a b s t r a c t

Background: Overcrowding in the Emergency Department is a problem with many strategies for intervention such as the Physician in triage . This brief evaluation is designed to minimize diagnostic uncertainty and ex- pedite the work up when the patient is seen in the Emergency Department. We hypothesized that this would in- crease CT imaging which would be increasingly negative as the pressure to maintain throughput rises on busy days in the Emergency Department.

Methods: We designed a retrospective study in which ordering patterns of Emergency physicians was explored using a group of patients with abdominal pain, presenting to triage in a 2 year period. We compared CT ordering rates on the 5% highest and lowest volume days (HD5 and LD5) and examined the bivariate relationship between volume and imaging utilization.

Results: There was no statistical significance in the rate of CT’s ordered collectively by PIT physicians on HD5 com- pared with LD5 with a p-value of 0.25. There is a trend toward increased utilization when each physician is com- pared to themselves on HD5 vs. LD5 but these were not statistically significant differences. The percentage of “clinically relevant” CTs was not determined to be different, but there was an increase in the LOS when a CT was ordered on both LD5 and HD5 (HD5 p-value 0.009; LD5 p-value 0.0004).

Conclusion: There is no difference in CT ordering patterns for abdominal pain by PIT between HD5 and LD5. Like- wise CT ordering patterns do not demonstrate a difference in percentage of clinically relevant CTs.

(C) 2017

Introduction

Overcrowding in the emergency department is a well-studied, inter- national phenomenon that is unfortunately increasing. As reported by Institute of Medicine in June 2006, while visits to the Emergency De- partment (ED) are increasing, the number of EDs and Hospital beds are decreasing, leading to extended lengths of stay in the emergency de- partment (also known as “boarding”) [1]. In 2008-2010 the mean wait time to see an Emergency Department physician increased from 45 min (1998-2000) to 55 min [2]. Overcrowding, increasing door-to-doctor (DTD) time, and length of stay (LOS) are concerning trends due to the association with decreased quality of care, higher morbidity and mortal- ity, and decreased Hospital reimbursement based on quality metrics [3, 4,5,6,7].

Specifically, length of stay has become an increasingly important

topic and focus for hospital administration as it correlates with mortal- ity as well as patient satisfaction [6,7,8]. Numerous factors have been cited and studied for possible intervention, including ED throughput. This refers to factors inherent to the visit itself, which increase length

* Corresponding author.

E-mail addresses: [email protected] (K. Matz), [email protected] (T. Britt), [email protected] (V. LaBond).

of stay. This represents opportunities for process improvement such as managing inadequate staffing, decreased ED beds, increasing acuity, and increase in use of imaging studies including computed tomography (CT) as well as increasing procedures [5].

One of the possible solutions to mitigate increased LOS has been the implementation of a physician in triage (PIT). This has proven to be able to reduce door-to-doc time, an important core measure tied to reim- bursement, as well as reduce length of stay in the emergency depart- ment [9-16]. The PIT provides an initial assessment of the chief complaint and cursory physical exam enabling the physician to effec- tively triage the patient as well as order lab work and imaging studies. This provides the treating physician with study results with the idea that a prolonged ED course may be shortened by resulted studies reduc- ing diagnostic uncertainty.

This provides an interesting dilemma with regard to imaging studies ordered by PIT as this abbreviated exam may lead to investigations with independent impact on LOS. Of specific scrutiny is the ED physicians’ in- creasing utilization of the CT scan and its correlation to longer LOS, as well as increase cost and radiation exposure [8,17]. Between the years 2000 and 2010 the utilization of advanced imaging studies, CT and MRI, increased by 3.1 times (from 5 to 17%) [2]. Eighteen percent of working age adults and 29% of ED visits by patients over 65 include ad- vanced imaging of some kind [2]. This increasing use not only increases

http://dx.doi.org/10.1016/j.ajem.2017.02.003

0735-6757/(C) 2017

K. Matz et al. / American Journal of Emergency Medicine 35 (2017) 974-977 975

cost, but also increases radiation exposure and cancer risk beyond what we estimate or may fully understand [18].

The ED offers immense diagnostic uncertainty, and ordering prac- tices for ED physicians are influenced by a variety of stressors and indi- vidual biases unique to the emergency medicine provider and the environment in which they work [19,20]. ED physicians endure a high cognitive load due to a continuous flow of patients being evaluated in parallel, often with incomplete information, and the responsibility to rule out high risk diagnosis while maintaining the safety and through- put of the department [20]. The speed, breadth, and volume of critical thinking with the added pressure of throughput requires the ED physi- cian to rely heavily on cognitive short cuts such as pattern recognition and heuristics [19,20,21]. And, while invaluable to the seasoned physi- cian, these tools lend themselves to associated cognitive errors [19-22]. Additionally, studies in cognitive psychology suggest that this high cognitive load actually shapes decision-making and may lead to a less systematic approach and more risk adverse behavior [20,23]. Over- crowded days in the emergency department can exert a tremendous stress on the ED physician as the pressure to maintain throughput rises [20]. A 2016 study performed by Gorski et al. found that volume of the ED waiting room positively affected admission rates [24]. With special consideration to the PIT, this abbreviated evaluation, may lend itself to increased dependence on quick judgments based on uncon- scious pattern recognition, which may impact ED ordering patterns.

Therefore, we designed a retrospective study to describe the CT or- dering patterns of the PIT with relation to the ED volume as an environ- mental stressor. We hypothesized that the pressure of increasing throughput by the PIT on the busiest days in the emergency department leads to an increase in utilization of CT and therefore an increase in the rate of negative CT results. Furthermore, we believe, with each physician acting as their own control, ordering patterns may change on a busy day in the ED from even the same physician. Physician in triage offers a unique opportunity to examine ordering patterns of individual physi- cians and the effect of volume and throughput pressure.

Methods

This study was approved by the hospital’s institutional review board on 2/9/15. It was conducted in the Emergency Department of a 410 bed, community hospital located in a suburban area servicing Mid-Michigan. The hospital is a level II trauma center and Stroke Center also housing multiple medical and surgical residencies and fellowships including a dually (ACGME and AOA) accredited Emergency Medicine Residency. The Emergency Department is a 47 bed ED with over 64 000 annual ED visits and a 28% admission rate. The Emergency Department features a Physician in Triage, which was implemented in July 9, 2012 and is ac- tive daily 12-10 pm.

The population to be studied was the emergency medicine physi- cians working in triage from January 1, 2013 to December 31, 2014. In order to define high and low volume shifts, ED census data was obtained from the ED quality dashboard and we defined a “high volume day” (HD5) as census being in the top 5% of the 2- year period and “low

Table 2

Charlson cormorbidity index scoring criteria.

Weight	Conditions	ICD-9 codes
1	Myocardial infarct	410, 411
	Congestive heart failure Peripheral vascular disease Dementia	398, 402, 428 440-447 290, 291, 294
	Cerebrovascular disease Chronic pulmonary disease Connective tissue disease Ulcer disease Mild liver disease	430-433, 435 491-493 710, 714, 725 531-534 571, 573
2	Hemiplegia Moderate or severe renal disease Diabetes	342, 434, 436, 437 403, 404, 580-586 250
	Any Tumor Leukemia Lymphoma	140-195 204-208 200, 202, 203
3	Moderate or severe liver disease	070, 570, 572
6	Metastatic solid tumor	196-199

volume day” (LD5) being in the bottom 5% of the 2-year period. We then used patient visit data from those days to determine whether the PIT physician ordered CTs, in order to evaluate their ordering patterns, as a group.

In order to standardize our study, we narrowed the presenting chief complaint to that of abdominal pain. In this study we included all pa- tients 18 to 100 years old presenting to the ED through triage with the chief complaint of abdominal pain during the hours of the PIT physician. Patients who arrived by EMS were excluded, as the triage physician does not evaluate them. Data points we collected for these patient en- counters and are listed in Table 1. The modification of the Charleston Co- morbidity Index is represented in Table 2 [25,26]. Patient’s who underwent a CT ordered by a physician other than the triage physician such as a resident, physician extender, or a main ED physician were counted as not having a CT ordered by the PIT physician.

Pearson Correlation coefficients were computed to examine the bi- variate relationship between ED volume and rate of CTs ordered in tri- age. In the initial power analysis, significance of p b 0.05 and power of at least 80% required a minimum of 300 cases for the sample size. We determined our study time period based on the power analysis and PIT volume.

Once this data was extracted, we compared the rate of CT ordering by PIT between LD5 and HD5. Additionally, CT results were obtained and coded as positive or negative in order to compare the rate of posi- tive CTs ordered by the PIT on LD5 and HD5. In order to do this, two sep- arate physicians individually evaluated the results and scored them as positive or negative based on “clinically relevant” findings. Each physi- cian determined this definition individually. Disagreement between the physicians was resolved by a third physician’s evaluation. CT posi- tive and negative rates were correlated with volume as secondary bivar- iate analysis.

We then coded each PIT physician who evaluated patients on both LD5 and HD5 with a letter (A-L) in order to compare the data with

Table 1

Primary study variables.

Table 3

Comparative findings between low and high volume groups.

Demographics Visit details

Gender	Visit chief complaint			Lowest 5%	Highest 5%	p-Value
Age at presentation Race	ED door time (date and time) Disposition time (date and time)		Volume cutoff Average ED volume	<= 145 patients 139.8 +- 4.6	>= 204 patients 216.5 +- 12.6	b0.001
Comorbidities by ICD-9 Codes			Age	42.9 +- 17.5 years	44.8 +- 17.1 years	0.28
included in Charleston Index ED Data	CT Details		Gender Race	33.1% male 66.9% female 88.7% White	30.6% male 69.4% female 87.5% White	0.61 0.72
ED Daily census Triage physician	CT type ordered Ordering physician		Charleston index score	9.3% Black 1.37 +- 2.2	11.2% Black 1.69 +- 2.8	0.24
	CT result		Average length of stay	240.0 +- 107.42 min	372.6 +- 153.1 min	b0.001

976 K. Matz et al. / American Journal of Emergency Medicine 35 (2017) 974-977

Table 4

Percentage of positive computed tomographies ordered by PIT.

Lowest 5% Highest 5% p-Value

Total positive CTs ordered by PIT Total CTs ordered by PIT

21 (51.2%)

41

36 (47.4%)

76

0.74

Table 6 reports on LOS in minutes for both LD5 and HD5. As might be expected, the HD5 group did have a longer LOS independently. After analysis of the patient’s length of stay was completed, we determined that those patients who received a CT did have increase LOS both on busy and slow days, even when ordered by the physician in triage (Table 7).

Fig. 1. CT utilization. CT rate among patients seen by PIT physicians for abdominal pain.

each physician acting as their own control. In this manner we were able to more closely compare individual physician ordering patterns in triage on LD5 compared to HD5 by determining their rate of CT ordering and rate of positive CT.

Lastly, we compared the length of stay (LOS) between patients who had a CT ordered by the PIT physician and those who had not. LOS was determined by the difference between the ED door time and the physi- cian Disposition decision time.

Results

From a review of PIT volumes, low-volume 5% days (LD5) was de- fined as 145 patients per day or less while high-volume 5% days (HD5) was defined as a total ED census of 204 patients or more. There were 383 patients who met these criteria, 151 presenting on LD5 and 232 presenting on HD5. There were no significant differences in demo- graphics between volume groups (Table 3). The Charleston Index, which is a predictor of inpatient death by representation of disease bur- den, of the groups LD5 and HD5 was also similar [25,26]. When PIT CT ordering patterns were compared between LD5 and HD5 there was no statistical difference in the rate of CT ordering (Fig. 1). In our analysis of the individual PIT physicians, the majority of these physicians did in- crease their use of this imaging modality on HD5, however there were some physicians who decreased utilization (C, E, J, L) and some who did not change ordering pattern at all (K). These differences were also not statistically different (Fig. 2). “Clinically relevant” abdominal CTs were found on 48.1% of studies ordered on LD5 as compared to 43.1% on HD5 (p = 0.49). Those ordered by PIT, similarly had no statistical dif- ference both collectively and individually (Tables 4 and 5).

Discussion

Our study was designed to provide insight regarding CT ordering patterns of ED PIT and the effect volume pressures have on these pat- terns. This is important because while PIT is a strategy to increase throughput, most specifically on high volume days, an increase in CT utilization would decrease throughput, increase cost, and increase radi- ation exposure. We predicted that the pressure of moving the depart- ment would increase the ordering of CT in efforts to provide decreased diagnostic uncertainty and would thereby increase the inci- dence of negative CT scans.

We utilized a specific group of patients to test our hypothesis. Abdominal pain represents 9% of all ED visits in patients over 18 and of- fers a vast differential with a high of diagnostic uncertainty [2]. Further- more, CT imaging in the ED associated with abdominal pain represents an even higher correlation with increased LOS than other CT imaging [15].

In this study, though CTs were ordered at a slightly higher rate on HD5 the difference was not statistically significant. This finding may indicate no association between volume and CT ordering or it may be that the volume cutoffs were not sufficient to discriminate be- tween the two groups. However, we identified the most disparate ends of our site’s volume days. To show that this slight difference in or- dering rates was significant would require huge volumes and such a small difference is not clinically significant enough to improve on costs and other resources.

Importantly, with increased awareness of CT appropriateness, there was no evidence to suggest we had less “clinically significant” positive results on HD5 to suggest relative overuse on HD5 at our facility. This has reassuring implications for resource utilization, as our hypothesis was that the truncated examination would lead to increased ordering of CTs and incidence of negative results.

Fig. 2. Individual PIT ordering patterns. CTs ordered/total patients seen for abdominal pain by PIT physicians on the 5% low/high volume days (LD5;HD5). *N values are noted at the base of the bar graph and are representative of the number of CTs ordered per physician.

K. Matz et al. / American Journal of Emergency Medicine 35 (2017) 974-977 977

Table 5

Percent positive computed tomographies ordered by group.

Lowest volume days Highest volume days

PIT provider	Number (%) positive	Total ordered		Number (%) positive	Total ordered	p-Value
A	3 (75%)	4		7 (100%)	7	NC
B	3 (75%)	4		3 (33.3%)	9	NC
C	3 (60%)	5		5 (35.7%)	14	0.36
D	1 (100%)	1		1 (100%)	1	NC
E	4 (40%)	10		3 (33.3%)	9	0.77
F	1 (100%)	1		3 (60%)	5	NC
G	1 (33.3%)	3		8 (66.7%)	12	NC
H	-	0		7 (0%)	7	NC
I	1 (100%)	1		-	0	NC
J	0 (0%)	2		-	0	NC
K	2 (66.7%)	3		2 (40%)	5	NC
L	2 (66.7%)	3		-	0	NC

n is reported as number of CTs ordered on triaged patients with abdominal pain when seen by a particular physician in triage on LD5 and HD5 days. Percent is reported as per- centage of “clinically relevant” CTs of reported ordered CTs. NC = Not Calculated.

Table 6

Average length of stay by group.

Lowest 5% Highest 5% p-Value

Conflict of interest/disclosure of funding

This research did not receive any specific grant from funding agen- cies in the public, commercial, or not-for-profit sectors.

References

1. Committee on the Future of Emergency Care in the United States Health System. Hospital-Based Emergency Care: At the Breaking Point. Washington, DC: National Academic Press; 2006.
2. NCHS. National Center for Health Statistics. Health, United States. With a Special Fea- ture on Emergency Care. Hyattsville, MD. http://www.cdc.gov.proxy2.cl.msu.edu/ nchs/data/hus/hus12.pdf; 2012. (Accessed 09.19.2016).
3. Richardson DB. Increase in patient mortality at 10 days associated with emergency department overcrowding. Med J Aust 2006;184:213-6.
4. Pines JM, Hollander JE, Localio AR, Metlay JP. The association between emergency department crowding and hospital preformance on antibiotic timing for pneumonia and percutaneous intervention for myocardial infarction. Acad Emerg Med 2006;13: 873-8. http://dx.doi.org/10.1197/j.aem.2006.03.568.
5. Hoot NR, Aronsky D. Systematic review of emergency department crowding: causes, effects, and solutions. Ann Emerg Med 2008;52(2):126-36. http://dx.doi.org/10. 1016/j.annemergmed.2008.03.014.
6. Bernstein SL, Aronsky D, Duseja R, Epstein S, Handel D, Hwang U, et al. The effect of emergency department crowding on clinically oriented outcomes. Acad Emerg Med 2009;16(1):1-10. http://dx.doi.org/10.1111/j.1553-2712.2008.00295.x.
7. Pines JM, Pollack CV, Diercks DB, Chang AM, Shofer FS, Hollander JE. The association between emergency department crowding and adverse cardiovascular outcomes in patients with chest pain. Acad Emerg Med 2009;16(7):617-25. http://dx.doi.org/10.

   LOS 240.0 +- 107.4 min

   n = 131

   372.6 +- 153.1 min

   n = 197

   b 0.0001

   1111/j.1553-2712.2009.00456.x.

   Gardner RL, Sarkar U, Maselli JH, Gonzales R. Factors associated with longer ED lengths of stay. Am J Emerg Med 2007;25(6):643-50. http://dx.doi.org/10.1016/j. ajem.2006.11.037.

8. Imperato J, Morris DS, Binder D, Fischer C, Patrick J, Sanchez LD, et al. Physician in

   Though not supported by the overall data, individual physician or- dering patterns in relation to volume were variable. A majority of phy- sicians ordered slightly more CTs on HD5 compared to LD5, however the difference was not significant. This trend may suggest that high-vol- ume days affect CT ordering at the physician level but our study was not powered to demonstrate this or there may be no association at the indi- vidual level. Larger studies are needed to explore this potential relation- ship and, if any, whether physician characteristics play a role.

   Lastly, though this study implies volume is not impacting the PIT’s col- lective ordering patterns, it is essential to note, that undergoing a CT or- dered by the physician in triage was associated with increased length of stay. There was fallout of 55 of the total 383 patients, who did not have a disposition time available for analysis, representing 14% of the popula- tion, which is an important limitation. However, mean LOS increased by

   65.69 min on LD5 and 57.25 min on HD5. Multiple variables influence ED throughput and length of stay. While employing the strategy of PIT re- duces LOS and ordering patterns between variable volumes appear simi- lar, CT imaging independently increases LOS. This highlights competing strategies between decreasing diagnostic uncertainty, thereby expediting the ED workup and ordering tests, which extend it.

   In conclusion, it does not appear that high volume days when com- pared with low volume days change ordering patterns of the physician in triage. While this is reassuring, there is more opportunity to explore decision making in the individual triage physician as it has defined im- pact on throughput. LOS will continue to be an opportunity to improve ED metrics and patient safety, and therefore careful consideration of time-consuming tests must continue to be part of an emergency evalu- ation in triage.

   Table 7

   Average Length of Stay based on CT Ordering by Group.

   CT Ordered Lowest 5% Highest 5% p-value No LOS

   triage improves emergency department Patient throughput. Intern Emerg Med 2012;7:457. http://dx.doi.org/10.1007/s11739-012-0839-0.

   Han JH, France DJ, Levin SR, Jones ID, Storrow AB, Aronsky D. The effect of physician triage on emergency department length of stay. J Emerg Med 2010;39(2):227-33. http://dx.doi.org/10.1016/j.jemermed.2008.10.006.

9. Partovi SN, Nelson BK, Bryan ED, Walsh MJ. Faculty triage shortens emergency de- partment length of stay. Acad Emerg Med 2001;8:990-5. http://dx.doi.org/10. 1111/j.1553-2712.2001.tb01099.x.
10. Holroyd BR, Bullard MJ, Latoszek K, Gordon D, Allen S, Tam S, et al. Impact of a triage liaison physician on emergency department overcrowding and throughput: a ran- domized controlled trial. Acad Emerg Med 2007;14:702-8. http://dx.doi.org/10. 1197/j.aem.2007.04.018.
11. Terris J, Leman P, O’Connor N. Making an IMPACT on emergency Department flow: improving patient processing assisted by consultant in triage. Emerg Med J 2004; 21:537-41. http://dx.doi.org/10.1136/emj.2002.003913.
12. Subash F, Dunn F, McNicholl B. Team triage improves emergency department effi- ciency. Emerg Med J 2004;21:542-4. http://dx.doi.org/10.1136/emj.2002.003665.
13. Choi YF, Wong TW, Lau CC. Triage rapid initial assessment by doctor (TRIAD) im- proves waiting time and processing time of the emergency department. Emerg Med J 2006;23:262-5. http://dx.doi.org/10.1136/emj.2005.025254.
14. Rowe BH, Guo X, Villa-Roel C, Schull M, Holroyd B, Bullard M, et al. The role of triage liaison physicians on mitigating overcrowding in emergency departments: a sys- tematic review. Acad Emerg Med 2001;18(2):111-20. http://dx.doi.org/10.1111/j. 1553-2712.2010.00984.x.
15. Kanzaria HK, Probst MA, Ponce NA, Hsai RY. The association between advanced diag- nostic imaging and ED LOS. Am J Emerg Med 2014;32(10):1253-8. http://dx.doi. org/10.1016/j.ajem.2014.07.038.
16. Brenner DJ, Hall EJ. Computed tomography-an increasing source of radiation expo- sure. N Engl J Med 2007;357:2277-84. http://dx.doi.org/10.1056/NEJMra072149.
17. Kovacs G, Croskerry P. Clinical decision making: an emergency medicine perspec- tive. Acad Emerg Med 1999;6(9):947-52. http://dx.doi.org/10.1111/j.1553-2712. 1999.tb01246.x.
18. Pines JM. What cognitive psychology tells us about ED physician decision-making and how to improve it. Acad Emerg Med 2016 Oct 5. http://dx.doi.org/10.1111/ acem.13110 [epub ahead of print].
19. Pines JM. Profiles in patient safety: confirmfation bias in emergency medicine. Acad Emerg Med 2006;13:90-4. http://dx.doi.org/10.1197/j.aem.2005.07.028.
20. Groopman J. How Doctors Think. Boston: Houghton Mifflin Company; 2007.
21. Deck C, Jahedi S. The effect of cognitive load on economic decision making: a survey and new experiments. Eur Econ Review 2015;78:97-119 (Doi: 1016/j.euroecorev. 2015.05.004).
22. Gorski JK, Batt RJ, Otles E, Shah MN, Hamedani AG, Patterson BW. The impact of emergency department census and the decision to admit. Acad Emerg Med 2016 Sept 19. http://dx.doi.org/10.1111/acem.13103 (Epub ahead of print).
23. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prog-

    Yes LOS

    204.3+-96.1 minutes n=60

    270.2+-107.8 minutes n=71

    342.4+-165.6 minutes n=93

    399.6+-136.2 minutes n=104

    0.0001

    0.0001

    nostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40(5):373-83. http://dx.doi.org/10.1016/0021-9681(87)90171-8.

    D’Hoore W, Bouckaert A, Tilquin C. Practical Considerations on the use of the Charleston Comorbidity Index with Administrative Data Bases. J Clin Epidemiol 1996;49(12):1429-33http://dx.doi.org/10.1016/S0895-4356.

    p-value 0.0004 0.009