Article, Emergency Medicine

Relationship between racial disparities in ED wait times and illness severity

a b s t r a c t

Background: Prolonged emergency department (ED) wait times could potentially lead to increased mortality. Studies have demonstrated that black patients waited significantly longer for ED care than nonblack patients. However, the disparity in wait times need not necessarily manifest across all illness severities. We hypothesize that, on average, black patients wait longer than nonblack patients and that the disparity is more pronounced as illness severity decreases.

Methods: We studied 34143 patient visits in 353 hospital EDs in the National Hospital Ambulatory Medical Care Sur- vey in 2008. In a 2-model approach, we regressed natural logarithmically transformED wait time on the race variable, other patient-level variables, and hospital-level variables for 5 individually stratified illness severity categories. We re- ported results as percent difference in wait times, with 95% confidence intervals. We used P b .05 for significance level. Results: On average, black patients experienced significantly longer mean ED wait times than white patients (69.2 vs 53.3 minutes; P b .001). In the multivariate model, black patients did not experience significant different wait times for the 2 most urgent severity categories; black patients experienced increasingly longer waits vs nonblack patients for the 3 least urgent severity categories (14.7%, P b .05; 15.9%, P b .05; 29.9%, P b .001, respectively).

Conclusion: Racial disparity in ED wait times between black and nonblack patients exists, and the size of the dis- parity is more pronounced as illness severity decreases. We do not find a racial disparity in wait times for criti- cally ill patients.

(C) 2015

Introduction

Emergency department (ED) crowding is a significant public health issue [1,2]. In recent years, the rate of ED utilization has increased while the number of available EDs has simultaneously decreased [3]. These 2 factors together exacerbate crowding in the ED [4]. Emergency depart- ment crowding extends the time patients have to wait to receive testing and treatment [2], potentially leading to adverse outcomes [5]. As such, ED wait time can be used as a method to measure ED crowding [6-8]. Patients with illnesses such as acute myocardial infarction, stroke, sep- sis, pneumonia, and traumatic injuries benefit greatly from rapid, timely medical interventions [4], and prolonged ED wait times could potential- ly lead to increased mortality [9], protracted pain and suffering, and poor patient satisfaction in hospital EDs [4,10,11].

Emergency department crowding may disproportionately affect mi- nority populations, raising questions of ED care equitability [11]. Racial

? Prior presentations: None.

?? Funding sources/disclosures: There are no conflicts of interest or financial disclosures to report.

* Corresponding author at: 211 E Ontario St, Suite 200, Chicago, IL 60611. Tel.:+1 312 694 7000.

E-mail address: [email protected] (W.P. Qiao).

disparity in ED wait times has been studied by prior authors [1,2,4,11- 13]. These studies demonstrated that black patients waited significantly longer for ED care than nonblack patients. However, to the best of our knowledge, none explicitly analyze the relationship between wait time disparities with respect to patients’ illness severity upon presenta- tion to the ED. Although studies generally show minority patients expe- rience longer wait times, the disparity in wait times need not necessarily manifest across all illness severities. Although we should not allow for any racial disparity in ED care regardless of illness severity, detection of such disparity for critically ill patients would warrant im- mediate intervention.

In this study, we evaluate whether racial disparity differs by patients’ illness severity as measured by ED triage level. We hypothesize that on average, black patients wait longer than nonblack patients and that the disparity is more pronounced as illness severity decreases.

Methods

Study design

This study is a retrospective cross-sectional study of data collected in the National Hospital Ambulatory Medical Care Survey in 2008. The NHAMCS is approved annually by the Ethics Review Board

http://dx.doi.org/10.1016/j.ajem.2015.08.052

0735-6757/(C) 2015

W.P. Qiao et al. / American Journal of Emergency Medicine 34 (2016) 1015 11

of National Center for Health Statistics (NCHS) with waivers of the re- quirements to obtain informed consent of patients and patient authori- zation for release of patient medical record data by health care providers [14]. The institutional review board determined that this study was ex- empt from informed consent.

Study setting and population

The NHAMCS is an annual, national probability sample of ambulato- ry visits made to nonfederal, general, and short-stay hospitals in the United States conducted by the Centers for Disease Control and Preven- tion, NCHS. Although the survey includes visits to selected ambulatory care departments, this analysis focuses solely on the visits to hospital EDs. The survey has been conducted annually since 1992. The multistaged sample design is composed of 3 stages for the ED compo- nent: (1) 112 geographic primary sampling units that comprise a prob- ability subsample of primary sampling units from the 1985 to 1994 National Health Interview Surveys; (2) approximately 480 hospitals within primary sampling units; and (3) patient visits within emergency service areas. Sample hospitals are randomly assigned to 16 panels that rotate across thirteen 4-week reporting periods throughout the year. The initial sample frame of hospitals was based on the 1991 Strategic Marketing Group hospital database [14].

The NHAMCS 2008 recordED patient visits to 353 US hospital EDs and recorded a total of 34143 patient visits. Hospital staff members were trained to complete surveys for the NHAMCS. Information for each visit was recorded on a variety of patient, visit, and hospital characteristics. The US Bureau of the Census oversees the data collection [14]. Specific methods of survey procedures were provided by the NCHS [15].

Study protocol

The data were modified before formal analysis. From the 34143 patient visits in the NHAMCS data set, we excluded visits with missing wait times (n = 6983) to form a sample of 27 160 patient visits with actual wait times available for analysis. We then excluded visits with missing triage cat- egories (n = 3092), unknown insurance payment information (n = 1182), and visits occurring in EDs that did not perform triage (n = 409).

We then stratified the data into 5 cohorts according to their illness se- verity presented to the ED. Illness severity presented to the ED was ap- proximated by the immediacy by which a patient needed to be seen and is an “immediacy variable” in the NHAMCS. This variable is broken down into 5 triage categories in the NHAMCS: immediately, 1 to 14 minutes, 15 to 60 minutes, greater than 1 hour to 2 hours, and greater than 2 hours to 24 hours. The determination of the triage categories was assigned based on the clinical judgment of an ED health provider (ie, triage nurse) upon arrival at the ED [15]. The immediacy variable in the NHAMCS com- bines different Triage systems used across institutions into one 5-level sys- tem. In cases where an ED used a 3- or 4-level triage system, the levels were mapped to the best corresponding category in the 5-level system in the NHAMCS during the editing process [15,16]. We used these 5 triage categories in the NHAMCS to indicate patient illness severity.

Key outcome measures

The primary outcome measure or dependent variable was wait time before being seen by a physician. Wait time is defined as difference be- tween the time the patient arrives in the ED and the time the patient is examined by a physician [15]. The primary independent variable was Black race (vs nonblack race).

Patient-level variables included, age (continuous variable), sex, insurance payment method (private, self-pay, other), season of visit (December- February, March-May, June-August, September-November), and day of week of visit (weekday vs weekend). Hospital-level variables included hospital ownership type (for-profit, government, nonprofit), census region of the United States (Northeast, West, South, Midwest),

metropolitan statistical area (urban vs nonurban), and teaching status (teaching vs nonteaching). Teaching hospital status was defined if a hospital had a patient seen by a resident physician [15].

Data analysis

We performed univariate analysis first to analyze whether black pa- tients experienced longer mean and median wait times than nonblack patients. We then analyzed the disparity in wait times by stratifying the patient population by illness severity. We used a log-linear model for bivariate and multivariate regression analyses to further evaluate the differences in wait times by race. We natural-logarithmically trans- formed ED wait times due to the 1-side skewed distribution of wait times in our sample. Natural-logarithmically transforming the wait time variable allowed us to obtain a more normal distribution for re- gression analysis, potentially yielding more reliable estimations than those allowed by a skewed distribution.

We decided to use a 2-model approach to illustrate the effects of adding patient- and hospital-level variables on wait time. In model 1, we regressed natural-logarithmically transformed wait time on the race vari- able for each individually stratified severity categories (1-5). In model 2, we added patient- and hospital-level variables to model 1 and again regressed on each individually stratified severity categories to control for confounding. We reported results as percent difference in wait times, with 95% confidence intervals. We used P b .05 for significance level.

We performed statistical analyses using Stata Version 12.0 (StataCorp, College Station, TX). The survey data were analyzed using the sampled visit weight that is the product of the corresponding sam- pling fractions at each stage in the sample design. The sampling weights have been adjusted by NCHS for survey nonresponse within time of year, geographic region, urban/rural, and ownership designations, yielding an unbiased national estimate of ED visit occurrences, percent- ages, and characteristics. Because of the complex sample design, sam- pling errors were determined using the appropriate survey procedure following the guidance of the NHAMCS documentation, which takes into account the clustered nature of the sample [14,15].

Results

The patient population: black vs nonblack patients

Our sample contained 5290 black vs 17178 nonblack patient visits (Table 1). Black patient visits encompassed 22.3% of the total visits, after accounting for survey methodology. Nationally, differences in patient and hospital characteristics existed across black vs nonblack patients (Table 1). Black patients tended to be younger on average (34.5 vs 39.5 years old; P b .001), more likely to be female (57.0% vs 54.0%; P b .05), less likely to use private insurance as source of payment (29.7% vs 38.7%; P b .001), less likely to visit EDs in the northeastern and western re- gions (13.4% vs 20.4% and 8.8% vs 23.6%; P b .001, respectively), more like- ly to visit EDs in the southern region (59.4% vs 36.5%; P b .001), and more likely to present with an illness with the lowest severity (category 5) (14.4% vs 7.5%; P b .05), when compared to nonblack patients. Mean ED wait time also differed across races. Black patients experienced aver- age wait times of 69.2 minutes, compared to only 53.3 minutes for non- black patients (P b .001). Similar pattern holds true for median ED wait times across races (43 minutes [interquartile range {IQR}, 20-88] for black patients vs 33 minutes [IQR, 15-68] for nonblack patients). No sig- nificant differences were found between races for variables such as season of the ED visit, day of visit, teaching hospital status, hospital ownership type, hospital region, and urban vs rural status.

Unadjusted analysis: mean and median wait time by race and severity

Unadjusted analyses of mean and median ED wait times (Table 2) demonstrated differences in wait times by race. Black patients

12 W.P. Qiao et al. / American Journal of Emergency Medicine 34 (2016) 1015

Table 1

Patient and hospital characteristics by race (% [95% confidence interval])

nonblack patients. Black patients experienced longer wait times vs non- black patients for severity categories 3, 4, and 5 (22.8%, P b .001; 27.9%, P

Black race

(n = 5290, 22.3%)

Nonblack race P

(n = 17178)

b .001; 39.1%, P b .001, respectively). multivariate regression analysis after the addition of patient- and hospital-level covariates (Table 3, model 2) yielded differences in wait times between black and nonblack patients again for severity categories 3, 4, and 5, albeit with the size of the differences decreased when compared to model 1 (14.7%, P b .05; 15.9%, P b .05; 29.9%, P b .001, respectively).

Multivariate analysis (Table 3, model 2) also revealed differences in wait times across other characteristics as well. Holding all else constant, fe- male patients experienced longer ED wait times across all severity catego- ries 1 to 5 (24.3%, P b .05; 21.3%, P b .001; 10.6%, P b .001; 11.5%, P b .05;

Age (y)

34.5 (32.6-36.4)

39.5 (38.3-40.6)

b.001

Sex

Female

57.0 (55.3-58.7)

54.0 (53.0-55.1)

b.05

Season

December-February

20.0 (13.5-26.6)

25.1 (18.7-31.5)

March-May

22.0 (15.4-28.7)

26.4 (20.8-32.0)

June-August

31.1 (22.4-39.7)

26.6 (21.1-32.0)

September-November

26.9 (15.4-38.4)

22.0 (16.0-28.0)

.172

Day of visit

Weekend

28.9 (27.6-30.1)

29.7 (28.9-30.4)

.274

Source of payment

Private insurance

29.7 (26.6-32.8)

38.7 (36.2-41.3)

Self-pay

17.6 (15.0-20.2)

13.4 (12.1-14.7)

Other

Teaching status

52.7 (50.0-55.4)

47.9 (45.4-50.3)

b.001

Teaching hospital

51.9 (40.7-63.0)

43.5 (35.2-51.9)

.094

Hospital ownership

For-profit

11.8 (4.0-19.7)

13.9 (6.9-20.9)

Government

13.0 (8.7-17.4)

10.2 (6.7-13.7)

Nonprofit Region

75.1 (67.0-83.2)

75.9 (67.9-84.0)

.420

Northeast

13.4 (7.8-19.1)

20.4 (17.0-23.9)

West

8.8 (5.1-12.5)

23.6 (17.3-29.9)

South

59.4 (50.1-68.8)

36.5 (30.7-42.2)

Midwest

18.3 (11.8-24.9)

19.6 (15.13-24.1)

b.001

Urban vs rural

18.6%, P b .05, respectively). For-profit hospitals yielded shorter wait times than government and nonprofit hospitals for severity categories 2 and 3 (54.3%, P b .05, and 31.9%, P b .05, respectively). Visits to EDs in the southern states experienced longer wait times for severity categories 4 and 5 when compared to the reference category (33.7%, P b .05, and 61.3%, P b .001, respectively). Finally, urban hospitals experienced longer waits than nonurban hospitals for severity categories 2 to 5 (37.4%, P b

.05; 35.2%, P b .001; 43.0%, P b .05; 60.8%, P b .001, respectively).

Discussion

Urban 84.7 (72.4-97.1) 81.9 (72.7-91.1) .584

Severitya

Immediate 4.6 (2.9-6.2) 4.0 (3.2-4.7)
  • 1-14 minutes 12.9 (10.5-15.3) 15.7 (13.2-18.3)
  • 15-60 minutes 45.4 (39.0-51.8) 48.5 (45.3-51.7)
  • N 1 to 2 h 22.7 (18.1-27.3) 24.3 (21.4-27.3)
  • N 2 to 24 h 14.4 (9.1-19.7) 7.5 (5.8-9.2) b.001
  • Mean ED wait time (min) 69.2 (61.6-76.9) 53.3 (49.2-57.4) b.001

    Median ED wait time (min) 43 (20-88)b 33 (15-68)

    Bold values indicate significance at p values b0.05.

    a Severity is approximated by the “immediacy with which patients should be seen” variable in NHAMCS.

    b Denotes IQR.

    experienced longer wait times compared to nonblack patients in severity categories 3, 4, and 5 (14.2 minutes, P b .05; 17.5 minutes, P b .05; and 22.9 minutes, P b .05, respectively). Mean ED wait times did not differ sig- nificantly for severity categories 1 and 2. Similar pattern holds true for median ED wait times: differences in median wait times were observed for severity categories 3, 4, and 5 (6, 16, and 26 minutes, respectively), and no differences were observed for severity categories 1 and 2.

    3.3. Adjusted analysis: percent difference in wait times by race and severity

    Bivariate analysis of percent difference in ED wait times (Table 3, Model 1) further illustrated differences in wait times for black vs

    This is the first nation-wide study to explicitly demonstrate an in- verse relationship between racial disparity in ED wait times and illness severity. Specifically, we found that black patients experienced increas- ingly longer wait times as illness severity decreased, with the propor- tion of disparity greatest for the least severe illnesses. Fortunately, we also found that black patients did not experience significantly longer wait times for critically severe illnesses.

    Consistent with results of other studies, we found that racial dispar- ities exist in ED wait times [1,2,4,11-13]. Before adjusting for any con- founders, we demonstrated that black patients experienced longer mean wait times by almost 16 minutes when compared to nonblack pa- tients (“The patient population: black vs nonblack patients” section and Table 1). Other work has further documented racial disparities in ED wait times while controlling for illness severity [1,12]. In a recent study of racial differences in wait times in the United States, the authors concluded black patients waited 4% to 9% longer than white patients after controlling for illness severity as well as other various confounders [12]. Similarly, a study of racial disparities in ED wait times in Pediatric populations concluded that black patients on average waited 23% longer vs non-Hispanic white patients after controlling for severity [1].

    However, we did not find studies that specifically addressed the re- lationship between racial disparity in ED wait times and illness severity by stratifying patients into cohorts according to their severity catego- ries. Stratifying our sample as such allows explicit examination of the ef- fect of race on ED wait time in each illness severity cohort. This is particularly important because our analysis demonstrated that black pa- tients used EDs differently than nonblack patients: a higher percentage of black patients visited EDs for lesser severity illnesses (14.4% vs 7.5%

    Table 2

    Unadjusted analysis: mean and median ED wait time by severity of illness (minutes [95% confidence interval or IQR])

    Severitya

    1

    2

    3

    4

    5

    Mean wait time

    Race

    Black

    21.1 (13.5-28.8)

    42.4 (32.4-52.4)

    68.0 (57.9-78.0)

    79.9 (68.3-91.5)

    95.8 (75.7-115.9)

    Nonblack

    17.2 (14.5-20.0)

    37.8 (31.9-43.6)

    53.7 (49.3-58.1)

    62.4 (55.7-69.2)

    72.8 (60.1-85.6)

    Difference

    3.9 (P = .306)

    4.64 (P = .317)

    14.2 (P b .05)

    17.5 (P b .05)

    23.0 (P b .05)

    Median wait time Race

    Black

    9 (2-27)

    19 (10-54)

    41 (22-82)

    60 (30-100)

    70 (30-139)

    Nonblack

    9 (2-24)

    19 (8-44)

    35 (17-66)

    44 (20-85)

    44 (23-101)

    Difference

    0

    0

    6

    16

    26

    a Severity is approximated by the “immediacy with which patients should be seen” variable in NHAMCS.

    W.P. Qiao et al. / American Journal of Emergency Medicine 34 (2016) 1015

    13

    Table 3

    Percent difference in ED wait time by severity of illness (% [95% confidence interval])

    Model 1 Bivariate regression: % difference in ED wait time as a function of race

    Severitya

    1

    P

    2

    P

    3

    P

    4

    P

    5

    P

    Race

    Black

    2.7 (-23.2 to 28.7)

    .835

    11.6 (-9.9 to 33.0)

    .288

    22.8 (11.6-33.9)

    b.001

    27.9 (13.5-42.4)

    b.001

    39.1 (19.7-58.5)

    b.001

    Nonblack

    Ref

    Ref

    Ref

    Ref

    Ref

    Model 2 Multivariate regression: % difference in ED wait time as a function of race plus all other confounders

    Severity

    1

    2

    3

    4

    5

    Race

    Black

    -7.9 (-32.7 to 17.0)

    .531

    1.0 (-15.4 to 17.4)

    .902

    14.7 (4.6-24.9)

    b.05

    15.9 (4.6-27.3)

    b.05

    29.9 (17.6-42.1)

    b.001

    Nonblack

    Ref

    Ref

    Ref

    Ref

    Ref

    Age (y)

    -0.1 (-0.4 to 0.3)

    .683

    -0.1 (-0.3 to 0.1)

    .435

    -0.2 (-0.3 to -0.03)

    b.05

    -0.3 (-0.5 to -0.1)

    b.05

    0.1 (-0.2 to 0.4)

    .536

    Sex

    Female

    24.3 (7.5-41.1)

    b.05

    21.3 (12.9-29.7)

    b.001

    10.6 (5.3-15.9)

    b.001

    11.5 (3.6-19.4)

    b.05

    18.6 (6.7-30.5)

    b.05

    Season

    Winter

    -27.0 (-67 to 13.0)

    .183

    8.9 (-29.7 to 46.9)

    .643

    -1.0 (-20.9 to 18.9)

    .922

    2.0 (-22.2 to 26.2)

    .869

    -1.7 (-36.1 to 32.6)

    .921

    Spring

    1.6 (-38.7 to 41.9)

    .936

    15.8 (-23.7 to 55.3)

    .432

    3.7 (-15.2 to 22.6)

    .701

    5.2 (-19.1 to 29.4)

    .674

    -13.1 (-40.2 to 13.9)

    .337

    Summer

    9.7 (-25.3 to 44.7)

    .584

    -6.6 (-36.9 to 23.8)

    .670

    1.8 (-17.2 to 20.9)

    .849

    19.4 (-1.4 to 40.2)

    .067

    2.1 (-23.5 to 27.7)

    .870

    Fall

    Ref

    Ref

    Ref

    Ref

    Ref

    Day of visit

    Weekend

    -7.2 (-30.7 to 16.3)

    .544

    -2.7 (-13.7 to 8.2)

    .626

    -5.7 (-10.5 to -1.0)

    b.05

    -1.7 (-7.9 to 4.5)

    .586

    -17.7 (-28.6 to -6.8)

    b.05

    Source of payment

    Private insurance

    -2.2 (-25.4 to 21)

    .850

    -7.7 (-23.2 to 7.9)

    .331

    -6.7 (-13.6 to 0.3)

    .059

    0.4 (-8.1 to 8.8)

    .926

    5.3 (-10.5 to 21.1)

    .506

    Self-pay

    7.0 (-21.3 to 35.4)

    .624

    -9.7 (-8.8 to 28.2)

    .301

    -0.8 (-10.4 to 8.9)

    .877

    4.6 (-4.7 to 13.9)

    .330

    4.6 (-11.1 to 20.2)

    .565

    Other

    Ref

    Ref

    Ref

    Ref

    Ref

    Teaching status

    Teaching hospital

    17.8 (-9.7 to 45.2)

    .203

    13.9 (-9.5 to 37.2)

    .242

    7.5 (-3.8 to 25.7)

    .143

    1.0 (-17.9 to 19.8)

    .919

    21.4 (-4.1 to 46.9)

    .099

    Hospital ownership

    For-profit

    17.4 (-27.0 to 61.9)

    .438

    -54.3 (-88.3 to -20.4)

    b.05

    -31.9 (-59.5 to -5.4)

    b.05

    -24.1 (-61.9 to 13.8)

    .211

    6.1 (-26.1 to 38.2)

    .709

    Government

    -30.4 (-66.6 to 5.7)

    .098

    -29.3 (-64.8 to 6.0)

    .103

    -1.3 (-19.5 to 16.8)

    .886

    1.2 (-19.9 to 22.3)

    .913

    -9.4 (-48.2 to 29.4)

    .633

    Nonprofit

    Ref

    Ref

    Ref

    Ref

    Ref

    Region

    Northeast

    -40.7 (-66.6 to -14.8)

    b.05

    -7.7 (-33.7 to 18.3)

    .560

    8.3 (-10.9 to 27.6)

    .394

    28.0 (-0.1 to 56.1)

    .051

    58.0 (28.2-87.8)

    b.001

    West

    -23.8 (-69.7 to 22.1)

    .307

    -48.3 (-81.4 to -15.2)

    b.05

    -3.5 (-28.1 to 21.1)

    .780

    -14.5 (-45.5 to 16.5)

    .356

    15.2 (-26.6 to 57.0)

    .472

    South

    -22.2 (-53.1 to 8.7)

    .157

    -5.6 (-37.0 to 25.9)

    .728

    17.0 (-4.3 to 38.2)

    .116

    33.7 (11.1-56.2)

    b.05

    61.3 (29.0-93.6)

    b.001

    Midwest

    Ref

    Ref

    Ref

    Ref

    Ref

    Urban vs rural

    Urban

    7.7 (-21.7 to 37.1)

    .604

    37.4 (6.8-68.0)

    b.05

    35.2 (18.9-51.7)

    b.001

    43.0 (18.9-67.0)

    b.05

    60.8 (27.6-94.0)

    b.001

    a Severity is approximated by the “immediacy with which patients should be seen” variable in NHAMCS.

    14 W.P. Qiao et al. / American Journal of Emergency Medicine 34 (2016) 1015

    for severity category 5). Because a higher proportion of black patients presented with lesser illness severity, then it could have justified the longer mean wait times black patients suffered.

    Still, we found racial disparities in wait times in some cohorts de- spite stratifying according to illness severity in the bivariate and multi- variate regression models in the adjusted analysis (Table 3, models 1 and 2). Specifically, disparity appeared starting with severity category 3, with the size of the disparity increasing as severity decreased; the dis- parity was the greatest for the least severe illnesses (category 5), and there was no racial disparity in wait times for critically severe illnesses (severity categories 1 and 2). Although the addition of patient and hos- pital covariates (Table 3, model 2) tempered the absolute percent differ- ence in ED wait times associated with race for severity categories 3, 4, and 5 visits, the differences remained significant.

    Potential drivers of longer waits experienced by black patients are multifaceted. One way by which differences in wait times can be broken down is by between-hospital effects and within-hospital effects [2,11]. Between-hospital effects refer to intrinsic differences that likely exist across different EDs and differences in ED utilization patterns that explain the variations in wait times experienced by black and nonblack patients [11]. Within-hospital effects refer to differences in wait times that depend on the specific race/ethnicity of patients seen in the ED, such as varying patient preferences, communication patterns, language barriers, literacy, or even differential treatment of minority patients by providers [1,11].

    We attempted to control for between-hospital variations in wait times by incorporating hospital-level variables such as ownership, teaching sta- tus, region, and urban vs rural status. Yet, racial disparity in wait times still persisted after controlling for these confounders, and we found that black patients experienced significantly longer waits for lesser severe illnesses in urban and southern hospitals. This leads us to believe that the remain- ing disparity could be secondary to unidentified between-hospital effects and/or other within-hospital effects. Emergency department volume is a potential between-hospital effect driving the difference in wait times be- tween races [12]. Prior studies have suggested a positive correlation be- tween ED volume and wait time/overcrowding [4,11,12]. A recent study by Sonnenfeld et al [12] further noted that black patients are more likely to receive care at high-volume EDs than white patients. These high- volume EDs may be safety net hospitals that care for a disproportionate high share of minority patients under suboptimal funding and constrained resources [11]. In addition, race is commonly associated with other factors such as Household income, education, and degree of residential segregation. These socioeconomic factors could poten- tially be associated with differences in ED utilization between black and nonblack patients, thus helping to explain between-hospital variations in wait times. The combination of these between- hospital effects and racial factors could contribute to the differences in waits between races.

    Aside from between-hospital effects, within-hospital effects could contribute to racial differences in wait times. The persistence of longer wait times experienced by black patients independent of our efforts to account for illness severity, insurance, and other between-hospital ef- fects could be a point of concern, as the differences in waits could sug- gest disparity secondary to within-hospital effects. Although there is still much clarification needed on this issue, other authors have raised the possibility of provider biases (both conscious and unconscious) leading to delays in care for minorities patients, even if presenting symptoms are the same across races/ethnicities [1]. This disparity can become more apparent when providers are pressured for time and/or are working with incomplete medical information [1].

    We also found both interesting and unexpected that female patients experienced longer waits when compared to male patients across all 5 illness severities, even after controlling for confounders. We hypothe- size that the female patients presenting to EDs with obstetrical or gyne- cological complaints or presenting to women’s hospitals could be associated with differences in wait times. Further examination is re- quired to characterize this finding.

    Alleviating health care disparity is a policy goal in Healthy People 2020 [17]. The results of our study have implications on direction of fu- ture research and policy making. First, our results call for more research efforts toward identifying between-hospital and within-hospital effects that might contribute to racial disparity in ED wait time. Second, because racial disparity in wait time does not seem to exist for critically severe ill- nesses presenting to EDs, resources ought to be allocated toward alleviat- ing disparities in care for patients presenting with less severe illnesses, for which the disparities appear to be much more pronounced. Resources di- rected toward alleviating disparity in critically severe illnesses could po- tentially be inefficient. Allocating resources to address longer waits experienced by EDs located in southern states and in urban metropolitan areas, regardless of race, could also be considered. Ultimately, alleviating ED crowding and burgeoning ED waits across the nation will likely re- quire multiple reforms [4,18].

    Limitations

    Our study is not without limitations. We excluded 6983 visits with missing wait times (20.4% of the original 34143 visits). We considered these exclusions necessary, as these visits did not contain our outcome of interest. From our sample of 27160 visits that contained actual wait times available for analysis, we excluded patients with unknown triage status (n = 3092), unknown insurance payment methods (n = 1182), and visits occurring in EDs that did not perform triage (n = 409) or 17.2% of the total sample with wait time data. This may have introduced a degree of selection bias into the regression analysis [4]. However, we could not conclude if these exclusions systemically biased the results. We did not account for ED volume in our study design, a potential between-hospital effect driving the differences in wait times across races. However, we believe that ED volume could conceivably be corre- lated with other hospital-level covariates we did account for in our anal- ysis, thus mitigating the impact of this limitation. We also could not examine data from individual hospitals in our data set; therefore, we could not draw explicit conclusions regarding within-hospital effects on ED wait time disparities. This is secondary to a limitation of the data set. In addition, the reliability of the triage categories as a proxy for illness severity might not be consistent throughout all hospitals, as the requirements for triage assignment can vary across different institu- tions [10]. However, 2 previous studies analyzing ED wait times and tri- age categories with NHAMCS have cited the reliability of the triage assessment as fair to excellent [10,19]. Issues of racial disparity in triage category assignment have also been cited as well, potentially further confounding the problem [20]. A less severe patient inappropriately re- ceiving a higher triage assignment could have exacerbated longer waits for more severely ill patients [1]. However, more evidence is required to confirm the confounding secondary to racial disparity in triage assign- ment, as the study supporting it was only conducted within a single in- stitution [20]. Measurement errors are also potentially inherent in the NHAMCS data set, as data are sometimes recorded based on perception by the hospital personnel [1]. Furthermore, measurement of wait time can vary by hospital, as different hospitals register its patients in a vari- ety of locations, such as in the treatment area, during triage assignment, or even before triage assignment [1]. Unfortunately, these data set lim- itations are out of our control. However, since 2005, the Centers for Dis- ease Control and Prevention has made efforts to manually check and ensure the accuracy of ED wait times in the NHAMCS [19]. We believe that this helps curtail errors in wait time measurements. Finally, we sought to analyze differences between only black and nonblack pa- tients. As a result, we could not draw conclusions for other minority groups, particularly Hispanic patients, who make up a significant por- tion of the patient population using EDs (9.3%-20.4%, depending on the study and studied population) [4,11]. James et al [1] have noted that Hispanic patients seem to experience longer waits when compared to white patients, which leads us to believe that the size of the true dis- parity between black and white patients could potentially be even

    W.P. Qiao et al. / American Journal of Emergency Medicine 34 (2016) 1015 15

    bigger than what our results suggest (if Hispanic patients were parsed out from our nonblack reference group).

    Conclusion

    Racial disparity in ED waits times between black and nonblack pa- tients exists, and the size of the disparity is more pronounced as illness severity decreases. Fortunately, we do not find a racial disparity in wait times for critically ill patients. Nonetheless, as prolonged ED wait times potentially lead to increased mortality [9], protracted pain and suffering, and poor patient satisfaction [4], efforts to further clarify and address this disparity for those with lesser illness severity are appropriate.

    Acknowledgments

    The corresponding author thanks Xiaoling Song, MD, and Xugang Qiao, MD, for their unconditional support; Brian Trinh for his encour- agement; and Yumei Wang, Deming Song, Lingmei Zhao, and Qingwen Qiao for instilling in the corresponding author values without which all this would be impossible.

    References

    1. James CA, Bourgeois FT, Shannon MW. Association of race/ethnicity with emergency department wait times. Pediatrics 2005;115:e310-5.
    2. Pines JM, Localio AR, Hollander JE. Racial disparities in emergency department length of stay for admitted patients in the United States. Acad Emerg Med 2009; 16:403-10.
    3. Tang N, Stein J, Hsia RY, Maseli JH, Gonzales R. Trends and characteristics of US emergency department visits, 1997-2007. JAMA 2010;304(6):664-70.
    4. Wilper AP, Woolhandler S, Lasser KE, McCormick D, Cutrona SL, Bor DH, et al. Waits to see an emergency department physician: U.S. trends and predictors, 1997-2004. Overcrowding by patients unable to obtain nonurgent care elsewhere might be worsening care for those with truly urgent conditions. Health Aff 2008;27(2): w84-95.
    5. Sun BC, Hsia RY, Weiss RE, Zingmond D, Liang LJ, Han W, et al. Effect of emergency department crowding on outcomes of admitted patients. Ann Emerg Med 2013;61: 605-11.
    6. Bernstein SL, Aronsky D, Duseja R, Epstein S, Handel D, Hwang U, et al. The effect of emergency department crowding on clinically oriented outcomes. Acad Emerg Med 2009;16:1-10.
    7. Hwang U, Concato J. Care in the emergency department: how crowded is over- crowded? Acad Emerg Med 2004;11:1097-101.
    8. Hwang U, McCarthy ML, Aronsky D, Asplin B, Crane PW, Craven CK, et al. Measures of crowding in the emergency department: a systematic review. Acad Emerg Med 2011;18:527-38.
    9. Plunkett PK, Declan GB, Breslin T, Bennett K, Bernard S. Increasing wait times predict increasing mortality for emergency medical admissions. Eur J Emerg Med 2011;18: 192-6.
    10. Horwitz LI, Green J, Bradley EH. US emergency department performance on wait time and length of visit. Ann Emerg Med 2010;55:133-41.
    11. Park CY, Lee MA, Epstein AJ. Variation in emergency department wait times for chil- dren by race/ethnicity and payment source. Health Serv Res 2009;44(6):2022-39.
    12. Sonnenfeld N, Pitts SR, Schappert SM, Decker SL. Emergency department volume and racial and ethnic differences in Waiting times in the United States. Med Care 2012;50:335-41.
    13. Wu BU, Banks PA, Conwell DL. Disparities in emergency department wait times for acute gastrointestinal illnesses: results from the National Hospital Ambulatory Med- ical Care Survey, 1997-2006. Am J Gastroenterol 2009;104:1668-73.
    14. Center for Disease Control and Prevention. Sample Text for Describing NHAMCS in a Research Article. Ambulatory health care data. Available at: http://www.cdc.gov/ nchs/ahcd/ahcd_research_tools.htm. [Accessed Date (3/31/2015)].
    15. Center for Disease Control and Prevention. 2008 NHAMCS micro-data file documen- tation. Ambulatory Health Care Data. Available at: http://www.cdc.gov/nchs/ahcd/ ahcd_questionnaires.htm. [Accessed Date (3/31/2015)].
    16. Center for Disease Control and Prevention. Understanding and Interpreting the Na- tional Hospital Ambulatory Medical Care Survey (NHAMCS): key questions and an- swers. Ambulatory Health Care Data. Available at: http://www.cdc.gov/nchs/ahcd/ ahcd_research_tools.htm. [Accessed Date (3/31/2015)].
    17. Office of Disease prevention and Health Promotion. Access to Health Services. Healthy People 2020. Available at: http://www.healthypeople.gov/2020/topics-ob- jectives/topic/Access-to-Health-Services. [Accessed Date (3/31/2015)].
    18. Trzeciak S, Rivers EP. Emergency department overcrowding in the United States: an emerging threat to patient safety and public health. Emerg Med J 2003;20:402-5.
    19. Horwitz LI, Bradley EH. Percentage of US emergency department patients seen with- in the recommended Triage time: 1997 to 2006. Ann Intern Med 2009;169(20): 1857-65.
    20. Schrader CD, Lewis LM. Racial disparity in emergency department triage. J Emerg Med 2013;44(2):511-8.

    Leave a Reply

    Your email address will not be published. Required fields are marked *