Professional Documents
Culture Documents
Abstract
Objectives: This was a point-prevalence study designed to quantify the magnitude of emergency depart-
ment (ED) overcrowding and inpatient boarding. Every ED in Michigan was surveyed at a single point
in time on a Monday evening. Given the high patient volumes on Monday evenings, the effect on
inpatient boarding the next morning was also reviewed.
Methods: All 134 EDs within the state of Michigan were contacted and surveyed on Monday evening,
March 16, 2009, over a single hour and again the following morning. Questions included data on annual
census, bed number, number of admitted patients within the ED, ambulance diversion, and ED length of
stay.
Results: Data were obtained from 109 of the 134 (81%) hospitals on Monday evening and 99 (74%) on
Tuesday morning. There was no difference in annual visits or ED size between participating and nonpar-
ticipating EDs. Forty-seven percent of EDs were boarding inpatients on Monday evening, compared
with 30% on Tuesday morning. The mean estimated boarding times were 3.7 hours (Monday evening)
and 7.2 hours (Tuesday morning). Twenty-four percent of respondents met the definition of over-
crowded during sampling times. There was a significant relationship between inpatient boarding and
ED overcrowding (p < 0.001). Only three EDs were actively diverting ambulances.
Conclusions: In this study on a single Monday evening, 47% of EDs in Michigan were actively boarding
inpatients, while 24% were operating beyond capacity. On the following morning (Tuesday), EDs had
fewer boarded inpatients than on Monday evening. However, these boarded inpatients remained in the
ED for a significantly longer duration.
ACADEMIC EMERGENCY MEDICINE 2011; 18:1386–1391 ª 2011 by the Society for Academic
Emergency Medicine
E
mergency department (ED) overcrowding is a ity to provide timely care to these severely ill patients.
worsening phenomenon across the United In addition to an increase in ED resource use, boarding
States. As many as 96% of ED directors report admitted patients who are awaiting placement into
overcrowding as a problem, with 28% claiming that this inpatient beds is a significant factor contributing to
is a daily occurrence.1 As the safety net for critically ill overcrowding.2 The secondary effects of overcrowding
patients, the ED serves a vital role in health care deliv- include an increase in patient wait times, an increase in
ery and public safety. Overcrowding threatens the abil- patients leaving without being seen, a decline in overall
patient satisfaction survey scores, and an increase in
ambulance diversion.3,4 For critically ill patients, inpati-
From the Department of Emergency Medicine (BMF, EJR), ent boarding often delays transfers to the intensive
Michigan State University, East Lansing, MI; Emergency Medi- care unit and can result in higher mortality rates and
cine, Bristol Regional Medical Center (CNK), Bristol, TN; and increased lengths of stay.5
Emergency Medicine, Willis-Knighton Medical Center (GAL), In 2008, EDs in Michigan had 4.4 million patient
Shreveport, LA. visits.6 At 44.4 ED visits per 100 population, Michigan’s
Received March 16, 2011; revision received May 18, 2011; ED use was slightly greater than the national average.
accepted June 6, 2011. In 2007, there were 39.9 ED visits per 100 population in
The authors have no relevant financial information or potential the United States and 39.7 ED visits per 100 population
conflicts of interest to disclose. in Ontario, Canada.7 Nationally, the number of ED visits
Supervising Editor: Michael Mello, MD. increased from 90 million in 1992 to 108 million visits
Address for correspondence and reprints: Brent Felton, DO; by year 2000,8 finally reaching 124 million in 2008.9
e-mail: feltonbr@gmail.com.
determined. Significance for all results was defined as (p < 0.001). Only three hospitals were on ambulance
a < 0.05. diversion during the study period. There was no
demonstrable relationship between inpatient boarding
RESULTS and ambulance diversion (p = 0.10). There was,
however, a significant relationship between ED over-
All 134 Michigan EDs listed in the MCEP directory crowding and ambulance diversion (p = 0.02).
were successfully contacted Monday evening within the On Tuesday morning, all 134 Michigan EDs were
60-minute survey period. Of these, 109 (81.3%) chose to again contacted within the 60-minute survey period; 99
participate in the survey. Of the 25 EDs that did not (73.9%) chose to participate. This was a similar
participate, 15 cited that they were ‘‘too busy’’ (60%), response to Monday’s participation (p = 0.14). Of the 35
eight did not feel ‘‘comfortable’’ answering the survey EDs that did not participate, 15 cited that they were
questions (32%), one ED hung up on the surveyor (4%), ‘‘too busy’’ (43%), 11 simply refused to participate
and one call was routed to an answering machine (4%). (31%), and nine did not participate for other and
There were no statistical differences in volume or ED unknown reasons (26%). In comparing participating
size (bed number) between hospitals that participated and nonparticipating hospitals, both had similar mean
in the study and those that did not (Table 1). The mean ED sizes (bed numbers) of 21.0 (95% CI = 17.7 to 24.3)
numbers of ED beds were similar (p = 0.65) for partici- and 26.3 (95% CI = 18.3 to 34.3) beds, respectively
pating hospitals (22.0 beds; 95% CI = 18.3 to 25.7 beds) (p = 0.23; Table 1). Likewise, the mean annual censuses
and nonparticipating hospitals (23.9 beds; 95% were the same (p = 0.84) for participating hospitals
CI = 16.6 to 31.3 beds). Likewise the mean annual ED (29,916 visits; 95% CI = 25,083 to 34,749 visits) and non-
censuses were similar (p = 0.60) for participating participating hospitals (31,073 visits; 95% CI = 21,088 to
(29,580 visits; 95% CI = 24,835 to 34,325 visits) and non- 41,058 visits).
participating hospitals (33,000 visits; 95% CI = 21,190 to Among the 99 participating hospitals, 30 EDs reported
44,810 visits). boarding inpatients (30.3%). There was a statistically sig-
Of the 109 participating EDs, 26 (23.9%) reported nificant reduction in ED boarding between Monday eve-
overcrowding during the survey period, and 51 (46.8%) ning and Tuesday morning (p = 0.04). The mean boarded
were actively boarding inpatients. For EDs boarding time for patients on Tuesday morning was 7.2 hours
patients, 32 (62.7%) were due to a lack of inpatient bed (95% CI = 5.1 to 9.3 hours), which was significantly
availability, 16 (15.8%) were due to a delay in the admit- longer than for Monday evening (p = 0.004). The times
ting process, and the remaining 21.5% were for other for the longest boarded patient ranged from 30 minutes
and unknown reasons. The mean reported boarding to 48 hours, with a mean boarding time of 13.1 hours
time per patient was 3.7 hours (95% CI = 2.5 to (95% CI = 8.9 to 17.3 hours).
4.9 hours). The times for the longest boarded patient
ranged from 15 minutes to 33 hours, with a mean maxi- DISCUSSION
mum boarding time of 8.3 hours (95% CI = 5.6 to
11.0 hours). Of the 26 departments reporting over- Several studies have attempted to quantify the extent of
crowding, 12.4% of all patients treated in these depart- ED overcrowding and inpatient boarding. Schneider et
ments were being treated in hall beds with a mean al.10 used a point-prevalence model in a mailed survey
(±SD) of 6.4 (±6.14) hall bed patients per ED. Of the 51 to assess overcrowding in 250 randomly selected EDs.
EDs boarding inpatients, the mean number of boarded Similar to our study, a Monday in March was chosen
patients per ED was 5.5 (SD ± 6.25), with a median of for sampling as Monday is the busiest day of the week
3.0 patients per ED (range = 1–30; third quartile = 8.0 for most departments. Schneider’s survey obtained a
and first quartile = 1.00). When EDs were subdivided response rate of 36%.10 Of these respondents, 73% of
into interquartile ranges according to bed number, the EDs reported inpatient boarding of two or more
quartile of EDs with the largest bed numbers (32 to 89 patients and an ambulance diversion rate of 11%. Of all
beds) represented 65% of all overcrowded EDs in the ED patients during this ‘‘snapshot’’ in time, 22% were
state (Table 2). Furthermore, 43% of all EDs in the state admitted and waiting transfer to an inpatient bed.10
boarding inpatients were represented by this same In another mail-based survey that involved ED direc-
quartile. There was a statistically significant relationship tors in Florida, New York, and Texas, significant condi-
between ED overcrowding and ED inpatient boarding tions of overcrowding were reported.12 The goal of that
Table 1
Mean Census and Bed Number for Participating and Nonparticipating EDs
hospitals were compared, they did not differ signifi- inpatient boarding to 120 minutes for all admitted
cantly in annual census or bed number. As a result, the patients during this period, an additional 3,175 patient
data obtained from the participating EDs in this study visits could have been accommodated without changes
are likely to represent the conditions that most, if not in staffing or patient care areas. The resultant revenue
all, Michigan EDs are facing. There were 24% of Michi- loss of these encounters totaled $3.9 million.
gan EDs that met the definition of overcrowding during The consequences of ED overcrowding are severe.
this 1-hour window. This means that nearly one in four One study examining mortality as an end point demon-
EDs were operating beyond capacity. Furthermore, strated a relative risk of 1.34 for inpatient death when
47% of EDs during this period were actively boarding patients presented to EDs during periods of over-
inpatients, with a mean boarding time of 3.7 hours. crowding.23 Furthermore, overcrowding results in
Unlike previous studies examining ED overcrowding poorer performance in pneumonia quality care mea-
and inpatient boarding that examined a single snapshot sures, as well as longer inpatient lengths of stay.24,25
in time, additional data were collected on the following Despite these outcome measures, patients are not the
morning. This second survey provided a more complete only ones affected by inpatient boarding and ED over-
picture of the relationship between ED overcrowding crowding. The lack of inpatient bed availability for
and inpatient boarding over time. The data showed a these busy EDs is a major contributing factor to ED
greater percentage of EDs boarding inpatients on Mon- physician dissatisfaction.26
day evening than on Tuesday morning. This suggests
there is a mild relief in the ‘‘bottlenecking’’ of inpatient
boarding overnight. Furthermore, the ED boarding LIMITATIONS
time was significantly longer on Tuesday morning when
compared with Monday evening. This suggests there There are certain limitations to this study. A single
are fewer ED-boarded inpatients overall on Tuesday Monday evening in March was selected, similar to
morning, but these patients remain in the ED awaiting Schneider et al.10 The date was selected to avoid
transfer to inpatient beds for significantly longer peri- specific holidays and occurred in the absence of any
ods of time. These findings likely reflect the saturation specific illness epidemics. As best as the authors could
of available inpatient beds that occurred on Monday anticipate, this would have been a typical Monday
evening, and that patients boarded on Tuesday morning evening in the ED. Some census variability would exist
were waiting for inpatient beds to become available. among Monday evenings. Still, this date was selected to
The responses from the current study demonstrated minimize any anticipated unusual circumstances.
that ED boarding did not have a significant effect on Furthermore, the study design being a point-in-time
ambulance diversion. This conclusion is based on the survey has limitations. A primary limitation is the
low number of EDs on diversion during the limited restricted ability to assess outcome variability over
time window. The results of this point-in-time study time. Without repeated sampling of a population, the
might underestimate the magnitude of ambulance prevalence of the given outcome is limited to the study
diversion. Schull et al.17 reported that increased ambu- window. Furthermore, the absence of repeated data
lance diversion times were associated with inpatient collection further confines the ability to establish cause-
boarding, whereas walk-in patient volume and indivi- and-effect relationships for a given outcome. In this
dual ED physician coverage had no significant effect on descriptive study, the goal was not to establish a cause-
diversion time. For patients with high-risk chief com- and effect relationship between ED overcrowding and
plaints such as chest pain, there was a 28% increase in inpatient boarding. Despite this, a relationship was
transport time during periods of ambulance diversion.18 identified between these two variables. Our response
As a solution to the problems of overcrowding and rate, while greater than those of previous studies based
boarding, several initiatives have been proposed. Using a on responses by mail, was 81% and therefore did not
computer simulation model, investigators examined ED include every possible ED. The majority of the nonparti-
length-of-stay after increasing the number of ED beds cipating facilities responded that they were ‘‘too busy’’
versus a reduction in inpatient boarding times.19 This to participate. It is likely that many of these EDs met
model showed that increasing the number of ED beds our definition for overcrowding and were operating
actually increased ED length of stay from 240 to 247 min- beyond capacity. Without data from EDs self-reported
utes, whereas a reduction in inpatient boarding times as busy, it is possible that the data obtained from
reduced ED length of stay from 240 to 218 minutes. Simi- participating departments underestimated the rates of
larly, Viccellio et al.20 retrospectively examined the safety overcrowding and inpatient boarding in Michigan.
of admitted ED inpatients transferred to inpatient hall- Furthermore, this study was based on a single snapshot
way beds during periods of ED overcrowding. Patients in time. To increase the reliability of the results,
admitted to hallway beds had lower rates of mortality repeated sampling on multiple Mondays would need to
then patients admitted to standard inpatient beds with- be performed. There were only three EDs on diversion
out compromising patient safety. Furthermore, when during the sampling time. This seems to underestimate
patients were given the option of boarding in an ED hall- the authors’ experiences and the anecdotal experiences
way or an inpatient hallway until an inpatient room was of others. Gathering data on repeated Mondays and
available, 59% opted to be transported to an inpatient capturing nonparticipating EDs might garner a more
hallway bed.21 A retrospective study at a community accurate assessment of ambulance diversion. Finally,
hospital in Pennsylvania reviewed 62,588 patient visits caution should be used in generalizing Michigan results
between July 2004 and June 2005.22 By reducing to other states.
ACADEMIC EMERGENCY MEDICINE • December 2011, Vol. 18, No. 12 • www.aemj.org 1391