a b s t r a c t

Introduction: Although consultations are essential for delivering safe, high-quality care to patients in emergency departments, they contribute to emergency department patient flow problems and overcrowding which is asso- ciated with several adverse outcomes, such as increases in patient mortality and poor quality care. This study aimed to investigate how time flow metrics including emergency department length of stay is influenced by changes to the internal medicine consultation policy.

Method: This study is a pre- and post-controlled interventional study. We attempted to improve the internal medicine consultation process to be more concise. After the intervention, only Attending emergency physicians consult internal medicine chief residents, clinical fellows, or junior staff of each internal medicine subspecialty who were on duty when patients required special care or an admission to internal medicine.

Results: Emergency department length of stay of patients admitted to the department of internal medicine prior to and after the intervention decreased from 996.94 min to 706.62 min. The times from consultation order to ad- mission order and admission order to emergency department departure prior to and after the intervention were decreased from 359.59 min to 180.38 min and from 481.89 min to 362.37 min, respectively. The inpatient mor- tality rates and Inpatient bED occupancy rates prior to and after the intervention were similar.

Conclusion: The improvements in the internal medicine consultation process affected the flow time metrics. Therefore, more comprehensive and cooperative strategies need to be developed to reduce the time cycle metrics and overcrowding of all patients in the emergency department.

(C) 2017

1. Introduction

Timely consultations of specialists are essential for safe, high-quality care for patients in the ED. Therefore, consultations occur frequently in the emergency department (ED). They involve an emergency physician who requests participation in a patient’s care from another special- ist. However, consultation delays can affects patient flow in the ED and aggravate the severity of overcrowding, which is a growing concern [1]. In addition, the delay in specialist consultation and lack of specialist cov- erage can result in delayED patient care and increased transfers to hos- pitals with a higher level of care [2,3].

ED overcrowding is associated with several adverse outcomes, such as an increase in patient mortality and poor quality care [4-8]. In

* Corresponding author at: Department of Emergency Medicine, Gyeongsang National University School of Medicine and Gyeongsang National University Hospital, Gangnam- ro 79, Jinju 52727, Gyeongsangnam-do, Republic of Korea.

E-mail address: [email protected] (S.H. Lee).

addition, ED overcrowding usually leads to a long emergency depart- ment length of stay (EDLOS), and therefore could be connected to pa- tient dissatisfaction and compromised Emergent care. Although a large number of solutions have been proposed to reduce ED overcrowding and EDLOS, they will require multifaceted approaches, including con- sultation strategies and inpatient bed occupancy [9,10].

There have been several studies on the contribution of consulta- tions to ED patient flow and overcrowding [11-14]. While specialty consultation has been associated with increased EDLOS, there is little data regarding the effect of improvements in the consultation pro- cess on ED LOS. Because a larger volume of patients are admitted to the department of internal medicine (IM), improvements in IM con- sultation process will exert more of an effect on patient flow in the ED than modifications of admission strategies to other departments. For this reason, we paid more attention to improving the consulta- tion policy with IM, and examined its impact on EDLOS. Our hypoth- eses were that specialty consultation delays are an important component of EDLOS and that improvement of the specialty

https://doi.org/10.1016/j.ajem.2017.09.041

0735-6757/(C) 2017

consultation process would have a substantial effect on the decrease in EDLOS.

1. Method
   1. Design

This study is a pre- and post-controlled interventional study with pre- and post-intervention cohorts. This study aimed to investigate how EDLOS is influenced by changes in IM consultation policy. There- fore, we implemented a new consultation process and compared the outcomes before and after the intervention.

Time periods

Pre-intervention consultation and admission cycle times were mea- sured for all ED patients from July 2014 to June 2015. The daily inpatient bED occupancy ratio, admission rate, and inpatient mortality were also measured. After the new consultation process was begun, consultation and admission cycle times, EDLOS, admission rates, the inpatient bed occupancy ratio, and inpatient mortality were monitored and measured from July 2015 to June 2016.

Setting

This study was conducted in the ED of an 885-bed, suburban, aca- demic teaching hospital with 45 beds and an annual ED census of ap- proximately 35,000 patients. The hospital is the sole tertiary medical care facility in the area and treats a large number of patients. The ED is staffed by board-certified emergency attending physicians and resi- dents who are supervised by them. Approximately 11,000 ED patients are admitted annually through the ED, with more than half admitted to the department of internal medicine.

Measures

The performance measures were the consultation and admission cycle times of ED patients. In electronic medical record (EMR) of our hospital, it can register the subspecialty and attending physician who were on duty when requesting consultation with internal medicine, and the registration time will be automatically entered as well. When the subspecialty of internal medicine changes, it is possible to register the additional subspecialties and attending physicians. Therefore, it is possible to comprehensively understand the contents of every consulta- tion with each subspecialty. To investigate how the new IM consultation and admission policy affected patients admitted to the department of IM differently from patients admitted to other departments, we esti- mated EDLOS for the group admitted to IM and other departments sep- arately in the pre- and post-intervention period, as well as the overall EDLOS for all ED patients. In addition, the EDLOS for patients discharged by the ED were also measured to investigate the relationship between the rapid admission process to the IM and the burden of the ED. Other time metrics, including arrival to consultation order, consultation order to admission order, arrival to admission order, and admission order to ED departure, were also measured during the study periods. EDLOS data were extracted from our hospital’s EMR databases. The data include the time stamps of all patients who visited our ED and bed occupancy data for inpatients. In addition, the inpatient bed occu- pancy ratio was also measured to investigate the relationship between inpatient bed occupancy and ED overcrowding during the study period. Inpatient mortality was also measured to investigate the relationship between the rapid admission process and insufficient evaluation and acute care in the ED.

Intervention

Prior to the intervention, attending EPs and senior EM residents consulted IM junior or senior residents of each IM subspecialty who were on duty when patients needed special care or admission to IM. When they were required for patient consultations, they came to the ED and discussed the patient’s problem with a number of the admitting physicians including staff from the IM subspecialty after examining the patient, and determined the final admission decision. Thus, all levels of IM physicians were involved in the admission process when an IM con- sultation for admission was requested by EPs in our hospital. This IM consultation model has an admitting process in which EPs make a re- quest for IM consultations for admission and transfer admission deci- sions to IM admitting physicians. During the consultation, EPs communicate with admitting physicians while reviewing patient data in real time through EMR. The IM and ED attending physicians involved in the admission process collaborated to design the new consultation and admission process to decrease EDLOS. Therefore, we attempted to improve the IM consultation and admission process to be more concise. After the intervention, only attending EPs consult IM chief residents, clinical fellows, or junior faculty of each IM subspecialty who were on duty when patients required special care or an admission to IM. They re- view the laboratory and radiologic results of the patient through EMR. If further evaluation is needed for admission, the admitting physicians on duty travel to the ED and make the final admission decision after exam- ining the patient. After making an admission decision, the admitting physician promptly assigns the patient to an on call IM resident to re- view the patient and facilitate the admission. The difference in the IM consultation and admission process between the pre- and post-inter- vention periods depends on how involved residents are in the final ad- mission decision (Fig. 1). During the post-intervention period, only high level IM physicians from the admitting physician service take responsi- bility for admitting patients. However, this new collaborative process may deprive IM residents of training opportunities to learn how to make appropriate decisions regarding patients who need to be admit- ted. Therefore, we decided to implement this process only during the daytime (7 am-6 pm) and maintain the pre-interventional consultation and admission process during the evening and nighttime (6 pm-07 am). We met to discuss improvements to the consultation and admis- sion process bimonthly during the study period.

Data analysis

Continuous data are reported as the mean +- standard deviation (SD) or median with IQRs and were compared using Student’s t- test. Binomial data are represented as the percentage frequency of occurrence and were compared using univariate analysis with the Pearson ?²-test. Admission cycle times including arrival to consulta- tion order, consultation order to admission order, arrival to admis- sion order, admission order to ED departure, EDLOS, and inpatient bed occupancy, were compared between the pre- and post-interven- tion groups using Student’s t-test. We used the two sample Mann- Whitney U test to compare each of the post-intervention periods with the pre-intervention periods for variables distributed non-normally among admission cycle times. The pre- and post-intervention IM ad- mission rates and mortality were also compared using the Pearson ?²- test for differences in proportions. Statistical significance was set at P b 0.05. All of the analyses were performed using SPSS statistical soft- ware (version 21.0, IBM, Chicago, IL, USA).

Ethics approval

The study was approved by the Institutional Review Board of Gyeongsang National University Hospital.

Fig. 1. A process map of internal medicine consultation prior to and after the intervention.

1. Results

Prior to the intervention, 34,740 patients had visited the ED, and 11,304 patients (32.5%) were admitted through the ED from July 2014 to June 2015. After the intervention, 34,910 patients had visited the ED, and 12,242 patients (35.1%) were admitted through the ED from July 2015 to June 2016. During the pre-intervention period, 6967 pa- tients had consultations with IM, and 5322 patients were admitted to the department of IM. During the post-intervention period, 7301 pa- tients had consultations with IM, and 6152 patients were admitted to the department of IM. The overall EDLOS was 506.28 min during the pre-intervention period, and 443.68 min during the post-intervention period. The inpatient mortality rates of patients admitted to IM through the ED prior to and after the intervention were similar: 4.8% and 4.8%, respectively. The inpatient bED occupancy rate prior to and after the in- tervention was also similar: 84.8% and 84.0%, respectively (Table 1).

The post-intervention EDLOS of patients admitted to IM decreased from 996.94 min to 706.62 min compared with the pre-intervention EDLOS. The time from arrival to consultation order prior to and after the intervention was similar: 155.46 min and 163.87 min. The time from consultation to admission order was decreased from 359.59 min to 180.38 min prior to and after the intervention. In addition, admission order to ED departure was decreased from 481.89 min to 362.37 min prior to and after the intervention (Table 2). Overall the difference in the EDLOS of patients admitted to specialties other than IM and the overall EDLOS of patients discharged by the ED was not significant

Table 1

Characteristics of patients during the study period.

Variables	Pre-intervention	Post-intervention	Total, N
	period, N (%)	period, N (%)
Total patients admitted to ED	34,740	34,910	69,650
Age (years), mean +- S.D.	45.93 +- 26.23	47.56 +- 26.03	46.8
Gender (male)	20,134 (58.0%)	19,963 (57.2%)	29,553
IM consultation	6967 (20.1%)	7301 (20.9%)	14,268
Admission to IM	5322 (15.3%)	6152 (17.6%)	11,474
Consultation with specialty	18,178 (52.3%)	17,078 (48.9%)	35,256
other than IM Admission to specialty other	5699 (16.4%)	5569 (16.0%)	11,268
than IM
Discharge by ED	9595 (27.6%)	10,531 (30.2%)	20,126
Admission to ED	283 (0.8%)	521 (1.5%)	804
total admission patients	11,304 (32.5%)	12,242 (35.1%)	23,546
EDLOS (minute), mean +- S.D.	506.29 +- 691.50	443.69 +- 625.92	474.91
Bed occupancy rate (mean, %)	84.8%	84.0%	84.4%
Mortality after IM admission	256/5322, 4.8%	295/6152, 4.8%	551/11474,
(n, %)			4.8%

ED, emergency department; IM, internal medicine; EDLOS, emergency department length of stay; S.D., standard deviation.

prior to and after the intervention. Fig. 2 shows how the flow time met- rics of the patients admitted to IM vary monthly prior to and after the intervention. Table 3 shows the statistical differences in in-hospital ad- mission length for each subspecialty of internal medicine prior to and after the intervention. Table 4 shows the statistical differences in medi- cine floor, step down unit, and intensive care unit admission length. The data about the diagnosis at admission were so extensive that they were summarized according to subspecialties and attached to the supple- ment file.

1. Discussion

Prolonged EDLOS is associated with poor clinical outcomes and lon- ger hospital stays [15-17]. Patients with long hospital stay also had an influence on longer EDLOS and ED overcrowding, which are associated with inadequate emergent care and increased mortality [4]. In this study, improvements in IM consultation and admission process reduced the EDLOS by approximately 290 min. They reduced the consultation to admission order, arrival to admission order, and admission order to ED departure times by 49.8%, 33.2%, and 24.8%, respectively, which led to an EDLOS decrease for patients admitted to IM of 29.1%. Overall, the EDLOS of all patients who visited the ED was reduced by 12.4%. These improve- ments were durable and Clinically meaningful. This improvement will have an effect on overall patient flow, patient satisfaction, EM physician workflow, and performance for evaluations.

We expected that only the time of consultation order to admission

order would decrease among the consultation and admission cycle time metrics. However, in this study, admission order to ED departure was also significantly decreased from 481.89 min to 362.37 min prior to and after the intervention. We think that this decrease occurred be- cause the reduction in waiting times of patients admitted to IM resulted from the flexible use of admitting units. Therefore, the use of several ad- mitting units could have a large effect on the overall ED patient flow.

Many EDs have commonly implemented two models for admission

processes according to the department that makes the final admission decision: a consultation model and EM decision model. The consultation model can take time, unlike the EM decision model, in which EPs make the final admission decision. However, the final admission decision may be made more efficiently and accurately when the admitting attending physicians have responsibility for it compared to when lower level phy- sicians, including residents, are involved as in our consultation model [18]. This consultation model prevents patients from being transferred to other subspecialties after they are admitted because the admission decision is made by the admitting attending physician. It can also reduce the burden on EPs so that they can pay more attention to other ED pa- tients. Therefore, our new consultation process is expected to have shorter discharge processing times from the ED than in the pre-

Table 2

Variables and flow time metrics during the daytime prior to and after the intervention.

Variables	Pre-intervention period, minute, mean +- S.D.	Post-intervention period, minute, mean +- S.D.	P value
IM admitted patient number, N	3058	3628
Age (years), mean +- S.D.	65.58 +- 15.16	66.66 +- 15.11	0.004
Gender (male, %)	1777 (58.1%)	2148 (59.2%)	0.385
IM overall EDLOS	996.94 +- 886.26	706.62 +- 1105.76	b0.001
Arrival to consultation order	155.46 +- 178.83	163.87 +- 178.46	0.063
Consultation to admission order	359.59 +- 348.29	180.38 +- 225.93	b0.001
Arrival to admission order	515.05 +- 524.84	344.25 +- 406.73	b0.001
Admission order to ED departure	481.89 +- 714.65	362.37 +- 1032.24	b0.001
Overall EDLOS of patients admitted to specialty other than IM	423.61 +- 537.18	416.05 +- 706.97	0.844
Overall EDLOS of patients discharged by ED	316.66 +- 700.92	305.76 +- 593.17	0.464
Overall EDLOS discharged by specialty other than IM	288.67 +- 395.09	280.35 +- 388.64	0.257

ED, emergency department; IM, internal medicine; EDLOS, emergency department length of stay; S.D., standard deviation.

intervention period. However, the EDLOS of patients discharged by the ED was not significantly decreased after vs. prior to the intervention in this study. Although this new consultation process had almost no effect on the flow of patients discharged by ED attending physicians, this new consultation and admission process will have an effect on patient satis- faction, EM physician workflow, and evaluated performances.

The effects of the consultation process on all ED patients were rela- tively lower than the effects on the admitted patients [19]. In this study, the effect of the new IM consultation process on the overall EDLOS was a slight decrease, ranging from 463.45 min to 403.35 min. This result can be explained by the small proportion of patients admit- ted to IM (approximately 15% to 17%) of the total ED patients. The lim- ited effects on patients discharged by the ED and overall patient flow were reasonable, considering the small proportion of patients admitted to IM compared with the total number of ED patients.

A previous study showed that improving the consultation process is not the only solution for an efficient admission process [15]. Although EM admission decision models can lead to a shorter EDLOS for admitted patients compared with that for consultation models, the burden on EPs can be increased because of admission-related activities, and patients discharged by the ED can also wait longer to see the EPs. A modified model, such as our new consultation process can improve the overall patient flow including the average EDLOS of admitting patients. There- fore, this result implies that an improvement of the consultation process can reduce the overall flow of patients admitted to IM as well as the bur- den of EPs associated with admission-related activities. It can also lead to the increased satisfaction in ED patients and an efficient workflow of EM physicians.

Fig. 2. The monthly variation and comparison of flow time metrics of the patients admitted to internal medicine between the pre-intervention and post-intervention period.

Although all EDs attempt to make consultation and admission pro- cesses more efficient, academic teaching hospitals face the problem of providing both clinical care for patients and an education for residents. Policies such as our new consultation process may deprive IM residents of the opportunity to learn appropriate decision making skills for pa- tients who need to be admitted. However, this change can be accom- plished without sacrificing the educational opportunity for residents. Appropriate changes in the basic consultation process can enable aca- demic teaching hospitals to meet the needs for clinical care and resident training. Therefore, we maintained the original consultation process to meet their clinical and educational needs during the night time.

1. Limitations

There are several limitations to this study. First, the results may not be generalizable as this study is a retrospective, observational, and a sin- gle-center study. Second, the consultation model used in the study may not be generalizable to other centers, which may use different consulta- tion approaches. In other words, while our center requires a consulta- tion in order to admit a patient, in other centers, an EP can admit a patient without the input of another specialist. Third, the inclusion of only daytime consultations may have resulted in an over-representa- tion of daytime consultations, and night time consultations involve dif- ferent issues. Fourth, since this study only measured the time cycle metrics of ED consultation processes during two different periods, we did not attempt to measure patient or Physician satisfaction or the effect on educational training for IM residents. Because the satisfaction of the patient or physician also depends on various factors, patients could not comment upon the effect of a single part of the complex process. Though not specifically measured in the study, it would seem reason- able that the decreased EDLOS would be accompanied by higher patient or physician satisfaction. Patient or physician satisfaction is worthy of inclusion as an outcome in a future study. Because policies such as our new consultation process may deprive IM residents of educational

Table 3

In-hospital admission length for each subspecialty of internal medicine prior to and after the intervention.

Subspecialty	Pre-intervention	Post-intervention	P value
	N Day (mean +- S.D.) N	Day (mean +- S.D.)
Cardiology	652 6.22 +- 7.82 735	6.45 +- 7.50	0.469
Endocrinology	26 10.39 +- 11.93 48	9.02 +- 11.27	0.520
Gastroenterology	925 9.21 +- 8.86 1059	8.82 +- 8.51	0.194
Hematology and	327 15.09 +- 15.61 459	14.28 +- 15.07	0.337
oncology Infection	281 11.16 +- 13.47 224	13.02 +- 15.23	0.061
Nephrology	293 13.59 +- 56.91 401	10.48 +- 12.50	0.226
Pulmonology	547 12.73 +- 13.67 691	12.23 +- 18.14	0.481
Rheumatology	7 5.69 +- 3.28 11	8.21 +- 12.61	0.489
Total	3058 10.43 +- 20.79 3628	10.13 +- 12.82	0.336

S.D., standard deviation.

Table 4 Medicine floor, step down unit and intensive care unit admission length prior to and after the intervention.

1. Rao MB, Lerro C, Gross CP. The shortage of on-call surgical specialist coverage: a na- tional survey of emergency department directors. Acad Emerg Med Off J Soc Acad Emerg Med 2010;17:1374-82.
2. Mohanty SA, Washington DL, Lambe S, Fink A, Asch SM. Predictors of on-call special-

Admission unit Pre-intervention Day (mean +- S.D.)

Post-intervention Day (mean +- S.D.)

P value

ist Response times in California emergency departments. Acad Emerg Med Off J Soc Acad Emerg Med 2006;13:505-12.

1. Hoot NR, Aronsky D. Systematic review of emergency department crowding: causes,

Medicine floor	10.16 +- 11.91	9.97 +- 11.05	0.486	effects, and solutions. Ann Emerg Med 2008;52:126-36.
Step down unit	10.23 +- 10.92	10.57 +- 16.88	0.551	[5] Guttmann A, Schull MJ, Vermeulen MJ, Stukel TA. Association between waiting times
Intensive care unit	11.87 +- 45.96	10.20 +- 13.67	0.314	and short term mortality and hospital admission after departure from emergency

S.D., standard deviation.

opportunities, the effect on educational training also needs to be inves- tigated in future studies.

1. Conclusion

This study showed that an improvement of the IM consultation and admission process affected the flow time metrics of IM patients admit- ted through the ED. Therefore, more comprehensive and cooperative strategies need to be developed to reduce the time cycle metrics for all ED patients and for overall ED overcrowding.

Acknowledgments

Competing interests

No authors have any competing interests to declare.

Funding and all other required statements

No authors have any funding to declare.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi. org/10.1016/j.ajem.2017.09.041.

References

[1] Asplin BR, Magid DJ, Rhodes KV, Solberg LI, Lurie N, Camargo Jr CA. A conceptual model of emergency department crowding. Ann Emerg Med 2003;42:173-80.

department: population based cohort study from Ontario, Canada. BMJ 2011;342: d2983.

1. Cowan RM, Trzeciak S. Clinical review: emergency department overcrowding and the potential impact on the critically ill. Crit Care 2005;9:291-5.
2. Bernstein SL, Aronsky D, Duseja R, et al. The effect of emergency department crowding on clinically oriented outcomes. Acad Emerg Med Off J Soc Acad Emerg Med 2009;16:1-10.
3. Jeong H, Jeong S, Oh J, et al. Impact of Middle East respiratory syndrome outbreak on the use of emergency medical resources in febrile patients. Clin Exp Emerg Med 2017;4:94-101.
4. Higginson I. Emergency department crowding. Emerg Med J 2012;29:437-43.
5. Munro J, Mason S, Nicholl J. Effectiveness of measures to reduce emergency depart- ment waiting times: a natural experiment. Emerg Med J 2006;23:35-9.
6. Woods RA, Lee R, Ospina MB, et al. Consultation outcomes in the emergency depart- ment: exploring rates and complexity. CJEM 2008;10:25-31.
7. Yoon P, Steiner I, Reinhardt G. Analysis of factors influencing length of stay in the emergency department. CJEM 2003;5:155-61.
8. Nelson KA, Boslaugh SE, Hodge III D. Risk factors for extremely long length-of-stay among pediatric emergency patients. Pediatr Emerg Care 2009;25:835-40.
9. Cho SJ, Jeong J, Han S, et al. Decreased emergency department length of stay by ap- plication of a computerized consultation management system. Acad Emerg Med Off J Soc Acad Emerg Med 2011;18:398-402.
10. Richardson DB. The access-block effect: relationship between delay to reaching an inpatient bed and inpatient length of stay. Med J Aust 2002;177:492-5.
11. Chalfin DB, Trzeciak S, Likourezos A, Baumann BM, Dellinger RP. Impact of delayed transfer of critically ill patients from the emergency department to the intensive care unit. Crit Care Med 2007;35:1477-83.
12. Pines JM, Hollander JE, Localio AR, Metlay JP. The association between emergency department crowding and hospital performance on antibiotic timing for pneumonia and percutaneous intervention for myocardial infarction. Acad Emerg Med Off J Soc Acad Emerg Med 2006;13:873-8.
13. DeBehnke D, O’Brien S, Leschke R. Emergency medicine resident work productivity in an academic emergency department. Acad Emerg Med Off J Soc Acad Emerg Med 2000;7:90-2.
14. Kang H, Nembhard HB, Rafferty C, DeFlitch CJ. Patient flow in the emergency depart- ment: a classification and analysis of admission process policies. Ann Emerg Med 2014;64:335-42 [e8].