Correlation between triage nurse and physician ordering of ED tests

Correlation Between Triage Nurse and Physician Ordering of ED Tests DAVID C. SEABERG, MD, BRUCE A. MAcLEOD, MD Triage nurse ordering of tests can reduce patient time in the emergency department. This benefit must be balanced with the extra expense and potential morbidity of unnecessary tests. This study examined whether triage nurse ordering of lab tests and X-rays correlated with physician ordering. In Phase I, a prospective study, triage nurses completed an order form for each patient triaged during randomly selected 2,hour blocks over a 3-week period. In Phase II, Phase I was repeated with test ordering guidelines derived from Phase I results. Correlations of nurse and physician orders were analyzed by kappa statistics and paired t tests. In Phase I, 176 patients were enrolled of a possible 204 (82%). In Phase II, 350 patients were enrolled of a possible 418 (84%). The percentage of nurse overordering of tests in Phase I was 35%; in Phase II, it was 34% (P NS). Kappa values were determined for blood testing (Phase I, 0.48; Phase II, 0.54 [P = .048]), for X-ray (Phase I, 0.68; Phase II, 0.65 [P = .523]), for urinalysis (Phase I, 0.76; Phase II, 0.67 [P = .338]), and for electrocardiogram (Phase I, 0.78; Phase II, 0.77 [P = .438]). The improved lab correlation in Phase II resulted from the nurses ordering fewer tests (P = .046). Nurses deviated from the test ordering guidelines in 37% of patients. Triage nurse ordering of X-rays, electrocardiogram, and urinalysis has moderate to substantial correlation with physician ordering. Correlation of triage nurse and physician ordering of tests can be improved with the use of test ordering guidelines. (Am J Emerg ied 1998;16:8-11. Copyright © 1998 by W.B. Saunders Company) Triage nurse evaluation protocols have been developed in many emergency departments (EDs) throughout the United States. Included in these protocols are prehospital/trauma triage criteria, 1,2 patient severity triage criteria, 3 as well as screening exam guidelines. 4 Triage nursing protocols have the potential advantages of facilitating patient care, improving patient satisfaction, and increasing the consistency of care and improved teamwork. Triage nurse evaluation protocols have the added potential to increase the efficiency of the emergency physician. Triage protocols have not been studied extensively. Most studies have examined the role of the triage nurse in radiograph ordering. 5-8 In these studies, triage nurses were found to have excellent agreement with physicians on the ordering of extremity radiographs. Although many EDs have triage guidelines for other patient diagnostic test ordering, From the Division of Emergency Medicine, University of Florida Health Science Center, Jacksonville, FL. Manuscript received April 30, 1996; accepted October 16, 1996. Supported by the Emergency Medicine Association of Pittsburgh. Presented at the American College of Emergency Physicians Scientific Assembly, February 1995, San Francisco, CA, and the Pennsylvania Chapter of the American College of Emergency Physicians Annual Scientific Meeting, April 1993, Harrisburg, PA. Address reprint requests to Dr Seaberg, Division of Emergency Medicine, University of Florida Health Science Center, 655 West 8th St, Jacksonville, FL 32209. Key Words:Triage, nurse, physician, emergency medicine, diagnostic test ordering. Copyright © 1998 by W.B. Saunders Company 0735-6757/98/1601-000358.00/0

there is a paucity of literature evaluating the triage nurse's efficacy in this role. With the current trend of ED overcrowding throughoutthe country, efforts are being made to increase the efficacy of patient evaluation time in the department. One method to increase efficiency is through triage nurse protocols for ordering of diagnostic tests before the patient is seen by the physician. The benefit of triage nurse ordering must be balanced with the extra expense, extra time, and potential morbidity from overordering diagnostic tests that would not have been ordered by the physician. The objective of Phase I of this study was to examine whether triage nurse ordering of laboratory and radiographic studies correlated with physician ordering. Phase II sought to determine whether written diagnostic test ordering guidelines improved this correlation.

METHODS This was a two-phase prospective study conducted at the Mercy Hospital of Pittsburgh's Emergency Department, which has an annual census of over 42,000. The ED is a clinical training site for the University of Pittsburgh Affiliated Residency in Emergency Medicine. All attending physicians are board certified/eligible in emergency medicine. Outpatient visits to the department last a mean time of 2 hours and 23 minutes, with a mean of 49 minutes from the time the patient arrives at triage until seen by a physician. Emergency nurses perform the triage function. Triage nurse training consists of an in-service on the use of triage protocols, as well as initially working in an apprentice-type relationship with a nursing instructor. After the initial assessment, the nurse can triage patients into an urgent-care (fast-track), pediatric, or main (general) ED. Each phase of this study was conducted daily during randomly selected 2-hour blocks over 3-week periods. All patients who presented to the triage desk during the study times were included. Patients excluded from the study were those who arrived by medic unit and required an immediate ED bed, either because of major trauma, unstable vital signs, or potential resuscitation. Physicians were blinded to study design and did not know the study was being performed. The nurse was not allowed to consult with the physician. There were a total of 18 triage nurses and 12 emergency physicians who participated in the study. In Phase I, the triage nurse completed a "sham" order form on each patient lriaged, indicating which laboratory tests or X-rays he or she thought appropriate for each patient. In Phase II, the emergency physicians developed guidelines by group consensus for triage nurse ordering of diagnostic tests (Table 1). These local guidelines consisted of what the emergency physicians thought were the minimum basic diagnostic tests to be ordered for the presenting complaint and were also based on available medical literature. The goal of these guidelines was to have triage nurses order appropriate diagnostic tests while minimizing the amount of unnecessary, unwanted tests. In Phase II, conducted 6 months after Phase I, the triage nurse again completed a "sham" order form on each patient triaged, this time indicating which diagnostic tests were indicated using the written test ordering guidelines. The triage

SEABERG AND MAcLEOD • CORRELATION OF ED TEST ORDERING

9

TABLE1. Test Ordering Criteria Abdominal Pain Female, reproductive age: UA, urine HCG Abnormal Vital Signs: CBC, L/B/C/Glu, UA, amylase Age >60: CBC, L/B/C/Glu, UA, amylase Pelvic Complaints: UA, urine HCG Altered Mental Status: CBC, L/B/C/Giu, pulse oximetry Chest Pain Age >40: ECG "Cardiac Symptoms" (Rm 5-8 Triage): ECG, CCU labs Fever (>38°C) Age >50: CBC, UA, chest X-ray Extremity Injury (if pain and swelling): extremity X-ray Seizure: glucose, anticonvulsant level (if on them) Shortness of Breath: pulse oximetry, theophylline level (if on it) Age >50: CXR, pulse oximetry theophylline level (if on it) Traumatic Neck Pain: Cervical spine X-rays Unclear or Nonspecific complaints: No lab or X-ray orders X-Rays Available to Order: Chest Fever, age >50 Shortness of breath, age >50 CCU labs

Extremity If pain and swelling of extremity

Cervical Spine Traumatic neck pain

DO NOT ORDER: Abdominal, T-Spine, LS Spine, Sinus Films Labs Available to Order: CBC Abdominal pain: Abnormal VS, age >60 Fever: age >50 Urinalysis Abdominal Pain: Reproductive female, abnormal VS, age >60 Pelvic Complaints: Reproductive female Urine HCG: Reproductive female with abdominal pain or pelvic c/o L/B/C/Gluc, Amylase: Abdominal Pain: abnormal VS, age >60 ECG: Chest pain, age >40 CCU Labs: Triage to Rm 5-8 w/"Cardiac c/o" ABBREVIATIONS:UA, urinalysis; L/B/C/Glu, electrolytes/blood urea nitrogen/creatinine/glucose; CCU, coronary care unit; VS, vital signs.

nurses were all given an in-service into the use of these guidelines, and the guidelines were posted at the triage desk. Data on each patient were entered into the dBase IlI (Borland International, Inc, Scotts Valley, CA) data management computer program. Diagnostic tests were compiled into four major categories: blood, X-ray, urinalysis, and electrocardiogram (ECG). In each category, separate analyses were done for each specific test. The blood test category consisted of complete blood count (CBC), electrolyte levels, glucose level, amylase level, coagulation studies, ethanol level, beta-human chorionic gonadotropin (I3-HCG), and other blood tests. The percentage of diagnostic tests with complete agreement (physician and nurse both ordered), nurse overordering (nurse ordered but physician did not), and nurse underordering (physician ordered test but nurse did not) were calculated. The total percentage could exceed 100% because some patients had both underordering and overordering of diagnostic tests. Since the physicians must ultimately make patient diagnostic and disposition decisions, their test ordering was considered the "reference criterion" to which nurse ordering was compared. In each phase, correlation of triage nurse orders with physician chart orders were analyzed using kappa statistics. Kappas measure agreement beyond chance between the responses of the triage nurse and the physician orders. Kappas between 0.4 and 0.6 indicate moderate agreement, 0.6 to 0.8 indicate substantial agreement, and 0.8 to 1.0 indicate excellent agreement. Any changes in correlation between the Phase

I and Phase II diagnostic testing was analyzed using paired two-tailed student t tests. In addition, the authors retrospectively evaluated whether the triage nurses followed the test ordering guidelines developed during Phase II. The percentage of protocol deviations was calculated. The correlation of the guidelines along with actual physician ordering was measured using kappa statistics.

RESULTS In Phase I, 167 patients were enrolled out of a possible 204 (82%). In Phase II, 341 patients out of a possible 418 (82%) were enrolled. The ED triage chief complaints were similar for both Phase I and Phase II (Figure 1). Physicians ordered a total of 340 diagnostic tests in Phase I and 644 tests in Phase II. In Phase I, there was complete agreement between tests ordered by the triage nurse and the physician for 41% of patients. Nurse overordering of tests occurred for 35% of patients and underordering of tests occurred for 37% of patients when compared with physician ordering. Fifteen percent of patients had both overordering and underordering of diagnostic tests by the triage nurse.

10

AMERICAN JOURNAL OF EMERGENCY MEDICINE • Volume 16, Number 1 • January 1998

[]

Cardio-pulmonary

II

Dermatologic ENT/Ophthalmology

Gastrointestinal

12 ,..

10

General Medical

Infectious Neurologic OB/GYN Orthopedic t

Psychiatric Trauma/Injury

i 0

5

10

15

20

25

FIGURE 1. Triagechief complaints: percentage of patients with each chief complaint category ([], Phase I; m, Phase II). In Phase II, there was complete agreement between the nurse and physician ordering for 57% of patients. This was statistically significant compared with Phase I complete agreement (P = .0042). Nurse overordering occurred for 34% of patients and underordering occurred for 24%. The percentage of triage nurse overordering of diagnostic tests did not decrease in Phase II (P = .90). However, the percentage of underordering did decrease (P = .0021) compared with Phase I. In Phase I, correlation between tests ordered by triage nurses and physicians, as measured by kappa values, was substantial for X-rays, urinalysis, and ECGs. There was moderate correlation for blood test ordering (Table 2). There was a statistically significant improvement in correlation during Phase II between nurse and physician blood test ordering. This improvement correlation was noted primarily for CBC, electrolyte levels, and glucose measurements (Table 3). This improved correlation for blood tests primarily resulted from decreased nurse overordering of these specific tests (Figure 2). In Phase II, nurses deviated from the test ordering guidelines in 37% of patients. When the test ordering guidelines alone were compared with physician ordering, there was complete agreement for test ordering for 61% of patients. The guidelines overordered tests for 5% of patients and underordered for 35% of patients. The percentage of overordering of lab tests using the guidelines alone was

CBC

Lytes

Glu

FIGURE 2. Blood test ordering: Percentage of nurses overordering (m, Phase I; [], Phase I/) and underordering ([7, Phase I; [], Phase II). Lytes, electrolytes; Glu, glucose. significantly reduced compared with nurse overordering (5% v 34%, P < .000001). This effect was mainly seen in those patients for whom the physician did not order any diagnostic tests. Of the 134 patients in Phase II without diagnostic testing ordered by the physician, nurses ordered tests on 22%, whereas the guidelines ordered tests on only 3% (P = .0000045). Using the written guidelines alone, correlation with physician test ordering would have been excellent for urinalysis and substantial for all other tests (blood, X-ray, and ECG). Compared with the triage nurse ordering, correlation with physician ordering was significantly higher for X-rays and urinalyses using the guidelines (Table 4). The improved X-ray correlations using the guidelines were primarily for chest, extremity, and cervical spine X-rays (Table 5). DISCUSSION

This study found triage nurse ordering of tests has moderate to substantial correlation with physician ordering. TABLE 4. Correlation (Kappa Values) Compared to Physician Ordering

TABLE 2. Correlation (Kappa Values) of Tests Ordered

Phase I Phase II P Value

Blood

X-Ray

Urinalysis

ECG

0.48 0.54 .048

0.68 0.65 .523

0.76 0.67 .338

0.78 0.77 .438

Phase II

Blood

X-Ray

Urinalysis

ECG

Nurse Guidelines Alone PValue

.54 .59 .i015

.65 .77 .0001

.67 .86 .0182

.77 .78 .2296

TABLE 5. Correlation (Kappa Values) for X-Ray Tests TABLE 3. Correlation (Kappa Values) for Blood Tests

Phase I Phase II P Value

CBC

Electrolytes

Glucose

Phase II

Chest

Extremity

Cervical Spine

.47 .51 .67

.38 .52 .13

.39 .53 .t3

Nurse Guidelines Alone P Value

.63 .78 .046

.70 .86 .036

.55 .82 .00036

:

SEABERG AND MAcLEOD • CORRELATION OF ED TEST ORDERING

This has been supported in other studies that have examined nurse ordering of extremity radiographs. 5-8 To have a costand time-effective system, nurse overordering of tests must be minimized. In our department, test ordering guidelines improved blood testing correlation between the triage nurse and physicians, with less nurse overordering of CBC and electrolyte levels. In this study, the mean waiting time in the department until the patient was seen by a physician was 49 minutes. If the nurses could order the appropriate tests from triage, the physicians would have most of the necessary diagnostic tests to make disposition decisions after the patient's history and physical. This would have the potential of decreasing patient time in the ED and decreasing overall costs. In 37% of Phase II patients, the triage nurse did not follow the test ordering guidelines. If the guidelines had been followed, the retrospective review found that even nigher correlation with physician testing would have been achieved with less test overordering. The reasons that the guidelines were not followed are unclear. It may be that the nurses required more in-servicing and education in the use of the guidelines. Perhaps if they had a larger role in the development of the guidelines, there would have been less deviation. It is also a possibility that the guidelines were difficult to follow or that the fight questions were not being asked in triage. Specifically, the guidelines still required the triage nurse to interpret the patients' chief complaints. This may be dependent on the level of triage nurse training and experience. Since the guidelines were not followed in 37% of patients, perhaps there is another way to deliver these guidelines. Computer-driven protocols have been used in other studies to reduce laboratory u s e . 9-11 With the total cost of clinical laboratory tests in the United States estimated to be about 10% of all health care costs, lz computerized systems have been shown to decrease test utilization and costs. Likewise, computerizing our test ordering guidelines may make them easier to use. By using computerized test ordering guidelines, the triage nurse could be prompted to obtain pertinent history and then be instructed to order the appropriate tests. The ultimate outcome would be to order the most useful studies, decrease overordering of tests, and improve the efficiency of the physician. Computerized algorithms have been used by nurses to direct triage patients, 13 but not for laboratory test ordering. Another useful feature of using triage guidelines for test ordering is the potential to add clinical decision rules into the guidelines. Decision rules, such as for ankle ~4 and knee 15 radiographs, can significantly reduce the number of necessary X-rays, and thereby decrease costs. Some of these rules have been shown to be effectively implemented by triage nurses. 5,7 Currently, the "gold standard" for diagnostic test ordering is the physician usage of such tests. However, as well-designed and validated clinical decision rules are developed, these may help support physician judgement. This study has the limitation of comparing the theoretical intention to order ("sham") a diagnostic test by the triage nurse with the actual ordering by the physician. If the triage nurse had to actually order the tests, this may have led to an alteration in test ordering usage, thereby skewing the results. We are currently planning further studies on the real-time ordering of tests by the triage nurse using computerized test ordering guidelines.

11

Another limitation of this study is that the test ordering guidelines were locally developed by the physicians who practice in our ED. Although published studies and research on clinical decision rules were used to help formulate these guidelines, they were, in fact, developed by local consensus. The homogeneity of the emergency physicians and their practice style, as well as the type of patient presentations, can influence the external validity of these guidelines. We did not attempt to define specific test ordering guidelines that would be generalizable to other practice settings. We merely sought to determine whether locally derived test ordering guidelines could be used by triage nurses for the eventual use of facilitating patient flow without the adverse effects of increased costs or morbidity. Therefore, the transportability of these guidelines into other clinical practices may be limited.

CONCLUSION Triage nurse ordering of diagnostic tests has substantial agreement with physician ordering, although 34% of patients had at least one test not ordered by a physician. Physician test ordering correlated better with locally developed written guidelines than actual triage nurse ordering. Further efforts are needed to improve triage nurse compliance with written test ordering guidelines.

REFERENCES 1. Bart WG, Jones G, Fortlage D: The trauma triage rule: A new, resource-based approach to the prehospital identification of major trauma victims. Ann Emerg Med 1990; 19:1401-1406 2. Gormican SP: CRAM Scale: Field triage of trauma victims. Ann Emerg Med 1982;11:132-135 3. Derlet RW, Nishio DA: Refusing care to patients who present to an emergency department. Ann Emerg Med 1990;19:262-267 4. Pardee DA: Decreasing the wait for emergency department patients: An expanded triage nurse role. J Emerg Nurs 1992;18:311315 5. Ropp L, Blouin R, Dulberg C, et al: Radiograph ordering: Agreement between the triage nurse and the physician in a pediatric emergency department. J Emerg Med 1990;8:697-700 6. Klassen TP, Rott LJ, Sutcliffe T, et al: A randomized, controlled trial of radiograph ordering for extremity trauma in a pediatric emergency department. Ann Emerg Med 1993;22:1524-1529 7. Stiell IG, Greenberg G, McKnight RD, et al: Triage nurse use of decision rules for radiography in ankle injuries. Ann Emerg Med 1993;22:948 (abstr) 8. McArthur CL, Thomas M: Comparison of triage nurses versus emergency physician ordering of extremity radiographs. Am J Emerg Med 1995;13:248-250 9. Fowkes FGR: Containing the use of diagnostic tests. BMJ 1985;290:488-489 10. Young DW: Improving laboratory usage: A review. Postgrad Med J 1988;64:283-289 11. Peters M, Broughton PMG, Nightingale PG: Use of information technology for auditing effective use of laboratory services. J Clin Pathol 1991;41:539-542 12. Spiecher CE: Decision oriented test request forms. A system for implementing practice parameters in laboratory medicine. Clin Lab Med 1991 ;11:255-265 13. Berman DA, Coleridge ST, McMurray TA: Computerized algorithm-directed triage in the emergency department. Ann Emerg Med 1989;18:141-144 14. Stiell IG, McKnight RD, Greenberg GH, et al: Implementation of the Ottowa Ankle Rules. JAMA 1994;271:827-832 15. Seaberg DC, Jackson R: Clinical decision rule for knee radiographs. Am J Emerg Meal 1994;12:541-543