Impact of trauma activation on the ED length of stay for nontraumatic patients

Unlabelled imagetrauma activation on the ED le”>American Journal of Emergency Medicine (2012) 30, 311-316

Original Contribution

Impact of trauma activation on the ED length of stay for nontraumatic patients

Rajiv Arya MD a,?, Frank Dossantos DO b, Pamela Ohman-Strickland PhD c,

Mark A. Merlin DO a

aDepartment of Emergency Medicine, UMDNJ Robert Wood Johnson Medical School, New Brunswick, NJ 08901, USA

bUnited States Navy, USA

cDepartment of Biostatistics, UMDNJ-SPH, New Brunswick, NJ, USA

Received 24 August 2010; revised 3 December 2010; accepted 8 December 2010


Introduction: Trauma activation prioritizes Hospital resources for the assessment and treatment of trauma patient over all patients in the emergency department (ED). We hypothesized that length of stay (LOS) is longer for nontrauma patients during a trauma activation.

Methods: A retrospective, case-control chart review was conducted in a level I trauma center. Cases consist of patients who present 1 hour before and after the presentation of the trauma activation. Controls were patients presenting to the ED during the same period exactly 1 week before and after the cases. Confounding variables measured included sex, age, arrivals, and census for the 3 areas.

Results: Two hundred ninety-four trauma events occurred from January 1 until September 30, 2009. A significant difference was found between LOS of patients seen during a trauma activation with an average increase of 10.7 minutes in LOS (P =.0082; 95% confidence interval [CI], 2.8-18.7). This difference is attributable to the middle acuity area of the ED, in which the average increase in LOS was

20.3 minutes (P = .0004; 95% CI, 9.1-31.5). Significant LOS difference was not found when a trauma activation had an LOS of less than 60 minutes (P = .30; 95% CI, -7.1-61.7 for trauma LOS b60 minutes vs P = .02; 95% CI, 1.6-18.0 for trauma LOS >=60 minutes).

Conclusion: This retrospective case-control chart review identified an increase in ED LOS for patient presenting during trauma activations. Resource prioritization should be accounted for during times when these critical patients enter the ED.

(C) 2012



Over the past decade, the federal government has increas- ingly scrutinized the emergency department’s (ED) manage- ment of specific diagnosis. This has resulted in time-sensitive,

* Corresponding author. Tel.: +1 732 235 8717.

E-mail address: [email protected] (R. Arya).

resource-intensive guidelines and protocols. Patients who pre- sent to an ED with ST-elevation myocardial infarction, stroke, community-aquired pneumonia, and sepsis are all prioritized with these new time-sensitive protocols. The impact of trauma activation protocols on all ED patients has not been studied.


The effect of trauma activation on the care and management of ED patients with acute coronary syndrome

0735-6757/$ – see front matter (C) 2012 doi:10.1016/j.ajem.2010.12.011

or acute cerebral Vascular injury has been examined in 2 previous studies [1,2]. The impact of trauma activations on length of stay (LOS) for all patients has been notably absent from the literature.

Goals of the investigation

We hypothesized that LOS will be longer for general ED patients presenting during a trauma activation than those who present when no trauma activation is present.


This was a retrospective case-control, nonrandomized, single-center comparative study. The university institutional review board approved the protocol.


This study was conducted in the adult ED at a university- based academic medical center treating 65,000 adult patients per year. The ED is an urban level I trauma center and a tertiary care center. Only board-certified or board-eligible emergency physicians evaluated and treated patients. The ED is divided into 3 different clinical areas based upon a patient’s acuity: level 1 (nonurgent), level 2 (urgent care), and level 3 (Emergent care) [3]. Patients with the greatest acuity (level 3) are triaged to the emergent area of the ED and those not as acutely ill to the urgent and nonurgent areas of the ED, respectively. The ED has 70 hours of physician coverage, 48 hours of physician assistant (PA) coverage, and 40 hours of scribe coverage a day with no emergency medicine residents during the study. A scribe assists physicians with the clerical aspects of their job such as transcribing the notes and following up on diagnostic results. Physician assistants provide coverage in the emergent and urgent care areas of the ED, and all patients seen by a PA were evaluated by physician during the study. Physician assistants do not provide coverage for the nonurgent area. Physician assistants evaluate approximately 50% of the patients who present to the ED. No patients are triaged to the waiting room.

Trauma activation prioritizes a significant amount of shared resources for the trauma patient. The trauma team consists of an attending trauma surgeon, 2 trauma residents, an orthopedic surgery resident, an anesthesia resident and/or attending, and an emergency medicine attending depending on need. Nursing resources that are required are a primary nurse, a secondary nurse or medic, and a scribe nurse. Radiology technicians conduct initial imaging, and respira- tory therapy routinely responds to every alert. A dedicatED pharmacist holds response to the trauma bay, the department of radiology holds a computed tomographic (CT) scanner open, and the operating room holds a room until cleared by the trauma team. The blood bank sends a representative to obtain a confirmatory blood sample.

Selection of participants

Data from the department electronic medical records (EMR) ED information management systems (Livingston, NJ) were exported to MS Access for analysis. trauma activation patients were identified by querying the keyword of “trauma alert.” Adult ED patients were identified as those presenting within 1 hour before and after a trauma activation. control patients were those that presented within the identical 2-hour time window but presented 1 week before and after the trauma activation. Exactly 1 week before or after was chosen to eliminate any variance in LOS explained by time of week and day as well as to ensure enough separation from the trauma event that there were no residual effects of the trauma activation. In addition, including controls both before and after the trauma event minimized the possibility of seasonal or time-sequence-related biases [4]. All nontrau- matic patients older than 21 years were included in the study. Patients with a primary psychiatric diagnosis were excluded because of routinely excessive LOSs and low overlapping resource use.

Methods and measures

The department EMR records a time stamp for each important interaction. The main outcome variable of LOS is calculated as the difference in time the patient is entered (arrival time) and excluded (discharge time) from the EMR. Walk-in patients are entered into the EMR by the ED greeter who records 4 pieces of information. Those arriving via emergency medical services are entered into the EMR by the charge registered nurse. All patients are manually discharged from the EMR by one of the following individuals: nurse, physician, physician assistant, clerk, or scribe. Peak census is the peak number of patients who were in the department for the hourly time block.

Outcome measures

The end point was LOS. Increases in the LOS results in crowding, which has lead to poorer patient care [5-7].

Primary statistical analysis

The statistical evaluation included the mean and SD calculations for LOS for the overall ED and within each area of the ED. Mixed linear models examined the predictive effect of presence of a trauma on LOS, with random effects for trauma activation, used to evaluate whether significant differences existed between age, number of arrivals, and peak census for cases vs controls [8]. Conditional logistic models, conditional on trauma, tested for significant differences between sexes presenting during the case and control periods.

Random effects of a trauma alert activation were included and matched by time block during which the trauma occurred (cases with controls that occurred either exactly 1 week before or after the trauma). Although the unit of analysis was the patient, these mixed linear models, through the use of random effects, adjusted for different correlations of out- comes within a time block. Additional analyses controlled for age and sex of the patient as well as the number of arrivals that occurred during that person’s hour of arrival and the peak census during that person’s hour of arrival.

Analyses were first conducted by including all patients and, then, stratifying patients by area of the ED: emergent, urgent, and nonurgent. When stratified, the number of arrivals and peak census for the appropriate areas was included as covariates rather than the overall number of arrivals and census in the ED. Adjusted differences in means

Table 1 Summary statistics of descriptive variables for cases and controls as well as the difference in means (95% CI and P value)

Sample sizes given under Arrivals apply to other variables.

a Difference of means, 95% CI, and P values derived from mixed linear model, adjusting for correlation between subjects matched by trauma event.

b Estimates and 95% CIs estimate the Mantel-Haenszel common odds ratio and are presented with the related P value for testing the common odds ratio equals one.

between LOS for cases and controls are presented. For both the overall and stratified analyses, the same overall mean age and proportion of males is used to create comparable adjusted mean LOSs.

Analyses were repeated stratifying by the patients’ discharge disposition: (1) discharged home vs (2) admitted to the hospital. In addition, analyses were repeated for the subset of case/control matches in which the trauma activation patients’ LOS was greater than 60 minutes.

Finally, using only the cases, mixed linear models similar to those above examined whether the LOS for the trauma patient was positively associated with the LOS for the patients who arrived in ED within 1 hour before or after the trauma activation.

Sensitivity analyses included reanalysis using the meth- ods described above, except using a natural logarithmic


ED area

Mean (SD)

Difference in means (95% CI) a






10.0 (4.1)

9.3 (3.9)

0.7 (0.6-0.8)

n = 3951

n = 7868



4.2 (2.0)

3.5 (1.8)

0.6 (0.5-0.7)

n = 1252

n = 2512



3.7 (1.9)

3.6 (1.7)

0.1 (0.0-0.3)

n = 758

n = 1500



4.8 (2.2)

4.7 (2.2)

0.1 (0.0-0.2)

n = 1941

n = 3856


Peak volume


47.0 (15.3)

45.3 (15.8)

1.6 (1.3-2.0)



21.3 (7.6)

20.1 (7.4)

1.0 (0.7-1.3)



6.7 (2.7)

6.5 (3.0)

0.2 (0.0-0.4)



19.8 (6.3)

19.5 (6.3)

0.4 (0.2-0.6)




46.9 (19.0)

47.1 (19.1)

-0.2 (-0.9-0.5)



58.8 (19.4)

59.9 (19.2)

-1.1 (-2.4-0.2)



39.1 (14.1)

38.9 (13.7)

0.3 (-0.9-1.5)



42.1 (16.6)

41.9 (16.5)

0.3 (-0.6-1.2)






1.05 (0.98-1.14) b


(% Of females)




1.00 (0.87-1.16) b





1.00 (0.83-1.20) b





1.11 (0.94-1.25) b


transformed value of LOS. Results from these models were compared with those where LOS was untransformed. This sensitivity analysis was done because the distribution of LOS was right-skewed, potentially resulting in unreliable results.


There were 294 trauma activations that occurred during our study period. Each trauma activation was associated with an

Table 2 Means and SD for LOS as well as estimates of the differences of means (with 95% CIs and P values for testing no difference), unadjusted and adjusted for sex, age, number of admissions, and peak volume

ED area discharge status of the

Mean (SD) Difference in

Difference in means, a

trauma patient




means a (95% CI)


adjusted for sex, age, no. of admissions, and peak volume (95% CI)



281.1 (223.0)

271.6 (216.7)

9.6 (1.3-18.0)

10.7 (2.8, 18.7)






393.0 (204.5)

395.5 (206.0)

-1.9 (-15.8-12.0)

-3.1 (-17.0-10.9)






94.6 (67.3)

94.1 (104.3)

1.0 (-7.2-9.1)

-0.2 (-8.2-7.9)






281.8 (225.5)

259.9 (203.9)

22.2 (10.7-33.7)

20.3 (9.1-31.5)







207.4 (181.4)

199.1 (170.3)

7.9 (0.0-15.8)

7.8 (0.0-15.7)






456.9 (214.4)

448.6 (216.0)

11.0 (-4.0-26.0)

11.0 (-4.0-26.0)







314.3 (191.7)

314.4 (186.2)

-0.1 (-21.9-21.6)

0.0 (-22.2-22.1)






435.7 (198.6)

435.7 (203.5)

1.7 (-15.1-18.5)

-1.9 (-18.8-15.0)







94.6 (67.4)

94.1 (104.5)

1.0 (-7.2-9.2)

-0.1 (-8.2-7.9)






87.3 (47.1)

79.6 (57.8)





231.1 (188.6)

217.7 (164.6)

13.1 (2.8-23.4)

12.0 (1.8-22.2)






489.9 (241.0)

512.3 (235.6)

20.6 (-10.1-51.3)

24.7 (-5.9-55.3)






ED LOS of the trauma patient LOS >=60 min

280.7 (221.6)

271.8 (217.4)

9.0 (0.4-17.6)

9.8 (1.6-18.0)





LOS b60 min

287.6 (246.3)

271.8 (217.4)

19.8 (-15.7-55.2)

27.3 (-7.0-61.7)






LOS >=60 min

394.7 (206.6)

394.7 (206.0)

0.7 (-13.6-15.0)

-0.5 (-14.8-13.9)





LOS b60 min

358.8 (155.7)

409.9 (206.5)

-54.1 (-112.9-4.7)

-59.3 (-119.2-0.6)






LOS >=60 min

95.5 (68.2)

93.9 (105.9)

2.1 (-6.3-10.6)

0.6 (-7.8-8.9)





LOS b60 min

70.9 (30.6)

97.8 (58.3)

-27.0 (-49.3-4.8)

-13.0 (-31.8-5.9)





Urgent b

LOS >=60 min

280.2 (221.1)

261.7 (205.3)

18.7 (6.9-30.6)

16.8 (5.2-28.3)





LOS b60 min

303.5 (278.7)

232.4 (178.9)

71.7 (24.9-118.6)

71.7 (25.7-117.7)





a Differences in means are estimated from the conditional model, estimating the average difference between cases and controls within stratum as defined by the trauma event.

b Effects of trauma on LOS, as measured by difference of adjusted means, differed by trauma length (>=60 minutes vs b60 minutes) with P = .017.

average patient arrival of 13.4 cases (SD, 6.9) and 26.8 controls (SD, 12.5).

Length of stay for matched unadjusted variables such as age and sex, as well as those for matched adjusted variables such as number of admissions and peak volume, are presented on Tables 1 and 2. Overall, there was a significant difference between LOSs of patients seen during a trauma activation vs those that were not, with an average increase of 10.7 minutes in LOS. This difference seems mostly attributable to the urgent area of the ED, in which the average increase in LOS was 20.3 minutes. These differences were not significantly affected by final disposition of patient (P = .50, P = .90, P = .95, and P = .34 for overall, within the emergent area, within the nonurgent area, and within the urgent area, respectively). The urgent care area of the ED was significantly affected by whether the trauma patient had an LOS of 60 minutes or greater (P = .017). In contrast, the emergent and nonurgent areas of the ED were not significantly affected by whether the trauma patient had an LOS of 60 minutes or greater (P = .30, P = .13, and P = .42).

In analyses of cases, trauma LOS was only significantly and positively associated with increased LOSs in the nonurgent area of the ED, with an additional 0.9 minutes (95% CI, 0.5-1.3; P b .0001) for every additional half hour of the trauma LOS. Overall, cases saw a decrease of 0.7 minutes in LOS (95% CI, -1.4-0.0; P = .061) for an additional half hour of trauma; within the emergent area, there was a decrease of 0.2 minutes (95% CI, -0.9-0.6; P = .65); and within the urgent area, there was a decrease of 0.5 minutes (95% CI, -1.0-0.1; P = .078).

Sensitivity analyses using a logarithmic transform for LOS (results not shown) yielded similar results to those presented.


To our knowledge, this is the first study that shows the effect of trauma activation on the LOS of nontraumatic patients. The findings reveal that those patient triaged to the middle acuity area are most affected with an average increase of 20 minutes. The low-acuity patients do not require a significant amount of shared resources, hence no significant effect. Those that are the highest acuity will have priority over all other patients in the ED and will be less affected. Those in the high-acuity area also have the longest LOS (Table 2), and differences due to impact of trauma activation will be more difficult to detect. Analysis of the LOS of the trauma patient only had a minor affect on those patients evaluated in the urgent care area (0.9 minutes) and had no affect on those evaluated in the other 2 areas. The authors believe that the initial evaluation and management seem to be the most resource intensive. This finding suggests it is the activation of the trauma team that matters more than the length of the trauma case.


Several limitations exist in the study. This was a single institution study at an urban level I trauma center without ED residents at the time of the analysis. Other centers without similar structures may have various results. Similarly, systems with different percentages of penetrating vs blunt trauma as well as response protocols of personnel may have different results. In addition, systems that have different members of the trauma team and Response times by various technicians might yield various results.

Baseline characteristics of trauma patients in the control group could have been different. We did not measure injury severity scores or Glasgow coma scale. Baseline character- istics of the nontrauma patients only include age and sex. There may have been a case-mix bias that may have con- tributed to the difference. At the time of the study, a 3-tiered triage system was in place. The ability to compare our data with other institutions may be affected by differing triage schemes. Lastly, the nature of the trauma and specific resource needs associated with it, rather than their length of ED stay, may have had an impact.


Improving ED length of stay continues to be a challenging goal. Time-critical interventions are being more commonly conducted in EDs, and quality-of-care measurements reflect the department’s ability to obtain these time goals. Emergency department crowding and increased LOS are associated with worsening outcomes [1,3,5]. With nation- wide increasing volumes and higher acuities, solutions are needed to address the overwhelming stresses. Care must be exercised when creating specialized Treatment protocols, and current protocols should be assessed for the impact they are having on nonprotocol patients. Opportunities exist to better define the trauma activation patient population as well as when the resources should arrive for the trauma. A move to a tiered response with varying levels of resource demands may have less of an impact on nontrauma patient LOS.


  1. Fishman PE. The impact of trauma activations on the care of emergency department patients with potential acute coronary syndromes. Ann Emerg Med 2006;48:347-53.
  2. Chen EH. The impact of a concurrent trauma alert evaluation on time to head computed tomography in patients with suspected stroke. Acad Emerg Med 2006;13(3):349-52.
  3. emergency nurses Association. Making the right decision: a triage curriculum; 1995. Saunders Elsevier: Chicago, IL.
  4. Lumley T, Levy D. Bias in the case-Crossover design: implications for studies in air pollution. Environmetrics 2000;11:689-704.
  5. Pines JM, Hollander JE. Emergency department crowding is associated with poor care for patients with severe pain. Ann Emerg Med 2008;51: 1-5.
  6. 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.
  7. Schull MJ, Vermeulen M, Slaughter G, Morrison L, Daly P. Emergency department crowding and thrombolysis delays in acute myocardial infarction. Ann Emerg Med 2004;44:577-85 Erratum in: Ann Emerg Med. 2005 Jan;45:84.
  8. Littell RC, Milliken GA, Stroup WW, Wolfinger RD, Schabenberber O. SAS System for Mixed Models. 2nd ed. Cary, NC: SAS Institute Inc; 2006.

Leave a Reply

Your email address will not be published. Required fields are marked *