysis of data from the TraumaRegister DGU(R). BMC Emerg Med. 2021;21(1):1-11. https://doi.org/10. 1186/S12873-021-00525-0/TABLES/5.
16. Kang H, Nembhard HB, Rafferty C, DeFlitch CJ. Patient flow in the emergency depart- ment: A classification and analysis of admission process policies. Ann Emerg Med. 2014;64(4):335-342.e8. https://doi.org/10.1016/j.annemergmed.2014.04.011.
17. Ward MJ, KC D, Jenkins CA, Liu D, Padaki A, Pines JM. Emergency department pro- vider and facility variation in opioid prescriptions for discharged patients. Am J Emerg Med. 2019;37(5):851-8. https://doi.org/10.1016/j.ajem.2018.07.054.
18. Mazer-Amirshahi M, Mullins PM, Rasooly I, van den Anker J, Pines JM. Rising opioid prescribing in adult U.S. emergency department visits: 2001-2010. Miner JR, ed. Acad Emerg Med. 2014;21(3):236-43. https://doi.org/10.1111/acem.12328.
19. McCaig LF, Burt CW. Understanding and interpreting the National Hospital Ambula- tory Medical Care Survey: key questions and answers. Ann Emerg Med. 2012;60(6): 716-721.e1. https://doi.org/10.1016/J.ANNEMERGMED.2012.07.010.
20. De Sola H, Duenas M, Salazar A, Ortega-Jimenez P, Failde I. Prevalence of therapeutic use of opioids in chronic non-Cancer pain patients and associated factors: a system- atic review and Meta-analysis. Front Pharmacol. 2020;11:1754. https://doi.org/10. 3389/FPHAR.2020.564412/BIBTEX.
21. Chen I, Kurz J, Pasanen M, et al. Racial differences in opioid use for chronic nonma- lignant pain. J Gen Intern Med. 2005;20(7):593. https://doi.org/10.1111/J.1525- 1497.2005.0106.X.
22. Chikani V, Brophy M, Vossbrink A, et al. Racial/ethnic disparities in rates of traumatic injury in Arizona, 2011-2012. Public Health Rep. 2016;131(5):704. https://doi.org/ 10.1177/0033354916663491.