Received: 3 November 2019 Revised: 31 March 2020 Accepted: 21 April 2020

DOI: 10.1002/emp2.12094

ORIGINAL RESEARCH
The Practice of Emergency Medicine

Evaluation of an advanced practice provider emergency

department critical care step-down unit

Lisa Murphy NP1 Gino Paolucci NP1 Laura Pittenger PA-C, MPH1
Morayo Akande MPH2 Sarah J. Marks MS3 Roland C. Merchant MD, MPH3

1 Department of Emergency Medicine, Rhode

Island Hospital, Providence, Rhode Island, USA Abstract

2 Brown University, Providence, Rhode Island, Objective: In response to concerns about patient care and safety, our urban, tertiary
USA
care, Level 1 trauma center adult emergency department (ED) created an advanced
3 Department of Emergency Medicine, Brigham

and Women’s Hospital, Boston, Massachusetts,

practice provider-staffed critical care step-down unit (CCSU). We conducted a compre-
USA hensive evaluation of the CCSU’s impact on patient care, safety, and ED operations.
Methods: We compared ED length of stay, return visits to the ED within 72 hours, billing
Correspondence
Roland C. Merchant, MD, MPH, ScD, Department code assignments (current procedural terminology evaluation and management [CPT
of Emergency Medicine, Brigham and Women’s
E&M] codes), and quality of electronic health record documentation per QNOTE for
Hospital, 75 Francis Street, Boston, MA 02115,
USA. the 2 years after the CCSU was initiated (CCSU period) versus before its initiation (pre-
Email: [email protected]
CCSU period).
Preliminary results from this investigation were Results: There were 31,418 critical care ED patient visits in the pre-CCSU period and
presented at the 2018 American College of 33,396 in the CCSU period. Median ED length of stay did not change overall between
Emergency Physicians Research Forum, San
Diego, CA, October 1–3, 2018. the CCSU versus pre-CCSU period (∆1 [95% confidence interval (CI) = −2.4, 4.4] min-
utes), but decreased for patients who remained in the critical care suites (∆-4 [95% CI =
Funding and support: This project was supported
by a Lifespan Risk Management grant. −7.8, −0.2] minutes). 72-hour return ED visits also did not change overall (∆0% [95% CI
= −0.1, 0]), but decreased for patients who remained in the critical care suites (∆0.4%
[95% CI = −0.05, −0.4]). CPT E&M billing increased for highest-level visits (99,291:
∆1.3% [95% CI= 0.5, 2.0]). Quality of electronic health record documentation as mea-
sured by QNOTE also improved (∆11.5% [95% CI = 4.9, 18.1]).
Conclusion: This ED’s CCSU performance metrics indicate at least moderate improve-
ment in ED length of stay, 72-hour return visits, critical care patient billing, and elec-
tronic health record documentation. EDs elsewhere can consider implementation of
this advanced practice provider-staffed solution to improvement in critical care in ED.

KEYWORDS

critical care, emergency service, hospitals, nurse practitioners, operations research, physician
assistants

Supervising Editor: Christian A. Tomaszewski, MD, MBA.

This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any
medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.
○
c 2020 The Authors. JACEP Open published by Wiley Periodicals LLC on behalf of the American College of Emergency Physicians.

JACEP Open 2020;1–11. wileyonlinelibrary.com/journal/emp2 1

2 MURPHY ET AL

1 INTRODUCTION
The Bottom Line
1.1 Background This emergency department looked at the impact of staffing
a critical care step-down unit staffed with advanced practice
Due to continuous presentation of new patients to the critical care providers. This strategy decreased length of stay (by 29 min-
suites during normal operations at a large, urban, Level 1 trauma center utes) and 72-hour return (by 0.4%) for patients within the
and tertiary referral adult emergency department, patients frequently unit while improving (by 11.5%) documentation overall.
were moved from the ED critical care suites to the urgent care areas
before their treatment was complete or their disposition was final-
ized. This arrangement strained the effectiveness of the ED critical care
attending physicians and residents because they were responsible for
new incoming patients as well as those moved to the urgent care areas.
As a result, medical care was fragmented and length of stay was pro- 1.3 Goals of this investigation
longed. Re-evaluation and disposition were delayed due to compet-
ing priorities. When care was transferred to clinical providers in the We conducted a comprehensive evaluation of the impact of the CCSU
urgent care areas, loss of information and multiple transitions of care on patient care, safety, and ED operations. We specifically examined
over time occurred. Anecdotes of lower than expected medical care changes after the CCSU was initiated (CCSU period) as compared to
and patient safety events of critical care ED patients were reported by before its initiation (pre-CCSU period) according to ED length of stay,
ED staff. return visits to the ED within 72 hours, patient safety concern reports,
morbidity and mortality review referrals, visit level of care and critical
care billing code assignments, and quality of electronic health record
1.2 Importance documentation.

In response to concerns about patient care and safety, this ED created

a critical care step-down unit (CCSU) in a former observation care 2 METHODS
area of the ED. The CCSU was created as an observation care area
specifically for patients who were initially evaluated and treated in the 2.1 Setting
ED critical care suites. Any ED critical care patient who was considered
no longer to have immediate critical care needs but needed further The study setting is a 100 patient-bed, Level 1 trauma center, ter-
evaluation and treatment before a patient disposition was made could tiary referral adult ED in the United States (Rhode Island Hospi-
be transferred to the CCSU. Patients who were intubated, receiving tal, Providence, Rhode Island). The ED’s annual census is ≈110,000
vasopressors, were considered to have autonomic instability, or other- adult patient visits, of which the critical care annual patient volume is
wise required higher level of medical or nursing care were not eligible ≈16,000 patients. The ED medical staff is comprised of attending physi-
for the CCSU. Advanced practice providers (nurse practitioners and cians, emergency medicine resident physicians, and advanced practice
physician assistants) provided care in the CCSU. The ultimate intent providers. The physician staff, emergency medicine residency program,
of the CCSU was to improve patient care and safety and reduce risk and hospital are affiliated with the local medical school (Alpert Medi-
through cooperative care with the physician critical care and advanced cal School of Brown University). The hospital institutional review board
practice provider staff. Advanced practice providers were respon- approved the study.
sible for executing the plans for treatment and disposition of CCSU
patients, which included coordination of care with the ED critical
care physicians, consultants, nursing, and ancillary staff; performing 2.2 Study design
procedures (eg, lumbar punctures, incision and drainage of abscesses,
suturing lacerations); arranging for, reviewing the results of, and This investigation consisted of analyses of administrative and clini-
responding appropriately to laboratory testing and radiologic imaging; cal data from the following sources: databases on ED length of stay,
continuously re-evaluating each patient’s medical condition and needs; ≤72-hour return ED visits, and staff-reported safety concerns; mor-
communicating with family and outpatient care providers; arranging bidity and mortality review referrals; and electronic health records
follow-up and future care; and providing discharge instructions and of ED visits. Analyses were constructed to compare data for 2 years
patient education. This advanced practice provider-staffed CCSU prior to the debut of the CCSU (March 2011–February 2013; the pre-
reflects the practice of other EDs in exploring and expanding the roles CCSU period), and following its initiation (March 2013–February 2015;
advanced practice providers play in ED clinical care, as well as calls for the CCSU period). The end study date of February 2015 was chosen
more formal analyses of these roles.1-7 Although initiatives such as the because hospital system changed its electronic health record system
CCSU appear to be promising, their value needs formal assessment the following month, which affected the clinical documentation pro-
before they can be recommended for implementation elsewhere. cess, and which in turn affected how data were interpreted.
MURPHY ET AL 3

length of stay for patients who remained in the critical care areas only,
as opposed to those moved from the critical care to the urgent care
areas. The CCSU could enable critical care physician staff to be able
focus more on fewer critical care patients leading to quicker disposi-
tions, but this arrangement might have a lesser effect for the urgent
care medical staff.

2.5 Review and analysis of administrative and

clinical data

For all analyses, missing data were not imputed. Median length of
ED stay for all critical care patients and for each of the 5 patient
bed movement groups was calculated by disposition (all dispositions,
admissions, discharges, other dispositions [eg, transferred, eloped, left
without being seen, left against medical advice]) separately for the
F I G U R E 1 Patient bed movements in the ED relative to critical pre-CCSU and CCSU study periods. Differences with 95% confidence
care evaluation and treatment
intervals (CIs) were estimated. To permit an examination of direct and
indirect effects of the CCSU on ED operations, these calculations were
2.3 Patient population and unit of analysis repeated for ED visits that occurred during the 16 hours/day the CCSU
was in operation and separately for the hours it was closed. Differences
Patients for this investigation included those who were evaluated in in the proportions of ED return visits within 72 hours were compared
the critical care suites at any point during their stay during the study in a similar manner. The ED billing records database was queried for
period. The unit of analysis was ED visit, instead of individual ED the frequencies of the following 4 current procedural terminology
patients. As such, patients who had multiple visits to the ED that (CPT) evaluation and management (E&M) critical care codes were
included a critical care stay were counted on a per visit rather than a per used: 99,283 (Level 3 visit), 99,284 (Level 4 visit), 99,285 (Level 5 visit)
patient basis. ED visits were identified using the ED’s electronic health and 99,291 (critical care provided). These CPT E&M codes are used to
record. The electronic health record was queried for visits during which indicate to billing agencies that more complex, time-sensitive care was
patients were evaluated in the critical care suites at any point during provided to such patients in the ED, and hence translate into requests
their ED stay. We eliminated ED visits for patients who died during for higher reimbursement from insurers. Differences in proportions of
their ED stay; whose ED length of stay was <60 minutes or >24 hours; visits when these 4 codes were used also were compared across study
or were admitted to the ED psychiatric, chest pain or transient ischemic periods, patient bed movement groups, and dispositions.
attack observation unit. These visits were eliminated because these Electronic health records were evaluated for changes in the quality
patients likely would not be affected by the CCSU’s function because of provider documentation for the ED visit using the QNOTE evalua-
such patients would not be evaluated in that unit. tion system.8,9 QNOTE is a systematic method of evaluating the qual-
ity of clinical notes according to 12 elements (eg, history of present ill-
ness, review of systems, physical findings) considered to form the basis
2.4 Stratification of ED visit study sample by patient of an outpatient visit note. Quality of documentation of each element is
bed movements assessed according to criteria appropriate to that element (eg, for phys-
ical findings, criteria are the notes’ completeness, clarity, and concise-
ED patient visits were stratified into 5 patient bed movement groups ness) and are judged as missing or present, and if present, are judged
that reflected their acuity level and ED course through bed movements to be fully acceptable, partially acceptable, or unacceptable. For this
(Figure 1). The 5 patient bed movement groups consisted of patients investigation, we modified the 12 QNOTE elements to the 7 that were
who: (1) remained in the ED critical care suites only, (2) were moved within the control of the ED clinicians. Random samples of critical care
from the critical care suites to urgent care areas of the ED, (3) were ED visits for the pre-CCSU and post-CCSU study periods stratified by
moved from the ED critical care suites to the CCSU, (4) were moved the 5 patient bed movement groups were selected and the respective
from urgent care areas to the critical care suites, or (5) experienced electronic health records were obtained. Research staff not involved
multiple bed movements (to/from ED critical care suites, urgent care in the QNOTE assessment removed all patient and staff identifiers as
areas, and/or CCSU). Analyses were stratified by patient bed move- well as dates of service from the retrieved visit notes. The 3 advanced
ments because we expected that the CCSU would affect patient groups practice provider members of the investigative team independently
differently. We anticipated that ED operations would be affected dif- reviewed randomly selected samples of the electronic health records.
ferentially according to patient areas (eg, critical care suites, urgent The advanced practice provider investigators were blinded to which
care areas). For example, the CCSU might have a greater effect on ED electronic health records were drawn from the 5 patient bed move-
4 MURPHY ET AL

FIGURE 2 ED critical care visits and bed movements

ment groups as well as which were pre-CCSU or post-CCSU visits. Prior time with estimated 𝛽 coefficients, standard errors (SEs), and 95% CIs
to conducting their reviews, the advanced practice provider investiga- for the coefficients.
tors underwent iterative training on the QNOTE method that included
a joint group review of 10 records with discussion to reach agreement
on scoring criteria considerations and methodology, then an indepen- 2.6 Patient safety and morbidity and mortality
dent pilot review of 20 records with subsequent group discussion to report review and analyses
reconcile scoring criteria differences. Based on this pilot review, sam-
ple sizes of ED visits needed by the 5 patient bed movement groups Staff-reported safety concerns submitted through the hospital’s
were calculated to detect a 75% improvement in QNOTE score, com- anonymous reporting system were queried for the study period. How-
paring the post-CCSU versus pre-CCSU period. The advanced prac- ever, there were none relevant to the CCSU and so they could not be
tice provider investigators independently evaluated the resultant total analyzed. Morbidity and mortality review requests for cases submit-
sample size (n = 955) of visit notes using the QNOTE scoring system. ted to the department by any hospital staff member were examined for
Given the skewness of the data, we calculated the proportion of ED vis- the pre-CCSU and CCSU periods by 2 investigators. Reports specific
its for which QNOTE scores were 90% or greater, and calculated the to critical care patients were assessed qualitatively and summarized
difference in proportions between the CCSU and pre-CCSU periods, by content. We initially intended to focus on morbidity and mortality
along with corresponding 95% CIs. review requests for cases related to the CCSU and advanced practice
In addition, we performed interrupted time series analyses for all providers. However, few cases concerned advanced practice providers
patient bed movements for all dispositions for the 4 outcomes of and none the CCSU. As a result, instead we summarized each case and
median ED length of stay, repeat ED visits within 72 hours, critical care provided commentary on when we believed there might have been
visit (99,291) coding and QNOTE scores >90%. For repeat ED visits opportunities for which advanced practice provider or CCSU involve-
within 72 hours, critical care visit (99,291) coding and QNOTE scores, ment potentially could have improved patient care, if, counter to fact,
we created models using a log-link function with a random effect term the advanced practice providers and CCSU had been utilized in the
to adjust for attending physician, whereas for ED median length of stay, care of those patients. We focused our commentary on opportunities
we created a model using quantile regression with a fixed effect term for which care could have been improved by reducing risk, improving
to adjust for attending physician. We plotted the point estimates over patient safety, or preventing the error.
MURPHY ET AL 5

TA B L E 1 ED median length of stay by patient movement and disposition, CCSU versus pre-CCSU periods

Pre-CCSU CCSU CCSU versus. pre-CCSU

periodmedian periodmedian period∆ median
Patient movement Disposition minutes minutes minutes (95% CI)
All n = 31,418 n = 33,396
All dispositions 328 329 1 (−2.4, 4.4)
Admitted 354 356 2 (−1.8, 5.8)
Discharged 277 287 10 (5.3, 14.7)
Other 239 258.5 19.6 (1, 38.1)
Critical care only n= n=
All dispositions 253 249 −4 (−7.8, −0.2)
Admitted 289 285 −4 (−8.9, 0.9)
Discharged 202 188 −14 (−19.9, −8.1)
Other 195 200 5 (−18.9, 28.9)
Critical care to urgent n= n=
All dispositions 372 343 −29 (−34.9, −23.1)
Admitted 38 365 −16 (−24.2, −7.8)
Discharged 357 332.5 −24.2 (−34, −14.4)
Other 308 290.5 −16.7 (−54.7, 21.3)
Critical care to CCSU n= n=
All dispositions NA 358 NA
Admitted NA 376 NA
Discharged NA 326 NA
Other NA 310.5 NA
Urgent to critical care n= n=
All dispositions 398 375 −23 (−39.5, −6.5)
Admitted 448 417 −31 (−49, −13)
Discharged 281 265.5 −15.3 (−33.5, 2.9)
Other 247 247 0.7 (−66.4, 67.7)
Multiple bed movements n= n=
All dispositions 449 438 −11 (−21.1, −0.9)
Admitted 480 465 −15 (−30.7, 0.7)
Discharged 377.5 377 −0.8 (−18.4, 16.7)
Other 344 320 −24.4 (−88.1, 39.3)

CCSU, critical care step-down unit; CI, confidence interval; NA, not applicable.

3 RESULTS increased relative 9% in the CCSU versus pre-CCSU periods (CCSU

2169 − pre-CCSU 1978/CCSU 2169).
3.1 ED length of stay, 72-hour return visits, critical Total median ED length of stay and as stratified by disposition
care billing, and QNOTE analyses changed little in the CCSU versus pre-CCSU periods with a few excep-
tions (Table 1). Critical care patients were discharged a median of 10
There was a relative 6.3% increase in patient visits in the ED critical minutes later in the CCSU period. However, dispositions of critical care
care suites in the CCSU period as compared to the pre-CCSU period patients who remained in the ED critical care suites and those trans-
(pre-CCSU 31,418 − CCSU 33,396/pre-CCSU 31,418) (Figure 2). In ferred to or from critical care had shorter lengths of stay in the CCSU
comparison to the pre-CCSU period, there was an 11.9% relative period. A comparison of CCSU versus pre-CCSU ED length of stay iso-
decrease in ED patients who remained in the critical care suites (pre- lated only to the hours when the CCSU was staffed by advanced prac-
CCSU 12,933- CCSU 11,395/pre-CCSU 12,999), and a 57.3% relative tice providers showed similar trends, although patients remained in the
decrease in those moved from the critical care suites to the urgent care CCSU considerably longer in the overnight hours when no advanced
areas in the CCSU period (pre-CCSU 13,372 − CCSU 5707/pre-CCSU practice providers were assigned to staff the CCSU (Supporting Infor-
13,372). Patient movement from urgent care to the critical care suites mation Table S1).
6 MURPHY ET AL

TA B L E 2 ED revisits within 72 hours by patient movement and disposition, CCSU versus pre-CCSU periods

CCSU versus pre-CCSU

Patient movement Disposition Pre-CCSU period % CCSU period % period ∆% (95% CI)
All n = 31,418 n = 33,396
All dispositions 2.6 2.5 0 (−0.1, 0)
Admitted 1.2 1.1 −0.1 (−0.1, −0.1)
Discharged 4.9 4.5 −0.4 (−0.5, −0.3)
Other 7.6 10.0 2.4 (1.6, 3.2)
Critical care only n= n=
All dispositions 2.3 1.8 −0.4 (−0.5, −0.4)
Admitted 0.8 0.5 −0.3 (−0.3, −0.3)
Discharged 4.5 4.2 −0.3 (−0.5, −0.1)
Other 7.5 8.1 0.6 (−0.5, 1.7)
Critical care to urgent n= n=
All dispositions 2.9 3.5 0.5 (0.4, 0.6)
Admitted 1.5 0.9 −0.7 (−0.7, −0.6)
Discharged 5.2 4.7 −0.5 (−0.7, −0.3)
Other 7.7 11.5 3.8 (2.2, 5.5)
Critical care to CCSU n= n=
All dispositions NA 2.7 NA
Admitted NA 1.6 NA
Discharged NA 4.4 NA
Other NA 5.9 NA
Urgent to critical care n= n=
All dispositions 2.2 2.5 0.3 (0.2, 0.5)
Admitted 0.9 1.2 0.3 (0.2, 0.3)
Discharged 5.3 6.3 1.1 (0.4, 1.7)
Other 12.8 8.6 −4.2 (−8.3, −0.1)
Multiple bed movements n= n=
All dispositions 2.6 2.9 0.3 (0.2, 0.4)
Admitted 1.3 1.6 0.3 (0.2, 0.3)
Discharged 5.5 4.8 −0.7 (−1.1, −0.3)
Other 4.4 22.0 17.6 (13.9, 21.3)

CCSU, critical care step-down unit; CI, confidence interval; NA, not applicable.

Repeat ED visits within 72 hours were infrequent in both the CCSU movements, but critical care visit (99,291) coding was greater for those
and pre-CCSU periods (Table 2), likely due to the high admission rates. moved from urgent care to the critical care suites. In regard to quality
72-hour return ED visits in the CCSU period for critical care patients of electronic health record ED visit documentation, the proportion of
who were admitted or discharged from the ED decreased slightly, and visits for which QNOTE scores were 90% or greater increased overall
decreased for those who remained in the critical care suites only or in the CCSU period and for patients who remained in the critical care
moved to critical care from the urgent care. However, 72-hour ED suites, but not for other critical care patient visits (Table 4).
return visits increased for patients moved from urgent to critical care. Table 5 and Figure 3 provide the results of the interrupted time
CPT E&M coding for Level 5 and critical care visits (99,285 and 99,291) series analyses for all patient movements and all dispositions from the
was greater overall during the CCSU period, while coding for Level 3 ED for the outcomes of median ED length of stay, repeat ED visits
visits (99,284) was decreased (Table 3). Across all patient movement within 72 hours, critical care visit (99,291) coding, and QNOTE scores.
groups, coding for critical care visits (99,291) increased the most for For median ED length of stay, there were trends of decreasing length
patients who remained in the critical care suites. Level 3 and 4 visit of stay in the pre-CCSU period but increasing length of stay in the
(99,283, 99,284) coding was greater for patients who moved from the CCSU period, but no change in level of length of stay. For repeat ED
critical care suites to the urgent care areas. Level 5 visit coding (99,285) visits within 72 hours, there were no trends, but a level increase in the
was greatly decreased for critical care to urgent care area patient CCSU period. For critical care visit (99,291) coding, there was a trend
MURPHY ET AL 7

TA B L E 3 CPT E&M and critical care codes by bed movement and mentary of how potential CCSU use might have mitigated or reduced
disposition, pre-CCSU versus CCSU time periods the concerns raised about these cases or the case outcomes. The pre-
CCSU versus dominant themes based on our assessment of how CCSU care might
Patient Pre-CCSU CCSU pre-CCSU period ∆% have led to better patient care and safety included the potential for
movement period % period % (95% CI)
re-evaluation of patient complaints, re-examination of patients, more
All n = 31,418 n = 33,396 careful reviews of diagnostic imaging orders and results, initiation
99,283 3.6 3.6 0 (−0.3, 0.4) of appropriate consultations, and changes in patient disposition from
99,284 18.3 16.1 −2.2 (−2.9, −1.6) the ED. Our assessment results further suggest that by reducing the
99,285 56.4 57.3 1.0 (0.1, 1.8) crowding of the critical care suites, some patients might have had more
99,291 21.7 23.0 1.3 (0.5, 2.0) rapid response to critical care needs, diagnoses could have been made
Critical care n= n= sooner, patients from the urgent care areas could have been evaluated
only sooner in the critical care areas, and medical errors could have been
99,283 4.6 3.5 −1.1 (−1.7, −0.6) reduced.
99,284 16.6 12.0 −4.7 (−5.6, −3.7)
99,285 48.3 47.9 −0.4 (−1.8, 0.9)
4 LIMITATIONS
99,291 30.4 36.6 6.2 (4.9, 7.5)
Critical care n= n= First, the CCSU was initiated and evaluated at 1 ED. As such, there
to urgent
was no comparison group except for the ED itself. The impact of imple-
99,283 3.0 9.3 6.2 (5.5, 7.2)
menting related initiatives cannot therefore be directly inferred to
99,284 20.1 30.0 9.9 (8.4, 11.4) other EDs, especially with dissimilar ED operations and physical plants,
99,285 63.9 47.8 −16.1 (−17.7, −14.4) staffing models, payor complements, and patient clinical and demo-
99,291 13.0 12.9 0.1 (−1.3, 1.0) graphic characteristics. Second, pre- versus post-initiation of the CCSU
Critical care n= n= was limited by the matching electronic health record periods; CCSU
to CCSU impact after the change in electronic health record was not directly
99,283 NA 1.0 NA comparable. Third, inadequate documentation in the electronic health
99,284 NA 14.1 NA records overall limited the QNOTE assessment. Furthermore, better
99,285 NA 71.8 NA documentation in the CCSU period does not necessarily mean bet-
99,291 NA 13.1 NA ter medical care was provided. Fourth, improvements might reflect
Urgent to n= n= changes in practice not related to the CCSU, such as providers who
critical care improved their documentation practices resulting in higher CPT E&M
99,283 3.2 4.2 1.0 (−0.3, 2.3) coding. Fifth, some changes by patient bed movement group might
99,284 17.8 11.6 −6.2 (−8.6, −3.8) reflect changes in the patient mix due to the CCSU, instead of improve-
99,285 48.8 48.1 −0.7 (−4.0, 2.7) ments due to the CCSU, such as higher billing codes for patients who

99,291 30.3 36.1 5.9 (2.7, 9.0)

remained in the critical care areas. Sixth, given that there were no mor-
bidity and mortality referral requests relevant to the CCSU, the anal-
Multiple bed n= n=
movements yses of these cases were limited to commentary, which reflected the

99,283 2.1 2.4 0.3 (−0.5, 1.0) beliefs of the study authors.

99,284 18.1 15.3 −2.8 (−4.7, 0.9)

99,285 64.0 65.3 1.3 (−1.2, 3.7) 5 DISCUSSION
99,291 15.8 17.0 1.3 (−0.6, 3.1)

CCSU, critical care step-down unit; CI, confidence interval; CPT, current Our assessment of morbidity and mortality referrals suggested that
procedural terminology; E&M, evaluation and management; NA, not appli- had the CCSU been used more, fewer adverse patient incidents might
cable. have occurred. However, decision makers typically are more interested
of decreasing coding in the pre-CCSU period, no trend in the CCSU in performance metrics as indicators of the value of new ED operation
period, but a level increase in the CCSU period. There were no trends initiatives. The data indicate at least a moderate positive benefit of the
or level changes for QNOTE scores. CCSU on ED operations. Median length of stay decreased in general for
critical care patients in the critical care suites and urgent care areas,
likely because patients were diverted to the CCSU and enabled the
3.2 Morbidity and mortality referrals review medical team to concentrate on fewer patients and facilitate their dis-
positions more quickly. Although low due to the higher acuity of these
The summary of the morbidity and mortality referrals for the pre- patients, 72-hour return visits did not increase in the CCSU period,
CCSU and post-CCSU cases is provided in Table 6, as well as our com-
8 MURPHY ET AL

TA B L E 4 ED critical care patient visits with QNOTE scores 90% or greater by patient movement, pre-CCSU versus CCSU time periods

CCSU versus
Pre-CCSU Post-CCSU pre-CCSU period
Patient movement n period % n period % ∆% (95% CI)
All 73.6 85.2 11.5 (4.9, 18.1)
Critical care only 70.1 87.4 17.3 (10.3, 24.3)
Critical care to urgent 75.4 82.5 7.0 (−8.1, 22.1)
Critical care to CCSU NA 84.2 NA
Urgent to critical care 76.4 89.8 13.5 (−0.003, 27.2)
Multiple bed moves 78.8 83.0 4.1 (−7.7, 15.9)

CCSU, critical care step-down unit; CI, confidence interval; NA, not applicable.

TA B L E 5 Interrupted time series analyses models for all patient movements and all dispositions for ED median length of stay, ED revisits within
72 hours, critical care (99,291) billing, and QNOTE scores >90%

𝜷 SE 95% CI
ED median length of stay Trend before CCSU −0.54 0.14 −0.82, −0.26
Level change 0.93 1.07 −3.03, 1.17
Trend after CCSU 0.04 0.02 0.002, 0.09
ED revisits within 72 hours Trend before CCSU −0.003 0.004 −0.011, 0.006
Level change 0.09 0.03 0.04, 0.14
Trend after CCSU −0.002 0.001 −0.003, −0.001
Critical care (99,291) billing Trend before CCSU −0.05 0.01 −0.06, −0.03
Level change 0.11 0.03 0.05, 0.17
Trend after CCSU −0.001 0.001 −0.001, 0.0004
QNOTE scores >90% Trend before CCSU −0.005 0.06 −0.13, 0.12
Level change 0.17 0.20 −0.22, 0.56
Trend after CCSU −0.004 0.003 −0.01, 0.003

n.b. 72-hour ED visits, critical care billing, and QNOTE scores modeled using a log-link function with a random effect for attending physician, while ED median
length of stay modeled using quantile regression with a fixed effect for attending physician. CCSU, critical care step-down unit; CI, confidence interval; ED,
emergency department; SE, standard error.

and in fact decreased slightly for admitted and discharged patients. dents, critical care fellows and physician assistants.10 Study authors
Critical care CPT E&M billing increased in the CCSU period, perhaps reported a decrease in risk-adjusted 30-day mortality among all ED
because medical staff in the CCSU and critical care areas could devote patients. In 2013, the R. Adams Cowley Shock Trauma Center of the
more time to medical record documentation. This explanation is sup- University of Maryland Medical Center created a 9-bed, in ED criti-
ported by the finding that documentation quality improved in the cal care resuscitation unit.11 The research team observed that time
CCSU period, as measured through QNOTE. There were as expected elapsed from ED presentation to the operating room or further inpa-
compensatory decreases in coding levels among those patients who tient care was reduced after the unit was initiated. Koran et al12 exam-
were moved to the urgent care areas instead of remaining in the crit- ined the effects of the creation of an admission unit for patients await-
ical care suites. ing inpatient beds at a community hospital. Comparing the 2 months
Previous research on improving critical care delivery in the after versus before the unit opened, there were fewer instances for
emergency medicine setting generally has concerned disease or which patients who had been destined for intensive care unit transfers
condition-specific initiatives (eg, coronary disease, sepsis, stroke, and received slower responses from nursing staff for critical care needs.
trauma). Structural changes for critical care in the ED, such as the The authors attributed this improvement to reduction in ED crowding
CCSU described in this manuscript, have been reported infrequently, due to this structural change.
although none involve step-down critical care provided by advanced The impact of advanced practice providers on ED operations has
practice providers. The University of Michigan Medical Center opened been investigated in several studies, although in the context of general
a 9-bed intensive care unit adjacent to its adult ED in 2015 with a instead of critical care. Carter et al13 performed a systematic review
medical staff consisting of emergency medicine physicians and resi- of studies from Australia, the United Kingdom, and the United States
MURPHY ET AL 9

TA B L E 6 Summary of morbidity and mortality referrals for pre-CCSU and CCSU periods and potential for CCSU use for mitigating its
occurrence

Morbidity and
mortality Description of patient
classification presentation Adverse outcome Potential for CCSU use
Pre-CCSU period
Missed treatment Patient with chainsaw injury to No tetanus administered. Patient If the patient was transferred to the
leg developed a possible tetanus CCSU once stabilized, the advanced
infection practice provider could have
recognized the failure to administer a
tetanus shot and provided it
Missed diagnosis Patient with fall down stairs Patient had a missed ulnar fracture Missed injuries might have been
while in wheelchair and incomplete spinal cord injury recognized during CCSU
re-evaluation and observation
Missed diagnosis Patient with facial laceration Wood in wound missed and delayed in Missed injuries might have been
after walking into tree branch removal until after discharge. CT in recognized, CT finding clarified, and
ED had shown air in the orbit appropriate consultation obtained
during CCSU re-evaluation and
observation
Missed diagnosis Patient fell from ladder Missed pneumothorax while patient Radiology review could have been
was in critical care. Delayed conducted during CCSU
recognition after patient re-evaluation and pneumothorax
transferred to urgent care areas of discovered sooner
ED
Missed diagnosis Restrained driver in motor No initial pelvic x-ray ordered. Patient CCSU re-evaluation of patient
vehicle collision with later found to have small bowel complaints, examination, nursing
abdominal pain, per nursing injury and splenic laceration notes and diagnostic studies ordered
notes but not physician notes requiring operation might have led to sooner diagnosis
Missed diagnosis Patient with recent myocardial CT scan in ED showed possible CCSU re-evaluation of CT findings and
infarction who returned to ED pericardial effusion. Patient had a initiation of appropriate consultations
with back pain= ventricular wall rupture might have led to sooner diagnosis
Missed diagnosis Patient who fell and diagnosed Patient returned to ED after discharge CCSU re-examination of patient might
with clavicular fracture, radius and diagnosed with a thoracic burst have led to sooner diagnosis
fracture, and subdural fracture
hemorrhage
CCSU period
Missed diagnosis Driver involved in motor vehicle Patient returned to ED 3 days later CCSU re-examination of patient and
collision with hemoptysis, pneumothorax, re-evaluation of radiologic imaging
and multiple rib fractures ordered might have led to sooner
diagnosis
Missed diagnosis Intoxicated patient with Initial CT normal, but patient later CCSU re-evaluation of patient after
suspected trauma returned for abdominal pain and sobriety might have led to repeat CT
subsequent CT revealed splenic scan and earlier diagnosis
rupture
Inadequate care Motorized scooter operator hit Patient did not receive full assessment CCSU stay could have permitted
by a moving car of his wounds, lack of adequate evaluation of wound care needs,
wound care, incomplete orthopedic appropriate consultation, and
evaluation, and lack of addressing admission
his mobility; required admission 3
days later

of nurse practitioners on cost, quality of care, and wait times in the assistant care provided in the ED. Of the 3 studies that examined
emergency medicine setting. Quality of care was concluded to be at patient waiting times, 2 indicated that physician assistants reduced
least on par with resident physicians, patient satisfaction generally was overall waiting time when added to staffing, but had slightly longer
high, and wait times reduced when nurse practitioners were added patient length of stays as compared to physicians. A systematic review
to current staffing. In their systematic review, Halter et al14 reported by Doan et al15 had similar results on patient waiting time as the
the findings from 6 US studies and 1 Canadian study on physician Halter et al14 investigation. The authors also noted there had been 2
10 MURPHY ET AL

FIGURE 3 Interupted time-series analyses graphs. CCSU, critical care step-down unit; LOS, length of stay; TIA, transient ischemic attack

studies indicating similar cosmetic outcomes and complications from 2. Doetzel CM, Rankin JA, Then KL. Nurse practitioners in the emer-
wound care provided by physician assistants and physicians, and no gency department: barriers and facilitators for role Implementation.
Adv Emerg Nurs J. 2016;38(1):43-55.
differences in essential medical documentation between these types
3. Ginde AA, Espinola JA, Sullivan AF, Blum FC, Camargo CA, Jr. Use of
of providers in another study.15 midlevel providers in US Eds, 1993 to 2005: implications for the work-
In conclusion, this ED’s performance metrics indicate at least mod- force. Am J Emerg Med. 2010;28(1):90-94.
erate improvement in ED length of stay, 72-hour return visits, patient 4. Silvestri A, McDaniel-Yakscoe N, O’Neill K, et al. Observation medicine:
the expanded role of the nurse practitioner in a pediatric emergency
billing, and electronic health record documentation. EDs elsewhere can
department extended care unit. Pediatr Emerg Care. 2005;21(3):199-
consider implementation of this advanced practice provider-staffed 202.
solution to improvement in critical care in ED. 5. Menchine MD, Wiechmann W, Rudkin S. Trends in midlevel provider
utilization in emergency departments from 1997 to 2006. Acad Emerg
Med. 2009;16(10):963-969.
CONFLICT OF INTEREST
6. Wiler JL, Rooks SP, Ginde AA. Update on midlevel provider utiliza-
The authors declare no conflict of interest. tion in U.S. emergency departments, 2006 to 2009. Acad Emerg Med.
2012;19(8):986-989.
AUTHOR CONTRIBUTIONS 7. Sawyer BT, Ginde AA. Scope of practice and autonomy of physician
LM, GP, and LP were responsible for the initiation of the project and assistants in rural versus urban emergency departments. Acad Emerg
Med. 2014;21(5):520-525.
assisted in the study design, data collection, and analysis. MA assisted
8. Hanson JL, Stephens MB, Pangaro LN, Gimbel RW. Quality of outpa-
in the data collection and analysis. SM conducted the biostatistical tient clinical notes: a stakeholder definition derived through qualita-
analyses. RM oversaw the study design through its analysis and takes tive research. BMC Health Serv Res. 2012;12:407.
final responsibility for this manuscript. All investigators assisted in the 9. Burke HB, Hoang A, Becher D, et al. QNOTE: an instrument for
measuring the quality of EHR clinical notes. J Am Med Inform Assoc.
production of the manuscript.
2014;21(5):910-916.
10. Gunnerson KJ, Bassin BS, Havey RA, et al. Association of an emergency
REFERENCES department-based intensive care unit with survival and inpatient
1. Abbott PD, Schepp KG, Zierler BK, Ward D. The use of nurse practi- intensive care unit admissions. JAMA Netw Open. 2019;2(7):e197584.
tioners and physician assistants in Washington and Oregon Emergency 11. Scalea TM, Rubinson L, Tran Q, et al. Critical care resuscitation unit: an
Departments: a descriptive study of current practice. Adv Emerg Nurs J. innovative solution to expedite transfer of patients with time-sensitive
2010;32(4):338-345. critical illness. J Am Coll Surg. 2016;222(4):614-621.
MURPHY ET AL 11

12. Koran Z. The impact of an admission unit on failure- and late-to-rescue SUPPORTING INFORMATION
rates in the emergency department. Adv Emerg Nurs J. 2007;29(4):339- Additional supporting information may be found online in the Support-
345.
ing Information section at the end of the article.
13. Carter AJ, Chochinov AH. A systematic review of the impact of nurse
practitioners on cost, quality of care, satisfaction and wait times in the
emergency department. CJEM. 2007;9(4):286-295.
14. Halter M, Wheeler C, Pelone F, et al. Contribution of physician assis-
tants/associates to secondary care: a systematic review. BMJ Open. How to cite this article: Murphy L, Paolucci G, Pittenger L,
2018;8(6):e019573. Akande M, Marks SJ, Merchant RC. Evaluation of an advanced
15. Doan Q, Sabhaney V, Kissoon N, Sheps S, Singer J. A systematic review: practice provider emergency department critical care
The role and impact of the physician assistant in the emergency depart-
step-down unit. JACEP Open. 2020;1-11.
ment. Emerg Med Australas. 2011;23(1):7-15.
https://doi.org/10.1002/emp2.12094

AUTHOR BIOGRAPHY

Roland C. Merchant, MD, MPH, is Asso-

ciate Professor of Emergency Medicine,
Brigham and Women’s Hospital, Harvard
Medical School, and Adjunct Professor of
Emergency Medicine and Epidemiology
at Brown University. His medical degree
is from the Emory University School of
Medicine, and he has a master’s degree in
public health (quantitative methods) and a doctorate in epidemiol-
ogy from the Harvard School of Public Health.