Differences in emergency medicine resident procedural reporting by race and ethnicity

Abstract

Background

The recruitment, retention, and training of physicians from groups underrepresented in medicine (UiM) is critically important to the practice of emergency medicine (EM). Studies across specialties have demonstrated disparities in operative experiences among UiM resident learners who are UiM; however, there are limited data on procedural disparities in EM.

Objective

We sought to quantify the association between racial and ethnic identities that are UiM and the number of procedures reported among EM residents.

Methods

We conducted a retrospective review of procedural differences by UiM status (using self‐identified race and ethnicity) among graduating EM residents at nine training programs over a 10‐year period. Sites were selected to ensure diversity of program length, program type, and geography. Data from residents in combined training programs, those who did not complete their full training at that institution, and those with missing data or electing not to report race/ethnicity were excluded. We calculated median and interquartile ranges for each procedure by UiM status. We conducted multivariable regression analyses accounting for UiM status, gender, and site as well as a sensitivity analysis excluding values >3 standard deviations from the mean for each procedure.

Results

We collected data from 988 total residents, with 718 (73%) being non‐UiM, 204 (21%) being UiM, 48 (5%) electing not to specific race/ethnicity, and 18 (2%) missing race/ethnicity data. While unadjusted data demonstrated a difference between UiM and non‐UiM resident numbers across several procedures, there were no significant differences in procedures reported after accounting for gender and site in the primary or sensitivity analyses.

Conclusions

We did not identify a statistically significant difference in reported procedures between UiM and non‐UiM residents in EM. Future work should include qualitative investigations of UiM resident experience surrounding procedures as well as mixed‐methods studies to examine how these data interact.

INTRODUCTION

There has been a growing recognition of the importance of increasing recruitment and retention of a diverse physician workforce including incorporation of physicians underrepresented in the medical field. It has been shown that physicians who are underrepresented in medicine (UiM) are more likely to practice in medically underserved areas, meeting the needs of a population that has decreased access to quality health care. ¹ Despite this growing recognition, only 6.1% of emergency medicine (EM) physicians are, for example, Hispanic or Latino compared with 18.7% of the U.S. population. ² , ³

Previous studies have identified bias in training environments; however, literature surrounding bias in procedural training is more limited. ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ In procedural‐based specialties, such as surgery and EM, procedural competence is required for independent clinical practice. This requirement is reflected in the Accreditation Council for Graduate Medical Education's (ACGME) key index procedures (KIPs) which defines the minimum number of core procedures that must be completed prior to graduation. ¹¹

While the KIPs serve as a minimum requirement, developing competence is more complex. Learning curves can vary across procedures and learners. Moreover, deliberate practice requires more focused efforts with larger numbers of procedures and structured training and feedback. ¹² , ¹³ , ¹⁴ , ¹⁵ As such, it is critical to identify any discrepancies in KIPs logged between UiM residents and non‐UiM residents. The presence of such disparities may have consequences preventing the ability for those residents to achieve maximum proficiency while in training.

In specialties outside of EM, literature has evaluated disparities in operative cases and procedure numbers for women and UiM residents. In general surgery, Eruchalu et al. ⁹ reported female and UiM residents performed fewer major surgical cases than male or non‐UiM residents, with a discrepancy also noted when comparing male UiM and male non‐UiM groups. However, a similar study in ophthalmology found no difference in cataract surgeries or procedures for women and UiM residents. ¹⁰ EM differs from these specialties given the unique environment with unpredictable procedural opportunities and smaller teams. To our knowledge, no prior literature has evaluated how these biases and variabilities may be operating within EM procedural education. The objective of this study was to quantify the associations between UiM and non‐UiM identities and the number of procedures reported among EM residents.

METHODS

Study design

This was a retrospective review of procedure logs from graduating residents at nine ACGME‐accredited EM residency programs from 2013 to 2022. The institutional review boards at all sites reviewed the study, determining it as either exempt or approved without required consent.

Study population

The sites were intentionally selected with attention to distribution by geography (both region and rural vs. urban), program length, program type (e.g., academic, community, county), and program size. Specific details regarding the characteristics of each program can be found in Table S1. All residents from all study sites who graduated from 2013 to 2022 were eligible for inclusion. Exclusion criteria included combined training programs (e.g., internal medicine/EM), incomplete training at a single institution (e.g., transfer in/out), or missing data.

Study protocol

The list of procedures studied was based on the ACGME KIPs for EM. ¹¹ Each site obtained resident‐specific totals for each procedure at the time of graduation, which was based on resident self‐report. Resident gender, race, and ethnicity were also collected based on resident self‐report (e.g., self‐declaration, Electronic Residency Application Service [ERAS] profile). Residents were categorized as UiM according to the definitions used in ERAS. ¹⁶ We defined UiM as residents who identified as Black/African American, Hispanic/Latino, Native American, or Pacific Islander/Native Hawaiian. We excluded data from residents who selected “prefer not to say” for race/ethnicity. Each site lead deidentified data and populated a standardized data collection spreadsheet.

Data analysis

We calculated median and interquartile ranges (IQR) for each procedure by UiM versus non‐UiM. We compared reported procedures by UiM versus non‐UiM status using linear regression. Comparison between UiM and non‐UiM groups was performed using multiple regression controlling for gender and institution. Three regression analyses were performed: (1) unadjusted, (2) adjusted for gender and institution (“adjusted”), (3) adjusted for gender and institution with values >3 standard deviations (SDs) from procedural means removed (“sensitivity analysis”). We performed all analyses using Stata version 16 (StataCorp, LLC).

RESULTS

We collected procedure log data from 988 residents, with 718 (73%) being non‐UiM, 204 (21%) being UiM, 48 (5%) electing not to specific race/ethnicity, and 18 (2%) missing race/ethnicity data. Our sample racial/ethnic distribution is similar to the proportion of residents who were UiM nationally (15.2%) as of the 2021–2022 academic year. ¹⁷ , ¹⁸ Median and IQR values by UiM status are depicted in Table 1. The unadjusted regression demonstrated higher predicted procedural counts for residents who were UiM for ultrasound, intubations, central lines, procedural sedations, lumbar punctures, and dislocation reductions. Those identifying as UiM had lower predicted vaginal deliveries and cricothyroidotomies (Figure 1). After adjusting for gender and institution, no procedures had statistically significant differences between UiM vs non‐UiM residents (Table 1, Figure 2). When we conducted a sensitivity analysis removing values >3 SDs from the mean, the findings remained consistent with the adjusted analysis (Figure 3).

TABLE 1.

Median and predicted procedural frequency by underrepresentation in medicine.

Procedure	ACGME requirement	UiM, median (IQR)	Non‐UiM, median (IQR)	Predicted procedures for residents who are UiM vs. non‐UiM, adjusted a (95% CI)	p‐value b
Point‐of‐care ultrasound	150	407.5 (245–575)	286 (181–431)	3.9 (−48.4 to 56.1)	0.89
Adult medical resuscitation	45	129.5 (89–192.5)	107 (74–174)	−5.7 (−38.6 to 27.3)	0.74
Intubation	35	82.5 (62.5–103)	71 (49–92)	0.9 (−4.8 to 6.5)	0.77
Adult trauma resuscitation	35	72.5 (56.5–102)	66 (42–94)	−4.1 (−15.6–7.4)	0.49
Central line	20	54 (34.5–71)	41 (27–54)	−1.3 (−5.1 to 2.6)	0.51
Pediatric medical resuscitation	15	28 (20–38)	22 (16–32)	−2.8 (−8.3 to 2.8)	0.33
Sedation	15	31 (20–40.5)	20 (15–29)	0.4 (−2.3 to 3)	0.79
Lumbar puncture	15	27.5 (18–36)	18 (16–23)	−1 (−2.7 to 0.7)	0.25
Pediatric trauma resuscitation	10	15 (12–19)	14 (10–22)	−0.7 (−2.9 to 1.4)	0.5
Dislocation reduction	10	21 (15–28)	16 (12–20)	0.7 (−1.2 to 2.5)	0.48
Chest tube	10	17 (14–22)	16 (12–21)	−0.9 (−2.7 to 1.0)	0.36
Vaginal delivery	10	11 (10–14)	13 (11–17)	−0.4 (−1.4 to 0.6)	0.39
Cardiac pacing	6	8 (7–10)	7 (6–9)	0.4 (−0.8 to 1.6)	0.49
Cricothyrotomy	3	5.5 (4–8)	6 (4–8)	0 (−1 to 1)	0.9
Pericardiocentesis	3	4 (3–5)	4 (3–6)	1.2 (−1.6 to 4.1)	0.4

Abbreviations: ACGME, Accreditation Council for Graduate Medical Education; UiM, underrepresented in medicine.

^a Regression coefficient controlled for institution and gender (95% CI). Positive values represent higher expected procedural counts for residents who are UiM.

^b p statistic for linear regression coefficient.

FIGURE 1.

Predicted differences in procedural counts by UiM status (regression coefficients and 95% CIs). UiM, underrepresented in medicine.

FIGURE 2.

Predicted differences in procedural counts by UiM status controlling for gender and institution (regression coefficients and 95% CIs). UiM, underrepresented in medicine.

FIGURE 3.

Predicted differences in procedural counts by UiM status with outliers excluded and controlling for gender and institution (regression coefficients and 95% CIs). *Excluded outliers are values >3 SD from the procedural mean. UiM, underrepresented in medicine.

DISCUSSION

This study assessed potential disparities in UiM EM resident procedural reporting compared to their non‐UiM peers. While there appeared to be differences in the absolute number of procedures between UiM and non‐UiM, these differences were no longer significant after accounting for gender and program. This is consistent with findings reported by Kong et al. ¹⁰ showing no difference in number of cataract surgeries or other procedures performed in ophthalmology residents.

It is reasonable to conclude that reduced access to procedures (or, perhaps, reduced reporting of those procedures) does not appear to be a feature of UiM marginalization in EM. This may be a result of differing access to procedures outside of the operating room and can be contingent on additional factors (e.g., procedural availability, presence of additional learners, severity of illness). Higher volumes of procedures combined with a small team size (often consisting of one to two residents paired with a single attending) in the ED setting may reduce the degree of bias in procedural opportunities.

Resident self‐selection into the field of EM may also have dampened differences in our study population compared to a more broad graduate medical education population. Prior literature has suggested that certain behaviors and personality characteristics such as procedural aggressiveness, assertiveness, and confidence can be influential factors in gaining access to procedures. ¹⁹ Persons who select EM as a specialty may possess these characteristics to a greater degree than the general graduate medical trainee population. ²⁰ However, this notion is further complicated by the experiences reported by Black female physicians, for example, regarding the negative consequences surrounding perceptions of “anger” and “intimidation” when attempting to lead and assert themselves. ²¹

Though it is tempting to consider these data as evidence of seeming equity within EM, one must consider the well‐documented impact of marginalization on UiM learners, which behooves us to take a closer look. ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ In keeping with the principles of antiracism, rather than concluding that an absence of difference suggests an absence of racism, we must instead inquire, “How is racism operating here?” ³⁰ For example, though our study examines discrepancies in procedural counts, it does not investigate the actual experience of UiM residents in the procedural context compared to non‐UiM residents. Even if the numbers are similar, there may be issues related to autonomy, inclusion in only “easy cases,” or other problems that have yet to be described. Future research directions should further elucidate this using additional models, such as a qualitative investigation of the UiM resident procedural experience or direct assessment via simulated or recorded cases.

LIMITATIONS

Our study was limited by relying on a more limited definition of UiM. The American Association of Medical Colleges has expanded to a broader definition consisting of “racial and ethnic populations that are underrepresented in the medical profession relative to their numbers in the general population.” ³¹ Given the breadth of races reported by applicants, we were not able to match all of them against census data and it is possible that we may have misclassified some residents who met the more expansive UiM criteria. Although we attempted to achieve varied representation in program length, type, and geography, our study population may not fully reflect the true distribution of EM residents and may not be generalizable to programs with significantly different characteristics. In addition, our data are limited by self‐report. Residents who preferred not to declare their racial or ethnic identity were not included in our data set and, as a result, it is impossible to know how their inclusion would have impacted these findings. Furthermore, our data may underrepresent the total number of procedures performed by residents as logging may fall off once minimum requirements are met. Certain sites were also unable to reliably determine which procedures were completed in a simulated versus clinical environment. The challenges and limitations faced by UiM residents in the clinical realm may differ from those faced in simulated environments, and it is possible we did not detect a difference in procedures performed in the clinical environment.

CONCLUSIONS

We found no significant difference in the number of logged procedures between underrepresented in medicine and non–underrepresented in medicine residents; however, more data are needed to further elucidate how racism may be impacting resident emergency medicine learners' procedural experiences.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

Supporting information

Table S1.

Arno K, Bradby C, Shappell E, et al. Differences in emergency medicine resident procedural reporting by race and ethnicity. AEM Educ Train. 2024;8:e10930. doi: 10.1002/aet2.10930