In response to the observation that the number of stress imaging procedures was rising annually at a very high rate, the American College of Cardiology Foundation proposed a method for evaluating the appropriateness of cardiovascular imaging using technical panels of experts who evaluated possible clinical scenarios where imaging would be used (1). The panels developed appropriate use criteria (AUC) for these varied clinical indications using a modified Delphi exercise. They combined evidence-based medical information and expert opinion. For example, the revised AUC for radionuclide imaging (2) lists 67 clinical scenarios scored by the panels on a scale of 1 to 9. Indications for radionuclide imaging deemed appropriate were those with scores ranging from 7 to 9; uncertain indications were those that yielded a median score of 4 to 6; and inappropriate indications have median scores from 1 to 3. The inappropriate indications fall into several categories, including testing of asymptomatic or low-risk patients, routine testing early after coronary revascularization, pre-operative testing of low-risk patients with good functional capacity before noncardiac surgery, and detection of coronary artery disease (CAD) in stable symptomatic patients who have a low pre-test probability of CAD, have an interpretable baseline electrocardiogram, and are able to exercise (2).

Several groups have sought to determine how well academic and community physicians fared with respect to adherence to the AUC for stress radionuclide imaging and stress echocardiography (3- 6). The rate of inappropriate indications in these studies ranges from 13% to 14%. A recent multicenter study (6) comprising 6 diverse clinical sites and 6,351 prospectively enrolled patients who underwent stress single-photon emission computed tomography myocardial perfusion imaging (MPI) found that 14% of indications were inappropriate, 15% were uncertain, and 71% were appropriate. As observed in the single-center studies, the most frequent inappropriate indications for stress single-photon emission computed tomography MPI was for detecting CAD in asymptomatic patients at low risk, followed by testing asymptomatic patients after revascularization <2 years after percutaneous coronary intervention (PCI) who had symptoms before PCI. More inappropriate studies were ordered by noncardiologists than cardiologists and more in women than in men. By multivariate analysis, asymptomatic status was the best predictor of an inappropriate classification.

In the present issue of the Journal, Shah et al. (7) retrospectively examined the use and timing of stress testing more than 90 days after coronary revascularization in patients age 18 to 64 years from a national health insurance claims database. They also looked at subsequent rates of coronary angiography and repeat revascularization after stress testing. They found that among 28,177 patients undergoing revascularization, 59% had at least 1 cardiac stress test within 24 months (61% of PCI and 51% of coronary artery bypass graft [CABG] patients). Nuclear imaging was the predominant testing method. Of interest was the observation that the major spikes for stress testing ordering were at 6 and 12 months after revascularization, presumably coinciding with outpatient follow-up visits. Although more than one-half of the patients underwent stress testing, only 11% had subsequent cardiac catheterization within 30 days of testing and only 5% of all patients who underwent stress testing had repeat revascularization. There was also marked regional variation in the rate of stress testing after revascularization, with up to a 50% difference in rates of testing between regions. This regional variation is typical for cardiac diagnostic and therapeutic procedures performed in the U.S. (8- 9). The study by Shah et al. (7) suggests excessive and inappropriate use of stress testing, mostly with imaging, after coronary revascularization. That is too much imaging too soon after revascularization. From clinical experience and published studies of outcomes after PCI and CABG, it is doubtful that 60% of PCI patients and 50% of CABG patients had significant ischemic symptoms that warranted stress imaging. Shah et al. (7) cite data from the National Heart, Lung, and Blood Institute's Dynamic registry indicating that 18% of PCI patients report angina symptoms at the 1-year follow-up, a figure considerably lower than the 60% rate of stress testing after PCI reported in their study. With more drug-eluting stents being used for PCI and greater experience of operators in the interventional laboratory, the percentage of patients with recurrent angina at 1 year after PCI has dropped considerably.

The data from Shah et al. (7) support prior reports of inappropriate routine imaging tests in asymptomatic patients after uncomplicated PCI or CABG. A major limitation of the study, noted by the authors, is the lack of information on the percentage of patients who had symptoms after revascularization for which a stress test would be appropriate. They report only the International Classification of Diseases-Ninth Edition codes for the common diagnoses at the time of testing, and although some of these codes provide potential symptom states, the most common (i.e., 414) reflects only known disease (CAD). Billing data for clinical syndromes may be inaccurate. Nevertheless, the fact that approximately only 10% of the patients who underwent testing had downstream cardiac catheterization implies that most of the stress tests studies were low risk and likely were not indicated in the first place. Because the investigators eliminated any tests performed before 90 days after revascularization, those stress tests performed for cardiac rehabilitation were probably not included in their data after 90 days.

The claims data reviewed by Shah et al. (7) spanned the period from July 2004 through June 2007. The AUC for use of stress imaging procedures were perhaps not yet fully appreciated in that time. Physicians today are more cognizant of AUC, as are health insurance companies pre-authorizing stress tests. The latest AUC for stress radionuclide imaging (2) indicate that it is appropriate to perform a stress MPI to evaluate symptomatic patients after revascularization with an ischemic equivalent of 5 years or more after CABG. It clearly states that it is inappropriate to perform stress imaging for risk assessment after revascularization <2 years after PCI in asymptomatic patients. Interestingly, the indication for stress MPI ≥2 years after PCI in an asymptomatic patient received an uncertain indication. This would not apply to the study by Shah et al. (7), because they report stress test use rates only within 24 months after PCI, a period when the AUC clearly state such testing is inappropriate in asymptomatic patients.

There is one caveat regarding the value of post-revascularization stress imaging that deserves mention. In the COURAGE (Clinical Outcomes Utilizing Revascularization and Aggressive Drug Evaluation) trial nuclear substudy, Shaw et al. (10) found that patients in either the optimal medical therapy arm or the optimal medical therapy plus PCI arm who did not have at least a ≥5% reduction in ischemia 6 to 18 months after assigned treatment had a worse cumulative event-free survival at follow-up compared with those who had a significant reduction in myocardial ischemia. Similarly, regardless of treatment assignment, the magnitude of residual ischemia at follow-up was proportional to the risk of death or nonfatal infarction. The prevalence and extent of residual ischemia were determined by serial imaging before and 6 to 18 months after PCI or after randomization to medical therapy without a PCI. Only 11% of the PCI patients who underwent serial imaging had angina at the time of the follow-up stress test. The number of patients in this substudy was too small to make definitive conclusions and was considered to be hypothesis generating. If, in the future, a larger randomized study comparing a serial imaging strategy versus no post-procedure imaging in asymptomatic patients who have undergone PCI shows a benefit of detecting silent residual ischemia that is associated with better long-term outcomes, the appropriateness of stress testing after PCI will have to be re-evaluated. This assumes that eliminating or reducing residual ischemia identified in such patients can be accomplished and that it will be associated with improved infarct-free survival. Outcomes will have to be substantially better than with just effective control of risk factors such as reducing low-density lipoprotein cholesterol levels, maintaining blood pressure in the normotensive range, weight reduction, exercise therapy, and smoking cessation, but without routine post-treatment serial single-photon emission computed tomography MPI.

In summary, nearly 25% of all inappropriate stress imaging studies are performed in asymptomatic patients who have undergone revascularization <2 years after PCI (6). The study by Shah et al. (7), despite its limitations as a retrospective analysis of insurance claims, highlights the high and probably excessive use of stress testing in the first year after revascularization, with marked regional variation in test use. This study serves as another wake-up call to cardiovascular specialists to be more diligent in adhering to evidence-based practice guidelines and AUC. The value of stress imaging is greatest in the evaluation of risk for future cardiac events in symptomatic patients to identify those who would benefit the most from revascularization strategies. Our goal as cardiovascular specialists is to educate patients and referral physicians regarding the appropriate indications for expensive stress imaging procedures and where diagnostic and prognostic value is greatest. If we fail in this duty, we will be coerced into constantly securing pre-authorization from payers for diagnostic imaging tests to be performed for our patients. The problems with pre-authorization for formal approval of use of testing have been well described (11). They include no evidence for improved quality of care, the favoring of indiscriminate volume reduction, the lack of transparency, the fact that such measures are not based on AUC, inconsistent processes often characterized by confusion and inefficiency, reduced timeliness, an unstated goal of steerage to the test of least resistance, labor intensiveness, and scant data available for feedback or education (11). Thus, it behooves cardiovascular specialists to advocate for and adhere to accepted AUC developed by our own scientific societies.