- Email: [email protected]
Incremental value of exercise echocardiography over exercise electrocardiography in a chest pain unit
EJINME-02985; No of Pages 6 European Journal of Internal Medicine xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
European Journal of Internal Medicine journal homepage: www.elsevier.com/locate/ejim
Original Article
Incremental value of exercise echocardiography over exercise electrocardiography in a chest pain unit Alberto Bouzas-Mosquera a,⁎, Jesús Peteiro a, Francisco J. Broullón b, Nemesio Álvarez-García a, Jorge L. Rodríguez-Garrido a, Víctor X. Mosquera c, Dolores Martínez a, Juan C. Yáñez a, José M. Vázquez-Rodríguez a a b c
Department of Cardiology, Hospital Universitario A Coruña, A Coruña, Spain Department of Health Information Technology, Hospital Universitario A Coruña, A Coruña, Spain Department of Cardiac Surgery, Hospital Universitario A Coruña, A Coruña, Spain
a r t i c l e
i n f o
Article history: Received 19 March 2015 Received in revised form 29 July 2015 Accepted 5 August 2015 Available online xxxx Keywords: Exercise echocardiography Exercise electrocardiography Prognosis Chest pain unit
a b s t r a c t Background: Limited data are available on the added value of exercise echocardiography (ExEcho) over exercise electrocardiography (ExECG) in patients with suspected acute coronary syndromes (ACS) referred to a chest pain unit. We aimed to assess the incremental value of ExEcho over ExECG in this setting. Methods: ExECG and ExEcho were performed in parallel in 1052 patients with suspected ACS, nondiagnostic but interpretable electrocardiograms, and negative serial troponin results. The primary outcome was a composite of coronary death, nonfatal myocardial infarction or unstable angina with angiographic documentation of significant coronary artery disease within 6 months. Results: The primary outcome occurred in 2/614 patients (0.3%) with both negative ExECG and ExEcho, 3/60 (5%) with positive ExECG and negative ExEcho, 73/135 (54.1%) with negative ExECG and positive ExEcho, 106/136 (77.9%) with both positive ExECG and ExEcho, and 8/107 (7.5%) with inconclusive results. The addition of ExEcho data to a model based on clinical and ExECG data significantly increased the c statistic from 0.898 to 0.968 (change +0.070, 95% confidence interval 0.052–0.092), with a continuous net reclassification improvement of 1.56 and an integrated discrimination improvement of 22% (p b 0.001). Decision curve analysis showed that a strategy of referral to coronary angiography based on ExEcho was associated with the highest net benefit and with the largest reduction in unnecessary coronary angiographies. Conclusion: ExEcho provides significant incremental prognostic information and higher net clinical benefit than a strategy based on ExECG in patients referred to a chest pain unit for suspected ACS and negative troponin levels. © 2015 European Federation of Internal Medicine. Published by Elsevier B.V. All rights reserved.
1. Introduction Chest pain is one of the most common causes of presentation to the emergency department [1]. Many of these patients are hospitalized for a possible acute coronary syndrome at a significant cost [2]. However, a cardiac etiology is eventually found in less than one third of these patients [3]. Although it is important to reduce unnecessary admissions, patients inappropriately discharged with an unnoticed acute coronary syndrome have a significantly worse prognosis; the mortality rates for patients with missed unstable angina (UA) have been reported to be more than twice as high as for those who are admitted and treated [4], and this constitutes a major potential source of malpractice liability in the emergency department [5]. ⁎ Corresponding author at: Department of Cardiology, Hospital Universitario A Coruña, As Xubias, 84. 15006, A Coruña, Spain. Tel.: +34 981178184; fax: +34 981178258. E-mail address: [email protected] (A. Bouzas-Mosquera).
Exercise testing soon after admission can help establish the safety of discharge. Exercise electrocardiography (ExECG) has been the preferred initial noninvasive test in this setting [6], but its lower accuracy, as compared with noninvasive imaging techniques, may have prognostic implications [7]. Exercise echocardiography (ExEcho) is an add-on to ExECG that provides several advantages. Its diagnostic accuracy is similar to myocardial perfusion imaging [8], but ExEcho is more versatile, faster and safer; it does not involve the use of radiation, and has a significantly lower cost. However, there is scarce data on the added value of ExEcho over ExECG in the setting of a chest pain unit. Furthermore, an improvement in predictive accuracy is not sufficient to establish whether ExEcho would actually benefit patients, since the consequences of clinical decisions, such as unnecessary coronary angiographies or inappropriate discharges of patients with missed UA, have to be taken into account. Thus, our aim was to evaluate the incremental value of ExEcho over ExECG and its ability to improve clinical decision-making in patients
http://dx.doi.org/10.1016/j.ejim.2015.08.002 0953-6205/© 2015 European Federation of Internal Medicine. Published by Elsevier B.V. All rights reserved.
Please cite this article as: Bouzas-Mosquera A, et al, Incremental value of exercise echocardiography over exercise electrocardiography in a chest pain unit, Eur J Intern Med (2015), http://dx.doi.org/10.1016/j.ejim.2015.08.002
2
A. Bouzas-Mosquera et al. / European Journal of Internal Medicine xxx (2015) xxx–xxx
Fig. 1. Flow chart of patients enrolled in the study. ACS denotes acute coronary syndrome.
referred to a chest pain unit for acute chest pain, nondiagnostic electrocardiograms and negative troponin levels. 2. Methods 2.1. Patients Patients who were referred to our chest pain unit from July 2007 to December 2012 and underwent treadmill ExEcho were initially considered for inclusion in the study. Eligibility criteria were as follows: nontraumatic acute chest pain suspected of having an ischemic origin (in the absence of any obvious alternative cause), non-diagnostic but interpretable electrocardiograms, normal serial troponin levels, and ability to exercise on a treadmill. Patients with repolarization abnormalities precluding a proper interpretation of ExECG (i.e., left bundle branch block, preexcitation, paced rhythm, left ventricular hypertrophy with
strain, other repolarization abnormalities or treatment with digoxin) and those with at least one cardiac troponin I value above the diagnostic threshold for myocardial necrosis were not included. Patients with a history of coronary artery bypass grafting and those with known, significant, unrevascularized coronary stenoses were excluded (Fig. 1). Patients with any missing covariate or outcome data were also excluded (Fig. 1); the baseline characteristics of the latter did not differ significantly from those of the remaining subjects. The final study population consisted of 1052 patients. The research protocol was approved by the Comité Autonómico de Ética da Investigación de Galicia, our regional Ethics Committee. 2.2. Clinical and laboratory data The work-up in the emergency department consisted of clinical history, 12-lead electrocardiogram, chest X-ray, and at least 2 cardiac
Table 1 Baseline characteristics of 1052 patients with acute chest pain in the whole cohort and according to the subsequent development of coronary events.
Male, n (%) Age (years) Smokers, n (%) Diabetics, n (%) Hypertension, n (%) Hypercholesterolemia, n (%) Family history of CAD, n (%) Prior MI, n (%) Prior coronary revascularization, n (%) Type of chest pain Typical angina, n (%) Atypical angina, n (%) Nonischemic chest pain, n (%) Medications β-Blockers, n (%) Calcium channel blockers, n (%) Nitrites, n (%) RAAS blockers, n (%) SBP, mm Hg Heart rate, bpm Left ventricular ejection fraction, %
All patients (n = 1052)
Without primary outcome (n = 860)
With primary outcome (n = 192)
p
675 (64.2) 61.7 ± 12.5 251 (23.9) 196 (18.7) 558 (53.0) 542 (51.5) 72 (6.8) 204 (19.4) 241 (22.9)
518 (60.2) 61.1 ± 12.6 195 (22.7) 140 (16.3) 444 (51.6) 425 (49.4) 54 (6.3) 155 (18.0) 176 (20.5)
157 (81.8) 64.4 ± 11.9 56 (29.2) 56 (29.2) 114 (59.4) 117 (60.9) 18 (9.4) 49 (25.5) 65 (33.9)
b0.001 0.001 0.06 b0.001 0.05 0.004 0.20 0.02 b0.001
109 (10.4) 441 (41.9) 493 (46.8)
51 (5.9) 348 (40.5) 455 (52.9)
58 (30.2) 93 (48.4) 38 (19.8)
b0.001 0.04 b0.001
259 (24.6) 78 (7.4) 56 (5.3) 381 (36.3) 128 ± 27 79 ± 16 60.9 ± 5.7
194 (22.6) 64 (7.4) 33 (3.8) 302 (35.1) 127 ± 28 80 ± 16 61.2 ± 5.5
65 (33.9) 14 (7.3) 23 (12.0) 79 (41.1) 131 ± 20 76 ± 13 59.5 ± 6.4
0.001 0.78 b0.001 0.12 0.12 0.001 b0.001
bpm, beats per minute; CABG, coronary artery bypass grafting; CAD, coronary artery disease; MI, myocardial infarction; PCI, percutaneous coronary intervention; RAAS, renin–angiotensin–aldosterone system; SBP, systolic blood pressure.
Please cite this article as: Bouzas-Mosquera A, et al, Incremental value of exercise echocardiography over exercise electrocardiography in a chest pain unit, Eur J Intern Med (2015), http://dx.doi.org/10.1016/j.ejim.2015.08.002
A. Bouzas-Mosquera et al. / European Journal of Internal Medicine xxx (2015) xxx–xxx
3
Table 2 Testing results in the whole cohort and in patients with and without the primary outcome.
Submaximal (b85% MAPHR) Peak SBP, mm Hg Peak heart rate, bpm Peak RPP, ×103 mm Hg bpm Exercise workload, METs Exercise-induced chest pain, n (%) Positive ExECG, n (%) Positive ExEcho, n (%)
All patients (n = 1052)
Without primary outcome (n = 860)
With primary outcome (n = 192)
p
193 (18.3) 159 ± 27 148 ± 20 23.5 ± 5.5 9.7 ± 3.1 175 (16.6) 196 (18.6) 269 (25.6)
126 (14.7) 160 ± 27 150 ± 19 24.1 ± 5.4 10.0 ± 3.1 72 (8.4) 87 (10.1) 89 (10.3)
67 (34.9) 151 ± 25 138 ± 20 21.1 ± 5.2 8.3 ± 2.9 103 (53.6) 109 (56.8) 180 (93.8)
b0.001 b0.001 b0.001 b0.001 b0.001 b0.001 b0.001 b0.001
bpm, beats per minute; ExECG, exercise electrocardiography; ExEcho, exercise echocardiography; MAPHR, maximum age-predicted heart rate; METs, metabolic equivalents; RPP, ratepressure product; SBP, systolic blood pressure.
troponin I determinations, i.e., on arrival of the patient at the emergency department and ≥6 h later. Demographics, clinical data and stress testing results were entered in our database in a prospective fashion. Chest pain was classified as typical angina, atypical/probable angina and nonischemic chest pain as previously described [9]. β-Blockers were withdrawn from the time of admission to the emergency department. The diagnostic thresholds for cardiac troponin assays followed recommendations in effect at the time of use [10,11]. 2.3. Exercise stress testing Exercise stress testing was performed the following working morning after admission to the emergency department. All tests were carried out on a treadmill, and ExECG and ExEcho were performed in parallel. Blood pressure measurements and 12-lead electrocardiograms were obtained at baseline and at each stage of the exercise protocol. A submaximal test was defined as the inability to achieve ≥85% of the agepredicted maximum heart rate [12]. A positive ExECG was defined as the development of ST-segment deviation of ≥ 1 mm which was horizontal or sloping away from the isoelectric line 80 ms after the J point. ExEcho was performed as previously described [7,13]. A positive ExEcho was defined as the appearance of new or worsening wall motion abnormalities with exercise, except exercise-induced dyskinesia of an akinetic segment or isolated hypokinesia of the basal inferior or inferoseptal segments [14]. The tests were considered negative in the absence of exercise-induced electrocardiographic and echocardiographic criteria for positivity at ≥85% of maximum age-predicted heart rate. Otherwise, the tests were considered inconclusive. Subsequent diagnostic and therapeutic decisions were left to the discretion of treating physicians, who were aware of tests results. 2.4. Follow-up and end-points Follow-up data were gathered from hospital databases, medical records, death certificates and phone interviews. The primary outcome
was a composite of coronary events, namely coronary death, nonfatal myocardial infarction, or unstable angina warranting coronary angiography and with angiographic documentation of significant coronary artery disease, which were ascertained during the index hospital stay or within the following 6 months after the stress tests. Coronary death was defined as death due to an acute myocardial infarction or any death not attributable to any identified noncoronary cause. Myocardial infarction was defined as the appearance of new symptoms of myocardial ischemia or ischemic ECG changes accompanied by an abnormal increase in cardiac troponin levels. Coronary angiographies were performed at the discretion of the attending physicians. A significant coronary stenosis was defined as a ≥ 50% lumen stenosis of the left main coronary artery or a ≥70% diameter stenosis of any other epicardial coronary artery. 2.5. Statistical analysis Categorical variables are expressed as frequencies with percentages and comparison between groups based on the χ2 test. Continuous variables are presented as mean ± standard deviation, and intergroup differences were assessed with the t-Student or the Mann–Whitney test, as appropriate. Multivariable associations of the different variables with outcomes were assessed with logistic regression models. To evaluate the added prognostic value of ExECG and ExEcho, 3 logistic regression models were fitted. The first model was based on clinical data (i.e., sex, age, cardiovascular risk factors, history of myocardial infarction, prior coronary revascularization, typical angina) and resting left ventricular ejection fraction (LVEF). The second model consisted in adding ExECG data (i.e., exercise-induced chest pain, ischemic electrocardiographic abnormalities, exercise workload and rate-pressure product). In a third model, ExEcho results (i.e., echocardiographic myocardial ischemia) were incorporated. Overall performance of the models was assessed by Nagelkerke's R2 and Brier's score, and calibration was evaluated by the Hosmer–Lemeshow goodness-of-fit test and calibration plots.
Table 3 Outcomes according to exercise electrocardiography and exercise echocardiography results. Outcomes within 6 months
ExECG(−) ExEcho(−) (n = 614)
ExECG(+) ExEcho(−) (n = 60)
ExECG(−) ExEcho(+) (n = 135)
ExECG(+) ExEcho(+) (n = 136)
Inconclusive test results (n = 107)
Primary outcome, n (%) All-cause mortality, n (%) Coronary death, n (%) Myocardial infarction, n (%)a Unstable angina, n (%)b Overall During index admission After discharge Coronary revascularization, n (%) Overall During index admission After discharge
2 (0.3) 1 (0.2) 0 (0) 0 (0)
3 (5) 0 (0) 0 (0) 0 (0)
73 (54.1) 1 (0.7) 0 (0) 1 (0.7)
106 (77.9) 0 (0) 0 (0) 4 (2.9)
8 (7.5) 0 (0) 0 (0) 0 (0)
2 (0.3) 0 (0) 2 (0.3)
3 (5) 2 (3.3) 1 (1.7)
72 (53.3) 66 (48.9) 6 (4.4)
102 (75) 93 (68.4) 9 (6.6)
8 (7.5) 6 (5.6) 2 (1.9)
2 (0.3) 0 (0) 2 (0.3)
3 (5) 2 (3.3) 1 (1.7)
63 (46.7) 57 (42.2) 6 (4.4)
98 (72.1) 84 (61.8) 14 (10.3)
7 (6.5) 5 (4.7) 2 (1.9)
a b
Before any revascularization procedure. With angiographic evidence of significant coronary artery disease.
Please cite this article as: Bouzas-Mosquera A, et al, Incremental value of exercise echocardiography over exercise electrocardiography in a chest pain unit, Eur J Intern Med (2015), http://dx.doi.org/10.1016/j.ejim.2015.08.002
4
A. Bouzas-Mosquera et al. / European Journal of Internal Medicine xxx (2015) xxx–xxx
Table 4 Performance measures of models before and after the addition of exercise electrocardiography and exercise echocardiography data.
Likelihood ratio χ2 Brier score Brier scaled, % Nagelkerke's R2 (95% CI) C-statistic (95% CI) Discrimination slope Hosmer–Lemeshow test χ2 (p value)
Model 1 (Clinical + LVEF)
Model 2 (Clinical + LVEF + ExECG)
Model 3 (Clinical + LVEF + ExECG + ExEcho)
157 0.125 0.162 0.226 (0.151–0.271) 0.768 (0.725–0.800) 0.165 2.35 (0.97)
394 0.088 0.410 0.510 (0.428–0.557) 0.898 (0.870–0.917) 0.420 11.4 (0.18)
628 0.054 0.637 0.733 (0.669–0.769) 0.968 (0.956–0.976) 0.643 6.89 (0.55)
CI, confidence interval; ExECG, exercise electrocardiography; ExEcho, exercise echocardiography; LVEF, left ventricular ejection fraction.
Discrimination of each model was assessed by receiver operator characteristic (ROC) curves and expressed by the c statistic (equivalent to the area under the ROC curve). The 95% confidence intervals (CI) of c statistics were estimated with a nonparametric bootstrap procedure, for which two thousand bootstrap samples were drawn with replacement. The improvement in prediction power was determined by the likelihood ratio test. We also explored the change in c statistics between the models by estimating the 95% CI of their pairwise differences with bootstrapping following the aforementioned approach; a statistically significant difference was verified when intervals did not embrace zero. Improvement in risk prediction was also evaluated by the net reclassification improvement (NRI) and the integrated discrimination improvement (IDI) [15–17]. The predicted probabilities of the models were stratified into three risk categories arbitrarily defined as low (b10%), intermediate (10–40%), and high risk (N 40%). A reclassification table was built and the corresponding categorical NRI estimated based on these thresholds. The event NRI was calculated as the difference between the proportion of patients with an upward movement in their risk category minus the proportion moving down among those who developed events, and the nonevent NRI was estimated as the difference between the proportion of subjects moving down minus the proportion moving up for individuals without events [15]. The categorical NRI [denoted as NRI (0.1, 0.4)] was calculated as the sum of the event and nonevent NRI. Given the lack of data in the literature supporting the selection of specific risk thresholds for the endpoint evaluated in this study, NRI was also calculated for any change in risk estimate, i.e., the continuous (category-free) version of the NRI [NRI (N0)] [16]. The predicted probabilities of the models were plotted in predictiveness curves [18], which display the range and distribution of the estimated risks. Finally, clinical usefulness of ExEcho over ExECG was evaluated by the difference in net benefit for referral to coronary angiography over a range of threshold probabilities of the outcome by means of decision curve analysis [19,20], which accounts for harms of false positive and false negative results. The threshold probability was defined as the probability for coronary events at which the expected benefit of referral for coronary angiography was considered equal to the expected benefit of avoiding it. The net benefit was calculated as the probability of a true positive result minus the probability of a false positive result (resulting in an unnecessary coronary angiography) weighted by the odds of the threshold probability at which a patient or clinician would consider referral for coronary angiography. Decision curves consisted in plotting the net benefit of each strategy as a function of the probability threshold. The net reduction in unnecessary coronary angiographies without missing any acute coronary syndromes was estimated by the difference in net benefit of a model strategy minus the net benefit from a strategy based on referring all patients for coronary angiography, divided by the odds of the threshold probability. Statistical analyses were performed with R Statistical Software version 3.2.0 (R Foundation for Statistical Computing, Vienna, Austria), and statistical significance was set at p b 0.05.
3. Results 3.1. Baseline characteristics, exercise testing results and events Mean age was 61.7 years, and 64% of the patients were male. Baseline characteristics of the 1052 patients are shown in Table 1. The primary outcome occurred in 192 patients, who tended to be older, were more frequently male, and were more likely to have diabetes, hypertension, hypercholesterolemia, prior myocardial infarction or coronary revascularization, typical angina and lower LVEF (Table 1). Table 2 summarizes the data obtained during the tests according to the occurrence of the primary endpoint, and Table 3 shows the outcomes stratified according to ExECG and ExEcho results. The occurrence of the primary outcome was 10 times more frequent in patients with positive ExEcho and negative ExECG than in those with positive ExECG and negative ExEcho (Table 3). 3.2. Model performance, incremental value of stress testing and decision curves Table 4 shows the performance measures of the 3 multivariate models before and after the sequential addition of ExECG and ExEcho data. Calibration of all models was satisfactory (Table 4 and Supplementary Fig. 1). Table 5 summarizes the incremental value of the models. The addition of ExEcho data (model 3) to the model based on clinical data, resting LVEF and ExECG results (model 2) significantly increased the c statistic from 0.898 to 0.968 (change + 0.070, 95% CI 0.052–0.092) (Fig. 2 and Tables 4 and 5). Discrimination slopes also increased from 0.42 to 0.64 (Supplementary Fig. 2), leading to an IDI of 22% (p b 0.001). Supplementary Fig. 3 plots the reclassification of predicted risks of the different modeling steps; overall, 88% of the patients with the primary outcome had a higher predicted risk using the model with Table 5 Incremental value of testing.
Likelihood ratio statistic (p value) Change in Nagelkerke's R2 (95% CI) Change in c statistics (95% CI) NRI (0.1, 0.4) (95% CI) NRI (N0) (95% CI) IDI (95% CI)
Model 2 vs. model 1
Model 3 vs. model 2
237 (b0.001)
234 (b0.001)
+0.284 (0.220–0.346)
+0.223 (0.169–0.282)
+0.130 (0.102–0.165) 0.57 (0.47–0.68) 1.12 (0.98–1.25) 0.25 (0.21–0.29)
+0.070 (0.052–0.092) 0.41 (0.33–0.50) 1.56 (1.46–1.66) 0.22 (0.19–0.26)
Model 1: Clinical data + left ventricular ejection fraction. Model 2: Clinical data + left ventricular ejection fraction + exercise electrocardiography data. Model 3: Clinical data + left ventricular ejection fraction + exercise electrocardiography data + exercise echocardiography data. CI, confidence interval.
Please cite this article as: Bouzas-Mosquera A, et al, Incremental value of exercise echocardiography over exercise electrocardiography in a chest pain unit, Eur J Intern Med (2015), http://dx.doi.org/10.1016/j.ejim.2015.08.002
A. Bouzas-Mosquera et al. / European Journal of Internal Medicine xxx (2015) xxx–xxx
Fig. 2. Receiver operating characteristic curves for the predicted probabilities of the 3 logistic regression models. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
ExEcho data (model 3) as compared with the model based on ExECG (model 2), whereas 12% had a lower predicted risk [event NRI (N0) 76%]. Similarly, 90.1% of patients without events had a lower prediction with ExEcho data than without, while in 9.9% of these patients the predicted risk increased [nonevent NRI (N 0) 80.2%], yielding a continuous NRI of 1.56 (95% CI 1.46–1.66). Supplementary Table 1 shows the reclassification table when risk cutoffs of 10% and 40% are considered. Predictiveness curves for the different models are displayed in Supplementary Fig. 4. Applying the thresholds of 10% and 40% for definition of risk categories, the percentage of patients with predicted probabilities falling in the intermediate risk category was lower for the model based on ExEcho data (6.7%) than for the model based on ExECG (20.3%), thereby indicating that the former performed better in discriminating patients into low vs. high risk categories. Decision curves are depicted in Fig. 3, which shows that a strategy of referral to coronary angiography based on ExEcho was associated with the highest net benefit (Fig. 3A) and the largest reduction in unnecessary coronary angiographies (Fig. 3B) for the range of threshold probabilities of the outcome below 80%. 4. Discussion This study demonstrates that treadmill ExEcho provides significant incremental prognostic information over ExECG for the prediction of
5
coronary events in patients with acute chest pain and normal troponin levels referred to a chest pain unit, and also shows that a strategy of referral to coronary angiography based on ExEcho is associated with the highest net benefit and with the largest reduction in avoidable coronary angiographies. The superior performance of ExEcho as compared with ExECG may be explained in part by an earlier appearance of exercise-induced wall motion abnormalities in the ischemic cascade as compared with angina or ST-segment changes, which are the end points of treadmill ExECG. In addition, our group has previously shown that imaging acquisition during the peak of exercise on treadmill provides higher sensitivity for the detection of significant coronary artery disease than imaging in the immediate post-exercise period [21,22]. Decision curve analysis in our study demonstrated that a strategy of referral to coronary angiography based on ExEcho was superior to all alternative strategies in terms of net benefit and reduction in avoidable coronary angiographies throughout the range of plausible, clinically relevant threshold probabilities, on the grounds that it would be unusual for a patient to refuse coronary angiography if his predicted risk of the primary outcome was higher than 80%. This threshold probability would be equivalent to consider an unnecessary coronary angiography as 4 times more harmful than a missed acute coronary syndrome diagnosis, since the relative weight for false-positive vs. true-positive diagnoses is equal to the odds of the decision threshold. Clinical practice guidelines recommend exercise testing as a class I indication in patients with suspected acute coronary syndrome who have normal ECGs and troponins [23,24]. European guidelines also recommend echocardiography in these patients to evaluate left ventricular wall motion at rest and to rule out differential diagnoses (also a class I indication) [23]. Unlike ExECG and myocardial perfusion imaging, ExEcho allows ruling out some alternative causes for acute chest pain, such as valvular, pericardial or aortic diseases, and may also detect indirect evidence of pulmonary embolism through the assessment of right heart chambers and estimation of systolic pulmonary pressure. Thus, ExEcho fulfills guidelines requirements in a single test. If resting echocardiographic data are not available in addition to ExECG, the incremental value of ExEcho over the latter may be significantly underestimated. Treadmill ExEcho is actually an add-on to ExECG and, as such, provides the same diagnostic and prognostic information that can be obtained with the latter. In addition, its ability to avoid unnecessary downstream testing, as compared with ExECG, may render it a more cost-effective diagnostic tool in the setting of a chest pain unit [25].
Fig. 3. (A) Decision curves showing the net benefit of the 3 models as a function of the threshold probability for referral to coronary angiography. The y axis indicates the net benefit (i.e., the probability of a true positive result minus the weighted probability of a false positive result from an unnecessary coronary angiography). The x axis represents the threshold probability at which a patient or clinician would opt for coronary angiography. The horizontal grey line indicates the strategy of referring no patients to coronary angiography, which has no net benefit. The blue line represents the strategy of referring all patients to coronary angiography with the assumption that all patients will experience coronary events. The remaining 3 lines indicate the strategies based on the different prediction models. (B) Net reduction in unnecessary referrals for coronary angiography with each predictive model as compared with the strategy of referring all patients to coronary angiography. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
Please cite this article as: Bouzas-Mosquera A, et al, Incremental value of exercise echocardiography over exercise electrocardiography in a chest pain unit, Eur J Intern Med (2015), http://dx.doi.org/10.1016/j.ejim.2015.08.002
6
A. Bouzas-Mosquera et al. / European Journal of Internal Medicine xxx (2015) xxx–xxx
The value of stress echocardiography for predicting outcome in a chest pain unit has been previously assessed [26,27]. However, the preferred modality of testing in these studies was pharmacological stress echocardiography and, therefore, a direct evaluation of its incremental value over treadmill ExECG could not be performed. As compared with drug-induced stress, exercise is more physiological and safer, and provides data with well-established diagnostic and prognostic value which either cannot be obtained with pharmacological stress (such as exercise workload) [28] or do not have the same significance when triggered by drugs (such as stress-induced drop in blood pressure, ventricular arrhythmias, angina or ischemic ECG changes), hence the importance of using exercise as the stress of choice whenever possible [29]. ExEcho has also additional advantages over other competing noninvasive imaging modalities, such as SPECT; although the diagnostic and prognostic values of both techniques are similar [8,30], the former is significantly faster, more versatile, has lower cost and does not involve the use of radiation. Our study has the limitations inherent to an observational singlecenter study. Thus, some residual confounding cannot be excluded. Given that ExECG and ExEcho were performed simultaneously, observers interpreting the results of one test were not blinded to the results of the other; nonetheless, this ensures that both tests were performed under the exact same conditions. In addition, only a small proportion of patients underwent coronary angiography and, thus, the possibility of referral bias should be taken into account. 5. Conclusions In this study, ExEcho provided significant prognostic information for the prediction of coronary events over clinical data, LVEF and ExECG results in patients with acute chest pain, nondiagnostic resting electrocardiograms and negative troponin results referred to a chest pain unit. Decision curve analysis also showed that a strategy of referral to coronary angiography based on ExEcho was associated with the highest net benefit and with the greatest reduction in unnecessary coronary angiographies. Thus, ExEcho proved to be a more powerful tool to identify patients at risk in this setting. Further prospective studies, preferably randomized-controlled trials, are warranted to confirm the impact of this strategy on patient outcomes. Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.ejim.2015.08.002. Conflicts of interest The authors state that they have no conflicts of interest. Financial disclosures The authors declare none. References [1] Niska R, Bhuiya F, Xu J. National Hospital Ambulatory Medical Care Survey: 2007 emergency department summary. Natl Health Stat Rep 2010:1–31. [2] Lee TH, Goldman L. Evaluation of the patient with acute chest pain. N Engl J Med 2000;342:1187–95. [3] Blomkalns AL, Gibler WB. Chest pain unit concept: rationale and diagnostic strategies. Cardiol Clin 2005;23:411–21 [v]. [4] Pope JH, Aufderheide TP, Ruthazer R, et al. Missed diagnoses of acute cardiac ischemia in the emergency department. N Engl J Med 2000;342:1163–70. [5] Brown TW, McCarthy ML, Kelen GD, Levy F. An epidemiologic study of closed emergency department malpractice claims in a national database of physician malpractice insurers. Acad Emerg Med 2010;17:553–60.
[6] Amsterdam EA, Kirk JD, Bluemke DA, et al. Testing of low-risk patients presenting to the emergency department with chest pain: a scientific statement from the American Heart Association. Circulation 2010;122:1756–76. [7] Bouzas-Mosquera A, Peteiro J, Alvarez-Garcia N, et al. Prediction of mortality and major cardiac events by exercise echocardiography in patients with normal exercise electrocardiographic testing. J Am Coll Cardiol 2009;53:1981–90. [8] Heijenbrok-Kal MH, Fleischmann KE, Hunink MG. Stress echocardiography, stress single-photon-emission computed tomography and electron beam computed tomography for the assessment of coronary artery disease: a meta-analysis of diagnostic performance. Am Heart J 2007;154:415–23. [9] Diamond GA. A clinically relevant classification of chest discomfort. J Am Coll Cardiol 1983;1:574–5. [10] Apple FS, Wu AH, Jaffe AS. European Society of Cardiology and American College of Cardiology guidelines for redefinition of myocardial infarction: how to use existing assays clinically and for clinical trials. Am Heart J 2002;144:981–6. [11] Thygesen K, Alpert JS, Jaffe AS, et al. Third universal definition of myocardial infarction. J Am Coll Cardiol 2012;60:1581–98. [12] Makani H, Bangalore S, Halpern D, Makwana HG, Chaudhry FA. Cardiac outcomes with submaximal normal stress echocardiography: a meta-analysis. J Am Coll Cardiol 2012;60:1393–401. [13] Peteiro J, Bouzas-Mosquera A, Broullon FJ, Garcia-Campos A, Pazos P, Castro-Beiras A. Prognostic value of peak and post-exercise treadmill exercise echocardiography in patients with known or suspected coronary artery disease. Eur Heart J 2010;31: 187–95. [14] Sicari R, Nihoyannopoulos P, Evangelista A, et al. Stress echocardiography expert consensus statement: European Association of Echocardiography (EAE) (a registered branch of the ESC). Eur J Echocardiogr 2008;9:415–37. [15] Pencina MJ, D'Agostino Sr RB, D'Agostino Jr RB, Vasan RS. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med 2008;27:157–72. [16] Pencina MJ, D'Agostino Sr RB, Steyerberg EW. Extensions of net reclassification improvement calculations to measure usefulness of new biomarkers. Stat Med 2011; 30:11–21. [17] Leening MJ, Vedder MM, Witteman JC, Pencina MJ, Steyerberg EW. Net reclassification improvement: computation, interpretation, and controversies: a literature review and clinician's guide. Ann Intern Med 2014;160:122–31. [18] Pepe MS, Feng Z, Huang Y, et al. Integrating the predictiveness of a marker with its performance as a classifier. Am J Epidemiol 2008;167:362–8. [19] Vickers AJ, Elkin EB. Decision curve analysis: a novel method for evaluating prediction models. Med Decis Making 2006;26:565–74. [20] Steyerberg EW, Vickers AJ. Decision curve analysis: a discussion. Med Decis Making 2008;28:146–9. [21] Peteiro J, Fabregas R, Montserrat L, Alvarez N, Castro-Beiras A. Comparison of treadmill exercise echocardiography before and after exercise in the evaluation of patients with known or suspected coronary artery disease. J Am Soc Echocardiogr 1999;12:1073–9. [22] Peteiro J, Garrido I, Monserrat L, Aldama G, Calvino R, Castro-Beiras A. Comparison of peak and postexercise treadmill echocardiography with the use of continuous harmonic imaging acquisition. J Am Soc Echocardiogr 2004;17:1044–9. [23] Hamm CW, Bassand JP, Agewall S, et al. ESC Guidelines for the management of acute coronary syndromes in patients presenting without persistent ST-segment elevation: The Task Force for the management of acute coronary syndromes (ACS) in patients presenting without persistent ST-segment elevation of the European Society of Cardiology (ESC). Eur Heart J 2011;32:2999–3054. [24] Anderson JL, Adams CD, Antman EM, et al. 2012 ACCF/AHA focused update incorporated into the ACCF/AHA 2007 guidelines for the management of patients with unstable angina/non-ST-elevation myocardial infarction: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol 2013;61:e179–347. [25] Jeetley P, Burden L, Stoykova B, Senior R. Clinical and economic impact of stress echocardiography compared with exercise electrocardiography in patients with suspected acute coronary syndrome but negative troponin: a prospective randomized controlled study. Eur Heart J 2007;28:204–11. [26] Shah BN, Balaji G, Alhajiri A, Ramzy IS, Ahmadvazir S, Senior R. Incremental diagnostic and prognostic value of contemporary stress echocardiography in a chest pain unit: mortality and morbidity outcomes from a real-world setting. Circ Cardiovasc Imaging 2013;6:202–9. [27] Bedetti G, Pasanisi EM, Tintori G, et al. Stress echo in chest pain unit: the SPEED trial. Int J Cardiol 2005;102:461–7. [28] Peteiro J, Bouzas-Mosquera A, Broullon F, Martinez D, Yanez J, Castro-Beiras A. Value of an exercise workload ≥10 metabolic equivalents for predicting inducible myocardial ischemia. Circ Cardiovasc Imaging 2013;6:899–907. [29] Pellikka PA, Nagueh SF, Elhendy AA, Kuehl CA, Sawada SG. American Society of Echocardiography recommendations for performance, interpretation, and application of stress echocardiography. J Am Soc Echocardiogr 2007;20:1021–41. [30] Maganti K, Rigolin VH. Stress echocardiography versus myocardial SPECT for risk stratification of patients with coronary artery disease. Curr Opin Cardiol 2003;18: 486–93.
Please cite this article as: Bouzas-Mosquera A, et al, Incremental value of exercise echocardiography over exercise electrocardiography in a chest pain unit, Eur J Intern Med (2015), http://dx.doi.org/10.1016/j.ejim.2015.08.002
Contents lists available at ScienceDirect
European Journal of Internal Medicine journal homepage: www.elsevier.com/locate/ejim
Original Article
Incremental value of exercise echocardiography over exercise electrocardiography in a chest pain unit Alberto Bouzas-Mosquera a,⁎, Jesús Peteiro a, Francisco J. Broullón b, Nemesio Álvarez-García a, Jorge L. Rodríguez-Garrido a, Víctor X. Mosquera c, Dolores Martínez a, Juan C. Yáñez a, José M. Vázquez-Rodríguez a a b c
Department of Cardiology, Hospital Universitario A Coruña, A Coruña, Spain Department of Health Information Technology, Hospital Universitario A Coruña, A Coruña, Spain Department of Cardiac Surgery, Hospital Universitario A Coruña, A Coruña, Spain
a r t i c l e
i n f o
Article history: Received 19 March 2015 Received in revised form 29 July 2015 Accepted 5 August 2015 Available online xxxx Keywords: Exercise echocardiography Exercise electrocardiography Prognosis Chest pain unit
a b s t r a c t Background: Limited data are available on the added value of exercise echocardiography (ExEcho) over exercise electrocardiography (ExECG) in patients with suspected acute coronary syndromes (ACS) referred to a chest pain unit. We aimed to assess the incremental value of ExEcho over ExECG in this setting. Methods: ExECG and ExEcho were performed in parallel in 1052 patients with suspected ACS, nondiagnostic but interpretable electrocardiograms, and negative serial troponin results. The primary outcome was a composite of coronary death, nonfatal myocardial infarction or unstable angina with angiographic documentation of significant coronary artery disease within 6 months. Results: The primary outcome occurred in 2/614 patients (0.3%) with both negative ExECG and ExEcho, 3/60 (5%) with positive ExECG and negative ExEcho, 73/135 (54.1%) with negative ExECG and positive ExEcho, 106/136 (77.9%) with both positive ExECG and ExEcho, and 8/107 (7.5%) with inconclusive results. The addition of ExEcho data to a model based on clinical and ExECG data significantly increased the c statistic from 0.898 to 0.968 (change +0.070, 95% confidence interval 0.052–0.092), with a continuous net reclassification improvement of 1.56 and an integrated discrimination improvement of 22% (p b 0.001). Decision curve analysis showed that a strategy of referral to coronary angiography based on ExEcho was associated with the highest net benefit and with the largest reduction in unnecessary coronary angiographies. Conclusion: ExEcho provides significant incremental prognostic information and higher net clinical benefit than a strategy based on ExECG in patients referred to a chest pain unit for suspected ACS and negative troponin levels. © 2015 European Federation of Internal Medicine. Published by Elsevier B.V. All rights reserved.
1. Introduction Chest pain is one of the most common causes of presentation to the emergency department [1]. Many of these patients are hospitalized for a possible acute coronary syndrome at a significant cost [2]. However, a cardiac etiology is eventually found in less than one third of these patients [3]. Although it is important to reduce unnecessary admissions, patients inappropriately discharged with an unnoticed acute coronary syndrome have a significantly worse prognosis; the mortality rates for patients with missed unstable angina (UA) have been reported to be more than twice as high as for those who are admitted and treated [4], and this constitutes a major potential source of malpractice liability in the emergency department [5]. ⁎ Corresponding author at: Department of Cardiology, Hospital Universitario A Coruña, As Xubias, 84. 15006, A Coruña, Spain. Tel.: +34 981178184; fax: +34 981178258. E-mail address: [email protected] (A. Bouzas-Mosquera).
Exercise testing soon after admission can help establish the safety of discharge. Exercise electrocardiography (ExECG) has been the preferred initial noninvasive test in this setting [6], but its lower accuracy, as compared with noninvasive imaging techniques, may have prognostic implications [7]. Exercise echocardiography (ExEcho) is an add-on to ExECG that provides several advantages. Its diagnostic accuracy is similar to myocardial perfusion imaging [8], but ExEcho is more versatile, faster and safer; it does not involve the use of radiation, and has a significantly lower cost. However, there is scarce data on the added value of ExEcho over ExECG in the setting of a chest pain unit. Furthermore, an improvement in predictive accuracy is not sufficient to establish whether ExEcho would actually benefit patients, since the consequences of clinical decisions, such as unnecessary coronary angiographies or inappropriate discharges of patients with missed UA, have to be taken into account. Thus, our aim was to evaluate the incremental value of ExEcho over ExECG and its ability to improve clinical decision-making in patients
http://dx.doi.org/10.1016/j.ejim.2015.08.002 0953-6205/© 2015 European Federation of Internal Medicine. Published by Elsevier B.V. All rights reserved.
Please cite this article as: Bouzas-Mosquera A, et al, Incremental value of exercise echocardiography over exercise electrocardiography in a chest pain unit, Eur J Intern Med (2015), http://dx.doi.org/10.1016/j.ejim.2015.08.002
2
A. Bouzas-Mosquera et al. / European Journal of Internal Medicine xxx (2015) xxx–xxx
Fig. 1. Flow chart of patients enrolled in the study. ACS denotes acute coronary syndrome.
referred to a chest pain unit for acute chest pain, nondiagnostic electrocardiograms and negative troponin levels. 2. Methods 2.1. Patients Patients who were referred to our chest pain unit from July 2007 to December 2012 and underwent treadmill ExEcho were initially considered for inclusion in the study. Eligibility criteria were as follows: nontraumatic acute chest pain suspected of having an ischemic origin (in the absence of any obvious alternative cause), non-diagnostic but interpretable electrocardiograms, normal serial troponin levels, and ability to exercise on a treadmill. Patients with repolarization abnormalities precluding a proper interpretation of ExECG (i.e., left bundle branch block, preexcitation, paced rhythm, left ventricular hypertrophy with
strain, other repolarization abnormalities or treatment with digoxin) and those with at least one cardiac troponin I value above the diagnostic threshold for myocardial necrosis were not included. Patients with a history of coronary artery bypass grafting and those with known, significant, unrevascularized coronary stenoses were excluded (Fig. 1). Patients with any missing covariate or outcome data were also excluded (Fig. 1); the baseline characteristics of the latter did not differ significantly from those of the remaining subjects. The final study population consisted of 1052 patients. The research protocol was approved by the Comité Autonómico de Ética da Investigación de Galicia, our regional Ethics Committee. 2.2. Clinical and laboratory data The work-up in the emergency department consisted of clinical history, 12-lead electrocardiogram, chest X-ray, and at least 2 cardiac
Table 1 Baseline characteristics of 1052 patients with acute chest pain in the whole cohort and according to the subsequent development of coronary events.
Male, n (%) Age (years) Smokers, n (%) Diabetics, n (%) Hypertension, n (%) Hypercholesterolemia, n (%) Family history of CAD, n (%) Prior MI, n (%) Prior coronary revascularization, n (%) Type of chest pain Typical angina, n (%) Atypical angina, n (%) Nonischemic chest pain, n (%) Medications β-Blockers, n (%) Calcium channel blockers, n (%) Nitrites, n (%) RAAS blockers, n (%) SBP, mm Hg Heart rate, bpm Left ventricular ejection fraction, %
All patients (n = 1052)
Without primary outcome (n = 860)
With primary outcome (n = 192)
p
675 (64.2) 61.7 ± 12.5 251 (23.9) 196 (18.7) 558 (53.0) 542 (51.5) 72 (6.8) 204 (19.4) 241 (22.9)
518 (60.2) 61.1 ± 12.6 195 (22.7) 140 (16.3) 444 (51.6) 425 (49.4) 54 (6.3) 155 (18.0) 176 (20.5)
157 (81.8) 64.4 ± 11.9 56 (29.2) 56 (29.2) 114 (59.4) 117 (60.9) 18 (9.4) 49 (25.5) 65 (33.9)
b0.001 0.001 0.06 b0.001 0.05 0.004 0.20 0.02 b0.001
109 (10.4) 441 (41.9) 493 (46.8)
51 (5.9) 348 (40.5) 455 (52.9)
58 (30.2) 93 (48.4) 38 (19.8)
b0.001 0.04 b0.001
259 (24.6) 78 (7.4) 56 (5.3) 381 (36.3) 128 ± 27 79 ± 16 60.9 ± 5.7
194 (22.6) 64 (7.4) 33 (3.8) 302 (35.1) 127 ± 28 80 ± 16 61.2 ± 5.5
65 (33.9) 14 (7.3) 23 (12.0) 79 (41.1) 131 ± 20 76 ± 13 59.5 ± 6.4
0.001 0.78 b0.001 0.12 0.12 0.001 b0.001
bpm, beats per minute; CABG, coronary artery bypass grafting; CAD, coronary artery disease; MI, myocardial infarction; PCI, percutaneous coronary intervention; RAAS, renin–angiotensin–aldosterone system; SBP, systolic blood pressure.
Please cite this article as: Bouzas-Mosquera A, et al, Incremental value of exercise echocardiography over exercise electrocardiography in a chest pain unit, Eur J Intern Med (2015), http://dx.doi.org/10.1016/j.ejim.2015.08.002
A. Bouzas-Mosquera et al. / European Journal of Internal Medicine xxx (2015) xxx–xxx
3
Table 2 Testing results in the whole cohort and in patients with and without the primary outcome.
Submaximal (b85% MAPHR) Peak SBP, mm Hg Peak heart rate, bpm Peak RPP, ×103 mm Hg bpm Exercise workload, METs Exercise-induced chest pain, n (%) Positive ExECG, n (%) Positive ExEcho, n (%)
All patients (n = 1052)
Without primary outcome (n = 860)
With primary outcome (n = 192)
p
193 (18.3) 159 ± 27 148 ± 20 23.5 ± 5.5 9.7 ± 3.1 175 (16.6) 196 (18.6) 269 (25.6)
126 (14.7) 160 ± 27 150 ± 19 24.1 ± 5.4 10.0 ± 3.1 72 (8.4) 87 (10.1) 89 (10.3)
67 (34.9) 151 ± 25 138 ± 20 21.1 ± 5.2 8.3 ± 2.9 103 (53.6) 109 (56.8) 180 (93.8)
b0.001 b0.001 b0.001 b0.001 b0.001 b0.001 b0.001 b0.001
bpm, beats per minute; ExECG, exercise electrocardiography; ExEcho, exercise echocardiography; MAPHR, maximum age-predicted heart rate; METs, metabolic equivalents; RPP, ratepressure product; SBP, systolic blood pressure.
troponin I determinations, i.e., on arrival of the patient at the emergency department and ≥6 h later. Demographics, clinical data and stress testing results were entered in our database in a prospective fashion. Chest pain was classified as typical angina, atypical/probable angina and nonischemic chest pain as previously described [9]. β-Blockers were withdrawn from the time of admission to the emergency department. The diagnostic thresholds for cardiac troponin assays followed recommendations in effect at the time of use [10,11]. 2.3. Exercise stress testing Exercise stress testing was performed the following working morning after admission to the emergency department. All tests were carried out on a treadmill, and ExECG and ExEcho were performed in parallel. Blood pressure measurements and 12-lead electrocardiograms were obtained at baseline and at each stage of the exercise protocol. A submaximal test was defined as the inability to achieve ≥85% of the agepredicted maximum heart rate [12]. A positive ExECG was defined as the development of ST-segment deviation of ≥ 1 mm which was horizontal or sloping away from the isoelectric line 80 ms after the J point. ExEcho was performed as previously described [7,13]. A positive ExEcho was defined as the appearance of new or worsening wall motion abnormalities with exercise, except exercise-induced dyskinesia of an akinetic segment or isolated hypokinesia of the basal inferior or inferoseptal segments [14]. The tests were considered negative in the absence of exercise-induced electrocardiographic and echocardiographic criteria for positivity at ≥85% of maximum age-predicted heart rate. Otherwise, the tests were considered inconclusive. Subsequent diagnostic and therapeutic decisions were left to the discretion of treating physicians, who were aware of tests results. 2.4. Follow-up and end-points Follow-up data were gathered from hospital databases, medical records, death certificates and phone interviews. The primary outcome
was a composite of coronary events, namely coronary death, nonfatal myocardial infarction, or unstable angina warranting coronary angiography and with angiographic documentation of significant coronary artery disease, which were ascertained during the index hospital stay or within the following 6 months after the stress tests. Coronary death was defined as death due to an acute myocardial infarction or any death not attributable to any identified noncoronary cause. Myocardial infarction was defined as the appearance of new symptoms of myocardial ischemia or ischemic ECG changes accompanied by an abnormal increase in cardiac troponin levels. Coronary angiographies were performed at the discretion of the attending physicians. A significant coronary stenosis was defined as a ≥ 50% lumen stenosis of the left main coronary artery or a ≥70% diameter stenosis of any other epicardial coronary artery. 2.5. Statistical analysis Categorical variables are expressed as frequencies with percentages and comparison between groups based on the χ2 test. Continuous variables are presented as mean ± standard deviation, and intergroup differences were assessed with the t-Student or the Mann–Whitney test, as appropriate. Multivariable associations of the different variables with outcomes were assessed with logistic regression models. To evaluate the added prognostic value of ExECG and ExEcho, 3 logistic regression models were fitted. The first model was based on clinical data (i.e., sex, age, cardiovascular risk factors, history of myocardial infarction, prior coronary revascularization, typical angina) and resting left ventricular ejection fraction (LVEF). The second model consisted in adding ExECG data (i.e., exercise-induced chest pain, ischemic electrocardiographic abnormalities, exercise workload and rate-pressure product). In a third model, ExEcho results (i.e., echocardiographic myocardial ischemia) were incorporated. Overall performance of the models was assessed by Nagelkerke's R2 and Brier's score, and calibration was evaluated by the Hosmer–Lemeshow goodness-of-fit test and calibration plots.
Table 3 Outcomes according to exercise electrocardiography and exercise echocardiography results. Outcomes within 6 months
ExECG(−) ExEcho(−) (n = 614)
ExECG(+) ExEcho(−) (n = 60)
ExECG(−) ExEcho(+) (n = 135)
ExECG(+) ExEcho(+) (n = 136)
Inconclusive test results (n = 107)
Primary outcome, n (%) All-cause mortality, n (%) Coronary death, n (%) Myocardial infarction, n (%)a Unstable angina, n (%)b Overall During index admission After discharge Coronary revascularization, n (%) Overall During index admission After discharge
2 (0.3) 1 (0.2) 0 (0) 0 (0)
3 (5) 0 (0) 0 (0) 0 (0)
73 (54.1) 1 (0.7) 0 (0) 1 (0.7)
106 (77.9) 0 (0) 0 (0) 4 (2.9)
8 (7.5) 0 (0) 0 (0) 0 (0)
2 (0.3) 0 (0) 2 (0.3)
3 (5) 2 (3.3) 1 (1.7)
72 (53.3) 66 (48.9) 6 (4.4)
102 (75) 93 (68.4) 9 (6.6)
8 (7.5) 6 (5.6) 2 (1.9)
2 (0.3) 0 (0) 2 (0.3)
3 (5) 2 (3.3) 1 (1.7)
63 (46.7) 57 (42.2) 6 (4.4)
98 (72.1) 84 (61.8) 14 (10.3)
7 (6.5) 5 (4.7) 2 (1.9)
a b
Before any revascularization procedure. With angiographic evidence of significant coronary artery disease.
Please cite this article as: Bouzas-Mosquera A, et al, Incremental value of exercise echocardiography over exercise electrocardiography in a chest pain unit, Eur J Intern Med (2015), http://dx.doi.org/10.1016/j.ejim.2015.08.002
4
A. Bouzas-Mosquera et al. / European Journal of Internal Medicine xxx (2015) xxx–xxx
Table 4 Performance measures of models before and after the addition of exercise electrocardiography and exercise echocardiography data.
Likelihood ratio χ2 Brier score Brier scaled, % Nagelkerke's R2 (95% CI) C-statistic (95% CI) Discrimination slope Hosmer–Lemeshow test χ2 (p value)
Model 1 (Clinical + LVEF)
Model 2 (Clinical + LVEF + ExECG)
Model 3 (Clinical + LVEF + ExECG + ExEcho)
157 0.125 0.162 0.226 (0.151–0.271) 0.768 (0.725–0.800) 0.165 2.35 (0.97)
394 0.088 0.410 0.510 (0.428–0.557) 0.898 (0.870–0.917) 0.420 11.4 (0.18)
628 0.054 0.637 0.733 (0.669–0.769) 0.968 (0.956–0.976) 0.643 6.89 (0.55)
CI, confidence interval; ExECG, exercise electrocardiography; ExEcho, exercise echocardiography; LVEF, left ventricular ejection fraction.
Discrimination of each model was assessed by receiver operator characteristic (ROC) curves and expressed by the c statistic (equivalent to the area under the ROC curve). The 95% confidence intervals (CI) of c statistics were estimated with a nonparametric bootstrap procedure, for which two thousand bootstrap samples were drawn with replacement. The improvement in prediction power was determined by the likelihood ratio test. We also explored the change in c statistics between the models by estimating the 95% CI of their pairwise differences with bootstrapping following the aforementioned approach; a statistically significant difference was verified when intervals did not embrace zero. Improvement in risk prediction was also evaluated by the net reclassification improvement (NRI) and the integrated discrimination improvement (IDI) [15–17]. The predicted probabilities of the models were stratified into three risk categories arbitrarily defined as low (b10%), intermediate (10–40%), and high risk (N 40%). A reclassification table was built and the corresponding categorical NRI estimated based on these thresholds. The event NRI was calculated as the difference between the proportion of patients with an upward movement in their risk category minus the proportion moving down among those who developed events, and the nonevent NRI was estimated as the difference between the proportion of subjects moving down minus the proportion moving up for individuals without events [15]. The categorical NRI [denoted as NRI (0.1, 0.4)] was calculated as the sum of the event and nonevent NRI. Given the lack of data in the literature supporting the selection of specific risk thresholds for the endpoint evaluated in this study, NRI was also calculated for any change in risk estimate, i.e., the continuous (category-free) version of the NRI [NRI (N0)] [16]. The predicted probabilities of the models were plotted in predictiveness curves [18], which display the range and distribution of the estimated risks. Finally, clinical usefulness of ExEcho over ExECG was evaluated by the difference in net benefit for referral to coronary angiography over a range of threshold probabilities of the outcome by means of decision curve analysis [19,20], which accounts for harms of false positive and false negative results. The threshold probability was defined as the probability for coronary events at which the expected benefit of referral for coronary angiography was considered equal to the expected benefit of avoiding it. The net benefit was calculated as the probability of a true positive result minus the probability of a false positive result (resulting in an unnecessary coronary angiography) weighted by the odds of the threshold probability at which a patient or clinician would consider referral for coronary angiography. Decision curves consisted in plotting the net benefit of each strategy as a function of the probability threshold. The net reduction in unnecessary coronary angiographies without missing any acute coronary syndromes was estimated by the difference in net benefit of a model strategy minus the net benefit from a strategy based on referring all patients for coronary angiography, divided by the odds of the threshold probability. Statistical analyses were performed with R Statistical Software version 3.2.0 (R Foundation for Statistical Computing, Vienna, Austria), and statistical significance was set at p b 0.05.
3. Results 3.1. Baseline characteristics, exercise testing results and events Mean age was 61.7 years, and 64% of the patients were male. Baseline characteristics of the 1052 patients are shown in Table 1. The primary outcome occurred in 192 patients, who tended to be older, were more frequently male, and were more likely to have diabetes, hypertension, hypercholesterolemia, prior myocardial infarction or coronary revascularization, typical angina and lower LVEF (Table 1). Table 2 summarizes the data obtained during the tests according to the occurrence of the primary endpoint, and Table 3 shows the outcomes stratified according to ExECG and ExEcho results. The occurrence of the primary outcome was 10 times more frequent in patients with positive ExEcho and negative ExECG than in those with positive ExECG and negative ExEcho (Table 3). 3.2. Model performance, incremental value of stress testing and decision curves Table 4 shows the performance measures of the 3 multivariate models before and after the sequential addition of ExECG and ExEcho data. Calibration of all models was satisfactory (Table 4 and Supplementary Fig. 1). Table 5 summarizes the incremental value of the models. The addition of ExEcho data (model 3) to the model based on clinical data, resting LVEF and ExECG results (model 2) significantly increased the c statistic from 0.898 to 0.968 (change + 0.070, 95% CI 0.052–0.092) (Fig. 2 and Tables 4 and 5). Discrimination slopes also increased from 0.42 to 0.64 (Supplementary Fig. 2), leading to an IDI of 22% (p b 0.001). Supplementary Fig. 3 plots the reclassification of predicted risks of the different modeling steps; overall, 88% of the patients with the primary outcome had a higher predicted risk using the model with Table 5 Incremental value of testing.
Likelihood ratio statistic (p value) Change in Nagelkerke's R2 (95% CI) Change in c statistics (95% CI) NRI (0.1, 0.4) (95% CI) NRI (N0) (95% CI) IDI (95% CI)
Model 2 vs. model 1
Model 3 vs. model 2
237 (b0.001)
234 (b0.001)
+0.284 (0.220–0.346)
+0.223 (0.169–0.282)
+0.130 (0.102–0.165) 0.57 (0.47–0.68) 1.12 (0.98–1.25) 0.25 (0.21–0.29)
+0.070 (0.052–0.092) 0.41 (0.33–0.50) 1.56 (1.46–1.66) 0.22 (0.19–0.26)
Model 1: Clinical data + left ventricular ejection fraction. Model 2: Clinical data + left ventricular ejection fraction + exercise electrocardiography data. Model 3: Clinical data + left ventricular ejection fraction + exercise electrocardiography data + exercise echocardiography data. CI, confidence interval.
Please cite this article as: Bouzas-Mosquera A, et al, Incremental value of exercise echocardiography over exercise electrocardiography in a chest pain unit, Eur J Intern Med (2015), http://dx.doi.org/10.1016/j.ejim.2015.08.002
A. Bouzas-Mosquera et al. / European Journal of Internal Medicine xxx (2015) xxx–xxx
Fig. 2. Receiver operating characteristic curves for the predicted probabilities of the 3 logistic regression models. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
ExEcho data (model 3) as compared with the model based on ExECG (model 2), whereas 12% had a lower predicted risk [event NRI (N0) 76%]. Similarly, 90.1% of patients without events had a lower prediction with ExEcho data than without, while in 9.9% of these patients the predicted risk increased [nonevent NRI (N 0) 80.2%], yielding a continuous NRI of 1.56 (95% CI 1.46–1.66). Supplementary Table 1 shows the reclassification table when risk cutoffs of 10% and 40% are considered. Predictiveness curves for the different models are displayed in Supplementary Fig. 4. Applying the thresholds of 10% and 40% for definition of risk categories, the percentage of patients with predicted probabilities falling in the intermediate risk category was lower for the model based on ExEcho data (6.7%) than for the model based on ExECG (20.3%), thereby indicating that the former performed better in discriminating patients into low vs. high risk categories. Decision curves are depicted in Fig. 3, which shows that a strategy of referral to coronary angiography based on ExEcho was associated with the highest net benefit (Fig. 3A) and the largest reduction in unnecessary coronary angiographies (Fig. 3B) for the range of threshold probabilities of the outcome below 80%. 4. Discussion This study demonstrates that treadmill ExEcho provides significant incremental prognostic information over ExECG for the prediction of
5
coronary events in patients with acute chest pain and normal troponin levels referred to a chest pain unit, and also shows that a strategy of referral to coronary angiography based on ExEcho is associated with the highest net benefit and with the largest reduction in avoidable coronary angiographies. The superior performance of ExEcho as compared with ExECG may be explained in part by an earlier appearance of exercise-induced wall motion abnormalities in the ischemic cascade as compared with angina or ST-segment changes, which are the end points of treadmill ExECG. In addition, our group has previously shown that imaging acquisition during the peak of exercise on treadmill provides higher sensitivity for the detection of significant coronary artery disease than imaging in the immediate post-exercise period [21,22]. Decision curve analysis in our study demonstrated that a strategy of referral to coronary angiography based on ExEcho was superior to all alternative strategies in terms of net benefit and reduction in avoidable coronary angiographies throughout the range of plausible, clinically relevant threshold probabilities, on the grounds that it would be unusual for a patient to refuse coronary angiography if his predicted risk of the primary outcome was higher than 80%. This threshold probability would be equivalent to consider an unnecessary coronary angiography as 4 times more harmful than a missed acute coronary syndrome diagnosis, since the relative weight for false-positive vs. true-positive diagnoses is equal to the odds of the decision threshold. Clinical practice guidelines recommend exercise testing as a class I indication in patients with suspected acute coronary syndrome who have normal ECGs and troponins [23,24]. European guidelines also recommend echocardiography in these patients to evaluate left ventricular wall motion at rest and to rule out differential diagnoses (also a class I indication) [23]. Unlike ExECG and myocardial perfusion imaging, ExEcho allows ruling out some alternative causes for acute chest pain, such as valvular, pericardial or aortic diseases, and may also detect indirect evidence of pulmonary embolism through the assessment of right heart chambers and estimation of systolic pulmonary pressure. Thus, ExEcho fulfills guidelines requirements in a single test. If resting echocardiographic data are not available in addition to ExECG, the incremental value of ExEcho over the latter may be significantly underestimated. Treadmill ExEcho is actually an add-on to ExECG and, as such, provides the same diagnostic and prognostic information that can be obtained with the latter. In addition, its ability to avoid unnecessary downstream testing, as compared with ExECG, may render it a more cost-effective diagnostic tool in the setting of a chest pain unit [25].
Fig. 3. (A) Decision curves showing the net benefit of the 3 models as a function of the threshold probability for referral to coronary angiography. The y axis indicates the net benefit (i.e., the probability of a true positive result minus the weighted probability of a false positive result from an unnecessary coronary angiography). The x axis represents the threshold probability at which a patient or clinician would opt for coronary angiography. The horizontal grey line indicates the strategy of referring no patients to coronary angiography, which has no net benefit. The blue line represents the strategy of referring all patients to coronary angiography with the assumption that all patients will experience coronary events. The remaining 3 lines indicate the strategies based on the different prediction models. (B) Net reduction in unnecessary referrals for coronary angiography with each predictive model as compared with the strategy of referring all patients to coronary angiography. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
Please cite this article as: Bouzas-Mosquera A, et al, Incremental value of exercise echocardiography over exercise electrocardiography in a chest pain unit, Eur J Intern Med (2015), http://dx.doi.org/10.1016/j.ejim.2015.08.002
6
A. Bouzas-Mosquera et al. / European Journal of Internal Medicine xxx (2015) xxx–xxx
The value of stress echocardiography for predicting outcome in a chest pain unit has been previously assessed [26,27]. However, the preferred modality of testing in these studies was pharmacological stress echocardiography and, therefore, a direct evaluation of its incremental value over treadmill ExECG could not be performed. As compared with drug-induced stress, exercise is more physiological and safer, and provides data with well-established diagnostic and prognostic value which either cannot be obtained with pharmacological stress (such as exercise workload) [28] or do not have the same significance when triggered by drugs (such as stress-induced drop in blood pressure, ventricular arrhythmias, angina or ischemic ECG changes), hence the importance of using exercise as the stress of choice whenever possible [29]. ExEcho has also additional advantages over other competing noninvasive imaging modalities, such as SPECT; although the diagnostic and prognostic values of both techniques are similar [8,30], the former is significantly faster, more versatile, has lower cost and does not involve the use of radiation. Our study has the limitations inherent to an observational singlecenter study. Thus, some residual confounding cannot be excluded. Given that ExECG and ExEcho were performed simultaneously, observers interpreting the results of one test were not blinded to the results of the other; nonetheless, this ensures that both tests were performed under the exact same conditions. In addition, only a small proportion of patients underwent coronary angiography and, thus, the possibility of referral bias should be taken into account. 5. Conclusions In this study, ExEcho provided significant prognostic information for the prediction of coronary events over clinical data, LVEF and ExECG results in patients with acute chest pain, nondiagnostic resting electrocardiograms and negative troponin results referred to a chest pain unit. Decision curve analysis also showed that a strategy of referral to coronary angiography based on ExEcho was associated with the highest net benefit and with the greatest reduction in unnecessary coronary angiographies. Thus, ExEcho proved to be a more powerful tool to identify patients at risk in this setting. Further prospective studies, preferably randomized-controlled trials, are warranted to confirm the impact of this strategy on patient outcomes. Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.ejim.2015.08.002. Conflicts of interest The authors state that they have no conflicts of interest. Financial disclosures The authors declare none. References [1] Niska R, Bhuiya F, Xu J. National Hospital Ambulatory Medical Care Survey: 2007 emergency department summary. Natl Health Stat Rep 2010:1–31. [2] Lee TH, Goldman L. Evaluation of the patient with acute chest pain. N Engl J Med 2000;342:1187–95. [3] Blomkalns AL, Gibler WB. Chest pain unit concept: rationale and diagnostic strategies. Cardiol Clin 2005;23:411–21 [v]. [4] Pope JH, Aufderheide TP, Ruthazer R, et al. Missed diagnoses of acute cardiac ischemia in the emergency department. N Engl J Med 2000;342:1163–70. [5] Brown TW, McCarthy ML, Kelen GD, Levy F. An epidemiologic study of closed emergency department malpractice claims in a national database of physician malpractice insurers. Acad Emerg Med 2010;17:553–60.
[6] Amsterdam EA, Kirk JD, Bluemke DA, et al. Testing of low-risk patients presenting to the emergency department with chest pain: a scientific statement from the American Heart Association. Circulation 2010;122:1756–76. [7] Bouzas-Mosquera A, Peteiro J, Alvarez-Garcia N, et al. Prediction of mortality and major cardiac events by exercise echocardiography in patients with normal exercise electrocardiographic testing. J Am Coll Cardiol 2009;53:1981–90. [8] Heijenbrok-Kal MH, Fleischmann KE, Hunink MG. Stress echocardiography, stress single-photon-emission computed tomography and electron beam computed tomography for the assessment of coronary artery disease: a meta-analysis of diagnostic performance. Am Heart J 2007;154:415–23. [9] Diamond GA. A clinically relevant classification of chest discomfort. J Am Coll Cardiol 1983;1:574–5. [10] Apple FS, Wu AH, Jaffe AS. European Society of Cardiology and American College of Cardiology guidelines for redefinition of myocardial infarction: how to use existing assays clinically and for clinical trials. Am Heart J 2002;144:981–6. [11] Thygesen K, Alpert JS, Jaffe AS, et al. Third universal definition of myocardial infarction. J Am Coll Cardiol 2012;60:1581–98. [12] Makani H, Bangalore S, Halpern D, Makwana HG, Chaudhry FA. Cardiac outcomes with submaximal normal stress echocardiography: a meta-analysis. J Am Coll Cardiol 2012;60:1393–401. [13] Peteiro J, Bouzas-Mosquera A, Broullon FJ, Garcia-Campos A, Pazos P, Castro-Beiras A. Prognostic value of peak and post-exercise treadmill exercise echocardiography in patients with known or suspected coronary artery disease. Eur Heart J 2010;31: 187–95. [14] Sicari R, Nihoyannopoulos P, Evangelista A, et al. Stress echocardiography expert consensus statement: European Association of Echocardiography (EAE) (a registered branch of the ESC). Eur J Echocardiogr 2008;9:415–37. [15] Pencina MJ, D'Agostino Sr RB, D'Agostino Jr RB, Vasan RS. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med 2008;27:157–72. [16] Pencina MJ, D'Agostino Sr RB, Steyerberg EW. Extensions of net reclassification improvement calculations to measure usefulness of new biomarkers. Stat Med 2011; 30:11–21. [17] Leening MJ, Vedder MM, Witteman JC, Pencina MJ, Steyerberg EW. Net reclassification improvement: computation, interpretation, and controversies: a literature review and clinician's guide. Ann Intern Med 2014;160:122–31. [18] Pepe MS, Feng Z, Huang Y, et al. Integrating the predictiveness of a marker with its performance as a classifier. Am J Epidemiol 2008;167:362–8. [19] Vickers AJ, Elkin EB. Decision curve analysis: a novel method for evaluating prediction models. Med Decis Making 2006;26:565–74. [20] Steyerberg EW, Vickers AJ. Decision curve analysis: a discussion. Med Decis Making 2008;28:146–9. [21] Peteiro J, Fabregas R, Montserrat L, Alvarez N, Castro-Beiras A. Comparison of treadmill exercise echocardiography before and after exercise in the evaluation of patients with known or suspected coronary artery disease. J Am Soc Echocardiogr 1999;12:1073–9. [22] Peteiro J, Garrido I, Monserrat L, Aldama G, Calvino R, Castro-Beiras A. Comparison of peak and postexercise treadmill echocardiography with the use of continuous harmonic imaging acquisition. J Am Soc Echocardiogr 2004;17:1044–9. [23] Hamm CW, Bassand JP, Agewall S, et al. ESC Guidelines for the management of acute coronary syndromes in patients presenting without persistent ST-segment elevation: The Task Force for the management of acute coronary syndromes (ACS) in patients presenting without persistent ST-segment elevation of the European Society of Cardiology (ESC). Eur Heart J 2011;32:2999–3054. [24] Anderson JL, Adams CD, Antman EM, et al. 2012 ACCF/AHA focused update incorporated into the ACCF/AHA 2007 guidelines for the management of patients with unstable angina/non-ST-elevation myocardial infarction: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol 2013;61:e179–347. [25] Jeetley P, Burden L, Stoykova B, Senior R. Clinical and economic impact of stress echocardiography compared with exercise electrocardiography in patients with suspected acute coronary syndrome but negative troponin: a prospective randomized controlled study. Eur Heart J 2007;28:204–11. [26] Shah BN, Balaji G, Alhajiri A, Ramzy IS, Ahmadvazir S, Senior R. Incremental diagnostic and prognostic value of contemporary stress echocardiography in a chest pain unit: mortality and morbidity outcomes from a real-world setting. Circ Cardiovasc Imaging 2013;6:202–9. [27] Bedetti G, Pasanisi EM, Tintori G, et al. Stress echo in chest pain unit: the SPEED trial. Int J Cardiol 2005;102:461–7. [28] Peteiro J, Bouzas-Mosquera A, Broullon F, Martinez D, Yanez J, Castro-Beiras A. Value of an exercise workload ≥10 metabolic equivalents for predicting inducible myocardial ischemia. Circ Cardiovasc Imaging 2013;6:899–907. [29] Pellikka PA, Nagueh SF, Elhendy AA, Kuehl CA, Sawada SG. American Society of Echocardiography recommendations for performance, interpretation, and application of stress echocardiography. J Am Soc Echocardiogr 2007;20:1021–41. [30] Maganti K, Rigolin VH. Stress echocardiography versus myocardial SPECT for risk stratification of patients with coronary artery disease. Curr Opin Cardiol 2003;18: 486–93.
Please cite this article as: Bouzas-Mosquera A, et al, Incremental value of exercise echocardiography over exercise electrocardiography in a chest pain unit, Eur J Intern Med (2015), http://dx.doi.org/10.1016/j.ejim.2015.08.002