a b s t r a c t

Objective: We hypothesized that resident characteristics impact patterns of patient self-assignment in the emer- gency department (ED). Our goal was to determine if male residents would be less likely than their female col- leagues to see patients with sensitive (e.g. breast-related or gynecologic) chief complaints (CCs). We also investigated whether resident specialty was associated with preferentially choosing patients with more familiar chief complaints.

Methods: We performed a retrospective cross-sectional study at a tertiary academic medical center using data from all adult patients presenting to the ED between 2010 and 2019 with one of six CC categories (vaginal bleed- ing, breast-related concerns, male genitourinary [GU] concerns, gastrointestinal bleeding, epistaxis, and lacera- tion). These CCs were chosen as they each require either an invasive medical exam or procedure, and cannot easily be evaluated with an exam in a hallway bed. We used logistic regression to assess the likelihood of being treated by a male resident compared to a female resident for each CC, adjusting for candidate variables of patient age, race, primary language, ESI score, bed location, time of day, day of week, calendar month, and res- ident specialty. We also similarly analyzed patterns of patient self-assignment according to resident specialty. Results: Male residents were significantly less likely than female residents to treat patients with breast-related CCs (adjusted OR 0.67, 95% CI 0.54-0.83, p < 0.001) or Vaginal bleeding (adjusted OR 0.73, 95% CI 0.63-0.84, p < 0.001, reference group: epistaxis). Off-service residents were more likely to assign themselves to familiar chief complaints, for example surgery residents were more likely to see patients with lacerations (adjusted OR 2.11, 95% CI 1.71-2.61, p < 0.001) and OB/GYN residents were less likely to see patients with male GU concerns (adjusted OR 0.21, 95% CI 0.05-0.85, p = 0.029), compared to emergency medicine residents.

Conclusion: In a single facility, resident characteristics were associated with preferential patient self-assignment. Further work is necessary to determine the underlying reasons for patient avoidance, and to create work environ- ments in which preferentially choosing patients is discouraged.

(C) 2021

1. Introduction

Prioritizing and evaluating patients appropriately are key milestones for emergency resident education [1]. Many emergency departments (EDs) rely on residents to assign themselves to patients at their own dis- cretion. The decision to pick up a patient may be made based on an in- ternal calculation of perceivED capacity to see a new patient, patient acuity, overall departmental volume, time remaining on shift, and other factors [2].

Recent evidence has corroborated concerns for cherry-picking be- havior, e.g. that patient prioritization may be influenced by physician preference [3]. In one study including over 250 emergency department chief complaints (CCs), there was significant variation in pickup time by

* Corresponding author.

E-mail address: [email protected] (I. Agarwal).

¹ Both authors contributed equally as senior authors.

CC, with the longest times for nebulous or complex CCs such as abdom- inal pain and numbness/tingling, and the shortest times for CCs with perceived quick or simple dispositions e.g. ankle injury, allergic reaction, and wrist injury [3]. Among the CCs with the longest pickup time was vaginal bleeding. Length of time between patient rooming and res- ident self-assignment for vaginal bleeding was found to be double the median of all other emergency department chief complaints (12 min vs. 6 min) [3].

The factors that influence patient self-selection are not well-studied, but are critical to understand. Particularly in academic training environ- ments, residents are learning how to prioritize patients while balancing efficiency and fairness. Due to preferential patient self-assignment, some emergency departments have abandoned self-assignment in favor of rotational patient assignment [2].

We hypothesized that resident characteristics, including sex and pri- mary specialty, could impact patterns of patient self-assignment. Specif- ically, we asked whether male residents would be less likely than female

https://doi.org/10.1016/j.ajem.2021.01.081

0735-6757/(C) 2021

Table 1

Patient and resident characteristics, by category of chief complaint.

	Vaginal bleeding	Breast concern	Male GU concern	GI bleeding	Epistaxis	Laceration
	(n = 3526)	(n = 513)	(n = 952)	(n = 5088)	(n = 1245)	(n = 4502)
Patient characteristics
Age (median, IQR)	37 (28, 54)	41 (31, 54)	36 (26, 55)	64 (49, 76)	67 (53, 79)	33 (23, 55)
Acuity (ESI score) (n, %)
1	93 (3)	3 (1)	7 (1)	158 (3)	20 (2)	68 (2)
2	1140 (32)	42 (8)	349 (37)	3324 (65)	292 (23)	541 (12)
3	2240 (64)	419 (82)	552 (58)	1600 (31)	828 (67)	2293 (51)
4	50 (1)	49 (10)	44 (5)	5 (0)	103 (8)	1567 (35)
5	3 (0)	0 (0)	0 (0)	1 (0)	2 (0)	33 (1)
Race (n, %)
White	1481 (42)	205 (40)	540 (57)	3246 (64)	802 (64)	2967 (66)
African American	1154 (33)	174 (34)	200 (21)	985 (19)	251 (20)	628 (14)
Hispanic/Latino/Caribbean	420 (12)	64 (12)	106 (11)	371 (7)	82 (7)	323 (7)
Asian	227 (6)	21 (4)	34 (4)	276 (5)	69 (6)	301 (7)
Native American/Hawaiian	13 (0)	0 (0)	1 (0)	22 (0)	2 (0)	12 (0)
Unknown/other	231 (7)	49 (10)	71 (7)	188 (4)	39 (3)	271 (6)
English as primary language (n, %)	3060 (87)	443 (86)	849 (89)	4505 (89)	1080 (87)	4162 (92)
In hallway bed (n, %)a	210 (6)	39 (8)	82 (9)	558 (11)	214 (17)	984 (22)
Resident characteristics
Door-to-resident time (min) (median, IQR)	49 (21, 97)	54 (26, 101)	40 (16, 82)	49 (20,101)	39 (17, 78)	37 (16, 80)
Male sex of resident (n, %)	2215 (63)	313 (61)	726 (76)	3536 (70)	872 (70)	3229 (72)
Shift assigned to resident (n, %)
7 am - 3 pm	1149 (33)	184 (36)	331 (35)	1920 (38)	475 (38)	1158 (26)
3 pm - 11 pm	1571 (45)	236 (46)	414 (43)	2260 (44)	443 (36)	2159 (48)
11 pm - 7 am	806 (23)	93 (18)	207 (22)	908 (18)	327 (26)	1185 (26)
Primary specialty of resident (n, %) Emergency Medicine	2785 (79)	397 (77)	793 (83)	3977 (78)	100 (80)	3785 (84)
Medicine	609 (17)	94 (18)	138 (15)	1033 (20)	207 (17)	507 (11)
OB/GYN	51 (1)	7 (1)	2 (0)	32 (1)	11 (1)	42 (1)
Surgery	81 (2)	15 (3)	19 (2)	46 (1)	27 (2)	168 (4)

^a 125 beds are unknown/missing (15 vaginal bleeding, 1 breast, 11 male GU, 10 GI bleeding, 5 epistaxis, 83 laceration).

residents to assign themselves to patients with breast-related or gyne- cologic CCs, and whether resident specialty would impact the CCs resi- dents assigned to themselves. Evidence of preferentially choosing patients could support switching from patient self-assignment to rota- tional assignment, to ensure a more egalitarian patient assignment process.

1. Methods

We performed a retrospective cross-sectional study at an academic medical center using data from all adult patients seen in the ED between 2010 and 2019 with one of six CC categories (vaginal bleeding, breast- related concerns, male genitourinary [GU] concerns, gastrointestinal [GI] bleeding, epistaxis, and laceration). Breast-related concerns in- cluded CCs such as breast pain, nipple discharge, and breast mass. Male GU concerns included CCs such as testicular or penile pain and/ or discharge. The CCs included in this study were chosen because they each require either an invasive medical exam or procedure and cannot easily be evaluated with an exam in a hallway bed. As the study in- volved the use of de-identified data, the host institutional review board granted a waiver of informed consent for data use.

Visits were excluded if no resident was involved in the patient care. Most residents (~80%) working in our ED are emergency medicine (EM). There are also off-service residents from other specialties, includ- ing internal medicine (IM), surgery, and OB/GYN. Pediatric EM fellows rotating in the department were grouped with EM residents. Approxi- mately 2/3 of the EM residents at our institution are male, with some year-to-year variation. Residents work 8- or 9-h shifts. Staffing is highest during peak hours (up to 7 residents) and lowest during off- peak hours (as few as 2 residents).

With the exception of critical “trigger” patients, which are automat- ically assigned to a second-year EM resident, residents are expected to self-assign themselves to patients in order of wait time in a room. While picking up patients out of order is strongly discouraged (and tracked), it is possible to delay picking up a patient if an undesirable pa- tient is at the top of the list, until another resident has assigned themself to that patient.

Time stamps of all patient encounters were automatically logged. The following data were abstracted for each patient visit: CC, time of ar- rival, time of resident self-assignment, age, race, primary language, ESI, bed location, day of week, calendar month, resident sex, and resident specialty. Pickup time was defined as the time interval (in minutes) be- tween arrival to the ED and when a resident physician assigned

Table 2

Median minutes from door-to-resident by chief complaint and resident sex.

Chief complaint	Overall (n = 15,824)	Male resident (n = 10,889)	Female resident (n = 4935)	p-value comparing males to females
Breast complaint	54 (26, 101)	52 (26, 102)	57 (26, 100)	0.724
Vaginal bleeding	49 (21, 97)	47 (20, 95)	53 (22,100)	0.038
Male GU complaint	40 (16, 82)	41 (16, 82)	38 (15, 82)	0.672
GI bleeding	49 (20, 101)	48 (20,100)	51 (21,102)	0.368
Epistaxis	39 (17, 78)	37 (16, 80)	42 (18, 73)	0.397
Laceration	37 (16, 80)	37 (15, 80)	37 (17, 79)	0.822

Table 3a

Odds ratio for being treated by a male physician (all residents), by chief complaint.

Chief complaint	OR (95% CI)	p-value	aOR (95% CI)	p-value
Epistaxis	Reference		Reference
Vaginal bleeding	0.72 (0.63-0.83)	<0.001	0.73 (0.63-0.84)	<0.001
Breast concern	0.67 (0.54-0.83)	<0.001	0.67 (0.54-0.83)	<0.001
Male GU concern	1.37 (1.13-1.67)	0.001	1.36 (1.12-1.65)	0.002
GI bleeding	0.97 (0.85-1.12)	0.709	0.98 (0.85-1.12)	0.743
Laceration	1.09 (0.95-1.24)	0.245	1.07 (0.93-1.23)	0.340

Bold values are statistically significant at the threshold of p < 0.05.

Table 3b Odds ratio for being treated by a male physician (emergency medicine residents only), by chief complaint.

Chief complaint	OR (95% CI)	p-value	aOR (95% CI)	p-value
Epistaxis	Reference		Reference
Vaginal bleeding	0.82 (0.70-0.96)	0.015	0.84 (0.71-0.98)	0.028
Breast concern	0.78 (0.61-1.00)	0.054	0.79 (0.61-1.02)	0.066
Male GU concern	1.41 (1.13-1.76)	0.002	1.39 (1.12-1.74)	0.003
GI bleeding	1.09 (0.94-1.28)	0.255	1.10 (0.94-1.29)	0.236
Laceration	1.11 (0.95-1.30)	0.180	1.11 (0.94-1.29)	0.214

Bold values are statistically significant at the threshold of p < 0.05.

themself to the patient. CCs meeting our criteria were identified using our existing concept-based ontology [4]. All analysis was performed using STATA 14.2.

We performed a Wilcoxon rank sum test (for non-normally distrib- uted data) to compare the median pickup time for each of the six CCs in our analysis, as well as to compare pickup time by sex. We used adjusted logistic regression to compare the ratio of male to female residents for each CC. We a priori chose the candidate variables of patient age, race, primary language, bed location, Emergency Severity Index , time of day, day of week, calendar month, and resident specialty and used likelihood ratio tests to create a parsimonious model. In our primary analysis, the CC of epistaxis was used as our reference category. In sen- sitivity analysis, we restricted our analysis to EM residents only (e.g. ex- cluded off-service residents). We also conducted a sensitivity analysis excluding high acuity ESI1 patients that may have been automatically assigned.

We conducted a multivariate logistic regression using the same can- didate variables above to compare the ratio of EM vs. IM, surgery, and OB/GYN residents for each CC.

1. Results

A total of 15,826 visits were included in this study. There was a sig- nificant difference in the overall door-to-resident time between CCs (p < 0.001), with breast-related concerns and vaginal bleeding having among the longest door-to-resident times (Table 1). For vaginal

bleeding, the door-to-resident time for female residents was longer than for male residents (female: median 53 min (IQR: 22, 100), male: median 47 (IQR: 20, 95), p = 0.038, Table 2).

Patients with breast-related concerns or vaginal bleeding were less likely to be seen by a male resident than a female resident (breast: ad- justed OR [aOR] 0.67, 95% CI 0.54-0.83, p < 0.001; vaginal bleeding: ad- justed OR 0.73, 95% CI 0.63-0.84, p < 0.001) compared to a reference group of patients with epistaxis (Table 3a). Patients with male GU con- cerns were more likely to be seen by a male resident than a female res- ident (aOR 1.36, 95% CI 1.12-1.65, p = 0.002). These results were not significantly affected by excluding off-service residents and restricting the analysis to EM residents (Table 3b). Our results were robust to the exclusion of high acuity ESI1 patients that may have been automatically assigned.

We found significant effects of resident specialty on patient prefer- ence, with off-service residents more likely to assign themselves to pa- tients with more familiar chief complaints (Table 4). For example, OB/ GYN residents were statistically more likely to pick up patients with vaginal bleeding compared to residents from other services (aOR 1.94, 95% CI 1,38-2.73, p < 0.001), and less likely to pick up a patient with male GU concerns (aOR 0.21, 95% CI 0.05-0.85, p = 0.029). Residents from a surgical background were more likely to assign themselves to pa- tients with lacerations (aOR 2.11, 95% CI 1.71-2.61, p < 0.001), while IM residents were less likely to assign themselves to these patients (aOR 0.58, 95% CI 0.52-0.64, p < 0.001).

1. Discussion

In our ED, resident characteristics were significantly associated with patterns of patient self-assignment. Our study builds upon prior work showing cherry-picking behavior in other settings [5].

We found that male residents were less likely to see patients with breast-related concerns and vaginal bleeding compared to other similar studied CCs. There are several potential explanations for these findings. Male residents may be more uncomfortable than their female col- leagues with performing breast or pelvic exams in the ED. They may be looking to avoid uncomfortable patient interactions, or they may be less confident with their own ability to perform the exam. Male medical students on OB/GYN rotations self-report significantly higher discrimi- nation against their gender by residents and attending physicians and also report a higher rate of patients declining consent for internal pelvic examination [6,7]. In the ED, breast and pelvic exams often require a chaperone (particularly for male providers), which can be logistically difficult to coordinate during a busy shift. Given these and other issues, female residents may unfairly feel pressure to see patients requiring a breast or pelvic exam.

Given that the majority of OB/GYN residents are female, this could have contributed to the association of female sex and patient self- assignment that we observed. However, excluding OB/GYN residents and limiting our study to EM residents did not have a statistically signif- icant effect on our conclusions.

Table 4

Adjusted odds of treating chief complaint by resident specialty.

Primary specialty	Vaginal bleeding		Breast concern		Male GU concern		GI bleeding		Epistaxis			Laceration
	OR (95% CI) p-value		OR (95% CI) p-value		OR (95% CI) p-value		OR (95% CI) p-value		OR (95% CI)	p-value		OR (95% CI) p-value
Emergency	Ref.		Ref.		Ref.		Ref.		Ref.			Ref.
Medicine
Internal Medicine	1.10 0.063		1.17 0.175		0.85 0.083		1.46 <0.001		1.02	0.800		0.58 <0.001
	(0.99-1.22)		(0.93-1.47)		(0.70-1.02)		(1.34-1.60)		(0.87-1.19)			(0.52-0.64)
OB/GYN	1.94 <0.001		1.58 0.244		0.21 0.029		0.62 0.019		0.96	0.906		0.96 0.844
	(1.38-2.73)		(0.73-3.39)		(0.05-0.85)		(0.42-0.93)		(0.52-1.79)			(0.76-1.38)
Surgery	1.05 0.690		1.37 0.244		0.85 0.493		0.33 <0.001		0.96	0.853		2.11 <0.001
	(0.82-1.35)		(0.81-2.32)		(0.53-1.36)		(0.24-0.45)		(0.65-1.43)			(1.71-2.61)

We found residents from different training backgrounds (e.g. inter- nal medicine, OB/GYN, and surgery) were likely to pick up patients with more familiar chief complaints. While this may be most comfort- able for residents, residency programs have a duty to provide residents with a well-balanced education and expose them to a range of patients [8]. Off-service residents working the ED should be encouraged to see a variety of CCs that they might not normally be exposed to [5]. Recogniz- ing potential bias towards seeing patients with more familiar CCs is a first step towards ensuring a balanced education for residents rotating in the ED.

Rotational patient assignment, in which predetermined criteria are used to assign patients to physicians or teams, has been proposed as an alternative model to physician self-assignment in order to counteract tendencies to preferentially choose patients [2]. In one ED, changing from self-assignment to rotational assignment led to a 44% decrease in door to physician time, a 14% decrease in ED length of stay, and a 40% decrease in patient complaints [2]. These findings raise the question of whether rotational patient assignment may be a better system than self-assignment for resident physicians, both for improving departmen- tal flow and for optimizing resident learning.

1. Limitations

Due to the limitations of our data set, we were unable to adjust our data by resident training level, the number of residents working in the department, or the male to female ratio of residents working in the de- partment at any given time. Furthermore, because our analysis was lim- ited to six candidate CC categories, we are unable to speculate more broadly on how resident sex affects cherry-picking behavior beyond the CCs included. Finally, our manuscript does not account for non- binary gender identification, as residents were classified according to bi- nary sex designations (male and female).

1. Conclusion

In a single academic center, male residents were less likely than fe- male residents to assign themselves to see patients with breast- related or gynecologic complaints compared to the other CCs we stud- ied. Off-service residents were also more likely to pick up patients

with familiar CCs. Whether this behavior is due to inadequate training, personal bias, discomfort with sensitive examinations, or other reasons remains to be determined. In the future, we hope to explore rotational patient assignment as a means to mitigate preferential patient selection.

Prior presentations

Accepted at SAEM 2020 (cancelled due to COVID-19).

Funding sources/disclosures

None.

Declaration of Competing Interest

We declare no conflicts of interest.

Acknowledgements

None.

References

1. The Accreditation Council for Graduate Medical Education. The emergency milestone project. East Lansing, MI: The American Board of Emergency Medicine; 2015.
2. Traub, et al. Emergency department rotational patient assignment. Ann Emerg Med. 2, Feb 2016;67.
3. Patterson, et al. Cherry picking patients: examining the interval between patient rooming and resident self-assignment. Acad Emerg Med. 6, June 2016;23.
4. Greenbaum, et al. Improving documentation of presenting problems in the emer- gency department using a domain-specific ontology and machine learning-driven user interfaces. Int J Med Inform. 2019 Sep, 27;132 ePub.
5. Whiteman, Foltin. physician bias during patient selection in the pediatric emergency department. Pediatr Emerg Care. 1998;14:5.
6. Wallbridge, et al. Does medical students’ gender affect their clinical learning of gynaecological examination? A retrospective cohort study. Postgrad Med J. 11, June 2018;94.
7. Zahid, et al. Gender bias in training of medical students in obstetrics and gynaecology: a Myth or reality? Eur J Obstet Gynecol Reprod Biol. Mar 2015;186.
8. Wartman, et al. Barriers and conflicts. The service/education conflict in residency pro- grams: a model for resolution. J Gen Intern Med. 1990;5(Suppl):S59-69.