a b s t r a c t

Objective: Evaluate an established scribe program on throughput and revenue capture in an Emergency Depart- ment (ED) undergoing an EMR transition.

Methods: A prospective cohort design comparing patients managed with and without scribes in an academic ED. Throughput metrics (medians, min) and relative value units (RVUs, means) were collected. Data was evaluated in its entirety (three months), as well as in two subsets: go live (immediate two weeks) and adoption (two weeks post implementation to end).

Results: All patients: There was no significant difference in throughput or RVUs during the three month period. During go-live, scribes showed improvement in total RVUs per patient (4.63 vs 4.40, p = 0.048). During adoption, scribed patients had decreased length of stay (LOS, 221 vs 231, p = 0.023).

Adults: Door to provider (28 vs 37, p = 0.014) and total RVUs (5.20 vs 4.92, p = 0.042) were improved with scribes in the go-live period. Scribes improved go-live morning and overnight shifts, while lengthening provider to disposition during afternoon shifts. No significant differences were seen in the adoption period, except for in- creased provider to disposition time overnight with scribes (154 vs 146, p = 0.030). Pediatrics: When all pedi- atric patients were compared, scribe patients had a decreased professional RVU charge (2.78 vs 2.90, p = 0.037). During go live and adoption, no significant differences were found in any other parameter or subgrouping.

Conclusions: A scribe’s ability to mitigate operational inefficiencies introduced by an EMR transition seems limited in an academic hospital. Previous research highlighting the impact of scribes on revenue was not replicated dur- ing this study.

(C) 2019

1. Introduction
   1. Background

medical scribes are an increasingly popular service utilized across the United States to decrease provider cognitive load and time providers spend focusing on clerical work [1-3]. In doing this, scribes also benefit the providers with whom they work by allowing them to spend more time focusing on the patient, creating a more personable patient-to-

? Presented at: SAEM Annual Meeting 2019, Las Vegas, Nevada.

* Corresponding author.

E-mail address: [email protected] (H.A. Heaton).

provider interaction [4]. Medical scribes document the patient history and exam findings directly into the electronic medical record (EMR) contemporaneously with the patient encounter, expediting provider workflow in the process [2,4,5].

Implementation of a new EMR results in increased time for numer- ous operational metrics within an emergency department (ED). These metrics remain elevated during the first six months after an EMR rollout before returning back to their baseline [6].Although research has evalu- ated the impact of a new EMR implementation on ED efficiency, there is no research on the impact medical scribes have during an EMR transi- tion. Previous studies have demonstrated that scribes have a positive impact on the throughput of an adult ED at baseline; therefore, discern- ing the level of impact, if any, scribes provide during this transition pe- riod, if any, would be useful [5,7-12].

https://doi.org/10.1016/j.ajem.2019.158433

0735-6757/(C) 2019

Importance

Understanding the utility of medical scribes during an EMR transi- tion is important; if utility can be shown, scribe programs could be de- ployed as a tool to help to not only improve operational metrics in the ED, but also increase the quality of patient care during an implementa- tion, and mitigate the Financial burden incurred by extended opera- tional inefficiencies.

Goals of this investigation

This study aims to evaluate the effect of an established scribe pro- gram on Patient throughput and revenue capture in an ED undergoing a transition between two EMRs.

1. Methods
   1. Study design and setting

This study was completed in the ED of a large academic hospital in a Midwest, urban community. Our ED volume averages 78,000 patients annually, with 82% adults (N17 years old) and 18% pediatrics. Thirty- five percent of adult patients and 16% of pediatric patients are admitted. Our hospital is a quaternary care center for multiple specialties and is a Level 1 Trauma Center. Our study is a prospective cohort analysis. This study was deemed exempt of review by our Institutional Review Board (IRB).

This study occurred during the transition between two EMRs. While our hospital was on an EMR prior to implementation, our product was a standalone ED documentation tool with the hospital and clinic on a dif- ferent system. The transition allowed our hospital system to move onto a single EMR. Our enterprise has 21 emergency departments; our de- partment was the 19th department to roll out this EMR.

Selection of participants

For this study, adult and pediatric patients registered between May 5th and July 31st, 2018 were eligible for inclusion. All adults roomed in ED Treatment Area A, a high Acuity area of the adult ED, open 24 h a day staffed with a board certified Emergency Medicine (EM) attending physician, a senior EM resident, and an intern, were included. All pedi- atric patients, roomed in Treatment Area B, a dedicated pediatric ED staffed with a board certified EM or Pediatric EM attending physician and two to three multi-specialty residents (EM, pediatrics, family med- icine), evaluated during the evening shift were included. Less than one- third of the residents participating in this study were from specialties other than the ED.

Intervention

At the time of this study, our department directly employed eight medical scribes in two full-time positions and six part-time positions. Our scribe program began 2015 with training received via an in-house program [13].

For the study period, we divided ED patients into two groups: (1) pa- tients managed by a “traditional” care team where providers used per- sonal preference to construct their own document in the medical record without scribe utilization; and (2) patients managed by a care team that included a medical scribe who documented for the attending physician. For traditional care teams with no scribes, the resident docu- mented a complete medical chart with the attending physician com- pleting an attestation note. For scribED patient encounters, the attending completed the medical chart, including the history of present illness (HPI), Review of Systems (ROS) and Physical Exam in addition to the medical decision making; residents were expected to document their Physical exam findings and plan, but did not duplicate the HPI or ROS documentation. Scribes also assisted the care team by increasing awareness of laboratory and radiology findings, in particular when the patient was noted to be ready for re-evaluation for disposition. This is the practice our department has followed for several years, and was not modified during the EMR transition.

Scribes were assigned to an attending physician during eight-hour

shifts, following an allocation pattern developed independently from providers’ schedules, with no preference given based on specific pro- viders. We developed an allocation scheme to allow for accurate com- parison between intervention and control patients. The pattern was followed without deviation throughout the study period. Patients seen during a day shift with a scribe were compared to patients seen during a day shift without a scribe, and so on.

Scribes underwent the same EMR training as the provider staff. Of note, one full time scribe was trained as a Super User and worked both as a scribe and Super User during this time period.

Methods and measurements

Investigators extracted patient demographics, patient-specific timestamps, and billing data from the EMR. We evaluated patient- specific throughput metrics and billing data for each included visit dur- ing the study period. Discrete fields in the EMR allowed us to identify all patients for whom a scribe was part of the care team.

In treatment areas where scribes covered more than one shift in a 24 hour period, patients were further categorized by their arrival time into shifts including: morning (0700-1500), afternoon (1500-2300), or overnight (2300-0700).

Three months of data were evaluated in entirety, as well as in two subsets: go live (immediate two weeks following implementation) and adoption (two weeks post implementation of EMR to end of the study period).

Table 1

Participant demographics (adults and pediatrics).

	Adults				Pediatrics
	Non-scribed N = 2312 Median or n (%)	Scribed N = 2317	p value		Non-scribed N = 885 Median or n (%)	Scribed N = 921	p value
Male	1153 (50)	1156 (50)	0.99		440 (50)	478 (52)	0.35
ESI 1	41 (2)	56 (2)	0.27		4 (b1)	3 (b1)	0.75
ESI 2	502 (22)	525 (23)			74 (9)	73 (8)
ESI 3	1381 (60)	1364 (59)			366 (42)	395 (44)
ESI 4	364 (16)	343 (15)			412 (47)	427 (47)
ESI 5	2 (b1)	15 (1)			12 (2)	10 (1)
Disposition: Admitted	920 (40)	892 (39)	0.37		128 (14)	94 (10)	0.006

ESI: Emergency Severity Index.

Outcomes

Measures analyzed for both the scribe and non-scribed groups

Table 2

Adult ED throughput metrics.

Non-scribed Scribed p

included:

1. Length of stay (LOS, min): arrival time until departure time from the department
2. Door to provider (min): arrival time until first seen by a provider
3. Treatment room time (min): total time spent in the treatment room (equals total ED LOS minus any time spent in the waiting room)
4. Provider to disposition (min): time first seen by a provider until the Disposition decision was made and entered in the EMR
5. Total relative value units (RVUs): financial metric coded by an exter- nal vendor (LogixHealth, Bedford, MA).
   1. Analysis

Continuous features were summarized with means and medians. Categorical features were summarized with frequency counts and per- centages. Comparisons between the non-scribe and scribe groups

were evaluated using Wilcoxon rank sum or chi-square tests. Statistical

Median minutes

value

analyses were performed using version 9.4 of the SAS software package (SAS Institute Inc.; Cary, NC). Any p-value b 0.05 was considered statis- tically significant.

All adult patients	N = 2312	N = 2317
Length of stay	272	267	0.34
Door to provider	27	25	0.064
Treatment room	221	222	0.67
Provider to disposition Adult patients, go live	163 N = 348	166 N = 370	0.32
Length of stay	295	290	0.25
Door to provider	37	28	0.014
Treatment room	251	247	0.96
Provider to disposition Adult patients;	161 N = 1964	179 N = 1947	0.52
adoption Length of stay	269	265	0.53
Door to provider	25	24	0.29
Treatment room	217	216	0.75
Provider to disposition	163	165	0.48

1. Results
   1. Characteristics of study subjects

A total of 6435 (4629 adults and 1806 pediatric) patients visits were included in this evaluation. All patients roomed in a study area were in- cluded. There was no significant difference in gender or Emergency Se- verity Index (ESI) between the non-scribed and scribed cohorts. In the adult group, there was no significant difference in the number of pa- tients discharged between the non-scribed and scribed groups. In con- trast, the pediatrics population contained a significantly higher number of patients discharged in the non-scribed group (p = 0.006). See Table 1.

Main results

Adult scribed patients had a significantly decreased door to provider time during the immediate period following go live (median 28 min vs 37 min, p = 0.014); however, when all adult patients were considered together, and when the adult patients seen in the adoption period were evaluated, that same significance was not found. In the adult population, scribes had no effect on LOS, treatment room time, or provider to dispo- sition decision, nor when the go live group and adoption group were evaluated. See Table 2.

Scribes made a significant impact during go live morning shifts for the adult patients with decreased LOS (median 258 vs 301 min, p =

All adult patients		N = 736	N = 772
	Length of stay	267	257	0.13
	Door to provider	20	19	0.64
	Treatment room	245	233	0.11
Adult patients, go live	Provider to disposition Length of stay	189 N = 107 301	179 N = 128 258	0.18 0.012
	Door to provider	21	25	0.87
	Treatment room	290	251	0.022
Adult patients; adoption	Provider to disposition Length of stay	203 N = 629 261	160 N = 644 256	0.007 0.53
	Door to provider	19	18	0.75
	Treatment room	241	231	0.39
	Provider to disposition	188	181	0.77

Scribes made no significant impact in throughput metrics for pediat- ric patients in the go live period, adoption period, or when all pediatric patients were considered together. See Table 6.

Scribes improved total RVU capture during the go live period (mean 5.20 vs 4.92, p = 0.042), but not in the adoption period nor when all adult patients were considered together (Table 7). Scribes made no sig- nificant impact on RVUs in the pediatric patients (Table 8).

Limitations

There are several limitations of this study to be considered when evaluating the outcomes. First, this study took place at an academic cen- ter with resident learners from multiple specialty fields. The study pe- riod spanned across the month of July; this is significant in an academic ED. The amount of new learners and new senior residents learning new roles could have affectED throughput measures; however, these effects may have been partially mitigated by the study design as the comparisons were done within the same groups of learners (e.g. go live learners compared to go live learners, adoption learners to adop- tion learners).

The study does not look at the impact of scribes with a specific pro- vider, but rather focuses on a specific shift. This study does not control for individual provider differences and their particular strategies for uti- lizing scribes. During the “go live” period there was readily available on- site EMR support staff. It is possible that scribes were not used to their full potential during this period as physician staff was prioritizing their

Table 3

Adult ED throughput metrics, morning shift.

Non-scribed Scribed p

0.012), reduced treatment room time (median 251 vs 290 min, p = 0.022), and shorter provider to disposition decision (median 160 vs 203 min, p = 0.007). They made no statistically significant impact in the adoption period (Table 3).

Scribes made no statistically significant improvement in adult throughput metrics during the afternoon. Provider to disposition was significantly longer in scribed patients (median 189 vs 140 min, p = 0.016) during go live. See Table 4.

Overnight, scribes were associated with improved door to provider metrics when all overnight adult patients were considered together (median 21 vs 28 min, p = 0.011) and in the go live period (22 vs 34 min, p = 0.019). Additionally, scribes were associated with signifi- cantly lengthened provider to disposition decision duration in the adop- tion period (154 vs 146 min, p = 0.030). See Table 5.

Median minutes

value

Table 4

Adult ED throughput metrics, afternoon shift.

Non-scribed Scribed p

Table 6

Pediatric ED throughput metrics.

Area Non-scribed Scribed p

Median minutes

value

Median minutes

value

All adult patients	Length of stay	N = 748 294	N = 788 291	0.86		All pediatric patients		N = 885	N = 921
	Door to provider	42	33	0.42			Length of stay	140	139	0.26
	Treatment room	223	224	0.91			Door to provider	15	15	0.53
	Provider to disposition	168	169	0.94			Treatment room	125	123	0.27
Adult patients, go live	Length of stay	N = 140 306	N = 131 320	0.52			Provider to disposition	95	91	0.68
	Door to provider	54	40	0.23		Pediatric patients, go live		N = 194	N =
	Treatment room	240	251	0.30					162
	Provider to disposition	140	189	0.016			Length of stay	147	154	0.22
Adult patients; adoption		N = 608	N = 657				Door to provider	19	24	0.19
	Length of stay	292	285	0.65			Treatment room	129	138	0.45
	Door to provider	39	33	0.80			Provider to	87	94	0.42
	Treatment room	220	216	0.66			disposition
	Provider to disposition	172	165	0.26		Pediatric patients,		N = 691	N =

own learning of the EMR system while at-elbow support was available rather than using scribes to improve efficiency in this setting.

Of note, due to scribe staffing limitations we were not able to use scribes to evaluate metrics in regards to attending only throughput values and RVUs (no resident). Attending only paired with a scribe would be more consistent with a community-style setting of Emergency Medicine. Additionally, the pediatric area only had scribes in the eve- ning, which may have skewed the results.

Finally, our practice skews toward higher admission rates than the majority of EM practices in the United States; the complexity and acuity of our patient population might impact the applicability of this study to other practices.

1. Discussion

Previous research has shown that emergency physicians spend the majority of their time on shift doing non-patient centered tasks [2]. Scribes have been hypothesized as a solution to increase throughput metrics for physicians and decrease clerical burden; however, the data thus far have been mixed. One study found no significant difference in LOS or time to disposition, but a small increase in the number of patients seen per hour using scribes [3]. Other studies have found that all throughput metrics improved with scribe implementation, including improved overall door-to-physician time [9].

This study sought to evaluate the usefulness of scribes during a tran- sition between two EMRs, both immediately in the “go live” period and then over an “adoption” period, for a total of 3 months. We evaluated throughput and RVU metrics over this time period in adult and pediatric

Table 5

Adult ED throughput metrics, overnight shift.

Non-scribed Scribed p

areas, which were further broken down by evaluating adult ED through- put metrics by different shift type, including morning, afternoon and overnight shifts.

adoption			759
	Length of stay	140	137	0.094
	Door to provider	14	14	0.72
	Treatment room	124	121	0.16
	Provider to disposition	97	90	0.39

Overall, scribes made a statistically significant impact during the “go live” period to decrease the door to provider time in the adult ED, re- gardless of shift type. The study would suggest that scribes made the most statistically significant impact during the morning shifts of the “go live” period. Scribes were associated with improved the LOS, treat- ment room time, and provider to disposition time. One hypothesis is that in the morning the system is not as burdened with accrued tasks, and typically there are fewer patients in the waiting room. Additionally, the department increased provider and nurse staffing during the first two weeks of go live. We attempted to control for this with the alternat- ing day pattern, however, comparing the results between the go live pe- riod and the adoption period is limited. The staffing model during the adoption period was closer to typical staffing levels.

There was no statistically significant impact on pediatric throughput

measures with scribes, regardless of time period evaluated, consistent with previous studies [12]. The reason for this lack of impact in the pe- diatric ED could be secondary to the variability of volume and Acuity of patients in the department, but further research would be required to understand this phenomenon.

Strengths of this study include the Methodological rigor by which it was conducted. The same two areas (adult and pediatric) were evalu- ated throughout the study period on alternating days with and without scribes. This study also evaluated scribes in a pediatric area, which has not been researched previously.

In summary, the presence of scribes in an academic hospital ED

Median minutes

All adult patients		N = 828	N = 757	ineffic
	Length of stay	264	265 0.86
	Door to provider	28	21 0.011
	Treatment room	198	210 0.092	Table 7

value

transitioning between EMRs did not significantly mitigate operational iencies. Another consideration is the other hospital areas that

Provider to disposition 146 156 0.011

Adult ED RVUs.^a

Adult patients, go live N = 101 N = 111

Length of stay 279 288 0.85 Area Non-scribed Scribed p value

a Mean summarized for this variable to illustrate the differences between the groups more clearly given the somewhat unique nature of this variable (limited options for vari- able output).

	Door to provider Treatment room	34 232	22 234	0.019 0.27		Mean
	Provider to disposition	146	178	0.23	All adult patients	4.72	4.79	0.76
Adult patients; adoption		N = 727	N = 646		Adult patients, go live	4.92	5.20	0.042
	Length of stay	263	263	0.77	Adult patients, adoption	4.69	4.71	0.56

	Door to provider	26	21	0.062
	Treatment room	195	202	0.21
	Provider to disposition	146	154	0.030

Table 8

Pediatric RVUs.^a

Area Non-scribed Scribed p value Mean

Declaration of competing interest

None.

References

All pediatric patients	3.32	3.22	0.20
Pediatric patients, go live	3.51	3.32	0.65	[1] Heaton HA, Wang R, Farrell KJ, et al. Time motion analysis: impact of scribes on pro-
Pediatric patients, adoption	3.27	3.19	0.26	vider time management. J Emerg Med. 2018;55(1):135-40 Jul. Epub ahead of print .

^a Mean summarized for this variable to illustrate the differences between the groups more clearly given the somewhat unique nature of this variable (limited options for vari- able output).

interface with the ED and affect throughput are also learning a new sys- tem and throughput metrics may be prolonged, regardless of scribe presence. There is likely a lot more interplay between other depart- ments and the ED affecting its throughput metrics; further research is required to determine the relative effects of EMR implementation on overall ED metrics with respect to supportive departments such as radi- ology and lab.

Grant support

None.

Author contributions

HAH, KAK, TRH were involved with study concept and design; HAH, RJM, KAK participated in acquisition of the data; HAH, EJS, WJG, and CML participated in analysis and interpretation of the data; CML did the statistical analysis; HAH, EJS, WJG were involved in the initial draft of the manuscript; all participated in critical revision of the manuscript.

https://doi.org/10.1016/j.jememed.2018.04.018.

1. Chisholm CD, Weaver CS, Whenmouth L, et al. A task analysis of emergency physi- cian activities in academic and Community settings. Ann Emerg Med. 2011;58: 117-22.
2. Heaton HA, Castaneda-Guarderas A, Trotter ER, Erwin PJ, Bellolio MF. Effect of scribes on patient throughput, revenue and patient and provider satisfaction: a sys- tematic review and meta-analysis. Am J Emerg Med. 2016;34:2018-28.
3. Joint Commission clarifies use of documentation assistants, Perspectives. 2018(8): 38.
4. Heaton HA, Nestler DM, Lohse CM, Sadosty AT. Impact of scribes on emergency de- partment patient throughput one year after implementation. Am J Emerg Med. 2017;35(2):311-4. https://doi.org/10.1016/j.ajem.2016.11.017 Feb.
5. Ward MJ, Landman AB, Case K, Berthelot J, Pilgrim RL, Pines JM. The effect of elec- tronic health record implementation on emergency department operational mea- sures of performance. Ann Emerg Med. 2014;63(6):723-30. https://doi.org/10. 1016/j.annemergmed.2013.12.019.
6. Allen BBB, Weeks E, Payton T. An assessment of emergency department throughput and provider satisfaction after the implementation of a scribe program. Adv Emerg Med. 2014;2014:7.
7. Arya R, Salovich DM, Ohman-Strickland P, Merlin MA. Impact of scribes on perfor- mance indicators in the emergency department. Acad Emerg Med. 2010;17:490-4.
8. Bastani A, Shaqiri B, Palomba K, Bananno D, Anderson W. An ED scribe program is able to improve throughput time and patient satisfaction. Am J Emerg Med. 2014; 32:399-402.
9. Hess JJ, Wallenstein J, Ackerman JD, et al. Scribe impacts on Provider experience, op- erations, and teaching in an academic emergency medicine practice. West J Emerg Med. 2015;16:602-10.
10. Walker K, Ben-Meir M, O’Mullane P, Phillips D, Staples M. Scribes in an Australian private emergency department: a description of physician productivity. Emerg Med Australas. 2014;26:543-8.
11. Heaton HA, Nestler DM, Jones DD, Lohse CM, Goyal DG, Kallis JS, et al. Impact of scribes on patient throughput in adult and pediatric academic EDs. Am J Emerg Med. 2016;34:1982-5.
12. Heaton HA, Samuel R, Farrell KJ, Colletti JE. Emergency department scribes; a two- step training program. Ann Emerg Med. 2015;66(4):S159.