Original Contribution

Factors affecting stress in emergency medicine residents while working in the ED

Keith Wrenn MD ^a,?, Brent Lorenzen BS ^a, Ian Jones MD ^a,b,

Chuan Zhou PhD ^c, Dominik Aronsky MD, PhD ^a,b

^aDepartment of Emergency Medicine, Vanderbilt University School of Medicine, Nashville, TN 37232-4700, USA ^bDepartment of BioMedical Informatics, Vanderbilt University School of Medicine, Nashville, TN 37232-4700, USA ^cDepartment of Biostatistics, Vanderbilt University School of Medicine, Nashville, TN 37232-4700, USA

Received 20 March 2009; revised 16 April 2009; accepted 11 May 2009

Abstract

Objectives: The aim of this study was to identify factors other than work hours in the emergency department (ED) work environment contributing to resident stress.

Methods: This study involved a prospective cohort evaluation of emergency medicine residents in the ED. Twelve surveys were collected from 18 subjects, 4 each from the day, evening, and night shifts. The perceived stress Questionnaire and a visual analog stress scale were administered. Data collected included the shift number of a given consecutive sequence of shifts, number of procedures performed, number of adverse events, average age of the patients seen by the resident, triage nurse-assigned acuities of the patients seen by the resident during the shift, the number of patients seen during a shift, the number of patients admitted by the resident during the shift, anticipated overtime after a shift, and shift-specific metrics related to overcrowding, including average waiting room time both for the individual residents and for all patients, average waiting room count for all patients, and average occupancy of the ED for all patients.

Results: Among the 216 studiED shifts, there was considerable variability in stress both within and between residents. In the multivariate mixed-effect regression analysis, only anticipated overtime and process failures were correlated with stress. Factors related to ED overcrowding had no significant effect on resident stress.

Conclusions: Resident stress was most impacted by anticipation of overtime and adverse events. Overcrowding in the ED and traditional measures of workload did not seem to affect stress as much.

(C) 2010

Introduction

Residency is stressful, and it may contribute to family problems [1-3], burnout [1-5], anger [6,7], mood disturbances

* Corresponding author. Tel.: +1 615 936 1157; fax: +1 615 936 1316.

E-mail address: [email protected] (K. Wrenn).

[6,8-10], and even posttraumatic stress disorder [11]. Multiple studies have demonstrated that fatigue, Sleep deprivation, and excessive hours on call contribute to stress, accidents, conflict with other health care workers, unhealthy lifestyle, and Medical errors [9,12,13]. Duty hour restrictions were intro- duced by the Accreditation Council for Graduate Medical Education in July 2003. The restrictions have helped alleviate chronic sleep deprivation and improved residents’ mood [8].

0735-6757/$ - see front matter (C) 2010 doi:10.1016/j.ajem.2009.05.001

Many other stressors exist for those in residency training. Indebtedness, financial pressures within academic medical centers, threats of residency closures, and job availability are some of these [10]. There are also role-related stressors such as heavy workloads, fear of mistakes, subordinate status, public humiliation, handling life-and-death situations, over- whelming clerical and administrative responsibilities, distrac- tions and interruptions in work flow, frustrated hopes and expectations, delay of gratification, moral conflicts with sense of duty, competing demands of others, safety issues, difficulties in balancing concern and detachment, and dealing with difficult patients, bureaucracy, dysfunctional systems, and political issues [2,5,7,12-18]. Attending physician stress is likely to be somewhat different for many reasons, not the least of which are higher salaries, greater Knowledge base, greater autonomy, generally lighter workload depending on experience, and academic vs private practice.

Recently, in emergency departments (EDs), overcrowding due to increasED volumes and the boarding of inpatients has become commonplace [19]. When patients wait or leave without being seen, it is a stress on the system. It is reasonable to think that it would be a stress on the providers who practice in EDs.

It is clear that there are environmental issues other than hours that contribute to stress [4,7,10,17,20,21]. The purpose of this study was to identify factors in our ED work environment that contribute to stress. We specifically wanted to examine how the issue of overcrowding impacts emergency medicine (EM) resident stress. We also postu- lated that EM residents doing shift work in our ED would have less stress related to hours and sleep deprivation because of the tightly regulated duty hours [22]. We therefore undertook a study to evaluate how Environmental factors occurring during a shift might impact EM resident stress level.

Methods

This was a prospective cohort evaluation of stress levels in 18 postgraduate year (PGY)-2 and PGY-3 EM residents while working shifts in the ED. The study was approved by the institutional review board of Vanderbilt University Medical Center.

Twelve surveys and questionnaires were collected from each subject, 4 each from the day, evening, and night shifts. Resident shifts were scheduled as 8 to 10 hours long and all involved staffing the higher acuity section of the ED. This area was chosen because of the uniformity of shift lengths and the types of patients seen, although patients of lower acuity were seen at times. The lower Acuity area had 12-hour shifts and much wider acuity variability. Information from each resident was de-identified and assigned a code number for tracking purposes. The research assistant (B.L.), a second year medical student with no other connection to the department, administered the survey and was the only one

who knew the tracking number. He was not involved in any sort of evaluation of residents. All investigators except the research assistant were blinded to the particular resident data. This ensured that subjects would not feel the pressure to alter reported stress out of concern that the measurements could potentially affect departmental evaluations.

This study occurred in the adult ED at a tertiary care level 1 trauma center with an annual volume of 55 000 patients. Data were collected between June 23, 2005, and May 30, 2006. Eligible participants included 10 PGY-2 and 8 PGY-3 EM residents in a PGY 1-3 residency program. We excluded a priori PGY-1 residents because it was felt that they were exposed to additional stressors associated with being an intern and also had only 4 ED months involving shifts during the year and more off-service obligations. The PGY-2 and PGY-3 residents have a much more uniform ED experience with 7 to 9 months of shifts in the ED. Before beginning the study, a research assistant met with the eligible subjects and explained the purpose of the study, the procedures that would be used for data collection, and potential risks and benefits of the study. The residents were not blinded to the fact that this was a study of ED stressors. All eligible subjects agreed to participate and signed an informed consent document.

Shifts were chosen at the convenience of the investigator (B.L.) collecting the data. Resident shifts are assigned during a month based on prior limited requests for specific time off. Shift numbers, types of shifts, and weekends/holiday shifts are distributed quite evenly among residents. At the scheduled end of a shift, the investigator had a subject complete the survey and questionnaire in the ED. The survey consisted of a modified version of the previously validated Perceived Stress Questionnaire (PSQ) [23] normalized to a possible range of 0 to 100 and, because of the modifications, a rating of self-perceived stress where subjects placed a mark on a 100-mm Visual analog scale labeled “none” at one end and “extreme” at the other. The survey took less than 5 minutes to complete.

The questionnaire reflected information about events occurring during the shift. Workload demographics included the shift number of a given sequence of shifts over consecutive days and the number of procedures performed, such as rapid- sequence intubation, lumbar puncture, central line placement, laceration repair, and splinting. Also included were the number of adverse events such as laboratory delays; imaging delays; consultation delays; unsuccessful procedures; negative inter- actions with patients, consultants, staff, or attending physi- cians; negative patient outcomes; unexpected deaths; or serious complications. Finally included was how long after the scheduled end of the shift the resident expected to stay to complete the necessary work (anticipated overtime).

The ED information system [24] provided data about the patients for whom the participants assumed primary responsibility during their shifts, including average waiting room (WR) time, average WR count, average length of stay (LOS), average patient age, triage nurse-assigned Acuity levels as measured by the 5-level Emergency Severity Index

Variable Mean (SD)	Median (IQR)		Range
Individual resident information during shifts
Total patients during shift 8.77 (2.94)	9	(7-11)	4-19
Patients per hour 1.09 (0.35)	1.13	(0.85-1.38)	0.5-1.9
Shifts in a row 3.67 (1.66)	4	(2-5)	0-10
Anticipated overtime (min) 47.02 (30.12)	45	(30-60)	0-180
Consultations during shift 3.01 (1.86)	3	(2-4)	0-9
Procedures 0.78 (1.02)	0	(0-1)	0-5
Process failures 1.07 (1.43)	1	(0-2)	0-7
ED operational characteristics during shifts-specific to residents
Average WR time (min) 29.67 (32.55)	16.5	(8-38)	2-172
Average age of patients (y) 43.85 (7.93)	43.82	(38.95-48.91)	11.50-65.11
Average Acuity of patients 2.32 (0.30)	2.33	(2.13-2.50)	1.5-3.5
Patients admitted 3.69 (1.91)	4	(2-5)	0-8
Overall operational characteristics during shifts
Average ED occupancy (%) 79.60 (15.73)	81.47	(69.76-93.02)	26.34-103.40
Average WR time (h) 1.19 (3.42)	0.55	(0.23-1.16)	0.01-32.09
Average WR count 4.48 (4.41)	2.94	(1.26-6.18)	0.23-19.57
Average ED LOS (h) 6.60 (2.99)	5.90	(4.56-7.75)	1.95-20.10
IQR indicates interquartile range.

version 3 (ESI) [25], and the number of patients admitted to the hospital. Because exposure to patients with severe conditions may lead to increased stress, we classified ESI level 1 patients as critically ill and ESI level 2 to 5 patients as not critically ill. To characterize the overall ED activity during a participant’s specific shift, we queried the ED information system as to the overall WR count and average ED occupancy rate as defined by the percentage of occupied patient rooms divided by the number of licensed beds (20 in the higher acuity area and 46 total). There are times when patients are put into 8 unlicensed hallway beds (4 in the higher acuity area) and the occupancy rate can exceed 100%. Descriptive statistics were calculated for each variable, including means, SDs, medians, and interquartile ranges. Because one of the primary objectives of the study was to evaluate the immediate impact of ED process on residents’ stress level, the authors identified 2 ED factors, average ED patients’ LOS and average WR patient counts during shifts, as primary covariates in the analysis. These variables were

Table 1 Shift-specific characteristics for 18 resident physicians during 256 shifts

included in the models regardless of their significance.

The authors carefully identified a list of ED variables that were potential predictors of stress. The list included variables related to workload and stressful events. Because of the relatively small sample size and potential correlation among many of the variables, the authors developed a strategy for selecting variables for the final models, which was a combination of careful reasoning of clinical significance and univariate screening. Each variable was analyzed using a univariate random effects model, and the variables with significance for either VAS or PSQ were used for the final multivariate analysis. In addition, variables not statistically significant but deemed scientifically important were also kept in the model.

The repeated measures allowed us to have a larger sample size and the ability to assess the within-resident variability. Because repeated measures were taken for the same types of shifts within the same resident, the outcomes would most likely be correlated with residents and shifts. Thus, for the regression analysis, we used linear mixed-effects models with a resident-specific random intercept to adjust for the nesting shift-in-resident correlation structure. The parameter estimation applied a restricted maximum likelihood method.

Fig. 1 PSQ measurements of individual resident stress.

Variable	VAS				PSQ
	Estimate	95% CI	P		Estimate	95% CI	P

Shift afternoon	-4.35	-10.88 to 2.19	.1912	-2.56	-7.33 to 2.22	.2925
Shift night	-9.10	-15.63 to -2.56	.0066 ?	-2.90	-7.68 to 1.87	.2320
Total patients during shifts	1.61	0.71 to 2.51	.0005 ?	0.35	-0.32 to 1.01	.3058
Patients per hour	12.89	5.43 to 20.36	.0009 ?	1.82	-3.73 to 7.36	.5196
Shift number	-0.54	-2.54 to 1.45	.5927	0.24	-1.20 to 1.69	.7427
Shifts in a row	-1.27	-2.89 to 0.35	.1237	-0.41	-1.58 to 0.75	.4890
Procedures	0.21	-2.43 to 2.86	.8741	2.16	0.26 to 4.06	.0259 ?
Process failures	6.95	5.30 to 8.61	b.0001 ?	3.90	2.64 to 5.17	b.0001 ?
Anticipated overtime	0.25	0.17 to 0.34	b.0001 ?	0.16	0.10 to 0.22	b.0001 ?
Consultations during shifts	1.45	0.01 to 2.90	.0489 ?	0.86	-0.18 to 1.90	.108
Average WR time	-0.01	-0.09 to 0.07	.7834	-0.02	-0.08 to 0.04	.5206
Average patient age	0.14	-0.20 to 0.48	.4295	0.03	-0.22 to 0.27	.8324
Average patient acuity	-9.68	-18.61 to -0.75	.0337 ?	-3.32	-9.81 to 3.17	.3140
Patients admitted	2.61	1.24 to 3.99	.0002 ?	1.06	0.04 to 2.07	.0413 ?
Average occupancy	0.04	-0.14 to 0.21	.683	-0.08	-0.20 to 0.05	.2135
Average WR time	-0.16	-0.95 to 0.63	.6891	-0.21	-0.79 to 0.36	.4597
Average LOS	-1.27	-2.16 to -0.39	.0053 ?	-0.91	-1.55 to -0.28	.0055 ?
Average WR count	0.69	0.08 to 1.30	.0258 ?	-0.03	-0.47 to 0.41	.8934
* P b .05.

We report both P values and confidence intervals. A 2-sided P value less than .05 was considered statistically significant. Statistical analyses were performed using the R statistical software [26].

Table 2 Results of univariate analysis of all variables with VAS and PSQ scores

Results

All 18 residents completed 100% of the surveys for a total of 216 shifts. There were 7 women and 3 men in the PGY-2 group and 3 women and 5 men in the PGY-3 group. The ages ranged from 26 to 37 years. There were 6 PGY-3 residents and 5 PGY-2 residents who were married. There were 2

PGY-3 residents with children and 3 PGY-2 residents with children. Table 1 displays the shift-specific characteristics of the participant group. The triage-assigned patient acuities for the patients seen by the 18 residents were level 1, 6%; level 2, 61%; level 3, 28%; level 4, 4%; and level 5, 1%.

The mean PSQ and VAS scores did not differ signifi- cantly: 25 (95% confidence interval [CI], 5-51) and 23 (95% CI, 0-63), respectively (P = .408). Stress scores varied considerably among residents. Within-resident stress demon- strated high variability from shift to shift. The highest range was from 3 to 70, whereas the lowest was from 4 to 21 on the PSQ (Fig. 1).

On univariate analysis, the PSQ correlated with the number of procedures, number of process failures, anticipated

Table 3 Results from multivariable analysis for selected variables with VAS score

Table 4 Results from multivariable analysis for selected variables with PSQ score

Variable Estimate	95% CI	P
Shift afternoon -1.11	-7.87 to 5.65	.7403
Shift night 4.62	-1.68 to	.1453
	10.92
Anticipated overtime 0.19	0.12 to 0.27	b.0001 ?
Total patients during shifts 0.89	-0.01 to 1.78	.0517
Consultations during shifts -0.80	-2.12 to 0.52	.2311
Patients admitted 0.77	-0.69 to 2.24	.2996
Average patient acuity -2.49	-10.59 to	.5436
	5.60
Procedures -0.73	-2.70 to 1.25	.4698
Process failures 6.14	4.66 to 7.62	b.0001 ?
Average LOS overall 0.06	-0.68 to 0.80	.8737

Variable	Estimate	95% CI	P
Shift afternoon	-0.90	-4.93 to 3.12	.6509
Shift night	3.28	-0.50 to 6.98	.0871
Anticipated overtime	0.08	0.04 to 0.13	.0004 ?
Total patients during shifts	0.11	-0.45 to 0.63	.7332
Consultations during shifts	0.01	-0.81 to 0.78	.9713
Patients admitted	0.83	-0.13 to 1.64	.0919
Average patient acuity	2.34	-2.91 to 6.88	.4234
Procedures	0.80	-0.46 to 1.95	.2225
Process failures	3.70	2.61 to 4.40	b.0001 ?
Average LOS overall	0.04	-0.44 to 0.46	.9658
Average WR count overall	0.05	-0.28 to 0.40	.7190
* P b .05.

overtime, number of patients admitted, and average LOS (Table 2). The VAS correlated with total patients during shifts, patients per hour, number of process failures, anticipated overtime, number of consultations, average patient acuity, number of patients admitted, average LOS, and average WR count. Using the VAS, night shifts seemed to be associated with less stress on univariate analysis.

On multivariate analysis, both the VAS (Table 3) and the PSQ (Table 4) were correlated with anticipated overtime and adverse events (Table 3).

Discussion

Stress may be best defined as the inability to cope or the fear of not coping [27]. Critical to coping are issues of internal and external control, like mood and the environment, for example [3]. This is an era of medicine that is defined by higher acuity patients, increasing patient volumes, shorter hospital stays, an emphasis on faster throughput, more uncertainty about the future, and decreased time for patients, peers, and faculty [15]. It seems that there is a trend toward more hassled residents and attending physicians and more dissatisfaction, cynicism, depersonalization, and anger due to a variety of stressors [4,5,7,10].

In this study, it seems that EM resident shift-specific stress was associated primarily with the perception of excessive time required to finalize or hand over patient care after the shift and with adverse events. Number of patients seen, patients per hour, number of consultations during a shift, patient acuity, and numbers of admitted patients did not affect stress in the multivariate analysis.

Night shifts are thought to impair physician wellness [21]. In this study, at least in the univariate analysis, night shifts correlated inversely with stress. Data from our ED informa- tion system and other EDs show that peak numbers of patients arriving to the ED are in the late morning to afternoon time periods [28]. The night shift may therefore be a quieter time in terms of patient numbers, patient acuity, patient flow, and anticipated overtime. Also, wellness is a long-term issue, and we were measuring shift-specific stress.

Environmental factors that reflect ED overcrowding such as high average occupancy, long LOS, long WR times, and high WR counts did not seem to impact resident stress to any great degree in multivariate analysis. When overcrowding is a chronic problem, it is possible that crowding has less impact on stress because it is the norm. In this study, however, the ED diversion time for each resident shift averaged 18% and varied among residents and shifts.

One coping strategy that residents often adopt is the “end- in-sight” attitude, allowing one to accept stressors as long as they are perceived to be temporary [15]. It is conceivable that ED overcrowding is a system’s factor that affects attending physicians and administrators more than resident physicians who interact directly with fewer patients and are less involved or exposed to ED operational factors. In other

words, overcrowding may actually protect residents to some extent. There is information that stressors affecting faculty physicians are different than those affecting residents [2]. When patients are in the WR, because there are no beds available in the ED in which to be seen, throughput slows and the number of patients actually seen may fall. Slow throughput translates to less time spent after the shift is officially over and to fewer distractions, which have repeatedly been shown to contribute to stress in other settings [4,7,12,14,17]. Overcrowding may ironically decrease feelings of being overwhelmed during a shift even while it may negatively impact resident education. On the other hand, if patients become angry because of long waiting times, this could increase the possibility of conflict with a resident, an adverse event. In any case, fixing the overcrowding problem and attendant gridlock in the ED may have no effect or could actually increase resident stress.

Although adverse events are not necessarily predictable or preventable, strategies to mitigate their impact on the resident are available. Retreats at the beginning of training to teach residents specifically about coping skills, ethics, communication, team work, and grief training seem to be successful in terms of both residents’ perceptions about their residency training and their ability to cope, a significant component of ameliorating resident stress [29]. After an adverse event in a busy ED environment, it is difficult but feasible to do focused, critical incident debriefings with residents, either immediately or at the end of a shift [30]. Having policies in place to ameliorate resident conflicts with other health care providers, mainly the early intervention of faculty physicians or higher level residents, also seems to be indicated.

Our study has several limitations. First, the findings from this small cohort of participants at a single institution providing care for a high-acuity patient population may not generalize to other settings. Any given institution may have more or fewer environmental stressors and stressors of different types.

The degree of overcrowding and gridlock may vary considerably from institution to institution as well. Further- more, the university hospital setting may also differ from other types of training facilities in other ways. There was wide variability in stress among different residents and even within each resident’s experiences. This accounted for a wide 95% CI but should have been controlled for by the repeated- measures methodology. There were differences in the home lives of residents, which may have affected the way residents handled stressful events and particularly the anticipated overtime variable. During the course of this study, however, we felt that overcrowding was a serious and ongoing issue. For the 216 resident shifts, the adult ED was on diversion for an average of 88 minutes (SD, 150 minutes), which is 18.3% of each shift (morning shift, 9.6% or 46 minutes; afternoon, 31.3% or 150 minutes; night, 14.1% or 68 minutes). During the 12 shifts for each resident, the ED diversion ranged from 45 to 178 minutes (or 9% to 37% of each shift time).

Although several instruments for measuring stress exist, we used the PSQ because it seemed to best fit our study design. It is a validated instrument but was slightly modified to fit the ED setting, and as a control measure, we also used a 100-mm VAS, which correlated well with the PSQ in the multivariate analysis. The timing of administration of the surveys at the end of a shift may have led some variables, such as anticipated overtime, to influence the results more than they would have if the instrument had been administered mid-shift. An end-of-shift survey may also underestimate the immediate impact of stressors occurring during a shift. We chose to use the end-of-shift time to standardize survey administration and because it was less disruptive to patient care. End-of-shift perception of stress may also represent the more lasting, and therefore harmful, stress that the resident takes home.

This is a convenience sample done to accommodate the research assistant and to be able to catch residents equally in all different types of shifts. We applied ED diversion as a surrogate marker for overcrowding, which may not always capture the condition of the ED accurately but has been used to characterize ED crowding [28].

Resident stress is impacted by many factors other than physical exhaustion and sleep deprivation from excessive duty hours. Overcrowding may be bad for patients and resident education, but it did not seem to cause stress. It is unlikely that solving the ED overcrowding issue will necessarily translate into less stress for the residents. In fact it, might increase resident stress as throughput pressure increases.

When the workload is heavy, the environment is intense, and the potential for conflict and mistakes is high, adverse events and extra work cause stress. Policies and procedures are likely to alleviate some of these stressors. When looking at resident stress, it is not just about time but also about time pressure and emotional intensity during a shift.

References

1. Martini S, Arfken CL, Churchill A, Balon R. Burnout comparison among residents in different medical specialties. Acad Psychiatry 2004;28:240-2.
2. Sargent MC, Sotile W, Sotile MD, Rubash H, Barrack RL. Stress and coping among orthopedic and surgery residents and faculty. J Bone Joint Surgery 2004;86:1579-86.
3. Rios A, Sanchez Gascon F, Martinez JF, Guerrero M. Influence of residency training on personal stress and impairment in family life: analysis of related factors. Med Princ Pract 2006;15:276-80.
4. Castelo-Blanco C, Figueras F, Eixarch E, Quereda F, Cancelo MJ, Gonzalez S, et al. Stress symptoms and burnout in obstetric and gynaecology residents. BJOG 2007;114:94-8.
5. Iacovides A, Fountoulakis KN, Kaprinis St, Kaprinis G. The relationship between job stress, burnout and clinical depression. J Affect Disord 2003;75:209-21.
6. Zare SM, Galanko J, Behrns KE, Koruda MJ, Boyle LM, Farley DR, et al. Psychological well-being of surgical residents before the 80 hour work week: a multiinstitutional study. J Am Coll Surg 2004;198:633-40.
7. Michels PJ, Probst JC, Godenick MT, Palesch Y. Anxiety and anger among family practice residents: a South Carolina Family Practice Consortium Study. Acad Med 2003;78:69-79.
8. Kiernan M, Civetta J, Bartus C, Walsh S. 24 Hours on-call and acute fatigue no longer worsen resident mood under the 80-hour work week regulations. Curr Surg 2006;63:237-41.
9. Yi MS, Mrus JM, Mueller CV, Luckhaupt SE, Peterman AH, Puchalski CM, et al. self-rated health of primary care house officers and its relationship to psychological and spiritual well-being. BMC Med Educ 2007;7:9.
10. Collier VU, McCue JD, Markus A, Smith L. Stress in medical residency: status quo after a decade of reform? Ann Intern Med 2002; 136:384-90.
11. Mills LD, Mills TJ. Symptoms of post-traumatic stress disorder among emergency medicine residents. J Emerg Med 2005;28:1-4.
12. Vidyarthi AR, Auerbach AD, Wachter RM, Katz PP. The impact of duty hours on resident self reports of errors. J Gen Intern Med 2007;22: 205-9.
13. Baldwin Jr DC, Daugherty SR, Tsai R, Scott Jr MJ. A national survey of residents’ self-reported work hours: thinking beyond specialty. Acad Med 2003;78:1154-63.
14. Salerno SM, Faestel PM, Mulligan T, Rosenblum MJ. Disruptions and satisfaction in internal medicine resident continuity clinic differ between inpatient and outpatient rotations. Teach Learn Med 2007; 19:30-4.
15. Cohen JJ. Heeding the plea to deal with resident stress. Ann Intern Med 2002;136:394-5.
16. Ratanawongsa N, Wright SM, Carrese JA. Well-being in residency: a time for temporary imbalance? Med Educ 2007;41:273-80.
17. Moreno MA. Resident stress revisited: a senior pediatric resident’s point of view. Pediatrics 2003;112:411-4.
18. Daugherty SR, Baldwin Jr DC, Rowley BD. Learning, satisfaction and mistreatment during medical internship. JAMA 1998;279:1194-9.
19. Institute of Medicine. IOM report: the future of emergency care in the United States health system. Acad Emerg Med 2006;13:1081-5.
20. Shanafelt TD, Bradley KA, Wipf JE, Back AL. Burnout and self- reportED patient care in an internal medicine residency program. Ann Intern Med 2002;136:358-67.
21. Houry DE, Shockley LW, Markovchick V. Wellness issues and the emergency medicine resident. Ann Emerg Med 2000;35:394-7.
22. Accreditation Council of Graduate Medical Education [home page on the Internet]. ACGME program requirements for graduate medical education in emergency medicine. Effective date 7-1-07 [cited 2008 Apr 10] Available from: http://www.acgme.org/acWebsite/ RRC_110/110_prIndex.asp.
23. Fliege H, Rose M, Arck P, Walter OB, Kocalevent RD, Weber C, et al. The Perceived Stress Questionnaire (PSQ) reconsidered: validation and reference values from different clinical and healthy adult samples. Psychosom Med 2005;67:78-88.
24. Aronsky D, Jones I, Lanaghan K, Slovis CM. Supporting patient care in the emergency department with a computerized whiteboard system. J Am Med Inform Assoc 2008;15(2):184-94.
25. Tanabe P, Gimbel R, Yarnold PR, Kyriacou DN, Adams JG. Reliability and validity of scores on the Emergency Severity Index version 3. Acad Emerg Med 2004;11:59-65.
26. R Development Core Team. R: a language and environment for statistical computing. Vienna (Austria): R Foundation for Statistical Computing3-900051-07-0; 2007. Available from: http://www.R-pro- ject.org.
27. Baldwin PJ, Dodd M, Wrate RW. Young doctors’ health: I. How do working conditions affect attitudes, health and performance? Soc Sci Med 1997;45:35-40.
28. McCarthy ML, Aronsky D, Jones ID, Miner JR, Band RA, Baren JM, et al. The emergency department occupancy rate: a simple measure of emergency department crowding? Ann Emerg Med 2008;51(1):15-24, 24.e1-2.
29. Reilly JM, Ring JM. Physician, heal thyself: tools for resident well- being. Med Educ 2007;41:522-3.
30. Mitchell JT. The history, status and future of critical incident stress debriefings. J Emerg Med Serv 1988;13:47-52.