pulmonary embolism rule-out criteria“>American Journal of Emergency Medicine (2011) 29, 1023-1027

Original Contribution

The Pulmonary Embolism Rule-Out Criteria rule in a community hospital ED: a retrospective study of its potential utility^?

Robert J. Dachs MD ^a,?, Divya Kulkarni MD ^b, George L. Higgins III MD ^c

^aDepartment of Emergency Medicine, Ellis Hospital, Schenectady, NY 12308, USA ^bDepartment of Family Medicine, Ellis Hospital, Schenectady, NY 12308, USA ^cDepartment of Emergency Medicine, Maine Medical Center, Portland, ME, USA

Received 2 May 2010; revised 18 May 2010; accepted 18 May 2010

Abstract

Background: The Pulmonary Embolism Rule-out Criteria rule identifies patients who can be safely discharged from the emergency department (ED) without undergoing laboratory or radiological investigation for possible pulmonary embolism (PE). It was shown to be 99% sensitive in a large validation series. Our objective was to assess the PERC rule’s performance in a representative US community hospital.

Methods: A chart review of ED patients receiving Computed tomographic scans (CTS) for possible PE during a 4-month study period was performed. The PERC rule was applied to this cohort, and its sensitivity and negative predictive value were determined.

Results: Two hundred thirteen patients underwent chest CTS to “rule out” PE. Forty-eight patients met PERC rule criteria, and all had negative CTS. Of the remaining 165 patients, 18 patients (11%) had scans positive for PE. The overall prevalence of PE was 8.45% (95% CI, 5.22-13.24%). The PERC rule’s sensitivity was 100% (95% CI, 78.12-100%), with a negative predictive value of 100% (95% CI, 90.80-100%). Application of the PERC rule at the point-of-care would have reduced CTS by 23%. Conclusions: In our community hospital, the PERC rule successfully identified ED patients who did not require CTS evaluation for PE. Had the PERC rule been applied, nearly one-quarter of all CTS performed to “rule out PE” could have been avoided.

(C) 2011

Introduction

In an attempt to identify patients with potentially life- threatening or disabling pulmonary embolism (PE), emer-

^? Disclosure: This work was not supported by any external grant or funding.

* Corresponding author. Tel.: +1 518 243 4183; fax: +1 518 243 1853.

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

gency physicians may expose many patients without the condition to excessive laboratory and radiological testing. The consequences of such unnecessary testing include, but are not limited to: increased costs, increased use of emergency department (ED) and Hospital resources, the risk of pursuing false positive results, and patient exposure to levels of computed tomography (CT)-associated radiation that could potentially result in future malignancies [1].

In order to decrease unnecessary testing in patients considered to have little or no risk of harboring PE, Kline et

0735-6757/$ - see front matter (C) 2011 doi:10.1016/j.ajem.2010.05.018

al [2] developed an 8-step clinical decision rule, the Pulmonary Embolism Rule-out Criteria (PERC rule), derived from 21 variables studied in 3148 patients. The rule was then validated in a prospective study of 8138 patients and reported to miss only 1% of patients with PE who were PERC rule negative [3]. These studies suggest that the rule could effectively and safely identify a group of patients who would not require any diagnostic testing beyond a history and physical examination.

To date, 4 small or limited reports have attempted to externally assess the PERC rule’s performance. Two European reports have been submitted as letters to the editor, and one US report has been published as an abstract [4-6]. The only complete report is a small study from a US community hospital with a predominately managed-care population [7]. No study has applied the PERC rule to ED patients being managed in a more representative full-service US community hospital setting with a patient population theoretically different from those of tertiary university-based institutions and managed care facilities. The purpose of our study was to define the sensitivity and negative predictive value of the PERC rule in our community hospital ED and, if found to be highly sensitive, to determine what percentage of Chest CT scans could have been avoided in our patient population if the PERC rule had been employed at the point of care.

Methods

Study design

This was a retrospective medical record review using a structured extraction tool for data collection. Study subjects were all ED patients who underwent a CT scan to rule out PE during the study period. The PERC rule was then applied to these patients to determine its accuracy in predicting the absence of PE. The institutional review board approved the study protocol before initiating the chart review. All data were maintained in a secure electronic database and accessed only by the primary investigator.

Study setting and population

The study institution is a 273-bed full service community hospital located in northeastern New York state that serves approximately 250 000 residents in a three county area. The ED serves 42 000 patients annually and admits approxi- mately 27% of these patients. A Family Medicine residency is the only full-time training program located on-site.

Study subjects were identified from a computerized log of ED patients who received a CT scan of the chest from June 1, 2008, to September 30, 2008. This list was obtained directly from the radiology department electronic files and included the name of the ordering physician. All CT requests included

a reason for the examination. Those performed to “rule out PE,” as well as those using a contrast protocol to identify PE, were included for further analysis. All CT studies were performed on a General Electric VCT 64 slice helical scanner (Milwaukee, Wis).

Study protocol

The two principle investigators (RD and DK) simulta- neously reviewed the electronic medical record of each patient undergoing chest CT to determine if it was (1) ordered by an emergency physician and (2) performed for the purpose of detecting the presence of PE. Patients were only included if their CT was ordered by an emergency physician. ED CT scans ordered by other members of the medical staff were not included in the analysis since these were often generated after the initial point of care because the patient had been admitted, but had not yet been transferred to the Inpatient unit.

Each variable in the PERC rule was then applied to all cases meeting both inclusion criteria. The investigators simultaneously extracted clinical data to answer the following eight questions:

Was the patient older than 49 years?

Was the pulse rate above 99 beats per minute?

Was the pulse oximetry less than 95% on room air?

Was there a history of hemoptysis?

Was the patient taking exogenous estrogen?

Was there a prior diagnosis of venous thromboembolism?

Had there been recent surgery or trauma in last

4 weeks?

Was there unilateral Leg swelling?

Cases in which the answer to all eight questions was “No” were considered meeting PERC rule criteria and classified as “PERC rule negative (-).” An affirmative answer to one or more of the 8 questions resulted in the case being classified as “PERC rule positive (+)”.

Both the emergency physician and nursing staff at the study institution consistently document patient-specific information in a commercially available template system (T-system, Inc, Dallas, Tex), and these records were accessed for data collection. The patient’s age was abstracted from the hospital registration form. If multiple sets of vital signs were obtained, all values were reviewed. If any pulse rate exceeded 99 beats per minute, or any pulse oximetry reading of less than 95% was recorded, then the patient was considered PERC rule+. Both nursing records and physician forms were reviewed for documentation of hemoptysis, prior diagnosis of venous thromboembolism, recent surgery or trauma, and presence of leg swelling. Documentation of any of these placed the patient in the PERC rule+ group. All records, including Medication reconciliation forms, were reviewed to assess the use of exogenous estrogen. Any notation documenting estrogen

use or positive pregnancy status resulted in the patient being declared PERC rule+.

The original PERC rule also required that patients be considered “low probability” for PE based on the clinician’s gestalt. This Clinical impression could not be captured in review of the medical record and only the eight objective clinical criteria were used to determine PERC rule classification.

All CT scans were interpreted by a member of the radiology staff at the study institution, all of whom are certified by the American Board of Radiology. The final report of each CT study was used to determine the presence or absence of pulmonary embolism.

Outcomes measured and data management

The main outcome measured was the sensitivity of the PERC rule. All data were entered in an electronic spreadsheet (Microsoft Excel, Microsoft Corp, Redmond, Wash). The sensitivity, specificity, positive predictive value, and negative predictive value, and their respective 95% CIs, were calculated from a 2 x 2 observed contingency table. No a priori sample size calculation or adjustment for multiple comparisons was performed. A P value of .05 or less was considered statistically significant.

Results

Three hundred eight chest CT scans were ordered from the ED during the four month study period, with 213 of these ordered by emergency physicians to determine the presence of PE. Fig. 1 demonstrates the study algorithm and final clinical assignment of all patients. 18 cases of PE were identified, for a prevalence of 8.4% (95% CI, 5.2-13.2%). Of the 213 study subjects, 48 (22.5%) were PERC rule-. All 48 of these cases were negative for PE, resulting in 100% sensitivity (95% CI, 78.1-100%), with a corresponding negative predictive value of 100% (95% CI, 90.8-100%). Application of the PERC rule at the point-of-care would have reduced CT scans by 23%.

The 2 x 2 observed contingency table shown in Table 1 displays sensitivity, specificity, positive predictive values, and negative predictive values. The corresponding 95% CIs are included.

Discussion

To appropriately use resources, various attempts at defining which patients require Radiographic investigation for PE have been made. As an example, D-dimer testing has been promoted as a screen for patients deemed to be at low risk for PE [8]. However, its poor specificity and the potential for its indiscriminate use may result in the unintended consequence of a paradoxical and unwarranted escalation in radiographic evaluation for PE [9,10].

Alternatively, Kline and colleagues [2] derived the PERC rule in order to identify patients with such a Low pretest probability of PE that diagnostic evaluation would not be necessary. The high sensitivity noted in their derivation study was subsequently validated in a large multicenter study published in 2008 [3].

Attempts to duplicate the PERC rule’s remarkable results have been limited. Table 2 summarizes the four previous reports that have attempted to apply the PERC rule. The only prospective study other than the original Kline et al investigation was reported in abstract form [3]. The other three reports retrospectively applied the PERC rule in a fashion similar to the model we present here. Two reports suggested the PERC rule’s sensitivity was good enough to use as a screening tool [4,7]. However, a European university study that retrospectively applied the PERC rule to their patient population noted the rule had inadequate sensitivity to circumvent radiographic evaluation for PE. However, this study included a patient population with a 25.7% prevalence of PE [5]. Kline et al has commented that the PERC rule is best suited to a patient population with a prevalence of PE of less than 10% [10].

The current study lends further support that the PERC rule may be useful in patient populations with a low prevalence of disease. It differs from all previous studies that evaluated the

Fig. 1 Study algorithm.

Table 1 2 x 2 Observed contingency table
Test	CT + for PE n = 18	CT - for PE n = 195
PERC rule+	18	147	PPV 10.91%
n = 165 PERC rule-	0	48	(95% CI, 6.8-16.9) NPV 100%
n = 48			(95% CI, 90.8-100)
	Sensitivity 100%	Specificity 24.6%
	(95% CI, 78.12-100)	(95% CI, 18.87-31.39)

utility of the PERC rule in that it exclusively examined the PERC rule’s sensitivity in a representative US community hospital population. Because 7 of every eight Americans are hospitalized in local, community hospitals for their acute illnesses [11], assessment of the PERC rule’s sensitivity in this patient population is critical before its widespread application can be supported. To our knowledge, our report represents the largest series to date that focuses on this important patient population.

In a smaller study, Wolf et al demonstrated the PERC rule’s high sensitivity in a group of patients from a community hospital in Colorado. However, their patient population was unique due to the high penetrance of managed care in the community [7]. Although Kline’s original validation study did include a few community hospitals among its thirteen sites, most sites were university or referral centers [3].

Our study is also unique since all cases in which the ED physicians ordered a CT scan to “rule out” a PE were included in our analysis and therefore the study population was not limited by any specific enrollment criteria. In contrast, the Manchester Investigation of Pulmonary Embo- lism Diagnosis study was limited to include only those patients with “pleuritic chest pain.” This resulted in the exclusion of those patients with only dyspnea as a chief complaint or other non-pleuritic clinical symptoms and signs suggestive of PE [4]. Two other publications were analyses

Table 2 Review of PERC validation studies

of previously prospective studies with specific inclusion and exclusion criteria [5,7]. Narrowing the inclusion criteria for patient enrollment in these studies introduces the potential for underestimating the prevalence of disease in their communities. By including all patients who underwent CT scanning for PE, we were able to enroll study subjects whom the clinician had determined to have “some” degree of suspicion for PE even if they did not meet more objective criteria. This is commonly practiced by emergency physi- cians and likely includes the proven diagnostic value of clinician “gestalt.”

Limitations

Limitations of this study include its retrospective design, based on a review of the ED medical record. Because our study site uses the T-system documentation record, all variables were captured and recorded on a standardized recording form. The only noteworthy deviation from the methodological principles set forth by Lowenstein [12] was that the abstractors were not blinded to the study hypothesis. Lack of funding prevented the use of independent abstractors to perform this function. In addition, the investigators abstracted the data simultaneously to familiarize themselves with a new hospital computer system. However, because all

Author	Study design	Location	Number (n) of patients		Prevalence of PE (%)	% of patients PERC (-)	PERC false (-) (95% CI)
Kline et al [3]	Prospective	Multicenter	n =	8138	6.9%	20%	15/1666: 1.0% (0.6%-1.6%)
(validation study)
Righini et al [5]	Retrospective	University of Geneva	n =	762	23%	11.7%	6/89: 6.7% (3%-14%)
(letter to editor)
Hogg et al [4]	Retrospective	Manchester Royal	n =	425	5.3%	51%	3/216: 1.39% (0.5%-4.0%)
(letter to editor)		Infirmary
Courtney et al [6]	Prospective,	Northwestern Univ.	n =	315	4.4%	42%	2/131: 1.5% (0.2%-5.4%)
(abstract)	Observational
Wolf et al [7]	Retrospective	Community-based,	n =	134	12%	14%	0/19: 0%
		Denver, Co
Dachs et al 2010	Retrospective	Community-based,	n =	213	8.4%	22.5% 0/48:
(current study)		Schenectady, NY				0% (0-9.2%)

data points collected were clear objective measures, any bias would likely be minimized.

Because of the retrospective design of the study, no follow-up was available on the study subjects. Therefore, we are unable to comment on the number of patients with a negative CT who might have subsequently been diagnosed as having a PE. However, we are unaware of any study subject returning to our ED within 3 months of discharge who was subsequently diagnosed with venous thromboembolism.

We chose not to include patients who underwent ventilation/perfusion (VQ) scanning for evaluation of possible PE. Ventilation/perfusion scans at our institution are reported as normal, low probability, indeterminate, and high probability. Because these results are not dichotomized, they could not be seamlessly correlated with the CT results. However, we believe that our decision not to include these patients did not affect the sensitivity calculations. The medical records of all sixteen patients for whom VQ scans were ordered in the ED during the study period were reviewed for PERC rule criteria. Only a single patient was classified as PERC rule-. The VQ scan in this patient was reported as “normal.” Inclusion of this one patient in the final analysis would not have changed the PERC rule’s 100% sensitivity.

In addition to the eight objective variables, the PERC rule requires the clinician to have a “gestalt” that the patient is at low risk for PE. This low index of suspicion was defined as less than a 15% chance of PE being present. Given the retrospective design of our investigation, it was impossible for us to extract this subjective indicator from the medical record. Published chart review studies by their nature cannot assess and account for a clinician’s degree of suspicion. However, with one exception, every study that has applied the PERC rule retrospectively has found it to retain a high level of sensitivity. This raises two interesting theoretical questions: (1) To what degree do the eight objective clinical criteria ultimately contribute to the clinician’s “gestalt”? (2) If the 8 clinical criteria strongly influence and represent a majority of clinician’s gestalt, how often is the clinician’s decision to override the objective criteria and proceed with radiographic testing appropriate?

Conclusion

Our study supports emerging evidence that the PERC rule’s sensitivity is clinically adequate in patient populations with a low prevalence of PE and suggests that the community hospital may be an appropriate location to implement the PERC rule. In our ED, nearly one quarter of all CT angiograms ordered to identify patients with a possible PE

could have been safely omitted had the PERC rule been employed. We would suggest that community hospitals similar to ours assess the prevalence of PE in their ED population before employing the PERC rule. Appropriate application of the PERC rule has the potential to significantly reduce the number of unnecessary chest CT scans.

Acknowledgment

The authors extend sincere appreciation to Tania D. Strout, RN, BSN, MS for her valuable assistance with statistical analysis.

References

1. Brenner DJ, Hall EJ. Computed tomography-an increasing source of radiation exposure. NEJM 2007;357:2277-84.
2. Kline C, Mitchell AM, Kabrhel C, et al. Clinical criteria to prevent unnecessary diagnostic testing in emergency department patients with suspected pulmonary embolism. J Thromb Haemost 2004;2: 1247-55.
3. Kline JA, Courtney DM, Kabrhel C, Moore CL, Smithline HA, Plewa MC, et al. Prospective multicenter evaluation of the pulmonary embolism rule-out criteria. J Thromb Haemost 2008;6:772-80.
4. Hogg KE, Dawson K, Kline JA. Application of pulmonary embolism rule-out criteria to the UK Manchester Investigation of Pulmonary Embolism Diagnosis (MIOPED) study cohort. J Thromb Haemost 2005;3:592-3.
5. Righini M, LeGal G, Perrier A, et al. More on: clinical criteria to prevent unnecessary diagnostic testing in emergency department patients with suspected pulmonary embolism. J Thromb Haemost 2005;3:188-9.
6. Courtney DM, Pribaz JR, Sehh AC, et al. Prospective evaluation of the Pulmonary Embolism Rule-out Criteria rule: an 8-variable block rule to identify subjects at Very low risk of pulmonary embolism. Acad Emerg Med 2006;13:S157-8 (Abstract).
7. Wolf SJ, McCubbin TR, Nordenholz KE, et al. Assessment of the pulmonary embolism rule-out criteria rule for evaluation of suspected pulmonary embolism in the emergency department. Am J Emerg Med 2008;26:181-5.
8. Roy PM, Colombet I, Durieux P, et al. Systematic review and meta- analysis of strategies for the diagnosis of suspected pulmonary embolism. BMJ 2005;331:259-68.
9. Kabrhel C, Matts C, McNamara M, et al. A highly sensitive ELISA D- dimer increases testing but not diagnosis of pulmonary embolism. Acad Emerg Med 2006;13:519-24.
10. Kline JA, Mitchell AM, Kabrhel C, et al. More on: clinical criteria to prevent unnecessary diagnostic testing in emergency department patients with suspected pulmonary embolism. J Thromb Haemost 2005;3:190-1.
11. Green LA, Fryer GE, Yawn BP, et al. The ecology of medical care revisited. NEJM 344;2021-25.
12. Lowenstein SR. Medical record reviews in emergency medicine: the blessing and the curse. Ann Emerg Med 2005;45:452-5.