Article, Emergency Medicine

Evaluation of the effectiveness of peer pressure to change disposition decisions and patient throughput by emergency physician

a b s t r a c t

Objectives: The aim of this study was to develop a strategy for imposing peer pressure on emergency physicians to discharge patients and to evaluate Patient throughput before and after intervention.

Methods: A before-and-after study was conducted in a medical center with more than 120 000 annual emergency department (ED) visits. All nontraumatic adult patients who presented to the ED between 7:30 and 11:30 AM Wednesday to Sunday were reviewed. We created a “team norm” imposed peer-pressure effect by announcing the patient discharge rate of each emergency physician through monthly e-mail reminders. Emergency department length of stay (LOS) and 8-hour (the end of shift) and final disposition of patients before (June 1, 2011-September 30, 2011) and after (October 1, 2011-January 30, 2012) intervention were compared.

Results: Patients enrolled before and after intervention totaled 3305 and 2945. No differences existed for age, sex, or average number of patient visits per shift. The 8-hour discharge rate increased significantly for all patients (53.5% vs 48.2%, P b .001), particularly for triage level III patients (odds ratio, 1.3; 95% confidence interval, 1.09-1.38) after intervention and without corresponding differences in the final disposition (P =

.165) or admission rate (33.7% vs 31.6%, P = .079). Patients with a final discharge disposition had a shorter LOS (median, 140.4 min vs 158.3 min; P b .001) after intervention.

Conclusions: The intervention strategy used peer pressure to enhance patient flow and throughput. More patients were discharged at the end of shifts, particularly triage level III patients. The ED LOS for patients whose final disposition was discharge decreased significantly.

(C) 2013

  1. Introduction

Emergency department (ED) crowding is a known critical health care problem internationally and is an evolving problem in Taiwan. The causes of crowding are complex, and the input-throughput- output conceptual model [1] has become a widely accepted model for evaluation of this problem. Possible causes for crowding include nonurgent visits, “frequent-flyer” patients, the influenza season, inadequate staffing, inPatient boarding, and hospital bed shortages [2]. Keeping patients in the ED ultimately affects the quality of patient care and has been documented to have ill effects including an increased risk of in-hospital mortality, longer times to treatment for patients with pneumonia or acute pain, and a higher probability of leaving the ED against medical advice (AMA) or without being seen [3].

* Corresponding author. Niaosong Township, Kaohsiung County 833, Taiwan (ROC). Tel.: +886 0911 303 270; fax: +886 07 7317123 8415.

E-mail address: [email protected] (C-W. Lee).

This crowded ED phenomenon may compromise the health care providers’ ability to care for patients, which may lead to decreased Job satisfaction, frustration, anger, depression, and, ultimately, physician burnout [4].

Under these circumstances, with increased vulnerability to possible Medical errors or adverse outcomes, emergency physicians (EPs) may become increasingly reluctant to discharge patients directly from the ED for fear of premature diagnosis, insufficient treatment, or even medicolegal liability. Prolonging ED observations or admitting patients with borderline criteria for inpatient care further aggravates ED crowding. To break this cycle and to save scarce medical resources for the most critically ill patients, it is reasonable to allocate priority care to patients with urgent needs over patients who can be discharged with little or no health risk [5].

Weiner et al [6] showed that timeliness of EPs’ compliance with a clinical pneumonia guideline was improved by weekly e-mail reminders through the “Audit and feedback” approach. This study used a behavior-modifying measure by the creation of team norms enhanced by e-mail reminders. Norms are the rules that the team agrees to follow and designate as a standard for performance by the

0735-6757/$ – see front matter (C) 2013

whole team. Once developed, team norms are used to guide and shape team members’ behavior. To improve patient throughput and ED crowding, a “team norm” used peer pressure to encourage EPs to discharge patients who could be safely discharged according to each EP’s clinical judgment. The discharge rate of each attending physician was announced through e-mails to the entire emergency team on a monthly basis. The present study evaluated the effect of this strategy.

  1. Methods
    1. Study design

To evaluate the effectiveness of peer pressure in changing EP decisions about disposition and patient throughput, we conducted a before-and-after retrospective case review of patients who visited the ED. This study intervention created a team norm that imposed an “unspoken peer-pressure effect” by announcing the patient discharge rate of each EP. The monthly discharge rate of each EP was calculated and announced by e-mail at the beginning of the following month to enhance the team norm and shape the behavior of each EP. E-mails included the detailed numbers of discharged patients and patient visits for each shift handled by each EP. Grading and listing of the discharge rate data for all EPs was also included, and the top 3 EPs were highlighted. Take December 2011 for example; certain EP had 4 day shifts with an average discharge rate of 43.1% (62 discharged patients divided by the 144 visited patients), which was the ninth in the ranking list, whereas the top discharge rate was 60.7%. The information was sent by e-mail on the beginning of January 2012 with detail analysis of each shift available on the attached file. Only statistical figures were reported to the EPs without any additional financial rewards or punishments. At a monthly meeting at the end of September 2011, all EPs agreed with the concept of alleviating ED crowding by discharging less severe patients and focusing on those patients who were critically ill. The preintervention study period was June 1, 2011, to September 30, 2011, and the postintervention period was October 1, 2011, to January 30, 2012. Mondays and Tuesdays and Chinese New Year (January 18- 29, 2012) were excluded from the study because of different EP scheduling. At the end of February, another intervention aiming to improve ED crowding developed by administrative department was proposed. The study period was limited to 8 months because the planned intervention might affect the discharge rate of EPs. The study was approved by our hospital’s institutional review board.

Study setting and population

This study was conducted in a tertiary academic medical center in Southern Taiwan with more than 3000 acute beds and 120 000 annual nontraumatic emergency visits during the study period from June 1, 2011, to January 31, 2012. All nontraumatic adult patients who presented to the ED between 7:30 to 11:30 AM from Wednesday to Sunday were included in the study. The Five Level Taiwan Triage and Acuity Scale formulated by the Department of Health were used for the evaluation of disease acuity. The baseline length of a duty shift for an attending physician was 8 hours (from 0730 to 1530 hours). During the shift, 2 EPs were in charge of all nontraumatic patients in the ED. Patients were alternately assigned to each attending physician. Theoretically, because of the nature of random distribution, the demographic factors and patterns of disease of patients assigned to each attending physician should ultimately be the same. Although emergency medicine or internal medicine residents assisted in the initial evaluation and management of patients, all final Disposition decisions were made independently by the on-duty EP. Patients who visited in the latter half of the shift were excluded because of the possibility that incomplete data or an examination report might influence and defer the disposition. We studied only 15 of the 16 EPs because a new member joined in October 2011, preventing a

comparison. The new EP had only 7 day shifts in the 8-month study period, and his patient data were excluded. We had 4 senior EPs with at least 10 years of postresident work experience, 7 intermediate EPs with 5 to 9 years of work experience, and 4 junior EPs with fewer than 5 years of work experience. The youngest EPs had at least 2 years of work experience.


Data including demographic factors (sex, age, triage level) and patients’ disposition were time stamped and abstracted from the electronic ED administrative database. Outcome measures were patients’ disposition and ED length of stay . To clarify the effect on each physician, the 8-hour (the end of the duty shift) and final dispositions of patients were documented. We collected these data because of the duty physician’s handover to a subsequent physician at that time. That is, the disposition of a patient with LOS for 8 hours was made by another physician. We categorized dispositions into 8 groups: discharge, ward admission, intensive care unit admission, outpatient referral, hospital transfer, died, discharge AMA, and observation only. The discharge rate of each triage level during each shift was calculated by the number of patients discharged within the 8-hour duty period divided by the number of all patient visits during that shift. For example, if 50 patients visited during a shift with 10 patients in each 5 triage levels and if the duty EP discharged 8 patients at the end of the shift with 4 of those in triage level III, then the overall final discharge rate for the shift would be 16 % (8/50) and the discharge rate in triage level III would be 40% (4/10). The degree of ED crowding was measured by the ED occupancy rate in the end of each shift because this has been suggested as the simplest and best overall indicator of crowding [7]. The ED occupancy rate was defined as the total number of patients in the ED at 1530 hours divided by the total number of licensed 150 treatment bays, including all Observation Units, in our hospital. The unscheduled 72-hour revisit rate, defined as the number of patients who returned within 72 hours divided by the total number of discharged patients, was regarded as the index for inappropriate discharge.

Data analysis

The results of descriptive analyses of independent variables are reported as percentages, mean +- SD or median +- interquartile range (IQR). The independent variables were analyzed using the ?2 test, Mann-Whitney U test, and Student t test. The statistical significance of the relationship between the discharge rates before and after intervention in terms of total patients and different triage groups was analyzed with logistic regression to obtain the odds ratio and 95% confidence interval. A P value less than .001 was regarded as statistically significant. SPSS version 12.0 (SPSS, Inc, Chicago, IL) was used for all statistical analyses.

  1. Results

During the 8-month study period, 3305 and 2945 patients were enrolled before and after intervention, respectively. During the 88 shifts with 176 duty EPs before intervention, senior, intermediate, and junior EPs were in charge of 51 (28.98%), 75 (42.61%), and 50 (28.41%), respectively. Senior, intermediate, and junior EPs were in charge of 44 (28.57%), 69 (44.81%), and 41 (26.62%) of the 77 shifts after intervention. There was no statistically significant difference in the distribution of duty attending physicians (P = .909). The average number of patient visits and the relevant basic demographic factors and disposition are shown in Table 1. No significant difference was found among the demographic factors, except for the triage level. Compared with the preintervention stage, the percentage of patients at triage level III was higher (75.96% vs 69.13%, P b .001) in the postintervention group.

Table 1

Demographic factors, ED occupancy rate, and 72-hour revisit rate before and after intervention





(n = 3305)

(n = 2945)





37.6 +- 6.34

38.3 +- 8.79


Agea (y)

59.5 +- 18.39

58.6 +- 17.83


Male, n (%)

1597 (48.3)

1422 (48.3)


Triage, n (%)



87 (2.6)

55 (1.9)


624 (18.9)

452 (15.3)


2285 (69.1)

2237 (76.0)


246 (7.4)

168 (5.7)


63 (1.9)

33 (1.1)

72-h revisit, n (%)

86 (2.6)

88 (3.0)


Occupancy rateb (%)

98.1 +- 14.0

94.8 +- 16.3


a Expressed as mean +- SD.

b Defined as the total number of patients in the ED divided by the total number of licensed 150 treatment bays in our hospital.

Fig. 1. Comparison of odds ratio of patient discharge rate for each triage level.

We found a statistically significant difference in the overall 8-hour disposition of patients (Table 2) after intervention. In the 8-hour disposition, at the end of the shift, more patients were discharged (53.5% vs 48.2%, P b .001) compared with the preintervention period. After analyzing the number and percentage of patients discharged from different triage levels, the greatest differences in discharge rates occurred in overall patients and in patients at triage level III (Fig. 1). The overall distribution of the final disposition of patients was the same (Table 3). The admission rates including ward and ICU admissions showed no statistically significant difference after inter- vention (33. 8% vs 31.6%, P = .079). Patients with a final disposition of discharge at the end of the shift had improvED throughput with shorter ED patient LOS (median, 140.4 vs 158.3 minutes; P b .001) (Fig. 2). The median LOS of patients with a final disposition of ward or ICU admission was 28.65 (IQR, 5.8-58.06) hours without a statistically significant difference after intervention (median, 29.86 vs 27.74 hours; P = .185). As shown in Table 1, the ED occupancy rate showed no statistically significant difference after intervention (94.83 +- 16.3 vs 98.13 +- 14.00, P = .164).

Twelve of the 18 studied EPs increased their overall 8-hour

discharge rate after intervention, with 4 having a statistically significant difference (P b .05). Four of the 6 authors of this work were part of the study, but only 1 author knew what was going to be introduced at the September meeting. The author has increase discharge rates without statistical significance after intervention. Unscheduled ED returns within 72 hours of ED discharge did not differ significantly after intervention (3.0% vs 2.6%, P = .354).

  1. Discussion

Under the current policy of National Health Insurance in Taiwan, a large proportion of ED beds are occupied by patients who are

ineligible for inpatient care. Emergency department crowding is a complex and multifaceted problem, and many types of managerial strategies have been implemented in EDs throughout the world and reported in the literature. Commonly studied solutions for crowding have included increased resources, demand management, and operations research. These are summarized in a review article by Hoot and Aronsky [8]. Among the root causes of ED crowding, the issue of output with inability to transfer emergency patients to inpatient beds and the resultant boarding of admitted patients in the ED is critical [9,10]. The creation of Multidisciplinary teams to facilitate hospital bed access [11,12] and increases in the number of acute managED observation units [13,14] or even inpatient beds [15] have been documented to improve crowding; however, EPs who experience crowding and its consequences have little power to implement these solutions because most of them require the collaboration of policy makers or institution managers. In an era of increasing crowding and blocked access for admitted patients, which drain ED resources, it is important that EDs explore initiatives to improve those aspects of efficiency that are within their control.

To alleviate crowding in the ED, this study used a novel strategy by creating a team norm to stimulate a peer effect to encourage EPs to discharge patients that do not require admission. The effect was further enhanced by the novel and innovative strategy published by Weiner et al [6]. By using weekly e-mail reminders listing perfor- mance on antibiotic administration, Hoot et al improved the timeliness of time from arrival to antibiotic administration (162-146 minutes, P = 0.018). Our strategy was an extension and application of this known audit and feedback methodology. In our work, the ultimate goal was to enhance patient flow and throughput as well as to preserve limited resources for the sickest patients. Focusing on timely discharge of patients is an important strategy for mitigating ED crowding. The intervention strategy in this study resulted in an overall decrease in the time patients spent in an ED bed, a practice that

Table 2

Eight-hour disposition of patients

Disposition Total

Intervention P

Table 3

Final disposition of patients

(n = 6250)


(n = 3305)


(n = 2945)

Disposition Total

(n = 6250)



1593 (48.2)

1577 (53.5)


Outpatient referral


15 (0.5)

13 (0.4)



1978 (59.8)

1848 (62.8)


Discharge AMA


71 (2.1)

47 (1.6)

Outpatient referral


18 (0.5)

13 (0.4)

Ward admission


314 (9.5)

260 (8.8)

Discharge AMA


121 (3.7)

86 (2.9)

ICU admission


59 (1.8)

33 (1.1)

Ward admission


1030 (31.2)

849 (28.8)



12 (0.4)

9 (0.3)

ICU admission


85 (2.6)

83 (2.8)



28 (0.8)

24 (0.8)



21 (0.6)

24 (0.8)

Pending disposition


1213 (36.7)

982 (33.3)



52 (1.6)

42 (1.4)

Intervention P

Before (n = 3305) After (n = 2945)

Fig. 2. Emergency department LOS of patients with a final disposition of discharge.

benefitted many other patients. The LOS for discharged patients decreased by a median of 17.9 minutes, which represented an 11.3% decrease. In a busy ED with more than 120 000 visits and a discharge rate of 50.72%, a 17.9-minute decrease in LOS for discharges amounts to a decrease of around 18 000 hours in overall time spent in the ED. Furthermore, this strategy added no additional cost to the adminis- tration of the ED, and the magnitude and importance were clear.

The widely used audit and feedback strategies could be effective in improving professional practice based on the modification of behavior if given feedback that their clinical practice was inconsis- tent with that of their peers or accepted guidelines [16]. Recent studies showed positive effect on the improvement of thrombolytic therapy administration in acute stroke [17] and Door-to-balloon times in patients with acute ST-elevation myocardial infarction [18] after intervention with the feedback. Weiner et al and our work found that the improvement of clinical practice were associated with e-mail reminders, as well. Contrary, one previous study using 1-time e-mail communication found that there was no difference in EP ordering rates of chest computed tomography studies to exclude pulmonary embolus in young patients [19]. The different outcome with similar intervention might relate to the frequency of feedback and suggest that the 1-time communication is possibly not adequate to raise awareness and to modify clinical behavior. The most appropriate frequent of audit and feedback remains to be further studied and clarified.

Economists are interested in how workers modify their behavior in the presence of information about the productivity of their coworkers. If workers respond to peer effect, then a firm can hopefully raise worker productivity by revealing publicly available information about worker productivity. Research results by Falk and Ichino [20] and Mas and Moretti [21] showed that peer pressure had a positive and significant effect on productivity in real time. Real-time interaction between workers allows peer pressure to be determined throughout the work period. Our setting was more static because an EP observed the final patient discharge rate of another EP only at the beginning of the following month.

Various fields in the medical profession have used financial incentives to improve Quality performance of health care [22]. Positive effect of bonus payments for emergency service performance has been found in the literature [23]; however, such incentives may have possible unintended effects on the quality of care. For example, Roski and colleagues [24] examined the effect of bonus payments for identifying patients with Tobacco use disorders and made tobacco cessation advice available to large multispecialty group practices. The incentive was associated with increased documentation of tobacco use but did not provide advice to quit smoking. This highlights the

problem of gaming behavior, whereby the incentive improves documentation rather than changing the quality of health care delivered to patients. Emergency physicians may face additional burdens and stress from pay for performance and find their sense of autonomy and professionalism eroded by a quality agenda established and dictated by payers [25], resulting in a decline in department morale and career satisfaction [26]. Our work emphasized that the intervention strategy was not based on either financial penalties or financial bonuses. In other words, a crude carrot-and-stick strategy does not move EPs toward specific behavior patterns.

In contrast to the improved LOS of discharged patients, we found that this strategy had a statistically insignificant effect on the ED occupancy rate. The ED at our hospital was gridlocked with admitted patients with a median LOS of 28.65 (IQR, 5.8-58.06) hours. Compared with the median 2.46-hour LOS of discharged patients, it is possible that the boarding time for admitted patients was so long that the beneficial effect from discharging patients was minimized and therefore did not improve the ED occupancy rate.

In addition to decreasing the LOS of discharged patients, the administrative strategy of this particular clinical strategy was to increase the discharge rate of ED patients, as well. In this study, improvement in discharge occurred overall and for triage level III patients. In general, triage levels I and II patients were more likely to be admitted or observed in the ED because of their relatively critical condition upon arrival. On the other hand, the severity of the condition of patients in triage levels IV and V was mostly trivial and did not require admission care. It was reasonable to expect no significant difference in the disposition of these extreme patients after intervention. Level III patients are generally more complex, and this may result in uncertainty and variation in their management and disposition by EPs. It is not surprising that these patients are most likely to show differences after intervention.

  1. Limitations

There were some limitations in this novel study. First, it has limited generalizability of its findings because it was conducted in only 1 medical center and included only 15 EPs for a limited time. Second, although the final disposition and number of unschedulED revisits showed no statistically significant difference after intervention, premature closures or incomplete diagnostic evaluations were not well studied because there was no follow-up to determine the final patient outcome, especially for patients who were discharged from the ED. Third, we did not evaluate other potential occult adverse effects of this intervention such as physician-patient interaction, suboptimal physician-to-patient communications, or patient dissat- isfaction. Finally, it is doubtful that the effect of this strategy could last permanently given these time-limited data. Future multiple-center and long-term studies should be conducted to determine how the effectiveness of peer pressure is related to the interactions among ED attending physicians and between physicians and patients and to clarify if this intervention can be sustained.

  1. Conclusions

The intervention strategy used peer pressure enhanced by e-mail reminders to improve patient flow and throughput. More patients were discharged at the end of shifts, particularly triage level III patients. The ED LOS decreased significantly by a median of 17.9 minutes for patients whose final disposition was discharge.


We gratefully acknowledge Dr I-Chuan Chen for her invaluable advice during the study.


  1. Asplin BR, Magid DJ, Rhodes KV, et al. A conceptual model of emergency department crowding. Ann Emerg Med 2003;42:173-80.
  2. Moskop JC, Sklar DP, Geiderman JM, et al. Emergency department crowding, part 1–concept, causes, and moral consequences. Ann Emerg Med 2009;53: 605-11.
  3. Bernstein SL, Aronsky D, Duseja R, et al. The effect of emergency department crowding on clinically oriented outcomes. Acad Emerg Med 2009;16:1-10.
  4. Rondeau KV, Francescutti LH. Emergency department overcrowding: the impact of resource scarcity on physician job satisfaction. J Healthc Manag 2005;50: 327-40.
  5. Kelen GD, Kraus CK, McCarthy ML, et al. Inpatient disposition classification for the creation of hospital surge capacity: a multiphase study. Lancet 2006;368: 1984-90.
  6. Weiner SG, Brown SF, Goetz JD, et al. Weekly e-mail reminders influence emergency physician behavior: a case study using the Joint Commission and Centers for Medicare and Medicaid Services Pneumonia Guidelines. Acad Emerg Med 2009;16:626-31.
  7. Hoot NR, Zhou C, Jones I, et al. Measuring and forecasting emergency department crowding in real time. Ann Emerg Med 2007;49:747-55.
  8. Hoot NR, Aronsky D. Systematic review of emergency department crowding: causes, effects, and solutions. Ann Emerg Med 2008;52:126-36.
  9. Moskop JC, Sklar DP, Geiderman JM, et al. Emergency department crowding, part 2: barriers to reform and strategies to overcome them. Ann Emerg Med 2009;53: 612-7.
  10. Boyle A, Beniuk K, Higginson I, et al. Emergency department crowding: time for interventions and policy evaluations. Emerg Med Int 2012 [Article ID 838610, 8 pages].
  11. Clark JJ. Unlocking hospital gridlock. Healthc Financ Manage. 2005;59:94-96, 98, 100-102.
  12. Greene J. Emergency Department flow and the boarded patient: how to get admitted patients upstairs. Ann Emerg Med 2007;49:68-70.
  13. Kelen GD, Scheulen JJ, Hill PM. Effect of an emergency department (ED) managed acute care unit on ED overcrowding and emergency medical services diversion. Acad Emerg Med 2001;8:1095-100.
  14. Moloney ED, Bennett K, O’Riordan D, et al. Emergency department census of patients awaiting admission following reorganisation of an admissions process. Emerg Med J 2006;23:363-7.
  15. McConnell KJ, Richards CF, Daya M, et al. Effect of increased ICU capacity on emergency department length of stay and Ambulance diversion. Ann Emerg Med 2005;45:471-8.
  16. Jamtvedt G, Young JM, Kristoffersen DT, et al. Audit and feedback: effects on professional practice and health care outcomes. Cochrane Database Syst Rev 2006(2):CD000259.
  17. Ghrooda E, Alcock S, Jackson AC. Improvement in thrombolytic therapy adminis- tration in acute stroke with feedback. Can J Neurol Sci 2012 Nov;39(6):789-92.
  18. Lai CL, Fan CM, Liao PC, et al. Impact of an audit program and other factors on door- to-balloon times in acute ST-elevation myocardial infarction patients destined for primary coronary intervention. Acad Emerg Med 2009 Apr;16(4):333-42.
  19. Hardin LV, Nguyen SA, Ravenel JG. Is e-mail communication effective in changing Ordering patterns in the emergency department? A case study of computed tomography for pulmonary embolus. Acad Radiol 2008;15:433-7.
  20. Falk A, Ichino A. Clean evidence on peer effects. J Labor Econ 2006;24:39-57.
  21. Mas A, Moretti E. Peers at work. Am Econ Rev 2009;99:112-45.
  22. Scott A, Sivey P, Ait QD, et al. The effect of financial incentives on the quality of health care provided by primary care physicians. Cochrane Database Syst Rev 2011;9:CD008451.
  23. Glaser WF. The effects of bonus payments on emergency service performance in Victoria. Med J Aust 2000;172:92-3.
  24. Roski J, Jeddeloh R, An L, et al. The impact of financial incentives and a patient registry on Preventive care quality: increasing Provider adherence to evidence- based smoking cessation practice guidelines. Prev Med 2003;36:291-9.
  25. Epstein AM, Lee TH, Hamel MB. Paying physicians for high quality care. N Engl J Med 2004;350:406-10.
  26. Sikka R. Pay for performance in emergency medicine. Ann Emerg Med 2007;49: 756-61.

Leave a Reply

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