Author + information
- Albert J. Sinusas, MD∗ ( and )
- Erica S. Spatz, MD, MHS
- Section of Cardiovascular Medicine, Department of Medicine, Yale University School of Medicine, New Haven, Connecticut
- ↵∗Reprint requests and correspondence:
Dr. Albert J. Sinusas, Section of Cardiovascular Medicine, Yale University School of Medicine, P.O. Box 8017, DANA 3, New Haven, Connecticut 06520.
Guidelines recommend that most patients being evaluated for ischemic heart disease (IHD) undergo exercise electrocardiography (ECG), provided that they are able to exercise (1). Yet, despite more than 3 decades of data and experience, the test continues to vex interpreting physicians and referring providers alike. Although ECG is recognized as providing valuable information, ambivalence arises in estimating the likelihood of disease on the basis of a test with a reported sensitivity of only 68% and specificity of 77%, far below the test performance of most other cardiovascular imaging modalities.
Inherent to the interpretation of the exercise ECG is the application of the Bayes theorem, or the process of probabilistic reasoning. Simply stated, the predictive value of any test result is conditional on the pre-test probability of disease. These concepts were illustrated in Diamond and Forrester's landmark paper of 1979, demonstrating the likelihood of coronary artery disease on the basis of an individual's age, sex, and symptoms as well as the magnitude of ST-segment depression with exercise. In the ensuing 35 years, researchers have leveraged other data associated with the exercise ECG to improve the accuracy for detection of IHD, assess severity of disease, and inform risk stratification, including the evaluation of hemodynamic changes, exercise time, and symptoms. For example, the presence of angina and duration of exercise, captured in the Duke Treadmill Score, have been demonstrated to correlate with the risk of future events (2). Many other electrocardiographic indexes beyond simple analysis of ST-segment depressions have been proposed for detection of IHD (e.g., heart rate recovery, heart rate adjustment for ST-segment depression, QRS duration and amplitude), although none of these indexes have been as widely applied as the analysis of discrete ST-segment depressions. These other previously established, more sensitive or specific ECG indexes are rarely reported, and it is not clear that they would change clinical practice (3,4).
In this issue of the Journal, Christman et al. (5) examine specific exercise ECG findings and their association with the use of additional diagnostic cardiovascular imaging tests as well as cardiovascular outcomes (1). From a 10,000-ft view, scoping the diagnostic evaluation of adults suspected of having IHD is incredibly helpful for: 1) observing referral patterns; 2) discovering knowledge gaps; and 3) identifying areas for quality improvement; it is indeed the feedback that every health system should provide. Unfortunately, the overview engenders more questions than conclusions. For example, it is impossible to judge whether the diagnostic referral patterns observed in this study are the result of probabilistic reasoning or the interpretation of test findings in isolation. Without considering the patient's pre-test probability of disease and risk of future events, only limited conclusions can be drawn about the diagnostic utility of exercise ECG and the appropriateness of downstream testing. With respect to new knowledge, although the study adds to the literature regarding the performance of specific exercise ECG findings in detecting disease, there continue to be salient questions about our ability to integrate test findings into calculations of disease probability and decisions regarding the next steps. Given these limitations, the study's impact may be greatest in identifying areas for improvement. Questions about test performance and the appropriateness of downstream testing are increasingly relevant for the cardiovascular imaging community, particularly in light of the recent advances in noninvasive diagnostic testing modalities that have further expanded options for test selection. It may be that interpreting physicians need to reconsider their role in informing Bayesian decision-making.
Outcomes following the diagnostic evaluation of IHD of 3,345 individuals referred for exercise ECG to a high-volume, academic stress laboratory were reviewed. The exercise ECG was reported as positive for ischemia in 3.7% of individuals; most were negative (67.7%) or inconclusive (28.5%). Additional downstream testing was performed in approximately 1 in 10 adults, reflecting the myriad of diagnostic modalities available for evaluating ischemia: nuclear myocardial perfusion imaging (obtained in 65% of individuals referred for a downstream test), stress echocardiogram (10%), coronary computed tomographic angiography (4%), stress magnetic resonance imaging (1%), and coronary angiography (20%). Despite the increased sensitivity of more advanced diagnostic modalities, downstream testing yielded relatively few positive test results, especially among individuals with negative and inconclusive test findings, perhaps suggesting referring providers' overestimation of disease probability or the limitations of all diagnostic tests. Moreover, as expected in this ambulatory population, the combined event-free survival from coronary revascularization, myocardial infarction, and cardiovascular death was low. These data provide important feedback; at the margins, the exercise ECG remains a useful initial strategy for the risk stratification of individuals suspected of having IHD.
A second aim of the study was to determine whether new insights may be gained from re-evaluating the reasons for inconclusive test findings, which generated the bulk of downstream testing. The authors found that individuals with inconclusive tests because of “rapid recovery of ECG changes” unanimously had normal findings on subsequent diagnostic imaging tests, whereas those with inconclusive tests because of “typical angina but no ECG changes” frequently had a positive downstream test. These findings are consistent with prior studies showing that transient ECG changes and chest pain that occur with exercise ECG testing are important diagnostic and prognostic indexes. The presence of chest pain was shown to be an important independent predictor of coronary artery disease along with outcomes in the 1970s and is a risk factor in the Duke Treadmill Score used for prognosis (6,7). The clinical value of analyzing the time to recovery of ST-segment depressions has also been established in earlier studies (4). In fact, these earlier studies suggested that it was necessary to perform a heart rate adjustment for the recovery of exercise-induced ST-segment depressions and that change in ST-segment/heart rate index or evaluation of the ST-segment rate-recovery loop during exercise testing improved the prognostic value (4).
Nonetheless, conclusions about the utility of specific exercise ECG findings in the “inconclusive” group must be interpreted with caution. For example, if only individuals with a high pre-test probability were referred for additional testing, then individuals with a low probability of disease would not be accounted for when assessing the yield of inconclusive test findings. This differential assessment leads to “verification bias,” or the inability to verify disease in all subjects (because not everyone had a downstream test) (8). Hence, the finding that 21% of individuals with “angina but no ECG changes” had a positive downstream test may be an overestimation of the true utility of this finding.
Another limitation in validating specific exercise ECG findings with noninvasive cardiovascular imaging tests is that most of the downstream tests performed in this study are not considered gold standards of coronary artery disease. All noninvasive imaging tests are inherently subject to delivering false-positive and -negative results and cannot be used to verify exercise ECG findings. This can be appreciated if we consider the clinical significance of exertional angina. As outlined in the previous text, the presence of angina during a physiological test to increase myocardial oxygen demand has long been recognized as a predictor of outcomes, independent of ECG findings (6,9). Similarly, angina without perfusion defects or wall motion defects is also associated with adverse events. These tests were designed to detect obstructive coronary artery disease. However, microvascular disease or epicardial endothelial dysfunction may result in angina with normal ECG, imaging, and even coronary catheterization findings. Indeed, women with angina or other evidence of ischemia but no obstructive coronary disease have poor outcomes (10,11). As such, one must wonder whether patients were harmed by receiving a negative downstream test result.
Developments in quantitative dynamic positron emission tomography imaging may help advance our understanding of this paradox. For example, diminished coronary flow reserve, a marker of microvascular disease, has been demonstrated through the use of dynamic positron emission tomography in patients with chest pain and normal epicardial arteries (12). These and other developments have the potential to transform test interpretation and risk assessment; however, they must be correlated with patient outcomes.
Indeed, as the field of advanced cardiac imaging grows, it is increasingly more complex to determine the optimal diagnostic work-up for patients suspected of IHD. Test selection and interpretation requires knowing more than just the pre-test probability of disease and the reported sensitivity and specificity of the test (13). Consideration must be given to the local quality of the test (both in its administration and interpretation), the test's performance in specific populations, and how the test would affect patient treatment and/or outcomes. Patient preferences must also be taken into account. These aspects of test selection are nuanced for each patient and must be considered when inferring the appropriateness of any downstream test, whether the optimal imaging modality was selected, or whether patients benefited (or were harmed) by additional testing.
How can we improve the yield of diagnostic tests and referral patterns to investigate IHD? The cardiovascular imaging community may be uniquely positioned to help shape the diagnostic landscape. To do so, however, cardiovascular imaging specialists may need to reconsider the reporting of test interpretations and their role in influencing disease management. For example, a more expanded reporting framework may consider questions such as: What is the patients' baseline risk for IHD? What are the specific test findings and what do they indicate, even if discordant? What are the options for next steps? What is at stake? Are there any risks in further testing? How might downstream testing affect management? What do the guidelines offer? What is the cost? Ideally, the consultation would serve as a clinical guide, facilitating more informed, patient-centered discussions and decisions regarding next steps. It would also serve as a roadmap for future investigative work and technology, identifying gaps in knowledge and opportunities for advancement. In this context, specific findings from the exercise ECG may prompt more thoughtful decision-making, actually improving their diagnostic yield.
↵∗ Editorials published in the Journal of the American College of Cardiology reflect the views of the authors and do not necessarily represent the views of JACC or the American College of Cardiology.
Both authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- American College of Cardiology Foundation
- Fihn S.D.,
- Gardin J.M.,
- Abrams J.,
- et al.
- Kligfield P.,
- Lauer M.S.
- Kligfield P.,
- Ameisen O.,
- Okin P.M.
- Christman M.P.,
- Sommer Bittencourt M.,
- Hulten E.,
- et al.
- Diamond G.A.
- Johnson B.D.,
- Shaw L.J.,
- Buchthal S.D.,
- et al.
- Geltman E.M.,
- Henes C.G.,
- Senneff M.J.,
- Sobel B.E.,
- Bergmann S.R.
- Fryback D.G.,
- Thornbury J.R.