Author + information
- Michael P. Thomas, MD∗ (, )
- Hitinder S. Gurm, MD and
- Brahmajee K. Nallamothu, MD, MPH
- Division of Cardiovascular Medicine, University of Michigan, Ann Arbor, Michigan
- VA Ann Arbor Healthcare System and Center for Clinical Management Research, Ann Arbor, Michigan
- ↵∗Reprint requests and correspondence:
Dr. Michael Thomas, University of Michigan, Cardiovascular Medicine, Cardiovascular Center, 1500 East Medical Center Drive, SPC 5869, Ann Arbor, Michigan 48109-5869.
Coronary angiography is a critical diagnostic tool for defining anatomy and guiding therapy in coronary artery disease. Not surprisingly, it has gained widespread use since Mason Sones first described it over 50 years ago with an estimated 2 million procedures performed each year in the United States alone (1). However, there are well-known, significant costs associated with coronary angiography, both to the patient (given its procedural risks) and to the healthcare system as a whole. In 2010, a highly publicized article by Patel et al. (2), in the New England Journal of Medicine, raised concerns about the potential, indiscriminate use of coronary angiography given its “low diagnostic yield.” Utilizing data on nearly 400,000 patients without known coronary artery disease who had been referred for elective procedures in the CathPCI Registry, only 38% were found to have obstructive disease, whereas 39% had little or no disease (i.e., “normal” coronary arteries). Adding to these troubling findings were the observations that a large number of patients were asymptomatic (∼30%) and noninvasive testing before the procedure did not improve diagnostic yield.
Ko et al. (3) further explored these issues in an intriguing report published earlier this year in the Journal of the American Medical Association in which cross-national comparison data were used between New York State and Ontario. In this study, the authors compared 18,114 patients in New York and 54,933 patients in Ontario who were undergoing elective coronary angiography (utilizing a government-funded, single-payer system) from 2008 to 2011. The overall rate of obstructive disease in New York was only 30% compared with 45% in Ontario, which is a finding primarily driven by a higher rate of referral of low-risk patients in New York. Using a risk model based on clinical factors and noninvasive testing, fewer than 1 in 5 patients in New York had greater than 50% likelihood of an obstructive coronary artery disease compared with more than 2 in 5 patients in Ontario. Importantly, no underdetection of patients with surgical coronary artery disease (left main disease or 3-vessel coronary artery disease) was noted, despite a historically 50% lower use of coronary angiography per capita in Ontario. Thus, a more restricted approach to patient selection for coronary angiography in Ontario did not appear to miss those with critical disease.
In this issue of the Journal, Bradley et al. (4) add to this discussion with a report from the Veterans Affair (VA) Healthcare System's Cardiovascular Assessment, Reporting and Tracking for Cath Labs (CART-CL) program. This study is important because the VA Healthcare System represents a large, integrated healthcare delivery system in the United States where financial incentives for performing coronary angiography and medico-legal concerns may be less than in the private sector. The authors utilized data from 76 VA cardiac catheterization laboratories between 2007 and 2010. Of the 22,538 patients who underwent elective coronary angiography during this time period, 4,829 had normal coronary arteries (21%) and 11,622 (52%) had obstructive disease. Patients with normal coronary arteries were more likely to have low Framingham risk scores and to have undergone a noninvasive test. To assess hospital-level variation, the hospitals were divided into quartiles based on the percentage of cases with normal coronary arteries with quartile 1 having a rate of normal coronary arteries of 11% and quartile 4 having a rate of 30%. Patients in quartile 1 were more likely to undergo noninvasive testing, but no consistent trends were noted across quartiles in patient demographics, cardiovascular risk factors, Framingham scores, or hospital characteristics.
This work by Bradley et al. (4) is important for several reasons. First, it suggests a higher referral threshold for coronary angiography within the VA. Given the possibility of less direct financial incentives for testing in an integrated healthcare delivery system, this finding may have implications for Accountable Care Organizations that will gain traction in the coming years. Second, their observation of 10-fold variation in hospital rates of normal coronary arteries is important. Despite finding an overall rate of normal coronary arteries that was almost one-half of what was reported from the CathPCI Registry, this inconsistency implies an ongoing need to improve patient selection across institutions and reminds us that factors beyond financial incentives are playing a role. Third, this report raises a real concern regarding studies that compare rates of normal coronary arteries across healthcare systems that many VA cardiologists will immediately recognize. Given a higher burden of baseline disease in the VA population, a poor decision to perform coronary angiography in a veteran (e.g., asymptomatic and low-risk stress test) may be statistically more likely to yield obstructive disease than an appropriate decision in other settings. As even 10% of patients with an acute coronary syndrome might have nonobstructive disease (5), we may be right, but for the wrong reasons, or wrong for the right reasons.
Thus, it remains unclear as to what we (as a clinical community) are to do collectively with these studies of rates of normal coronary arteries (and the others that may potentially follow). Yet, the questions that they raise are potentially enormous. To what extent do high rates of coronary arteries indicate poor quality or suggest that we are performing too many procedures? Do we need to become more adept at risk stratification or do we need more or better noninvasive testing? How are financial incentives driving these decisions and what other factors—such as medico-legal concerns—are playing a role in patient selection? And finally, what is a reasonable rate of normal coronary arteries that should be expected for cardiologists, realizing that 0% is neither possible nor desirable?
Of course, many of these questions deal with the overall quality of current clinical assessments and noninvasive testing. These issues were highlighted over 3 decades ago in the seminal work of Diamond and Forrester (6) with their application of Bayes' theorem to coronary angiography. Results of any clinical finding or diagnostic test must be placed into the context of a patient, as their interpretation inherently depends on the pretest probability of disease. Diamond and Forrester (6) demonstrated that the probability of coronary artery disease may be obtained in large part through assessment of the patient's age, sex, and symptoms. Despite significant advancements in noninvasive tests since that time, it is disappointing that these tests only marginally improve the diagnostic yield of coronary angiography over these clinical factors (7).
Despite the clear need to improve our decision-making process for coronary angiography, it also is important to acknowledge that some elective procedures will undoubtedly result in the finding of normal coronary arteries. So when is it right for us to be wrong? Is the rate of normal coronary arteries found in the CathPCI Registry of 39% too high or perhaps the rate of 21% in the VA population too low? Too high a rate suggests waste and the danger of unnecessary procedures, whereas too low a rate implies we may be causing harm by missing patients appropriately referred for this diagnostic test. Although the study by Ko et al. (3) suggests this latter concern may be minimized, it is clear (even from that study) that we must be able to accept a few false-positive test results as part of the process. In some circumstances, there may be great value to a negative study that identifies normal coronary arteries, given the concerns many patients have with the possibility of cardiac conditions as a cause of their symptoms. In fact, the value of any diagnostic test lies not only in its ability to “rulein” disease, but in how it helps clinicians to “ruleout” disease as well because that also strongly influences subsequent management.
We believe many forces will push this debate even further in the coming years. Rates of normal coronary angiographies have been discussed as a performance measure for over a decade now, but there has been little pursuit of it (8). However, the emerging data highlighted by Patel et al. (2), Ko et al. (3), and Bradley et al. (4) suggest that a greater interest will and should be placed on risk stratification and patient selection in the coming years. This is further supported by the recent publication of appropriate use criteria (AUC) for coronary angiography. Using rates of normal coronary arteries may supplement AUC to fully inform us on how well an entire system is doing in this regard and ensure the validity in quality comparisons across hospitals. Bradley et al. (4) eloquently raise these points in their article, but they also warn us about potential limitations with its use in isolation. For example, before rates of normal coronary arteries become a performance measure, we clearly need more empirical work as ranking of hospitals in the VA Healthcare System was highly sensitive to how “normal” was defined. In Figure 3 of the article by Bradley et al. (4), the top hospital ranked by its rate of normal coronary arteries was approximately 50th by its rate of nonobstructive coronary artery disease.
Although the extent to which the use of rankings and performance measurement of rates of normal coronary arteries may influence clinicians is unclear, it may be consequential. A prominent example of the real-world implications of these decisions was recently illustrated. In a $4 million settlement by a physician and healthcare system for allegedly performing unnecessary coronary angiography, it was purported that 75% of patients had “no significant heart blockages” (9). This case is obviously complex and raised multiple issues, including the improper reading of nuclear stress tests prior to coronary angiography. Yet, it is telling that this case is fundamentally different from prior reports of inappropriate coronary stenting or cardiac surgery because it involves the claim that a diagnostic test, and not a therapeutic procedure, was overused. This also raises the natural question as to whether this logic may be extended to other diagnostic tests, such as measures of normal rates of echocardiograms, computed tomography scans, ultrasounds, and even some expensive laboratory tests.
Improving our understanding of all these issues surrounding rates of normal coronary arteries will be fundamental as we move forward in an era of AUC, quality improvement initiatives, performance measures, and escalating costs. This must be done carefully with recognition that large differences will exist across the populations that we serve. This should influence how these data are collected, interpreted, and reported. As clinicians, we certainly need to become better at how we utilize expensive and sometimes risky tests like coronary angiography. This desire for improvement, however, must be balanced with the knowledge that there remains an important role for judgment in such decisions. That is we need to hold on to the right to be wrong.
↵∗ 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.
Dr. Gurm has received research funding from the National Institutes of Health, Agency for Healthcare Research and Quality, and Blue Cross Blue Shield of Michigan; has consulted for Osprey Medical; and has served as a peer reviewer of PCI quality and appropriateness. Drs. Thomas and Nallamothu have reported that they have no relationships relevant to the contents of this paper to disclose.
- American College of Cardiology Foundation
- Ryan T.J.
- Bradley S.M.,
- Maddox T.M.,
- Stanislawski M.A.,
- et al.
- Maddox T.M.,
- Ho P.M.,
- Roe M.,
- Dai D.,
- Tsai T.T.,
- Rumsfeld J.S.
- Bashore T.M.,
- Bates E.R.,
- Berger P.B.,
- et al.
- ↵Baldas T. Cardiology practice, Jackson hospital to pay $4M in lawsuit alleging medically inappropriate heart procedures. Detroit Free Press. July 2013. Available at: http://www.freep.com/article/20130710/NEWS05/307100126/cardiology-settlement-jackson-hospital-medically-inappropriate. Accessed September 17, 2013.