(Chest. 2004;125:1651-1656.)
© 2004
American College of Chest Physicians
ECG Score Predicts Those With the Greatest Percentage of Perfusion Defects Due to Acute Pulmonary Thromboembolic Disease*
Stephen Iles, MRCP;
Campbell J. Le Heron;
Gwyn Davies, MRCP;
John G. Turner, MD and
Lutz E. L. Beckert, MD, FCCP
* From the Canterbury Respiratory Research Group (Dr. Iles and Mr. Le Heron); and Department of Medicine (Drs. Davies, Turner, and Beckert), Christchurch School of Medicine and Health Sciences, University of Otago, Christchurch, Otago, New Zealand.
Correspondence to: Stephen Iles, MRCP, Department of Respiratory Medicine, Royal Cornwall Hospital, Treliske, Cornwall TR4 3LJ, United Kingdom; e-mail: steveiles{at}doctors.org.uk
 |
Abstract
|
|---|
Background: More aggressive management may be warranted for patients with acute pulmonary embolism (PE) and the greatest pulmonary vascular obstruction. We hypothesized that a scoring system based on the ECG might identify such patients.
Methods: Consecutive patients investigated for PE at Christchurch Hospital between 1997 and 2002 with high-probability ventilation/perfusion (
/
) scan findings were studied. The ECG obtained closest to and within 48 h of the scan was scored by two independent observers, and the mean ECG score was calculated. / scan findings were categorized into those with < 30%, 30 to 50%, and > 50% perfusion defect by two independent observers experienced in / interpretation. A consensus score was taken when disagreement occurred.
Results: Two hundred twenty-nine patients were included in the study. The interobserver agreement for ECG score was 0.96 (Cronbach 
) and / score was 0.55 (
). The ECG predicted those with the greatest amount of perfusion defects. Mean ECG score was 2.6 (SD 2.8) in patients with < 30% perfusion defect, 3.2 (SD 2.9) in patients with 30 to 50% perfusion defect, and 5.3 (SD 3.7) in patients with > 50% perfusion defect. The area under the receiver operating characteristic curve for ECG score and those with > 50% perfusion defect was 0.71 (SE 0.04). An ECG score of 
3 predicted those with > 50% perfusion defect with a sensitivity of 70% (95% confidence interval [CI], 59 to 81%), and a specificity of 59% (95% CI, 51 to 67%).
Conclusion: An ECG score, simple to derive, predicts those with the greatest percentage of perfusion defect. Using the ECG for management warrants prospective evaluation.
Key Words: ECG score • percentage perfusion defect • pulmonary embolism • severity assessment
|
Introduction
|
|---|
Applying recent therapeutic advances in the management of pulmonary embolism (PE) depends on identifying patients at low or high risk of adverse events. Those with acute massive or submassive PE may benefit from thrombolysis.1 Conversely, patients with non–life-threatening PE can be treated with anticoagulation as outpatients, thereby obviating the need for admission to the hospital.2 In order to expedite effective management, the clinician relies on clinical judgement prior to confirmatory imaging of PE, so that those most at risk might be recognized and treated early. Similarly, definitive investigation might not need be done after hours, if those with non–life-threatening PE could be identified. Methods used to define the severity of PE, such as the finding of right ventricular dysfunction on echocardiography are not routinely available in the early management of those with possible PE in many centers.
An ECG scoring system has recently been derived from a group of 26 patients with PE confirmed by pulmonary angiography.3 In this small group of patients, the ECG score correlated with the degree of pulmonary hypertension, and a score > 9 was suggestive of systolic pulmonary artery pressure > 50 mm Hg. Ventilation/perfusion (/) scanning is less invasive than pulmonary angiography, and is the most common method of investigating PE in some clinical centers.4 In this study, we used a previously published method5 of quantifying the extent of perfusion defects on / scanning as the reference standard to judge the severity of PE. We hypothesized that the ECG score could identify those patients with the greatest amount of perfusion defects on / scanning due to PE.
|
Materials and Methods
|
|---|
Patients
The study was performed at Christchurch Hospital, a tertiary-level institution with 600 beds, serving a population of 375,000. Approximately 800 / scans are performed annually as the primary method of investigating possible PE. All medical and surgical patients who were investigated for possible PE with / scans between September 1997 and August 2002, and in whom a high probability result was found, were eligible for the study. Patients were excluded if the lung scan was performed for reasons other than for PE, if the case records could not be located or if no record was found of an ECG obtained within 48 h, or if the / scan was not found.
Investigations
Case notes of study patients were scrutinized carefully for demographic data and risk factors for PE (British Thoracic Society criteria6), and a copy was made of the ECG obtained nearest the time of the lung scan. The index test, the ECG score, was calculated according to method of the original authors3 (Table 1 ). Two individuals trained in the interpretation of ECGs scored the ECG independently. Both were blinded to the / scan interpretation of percentage of the lung involved, although they were aware that the scan result had previously been categorized as high probability. Data collection and the index test were performed retrospectively between November 2002 and January 2003.
Between 1997 and 2002, all / scans at Christchurch Hospital were interpreted by one of two experienced nuclear physicians and classified prospectively at the time of the scan, according to the modified Prospective Investigation of Pulmonary Embolism Diagnosis7 criteria. Two physicians experienced in the interpretation of / scans then reevaluated all scans classified as high probability independently. Both were blinded to the ECG score. Scans were categorized into scintigraphic extent of PE: < 30%, 30 to 50%, and > 50% of perfusion defect of the total lung area by the semiquantitative methods described previously.5 No other clinical information was available to the readers, and when a disagreement occurred a consensus was taken. Ventilation studies were performed using an aerosol technique with 99mTc diethylene triamine penta-acetic acid. Perfusion imaging was performed using 99mTc macroaggregated albumin technique. On both the ventilation and perfusion studies, six standard views were obtained with a gamma camera.
Statistical Methods
Analysis of variance (ANOVA) and 
2 tests were used to compare ECG scores and clinical characteristics between percentage perfusion defect groups as appropriate. For nominal variables, the 2 test was used. Interrater reliability was assessed by the Cronbach for the ECG scores and for the perfusion subgroups. The diagnostic accuracy of the ECG score in predicting those with > 50% perfusion defect was expressed as a receiver operator characteristic (ROC) curve. This enabled sensitivities and specificities to be calculated for different ECG scores.
|
Results
|
|---|
During the study period, 319 high-probability / scan findings were found in those investigated for PE at Christchurch Hospital (Fig 1
). Ninety scans were excluded from study: no ECG was recorded within 48 h of the scan in 62 cases, case notes could not be found in 21 cases, and the / scan could not be located in 7 cases. This left 229 cases (72%) in which both the / scan and ECG could be evaluated.
Clinical Characteristics
The clinical characteristics of the study group were described by the presence or absence of established risk factors for PE, noted in the case records prior to investigation for PE (Table 2 ). There was no significant difference between excluded cases where the clinical records could be found and those included in the study (all characteristics p > 0.05). Clinical characteristics between / scan categories were similar, except that those with prior venous thromboembolic disease were more likely to have the greatest percentage perfusion defect (24% < 30% perfusion defect, 33% 30 to 50% perfusion defect, and 43% > 50% perfusion defect, 2 p = 0.04).
Percentage Perfusion Defects on / Scan
By consensus, 78 scans, 84 scans, and 67 scans were categorized as < 30%, 30 to 50%, and > 50% perfusion defect, respectively (Table 2). The level of interobserver agreement for categorization of the / scan result was moderate ( = 0.55). Only 3 of a total of 67 disagreements occurred when one observer categorized the scan as < 30% and the other observer as > 50% perfusion defect.
ECG Scoring
The ECG scores differed significantly between percentage perfusion defect groups (mean ECG score in those with > 50% perfusion defect vs those with < 30% perfusion defect and those with 30 to 50% perfusion defect; ANOVA, p < 0.001; Fig 2
). The mean ECG scores were 2.6 (SD 2.8) in patients with < 30% perfusion defect, 3.2 (SD 2.9) in patients with 30 to 50% perfusion defect, and 5.3 (SD 3.7) in patients with > 50% perfusion defect. Interobserver agreement for ECG score was excellent (Cronbach = 0.96).
In order to assess the ability of the mean ECG score to predict those with > 50% perfusion defect, a ROC curve was constructed (Fig 3
). The area under the curve was 0.71 (SE 0.04). An ECG score of 3 predicted those with > 50% perfusion defect with a sensitivity of 70% (95% confidence interval [CI], 59 to 81%) and a specificity of 59% (95% CI, 51 to 67%). An ECG score of 6 was 40% sensitive (95% CI, 28 to 52%) and 90% specific (95% CI, 85 to 95%) in predicting those cases with > 50% perfusion defect.
|
Discussion
|
|---|
This study shows that a simple ECG scoring system correlates well with the percentage of the pulmonary vascular system obstructed on the basis of the defects on the / in the presence of PE. An ECG score of 3 indicated those with > 50% perfusion defect on a high-probability lung scan with a sensitivity of 70% (95% CI, 59 to 81%) and specificity of 59% (95% CI, 51 to 67%). In addition, the ECG scoring system has minimal interobserver error (Cronbach = 0.96) and could be applied at initial assessment of those with suspected PE.
There are several limitations of the present study. Firstly, patients were not tested prospectively. However, the only bias introduced was that all patients who were screened had already had a lung scan showing high probability for pulmonary embolus. Unfortunately, 90 of 319 of those screened (28%) could not be included because of missing data; again we do not feel significant bias was introduced, as the clinical characteristics of 69 of those excluded were no different. Our reference standard for quantifying the perfusion defects due to PE could be criticized, as the interobserver agreement between scans was only moderate ( = 0.55). However, only 3 of 67 disagreements occurred between < 30% perfusion defect and > 50% perfusion defect, and we feel that the level of disagreement is acceptable given the semiquantitative method of determining perfusion defect. It must be emphasized that only those with a high-probability lung scan were studied, and the ECG score has yet to be assessed in all those investigated for PE. The ECG score is likely to have its own limitations. The score is based largely on T-wave inversion, which can vary in character, may be secondary to right bundle-branch block, and are frequently found in leads V1 and V3 in normal ECGs.
Although the management of those with suspected PE requires that an initial assessment of adverse risks be made,6 there is no agreed method of defining the severity of PE. In the literature, the definition of PE severity ranges from the subjective description of symptoms8 to the more objective determination of right ventricular dysfunction by echocardiography.1 However, although right ventricular dysfunction has been associated with an increase in mortality,9 the availability of echocardiography as a prognostic tool may be limited. Consequently, a more accessible and rapid method of determining the severity of PE would be highly desirable.
The utility of an ECG-based predictive rule in the diagnosis of PE is limited, in part due to poor positive predictive value and sensitivity.10 However, the prevalence of ECG abnormalities are greater in those with the most extensive amount of embolization found on imaging,1112 and consequently a normal ECG is rarely found in those with extensive PE.12 The ECG scoring system used in the present study has been shown to predict those with pulmonary artery pressures of > 50 mm Hg.2 This is the first published record of the utility of the ECG scoring system in a large cohort of patients investigated for PE by lung scintigraphy. An example of an ECG showing the findings associated with > 50% perfusion defect is shown in Figure 4
. Our data demonstrate that the ECG score predicts those with the greatest amount of perfusion defects (> 50% perfusion defect vs 30 to 50% perfusion defect and < 30% perfusion defect; ANOVA, p < 0.001). However, it is less able to differentiate those with < 30% perfusion defect vs those with 30 to 50% perfusion defect due to PE. The low ECG score in latter groups is likely to reflect compensatory cardiac mechanisms that limit right ventricular dysfunction in those with low clot burden. We would hypothesize that once the majority of the pulmonary vascular bed is occluded by thromboemboli, cardiac decompensation may occur with resultant ECG abnormality and therefore a higher ECG score. A correlation should be sought between echocardiographic changes of right ventricular dysfunction and echocardiography-diagnosed pulmonary hypertension and ECG changes in future studies.
In conclusion, the ECG score predicts those with the greatest percentage of perfusion defects due to PE and is highly reproducible. An ECG score of > 6 is strongly suggestive of > 50% perfusion defect. Although the diagnostic accuracy of the ECG score does not allow identification of all those with > 50% perfusion defect, we feel that it may be of use in combination with other clinical parameters and warrants prospective evaluation.
|
Footnotes
|
|---|
Abbreviations: ANOVA = analysis of variance; CI = confidence interval; PE = pulmonary embolism; ROC = receiver operating characteristic; / = ventilation/perfusion
Dr. Iles is supported by an educational grant from GlaxoSmithKline New Zealand Ltd.
Received for publication June 18, 2003.
Accepted for publication December 18, 2003.
|
References
|
|---|
- Konstantinides, S, Geibel, A, Heusel, G, et al (2002) Heparin plus alteplase compared with heparin alone in patients with submassive pulmonary embolism. N Engl J Med 347,1143-1150[Abstract/Free Full Text]
- Yacovella, T, Alter, M Anticoagulation for venous thromboembolism: what are the current options? Postgrad Med 2000;108,43-46,51–54[CrossRef]
- Daniel, KR, Courtney, DM, Kline, JA Assessment of cardiac stress from massive pulmonary embolism with 12-lead ECG. Chest 2001;120,474-481[Abstract/Free Full Text]
- Burkill, GJ, Bell, JR, Padley, SP Survey on the use of pulmonary scintigraphy, spiral CT and conventional pulmonary angiography for suspected pulmonary embolism in the British Isles. Clin Radiol 1999;54,807-810[CrossRef][ISI][Medline]
- Galle, C, Papazyan, JP, Miron, MJ, et al Prediction of pulmonary embolism extent by clinical findings, d-dimer level and deep vein thrombosis shown by ultrasound. Thromb Haemost 2001;86,1156-1160[Medline]
- British Thoracic Society guidelines for the management of suspected acute pulmonary embolism. Thorax 2003;58,470-483[Free Full Text]
- The PIOPED Investigators. Value of the ventilation/perfusion scan in acute pulmonary embolism: results of the Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED). JAMA 1990;263,2753-2759[Abstract]
- Janata, K, Holzer, M, Laggner, AN, et al Cardiac troponin T in the severity assessment of patients with pulmonary embolism: cohort study. BMJ 2003;326,312-313[Free Full Text]
- Ribeiro, A, Lindmarker, P, Juhlin-Dannfelt, A, et al Echocardiography Doppler in pulmonary embolism: right ventricular dysfunction as a predictor of mortality rate. Am Heart J 1997;134,479-487[CrossRef][ISI][Medline]
- Rodger, M, Makropoulos, D, Turek, M, et al Diagnostic value of the electrocardiogram in suspected pulmonary embolism. Am J Cardiol 2000;86,807-809,A810[CrossRef][ISI][Medline]
- Stein, PD, Dalen, JE, McIntyre, KM, et al The electrocardiogram in acute pulmonary embolism. Prog Cardiovasc Dis 1975;17,247-257[CrossRef][ISI][Medline]
- Ferrari, E, Imbert, A, Chevalier, T, et al The ECG in pulmonary embolism: predictive value of negative T waves in precordial leads; 80 case reports. Chest 1997;111,537-543[Abstract/Free Full Text]
TOP
Abstract
Introduction
