General characteristics of included articles

General characteristics of articles are reported in Table 2. Eighty-three articles were published over 28 years, from 1990 to 2018; 68% of studies were published after 2008. Studies are clustered for analysis by ED process (triage, wait times, analgesia, diagnosis, treatment, leaving without being seen (LWBS)/against medical advice (AMA), and patient experiences). Categories are not mutually exclusive, as some studies report on multiple processes. Citations of all included studies, and notes on findings and methodologies are given in the Supplemental Online Appendix B. See Figure 2 for an analysis of outcomes by study category.

Figure 2. Outcomes by study category.

Table 2. General characteristics of included articles

Triage scores

Seven studies (100%) considering triage scores found differences between racial and ethnic minority v. majority groups. Six (86%) of seven found minority groups had less acute scores, while one (14%) found that the minority group had more acute scores.^{Reference Vigil, Alcock and Coulombe14–Reference Johnston-Leek, Spirvulis, Stella and Palmer16} Some investigators found minority groups had less acute triage scores despite similar presenting complaints or pain ratings.^{Reference Lopez, Wilper, Cervantes, Betancourt and Green15,Reference Vigil, Coulombe and Alcock17}

Waiting times

Twelve studies examined waiting times. Eleven (91%) found that minority groups waited longer on average to be assessed than white patients.^{Reference Sonnenfeld, Pitts, Schappert and Decker18}

Analgesia

Twenty-four studies examined differences in analgesic administration. Fifteen (63%) of 24 found differences in analgesic administration between racial and ethnic groups. Six (35%) of 17 studies found minority group patients were less likely to receive any analgesics compared with the majority group. Eleven (85%) of 13 studies found that racial/ethnic minority groups were less likely to receive opioids. Three (100%) of three studies found that racial/ethnic minority groups were less likely to receive a prescription for analgesics. Two (67%) of three studies found minority groups waited longer for analgesics. Nine (38%) of 24 studies found no differences between racial and ethnic groups for any process studied.

Diagnostic testing and treatment

Thirty studies examined differences in diagnostic testing and treatment between racial/ethnic groups. Differences in administration of diagnostic testing and treatment for chest pain, asthma, psychiatric disorders, gastro-intestinal complaints, trauma, and strokes were investigated. Twenty-five (83%) of 30 found differences between racial/ethnic minority groups on at least one of the evaluated processes.^{Reference Pezzin, Keyl and Green19–Reference Payne and Puumala21}

Studies that found no differences in diagnostics were largely focused on trauma. Five studies examined administration of diagnostic testing in trauma, with four (80%) finding no differences.^{Reference Shafi and Gentilello22} See Appendix B for details regarding disparities seen in diagnosis and treatment.

LWBS and leaving AMA

Four (100%) studies investigating LWBS/AMA found LWBS/AMA more likely in minority group than majority group patients.^{Reference Harrison, Finkelstein, Puumala and Payne23,Reference Wright24} All focused on Indigenous populations. Harrison et al. found that minority patients other than Indigenous patients were not more likely than their white counterparts to LWBS after controlling for insurance status.^{Reference Harrison, Finkelstein, Puumala and Payne23} This finding was recently replicated by Weber et al. 2018, in a study published after our search.^{Reference Weber, Ziegler, Kharbanda, Payne, Birger and Puumala25}

Subjective experiences

Two studies analyzed patient trust in providers. Both found nonwhite patients had lower trust scores for providers than white patients.^{Reference Fields, Abraham, Gaughan, Haines and Hoehn26,Reference Lee, Tamayo-Sarver, Kinneer and Hobgood27} In qualitative studies of minority patient experiences, themes included discrimination, stereotyping, language barriers, and judgment.^{Reference Pickner, Ziegler and Hanson28,Reference Chapman, Smith and Martin29}

Author explanations of causes of disparities

Using inductive thematic analysis, we categorized explanations of disparities that authors of studies proposed. See Table 3 for definitions of categories. Proposed explanations fall into categories of: (1) provider bias, (2) patient-provider interactions, (3) facility characteristics, (4) clinical presentations, and (5) provider judgement v. standardized care. In some cases, these possible causes appear in the discussion sections of articles, and are not investigated. Authors sometimes cite studies to support proposed explanations of disparities, but it is beyond the scope of this review to follow up on the citations authors relied on.

Table 3. Author-proposed explanations of disparities

Provider bias

Provider biases were suggested as a cause of disparities in analgesia provision,^{Reference Pletcher, Kertesz, Kohn and Gonzales30,Reference Singhal, Tien and Hsia31} triage scores,^{Reference Vigil, Alcock and Coulombe14,Reference Vigil, Coulombe and Alcock17} wait times,^{Reference Haywood, Tanabe, Naik, Beach and Lanzkron32,Reference James, Bourgeois and Shannon33} and patient experiences^{Reference Pickner, Ziegler and Hanson28} in article discussion sections. These author-proposed explanations are based in discussion and do not have supporting data. One study asserted the possibility of overt discrimination as a cause for wait time differences.^{Reference James, Bourgeois and Shannon33} Authors finding differences in opioid prescribing suggested that physicians may be more worried about opioid abuse and diversion in certain racial/ethnic groups.^{Reference Singhal, Tien and Hsia31,Reference Joynt, Train, Robbins, Halterman, Caiola and Fortuna34}

Patient-provider interactions

Patient-provider interactions were posited to impact analgesia provision,^{Reference Pletcher, Kertesz, Kohn and Gonzales30,Reference Ortega, Velden, Lin and Reid35} diagnostics,^{Reference Levas, Dayan and Mittal20} and patient experiences.^{Reference Fields, Abraham, Gaughan, Haines and Hoehn26,Reference Pickner, Ziegler and Hanson28} Authors proposed that trust for providers may be affected by language barriers, as language differences between providers and patients make it difficult for the patient to articulate their concerns.^{Reference Fields, Abraham, Gaughan, Haines and Hoehn26,Reference Lee, Tamayo-Sarver, Kinneer and Hobgood27} Another proposed explanation for disparities in trust is a history of discrimination and mistreatment of minority groups.^{Reference Lee, Tamayo-Sarver, Kinneer and Hobgood27} Authors theorized that patients feel less enabled in decisions about health care when “social distance” is greater between patients and providers.^{Reference Lee, Tamayo-Sarver, Kinneer and Hobgood27}

Facility characteristics

Characteristics of EDs that minority groups present to were theorized to influence differences in triage scores^{Reference Lopez, Wilper, Cervantes, Betancourt and Green15} and wait times.^{Reference Sonnenfeld, Pitts, Schappert and Decker18} Hospitals where minority patients present may be more likely to be in urban centers, serve larger populations, have longer wait times, and have higher average acuity.^{Reference James, Bourgeois and Shannon33} Seven studies examined wait times within hospitals compared with between hospitals by race and ethnicity. Five (71%) of these studies found differences in wait times between races/ethnicities was present both within and between hospitals, suggesting that facility characteristics alone do not explain this disparity.^{Reference Sonnenfeld, Pitts, Schappert and Decker18}

Differences in clinical presentations

Numerous authors mention that differences in clinical presentation may impact care. Lower acuity at presentation due to lack of access to primary care was identified as a possible mechanism for differences in LWBS,^{Reference Wright24} wait times,^{Reference Sonnenfeld, Pitts, Schappert and Decker18} and diagnostics.^{Reference Levas, Dayan and Mittal20} Notably, select studies on wait times^{Reference Haywood, Tanabe, Naik, Beach and Lanzkron32,Reference James, Bourgeois and Shannon33} and diagnostics^{Reference Pezzin, Keyl and Green19,Reference Levas, Dayan and Mittal20} found disparities between clinically similar cohorts.

Provider judgment and standardized care

Analgesia, diagnostics, and treatments received by minority patients were theorized to be impacted by the degree to which care is standardized.^{Reference Shafi and Gentilello22,Reference Ortega, Velden, Lin and Reid35,Reference Bazarian, Pope, McClung, Cheng and Flesher36,Reference Dickason, Chauhan and Mor37} Authors found that where provider judgment plays a larger role, disparities in treatment were larger (e.g., analgesic prescribing for “subjective” pain, such as back pain, v. “objective” pain associated with renal calculi).^{Reference Singhal, Tien and Hsia31,Reference Dickason, Chauhan and Mor37} Conversely, authors theorize that a lack of disparities in diagnostics and treatment for trauma patients may be explained by standardization of trauma care, although this was not investigated.^{Reference Shafi and Gentilello22,Reference Bazarian, Pope, McClung, Cheng and Flesher36}