- Email: [email protected]
Prognosis of patients with vascular disease after clinical evaluation and dobutamine stress echocardiography
Prognosis of patients with vascular disease after clinical evaluation and dobutamine stress echocardiography Raj S. Ballal, MD, Samir Kapadia, MD, Maria-Anna Secknus, MD, David Rubin, MD, Kristopher Arheart, EdD, and Thomas H. Marwick, MD, PhD Cleveland, Ohio
Background Coronary disease is an important cause of long-term morbidity in patients needing major vascular surgery. We sought to assess the efficacy of preoperative clinical evaluation and the detection of inducible ischemia for prediction of immediate and long-term cardiac outcomes of patients undergoing vascular surgery.
Methods In 233 patients undergoing vascular procedures, we assessed risk clinically on the basis of Eagle’s criteria. Dobutamine echocardiography was performed with a standard protocol and results were classified as showing ischemia, scar, or a normal response. Patients were observed perioperatively, and late follow-up (28 ± 13 months) was completed in all surgical survivors. A composite end point of cardiac death, myocardial infarction, and unstable and progressive angina requiring late revascularization was used to judge event-free survival.
Results Of 233 patients undergoing preoperative dobutamine echocardiography, 39 (17%) had inducible ischemia and 36 (15%) had scar. Perioperative events occurred in 8 patients (3%). None of the patients with ischemia had perioperative events, reflecting the effect of revascularization in 9 patients. Late events occurred in 36 patients; ischemia on preoperative stress testing was a predictor of these events even after adjusting for clinical variables and left ventricular dysfunction (relative risk = 3.3; 95% confidence interval 1.6 to 6.8; P = .001). The association of ischemia with clinical predictors was associated with incrementally worse outcome.
Conclusion In addition to perioperative assessment, the combined use of clinical and dobutamine echocardiographic evaluation may stratify the risk of late cardiac events. (Am Heart J 1999;137:469-75.)
Pharmacologic stress testing is of value for cardiac risk stratification before vascular surgical procedures.1-4 Among the alternatives, dobutamine stress echocardiography (DSE) is feasible and safe,5 with a high negative predictive value, and has the additional advantages of having relatively low cost and providing additional information regarding ventricular and valvular function. However, improvements in anesthetic and surgical techniques have made perioperative cardiac events uncommon.6 Indeed, even if DSE is positive, perioperative risk is still <30%.7-9 The predictive value of a positive test may be enhanced by combination with clinical assessment, which is also useful for initial selection of patients for DSE.10
Conversely, correctable coronary artery disease (CAD) is present in up to one third of the patients undergoing vascular procedures.11 Although it is unclear if identification and management of CAD before surgery decreases perioperative risk,12 revascularization improves long-term outcome in patients with severe CAD.13 The use of initial clinical evaluation may add incremental information to DSE for stratification of long-term risk; however, its use to select patients for no DSE on the basis of low perioperative risk might preclude identification of CAD and influence long-term cardiac outcome. Thus the purpose of our study was to determine the perioperative and long-term significance of risk stratification on the basis of clinical evaluation and DSE.
From the Cleveland Clinic Foundation. Submitted March 11, 1998; accepted May 21, 1998. Reprint requests: Thomas H. Marwick, MD, University Department of Medicine, Princess Alexandra Hospital, Ipswich Rd, Brisbane, QLD 4102, Australia. Copyright © 1999 by Mosby, Inc. 0002-8703/99/$8.00 + 0 4/1/91937
Methods Patient population The study group was composed of 233 consecutive patients referred for DSE before a vascular procedure between 1992 and
American Heart Journal March 1999
470 Ballal et al
Table I. Clinical characteristics in patients with normal, ischemic, and scar responses to DSE Characteristics
Normal
Ischemia
Scar, no ischemia
N Age (y) Men Hypercholesterolemia Diabetes mellitus Hypertension Smoking Family history of CAD PVD H/O CAD H/O CABG H/O PTCA H/O CVA Chronic renal failure Eagle criteria None 1 2 >3
158 70 ± 9 95 (60) 68 (43) 39 (25) 118 (75) 121 (77) 22 (14) 156 (99) 65 (41) 21 (13) 11 (7) 33 (21) 13 (8)
39 69 ± 9 26 (67) 20 (51) 15 (38) 26 (67) 26 (67) 5 (13) 39 (100) 30 (77) 16 (41) 3 (8) 10 (26) 5 (13)
36 69 ± 12 28 (78) 18 (50) 9 (25) 19 (53) 29 (81) 8 (22) 34 (94) 32 (89) 14 (39) 4 (11) 7 (19) 2 (6)
32 (20) 78 (49) 36 (23) 12 (8)
5 (13) 13 (33) 12 (33) 8 (22)
3 (6) 14 (39) 12 (36) 8 (18)
P value
.43 .13 .55 .21 .02 .33 .42 .13 < .01 < .01 .71 .77 .51 .01
Values in parentheses are percentages. PVD, Peripheral vascular disease; H/O, history of; CAD, coronary artery disease; CABG, coronary artery bypass grafting; PTCA, percutaneous transluminal coronary angioplasty; CVA, cerebrovascular accident.
1995. Clinical data regarding the past cardiac history, cardiac risk factors, and medical therapy were obtained prospectively. Preoperative clinical risk was scored on the basis of angina, prior infarction or Q waves, diabetes, and heart failure.1
Dobutamine stress test protocol After at least 4 hours of fasting, baseline echocardiographic images were obtained in the left lateral decubitus position. Resting images were obtained in parasternal long- and shortaxis, and apical 4- and 2-chamber views. Dobutamine infusion was started at 5 mg/kg/min and increased every 3 minutes to 10, 20, 30, and a maximum of 40 mg/kg/min. Atropine (maximum of 2 mg) and handgrip were used in patients who failed to achieve at least 85% of maximum predicted heart rate after the peak dose of dobutamine. Echocardiographic images were obtained at each dose increment and digitized on-line (ImageVue, Eastman Kodak Health Imaging) with electrocardiographic gating. Images were also stored on videotape. Continuous electrocardiographic monitoring was done using a 3-channel system, and a 12-lead electrocardiogram (ECG) was obtained at the beginning of each stage or with symptoms. The test was terminated at the end of the protocol (40 mg/kg/min dobutamine, 2 mg atropine) or if there was severe ischemia or other intolerable symptoms or side effects (most commonly hypotension, severe hypertension, or arrhythmias).
Echocardiographic analysis Digitized quad-screen and videotaped images were analyzed by 2 experienced observers blinded to the clinical
data. By use of a 16-segment model,14 regional wall motion was scored qualitatively at rest, low-dose, and peak stress. Myocardial ischemia was defined as a new or worsening wall motion abnormality provoked by pharmacologic stress. A biphasic response, with improvement at low doses of dobutamine and then subsequent worsening of wall motion at higher doses, was also considered to represent an ischemic response. Scar was identified if a myocardial segment showed no evidence of ischemia but had a resting wall motion abnormality. A “normal” response was identified in the absence of rest or stress-induced regional wall motion abnormalities but included patients with global left ventricular dysfunction if this was not focal. Left ventricular ejection fraction (EF) was classified as normal (≥55%), mild (EF 40% to 54%), moderate (EF 30% to 39%), or severe (EF <30%).
Patient outcome analysis Patients were prospectively observed to the time of hospital discharge to identify perioperative cardiac events. Late events were identified by clinic review or telephone followup if this was not possible. The predetermined composite end point was composed of cardiac death, myocardial infarction, and unstable or progressive angina requiring revascularization. Cardiac death was defined as a death from myocardial infarction, congestive heart failure, fatal arrhythmia, or sudden death. Myocardial infarction was defined with the usual clinical, enzymatic, and ECG criteria. Unstable angina was defined as an accelerated pattern of chest pain requiring hospital admission because of
American Heart Journal Volume 137, Number 3
Ballal et al 471
Table II. Hemodynamic and echocardiographic findings in patients with diagnoses of normal studies, ischemia, and scar during DSE Characteristics
Normal
Ischemia
Scar, no ischemia
Rest HR (beats/min) Rest SBP (mm Hg) Rest DBP (mm Hg) Peak SBP (mm Hg) Peak DBP (mm Hg) MPHR (%) Maximum dose Early termination Resting LV dysfunction Mild Moderate Severe
74 ± 14 156 ± 25 81 ± 11 147 ± 33 71 ± 15 86 ± 13 36 ± 7 12 (8)
71 ± 13 159 ± 24 78 ± 11 152 ± 35 68 ± 13 83 ± 12 36 ± 8 3 (8)
74 ± 14 143 ± 23 79 ± 10 146 ± 32 71 ± 15 84 ± 13 36 ± 10 3 (8)
16 (10) 7 (4) 2 (1)
10 (26) 10 (26) 2 (5)
7 (19) 10 (28) 7 (19)
P value .57 .01 .30 .66 .44 .28 .82 .99 <.01
Values in parentheses are percentages. HR, Heart rate; SBP, systolic blood pressure; DBP, diastolic blood pressure; Mild, mild systolic dysfunction; Mod, moderate systolic dysfunction; MPHR, maximum predicted heart rate.
increased frequency, longer duration, or increased refractoriness of pain to current therapy.
Table III. Relation of perioperative and late cardiac events to results of dobutamine echocardiography
Statistical analysis Differences in continuous variables were compared with Student’s t test or analysis of variance, whereas differences in categoric variables were compared with the chi-square analysis. The null hypothesis was rejected when there was <5% chance of a result being from chance alone. Kaplan-Meier survival curves were used to express cumulative event-free survival from the time of surgery. The log-rank statistic was used to compare the differences between the survival curves. Cox regression was used to evaluate the significance of risk factors for late events.
Results Patient characteristics Two hundred thirty-three patients (aged 70 ± 9 years, 159 men) underwent DSE and proceeded to vascular surgery. These patients had a high prevalence of known coronary disease (55%), including 65 (28%) with prior myocardial revascularization and 71 (31%) with left ventricular dysfunction. Significant risk of perioperative events was identified on clinical grounds (≥1 Eagle factor) in 194 patients (83%). Aortic aneurysm repair was performed in 94 patients (40%), carotid artery surgery in 91 (39%), and infrainguinal vascular repair in 48 (21%).
Number of patients Early revascularization Perioperative events Cardiac death MI or unstable angina Late events Cardiac death MI or unstable angina CABG-PTCA
Normal
Ischemia
Scar without ischemia
158 0 6 (4) 1 (1) 5 (3) 20 (13) 11 (7) 6 (4) 3 (2)
39 9 0 (0) 0 0 13 (33) 7 (18) 6 (15) 0
36 0 2 (6) 0 2 3 (8) 1 (3) 1 (3) 1 (3)
Values in parentheses are percentages. MI, Myocardial infarction.
marizes the clinical characteristics of the patients categorized according to the results of their stress test: normal, inducible ischemia, or scar but no ischemia. Compared with the patients with normal stress tests, the patients with inducible ischemia had higher Eagle criteria, hypertension, a history of CAD, or bypass surgery. Hemodynamic and echocardiographic responses to dobutamine are summarized in Table II. Patients with inducible ischemia had higher resting blood pressure than the rest of the patients, but the other responses were indistinguishable between the groups. Patients with ischemia also had worse left ventricular function than the other patients.
Dobutamine echocardiography DSE was performed an average 38 days (interquartile range 4 to 34 days) before vascular surgery. Thirty-nine patients (17%) were identified as having inducible ischemia and 36 patients (15%) had scar. Table I sum-
Cardiac events Perioperative events occurred in 8 patients (3%), of whom 1 died of cardiac causes, 6 had myocardial infarction, and 1 had unstable angina develop.
American Heart Journal March 1999
472 Ballal et al
Figure 1
Event-free survival after vascular surgery of patients with and without ischemia provoked by DSE.
Patients were monitored for 28 ± 13 months after vascular surgery. Cardiac events occurred in 36 patients (16%): 19 died of cardiac causes, 7 had a myocardial infarction, 6 were admitted to the hospital with unstable angina, and 4 required late revascularization for progressive cardiac symptoms. Patients undergoing early revascularization (<3 months after the DSE) were censored from the analysis at the time of the procedure, as were 16 patients who died of noncardiac causes.
Relation of cardiac events to DSE Patients with perioperative events had normal stress test results or evidence of scar alone (Table III); none occurred in the patients with inducible ischemia. This finding probably reflected the performance of early myocardial revascularization in 9 patients, 3 with percutaneous intervention and 6 with coronary bypass surgery. Patients with early events had a higher prevalence of global left ventricular dysfunction than those without (100% vs 50%, P = .01), but no other risk factors, historical or stress features were significantly different. The relation between the results of DSE and longterm outcome is illustrated in Fig 1. A higher event rate was associated with patients with inducible
ischemia compared with patients without inducible ischemia (33% vs 12%, P = .001). However, the presence of scar without ischemia was not associated with higher event rate (8% vs 17%, P = .199).
Relation of cardiac events to clinical risk The relation between the clinical risk factor status and early and late events is summarized in Table IV. Only 1 patient without any of Eagle’s clinical risk factors had a perioperative event, this being a postoperative myocardial infarction. The distribution of Eagle criteria was not significantly different between the patients who had or did not have early events (P = .74) (Table IV), possibly reflecting preoperative myocardial revascularization in the patients at highest risk. Preoperative clinical findings, as expressed by the Eagle criteria, did not predict adverse long-term outcome (Table IV). Of patients who were not revascularized before surgery and survived vascular surgery, late cardiac events occurred in 5 of 37 (13%) without clinical risk factors, compared with 31 of 185 (17%) with risk factors (P = .64). Nonetheless, Fig 2 shows that the presence of multiple risk factors increased the predictive value of a positive test result from 33% (1 risk factor) to 67% (>3 risk factors). Moreover, within each
American Heart Journal Volume 137, Number 3
Ballal et al 473
Figure 2
Event rate of patients according to their clinical risk factor score and DSE results. DbE, DSE.
Table IV. Relation of perioperative and late cardiac events to results of clinical evaluation Eagle criteria 0 (n = 39)
1 (n = 105)
2 (n = 62)
>3 (n = 27)
DSE (inducible ischemia)
+
–
+
–
+
–
+
–
Number of patients Early revascularization Perioperative events Cardiac death MI or unstable angina Late events Cardiac death MI or unstable angina PTCA-CABG
27 2 0 0 0 4 1 2 1
12 0 1 0 1 1 1 0 0
74 4 0 0 0 5 3 1 1
31 0 2 0 2 5 2 2 1
42 2 2 0 2 3 2 1 0
20 0 1 0 1 9 5 4 0
20 1 2 1 1 7 5 2 0
7 0 0 0 0 2 0 1 1
risk group, the presence of ischemia at DSE increased the level of risk. In a multivariate analysis of predictors of long-term risk, ischemia was an independent predictor of worse outcome when normalized for the clinical findings. Even when the patient population was adjusted for sex, Eagle score, and left ventricular dysfunction, ischemia on the dobutamine stress test was predictive of cardiac events at long-term follow-up (Table V).
Discussion The results of this study show that immediate postoperative cardiac events after vascular surgery are uncommon. The perioperative predictive value of a
negative DSE in this study (94%) is comparable to previous reports (95% to 100%).7 Interestingly, the presence of global left ventricular dysfunction in patients with otherwise normal studies may correlate with events. Moreover, inducible ischemia is even more associated with adverse long-term cardiac outcome than perioperative cardiac complications.
Selection of patients for risk stratification Patients undergoing surgery for peripheral vascular disease are widely recognized as having a high prevalence of anatomic evidence of coronary disease.11 A large experience has now been accumulated regarding the accuracy of DSE for preoperative risk stratification
American Heart Journal March 1999
474 Ballal et al
Table V. Results of multivariate analysis with Cox regression for late events after DSE Characteristic
P value
Relative risk
95% CI
Age Sex Eagle criteria LV function Ischemia on DSE
.36 .53 .86 .29 < .01
1.02 0.80 1.09 1.47 3.30
0.98-1.07 0.40-1.60 0.41-2.92 0.72-3.01 1.60-6.82
CI, Confidence interval.
of patients undergoing vascular surgery,3,9,15-17 and there is good evidence indicating that preoperative testing does identify patients at low risk of perioperative cardiac events. The frequency of perioperative cardiac events in the absence of intervention has usually been reported as <10%,18 and this frequency is diminishing with improvements in risk stratification, anesthetic and hemodynamic management, and improved surgical techniques. These factors probably contributed to the low (4%) early event rate in this study. In particular, it should be emphasized that the study group focused on patients who underwent vascular surgery, but does not account for patients who underwent DSE and had surgical plans canceled or altered on the basis of an abnormal test result. Indeed, given the integration of stress testing in clinical decision making in this group, it is difficult to establish the consequences of ischemia in these patients. Thus 75 (32%) of the 233 patients studied had inducible ischemia or scar without ischemia, of whom the 25 thought to be at the highest risk (33% of patients with abnormal test) underwent coronary arteriography, and only 9 (36% of patients undergoing angiography) proceeded to early revascularization. The predictive value of a positive (ischemic) test result in other studies has been reported to be 17% to 30%7; removal of the highest risk patients in this study very likely accounts for the lower apparent predictive value of both the Eagle criteria and DSE findings for early events. Decreasing rates of perioperative events has recently led investigators to question the usefulness of preoperative stress testing,6 especially as the requirement for pharmacologic stress in most of these patients means that relatively costly stress imaging tests are needed for this purpose. Studies with both dipyridamole stress myocardial perfusion imaging and DSE have shown that the investigation of patients at low clinical risk is generally not fruitful,9,19 and algorithms combining
operative and clinical risk have been derived to reduce the numbers of patients referred for risk stratification.10 Indeed, all of the events in patients without clinical risk factors occurred in patients with negative DSE.
Implications of cardiac risk stratification on long-term outcome The high correlation of atherosclerotic vascular disease with CAD has implications that exceed those related to perioperative risk stratification alone. Indeed, other authors have suggested that the presentation of a patient with vascular disease offers the physician an opportunity to diagnose coronary disease in high-risk patients. However, the long-term significance of a positive stress test in this group has been delineated in few studies.20,21 The results of this study confirm that inducible ischemia is associated with adverse long-term outcome. Rather than increasing the risk of late events because of selection of some patients to proceed without DSE, use of a standardized clinical risk profile enhanced the predictive power of DSE for predicting late events (Fig 2). Indeed, of the patients without clinical risk factors who suffered events, all had no ischemia at DSE, suggesting the possibility of no significant coronary disease at their original presentation and progression of their disease in the course of the study.
Limitations There is extensive support for the use of risk stratification in patients undergoing vascular surgery, and making the clinician blinded to these data is no longer ethical. Therefore this analysis is affected by practitioners’ bias regarding patient management. However, this influenced mainly preoperative management; a relatively small number of patients underwent revascularization and a sufficiently large number of end points occurred to make this analysis credible. DSE is currently interpreted qualitatively. The reproducibility of wall motion scoring is limited22 and was not used for this reason. Quantitative techniques, such as the center-line approach, have limited feasibility because of inadequate endocardial visualization and translational cardiac movement. New technologies such as myocardial Doppler imaging or color kinesis may improve the quantitation, reproducibility, and accuracy of the test but were not available when the data were gathered. This is an experience derived from a single large center. The risk profiles reported here may vary based on case mix and may not be generalizable to all hospitals.
American Heart Journal Volume 137, Number 3
Clinical implications In the face of increasing cost constraints, the expense of performing stress imaging studies in patients undergoing vascular surgery, the low frequency of positive results, and the low predictive value of positive tests have caused the value of preoperative risk assessment to be questioned. The results of this study emphasize that initial clinical evaluation is appropriate; although late events are uncommon in those without clinical risk variables, the presence of these factors in combination with DSE may be used to further stratify the likelihood of events in those with an ischemic response.
References 1. Eagle KA, Singer DE, Brewster DC, Darling RC, Mulley AG, Boucher CA. Dipyridamole thallium scanning in patients undergoing vascular surgery. Optimizing preoperative evaluation of cardiac risk. JAMA 1987;257:2185-9. 2. Boucher CA, Brewster DC, Darling RC, Okada RD, Strauss HW, Pohost GM. Determination of cardiac risk by dipyridamole thallium imaging before peripheral vascular surgery. N Engl J Med 1985;312:389-94. 3 Poldermans D, Fioretti PM, Forster T, Thomson IR, Boersma E, el-Said EM, et al. Dobutamine stress echocardiography for assessment of perioperative cardiac risk in patients undergoing major vascular surgery. Circulation 1993;87:1506-12. 4. Shaw LJ, Eagle KA, Gersh BJ, Miller DD. Metaanalysis of intravenous dipyridamole-thallium-201 imaging (1985 to 1994) and dobutamine echocardiography (1991 to 1994) for risk stratification before vascular surgery. J Am Coll Cardiol 1996;27:787-98. 5. Secknus M, Marwick TH. Evolution of dobutamine echocardiography protocols and indications: safety and side-effects in 3011 studies over 5 years. J Am Coll Cardiol 1997;29:1234-40. 6. D’Angelo AJ, Puppala D, Farber A, Murphy AE, Faust GR, Cohen JR. Is preoperative cardiac evaluation for abdominal aortic aneurysm repair necessary? J Vasc Surg 1997;25:152-6. 7. Poldermans D, Rambaldi R, Fioretti PM, Boersma E, Thomson IR, van Sambeek MR, et al. Prognostic value of dobutamine-atropine stress echocardiography for peri-operative and late cardiac events in patients scheduled for vascular surgery. Eur Heart J 1997;18:D86-96. 8. Eagle KA, Froehlich JB. Reducing cardiovascular risk in patients undergoing noncardiac surgery. N Engl J Med 1996;335:1761-3. 9. Poldermans D, Arnese M, Fioretti PM, Salustri A, Boersma E, Thomson IR, et al. Improved cardiac risk stratification in major vascular surgery with dobutamine-atropine stress echocardiography. J Am Coll Cardiol 1995;26:648-53.
Ballal et al 475
10. Rubin DN, Ballal RS, Marwick TH. Outcomes and cost implications of a clinical-based algorithm to guide the discriminate use of stress imaging before noncardiac surgery. Am Heart J 1997;134:83-92. 11. Hertzer NR, Beven EG, Young JR, O’Hara PJ, Ruschhaupt WF, Graor RA, et al. Coronary artery disease in peripheral vascular patients. A classification of 1000 coronary angiograms and results of surgical management. Ann Surg 1984;199:223-33. 12. Seeger JM, Rosenthal GR, Self SB, Flynn TC, Limacher MC, Harward TR. Does routine stress-thallium cardiac scanning reduce postoperative cardiac complications. Ann Surg 1994;219:654-61. 13. Alderman EL, Fisher LD, Litwin P, Kaiser GC, Myers WO, Maynard C, et al. Results of coronary artery surgery in patients with poor left ventricular function (CASS). Circulation 1983;68:785-95. 14. Schiller NB, Shah PM, Crawford M, DeMaria A, Devereux R, Feigenbaum H, et al. Recommendations for quantitation of the left ventricle by two-dimensional echocardiography. American Society of Echocardiography Committee on Standards, Subcommittee on Quantitation of Two-Dimensional Echocardiograms. J Am Soc Echocardiogr 1989;2:358-67. 15. Eichelberger JP, Schwarz KQ, Black ER, Green RM, Ouriel K. Predictive value of dobutamine echocardiography just before noncardiac vascular surgery. Am J Cardiol 1993;72:602-7. 16. Poldermans D, Fioretti PM, Forster T, Boersma E, Arnese M, du Bois NAR, et al. Dobutamine-atropine stress echocardiography for assessment of perioperative and late cardiac risk in patients undergoing major vascular surgery. Eur J Vasc Surg 1994;8:286-93. 17. Davila-Roman VG, Waggoner AD, Sicard GA, Geltman EM, Schechtman KB. Dobutamine stress echocardiography predicts surgical outcome in patients with an aortic aneurysm and peripheral vascular disease. J Am Coll Cardiol 1993;21:957-63. 18. Mangano DT, London MJ, Tubau JF, Browner WS, Hollenberg M, Krupski W, et al, SPI Research Group. Dipyridamole thallium-201 scintigraphy as a preoperative screening test: a reexamination of its predictive capacity. Circulation 1991;84:493-502. 19. Eagle KA, Coley CM, Newell JB, Brewster DC, Darling RC, Strauss HW, et al. Combining clinical and thallium data optimizes preoperative assessment of cardiac risk before major vascular surgery. Ann Intern Med 1989;110:859-66. 20. Poldermans D, Arnese M, Fioretti PM, Boersma E, Thomson IR, Rambaldi R, et al. Sustained prognostic value of dobutamine stress echocardiography for late cardiac events after major noncardiac vascular surgery. Circulation 1997;95:53-8. 21. Younis LT, Aguirre F, Byers S, Dowell S, Barth G, Walker H, et al. Perioperative and long-term prognostic value of intravenous dipyridamole thallium scintigraphy in patients with peripheral vascular disease. Am Heart J 1990;119:1287-92. 22. Hoffmann R, Lethen H, Marwick T, Arnese M, Fioretti P, Pingitore A, et al. Analysis of interinstitutional observer agreement in interpretation of dobutamine stress echocardiograms. J Am Coll Cardiol 1996;27:330-6.
Background Coronary disease is an important cause of long-term morbidity in patients needing major vascular surgery. We sought to assess the efficacy of preoperative clinical evaluation and the detection of inducible ischemia for prediction of immediate and long-term cardiac outcomes of patients undergoing vascular surgery.
Methods In 233 patients undergoing vascular procedures, we assessed risk clinically on the basis of Eagle’s criteria. Dobutamine echocardiography was performed with a standard protocol and results were classified as showing ischemia, scar, or a normal response. Patients were observed perioperatively, and late follow-up (28 ± 13 months) was completed in all surgical survivors. A composite end point of cardiac death, myocardial infarction, and unstable and progressive angina requiring late revascularization was used to judge event-free survival.
Results Of 233 patients undergoing preoperative dobutamine echocardiography, 39 (17%) had inducible ischemia and 36 (15%) had scar. Perioperative events occurred in 8 patients (3%). None of the patients with ischemia had perioperative events, reflecting the effect of revascularization in 9 patients. Late events occurred in 36 patients; ischemia on preoperative stress testing was a predictor of these events even after adjusting for clinical variables and left ventricular dysfunction (relative risk = 3.3; 95% confidence interval 1.6 to 6.8; P = .001). The association of ischemia with clinical predictors was associated with incrementally worse outcome.
Conclusion In addition to perioperative assessment, the combined use of clinical and dobutamine echocardiographic evaluation may stratify the risk of late cardiac events. (Am Heart J 1999;137:469-75.)
Pharmacologic stress testing is of value for cardiac risk stratification before vascular surgical procedures.1-4 Among the alternatives, dobutamine stress echocardiography (DSE) is feasible and safe,5 with a high negative predictive value, and has the additional advantages of having relatively low cost and providing additional information regarding ventricular and valvular function. However, improvements in anesthetic and surgical techniques have made perioperative cardiac events uncommon.6 Indeed, even if DSE is positive, perioperative risk is still <30%.7-9 The predictive value of a positive test may be enhanced by combination with clinical assessment, which is also useful for initial selection of patients for DSE.10
Conversely, correctable coronary artery disease (CAD) is present in up to one third of the patients undergoing vascular procedures.11 Although it is unclear if identification and management of CAD before surgery decreases perioperative risk,12 revascularization improves long-term outcome in patients with severe CAD.13 The use of initial clinical evaluation may add incremental information to DSE for stratification of long-term risk; however, its use to select patients for no DSE on the basis of low perioperative risk might preclude identification of CAD and influence long-term cardiac outcome. Thus the purpose of our study was to determine the perioperative and long-term significance of risk stratification on the basis of clinical evaluation and DSE.
From the Cleveland Clinic Foundation. Submitted March 11, 1998; accepted May 21, 1998. Reprint requests: Thomas H. Marwick, MD, University Department of Medicine, Princess Alexandra Hospital, Ipswich Rd, Brisbane, QLD 4102, Australia. Copyright © 1999 by Mosby, Inc. 0002-8703/99/$8.00 + 0 4/1/91937
Methods Patient population The study group was composed of 233 consecutive patients referred for DSE before a vascular procedure between 1992 and
American Heart Journal March 1999
470 Ballal et al
Table I. Clinical characteristics in patients with normal, ischemic, and scar responses to DSE Characteristics
Normal
Ischemia
Scar, no ischemia
N Age (y) Men Hypercholesterolemia Diabetes mellitus Hypertension Smoking Family history of CAD PVD H/O CAD H/O CABG H/O PTCA H/O CVA Chronic renal failure Eagle criteria None 1 2 >3
158 70 ± 9 95 (60) 68 (43) 39 (25) 118 (75) 121 (77) 22 (14) 156 (99) 65 (41) 21 (13) 11 (7) 33 (21) 13 (8)
39 69 ± 9 26 (67) 20 (51) 15 (38) 26 (67) 26 (67) 5 (13) 39 (100) 30 (77) 16 (41) 3 (8) 10 (26) 5 (13)
36 69 ± 12 28 (78) 18 (50) 9 (25) 19 (53) 29 (81) 8 (22) 34 (94) 32 (89) 14 (39) 4 (11) 7 (19) 2 (6)
32 (20) 78 (49) 36 (23) 12 (8)
5 (13) 13 (33) 12 (33) 8 (22)
3 (6) 14 (39) 12 (36) 8 (18)
P value
.43 .13 .55 .21 .02 .33 .42 .13 < .01 < .01 .71 .77 .51 .01
Values in parentheses are percentages. PVD, Peripheral vascular disease; H/O, history of; CAD, coronary artery disease; CABG, coronary artery bypass grafting; PTCA, percutaneous transluminal coronary angioplasty; CVA, cerebrovascular accident.
1995. Clinical data regarding the past cardiac history, cardiac risk factors, and medical therapy were obtained prospectively. Preoperative clinical risk was scored on the basis of angina, prior infarction or Q waves, diabetes, and heart failure.1
Dobutamine stress test protocol After at least 4 hours of fasting, baseline echocardiographic images were obtained in the left lateral decubitus position. Resting images were obtained in parasternal long- and shortaxis, and apical 4- and 2-chamber views. Dobutamine infusion was started at 5 mg/kg/min and increased every 3 minutes to 10, 20, 30, and a maximum of 40 mg/kg/min. Atropine (maximum of 2 mg) and handgrip were used in patients who failed to achieve at least 85% of maximum predicted heart rate after the peak dose of dobutamine. Echocardiographic images were obtained at each dose increment and digitized on-line (ImageVue, Eastman Kodak Health Imaging) with electrocardiographic gating. Images were also stored on videotape. Continuous electrocardiographic monitoring was done using a 3-channel system, and a 12-lead electrocardiogram (ECG) was obtained at the beginning of each stage or with symptoms. The test was terminated at the end of the protocol (40 mg/kg/min dobutamine, 2 mg atropine) or if there was severe ischemia or other intolerable symptoms or side effects (most commonly hypotension, severe hypertension, or arrhythmias).
Echocardiographic analysis Digitized quad-screen and videotaped images were analyzed by 2 experienced observers blinded to the clinical
data. By use of a 16-segment model,14 regional wall motion was scored qualitatively at rest, low-dose, and peak stress. Myocardial ischemia was defined as a new or worsening wall motion abnormality provoked by pharmacologic stress. A biphasic response, with improvement at low doses of dobutamine and then subsequent worsening of wall motion at higher doses, was also considered to represent an ischemic response. Scar was identified if a myocardial segment showed no evidence of ischemia but had a resting wall motion abnormality. A “normal” response was identified in the absence of rest or stress-induced regional wall motion abnormalities but included patients with global left ventricular dysfunction if this was not focal. Left ventricular ejection fraction (EF) was classified as normal (≥55%), mild (EF 40% to 54%), moderate (EF 30% to 39%), or severe (EF <30%).
Patient outcome analysis Patients were prospectively observed to the time of hospital discharge to identify perioperative cardiac events. Late events were identified by clinic review or telephone followup if this was not possible. The predetermined composite end point was composed of cardiac death, myocardial infarction, and unstable or progressive angina requiring revascularization. Cardiac death was defined as a death from myocardial infarction, congestive heart failure, fatal arrhythmia, or sudden death. Myocardial infarction was defined with the usual clinical, enzymatic, and ECG criteria. Unstable angina was defined as an accelerated pattern of chest pain requiring hospital admission because of
American Heart Journal Volume 137, Number 3
Ballal et al 471
Table II. Hemodynamic and echocardiographic findings in patients with diagnoses of normal studies, ischemia, and scar during DSE Characteristics
Normal
Ischemia
Scar, no ischemia
Rest HR (beats/min) Rest SBP (mm Hg) Rest DBP (mm Hg) Peak SBP (mm Hg) Peak DBP (mm Hg) MPHR (%) Maximum dose Early termination Resting LV dysfunction Mild Moderate Severe
74 ± 14 156 ± 25 81 ± 11 147 ± 33 71 ± 15 86 ± 13 36 ± 7 12 (8)
71 ± 13 159 ± 24 78 ± 11 152 ± 35 68 ± 13 83 ± 12 36 ± 8 3 (8)
74 ± 14 143 ± 23 79 ± 10 146 ± 32 71 ± 15 84 ± 13 36 ± 10 3 (8)
16 (10) 7 (4) 2 (1)
10 (26) 10 (26) 2 (5)
7 (19) 10 (28) 7 (19)
P value .57 .01 .30 .66 .44 .28 .82 .99 <.01
Values in parentheses are percentages. HR, Heart rate; SBP, systolic blood pressure; DBP, diastolic blood pressure; Mild, mild systolic dysfunction; Mod, moderate systolic dysfunction; MPHR, maximum predicted heart rate.
increased frequency, longer duration, or increased refractoriness of pain to current therapy.
Table III. Relation of perioperative and late cardiac events to results of dobutamine echocardiography
Statistical analysis Differences in continuous variables were compared with Student’s t test or analysis of variance, whereas differences in categoric variables were compared with the chi-square analysis. The null hypothesis was rejected when there was <5% chance of a result being from chance alone. Kaplan-Meier survival curves were used to express cumulative event-free survival from the time of surgery. The log-rank statistic was used to compare the differences between the survival curves. Cox regression was used to evaluate the significance of risk factors for late events.
Results Patient characteristics Two hundred thirty-three patients (aged 70 ± 9 years, 159 men) underwent DSE and proceeded to vascular surgery. These patients had a high prevalence of known coronary disease (55%), including 65 (28%) with prior myocardial revascularization and 71 (31%) with left ventricular dysfunction. Significant risk of perioperative events was identified on clinical grounds (≥1 Eagle factor) in 194 patients (83%). Aortic aneurysm repair was performed in 94 patients (40%), carotid artery surgery in 91 (39%), and infrainguinal vascular repair in 48 (21%).
Number of patients Early revascularization Perioperative events Cardiac death MI or unstable angina Late events Cardiac death MI or unstable angina CABG-PTCA
Normal
Ischemia
Scar without ischemia
158 0 6 (4) 1 (1) 5 (3) 20 (13) 11 (7) 6 (4) 3 (2)
39 9 0 (0) 0 0 13 (33) 7 (18) 6 (15) 0
36 0 2 (6) 0 2 3 (8) 1 (3) 1 (3) 1 (3)
Values in parentheses are percentages. MI, Myocardial infarction.
marizes the clinical characteristics of the patients categorized according to the results of their stress test: normal, inducible ischemia, or scar but no ischemia. Compared with the patients with normal stress tests, the patients with inducible ischemia had higher Eagle criteria, hypertension, a history of CAD, or bypass surgery. Hemodynamic and echocardiographic responses to dobutamine are summarized in Table II. Patients with inducible ischemia had higher resting blood pressure than the rest of the patients, but the other responses were indistinguishable between the groups. Patients with ischemia also had worse left ventricular function than the other patients.
Dobutamine echocardiography DSE was performed an average 38 days (interquartile range 4 to 34 days) before vascular surgery. Thirty-nine patients (17%) were identified as having inducible ischemia and 36 patients (15%) had scar. Table I sum-
Cardiac events Perioperative events occurred in 8 patients (3%), of whom 1 died of cardiac causes, 6 had myocardial infarction, and 1 had unstable angina develop.
American Heart Journal March 1999
472 Ballal et al
Figure 1
Event-free survival after vascular surgery of patients with and without ischemia provoked by DSE.
Patients were monitored for 28 ± 13 months after vascular surgery. Cardiac events occurred in 36 patients (16%): 19 died of cardiac causes, 7 had a myocardial infarction, 6 were admitted to the hospital with unstable angina, and 4 required late revascularization for progressive cardiac symptoms. Patients undergoing early revascularization (<3 months after the DSE) were censored from the analysis at the time of the procedure, as were 16 patients who died of noncardiac causes.
Relation of cardiac events to DSE Patients with perioperative events had normal stress test results or evidence of scar alone (Table III); none occurred in the patients with inducible ischemia. This finding probably reflected the performance of early myocardial revascularization in 9 patients, 3 with percutaneous intervention and 6 with coronary bypass surgery. Patients with early events had a higher prevalence of global left ventricular dysfunction than those without (100% vs 50%, P = .01), but no other risk factors, historical or stress features were significantly different. The relation between the results of DSE and longterm outcome is illustrated in Fig 1. A higher event rate was associated with patients with inducible
ischemia compared with patients without inducible ischemia (33% vs 12%, P = .001). However, the presence of scar without ischemia was not associated with higher event rate (8% vs 17%, P = .199).
Relation of cardiac events to clinical risk The relation between the clinical risk factor status and early and late events is summarized in Table IV. Only 1 patient without any of Eagle’s clinical risk factors had a perioperative event, this being a postoperative myocardial infarction. The distribution of Eagle criteria was not significantly different between the patients who had or did not have early events (P = .74) (Table IV), possibly reflecting preoperative myocardial revascularization in the patients at highest risk. Preoperative clinical findings, as expressed by the Eagle criteria, did not predict adverse long-term outcome (Table IV). Of patients who were not revascularized before surgery and survived vascular surgery, late cardiac events occurred in 5 of 37 (13%) without clinical risk factors, compared with 31 of 185 (17%) with risk factors (P = .64). Nonetheless, Fig 2 shows that the presence of multiple risk factors increased the predictive value of a positive test result from 33% (1 risk factor) to 67% (>3 risk factors). Moreover, within each
American Heart Journal Volume 137, Number 3
Ballal et al 473
Figure 2
Event rate of patients according to their clinical risk factor score and DSE results. DbE, DSE.
Table IV. Relation of perioperative and late cardiac events to results of clinical evaluation Eagle criteria 0 (n = 39)
1 (n = 105)
2 (n = 62)
>3 (n = 27)
DSE (inducible ischemia)
+
–
+
–
+
–
+
–
Number of patients Early revascularization Perioperative events Cardiac death MI or unstable angina Late events Cardiac death MI or unstable angina PTCA-CABG
27 2 0 0 0 4 1 2 1
12 0 1 0 1 1 1 0 0
74 4 0 0 0 5 3 1 1
31 0 2 0 2 5 2 2 1
42 2 2 0 2 3 2 1 0
20 0 1 0 1 9 5 4 0
20 1 2 1 1 7 5 2 0
7 0 0 0 0 2 0 1 1
risk group, the presence of ischemia at DSE increased the level of risk. In a multivariate analysis of predictors of long-term risk, ischemia was an independent predictor of worse outcome when normalized for the clinical findings. Even when the patient population was adjusted for sex, Eagle score, and left ventricular dysfunction, ischemia on the dobutamine stress test was predictive of cardiac events at long-term follow-up (Table V).
Discussion The results of this study show that immediate postoperative cardiac events after vascular surgery are uncommon. The perioperative predictive value of a
negative DSE in this study (94%) is comparable to previous reports (95% to 100%).7 Interestingly, the presence of global left ventricular dysfunction in patients with otherwise normal studies may correlate with events. Moreover, inducible ischemia is even more associated with adverse long-term cardiac outcome than perioperative cardiac complications.
Selection of patients for risk stratification Patients undergoing surgery for peripheral vascular disease are widely recognized as having a high prevalence of anatomic evidence of coronary disease.11 A large experience has now been accumulated regarding the accuracy of DSE for preoperative risk stratification
American Heart Journal March 1999
474 Ballal et al
Table V. Results of multivariate analysis with Cox regression for late events after DSE Characteristic
P value
Relative risk
95% CI
Age Sex Eagle criteria LV function Ischemia on DSE
.36 .53 .86 .29 < .01
1.02 0.80 1.09 1.47 3.30
0.98-1.07 0.40-1.60 0.41-2.92 0.72-3.01 1.60-6.82
CI, Confidence interval.
of patients undergoing vascular surgery,3,9,15-17 and there is good evidence indicating that preoperative testing does identify patients at low risk of perioperative cardiac events. The frequency of perioperative cardiac events in the absence of intervention has usually been reported as <10%,18 and this frequency is diminishing with improvements in risk stratification, anesthetic and hemodynamic management, and improved surgical techniques. These factors probably contributed to the low (4%) early event rate in this study. In particular, it should be emphasized that the study group focused on patients who underwent vascular surgery, but does not account for patients who underwent DSE and had surgical plans canceled or altered on the basis of an abnormal test result. Indeed, given the integration of stress testing in clinical decision making in this group, it is difficult to establish the consequences of ischemia in these patients. Thus 75 (32%) of the 233 patients studied had inducible ischemia or scar without ischemia, of whom the 25 thought to be at the highest risk (33% of patients with abnormal test) underwent coronary arteriography, and only 9 (36% of patients undergoing angiography) proceeded to early revascularization. The predictive value of a positive (ischemic) test result in other studies has been reported to be 17% to 30%7; removal of the highest risk patients in this study very likely accounts for the lower apparent predictive value of both the Eagle criteria and DSE findings for early events. Decreasing rates of perioperative events has recently led investigators to question the usefulness of preoperative stress testing,6 especially as the requirement for pharmacologic stress in most of these patients means that relatively costly stress imaging tests are needed for this purpose. Studies with both dipyridamole stress myocardial perfusion imaging and DSE have shown that the investigation of patients at low clinical risk is generally not fruitful,9,19 and algorithms combining
operative and clinical risk have been derived to reduce the numbers of patients referred for risk stratification.10 Indeed, all of the events in patients without clinical risk factors occurred in patients with negative DSE.
Implications of cardiac risk stratification on long-term outcome The high correlation of atherosclerotic vascular disease with CAD has implications that exceed those related to perioperative risk stratification alone. Indeed, other authors have suggested that the presentation of a patient with vascular disease offers the physician an opportunity to diagnose coronary disease in high-risk patients. However, the long-term significance of a positive stress test in this group has been delineated in few studies.20,21 The results of this study confirm that inducible ischemia is associated with adverse long-term outcome. Rather than increasing the risk of late events because of selection of some patients to proceed without DSE, use of a standardized clinical risk profile enhanced the predictive power of DSE for predicting late events (Fig 2). Indeed, of the patients without clinical risk factors who suffered events, all had no ischemia at DSE, suggesting the possibility of no significant coronary disease at their original presentation and progression of their disease in the course of the study.
Limitations There is extensive support for the use of risk stratification in patients undergoing vascular surgery, and making the clinician blinded to these data is no longer ethical. Therefore this analysis is affected by practitioners’ bias regarding patient management. However, this influenced mainly preoperative management; a relatively small number of patients underwent revascularization and a sufficiently large number of end points occurred to make this analysis credible. DSE is currently interpreted qualitatively. The reproducibility of wall motion scoring is limited22 and was not used for this reason. Quantitative techniques, such as the center-line approach, have limited feasibility because of inadequate endocardial visualization and translational cardiac movement. New technologies such as myocardial Doppler imaging or color kinesis may improve the quantitation, reproducibility, and accuracy of the test but were not available when the data were gathered. This is an experience derived from a single large center. The risk profiles reported here may vary based on case mix and may not be generalizable to all hospitals.
American Heart Journal Volume 137, Number 3
Clinical implications In the face of increasing cost constraints, the expense of performing stress imaging studies in patients undergoing vascular surgery, the low frequency of positive results, and the low predictive value of positive tests have caused the value of preoperative risk assessment to be questioned. The results of this study emphasize that initial clinical evaluation is appropriate; although late events are uncommon in those without clinical risk variables, the presence of these factors in combination with DSE may be used to further stratify the likelihood of events in those with an ischemic response.
References 1. Eagle KA, Singer DE, Brewster DC, Darling RC, Mulley AG, Boucher CA. Dipyridamole thallium scanning in patients undergoing vascular surgery. Optimizing preoperative evaluation of cardiac risk. JAMA 1987;257:2185-9. 2. Boucher CA, Brewster DC, Darling RC, Okada RD, Strauss HW, Pohost GM. Determination of cardiac risk by dipyridamole thallium imaging before peripheral vascular surgery. N Engl J Med 1985;312:389-94. 3 Poldermans D, Fioretti PM, Forster T, Thomson IR, Boersma E, el-Said EM, et al. Dobutamine stress echocardiography for assessment of perioperative cardiac risk in patients undergoing major vascular surgery. Circulation 1993;87:1506-12. 4. Shaw LJ, Eagle KA, Gersh BJ, Miller DD. Metaanalysis of intravenous dipyridamole-thallium-201 imaging (1985 to 1994) and dobutamine echocardiography (1991 to 1994) for risk stratification before vascular surgery. J Am Coll Cardiol 1996;27:787-98. 5. Secknus M, Marwick TH. Evolution of dobutamine echocardiography protocols and indications: safety and side-effects in 3011 studies over 5 years. J Am Coll Cardiol 1997;29:1234-40. 6. D’Angelo AJ, Puppala D, Farber A, Murphy AE, Faust GR, Cohen JR. Is preoperative cardiac evaluation for abdominal aortic aneurysm repair necessary? J Vasc Surg 1997;25:152-6. 7. Poldermans D, Rambaldi R, Fioretti PM, Boersma E, Thomson IR, van Sambeek MR, et al. Prognostic value of dobutamine-atropine stress echocardiography for peri-operative and late cardiac events in patients scheduled for vascular surgery. Eur Heart J 1997;18:D86-96. 8. Eagle KA, Froehlich JB. Reducing cardiovascular risk in patients undergoing noncardiac surgery. N Engl J Med 1996;335:1761-3. 9. Poldermans D, Arnese M, Fioretti PM, Salustri A, Boersma E, Thomson IR, et al. Improved cardiac risk stratification in major vascular surgery with dobutamine-atropine stress echocardiography. J Am Coll Cardiol 1995;26:648-53.
Ballal et al 475
10. Rubin DN, Ballal RS, Marwick TH. Outcomes and cost implications of a clinical-based algorithm to guide the discriminate use of stress imaging before noncardiac surgery. Am Heart J 1997;134:83-92. 11. Hertzer NR, Beven EG, Young JR, O’Hara PJ, Ruschhaupt WF, Graor RA, et al. Coronary artery disease in peripheral vascular patients. A classification of 1000 coronary angiograms and results of surgical management. Ann Surg 1984;199:223-33. 12. Seeger JM, Rosenthal GR, Self SB, Flynn TC, Limacher MC, Harward TR. Does routine stress-thallium cardiac scanning reduce postoperative cardiac complications. Ann Surg 1994;219:654-61. 13. Alderman EL, Fisher LD, Litwin P, Kaiser GC, Myers WO, Maynard C, et al. Results of coronary artery surgery in patients with poor left ventricular function (CASS). Circulation 1983;68:785-95. 14. Schiller NB, Shah PM, Crawford M, DeMaria A, Devereux R, Feigenbaum H, et al. Recommendations for quantitation of the left ventricle by two-dimensional echocardiography. American Society of Echocardiography Committee on Standards, Subcommittee on Quantitation of Two-Dimensional Echocardiograms. J Am Soc Echocardiogr 1989;2:358-67. 15. Eichelberger JP, Schwarz KQ, Black ER, Green RM, Ouriel K. Predictive value of dobutamine echocardiography just before noncardiac vascular surgery. Am J Cardiol 1993;72:602-7. 16. Poldermans D, Fioretti PM, Forster T, Boersma E, Arnese M, du Bois NAR, et al. Dobutamine-atropine stress echocardiography for assessment of perioperative and late cardiac risk in patients undergoing major vascular surgery. Eur J Vasc Surg 1994;8:286-93. 17. Davila-Roman VG, Waggoner AD, Sicard GA, Geltman EM, Schechtman KB. Dobutamine stress echocardiography predicts surgical outcome in patients with an aortic aneurysm and peripheral vascular disease. J Am Coll Cardiol 1993;21:957-63. 18. Mangano DT, London MJ, Tubau JF, Browner WS, Hollenberg M, Krupski W, et al, SPI Research Group. Dipyridamole thallium-201 scintigraphy as a preoperative screening test: a reexamination of its predictive capacity. Circulation 1991;84:493-502. 19. Eagle KA, Coley CM, Newell JB, Brewster DC, Darling RC, Strauss HW, et al. Combining clinical and thallium data optimizes preoperative assessment of cardiac risk before major vascular surgery. Ann Intern Med 1989;110:859-66. 20. Poldermans D, Arnese M, Fioretti PM, Boersma E, Thomson IR, Rambaldi R, et al. Sustained prognostic value of dobutamine stress echocardiography for late cardiac events after major noncardiac vascular surgery. Circulation 1997;95:53-8. 21. Younis LT, Aguirre F, Byers S, Dowell S, Barth G, Walker H, et al. Perioperative and long-term prognostic value of intravenous dipyridamole thallium scintigraphy in patients with peripheral vascular disease. Am Heart J 1990;119:1287-92. 22. Hoffmann R, Lethen H, Marwick T, Arnese M, Fioretti P, Pingitore A, et al. Analysis of interinstitutional observer agreement in interpretation of dobutamine stress echocardiograms. J Am Coll Cardiol 1996;27:330-6.