Abstract
Objectives: Prolonged admit wait times in the emergency department (ED) for patients who require
hospitalization lead to increased boarding time in the ED, a significant cause of ED congestion. This is
associated with decreased quality of care, higher morbidity and mortality, decreased patient satisfaction,
increased costs for care, ambulance diversion, higher numbers of patients who leave without being seen
(LWBS), and delayed care with longer lengths of stay (LOS) for other ED patients. The objective was to
assess the effect of a leadership-based program to expedite hospital admissions from the ED.
Methods: This before-and-after observational study was undertaken from 2006 through 2011 at one
community hospital ED. A team of ED and hospital leaders implemented a program to reduce admit wait
times, using a computerized hospital-wide tracking system to monitor inpatient and ED bed status. The
team collaboratively and consistently moved ED patients to their inpatient beds within an established
goal of 60 minutes after an admission decision was reached. Top leadership actively intervened in real
time by contacting staff whenever delays occurred to expedite immediate solutions to achieve the
60-minute goal. The primary outcome measures were the percentage of ED patients who were admitted
to inpatient beds within 60 minutes from the time the beds were requested and ED boarding time. LOS,
patient satisfaction, LWBS rate, and ambulance diversion hours were also measured.
Results: After ED census, hospital admission rates, and ED bed capacity were controlled for using a
multivariable linear regression analysis, the admit wait time reduction program contributed to an
increase in patients being admitted to the hospital within 60 minutes by 16 percentage points (95%
confidence intervals [CI] = 10 to 22 points; p < 0.0001) and a decrease in boarding time per admission of
46 minutes (95% CI = 63 to 82 minutes; p < 0.0001). LOS decreased for admitted patients by 79 minutes
(95% CI = 55 to 104 minutes; p < 0.0001), for discharged patients by 17 minutes (95% CI = 12 to
23 minutes; p < 0.0001), and for all patients by 34 minutes (95% CI = 25 to 43 minutes; p < 0.0001).
Patient satisfaction increased 4.9 percentage points (95% CI = 3.8 to 6.0 points; p < 0.0001). LWBS
patients decreased 0.9 percentage points (95% CI = 0.6 to 1.2 points; p < 0.0001) and monthly ambulance
diversion decreased 8.2 hours (95% CI = 4.6 to 11.8 hours; p < 0.0001).
Conclusions: A leadership-based program to reduce admit wait times and boarding times was
associated with a significant increase in the percentage of patients admitted to the hospital within
60 minutes and a significant decrease in boarding time. Also associated with the program were
decreased ED LOS, LWBS rate, and ambulance diversion, as well as increased patient satisfaction.
ACADEMIC EMERGENCY MEDICINE 2014; 21:266–273 © 2014 by the Society for Academic Emergency
Medicine
I
nternationally, emergency medical care systems are
overburdened with patients backing up in the emer- This predicament of overcrowding led the Institute of
gency department (ED) because they cannot get Medicine (IOM) to describe emergency medicine (EM)
From the Department of Emergency Medicine, The Permanente Medical Group, Kaiser Permanente Medical Centers (PBP, DRV),
Sacramento and Roseville, CA; and the Biostatistical Consulting Unit, Division of Research, Kaiser Permanente (MAC), Oakland, CA.
Received June 16, 2013; revision received August 16, 2013; accepted September 12, 2013.
Funded by a Kaiser Permanente Northern California Central Research Committee Community Benefit Grant, The Permanente
Medical Group, Division of Research, Kaiser Permanente. The authors have no relevant financial information or potential conflicts
of interest to disclose.
Supervising Editor: Sandra M. Schneider, MD.
Address for correspondence and reprints: Pankaj B. Patel, MD; e-mail: Pankaj.Patel@KP.org.
as in crisis, even at the breaking point.1 Prolonged vices Institutional Review Board reviewed this study
admit wait times for patients being held in the ED until and granted it an exemption from full review.
the inpatient ward can accept them (a practice known
as “boarding”) is a significant cause of ED congestion Study Setting and Population
and crowding.2–11 The adverse effects of prolonged We conducted this study between January 2006 and
admit wait times and the concomitant ED crowding are December 2011 at one urban community hospital within
profound and include measurable increases in patient a large integrated health care delivery system serving
morbidity and mortality, longer lengths of stay (LOS), approximately 3.3 million members at 21 hospitals and
more patients who leave without being seen (LWBS), over 160 medical offices. The study hospital has 287
and increased costs.5,6,12–26 Prolonged admit wait time licensed beds with units for intensive care, telemetry,
is a major cause for ambulance diversion, which directs general medical–surgical, neuroservices (includes neuro-
patients away from their desired hospital destinations surgery), and oncology, but is not a designated trauma
and delays their care.1,19,27–29 Patient satisfaction is also center. The hospital is affiliated with the University of
adversely affected by ED crowding and admit wait time California, Davis, School of Medicine with rotating resi-
delays.30,31 dents from allergy/immunology, EM, family medicine,
Because boarding admitted patients contributes sig- internal medicine, neurosurgery, obstetrics/gynecology,
nificantly to ED crowding, the IOM has recommended ophthalmology, orthopedics, pediatrics, plastic surgery,
an end to this practice.1,32 Several studies have demon- podiatry, radiology, surgery, and urology. The ED is
strated the significant advantages that result from staffed by one group of EM board-certified or board-eli-
reducing admit wait times, which can relieve ED con- gible physicians and provides care to a broad spectrum
gestion by decreasing ED LOS,13,27,33 ambulance diver- of patients that include pediatric and obstetric patients
sion hours,27,29,33 patients who LWBS,29 and costs.34 (even though the hospital does not have inpatient pedi-
Although various solutions to reduce ED boarding have atric or obstetric services). There were no significant
been identified, these proven strategies appear to be changes to ED staffing during the study period. The
underutilized, as most hospitals report having imple- average annual ED census during the study period was
mented very few of them.35,36 76,169 patients.
ED crowding can be reduced by expediting the trans-
fer of admitted patients out of the ED, thereby reducing Study Protocol
boarding times for patients held in the ED awaiting A process improvement plan was developed to reduce
transfer to inpatient units. Our ED implemented a lead- admit wait times, including all hours of the day and
ership-based admit wait time reduction program and week (nights, weekends, and holidays). A team of ED
then measured and analyzed the effect of this approach and hospital leaders was convened to work toward
to expediting ED hospital admissions on admit wait admit wait time reduction. A computerized tracking sys-
times, boarding times, LOS, patient satisfaction, LWBS tem called HealthConnect (Epic Systems Corporation,
patients, and ambulance diversion hours. Verona, WI) was used to monitor inpatient and ED bed
status in real time to assess admit wait times for every
METHODS ED patient who required admission to the hospital.
Measurable steps, goals, and reports were identified
Study Design and developed. Top leadership collaboratively and con-
This was a retrospective, observational, before-and- sistently tracked wait time metrics to assure progress
after study to compare the effects of a leadership-based toward the goal of admit wait time reduction (see
intervention on admit wait times and boarding time. Figure 1). No significant changes were made in ED
The Kaiser Permanente Northern California Health Ser- work flow or emergency physician staffing during the
1. Patient enters ED
2. Evaluation by ED physician
LOS for admitted
patient = from step #1
to step #7
3. Evaluation by consultant/hospitalist
Figure 1. Flow diagram for ED process time intervals. LOS = length of stay.
268 Patel et al. • ADMISSION WAIT TIME REDUCTION
study period. Consulting physicians were responsible vention period was from February 2008 through
for determining which patients required admission to December 2011.
the hospital and alerted the ED staff to their decisions
to admit by entering the order “admit to hospital” into Outcome Measures
the tracking system, which initiated the request for the Emergency department bed capacity, ED census, and
appropriate inpatient bed. This electronic entry became overall admission rates were collected. The primary out-
the starting time for the admit wait time interval. Once come measures were the percentage of ED patients
the overall goal was established to admit ED patients admitted to inpatient beds within 60 minutes from the
within 60 minutes of the bed requests, the overall admit time the beds were requested and ED boarding time
wait time was divided into two interval segments that per admission. Explicit measures of ED throughput and
were captured on the tracking system: 1) from the time patient satisfaction were collected, both before and after
an inpatient bed was requested to the time the hospital the implementation of the admit wait time reduction
ward accepted the ED patient (this was the primary program. Summarized monthly outcome data included
responsibility of hospital staff) and 2) from the time the percentage of patients admitted within 60 minutes,
ED patient was accepted by the inpatient ward to the boarding time per admission, LOS (for admitted
time the ED patient left the ED (this was the primary patients, discharged patients, and all patients), overall
responsibility of ED staff). Hospital and ED staff each boarding time, patient satisfaction scores, percentage of
sought to achieve their respective interval segments patients who LWBS, and ambulance diversion hours.
within 30 minutes. If hospital staff and ED staff each The patient satisfaction survey used in this study was
met their target goals, admission was achievable within not specifically designed for this project and has been
the overall 60-minute goal (less than 30 minutes by hos- previously described as a proprietary survey for our
pital and less than 30 minutes by ED). organization.37 Briefly, our medical group’s regional
The period from January 2006 through January 2008 Department of Access and Service Assessment sends
was the preintervention period. In February 2008, the out surveys to patients who have been seen in each of
admission wait time reduction program was implemented. the 21 hospital EDs in our Northern California region.
Interventions included the following: hospital and ED lead- Questions on the survey cover a variety of parameters
ership meetings with nurse managers, charge nurses, of ED care, including evaluation of the ED physician,
admitting department, discharge planning, information the coordination of care during the ED visit, and cour-
technology, housekeeping and environmental services, tesy and helpfulness of the staff. The patients are asked
physicians, and pharmacy; the systematic dissemination of to rate each item using a five-point Likert scale: poor,
tracked admit wait time data for everyone to review; and fair, good, very good, and excellent. The reported score
the real-time contact between leaders and their staff, to is the total percentage of patients who rank their overall
improve the flow of admitted patients from the ED to the ED care as very good or excellent. Annually, the regio-
inpatient wards. At regular meetings and via e-mail distri- nal department sent approximately 10,000 surveys to
bution, daily, weekly, and monthly tracking data were patients who had been treated in the study ED, with
made available for everyone to review. Top leadership approximately 2,500 returned surveys from which
reviewed this information frequently and often made patient satisfaction scores are reported for this study.
direct contact with their staff in real time if delays were
observed in either of these two interval segments, allowing Data Analysis
immediate resolution of issues that were causing the Descriptive measures of the variables were calculated.
delays. The admit wait time reduction program required We present continuous data as medians with interquar-
collaborative staff involvement from the ED, admitting tile ranges (IQRs). Variables were compared for signifi-
department, hospital inpatient ward, discharge planning, cance pre- and post–program implementation using the
housekeeping and environmental services (to get beds Wilcoxon-Mann-Whitney rank-sum test. Spearman cor-
cleaned and ready for newly admitted patients), and infor- relations and significance were calculated to assess the
mation technology (to assure tracking system functioning). bivariate influence each continuous confounder had on
The ED had 36 rooms and nine urgent care rooms each outcome. A categorical variable indicating the
which were physically separate from the main ED, for a decrease in ED bed capacity from 45 to 36 beds after
total of 45 rooms. In January 2010, the urgent care sec- January 2010 was used in the analysis. The bivariate
tion was closed, resulting in a total ED bed capacity of effects of this categorical variable on each outcome
36 rooms, with all patients managed thereafter in the were calculated using the Wilcoxon-Mann-Whitney
36-room ED (including those who had previously pre- rank-sum test.
sented to the ED and had been triaged to the urgent To determine if there were any trending, seasonal, or
care section). The urgent care closure did not affect the autoregressive effects that needed to be controlled for
processes for admission, as the admission rate for this in the outcomes, we first chose to conduct an inter-
group of patients was well below 1%. All of these rupted time series analysis. We used an autoregressive,
patients who previously came through the ED had med- integrated, moving average (ARIMA) model for each
ical screening examinations performed by credentialed outcome. Ultimately, multivariable linear regression
ED nurses and were then sent to the urgent care models were used to assess the effect of program imple-
located on site, but outside the footprint of the ED. As mentation on the outcome measures when controlling
the admit wait time process was streamlined, these for continuous and categorical confounders. Data were
patients were kept within the ED itself, incorporating analyzed using SAS 9.3 for Windows (SAS Institute
the urgent care staff into the ED as well. The postinter- Inc., Cary, NC).
ACADEMIC EMERGENCY MEDICINE • March 2014, Vol. 21, No. 3 • www.aemj.org 269
RESULTS
Ambulance
Diversion
(Hours)
(54)
During the study period (2001 to 2011), average annual
68
116
138
107
31
13
0
census was 76,169 with average annual admissions of
9,364 (12.3%). Annual mean data are presented in Table 1.
Actual pre- and postimplementation data (median,
Satisfaction
(3.3)
IQR) at the program site on patients admitted within
Patient
67.6
67.1
69.7
73.4
73.4
76.0
71.2
(%)
60 minutes are shown in Figure 2. Boarding time per
admission, LOS, patient satisfaction, LWBS patients,
ambulance diversion hours, ED census, hospital admis-
sion rates, and ED bed capacity are noted in Table 2.
(0.5)
ing variables and the outcomes to directly assess the
1.7
1.1
1.1
0.3
0.5
0.6
0.9
effect on ED throughput that can be attributed to these
factors. Spearman’s correlation coefficients and corre-
sponding p-values were calculated between the three
confounding variables (ED census, hospital admission
Patients (hr:min)
LOS Discharged
rates, and ED bed capacity) and each of the outcome
(0:28)
variables (results not reported).
3:57
3:52
3:55
3:33
2:53
2:50
3:30
We used an ARIMA model to assess for trends and
autoregressive factors. When no autoregressive factors
were identified (including seasonal trends), we contin-
ued the analysis using multivariable linear regression
for parsimony. In our multivariable linear regression
(hr:min)
Patients
LOS All
(0:41)
4:57
4:46
4:48
4:09
3:21
3:14
4:12
analysis (see Table 3), after controlling for the con-
founding variables listed above, there was an increase
in patients being admitted within 60 minutes to the hos-
LOS Admits
pital of 16 percentage points (95% CI = 10 to 22;
(hr:min)
(0:56)
p < 0.0001), an average decrease in boarding time per
8:35
8:35
8:26
7:10
6:40
6:19
7:37
admission of 46 minutes, and an average decrease in
LOS for admitted patients of 79 minutes. There was also
an average increase in patient satisfaction of 4.9
Minutes/Admit
(29)
93
85
88
41
32
21
60
sion of 8.2 hours per month
Staff from the urgent care were absorbed into the ED
after the urgent care was closed. A review of physician
and nonphysician staffing during the entire study per-
Time (Hours)
15,627
13,365
14,539
DISCUSSION
Admit*
(13.2)
Within
33.5
34.4
36.4
52.6
59.7
66.9
47.3
(%)
(2,208)
Census
74,164
75,044
78,818
79,551
73,907
75,532
76,169
(SD)
Mean
2007
2008
2009
2010
2011
2006
Year
Table 2
Pre- and Postintervention Outcomes
IQR = interquartile range; LOS = length of stay; LWBS = leave without being seen; NS = not significant.
*Program implemented February, 2008.
†Patients admitted within 60 minutes data not available January 2006–April 2006.
‡Patient satisfaction reported as percentage of patients who rate their care very good or excellent.
§ED bed capacity decreased January 2010 from 45 to 36 beds.
ACADEMIC EMERGENCY MEDICINE • March 2014, Vol. 21, No. 3 • www.aemj.org 271
Table 3
Multivariable Regression Results: Average Contribution of Program to Outcomes*
Point. Washington, DC: National Academies Press, for emergency medical admissions. Eur J Emerg
2006. Med 2011;18:192–6.
2. Schneider S. A letter from ACEP president Sandra 19. Falvo T, Grove L, Stachura R, et al. The opportunity
Schneider. Emerg Phys Monthly 2011;18:3. loss of boarding admitted patients in the emergency
3. Viccellio A, Santora C, Singer AJ, Thode HC Jr, department. Acad Emerg Med 2007;14:332–7.
Henry MC. The association between transfer of 20. Huang Q, Thind A, Dreyer JF, Zaric GS. The impact
emergency department boarders to inpatient hall- of delays to admission from the emergency depart-
ways and mortality: a 4-year experience. Ann Emerg ment on inpatient outcomes. BMC Emerg Med
Med 2009;54:487–91. 2010;10:16.
4. Moskop JC, Sklar DP, Geiderman JM, Schears RM, 21. Pines JM, Batt RJ, Hilton JA, Terwiesch C. The
Bookman KJ. Emergency department crowding, financial consequences of lost demand and reducing
part 1–concept, causes, and moral consequences. boarding in hospital emergency departments. Ann
Ann Emerg Med 2009;53:605–11. Emerg Med 2011;58:331–40.
5. Stolte E, Iwanow R, Hall C. Capacity-related interfa- 22. Bernstein SL, Aronsky D, Duseja R, et al. Society
cility patient transports: patients affected, wait times for Academic Emergency Medicine, Emergency
involved and associated morbidity. CJEM 2006; Department Crowding Task Force. The effect of
8:262–8. emergency department crowding on clinically ori-
6. Clark K, Normile LB. Influence of time-to-interven- ented outcomes. Acad Emerg Med 2009;16:1–10.
tions for emergency department critical care 23. Verdile VP. Sutton’s Law need not apply. Ann
patients on hospital mortality. J Emerg Nurs Emerg Med 2011;58:341–2.
2007;33:6–13. 24. ACEP Board of Directors. Policy statement: defini-
7. Garson C, Hollander JE, Rhodes KV, Shofer FS, tion of boarded patient. Ann Emerg Med
Baxt WG, Pines JM. Emergency department patient 2011;57:548.
preferences for boarding locations when hospitals 25. Sills MR, Fairclough D, Ranade D, Kahn MG. Emer-
are at full capacity. Ann Emerg Med 2008;51:9–12. gency department crowding is associated with
8. Hoot NR, Aronsky D. Systematic review of emer- decreased quality of care for children with acute
gency department crowding: causes, effects, and asthma. Ann Emerg Med 2011;57:191–200.
solutions. Ann Emerg Med 2008;52:126–36. 26. Singer AJ, Thode HC Jr, Viccellio P, Pines JM. The
9. Olshaker JS, Rathlev NK. Emergency department association between length of emergency depart-
overcrowding and ambulance diversion: the impact ment boarding and mortality. Acad Emerg Med
and potential solutions of extended boarding of 2011;18:1324–9.
admitted patients in the emergency department. 27. Quinn JV, Mahadevan SV, Eggers G, Ouyang H,
J Emerg Med 2006;30:351–6. Norris R. Effects of implementing a rapid admission
10. Derlet RW, Richards JR. Ten solutions for emer- policy in the ED. Am J Emerg Med 2007;25:559–63.
gency department crowding. West J Emerg Med 28. Patel PB, Derlet RW, Vinson DR, Williams M, Wills
2008;9:24–7. J. Ambulance diversion reduction: the Sacramento
11. Millard WB. Stand by to repel boarders. Ann solution. Am J Emerg Med 2006;24:206–13.
Emerg Med 2011;57:A15–19. 29. Kelen GD, Scheulen JJ, Hill PM. Effect of an emer-
12. Sprivulis PC, Da Silva JA, Jacobs IG, Frazer AR, gency department (ED) managed acute care unit on
Jelinek GA. The association between hospital ED overcrowding and emergency medical services
overcrowding and mortality among patients admit- diversion. Acad Emerg Med 2001;8:1095–100.
ted via Western Australian emergency departments. 30. Howell EE, Bessman ES, Rubin HR. Hospitalists
Med J Aust 2006;184:208–12. and an innovative emergency department admission
13. Lucas R, Farley H, Twanmoh J, et al. Emergency process. J Gen Intern Med 2004;19:266–8.
department patient flow: the influence of hospital 31. Pines JM, Iyer S, Disbot M, Hollander JE, Shofer
census variables on emergency department length FS, Datner EM. The effect of emergency department
of stay. Acad Emerg Med 2009;16:597–602. crowding on patient satisfaction for admitted
14. Gilligan P, Winder S, Singh I, Gupta V, Kelly PO, patients. Acad Emerg Med 2008;15:825–31.
Hegarty D. The Boarders in the Emergency Depart- 32. Asplin BR, Magid DJ. If you want to fix crowding,
ment (BED) study. Emerg Med J 2008;25:265–9. start by fixing your hospital. Ann Emerg Med
15. Schull MJ, Vermeulen M, Slaughter G, Morrison L, 2007;49:273–4.
Daly P. Emergency department crowding and 33. Fatovich DM, Nagree Y, Sprivulis P. Access block
thrombolysis delays in acute myocardial infarction. causes emergency department overcrowding and
Ann Emerg Med 2004;44:577–85. ambulance diversion in Perth. Western Australia.
16. Phua J, Ngerng WJ, Lim TK. The impact of a delay Emerg Med J 2005;22:351–4.
in intensive care unit admission for community- 34. Moloney ED, Smith D, Bennett K, O’riordan D, Silke
acquired pneumonia. Eur Respir J 2010;36:826–33. B. Impact of an acute medical admission unit on
17. McCarthy ML, Zeger SL, Ding R, et al. Crowding length of hospital stay, and emergency department
delays treatment and lengthens emergency depart- ‘wait times’. QJM 2005;98:283–9.
ment length of stay, even among high-acuity 35. Rabin E, Kocher K, McClelland M, et al. Solutions
patients. Ann Emerg Med 2009;54:492–503. to emergency department ‘boarding’ and crowding
18. Plunkett PK, Byrne DG, Breslin T, Bennett K, Silke are underused and may need to be regulated.
B. Increasing wait times predict increasing mortality Health Aff 2012;31:1757–66.
ACADEMIC EMERGENCY MEDICINE • March 2014, Vol. 21, No. 3 • www.aemj.org 273