a b s t r a c t

Study objectives: To compare diagnostic test ordering practices of NPs with those of physicians in the role of Pro- vider in Triage (PIT).

Methods: This was a secondary analysis of data from a prospective RCT of waiting room diagnostic testing, where 770 patients had diagnostic studies ordered from the waiting room. The primary outcome was the number of test categories ordered by provider type. Other outcomes included total tests ordered by the end of ED stay, and time in an ED bed. We compared variables between groups using t-test and chi-square, constructed logistic regression models for individual test categories, and univariate and multivariate negative binomial models.

Results: Physicians ordered significantly more diagnostic test categories than NPs (1.75 vs. 1.54, p b 0.001). By the end of their ED stay, there was no significant difference in total test categories ordered between provider type: physician 2.67 vs. NP 2.53 (p = 0.08), using a nonbinomial model, Incidence rate ratio (IRR) 1.07 (0.98-1.17). Pa- tient time in an ED bed was not significantly different between physicians and NPs (NP 244 min, SD = 133, Phy- sicians 248 min, SD = 152) difference 4 min (-24.3-16.1) p = 0.688.

Conclusion: NPs in the PIT role ordered slightly less diagnostic tests than attending physicians. This slight differ- ence did not affect time spent in an ED bed. By the end of the ED stay, there was no significant difference in total test categories ordered between provider types. PIT staffing with NPs does not appear to be associated with ex- cess test ordering or prolonged ED patient stays.

Background

nurse practitioners are increasingly staffing emergency de- partments (EDs) [1]. Initially delegated to “fast track” areas seeing lower acuity patients, their roles are expanding [2,3]. One place where NPs may be valuable is as Providers in Triage (PIT) - Screening patients and initiating diagnostic evaluations when beds are unavail- able in the ED and waits are long [3]. The concept of PIT involves placing a provider in or near the waiting room in order to perform patient assessments and initiate diagnostic evaluations and thera- peutic interventions prior to patients being brought in to the ED and evaluated by their definitive provider. The PIT role is evolving as ED wait times inexorably increase [4]. There is evidence that screening patients while in the waiting room provides an opportuni- ty to expedite their entire visit [5,6]. While one review suggested that using NPs in the PIT process “is an intelligent strategy for both

* Corresponding author at: Olive View UCLA Emergency Medicine, 14445 Olive View Drive, Sylmar, CA 91342, United States.

E-mail address: [email protected] (T. Begaz).

patients and institutions,” that same article stated that further re- search is necessary to demonstrate the benefits NPs may provide, and that comparison between NPs and physicians in this role would be valuable [3].

One controversy relates to possible differences in resource utili- zation between NPs and physicians. Studies in a variety of clinical settings have reported that NPs tend to order more tests [7-9], or order specialist consults at a higher frequency than physicians [9]. Other studies have not shown these differences [10,11], and some have demonstrated lower resource utilization by NPs as compared to physicians [12].

Determining practice differences between NPs and physicians in test ordering at triage has important implications for staffing busy emergency departments. We recently completed a large random- ized controlled trial (RCT) of waiting room diagnostic testing using PIT in a Los Angeles County ED, where approximately half of the encounters were with an NP and the rest were with an attending emergency physician. The objective of this study was to compare test ordering practices of NPs with those of physicians in the PIT role.

http://dx.doi.org/10.1016/j.ajem.2017.04.027 0735-6757/

Materials and methods

Study design

This was a secondary analysis of data from our previously published RCT which evaluated the initiation of diagnostic testing on patients in the waiting room [5]. In the original study, either an attending emergen- cy physician or an NP screened 1659 nonpregnant adults presenting with a chief complaint of abdominal pain and an Emergency Severity Index of 3 while they were in the waiting room of a Los Angeles County ED. This waiting room screen was called rapid medical evaluation (RME). These patients were then randomized to RME only or RME

+ waiting room diagnostic testing (RME + WRDT). Patients random- ized to the RME + WRDT group had laboratory tests and imaging or- dered at the discretion of the screening provider while they were still in the waiting room. Patients in the RME-only group had the same screening and were also sent back to the waiting room, but testing was deferred until the patient was assigned a bed in the ED and evalu- ated by his or her definitive provider. In the current study, we analyze data from the 770 patients in the RME + WRDT group to compare test ordering between ED attending physicians and NPs in the role of PIT.

Setting

We collected data during a 10-month period beginning July 23, 2014, and ending May 4, 2015, at a Los Angeles County ED with an an- nual census of approximately 55,000 visits per year. This ED is part of a university-affiliated teaching hospital with an emergency medicine residency. All patients who arrived in the ED were initially triaged by a trained nurse using the Emergency Severity Index , a validated triage system with a scale of 1 (critical) to 5 (nonurgent) [13]. Any pa- tients deemed unstable to wait were brought immediately to an ED room at the discretion of the triage nurse. Additionally, if there were ample ED rooms available, patients were immediately brought to a room regardless of ESI score. During times when all ED beds were occu- pied, stable patients were sent to the waiting room after initial nurse tri- age. PITs, who were either an attending emergency physician or nurse practitioner, were stationed near the waiting room to provide brief medical screening to all waiting room patients. Monday through Friday from 8 AM to 8 PM, we staffed an NP whose only clinical responsibility was as PIT. Additionally, there was an attending physician from noon to midnight daily whose only responsibility was PIT. In the late night or early morning hours, the PIT role was covered by either an NP or attend- ing physician who also had other clinical responsibilities. Nine PIT NPs with a median of 5 years ED nurse practitioner experience participated in the study. Resident physicians were not involved in the PIT process. There were 26 Attending emergency physicians in the study, with a me- dian of 9 years post-residency experience.

Selection of participants

We enrolled all nonpregnant adult patients with a chief complaint of abdominal pain and an Emergency Severity Index of 3 who were deemed stable to return to the waiting room. Exclusion criteria were: pregnancy (either known at triage or discovered during evaluation), age younger than 18 years, and screening provider judgment that the patient was unstable to return to the waiting room. The Olive View UCLA institutional review board approved the study and waived the need for informed consent.

Methods of measurement

A team of trained research assistants recorded patient demo- graphics. We recorded the following patient information: age; sex; presence or absence of hypertension, diabetes, or liver disease; history

of appendectomy, cholecystectomy, or other abdominal surgery; final diagnosis; disposition (home, admit, or leaving before completion of service); ordering provider type; and test categories ordered at screen- ing and in the ED. The test categories were blood, urine, radiograph, CT scan, and ultrasonography. We recorded location of test ordering (PIT or ED), the number of test categories ordered at each location, and time stamps from the medical record (ED arrival, PIT screening, bed assign- ment, ED provider assignment, and disposition time). Day and night ar- rival times were defined as 7 AM to 7 PM and 7:01 PM to 6:59 AM, respectively. Day of week was extracted from the date field.

Outcome measures

The primary outcome of this study was the total number of test cat- egories ordered by PIT provider type. Other outcomes included the fre- quency with which individual test categories were ordered by PIT provider type, total tests ordered by the end of ED stay, time in an ED bed, total ED time from patient arrival to ultimate disposition.

Statistical analysis

We summarized patient characteristics and study variables for pro- vider type using means and SDs, as well as frequencies and percentages. We compared these variables between groups using t-test and chi- square test for continuous and categorical measures, respectively. We then calculated 95% confidence intervals between groups using the Clopper-Pearson method [14]. We constructed negative binomial regres- sion models to evaluate predictors of the total number of test categories ordered for each patient. For each potential predictor (demographics, pa- tient history and arrival times/dates) we constructed a univariable model. Next, we included the full set of predictors in a multivariable model. For each model, we report the incidence rate ratios (IRRs) and with their corresponding 95% confidence intervals. Provider type was in- cluded to determine if the provider effect remained significant after ac- counting for the additional predictor variables. Next, we constructed logistic regression models for each individual test category. Separate models were created for urine test, blood test, and ultrasound but not for CT scan or X-ray as these tests were infrequently ordered. These models assessed the same set of predictors as the negative binomial models. Day of arrival was included as a seven-level categorical variable in the models, with Monday used as the referent category. Statistical analyses were performed using SAS V9.4 (Cary, NC). p-Values b 0.05 were considered statistically significant.

Results

NPs tended to see slightly older patients (42.8 years vs. 40.6 years) compared to physicians. NPs were also more likely to see patients dur- ing the day shift and during the regular work week (m-f) than physi- cians. Other patient characteristics were not significantly different between provider types (see Table 1).

Overall, physicians ordered more diagnostic test categories than NPs (1.75 vs. 1.54, p b 0.001). Physicians ordered significantly more Urine tests (77.8% vs. 71.4% p = 0.042) and ultrasounds (12.3% vs. 4.1%, p b 0.001) than NPs. Physicians also tended to order more blood tests (82.6% vs. 77.3%) and CTs (0.6% vs. 0%), but these differences were not

statistically significant (see Table 2).

Our univariable negative binomial model also demonstrated that physicians ordered more diagnostic test categories than NPs. The IRR was estimated to be 1.16 (1.03-1.29) (p = 0.011) and the result was es- sentially the same in the multivariable negative binomial model (IRR

1.15 (1.00-1.32)) (p = 0.049) (see Table 3a).

Next, we constructed a logistic regression model for each of the fol- lowing 3 test categories: urine, blood and ultrasound. After accounting for patient characteristics, the provider type (NP vs. physician) remained

Table 1

Summary of patient characteristics by PIT provider type.

Table 3a

Negative binomial regression for predicting how many test categories were ordered.

Patient characteristics	NP	Physicians	Diff (95% CI)		Characteristicsa	Uni. rate ratio (95% CI)	Mul. rate ratio (95% CI)
	(n = 419)	(n = 351)			PIT provider: Physician	1.16 (1.03, 1.29)	1.15 (1.00, 1.32)
Age	42.8 (13.7)	40.6 (13.5)	2.2 (0.2, 4.1)		Age	1.00 (0.96, 1.04)	1.00 (1.00, 1.01)
Male	138 (32.9%)	125 (35.6%)	-2.7% (-9.4%, 4.1%)		Male	0.82 (0.73, 0.93)	0.80 (0.71, 0.91)
Diabetes	46 (11.0%)	39 (11.1%)	-0.1% (-4.6%, 4.3%)		Diabetes	0.99 (0.83, 1.19)	0.97 (0.80, 1.17)
Hypertension	77 (18.4%)	62 (17.7%)	0.7% (-4.7%, 6.2%)		Cholecystectomy	1.18 (0.96, 1.44)	1.13 (0.92, 1.39)
Liver disease	12 (2.9%)	9 (2.6%)	0.3% (-2.0%, 2.6%)		Other surgery	0.99 (0.88, 1.27)	0.92 (0.81, 1.06)
Appendectomy	29 (6.9%)	16 (4.6%)	2.4% (-0.9%, 5.6%)		HTN	1.04 (0.90, 1.20)	1.02 (0.87, 1.20)
Cholecystectomy	32 (7.6%)	22 (6.3%)	1.4% (-2.2%, 5.0%)		Liver	1.12 (0.81, 1.55)	1.14 (0.82, 1.59)
Other abdominal surgery	119 (28.4%)	86 (24.5%)	3.9% (-2.3%, 10.1%)		Appendectomy	1.07 (0.85, 1.34)	1.09 (0.86, 1.37)
Arrival: 7 AM-7 PM	380 (90.7%)	161 (45.9%)	44.8% (38.9%, 50.7%)		Arrival: 7 AM-7 PM	0.87 (0.77, 0.98)	0.92 (0.80, 1.06)
Day of arrival a Results not shown for day of arrival.

statistically significant for urine tests and ultrasounds (p = 0.013 and p

Monday	81 (19.3%)	55 (15.7%)	3.7% (-1.7%, 9.0%)
Tuesday	87 (20.8%)	61 (17.4%)	3.4% (-2.2%, 8.9%)
Wednesday	80 (19.1%)	59 (16.8%)	2.3% (-3.1%, 7.7%)
Thursday	83 (19.8%)	40 (11.4%)	8.4% (3.4%, 13.5%)
Friday	82 (19.6%)	47 (13.4%)	6.2% (1.0%, 11.4%)
Saturday	6 (1.4%)	51 (14.5%)	-13.1% (-17.0%, -9.2%)
Sunday	0 (0%)	38 (10.8%)	-10.8% (-14.1%, -7.6%)

= 0.007, respectively) but not blood tests (see Table 3b).

By the end of their ED stay, including tests ordered in the main ED by their definitive provider teams, there was no statistically significant dif- ference in total test categories ordered between provider groups: physi- cian 2.67 vs. NP 2.53, p = 0.08, or using NB model, IRR 1.07 (0.98-1.17) (p = 0.118).

Patient time in an ED bed was not significantly different between NPs and physicians (NP 244 min, SD = 133, Physicians 248 min, SD

= 152), difference 4 min (-24.3-16.1). The total ED LOS from arrival at triage to ultimate disposition was shorter for NPs as compared to phy- sicians (NP 438 min, SD = 18, Physicians 487 min, SD = 213), differ- ence 49 min (20.9-76.9).

Discussion

In this paper we report the findings of a large retrospective analysis comparing test ordering between NPs and attending emergency physi- cians in the PIT role. We found that NPs ordered fewer diagnostic tests compared with attending emergency physicians. They ordered fewer test categories in aggregate, and specifically ordered fewer of each indi- vidual test category we examined. Despite this difference, by the end of the ED stay, there was no significant difference in total test categories ordered between groups. Thus, the total number of, for example, ultra- sounds, was similar between groups, even though fewer were ordered by NPs in triage. This suggests that some necessary studies that were not ordered by the PIT provider were subsequently ordered in the ED by the patient’s definitive provider, and that attending physicians were better able to “predict” what studies would be required by the de- finitive team in working up their patients. By the end of their ED stay, however, there was no significant difference in time spent in an ED bed based on screening provider type. Time in an ED bed is the most clinically relevant outcome in this study. Patients screened by a physi-

evening and night time, when time spent in the waiting room before placement in an ED bed was longer.

The goal of the PIT provider is to reduce length of stay by expediting appropriate testing during the previously underutilized waiting room time, so that faster Disposition decisions can be made when patients reach their definitive ED provider. There is good evidence that having a Provider in Triage when there is a shortage of ED beds does expedite care and improve throughput in busy emergency departments [6,15]. Our study builds on prior evidence to demonstrate that NPs function well as PIT [3,16].

Our research has some limitations. This was a retrospective analysis of a previously reported RCT. For this analysis, we looked at test order- ing by NPs and physicians among the subgroup of study participants for whom test ordering was allowed. Therefore, allocation of patients was not randomized. However, patient characteristics were similar be- tween provider types, except for a slight difference in age. This study only evaluated non-pregnant adult patients presenting with abdominal pain. It is possible that NP and physician test ordering would have been different for patients with other complaints. The range of complaints that present to the emergency department is very broad. We chose to include abdominal pain patients in this study because we felt that these patients tend to get more resource-intensive diagnostic workups, with broad differential diagnoses and a relatively wide range of possible diagnostic modalities, leaving ample room for discretion in what initial tests are most appropriate. Other complaints have a more limited reper- toire of possible diagnostic tests, which may influence PIT ordering pat- terns. Our measure of test ordering was somewhat rough and therefore limits our conclusions. We measured the number of test categories or- dered. So, for example, if a patient had a complete blood count ordered at either location (the waiting room or the ED), this was recorded as a “yes” for blood tests ordered at that location. A different patient may have had several different blood tests ordered (i.e., complete blood count, metabolic panel, and lipase) and this would identically be record- ed as “yes” for blood testing at that location.

In summary, in our busy county emergency department, nurse prac- titioners in the PIT role did not over-order diagnostic tests in compari- son with attending physicians. In fact, they ordered slightly less. This

Table 3b

Logistic regression for predicting test categories ordered.

cian did have a longer total ED LOS, which is best explained by the fact the physician RME screens were more likely to take place later in the

Diabetes	0.95 (0.53, 1.68)	0.82 (0.43, 1.57)	0.81 (0.29, 2.26)
Cholecystectomy	1.60 (0.71, 3.60)	2.70 (1.03, 7.05)	0.55 (0.16, 1.89)

Characteristics^a Urine test OR (95% CI)

Blood test OR (95% CI)

Ultrasound OR (95% CI)

Table 2	PIT: Physician Age	1.74 (1.12, 2.71) 0.99 (0.98, 1.01)	1.25 (0.79, 1.97) 1.01 (1.00, 1.03)	2.63 (1.30, 5.33) 0.99 (0.97, 1.02)
Frequency of test categories ordered by PIT provider type.	Male	0.28 (0.19, 0.40)	1.05 (0.70, 1.57)	0.26 (0.12, 0.55)

Type of test NP encounters

(n = 419)

				Other abd. surgery	0.85 (0.56, 1.30)	0.66 (0.44, 1.00)	1.04 (0.56, 1.95)
Urine	299 (71.4%)	273 (77.8%)	-6.4% (-12.6%, -0.3%)	HTN	0.74 (0.46, 1.19)	1.75 (0.98, 3.15)	0.75 (0.31, 1.78)
Blood	324 (77.3%)	290 (82.6%)	-5.3% (-10.9%, 0.3%)	Liver disease	1.40 (0.48, 4.12)	1.88 (0.42, 8.41)	2.32 (0.48, 11.17)
X-ray	5 (1.2%)	7 (2%)	-0.8% (-2.6%, 1%)	Appendectomy	1.25 (0.59, 2.63)	1.69 (0.69, 4.14)	1.35 (0.44, 4.15)
Ultrasound	17 (4.1%)	43 (12.3%)	-8.2% (-12.1%, -4.3%)	Arrival: 7 AM-7 PM	1.02 (0.65, 1.59)	0.64 (0.40, 1.04)	0.53 (0.28, 1.01)
CT 0 (0%) 2 (0.6%) -0.6% (-1.4%, 0.2%) a Results not shown for day of arrival.

Physician encounters (n = 351)

Diff (95% CI)

grant funding“>slight difference in practice between NPs and physicians did not affect the overall time spent in an ED bed. Therefore, with respect to the out- comes measured, NPs in the PIT role function similarly to attending phy- sicians. Given limited budgets and a shortage of ED physicians, using NPs as providers in triage is an effective strategy to address growing pa- tient care needs.

Prior presentations

None.

Grant funding

Statistical analyses for this research were supported by NIH/National Center for Advancing Translational Science (NCATS) UCLA CTSI Grant Number UL1TR000124.

Author contributions

Begaz: PI; study conceptualization and design; data acquisition; data analysis; Data interpretation; primary author of manuscript; final approval of version to be published; agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately inves- tigated and resolved.

Grogan: Data analysis; data interpretation; drafting the work and revising it critically for important intellectual content (primarily methods, analysis, results and discussion); final approval of version to be published; agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Elashoff: Data analysis; data interpretation; drafting the work and revising it critically for important intellectual content (primarily methods, analysis, results and discussion); final approval of version to be published; agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Talan: Study conceptualization and design; significant contribution to manuscript; final approval of version to be published; agreement to be accountable for all aspects of the work in ensuring that ques- tions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Taira: Study conceptualization and design; data analysis; data inter- pretation; significant contribution to manuscript; final approval of

version to be published; agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or in- tegrity of any part of the work are appropriately investigated and re- solved.

Acknowledgements

We would like to thank Neha Agarwal, Scott Lundberg, and Roee Sa- lem for their invaluable assistance on this project.

References

1. Brown DF, Sullivan AF, Espinola JA, Camargo Jr CA. Continued rise in the use of mid- level providers in US emergency departments, 1993-2009. Int J Emerg Med 2012; 5(1):21.
2. Ganapathy S, Zwemer Jr FL. Coping with a crowded ED: an expanded unique role for midlevel providers. Am J Emerg Med 2003;21(2):125-8 [0735-6757 (Print)].
3. Bahena D, Andreoni C. Provider in triage: is this a place for nurse practitioners? Adv Emerg Nurs J 2013;35(4):332-43 [1931-4493 (Electronic)].
4. Hing EBF. Wait time for treatment in hospital emergency departments: 2009. NCHS data brief; 2012 [accessed 12/02/2016].
5. Begaz T, Elashoff D, Grogan TR, Talan D, Taira BR. Initiating diagnostic studies on pa- tients with abdominal pain in the waiting room decreases time spent in an emer- gency department bed: a randomized controlled trial. Ann Emerg Med 2017; 69(3):298-307.
6. Rowe BH, Guo X, Villa-Roel C, Schull M, Holroyd B, Bullard M, Vandermeer B, Ospina M, Innes G. The role of triage liaison physicians on mitigating overcrowding in emer- gency departments: a systematic review. Acad Emerg Med 2011;18(2):111-20.
7. James MR, Pyrgos N. Nurse practitioners in the accident and emergency department. Arch Emerg Med 1989;6(4):241-6.
8. Hughes DR, Jiang M, Duszak Jr R. A comparison of diagnostic imaging ordering pat- terns between advanced practice clinicians and primary care physicians following office-based evaluation and management visits. JAMA Intern Med 2015;175(1): 101-7.
9. Hemani A, Rastegar DA, Hill C, al-Ibrahim MS. A comparison of resource utilization in nurse practitioners and physicians. Eff Clin Pract 1999;2:258-65 [1099-8128 (Print)].
10. Considine J, Martin R, Smit D, Jenkins J, Winter C. Defining the Scope of practice of the emergency nurse practitioner role in a metropolitan emergency department. Int J Nurs Pract 2006;12(4):205-13.
11. Lenz ER, Mundinger MO, Kane RL, Hopkins SC, Lin SX. Primary care outcomes in pa- tients treated by nurse practitioners or physicians: two-year follow-up. Med Care Res Rev 2004;61(3):332-51.
12. Liu H, Robbins M, Mehrotra A, Auerbach D, Robinson BE, Cromwell LF, Roblin DW. The impact of using mid-level providers in face-to-face primary care on health care utilization. Med Care 2017;55(1):12-8.
13. Wuerz RC, Milne LW, Eitel DR, Travers D, Gilboy N. Reliability and validity of a new

    five-level triage instrument. Acad Emerg Med 2000;7(3):236-42.

    Clopper CJPE. The use of confidence or fiducial limits illustrated in the case of the bi- nomial. Biometrika 1934;26(4):404-14.

14. Begaz T, Elashoff D, Grogan TR, Talan D, Taira BR. Initiating diagnostic studies on pa- tients with abdominal pain in the waiting room decreases time spent in an emer- gency department bed: a randomized controlled trial. Ann Emerg Med 2016.
15. Love RA, Murphy JA, Lietz TE, Jordan KS. The effectiveness of a provider in triage in the emergency department: a quality improvement initiative to improve patient flow. Adv Emerg Nurs J 2012;34(1):65-74 [1931-4493 (Electronic)].