a b s t r a c t

Background: Physician in triage (PIT) has been used as a potential solution to emergency department (ED) over- crowding and to decrease ED length of stay . This study examined the relationship between Computerized tomography utilization of PIT and ED patient volumes. We hypothesized that despite the pressure on PIT to improve throughput on the busiest days, they will continue to utilize CT at the same rate.

Methods: This retrospective chart review evaluated CT ordering patterns of PIT on patients with abdominal pain who presented to the ED over a 6-year period. CT utilization rate was calculated on days with the lowest 5% (LD5) and highest 5% (HD5) volumes based on average yearly volume. CT positive and negative rates were correlated with volume using Chi square analysis. Odds ratio and confidence intervals were calculated for the magnitude of effect difference.

Results: We found no statistically significant difference in CT utilization rate on HD5 vs LD5 (p = 0.833). There was a statistically significant increase in the rate of negative CT scans on HD5 (p = 0.046) which represented a 17% relative difference. LOS was longer on HD5 (p = 0.013) and when a CT scan was ordered (p b 0.001).

Conclusion: No difference was found in the rate at which the PIT ordered CT scans on high volume vs low volume days. The rate of CT scans without clinically relevant findings did increase slightly on high volume days. LOS was longer on high volume days and when a CT was ordered.

(C) 2020 The Authors. This is an open access article under the CC BY-NC-ND license (http://

creativecommons.org/licenses/by-nc-nd/4.0/).

Introduction

With the introduction of systems for early physician assessment at triage, waiting and processing times have been shown to decrease [1]. Additionally, they reduce door-to-doctor times (an important core mea- sure) and also appear to decrease length of stay in the ED. [2-9] The phy- sician in triage system (PIT) is of great benefit since the number of patients that present to EDs has continued to increase at a faster rate than number of ED and Hospital beds have increased in most industrial- ized nations [10,11]. IncreasED length of stay influences patient centered outcomes and has been shown to lead to increased mortality, decreased access to care due to patient elopement or ambulance diver- sion, and an overall decreased quality of care [12-16]. Length of stay has been one focus of hospital administration since it has been shown to correlate with mortality and patient satisfaction [3,13,16,17].

However, of specific concern and scrutiny is the potential for CT scan “over ordering” by the PIT. CT scan ordering correlates positively and in- dependently to a longer length of stay for patients [1,3,9]. There have been many studies examining the CT Ordering patterns of emergency

* Corresponding author.

E-mail address: [email protected] (V. LaBond).

physicians (EPs) focusing on a variety of variables thought to influence the decision to order a CT scan [3,10,17,18]. Busy, overcrowded days in the ED can exert a tremendous stress on the ED physician as the pres- sure to maintain throughput rises [17-19]. We hypothesized that de- spite this increased pressure on the physician in triage, they would continue to utilize CT at the same rate.

The current study was an extension to a previously performed study at our institution by Matz, Britt, and LaBond to assess CT ordering pat- terns of the physician in triage for abdominal pain [10]. That study ob- served a difference in the relative number of CT scans that were ordered when comparing the lowest 5% census days and the highest 5% census days but it was not sufficiently powered to show whether this dif- ference was statistically significant. The current study was performed with a larger sample size in order to more clearly evaluate the effect of high-volume days on CT ordering patterns by the physician in triage.

Methods

This study was approved by the hospital’s institutional review board on June 13, 2019. The study was conducted at a 410-bed community hospital located in a suburban area servicing the Mid-Michigan region. The hospital is a level II trauma center, STEMI center and stroke center

https://doi.org/10.1016/j.ajem.2020.04.085

0735-6757/(C) 2020 The Authors. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

12 M. Ullrich et al. / American Journal of Emergency Medicine 39 (2021) 11–14

that is host to multiple medical and surgical residency and fellowship programs including an ACGME accredited emergency medicine resi- dency. The ED has 47 beds with over 64,000 annual visits and an admis- sion rate of approximately 28%. The ED features a physician in triage daily from 12 pm to 10 pm.

The population to be studied was the emergency medicine physi- cians working in the triage area from May 1, 2014 to April 30, 2019. Daily census data for the ED was obtained from the ED quality dash- board and was used to determine our definition of high volume and low volume days for the 6-year study period. We defined a “high vol- ume day” (HD5) as a daily census in the 95th percentile or higher and a “low volume day” (LD5) as a daily census in the 5th percentile or lower, percentiles were calculated based on the average yearly volume for each calendar year in the study period. The patient data from the top 5% and bottom 5% days from each year during the study period were used to evaluate the CT ordering patterns of the physicians in tri- age as a group. We chose to use dates from each calendar year instead of the whole study period to minimize period effect.

To standardize the study, we narrowed the patients included to those with a chief complaint or CT indication that would fall under the ICD-10 parent code R10, abdominal and pelvic pain. We included all pa- tients 18 years old or greater who were seen by a triage physician with a chief complaint that would correspond to one of the R10 ICD-10 codes. Patients that were seen by a PIT physician but had a CT ordered by a pro- vider other than the PIT physician themselves were excluded.

Data extracted for patients who met the above inclusion criteria can be found in Table 1. Data extraction was performed by the principal in- vestigator and accuracy was checked randomly by the statistician. We calculated the rate at which CT scans had been ordered on the high and low volume days. If a patient had a CT ordered, the result was coded as either positive or negative in order to determine the rate of positive CT scans in both the high and low volume groups. A CT result was considered to be a positive result if it showed a finding that was considered to be clinically relevant. We defined clinically relevant as any CT result that met one or more of the following criteria: (1) an acute process was found that resulted in a Change in management (i.e. new medications given, STAT additional work up that revealed an acute process requiring immediate treatment or hospital admission),

(2) a process found that led to the patient being admitted, (3) an acute finding that could explain the reason the patient was having ab- dominal pain, (4) equivocal finding indicated by radiology that corre- lated with the chief complaint.

The determination of clinically relevant results required agreement between two EPs who reviewed all CT results. The reviewers were aware of the study purpose but blinded to the group designation of each patient. A “tiebreaker” determination was sought from a third in- vestigator if there was disagreement between the original reviewers.

The rate at which CT scans were ordered and the CT positive/nega- tive rates were then tested for association with ED volume using Chi square analysis. Odds Ratio and confidence intervals were then calcu- lated for the magnitude of effect difference. A total of 1000 patients were required to determine if a relative difference between groups of 16% was statistically significant with a power of N90% at p = .05. Length of stay (LOS) was also compared between patients receiving a CT scan

and those who did not. Length of stay was calculated using the ED door time and the physician Disposition decision time.

Results

Twenty-nine different attending physicians worked as physician in triage during our 6- year study period. Two of them worked on only the high-volume days and one worked on only the low volume days. The remaining twenty-six attending physicians worked on both high and low volume days.

The total number of patients in the study was 995 (N = 995). Table 2 shows patient age, gender, race, and length of stay. There were no signif- icant differences in age, gender, or race noted. The overall length of stay (LOS) for all patients was 461.8 min. Low-volume days had a mean LOS of 379.8 (SD: 672.6) and high-volume days had a mean LOS of 500.6 (SD:740.0), a relative 24% difference (p = 0.01). Length of stay was found to be longer on HD5 (500.6 min, SD:740.0) compared to LD5 (379.8 min, SD:672.6).

A CT scan was performed on 530 patients (53.3%) and of these, 275 (27.6%) showed clinically relevant findings (Table 3). Fig. 1 shows the difference found in the rate at which CT scans were ordered when com- paring HD5 (n = 358, 53.8%) to LD5 (n = 172, 53.0%). This difference in CT utilization rate was not statistically significant (p = 0.83).

As shown in Table 3, the percentage of CT scans without clinically relevant findings did differ for patients presenting on HD5 (n = 183, 27.1%) compared to those presenting on LD5 (n = 72, 22.5%). This find- ing represents a 17% relative difference between HD5 and LD5 and was found to be statistically significant (p = 0.046). When observing LOS, it was found that LOS was longer in patients who received a CT scan (542.8 min, SD: 915.5) compared to those who did not (369.4 min, SD: 376.3) (p b 0.001).

Discussion

The intention of our study was to examine ordering patterns of the physician in triage to attempt to determine if ED volumes had an effect on the rate of CT utilization. Physician in triage was intended to decrease ED length of stay by decreasing the door to doctor time, allowing the work up of the patient to begin sooner than it would if there were no phy- sician in triage. This would be most important when the ED was busy since it has been shown that length of stay increases as ED volume in- creases [1]. Previous studies have also demonstrated that on average, a patient that needs to wait for a CT abdomen/pelvis will have a longer length of stay [1,18]. This was confirmed in our study as well, any patient who had a CT performed had a significantly longer length of stay on aver- age compared to the patients that did not have a CT ordered (p b 0.001). Thus, if physician in triage was found to lead to an increased CT utilization rate then this could potentially negate the intended benefits.

Our prediction was that the physician in triage would utilize the CT

scanner at the same rate regardless of ED volume and that this would lead to no change in the rate of negative CT scans. Our results indicated that the small difference (LD5 53.0% vs HD5 53.8%) in CT utilization

Table 2

Comparative findings between low and high volume groups.

Table 1				Lowest 5% days	Highest 5% days	p-value
Primary study variables.				n = 320 (%)	n = 675 (%)
Demographics	Visit details		Age (mean, SD)	43.8 (20.4)	42.8 (19.8)	0.94
Gender Age at presentation	Visit chief complaint ED door date/time		Gender (n, %) Male	119 (37.2)	240 (35.6)	0.62
Race Descriptives	Disposition date/time CT details		Female Race (n, %) White	201 (62.8) 286 (89.4)	435 (64.4) 611 (90.5)	0.58
ED data	CT type ordered		Black/African American	30 (9.4)	53 (7.9)
ED daily census	Ordering physician		Length of stay (mean, SD)	379.8 (672.6)	500.6 (740.0)	0.01
Triage physician	CT result		Rate of CT utilization	172 (53.8)	358 (53.0)	0.83

M. Ullrich et al. / American Journal of Emergency Medicine 39 (2021) 11–14 13

Table 3

Results of computed tomography scans Ordered by PIT

Lowest 5% (LD5)

n = 172 n (%)

Highest 5% (HD5)

n = 358 n (%)

p-value

that there is no difference in CT utilization based on volume but that the true determinant of changes in utilization is actually patient acuity. We also did not examine the physicians in triage individually, this leaves the possibility that there are differences in CT utilization rate re- lated to the behavior patterns of individual physicians but that averag-

Total negative CTs ordered by PIT 72 (41.8) 183 (51.1) 0.05

Total positive CTs ordered by PIT 100 (58.2) 175 (48.9)

when comparing LD5 to HD5 was not statistically significant (p = 0.833) which seems to confirm our hypothesis that utilization rates would not change based on volume. The rate of negative findings among CT scans ordered was lower in the low-volume days (42%) then in the high-volume days (51%) and the difference was statistically significant at p = 0.05. This finding deserves further exploration in fu- ture studies. We had speculated that an increase in the rate of negative CT scans on HD5 might indicate that the physician in triage was order- ing CT scans that were potentially unnecessary. Making the assumption that a CT scan was not necessary because it did not show any clinically relevant findings is an example of hindsight bias. For example, a 20 year old male with anorexia, fever, and leukocytosis who presented to the ED with periumbilical pain that has now migrated to the right lower quadrant would prompt almost every physician to order a CT ab- domen/pelvis with contrast to evaluate for appendicitis. The result of the CT scan being negative would not mean it was any less necessary based on the high pretest probability of the patient having appendicitis. There are many other potential explanations for the increased number of negative CT scans found on busy days. A difference in average patient acuity related to ED volumes is one of the possible confounding vari- ables that could explain the increase in negative CT scans on busy days. This leaves opportunity for future study that could try to correlate our findings with patient acuity on busy compared to less busy days.

A strength of our study included controlling for a period effect. We had initially planned to pull our dates of interest based on the average volumes for the entire data set, but this resulted in a significant period effect. The majority of high-volume days would have come from 2014 to 2015 and the majority of low volume days from 2017 to 2018. Our so- lution was to choose our dates of interest based on the average volumes for each calendar year and this resulted in a much more equal distribu- tion of high and low volume days in any given year.

As described above, one limitation of our study is that we did not correlate acuity with volume and rate of CT utilization. It is possible

ing all providers together masked this difference. Matz et al.¹ did demonstrate a difference in CT utilization rate when comparing individ- ual physicians in triage to their peers. If we assume there really is a dif- ference in CT utilization rate among individual physicians, then which physician was working on which day would potentially have a major ef- fect on the CT utilization rate regardless of volume. For example, if the physicians less likely to use CT scans were also more likely to be the ones working during the high-volume days then this decrease in utiliza- tion rate based on individual physician behaviors could mask any po- tential difference due to another reason.

Of interest, the rate of CTs ordered in both groups was N50% (54 vs 53%). This finding seems unusually high, but we must consider that this data has been taken from only 10% (5% lowest and 5% highest vol- umes) of our yearly census. Therefore, we cannot determine our overall rate of CT ordering for patients with abdominal pain.

In conclusion, increased ED volumes do not appear to significantly change the rate at which the physician in triage utilizes CT scans for pa- tients with abdominal pain. There are several potential confounding variables that could explain the increased rate of negative CT scans on high volume days other than an increase in the ordering of unnecessary CTs. Our findings leave opportunity to further examine physician in tri- age ordering patterns while accounting for additional potential vari- ables such as patient acuity.

Credit author statement

Matthew Ullrich, DO, MHSA: Data Curation, Formal Analysis, Investi- gation, Methodology, Validation, Visualization, Original Draft, Review & Editing,

Virginia LaBond, MD, MS, FACEP: Conceptualization, Data Curation, Formal Analysis, Investigation, Methodology, Supervision, Validation, Review & Editing,

Todd Britt, DO: Conceptualization, Investigation, Methodology, Su- pervision, Validation, Review & Editing,

Kaitlyn Bishop, DO: Conceptualization, Data Curation, Investigation, Methodology, Validation, Review & Editing,

Total Patients Seen

Fig. 1. CT utilization rate for patients with abdominal pain.

ED Volume

Pts without CTs Ordered Pts with CTs Ordered

Highest 5% days (n=675)

Lowest 5% days (n=320)

0

148 (46%)

100

317 (47%)

400

300

200 172 (54%)

358 (53%)

600

500

P=0.833

800

700

14 M. Ullrich et al. / American Journal of Emergency Medicine 39 (2021) 11–14

Kimberly Barber PhD: Conceptualization, Formal Analysis, Investiga- tion, Methodology, Project Administration, Resources, Supervision, Val- idation, Visualization, Review & Editing,

References

1. 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. https://doi.org/10.1136/emj.2005.025254.
2. 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. https://doi.org/10. 1111/j.1553-2712.2009.00456.x.
3. 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. https://doi.org/10.1016/j. Ajem.2006.11.037.
4. Imperato J, Morris DS, Binder D, Fischer C, Patrick J, Sanchez LD, et al. Physician in triage improves emergency department Patient throughput. Intern Emerg Med 2012; 7:457. doi:10/1007/s11739-012-0839-0.
5. 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. https://doi.org/10.1016/j.jemermed.2008.10.006.
6. Partovi SN, Nelson BK, Bryan ED, Walsh MJ. Faculty triage shortens emergency de- partment length of stay. Acad Emerg Med. 2001;8:990-5. https://doi.org/10.1111/ j.1553-2712.2001.tb01099.x.
7. 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. https://doi.org/10. 1197/j.aem.2007.04.018.
8. 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. https://doi.org/10.1136/emj.2002.003913.
9. Subash F, Dunn F, McNicholl B. Team triage improves emergency department effi- ciency. Emerg Med J. 2004;21:542-4. https://doi.org/10.1136/emj.2002.003665.
10. Matz K, et al. CT ordering patterns for abdominal pain by physician in triage. American Journal of Emergency Medicine. 2017. https://doi.org/10.1016/j.ajem. 2017.02.003 [Accessed May 2, 2019].
11. Wargon M, Casalino E, Guidet B. From model to forecasting: a multicenter study in emergency departments. Acad Emerg Med. 2019;17:970-8. https://doi.org/10. 1111/j.1553-2712.2010.00847.x.
12. NCHS. National Center for Health Statistics. Health, United States, 2017. With a spe- cial feature on mortality. Hyattsville, MD. . https://www.cdc.gov/nchs/data/hus/hus1

    7.pdf; 2017. (Accessed 09.19.2016).

    Singer AJ, Thode Jr HC, Viccellio P, Pines JM. The association between length of emer- gency department boarding and mortality. Acad Emerg Med. 2011;18:1324-9. https://doi.org/10.1111/j.1553-2712.2011.01236.x.

13. Pines JM, Hollander JE, Localio AR, Metlay JP. The association between emergency department crowding and hospital performance on antibiotic timing for pneumonia and percutaneous intervention for myocardial infarction. Acad Emerg Med. 2006; 13:873-8. https://doi.org/10.1197/j.aem.2006.03.568.
14. Hoot NR, Aronsky D. Systematic review of emergency department crowding: causes, effects, and solutions. Ann Emerg Med 2008;52(2):126-36. doi:10/1016/j. annemergmed.2008.03.014.
15. 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. https://doi.org/10.1111/j.1553-2712.2008. 00295.x.
16. 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. https://doi.org/ 10.1016/j.ajem.2014.07.038.
17. Brenner DJ, Hall EJ. Computed tomography–an increasing source of radiation expo- sure. N Engl J Med. 2007;357:2277-84. https://doi.org/10.1056/NEJMra072149.
18. Kovacs G, Croskerry P. Clinical decision making: an emergency medicine perspec- tive. Acad Emerg Med. 1999;6(9):947-52. https://doi.org/10.1111/j.1553-2712. 1999.tb01246.x.