You are on page 1of 9

ORIGINAL RESEARCH CONTRIBUTION

Decreasing Length of Stay in the Emergency


Department With a Split Emergency Severity
Index 3 Patient Flow Model
Rajiv Arya, MD, Grant Wei, MD, Jonathan V. McCoy, MD, Jody Crane, MD, MBA, Pamela Ohman-
Strickland, PhD, and Robert M. Eisenstein, MD

Abstract
Objectives: There has been a steady increase in emergency department (ED) patient volume and wait
times. The desire to maintain or decrease costs while improving throughput requires novel approaches
to patient flow. The break-out session “Interventions to Improve the Timeliness of Emergency Care” at
the June 2011 Academic Emergency Medicine consensus conference “Interventions to Assure Quality in
the Crowded Emergency Department” posed the challenge for more research of the split Emergency
Severity Index (ESI) 3 patient flow model. A split ESI 3 patient flow model divides high-variability ESI 3
patients from low-variability ESI 3 patients. The study objective was to determine the effect of
implementing a split ESI 3 flow model has on patient length of stay (LOS) for discharged patients.
Methods: This was a retrospective chart review at an urban academic ED seeing over 70,000 adult
patients a year. Cases consisted of adults who presented from 9 a.m. to 11 p.m. from June 1, 2011, to
December 31, 2011, and were discharged. Controls were patients who presented on the same times
and days, but in 2010. Visit descriptors included age, race, sex, ESI score, and first diagnosis. The first
diagnosis was coded based on methods used by the Agency for Healthcare Research and Quality to
codify International Classification of Diseases, ninth version, into disease groups. Linear models
compared log-transformed LOS for cases and controls. A front-end ED redesign involved creating
guidelines to split ESI 3 patients into low and high variability, a hybrid sort/triage registered nurse, an
intake area consisting of an internal results waiting room, and a treatment area for patients after initial
assessment. The previous low-acuity area (ESI 4s and 5s) began to see low-variability ESI 3 patients as
well. This was done without additional beds. The intake area was staffed with an attending emergency
physician (EP), a physician assistant (PA), three nurses, two medical technicians, and a scribe.
Results: There was a 5.9% decrease, from 2.58 to 2.43 hours, in the geometric mean of LOS for
discharged patients from 2010 to 2011 (95% confidence interval CI = 4.5% to 7.2%; 2010, n = 20,215;
2011, n = 20,653). Abdominal pain was the most common diagnostic grouping (2010, n = 2,484; 2011,
n = 2,464) with a reduction in LOS of 12.9%, from 4.37 to 3.8 hours (95% CI = 10.3% to 15.3%).
Conclusions: A split ESI 3 patient flow model improves door-to-discharge LOS in the ED.
ACADEMIC EMERGENCY MEDICINE 2013; 20:1171–1179 © 2013 by the Society for Academic
Emergency Medicine

From the Department of Emergency Medicine (RA, GW, JVM, RME) and the Department of Biostatistics (PO), UMDNJ/ Robert
Wood Johnson Medical School, New Brunswick, NJ; and the Mid-Atlantic Permanente Medical Group (JC), Rockville, MD.
Received April 19, 2013; revision received June 12 and June 18, 2013; accepted June 20, 2013.
Presented at the Society for Academic Emergency Medicine Annual Meeting, Atlanta, GA, September 2013.
The authors have no relevant financial information or potential conflicts of interest to disclose.
Supervising Editor: Lowell Gerson, PhD.
Address for correspondence and reprints: Rajiv Arya, MD; e-mail: aryara@umdnj.edu.
Consensus Conference Follow Up 2011.
Editor’s Note: Academic Emergency Medicine highlights articles that follow up on the research agendas created at the journal’s
annual consensus conferences. This article relates to the 2011 AEM consensus conference, “Interventions to Assure Quality in the
Crowded Emergency Department.” All prior consensus conference proceedings issues are available open-access at www.aemj.
org, and authors interested in submitting consensus conference follow-up papers should consult the author guidelines. http://
onlinelibrary.wiley.com/journal/10.1111/%28ISSN%291553-2712/homepage/ForAuthors.html.

© 2013 by the Society for Academic Emergency Medicine ISSN 1069-6563 1171
doi: 10.1111/acem.12249 PII ISSN 1069-6563583 1171
1172 Arya et al. • DECREASING LOS WITH SPLIT ESI PATIENT FLOW

T
here has been a steady increase in emergency adults, age greater than 21 years, who presented from 9
department (ED) patient visits and a decrease in a.m. to 11 p.m. from June 1, 2011, to December 31 2011.
the number of EDs over the past 20 years.1 Controls were patients who presented the same time
Recently enacted legislation aimed at reducing costs and the same calendar days in 2010.
and increasing access will likely further increase ED
demand while driving the necessity for reducing cost Study Protocol
per patient. This need to decrease cost and improve Our site implemented a split ESI 3 flow model to effi-
throughput requires novel approaches to patient flow. ciently manage patients presenting during the busi-
The breakout session “Interventions to Improve the est hours of the day, 9 a.m. to 11 p.m. The
Timeliness of Emergency Care” at the June 2011 implementation of a split ESI 3 flow model required six
Academic Emergency Medicine consensus conference major changes: splitting of ESI 3 patients, infrastruc-
“Interventions to Assure Quality in the Crowded Emer- ture, staffing, responsibilities, purpose of the area, and
gency Department” posed the challenge for more a roll-out plan. Each will be described in detail, and
research on the split Emergency Severity Index (ESI) 3 then a few patient examples will be presented to dem-
patient flow model and greater efficiency.2,3 onstrate the application.
The ESI is reported to be the most commonly adopted
triage acuity system in the United States.4,5 Similar Splitting ESI 3 Patients. The intake17 area saw all ESI
scales are in use in Canada, the United Kingdom, Aus- 4 and 5 patients as well as low-variability ESI 3. Low-
tralia, and other countries. The boards of the American variability ESI 3 patients required two or more
College of Emergency Physicians and the Emergency resources and typically follow a standardized work flow:
Nurses Association support the adoption of five-level assessment, diagnostics, medications, and one reassess-
triage scales. ESI’s five-level system is driven by patient ment prior to disposition. They rarely require significant
acuity and resource utilization.6 Levels 1 and 2 are physician or physician assistant (PA) attention above
distinguished by instability of vital signs or severity of and beyond the initial assessment and final disposition
presenting complaint, while levels 3, 4, and 5 are con- reassessment. The physician and nursing leadership
sidered less acute and thus are differentiated by the worked together and created a guide to determine who
number of resources required. Examples of resources was considered a low-variability ESI 3. Some of the rec-
include diagnostic testing and interventions. Level 4 and ommendations are institution-specific; e.g., we are
5 patients require one or no resources, respectively, and unable to perform a private pelvic examination in the
are traditionally treated in designated low-acuity areas intake area (Table 1). The team determined that the
of the ED such as “fast tracks.” A split patient flow intake area would see about six patients per hour,
model for ESI Level 3 patients segments high-variability between a physician and PA, and wanted to limit the
and low-variability ESI 3 patients into separate streams patients to those who required no more than 20 min-
with the intent of driving them through a more custom- utes of provider time for assessment, orders, and reas-
ized and efficient process. The ultimate goal is to reduce sessment.
variability in the care of these patients, realizing
improvements in metrics such as door-to-doctor times, Infrastructure. Intake was divided into four distinct
length of stay (LOS), left without being seen, and patient zones: assessment, treatment area, results waiting room,
satisfaction.7 The benefits of reducing variability have and the discharge area. The intake assessment area was
been described in fast track settings.8–10 The split-flow equipped with six reclining chairs where the staff per-
model works in a similar manner, reducing variability, formed assessments and procedures and administered
reducing LOS, and enhancing service responsiveness. initial treatments. In addition to the six assessment
An ever-growing body of literature demonstrates the chairs, a discharge chair was designated to review
many negative consequences of the ED overcrowding results with patients and perform the discharge encoun-
and boarding crisis.11,12 Downstream effects include ters. The assessment area was our prior fast track and
delays in care, decreased patient and physician satisfac- required no construction. The treatment area consisted
tion, and increased patient mortality.13–16 Our objective of seven curtained treatment bays with stretchers,
was to determine the effect of implementing a split ESI 3 which were previously part of the main ED. During the
flow model has on patient LOS for all discharged patients. times intake was open it was used for those patients
and returned to the main ED after intake closed. The
bays were the closest to the intake assessment area and
METHODS
no construction was required. The results waiting room
Study Design was defined as an internal waiting room and contained
We conducted a retrospective chart review to examine five reclining chairs with a TV, a desk, and a computer.
the effect of a split ESI 3 flow model on the LOS for The results waiting room was a conference room that
patients who were discharged from the ED. The process was repurposed to clinical space and only required
had a go-live date of May 31, 2011. The study protocol cosmetic changes. The discharge area was a pediatric
was approved by the institutional review board of the triage room near the front entrance of the ED that was
medical school. no longer in use and was converted to a registration
work space.
Study Setting and Population
This study was conducted in an urban, academic ED that Staffing. The intake assessment area was staffed with
sees over 70,000 adult patients a year. Cases consisted of one emergency physician (EP), one PA, one scribe, two
ACADEMIC EMERGENCY MEDICINE • November 2013, Vol. 20, No. 11 • www.aemj.org 1173

Table 1
Guide to Sort for Split of ESI 3

Sort to Intake Do Not Sort to Intake


Head and neck Abdominal complaint (in extremis)
Dental complaint Headache, rule out meningitis
Ear complaint Headache, “worst headache of life”
Eye complaint Chest pain (except as described in intake)
Sore throat Shortness of breath (except as described in intake)
Facial complaint Altered mental status
Head injury, normal MS Weakness, >40 yr
Neck pain Dizziness, >40 yr
Headache—history of migraines not “worst headache of life” Syncope
Nose bleeds Cerebrovascular accident
Chest Hypotension
Cough/congestion age < 60 yr Symptomatic hypertension
Asthma Fall, >60 yr (if not able to ambulate)
Shortness of breat—possible bronchitis, respiratory Vaginal complaint (bleeding, discharge, etc)
Breast complaint Pelvic pain
Genitourinary Oncology patients
Groin complaint Transplant patients
Urinary complaint (level 3) Any condition requiring isolation (i.e., chicken pox, shingles)
Penile discharge/pain/complaint Pregnancy related
Ortho Any patient requiring cardiac monitoring
Extremity pain—stable, no obvious deformity, no open wound.
Back pain, ambulatory, minor accident/injury
Leg swelling possible deep venous thrombosis (level 3)
Skin
Abscess
Laceration
Rash
Cellulitis—exclude extensive cellulitis
Allergic reaction
Wound check
Skin problem
Puncture wound
Foreign body
Insect bite
Animal bite
General
Flu-like symptoms—age < 40 yr
Other
Rabies
Medication refill
Superficial bleeding
Extremity swelling—trauma (level 4)
Motor vehicle crash—only musculoskeletal
Abdomen
Abdominal pain—not in extremis, age < 55 yr
Flank pain—age <40 yr
Nausea or vomiting
Rectal complaint
Minor gastrointestinal bleed

ESI = Emergency Severity Index.

nurses, and one medical technician. The results waiting assessment area for the patient. The physician or PA
room was staffed by one medical technician. The treat- would perform the medical evaluation and assessment.
ment area was staffed by one nurse, and the discharge The scribe would prepare the medical record, update
kiosk was staffed by a registration clerk who completed the providers on results, and prepare discharge instruc-
the final registration paperwork. There was no increase tions and prescriptions. After placing the patient in a
in the physician or PA staffing hours. Sixteen additional treatment bed, the assessment nurse would place the
nurse and medical technician hours were added. initial intravenous (IV) line, draw blood, and medicate
the patient as needed. The assessment medical techni-
Responsibilities and Purpose. Each provider was cian would transport the patient to radiology and per-
accountable to ensure that each patient flowed smoothly form general medical technician duties. The treatment
through the process (Table 2). The triage nurse or the nurse would provide ongoing management of the
assessment nurses would direct-bed the patient based patient and would update the physician and PAs to the
on initial presenting complaint, which was compared to patient’s progress. The results waiting room medical
the list created by the team to determine the proper technician would transport the patient in and out of the
1174 Arya et al. • DECREASING LOS WITH SPLIT ESI PATIENT FLOW

Table 2
Intake Infrastructure, Staffing, and Purpose

Area Staffing Beds Purpose


Intake assessment One physician Six reclining chairs, Assessment, initial medications, phlebotomy, and procedures.
One PA one discharge chair
Two nurses
One medical
technician
One scribe
Results waiting One medical Five lounge chairs Decompress the clinical assessment area while patients wait
technician for consultants, ancillary services, and results.
Treatment One nurse Seven stretchers Provide medical treatment for patients that are not safe to sit
in a lounge chair.
Discharge kiosk One registration Completion of the registration process that does not occupy
clerk a clinical space.

PA = physician assistant.

room, would observe each patient for change in condi- the intake area were often seen by the nurse and pro-
tions, and would keep the patient informed as to the vider simultaneously, further streamlining the process.
status of his or her test results and treatment progress. For example, a suspected renal colic patient, an ideal
Each area within intake had a unique role relative to example of a low-variability ESI 3 patient, would be
the overall flow. The assessment area is where the phy- seen by the physician and initial medications and fluids
sician and PA would see the patients and the nurses started in intake. From there the patient would go to
would perform initial nursing tasks. The treatment area radiology for a computed tomography scan and would
enabled patients who could not stay vertical during return to either the results waiting room or the treat-
their entire clinical encounter (i.e., those medicated with ment area, depending on the type of medication he or
IV opiates) to assume treatment beds while undergoing she was given in intake. High-risk medications, such as
evaluation and management. The results waiting room IV opiates, required the patient to go to a treatment
provided an additional, more efficient location for bed, while patients receiving oral pain relievers, which
patients to await ancillary results, enabling the assess- they could otherwise take at home, could be managed
ment area to maintain flow and enhancing overall bed from the results waiting room. For a patient having
capacity. Finally, the discharge area allowed the collec- continued symptoms, the treatment nurse would contact
tion of the financial information and the completion of the intake physician for additional medication orders.
the registration process to occur in a nonclinical setting, When all of the diagnostic testing was complete and
further improving bed capacity. The main ED had bed- resulted and the nurse felt the patient was clinically
side registration. ready for discharge, the treatment nurse would contact
the physician to discharge the patient.
Patient Flow. In this flow model, patients initially pre-
sented to the registration clerk, who performed quick Rollout Plan. The ED redesign team was composed of
registration (name, date of birth, chief complaint). Based EP and nursing leadership along with three staff physi-
on the chief complaints, patients were pulled into the cians, one staff PA, two staff medical technicians, and
intake area or waited for a more traditional triage six staff nurses who worked to tailor the split ESI 3 flow
encounter (Figure 1). Those patients pulled directly to model to our department. A rapid-cycle testing model
for process improvement was employed with 10
pilot days that were performed prior to going live on
Quick May 31, 2011.18 The test pilots solidified the timing,
Registration staffing, responsibilities, patient flow, and education of
the splitting of the ESI 3. To differentiate between low-
Triage/Sort Intake Area
and high-variability ESI 3 patients, the nursing staff
composed a list of qualifying chief complaints. Where
there were disagreements, the nursing design team
Results Treatment
Waiting Room reviewed selected triage encounters in a blinded fash-
Area
ion, each rating the patient encounter as intake or main.
Discharge The cases where there was a definite majority opinion
Discharge
Kiosk were incorporated into the final criteria, and the staff
To the Main was subsequently educated on these criteria (Table 1).
ED
Admission
Data Extraction. An electronic medical record (EMR)
was used to extract patient demographics, clinical data,
Figure 1. Process for split ESI 3. ESI = Emergency Severity arrival and discharge times, and the final disposition.
Index. The EMR recorded a time stamp for each significant
ACADEMIC EMERGENCY MEDICINE • November 2013, Vol. 20, No. 11 • www.aemj.org 1175

patient milestone. The main outcome variable, LOS for overall results are not strongly influenced by the extreme
discharged patients, was calculated as the difference in values in the tail of the positively skewed distribution.
time between when the patient was entered (arrival Additional exploratory analyses evaluated whether
time) and removed (discharge time) from the EMR. ESI score or diagnosis code (where analyses were
Patients were entered into the EMR by the quick regis- restricted to the 12 most common diagnoses) modified
tration clerk. All patients were manually removed from the difference between cases and controls. In these
the EMR by one of the following individuals: nurse, analyses, indicator variables were used to represent the
physician, PA, registration clerk, or scribe. Visit de- categories of ESI or diagnosis code. Interactions
scriptors included age, race, sex, ESI score, and first between case/control status and the modifiers were
diagnosis. The first diagnosis was coded based on used to assess effect modification. With two modifying
methods used by the Agency for Healthcare Research variables of interest, it was noted whether they were
and Quality to codify International Classification of Dis- significant while maintaining a familywise error rate of
eases, ninth version, into disease groups.19 0.05 using the Bonferroni correction.

Data Analysis RESULTS


Descriptive statistics summarized the distributions of
both the outcome of interest LOS for each visit and the We analyzed 20,653 cases from 2011 and 20,215 con-
visit descriptors. Independent-sample t-tests and trols from 2010 for the study period. ESI level differed
chi-square tests were used, as appropriate, to compare significantly across years. No other visit descriptors dif-
distributions of the visit descriptors across years. fered significantly across years (Table 3). Linear models
Since LOS was positively skewed, a log-transforma- found significant differences between cases and con-
tion was taken to stabilize the variance before more trols, both unadjusted and adjusted for sex, age, and
formal inferential statistics were conducted. Additional race (p < 0.0001). Using the unadjusted regression coef-
histograms confirmed that this resulted in bell-shaped ficients, we estimate that there was a 5.9% decrease
distributions for the LOS, with homogeneous variances (calculated as [exp(beta for year) – 1) 9 100%]) in the
between cases and controls. Linear models compared geometric mean of LOS from controls relative to cases
log-transformed LOS for cases and controls, both (95% confidence interval [CI] = 4.5% to 7.2%). The
unadjusted as well as adjusted for age, race, and sex. adjusted estimates were similar.
These analyses included a random effect for patient to We found that ESI significantly modified the differ-
account for correlations between visits by the same ence in LOS between cases and controls (p = 0.0086).
patient. That is, the difference varied by levels of ESI. Thus,
All analyses were conducted using SAS for Windows, these results were further stratified by ESI and first
version 9.3 (SAS Institute Inc., Cary, NC). Results are diagnosis code (12 most frequent; Table 4). For visits
expressed as geometric means and percent changes in with ESI 1 (most critical), the difference between cases
these means, calculated as the exponential of the and controls was not significant (p = 0.40). For ESI 2,
least-square means of the adjusted means of the log- the difference was marginally significant (p = 0.033);
transformed values or the regression coefficients sum- however, the lower endpoint of the 95% CI was very
marizing the difference between adjusted means, close to zero. For ESI 3 or 4, LOS for cases were signifi-
respectively. Use of the log-transformed values and geo- cantly lower than those for controls (p = 0.0011 and
metrics means to summarize results ensures that the 0.0060). In particular, the cases had LOS that were

Table 3
Distribution of Visit Characteristics and p-values for Detecting Differences Between Controls and Cases

Characteristic Controls Cases p-value Comparing Controls With Cases


Age (yr), mean (SD) 43.4 (16.9) 43.1 (16.9) 0.1631
Sex
Male 9,069 (44.9) 11,146 (49.4) 0.93
Female 9,256 (55.1) 11,397 (50.6)
Race
Asian 1,099 (5.4) 1,130 (5.5) 0.98
Black or African American 4,895 (24.2) 4,946 (24.0)
Multiracial 30 (0.2) 29 (0.14)
White non-Hispanic 8,987 (44.5) 9,204 (44.6)
Unknown/unavailable 5,204 (25.7) 5,344 (25.9)
ESI score (missing n = 51)
1 8 (0.04) 20 (0.1) <0.0001
2 1,912 (9.5) 1,671 (8.1)
3 9,373 (46.4) 8,978 (43.6)
4 7,544 (37.3) 8,311 (40.3)
5 1,364 (6.8) 1,636 (7.9)

Data are reported as n (%) unless otherwise noted.


ESI = Emergency Severity Index.
1176 Arya et al. • DECREASING LOS WITH SPLIT ESI PATIENT FLOW

Table 4
Unadjusted Differences in LOS Between Controls and Cases by ESI and First Diagnosis Code

N Geometric Mean (95% CI) Linear Model Results


Diff % 95% CI for%
Variable Controls Cases Controls (Min) Cases (Min) (Min) p-value Difference Difference
Overall 20,215 20,653 155.3 (153.8 to 156.9) 146.2 (144.7 to 147.7) –9.1 <0.0001 –5.9 –7.2 to –4.5
ESI score*
1 8 20 361.7 (241.9 to 540.7) 295.2 (232.7 to 374.7) –66.5 0.40 –18.4 –48.8 to 30.3
2 1,912 1,671 275.7 (268.6 to 282.9) 286.9 (279.4 to 294.5) 11.2 0.033 4.1 0.3 to 8.0
3 9,373 8,978 210.8 (208.1 to 213.4) 204.4 (201.5 to 207.3) –6.4 0.0011 –3.0 –4.8 to –2
4 7,544 8,311 105.1 (103.5 to 106.7) 102.1 (100.6 to 103.6) –3.0 0.0060 –2.9 –4.8 to –0.8
5 1,364 1,636 73.6 (71.2 to 76.1) 70.6 (68.6 to 72.6) –3.0 0.056 –4.2 –8.2 to 0.1
Diagnosis†
Abdominal 2,484 2,464 262.0 (256.8 to 267.2) 228.2 (223.2 to 233.4) –33.8 <0.0001 –12.9 –15.3 to –10.3
pain
Back problem 1,137 1,015 112.6 (107.7 to 117.7) 109.7 (107.7 to 114.9) –2.9 0.42 –2.5 –8.5 to 3.8
Chest pain 959 974 249.8 (240.3 to 259.7) 247.7 (237.4 to 258.6) –2.1 0.77 –0.8 –6.3 to 5.0
Headache/ 741 744 180.1 (171.0 to 189.7) 148.0 (140.8 to 155.5) –32.1 <0.0001 –17.8 –23.3 to –11.9
migraines
Open wound 653 704 106.4 (101.8 to 111.2) 107.8 (103.1 to 112.8) 1.4 0.67 1.4 –4.8 to 7.7
extremities
Other 611 647 143.6 (136.9 to 150.7) 139.1 (132.1 to 146.5) –4.5 0.38 –3.1 –9.8 to 4.0
connective
tissue
Other joint 510 553 110.5 (104.3 to 117.0) 104.1 (98.4 to 110.0) –6.5 0.13 –5.8 –12.9 to 1.8
disease
Other lower 692 755 166.6 (158.7 to 174.8) 160.8 (152.7 to 169.2) –5.7 0.32 –3.5 –9.9 to 3.4
respiratory
Other injury 558 589 137.7 (130.4 to 145.4) 137.4 (130.1 to 145.1) –0.3 0.96 –0.2 –7.6 to –7.8
Skin infection 533 469 101.8 (96.3 to 107.7) 92.7 (87.6 to 98.1) –9.1 0.022 –9.0 –16.0 to –1.4
Sprain 1,372 1,411 115.0 (111.6 to 118.5) 102.5 (99.6 to 105.5) –12.5 <0.0001 –10.8 –14.5 to –7.1
Superficial 1,072 1,161 124.9 (120.4 to 129.5) 113.9 (110.0 to 118.1) –11.0 0.0004 –8.7 –13.2 to –4.0
injury

ESI = Emergency Severity Index; LOS = length of stay.


*The difference between cases and controls differed significantly across ESI (p = 0.0068).
†The difference between cases and controls differed significantly across diagnosis code (p < 0.0001; analyses included data only
with 12 most frequent diagnosis codes).

estimated to be 3.0 and 2.9% smaller than those for con- significant at the 0.025 significance level, the level that
trols, respectively. The difference in LOS for patients would be required to maintain a familywise error rate
with ESI values of 5 was similar to those with values of of 0.05 using the Bonferroni correction for the effect
3 or 4, but not significantly. modifications.
Diagnosis was also found to significantly modify the
difference between cases and controls (p < 0.0001; DISCUSSION
Table 4). Among the top 12 most common diagnoses,
the largest effects were seen for the following visits: We found an 9% to 18% (9 to 34 minutes) reduction in
headache/migraine (%diff = –17.8%; p < 0.0001), LOS for five of the 12 most common complaints (the five
abdominal pain (%diff = -12.9%, p < 0.0001), skin infec- account for about one-third of all ED visits) and a 5.9%
tion (%diff = –9.0%, p = 0.022), sprains (%diff = –10.8%; reduction in LOS for all patients. It is important to note
p < 0.0001), and superficial injury (%diff = -8.7%, that this improvement occurred in the setting of a sig-
p = 0.0004). In other words, for visits in which the diag- nificant increase in volume of 3.94% (from 40,564 to
nosis was headache/migraine, the LOS for cases was 42,162) during the study period. The central premise of
estimated to be 17.8% less than for controls. Differences splitting ESI 3 visits into high and low variability is to
between cases and controls were not significant for the take advantage of the segmented flow process of a
remaining seven of the 12 most common diagnoses traditional fast track area. Segmenting patient flow pro-
(back problem; chest pain; open wound, extremity; cesses is beneficial when there are significant differ-
other, connective tissue; other, joint disease; other, ences in patient characteristics, number or types of
lower respiratory; and other injury). Figure 2 shows activities in the process, or provider demands that can
these estimated differences with their 95% CIs. The potentially result in enhanced throughput in one or both
diagnoses for which significant effects were found of the segmented streams. The low-variability ESI 3
accounted for 30.7 and 30.3% of all controls and cases, patients all followed a standardized work flow: assess-
respectively, while the diagnoses for which nonsignifi- ment, diagnostics, medications, and reassessment to
cant effects were found accounted for 25.3 and 25.4% of disposition. The higher-variability ESI 3 patients had
controls and cases. Note that both interactions were more complex initial presentations and required
ACADEMIC EMERGENCY MEDICINE • November 2013, Vol. 20, No. 11 • www.aemj.org 1177

% Difference (with 95% CI) of Cases Relative to Controls


for Top 12 Most Common Diagnoses
Headache/Migraines
Abdominal Pain
Sprain
Skin Infection
Superficial Injury
Other Joint Disease
Other Lower Respiratory
Other Connective Tissue
Back Problem
Chest Pain
Other Injury
Open Wound Extremities
-30 -20 -10 0 10
% Difference of Cases
Relative to Controls

Figure 2. Percent difference (with 95% CI) of cases relative to controls for top 12 most common diagnoses.

multiple reassessments throughout their stays by vari- Improved flow can affect not only the quantity but
ous staff members. also the quality of care. For example, Sun et al.21
The intake area provided a more efficient, fast track– recently found increased LOS, cost, and mortality in a
like setting for a stream of patients who are not tradi- claims-based review of patients admitted through Cali-
tionally considered appropriate for this type of process. fornia EDs during times of high crowding. Delays in
Our results show that there are other presenting patient antibiotic or pain medication administration and
streams that can benefit from an enhanced process increased errors associated with increased crowding
design that is similar to a fast track environment. These have also been previously reported.21–24 While this
patient streams have predictable, low-variation evalua- study did not focus on these outcomes, its presentation
tion and treatment pathways with low physician and of a possible means to adapt to overcrowding can be
nursing demands. These patients are relatively well and beneficial.25
unlikely to require admission. Staff and resources can There was not a significant change in ESI 1 or 2 dis-
be better aligned to the patient arrivals during specific charged LOS. While this result is complicated by the
time periods based on expected service times for the presence of many confounding variables, we believe
nurse, physician, and the treatment space (recliner that this is due to the very high demand of admitted
chair). The results waiting room allowed us to minimize patients, resulting in a nursing bottleneck that was not
the occupancy of treatment spaces by patients who are affected by the process changes. ESI 1 or 2 patients
clinically well and reduced the effect of laboratory and were typically treated in the emergent area of the ED,
radiology turnaround times and consultant response which has the highest number of admissions, the sickest
time on the capacity of ED treatment beds. The dis- patients, and little utility for vertical patient flow. Thus
charge area permitted the registration to occur in a the demand for treatment spaces in this area did not
nonclinical space, improving bed capacity in a similar change as it did in the urgent area of the main ED, and
manner. This conservation of bed capacity likely this may represent a location where crowding as
explains the improved LOS of low-variability level 3 and described by Pines25 is more a result of hospital crowd-
level 4 patients. Level 5 patients also had an improve- ing than ED processes.
ment, although the number was too small to be statisti- The LOS of ESI 5 patients was likely improved, but
cally significant. This study adds evidence to the the small sample size prevented us from demonstrating
effectiveness of “vertical patient flow” models described a statistically significant improvement. Also, our historic
in a recent survey of ED medical directors by Liu ESI 5 LOS was already quite short, about 1 hour, so to
et al.,20 which found that 29% had adopted these in see a statistically significant improvement may have
response to the American College of Emergency Physi- required drastic change. In fact, our team anticipated a
cians Task Force on Boarding. potential increase in LOS for ESI 5 patients as these
By placing lower-variability ESI 3 patients in intake, patients were segmented into our fast track in our his-
we decreased the total number of patients going to the torical process, and they would now “compete” with the
urgent area of the main ED. Decompressing the urgent low-variability ESI 3 patients for resources in intake. If
area of the main ED, where all level 3 patients were pre- our ESI 5 LOS had increased significantly, we would
viously evaluated, predictably resulted in improved LOS then have considered treating these patients in an opti-
of high-variability level 3 patients treated in this area in mized environment such as a “super track.”26
the new process. This result can likely be attributed to Flow for some diagnoses improved more compared
the reduced staff workload and decreased the use of to others. Throughput times for headache/migraine,
treatment spaces in this area in the improved process. abdominal pain, and sprains were statistically improved
1178 Arya et al. • DECREASING LOS WITH SPLIT ESI PATIENT FLOW

compared to those of open wounds, chest pain, and and regional supply management plan was created. We
back pain. Intake facilitated a standardization of treat- believe that this may have improved the process for
ment and augmented workflow for some conditions that patients going to the main ED, but do not believe this
was not available before. For instance, abdominal pain had a meaningful effect on those patients managed in
and sprain evaluations typically involve treatment and/ intake. Given the retrospective nature of this study, we
or imaging. In intake, the whole team knew the general acknowledge there may be other unmeasured changes
approach even if not the physician’s specific prefer- that influenced the results. However, we do not believe
ences. that there were any other significant process or policy
For example, during an abdominal pain or kidney changes during the study period. Staff were not blinded
stone evaluation, the physician would see the patient, during the study; however, this effect may be minimal
the intake nurse would place the IV, and the technician due to the retrospective nature of data collection and
would transport the patient to imaging on the way to the fact that there we were not anticipating publishing
the treatment area for IV fluids or medications that the our work. Finally, the discharge times were generated
physician by then would have ordered. The patient when each patient was removed from the EMR, and
moved to the various processes that, in essence, were that was done by the staff soon after the patient left the
ready and waiting for him or her, rather than the vari- ED. This could have artificially increased the patient
ous processes finding the patient and completing serial LOS. This part of the process was not changed between
assessments prior to task completion. We saw improve- the 2 years, and as such we do not believe that it
ments for order-to-in-lab times for phlebotomy, and affected the results.
improvement in order-to-radiology completion for plain
films, as the staff were ready and assigned to those par- CONCLUSIONS
ticular tasks. The intake nurse only performed initial
assessments, gave medications, and gathered diagnos- The use of a split Emergency Severity Index level 3
tics. The same nurse, when working in the main ED, triage flow system reduced length of stay for five of the
would be required to perform the entire nursing task 12 most common ED complaints by 9% to 18% without
on the patient, e.g., one patient may require medications increasing length of stay for the most or least ill, despite
while the other waits for phlebotomy. The approach to a 3.9% increase in overall ED patient volume. Separat-
these patients also changed in terms of streamlined ing high- and low-variability ESI 3 visits improved
treatment. Traditionally, most patients with abdominal throughput and reduced length of stay in our ED. This
pain received IV fluids. However, many of these patients technique is recommended for EDs experiencing exces-
can take fluids by mouth so they do not need IV access, sive length of stay for middle- and low-acuity patients
another opportunity for improved turnaround time. or those EDs that have bed capacity constraints and
LOS for diagnoses like open wounds, chest pain, and have sufficient volumes to warrant segmentation.
back pain improved less, perhaps because they required
more lengthy treatment procedures such as suturing References
that could not be overcome by improved transport flow
to imaging. Also, while their evaluations may have been 1. American Hospital Association. Chartbook: Trends
relatively standard, treatment modalities used may have Affecting Hospitals and Health Systems. Table 3.3
had much greater variability so that staff could less suc- and Chart 3.7. Available at: http://www.aha.org/
cessfully anticipate providers’ plans. research/reports/tw/chartbook/ch3.shtml. Accessed
Further study is needed to assess the relationship Sep 6, 2013.
between a resource use–based triage system (ESI) and 2. Handel D, Epstein S, Khare R, et al. Intervention to
diagnoses. While controversial, clinicians may consider improve the timeliness of emergency care. Acad
generating consensus for evaluation and treatment of Emerg Med. 2011; 18:1295–302.
the most common diagnoses to reduce variability. 3. Ward MJ, Farley J, Khare RK, et al. Achieving
efficiency in crowded emergency departments: a
LIMITATIONS research agenda. Acad Emerg Med. 2011; 18:1303–
12.
We had 10 pilot days over a 3-month period to adjust 4. American College of Emergency Physicians. Policy
and hardwire our process, but the group felt that we statement. Triage scale standardization. Ann Emerg
may have still had a learning curve for both nurses and Med. 2004; 43:154.
providers that may have blunted the early results. There 5. Mchugh M, Tanabe P, McClelland M, Kahre RK.
was an initial resistance to placing ESI 3 patients into a More patients are triaged using the emergency
physical area that had formerly been a fast track. severity index than any other triage acuity system in
Despite the rollout of the new paradigm, there was dis- the United States. Acad Emerg Med. 2012; 19:106–9.
cussion of specific policy stating the ESI 3 patients could 6. Wuerz RC, Milne LW, Eitel DR, Travers D, Gilboy
go to this area, and many patients who should have N. Reliability and validity of a new five-level triage
been ESI 3 were given acuities of ESI 4 until the policy instrument. Acad Emerg Med. 2000; 7:236–42.
was changed (Table 3). This may have resulted in a sig- 7. Harris M, Wood J. Resuscitate ED metrics with
nificant difference in the ESI distributions between the split-flow design. Healthc Financ Manage. 2012;
study year and the control year. Prior to, and in con- 66:76–9.
junction with, this implementation the nurse and physi- 8. Considine J, Kropman M, Kelly E, Winter C. Effect
cian staffing were aligned and an improved inventory of emergency department fast track on emergency
ACADEMIC EMERGENCY MEDICINE • November 2013, Vol. 20, No. 11 • www.aemj.org 1179

department length of stay: a case-control study. the second performance measures and benchmark-
Emerg Med J. 2008; 25:815–19. ing summit. Acad Emerg Med. 2011; 18:539–44.
9. Sanchez M, Smally AJ, Grant RJ, Jacobs LM. Effects 18. Langley R. The Improvement Guide. Hoboken, NJ:
of a fast-track area on emergency department per- Wiley, 2009.
formance. J Emerg Med. 2006; 31:117–20. 19. Agency for Healthcare Research and Quality.
10. Soremekun AO, Terwiesch C, Pines JM. Emergency Healthcare Cost and Utilization Project Tools and
medicine: an operations management view. Acad Software. Available at: http://www.hcup-us.ahrq.
Emerg Med. 2011; 18:1262–8. gov/tools_software.jsp. Accessed Sep 6, 2013.
11. Institute of Medicine. Hospital-based Emergency 20. Liu SW, Hamedani AG, Brown DF, Asplin B, Cam-
Care: At the Breaking Point. Washington, DC: argo CA Jr. Established and novel initiatives to
National Academies Press, 2006. reduce crowding in emergency departments. West J
12. Singer AJ, Thode HC Jr, Viccellio P, Pines JM. The Emerg Med. 2013; 14:85–9.
association between length of emergency depart- 21. Sun BC, Hsia RY, Weiss RE, et al. Effect of emer-
ment boarding and mortality. Acad Emerg Med. gency department crowding on outcomes of admit-
2011; 18:1324–9. ted patients. Ann Emerg Med. 2013; 6:605–11.
13. McCarthy ML, Zeger SL, Ding R, et al. Crowding 22. Pines JM, Localio AR, Hollander JE, et al. The
delays treatment and lengthens emergency depart- impact of emergency department crowding
ment length of stay, even among high acuity measures on time to antibiotics for patients with
patients. Ann Emerg Med. 2009; 54:492–503. community-acquired pneumonia. Ann Emerg Med.
14. Pines JM, Luyer S, Disbot M, Hollander JE, Shofer 2007; 50:510–6.
FS, Datner EM. The effect of emergency department 23. Pines JM, Hollander JE. Emergency department
crowding on patient satisfaction for admitted crowding is associated with poor care for patients
patients. Acad Emerg Med. 2008; 15:825–31. with severe pain. Ann Emerg Med. 2008; 51:1–5.
15. Pines JM, Pollack CV Jr, Diercks DB, Chang AM, 24. Kulstad EB, Sikka R, Sweis RT, Kelley KM, Rzechula
Shofer FS, Hollander JE. The association between KH. ED overcrowding is associated with an
emergency department crowding and adverse car- increased frequency of medication errors. Am J
diovascular outcomes in patients with chest pain. Emerg Med. 2010; 28:304–9.
Acad Emerg Med. 2009; 16:617–25. 25. Pines JM. Emergency department crowding in Cali-
16. Bernstein SL, Aronsky D, Duseja R, et al. The effect fornia, a silent killer? Ann Emerg Med. 2013;
of emergency department crowding on clinically 61:612–4.
oriented outcomes. Acad Emerg Med. 2009; 16:1–10. 26. Crane J, Noon C. The Definitive Guide to Emer-
17. Welch SJ, Stone-Griffith S, Asplin B, et al. Emer- gency Department Operational Improvement. Boca
gency department operations dictionary: results of Raton, FL: CRC Press, 2011.

You might also like