ORIGINAL CONTRIBUTION

The Effect of an Emergency Department

Dedicated Midtrack Area on Patient Flow
Olanrewaju A. Soremekun, MD, MBA, Frances S. Shofer, PhD, David Grasso, MD,
Angela M. Mills, MD, Jessica Moore, RN, and Elizabeth M. Datner, MD

Abstract
Background: Emergency department (ED) crowding negatively affects quality of care and
disproportionately affects medium-acuity (Emergency Severity Index [ESI] level 3) patients. The effect of a
dedicated area in the ED focused on these patients has not been well studied.
Objectives: The objective was to find out the operational effect of a midtrack area dedicated to the
evaluation and safe disposition of uncomplicated medium-acuity (ESI 3) patients.
Methods: This was a 24-month pre-/postintervention study to evaluate the effect of implementation of a
dedicated midtrack area at an urban tertiary academic adult ED. The midtrack had three examination
rooms and three hallway stretchers for ongoing treatment staffed by an attending physician and two
registered nurses (RNs). Besides the two additional RNs representing a 3.4% increase in total daily
nursing hours, the intervention required no additional ED resources. The midtrack area was open from 1
p.m. to 9 p.m. on weekdays, corresponding to peak ED arrival rates. All patients presenting during
weekdays were included, excluding patients triaged directly to the trauma bay or psychiatric unit or who
expired in the ED. The main outcomes were left without being seen (LWBS) rates and ED length of stay
(LOS), adjusting for patient volume, daily total patient hours (a proxy for ED crowding), and acuity.
Results: A total of 91,903 patients were included for analysis during the study period including 261 pre-
and 256 postintervention days. Comparing the pre- and postintervention periods, mean ED daily visits
(173 vs. 182) and mean total daily patient hours (889 vs. 942) were all significantly higher in the
postintervention period (p < 0.0001). There was no significant change in percentage of patients with high
triage acuity levels. Despite this increase in volume and crowding, the unadjusted and adjusted LWBS
rates decreased from 6.85% to 4.46% (p < 0.0001) and from 7.33% to 3.97% (p < 0.0001), respectively. The
mean LOS for medium-acuity patients also decreased by 39.2 minutes (p < 0.0001). For high-acuity
patients, there was no significant change in the mean time to room (14.69 minutes vs. 15.21 minutes,
p = 0.07); however, their mean LOS increased by 24 minutes (331 minutes vs. 355 minutes, p < 0.0001).
Conclusions: Implementation of a midtrack area dedicated to caring for uncomplicated medium-acuity
(ESI 3) patients was associated with a decrease in overall ED LWBS rates and ED LOS for medium-
acuity patients.
ACADEMIC EMERGENCY MEDICINE 2014;21:434–439 © 2014 by the Society for Academic Emergency
Medicine

E
mergency department (ED) crowding has been and Medicaid Services (CMS) hospital reporting guide-
steadily increasing over time and has been lines.5 Various interventions have been studied to allevi-
shown to negatively affect several domains of ate ED crowding. ED-level interventions such as
quality such as timeliness, safety, and patient-centered- bedside registration, immediate bedding, and placement
ness.1–4 Given the high prevalence and negative effect of a physician in triage have been successful in improv-
on quality, ED crowding has been recognized as an ing patient flow.2,6–8 Other interventions such as dedi-
important public health issue, leading to the inclusion of cated fast-track areas, usually staffed by midlevel
ED crowding metrics in the 2013 Centers for Medicare providers, have also been shown to decrease wait times

From the Department of Emergency Medicine, University of Pennsylvania, Philadelphia, PA.

Received July 21, 2013; revisions received October 7 and November 4, 2013; accepted November 7, 2013.
Presented at the American College of Emergency Physicians Scientific Assembly, Denver, CO, October 2012.
The authors have no relevant financial information or potential conflicts of interest to disclose.
Supervising Editor: Richard T. Griffey, MD, MPH.
Address for correspondence and reprints: Olan A. Soremekun, MD, MBA; e-mail: olanmd@gmail.com.

ISSN 1069-6563 © 2014 by the Society for Academic Emergency Medicine

434 PII ISSN 1069-6563583 doi: 10.1111/acem.12345
ACADEMIC EMERGENCY MEDICINE • April 2014, Vol. 21, No. 4 • www.aemj.org 435

and length of stay (LOS) for the lowest-acuity ED studying the effect of an intervention on ED flow, the
patients.9,10 following groups of patients were excluded from analy-
Given the high prevalence of dedicated fast-track sis: patients triaged directly to the PEEC or trauma bay,
areas and the ability of EDs to create treatment space patients transferred from the ED to other institutions,
for the highest-acuity patient even during high levels of and patients who died or were dead on arrival in the
crowding, the groups of patients who have the longest ED. We excluded these patients either because they
wait times are the medium-acuity Emergency Severity received their care in other areas separate from the ED
Index (ESI) level 3 patients.11,12 Based on acuity or with their own dedicated resources or because their
resources required, these patients cannot be cared for LOS was dependent on the transfer mechanisms to
in most fast-track areas and are often placed lower in other institutions or to the morgue. We also excluded
priority for regular ED beds. patients who presented during a brief 4-day pilot of the
These medium-acuity patients may vary widely in intervention prior to full implementation. Although
terms of resources needed, complexity, and need for ED the hours of the midtrack intervention were from 1 p.m.
beds. A subset of these patients may need relatively few to 9 p.m., we included all patients because the benefits
resources. A dedicated area of the ED focused on the of the intervention may carry over to all patients arriv-
identification and disposition of this subset of medium- ing later in the day, thus affecting overall patient flow.
acuity patients may decrease the effect of ED crowding
on this group of patients and improve overall patient Study Protocol
flow. A dedicated midtrack area was fully implemented in
In this investigation, our primary objective was to January 2012 after the pilot period in December 2011.
study the effect of a dedicated “midtrack” area on key The area included three dedicated examination rooms
CMS ED patient flow metrics, ED LOS, and overall left and three hallway spaces. The fast-track area previously
without being seen (LWBS) rates. Secondarily, we esti- used this space, thus the implementation of the mid-
mated the effect of this intervention on higher- and track area reduced the dedicated fast-track area with
lower-acuity patients. Our key hypothesis was that a no change in the overall number of treatment spaces in
dedicated midtrack area will lead to a reduction in over- the study center. An attending physician who was reas-
all ED LOS and LWBS rates. signed from the main ED staffed the midtrack area so
the total number of attending hours per day was also
METHODS unchanged. The midtrack area, however, was staffed
with two additional registered nurses (RNs) for an addi-
Study Design tional 16 hours or a 3.4% increase in total nursing clini-
This was a retrospective 24-month pre- and postinter- cal hours per day. The hours of operation were
vention study of a dedicated midtrack area in the ED. weekdays from 1 p.m. to 9 p.m., corresponding to
The institutional review board at the study hospital historic peak patient arrival rates and waiting room
approved the study protocol. census. The midtrack physician identified eligible
patients from the waiting room by reviewing the RN
Study Setting and Population triage note, prior records, and overall triage evaluation.
The study period ran from January 1, 2010, to Decem- All nurses who perform triage at the study center have
ber 30, 2012. The study was performed at a tertiary care a minimum level of ED experience and undergo formal
adult urban ED with approximately 68,000 annual visits, ESI training and triage orientation. There was no
with a 4-year emergency medicine (EM) residency pro- change in the RN triage process postimplementation of
gram, a dedicated fast-track area staffed by midlevel midtrack. Physicians were instructed to identify patients
providers, and separate psychiatric emergency evalua- who they deemed to be uncomplicated medium-acuity,
tion center (PEEC), as well as trauma resuscitation bays. ESI level 3 patients with a high likelihood of discharge
The allocation of physician and nursing resources in the home.
ED during the entire study period was based on geo- The study center uses a computerized ED information
graphic assignment with nurses covering contiguous system (EDIS) with integrated charting, order entry,
rooms that align with the same physician care team and and patient tracking (EMTrac; University of Pennsylva-
goal patient-to-nurse ratio of four to one. The PEEC is a nia, Philadelphia, PA). This system allows, via a Micro-
separate area from the ED, with a dedicated waiting soft Access (Microsoft Corporation, Redmond, WA)
area and treatment bays that are staffed by psychiatry database query, collection of patient demographic infor-
attendings and residents as well as a dedicated group of mation, patient flow time stamps, disposition, and pre-
nurses. Patients who arrive to the ED with primary psy- cise reconstruction of state of the ED at the time of each
chiatric complaints are transferred directly to the PEEC individual patient arrival.
after initial triage. The trauma bays, while contiguous The primary operational outcomes measured were
with the ED, are staffed by a trauma surgery team and daily LWBS rates and patient ED LOS. LOS was defined
nurses that are separate from the ED. While the ED as the time from intake (first entry in the EDIS system)
senior resident and ED attending assist with trauma air- to the time of patient leaves the ED (removed from the
ways, the trauma bay does not use any of the ED EDIS system).
resources. No other major interventions occurred dur-
ing the study period. Data Analysis
All patients who presented to the study center on To determine differences in daily LWBS rates pre- and
weekdays were eligible for inclusion. Because we were postintervention, analysis of variance (ANOVA),
436 Soremekun et al. • EFFECT OF MIDTRACK ON PATIENT FLOW

adjusted for high triage acuity level (daily percentage of cared for in the midtrack area was 13 (IQR = 10 to 16;
patients with ESI levels 1 or 2), and daily total patient see Table 2 for midtrack patient demographics and ED
volume was performed. We examined the effect of the disposition).
midtrack area on the percentage of days with high The overall ED unadjusted LWBS rate decreased
LWBS rates (>5%) using Fisher’s exact test. This analy- from 6.9% to 4.5% (p < 0.0001) after implementation of
sis was stratified by normal and high volume days (>190 the midtrack area. Adjusting for ED volume, crowding,
patients per day). The thresholds for high LWBS rate or and acuity, the LWBS rates decreased from 7.3% to
high volume day was determined by ED leadership 4.0% (p < 0.0001; Figure 1). The decrease in LWBS rates
based on sensitivity analysis on historic data and our primarily affected medium-acuity patients, with the
goal LWBS rate of less than 2%. Data are presented as rates decreasing from 9.5% to 5.3% (p < 0.0001) in this
percent change with 95% confidence intervals (CIs). To subset of patients. The LWBS rates for high- and low-
assess differences in patient LOS pre-/postmidtrack, acuity patients decreased from 1.7% to 1.3%
ANOVA on the log-transformed LOS adjusting for total (p < 0.0001) and 4.37% to 4.30% (p < 0.0001), respec-
patient hours (a proxy for ED crowding), ED volume, tively. In addition to a decrease in the overall LWBS
and patient acuity level was performed. Because LOS rates, the percentage of days with high LWBS rates (>
was not normally distributed, a log transformation was 5%) on normal-volume and high-volume days decreased
performed prior to analysis. All statistical analyses were significantly (Figure 2).
performed using SAS statistical software (Version 9.3, Finally, the mean ED LOS for medium-acuity (ESI 3)
SAS Institute, Cary NC). The Tukey-Kramer method patients decreased by 36 minutes (343 minutes vs.
was used to adjust for multiple comparisons within the 307 minutes, p < 0.0001). There was no change in the
ANOVAs. A probability of p < 0.05 was considered sta- LOS for low-acuity patients despite the midtrack using
tistically significant. the fast-track rooms that were previously dedicated to
the care of these patients. The LOS of high-acuity (ESI 1
RESULTS and 2) patients increased by 24 minutes (331 minutes vs.
355 minutes, p < 0.0001); however, there was no signifi-
A total of 133,094 patients presented to the ED during cant change (14.69 minutes vs. 15.21 minutes, p = 0.07)
the 24-month study period. Those meeting exclusion cri- in the time to room for these patients (Figure 3).
teria were as follows: 1) 35,032 patients who presented
to the ED on a weekend, 2) 2,186 patients triaged
directly to PEEC, 3) 2,680 patients triaged directly to the
trauma bay, 4) 450 patients transferred to another medi- Table 2
Demographics of Patients Seen in Midtrack Area
cal facility, and 5) 72 patients who were dead on arrival
or expired in the ED. Additionally, 771 patients who
presented to the ED during the December pilot phase of Characteristic Percentage
midtrack were also excluded. The final sample included Age (yr), mean (SD) 32.1 (12.7)
91,903 patients, seen on 261 days pre- and 256 days % women 75
postintervention. Race (%)
White 21.3
After implementation of the midtrack, there was no Black 70.9
clinically meaningful change in proportion of patients Other 7.8
assigned high, medium, or low triage acuity, nor a Disposition (%)
change in the proportion of admitted patients (Table 1). Admitted 3.5
However, as shown in Table 1, there was an increase in Placed in observation 2.9
Treated & discharged 93.6
the mean daily ED volume (and number of days ED vol-
ume exceeded 190 patients) postimplementation of the N = 3,489 patients.
midtrack area. The median daily number of patients

Table 1
Daily ED Population Characteristics

Pre Midtrack Post Midtrack

Characteristic (n = 261 days) (n = 256 days) p-value
Triage acuity level—mean % of patients/day classified as:
High, ESI 1 and 2 23.4 (22.9–23.9) 24.1 (23.6–24.6) 0.052
Medium, ESI 3 44.7 (44.1–45.2) 45.3 (44.7–45.8) 0.14
Low, ESI 4 and 5 31.9 (31.5–32.4) 30.6 (30.2–31.1) < 0.0001
Daily % of patients admitted to observation or hospital 26.7 (26.3–27.1) 24.7 (24.3–25.1) < 0.0001
Mean daily patient hours 889 (872–906) 942 (923–962) < 0.0001
Mean daily ED volume 173 (171–175) 182 (180–185) < 0.0001
Number of days ED volume >190 (%) 33 (12.6%) 79 (30.9%) < 0.0001

Number in parentheses is 95% CI.

ESI = Emergency Severity Index.
ACADEMIC EMERGENCY MEDICINE • April 2014, Vol. 21, No. 4 • www.aemj.org 437

12
11 Mid-Track
implemented
10
9

Percent LWBS 8
7
6
5
4
3
2
1
0

2011 2012

Figure 1. Monthly LWBS rates adjusted for volume, acuity, and crowding. LWBS = left without being seen.

100 60

90 50
88%
40
80
Percent Days LWBS rate >5%

30
Change in LOS (minutes)

70 24
Post-Pre Mid-Track

20
60 61%
10
54%
50 0 0

40 -10

-20
30
-30
20 20% -36
-40
10 Normal Volume Days High Volume Days -50
(Daily Census <190 patients) (Daily Census ≥ 190 patients)
-60
Total Days 228 177 33 79 1-2 3 4-5

Control (Pre) Mid-Track (Post) ESI Classification

Figure 3. Postimplementation minus preimplementation

Figure 2. Percentage of days LWBS rates exceeded 5% during change in LOS (minutes) adjusted for volume and crowding.
pre- and postimplementation periods. LWBS = left without ESI = Emergency Severity Index; LOS = length of stay.
being seen.

DISCUSSION ED, the benefits of reducing service time variability by

cohorting patients with similar needs can outweigh the
Emergency department crowding negatively affects sev- reduction in flexibility, especially when the volume of
eral of the Institute of Medicine quality domains, and these patients is high.13 This finding explains the suc-
EDs with high levels of crowding face the challenge of cess of dedicated fast-track areas as well as our dedi-
reducing their wait times and LWBS rates. Similar to cated midtrack area in improving patient flow. At our
other EDs, our medium-acuity patients have the longest study center, having a high volume of younger ESI 3
observed wait times and are more likely to LWBS.11,12 patients who require fewer resources (shorter service
Counterintuitive to the goals of triage, dedicated fast- times) relative to older high-acuity patients was a key
track areas providing care for the lowest-acuity patients factor that allowed us to gain significant operational
result in shorter wait times for these patients when benefits in creating a cohort of these patients. In addi-
compared to medium-acuity patients. To improve over- tion, given the heterogeneity of ESI 3 patients, to ensure
all LWBS rates, targeted interventions are needed to large enough volumes to justify the reduction in flexibil-
reduce wait times and improve overall patient flow of ity, we assigned responsibility for identification of mid-
medium-acuity patients. track patients to the midtrack attending physician.
While dedicating an area of the ED to a care of cer- These physicians have different risk profiles and skill
tain type of patients may reduce overall flexibility of the sets and are ultimately responsible for determining the
438 Soremekun et al. • EFFECT OF MIDTRACK ON PATIENT FLOW

medical complexity and resource requirements of their high-volume day was determined by ED management
patients. These operational improvements occurred in based on sensitivity analysis on historic data, where
spite of an increase in overall patient volume and elimination of days when LWBS rates was greater than
inpatient boarding in the postimplementation period. 5% would lead our overall LWBS rate to be less than
In addition to the operational benefits of cohorting the national benchmark LWBS rate of 2%. While most
patients with similar service times, our implementation operational papers report aggregate numbers, most
of a midtrack area using only three examination rooms operational managers recognize that there are days
demonstrates a significant concept discussed often in when ED operations are working very well, and then
the ED operations literature. While an attending physi- there are “out-of-control” days usually driven by high
cian and two nurses covering three examination rooms census (facility specific) and boarding. Analyzing data
may seem overresourced, using fewer rooms ensures from this perspective allows for better understanding of
that patients are kept “vertical” during their stays in the the frequency of “out-of-control” days and better target-
ED and a high turnover of rooms. Using the examina- ing of operational interventions. For example, if the
tion rooms for initial evaluations and then transferring data show the aggregate LWBS rate at a study center
patients to hallway stretchers or back to the waiting to be 4%, but this rate is driven by Mondays where the
room ensured rapid turnover of rooms. This practice is LWBS rate averages 10%, then interventions should be
usually not the case in the main section of our ED, designed for Mondays and not for the other days of the
where as is common in many EDs, patients usually week.
occupy rooms for their entire stays. While the concept Although there are limitations in our particular study,
of keeping patients “vertical” during their stays in the the fact that patient volume and acuity increased after
ED has been described in the ED design literature, mid- midtrack was implemented would be expected to bias
track and physician in triage with an internal waiting the study toward the null. However, we found a clinical
area are two interventions that leverage this idea to and statistical difference in both LWBS rates and LOS
improve the utilization of ED beds and improve patient for all patients, not just those seen in midtrack. There-
flow.14 Recognizing that the availability of ED beds is a fore, although our study design is not perfect, we
key bottleneck, especially in EDs with high inpatient believe that the positive results in the context of bias
boarding rates and limited space to expand, redesigning towards the null and minimal increase in resources
EDs such that patients use ED bays only when clinically suggests utility for this approach.
appropriate will be critical to maximizing patient flow in
crowded EDs, as we demonstrate with our midtrack LIMITATIONS
intervention.
We were able to implement our midtrack area using The retrospective study design, lack of randomization,
ED rooms as well as physician staffing reallocated from and single intervention site with dedicated trauma bays
the fast-track area and main ED, respectively. By and a PEEC may affect the generalizability of our con-
performing analysis to identify spare capacity within clusions and our ability to fully exclude all potential con-
our current system, we were able to operationalize this founding factors and secular trends over the 24-month
intervention and reduce overall LWBS with minimal study period. Although a randomized controlled trial of
incremental investments. While reallocation of two different ED design strategies (midtrack vs. no mid-
resources maximizes our financial return of the inter- track) might be scientifically more valid, these types of
vention, we estimate that the financial return would still studies are not likely to ever happen in the “real-world”
have been positive if it required adding additional phy- setting where we try to improve clinical operations.17
sician coverage. Reallocation of physician resources There was a 3.4% increase in the daily total nursing
from the main ED and siphoning off easy ESI 3 patients clinical hours that was required to staff midtrack. While
to the midtrack area (and thus increasing complexity of this increase is minimal compared to other operational
patients in the main ED) may have contributed to the interventions, our study design limits our ability to attri-
increase in the LOS of high acuity (ESI 1 and 2) bute operational improvements to increase in resources
patients, because we did not see an increase in the time versus the midtrack intervention and redesign in the
to room of these patients. We did have to increase our care process. An additional limitation is our failure to
total nursing hours by 16 hours per day, representing a track any countermeasures that may be associated with
3.4% increase. We believe our incremental resources to improving patient flow, such as 72-hour returns of dis-
be minimal compared to other operational interven- charged patients or upgrades in care of admitted
tions, such as physician in triage as described by Rogg patients. Finally, the selection of the patients cared for
et al.15 Their intervention involved redesign of the front in the midtrack area was not standardized and was at
end to add four additional examination rooms and was the discretion of the attending physician.
staffed by 16 hours of additional attending physician
coverage as well as significant increase in nurs- CONCLUSIONS
ing hours. Interventions published by Holroyd et al.16
and Han et al.6 also involved an additional 8 to 9 hours Implementation of a midtrack area dedicated to the
of attending coverage per day and do not report on evaluation of uncomplicated medium-acuity patients
nursing hours. was associated with reduction in overall left without
Finally, in reporting our results, we presented data on being seen rates and ED length of stay for medium-acu-
number of days when the LWBS rate was beyond a ity patients. While our study describes the operational
certain threshold. The threshold for high LWBS rate or effect, further study is needed to evaluate the effect of
ACADEMIC EMERGENCY MEDICINE • April 2014, Vol. 21, No. 4 • www.aemj.org 439

this intervention on other quality measures, patient sat- 9. Ginde AA, Espinola JA, Sullivan AF, Blum FC, Cam-
isfaction, and resident education. argo CA Jr. Use of midlevel providers in US EDs,
1993 to 2005: implications for the workforce. Am J
Emerg Med 2013;28:90–4.
References
10. Sanchez M, Smally AJ, Grant RJ, Jacobs LM. Effects
1. Bernstein SL, Aronsky D, Duseja R, et al. The effect of a fast-track area on emergency department per-
of emergency department crowding on clinically formance. J Emerg Med 2006;1:117–20.
oriented outcomes. Acad Emerg Med 2009;16:1–10. 11. McCarthy ML, Zeger SL, Ding R, et al. Crowding
2. Moskop JC, Sklar DP, Geiderman JM, Schears RM, delays treatment and lengthens emergency depart-
Bookman KJ. Emergency department crowding, ment length of stay, even among high-acuity
part 1–concept, causes, and moral consequences. patients. Ann Emerg Med 2009;54:492–503.
Ann Emerg Med 2009;53:605–11. 12. Bullard MJ, Villa-Roel C, Bond K, Vester M, Holroyd
3. Pines JM, Iyer S, Disbot M, Hollander JE, Shofer BR, Rowe BH. Tracking emergency department
FS, Datner EM. The effect of emergency department overcrowding in a tertiary care academic institution.
crowding on patient satisfaction for admitted Healthcare Q (Toronto, Ont.) 2009;12:99–106.
patients. Acad Emerg Med 2008;15:825–31. 13. Soremekun OA, Terwiesch C, Pines JM. Emergency
4. Hing E, Bhuiya F. Wait time for treatment in hospi- medicine: an operations management view. Acad
tal emergency departments: 2009. NCHS Data Brief Emerg Med 2011;18:1262–8.
2012;102:1–8. 14. Liu SW, Hamedani AG, Brown DF, Asplin B, Cam-
5. Institute of Medicine. Hospital-based Emergency argo CA Jr. Established and novel initiatives to
Care: At the Breaking Point. Washington, DC: reduce crowding in emergency departments. West J
National Academies Press, 2007. Emerg Med 2013;14:85–9.
6. Han JH, France DJ, Levin SR, Jones ID, Storrow 15. Rogg JG, White BA, Biddinger BD, Chang Y,
AB, Aronsky D. The effect of physician triage on Brown DF. A long-term analysis of physician triage
emergency department length of stay. J Emerg Med screening in the emergency department. Acad
2010;39:227–33. Emerg Med 2013;20:374–80.
7. Wiler JL, Gentle C, Halfpenny JM, et al. Optimizing 16. Holroyd BR, Bullard MJ, Latoszek K, et al. Impact of
emergency department front-end operations. Ann a triage liaison physician on emergency department
Emerg Med 2009;55:142–60. overcrowding and throughput: a randomized con-
8. Soremekun OA, Biddinger PD, White BA, et al. trolled trial. Acad Emerg Med 2007;14:702–8.
Operational and financial impact of physician 17. Asplin BR, Yealy DM. Key requirements for a new
screening in the ED. Am J Emerg Med 2012;30: era of emergency department operations research.
532–9. Ann Emerg Med 2011;57:101–3.