NCDR indicates National Cardiovascular Data Registry; PCI, percutaneous coronary intervention.
Shown is the variation in hospital rate of asymptomatic patients at angiography.
PCI indicates percutaneous coronary intervention.
eTable. Hospital PCI Appropriateness for Nonacute Indications by Hospital Quartile of Asymptomatic Patients at Angiography Among Hospitals Reporting Twice as Many Elective Angiograms Than Elective PCIs
Bradley SM, Spertus JA, Kennedy KF, Nallamothu BK, Chan PS, Patel MR, Bryson CL, Malenka DJ, Rumsfeld JS. Patient Selection for Diagnostic Coronary Angiography and Hospital-Level Percutaneous Coronary Intervention AppropriatenessInsights From the National Cardiovascular Data Registry. JAMA Intern Med. 2014;174(10):1630-1639. doi:10.1001/jamainternmed.2014.3904
Diagnostic coronary angiography in asymptomatic patients may lead to inappropriate percutaneous coronary intervention (PCI) due to a diagnostic-therapeutic cascade. Understanding the association between patient selection for coronary angiography and PCI appropriateness may inform strategies to minimize inappropriate procedures.
To determine if hospitals that frequently perform coronary angiography in asymptomatic patients, a clinical scenario in which the benefit of angiography is less clear, are more likely to perform inappropriate PCI.
Design, Setting, and Participants
Multicenter observational study of 544 hospitals participating in the CathPCI Registry between July 1, 2009, and September 30, 2013.
Main Outcomes and Measures
Hospital proportion of asymptomatic patients at diagnostic coronary angiography and hospital rate of inappropriate PCI as defined by 2012 appropriate use criteria for coronary revascularization.
Of 1 225 562 patients who underwent elective coronary angiography, 308 083 (25.1%) were asymptomatic. The hospital proportion of angiography among asymptomatic patients ranged from 1.0% to 73.6% (median, 24.7%; interquartile range, 15.9%-35.9%). By hospital quartile of asymptomatic patients at angiography, hospitals with higher rates of asymptomatic patients at angiography had higher median rates of inappropriate PCI (14.8% vs 20.2% vs 24.0 vs 29.4% from lowest to highest quartile, P < .001 for trend). This outcome was attributable to more frequent use of inappropriate PCI in asymptomatic patients at hospitals with higher rates of angiography in asymptomatic patients (5.4% vs 9.9% vs 14.7% vs 21.6% from lowest to highest quartile, P < .001 for trend). Hospitals with higher rates of asymptomatic patients at angiography also had lower rates of appropriate PCI (38.7% vs 33.0% vs 32.3% vs 32.9% from lowest to highest quartile, P < .001 for trend).
Conclusions and Relevance
In a national sample of hospitals, performance of coronary angiography in asymptomatic patients was associated with higher rates of inappropriate PCI and lower rates of appropriate PCI. Improving preprocedural risk stratification and thresholds for coronary angiography may be one strategy to improve the appropriateness of PCI.
Increasing attention is being given to proper patient selection for coronary procedures to avoid unnecessary procedural risks and costs. This attention is evident in the proliferation of appropriate use criteria to assess patient selection for coronary procedures.1,2 Application of the appropriate use criteria to coronary revascularization has demonstrated wide facility-level variation in the quality of patient selection for elective percutaneous coronary intervention (PCI), with the hospital rate of inappropriate PCI ranging from 0% to 55%.3 These findings suggest opportunities to improve patient selection for coronary procedures as part of improving health care quality.
In light of these findings, strategies to minimize inappropriate PCI have emphasized the interventional cardiologist by ensuring that revascularization is warranted after completion of the diagnostic angiogram, particularly when PCI is considered in the same session (ie, ad hoc PCI).4,5 This approach fails to address the potential importance of patient selection for diagnostic coronary angiography, an invasive procedure requested by a range of health provider types and specialties. As an example, the clinical benefit of coronary angiography and PCI is unclear among patients without ischemic symptoms.1,2,6 Given the potential for a diagnostic-therapeutic cascade, in which an initial diagnostic test triggers subsequent treatments regardless of anticipated clinical benefit,7- 9 it is possible that greater use of angiography in asymptomatic patients leads to more frequent appropriate use criteria–defined inappropriate PCI. Alternatively, because the decision to proceed with PCI can be made independent of the decision to undertake coronary angiography, patient selection for angiography and PCI may be unrelated. Understanding the association between patient selection for coronary angiography and appropriate use of PCI for stable ischemic heart disease may guide future strategies to improve patient selection for both procedures.
We sought to determine if hospitals’ rates of performing elective coronary angiography in asymptomatic patients are associated with their rates of PCI appropriateness in a national sample of hospitals participating in the CathPCI Registry.10 We determined the hospital proportion of asymptomatic patients at coronary angiography as a facility-level measure of patient selection for elective angiography. We emphasized symptom status in assessing patient selection for coronary angiography given the implications of symptoms on the appropriateness of PCI.
Authorization for this study and waiver of written informed consent were granted by Chesapeake IRB (Columbia, Maryland). The CathPCI Registry is the largest register of diagnostic cardiac catheterization and PCI in the United States, with more than 1400 participating centers.11,12 Captured data include patient and hospital characteristics, procedural indication, findings, interventions, and outcomes based on prespecified data elements.10 Data quality assurance is achieved through automatic system validation and reporting of data completeness, education and training for site data managers, and random on-site auditing. The audit process includes more than 50 fields (with fields rotating in a 3-year cycle), with 300 to 625 randomly selected records audited annually at 25 randomly identified sites (ie, 12-25 records per audited site). Details on the National Cardiovascular Data Registry Data Quality Program have been previously described.13,14
We identified 2 239 720 patients undergoing elective (nonacute) diagnostic coronary angiography and 660 932 patients undergoing elective PCI at 1516 CathPCI Registry participating hospitals between July 1, 2009, and September 30, 2013 (Figure 1). Coronary angiography and PCI for acute indications, including acute coronary syndromes, acute myocardial infarction, and cardiogenic shock, were not included. Of patients with elective indications for coronary angiography, we excluded 521 437 patients (23.3%) undergoing angiography performed in consideration of transplantation, prior cardiac transplantation, or the evaluation of cardiomyopathy because the clinical indication for angiography differs in these patient populations relative to patients with stable ischemic heart disease. Of patients undergoing elective PCI, we excluded 272 733 (41.3%) who could not be mapped to the appropriate use criteria because of missing necessary data elements (eg, noninvasive stress test results). We also excluded 393 sites (25.9%) with annual nonacute PCI volume of less than 50 patients to avoid inflation of variance due to small numbers. In addition, because not all participating hospitals report angiography, we excluded 579 hospitals (38.2%) reporting less than 50% more elective diagnostic coronary angiograms relative to elective PCIs (<1.5:1 ratio of coronary angiography to PCI). Our final analytic cohort included 544 hospitals that performed 1 225 562 elective coronary angiograms and 203 158 elective PCIs.
Each PCI in our cohort was mapped to an appropriate use criteria clinical indication using algorithms to assign procedural appropriateness as appropriate, uncertain, or inappropriate based on 2012 appropriate use criteria.1 In these criteria, PCIs are considered inappropriate when the procedure is unlikely to improve the patient’s health status (symptoms, function, or quality of life) or survival.1,3,15
For each hospital, we determined the proportion of asymptomatic patients (“no symptoms, no angina” per CathPCI Registry data element 5000) among those undergoing elective diagnostic coronary angiography. We compared patient-level and hospital-level characteristics across quartiles of hospital proportion of asymptomatic patients at angiography. We also compared baseline patient characteristics at elective PCI by hospital quartile of asymptomatic patients at angiography. Comparisons of patient-level characteristics were performed using linear trend test for continuous variables and Mantel-Haenszel trend test for categorical variables. Comparisons of hospital-level characteristics were performed using Mantel-Haenszel trend test except for the median procedural volumes, which were compared using Kruskal-Wallis test.
We then plotted each hospital’s proportion of inappropriate PCI against its proportion of angiography performed in asymptomatic patients and assessed the association using Spearman correlation coefficient. We compared the hospital median PCI appropriateness ratings by hospital quartile of asymptomatic patients at angiography using Kruskal-Wallis test. Furthermore, we compared the proportion of inappropriate PCI attributable to performance in asymptomatic patients by hospital quartile of asymptomatic patients at angiography using Mantel-Haenszel trend test.
Given that a proportion of PCI could not be mapped to the appropriate use criteria because of a lack of preprocedural stress test results, we performed sensitivity analyses in which we evaluated the influence of the assumption thatstress tests were low risk (worst-case scenario) or high risk (best-case scenario) when noninvasive stress tests results were unavailable. To ensure that findings from our primary analysis did not reflect the inclusion of hospitals that did not report all angiograms to the CathPCI Registry (in the absence of performing PCI), we repeated our analyses after excluding facilities with less than twice as many elective angiograms reported than elective PCIs (<2:1 ratio).
All analyses were performed with statistical software (SAS 9.2; SAS Institute Inc). They were evaluated at a significance level of .05.
Of 1 225 562 patients who underwent elective diagnostic coronary angiography, 308 083 (25.1%) were asymptomatic at the time of angiography. The median hospital proportion of angiography performed among asymptomatic patients was 24.7%, with an interquartile range of 15.9% to 35.9% and a range of 1.0% to 73.6% (Figure 2). Categorized by quartile, the median proportions of asymptomatic patients were 12.1% in the lowest-quartile hospitals, 20.3% in thesecond lowest–quartile hospitals, 30.2% in the second highest–quartile hospitals, and 43.2% in the highest-quartile hospitals (Table 1).
Comparisons across hospital quartiles of asymptomatic patients at angiography were statistically significant for all baseline characteristics given our large sample size; however, most differences were small. Patients at hospitals with higher proportions of asymptomatic patients at angiography were more likely to be male (55.0% vs 56.7% vs 57.7% vs 58.4% from lowest to highest quartile), less likely to be of white race/ethnicity (89.3% vs 87.6% vs 85.8% vs 85.7% from lowest to highest quartile), and more likely to have private insurance (65.6% vs 66.9% vs 69.4% vs 70.8% from lowest to highest quartile) (P < .001 for all). Results of coronary angiography demonstrated slightly higher rates of obstructive coronary artery disease (CAD) among patients at hospitals with higher rates of angiography in asymptomatic patients (43.6% vs 44.7% vs 45.6% vs 45.8% from lowest to highest quartile, P < .001 for trend).
In the evaluation of hospital characteristics, hospitals with a higher proportion of asymptomatic patients at angiography were more likely to be a teaching hospital (32.4% vs 43.4% vs 47.1% vs 53.7% from lowest to highest quartile, P < .001) (Table 2). Procedural volumes were slightly lower at hospitals with higher rates of asymptomatic patients at angiography, although this was statistically significant only for elective angiography volumes.
Among 203 158 patients who underwent elective PCI, comparisons across hospital quartiles of asymptomatic patients at coronary angiography were statistically significant for most comorbidities and risk factors given our large sample size; however, differences were small (Table 3). Patients receiving PCI at hospitals with more asymptomatic patients were slightly less likely to receive at least 2 antianginal medications before PCI (25.1% vs 23.9% vs 23.7% vs 22.9% from lowest to highest quartile, P < .001). The use of preprocedural stress testing was higher among patients receiving PCI at hospitals with higher rates of angiography in asymptomatic patients (61.4% vs 66.8% vs 69.6% vs 68.8% from lowest to highest quartile, P < .001 for trend).
The hospital proportion of angiography performed in asymptomatic patients was positively associated with rates of inappropriate PCI (Figure 3) (Spearman ρ = 0.51, P < .001). Similarly, by hospital quartile of asymptomatic patients at angiography, hospitals with higher rates of asymptomatic patients at angiography had higher median rates of inappropriate PCI (14.8% vs 20.2% vs 24.0% vs 29.4% from lowest to highest quartile, P < .001 for trend) (Table 4). Hospitals with higher rates of asymptomatic patients at angiography also had lower rates of uncertain PCI appropriateness (45.1% vs 44.2% vs 40.2% vs 36.8% from lowest to highest quartile) and lower rates of appropriate PCI (38.7% vs 33.0% vs 32.3% vs 32.9% from lowest to highest quartile) (P < .001 for trend for both). At the patient level, the rate of inappropriate PCI was attributable to more frequent use of inappropriate PCI in asymptomatic patients at hospitals with higher rates of angiography performed in asymptomatic patients (5.4% vs 9.9% vs 14.7% vs 21.6% from lowest to highest quartile, P < .001 for trend) (Figure 4).
Sensitivity analyses demonstrated that the hospital proportion of angiography performed in asymptomatic patients remained associated with rates of inappropriate PCI when missing stress test results were assumed to be high risk (best-case scenario) (Spearman ρ = 0.21) or low risk (worst-case scenario) (Spearman ρ = 0.39) (P < .001 for both). Furthermore, the findings were unchanged in our sensitivity analysis of hospitals reporting at least twice as many elective coronary angiograms as elective PCIs (eTable in the Supplement).
We sought to determine if hospital-level patient selection for diagnostic coronary angiography, as assessed by symptom status at the time of the procedure, is associated with PCI appropriateness. Among 544 hospitals participating in the CathPCI Registry that performed elective coronary angiography in more than 1 million patients, 25.1% of patients were asymptomatic at the time of coronary angiography. We observed marked variation in the hospital rate of angiography performed in asymptomatic patients ranging from 1.0% to 73.6%. Hospitals with higher rates of asymptomatic patients at angiography also had higher rates of inappropriate PCI due to greater use of PCI in asymptomatic patients. Hospitals with higher rates of asymptomatic patients at angiography also had lower rates of appropriate PCI. These findings suggest that patient selection for coronary angiography is associated with the quality of patient selection for PCI as determined by the appropriate use criteria.
In a prior study3 from the CathPCI Registry, the hospital rate of inappropriate PCI was observed to vary from 0% to 55%. This variation has subsequently been demonstrated in other regional PCI quality improvement programs.16,17 In these studies, PCI performed in asymptomatic patients was found to account for almost half of the procedures categorized as inappropriate.3 Other factors accounting for the classification of PCI as inappropriate included submaximal antianginal therapy or low ischemic risk by preprocedural stress testing.3,16 These findings prompted suggestions by interventional cardiology associations to ensure that the patient’s symptom status, medical regimen, and ischemic risk are assessed on completion of the coronary angiogram to confirm that revascularization is warranted before proceeding with PCI.4
Although a strategy of clinical assessment immediately before PCI may minimize inappropriate use of the procedure, our findings suggest an opportunity to address patient selection before proceeding to the catheterization laboratory to optimize the use of angiography and PCI. Our findings of hospital variation in patient selection for angiography as determined by symptom status complement results ofa prior study18 in which the hospital rate of obstructive coronary disease was used as a measure of patient selection for the procedure. In that study, the median hospital rate of obstructive coronary angiography (defined as stenosis >50% in the left main coronary artery or >70% in any other epicardial coronary artery) was 30% and ranged from 15% to 100%. Together, these findings suggest that strategies are needed to improve patient selection for coronary angiography, a procedure requested by a range of health provider types and specialties for more than 1 million US patients annually at a mean cost of $9000 per procedure.19,20
Concerns about a diagnostic-therapeutic cascade have long surrounded the use of coronary angiography and PCI.7- 9 In this cascade, PCI is performed for obstructive coronary lesions identified at angiography regardless of whether revascularization is indicated.21 However, prior investigations were limited to comparison of rates of diagnostic and therapeutic coronary procedures as indirect evidence of the cascade. Our study provides more direct evidence of this cascade based on assessment of the clinical scenarios for angiography and PCI.
While proposals to reassess the indication for PCI at completion of angiography may prevent a diagnostic-therapeutic cascade,4,5,22 our study findings that emphasized patient symptoms in assessment of patient selection for coronary angiography suggest that optimal patient selection for coronary angiography may reduce the occurrence of this cascade altogether. As a corollary, ensuring that the indications for both coronary angiography and potential PCI are addressed before proceeding to the catheterization laboratory may reduce barriers to appropriate use of ad hoc PCI, a procedural strategy that reduces patient inconvenience and cost.23,24 Therefore, the onus of proper patient selection for PCI rests not only with the interventional cardiologist but also with the referring physicians (eg, cardiologists, internists, and family physicians) for coronary angiography.
Possible reasons for a diagnostic-therapeutic cascade in the use of coronary procedures include perceived patient expectations,25 a belief in the benefits of PCI for ischemia, and the open artery hypothesis.21 We observed greater use of preprocedural stress testing among patients receiving PCI at hospitals with higher rates of angiography in asymptomatic patients, a finding that is consistent with the use of screening stress testing to identify silent ischemia leading to angiography and PCI despite uncertain benefit.1,26,27 An emphasis on screening of asymptomatic patients with clinical risk factors may also explain the minimal differences in patient characteristics across hospital quartiles. The appropriate use criteria for coronary revascularization consider PCI in asymptomatic patients with high-risk CAD (ie, left main coronary artery, proximal left anterior descending coronary artery, or 3-vessel CAD) to be at worst of uncertain appropriateness.1 Therefore, our findings do not reflect greater use of PCI in patients with high-risk CAD at hospitals performing more angiography in asymptomatic patients. Potential approaches to addressing gaps in proper patient selection for invasive coronary procedures include greater patient involvement in the decision process,28 patient decision support,29,30 and application of the appropriate use criteria in measurement, reporting, and clinical decision support.
Although the rate of normal coronary angiography findings has been proposed as an indirect measure of the quality of patient selection for angiography,31- 33 our study highlights the limitations of using coronary angiography results to reflect procedural indication. We observed a rate of obstructive coronary disease that was slightly higher at hospitals with a larger proportion of asymptomatic patients at angiography. This suggests that results of angiography may not accurately reflect the quality of patient selection as determined by preprocedural characteristics of clinical decision making. In addition, the facility rate of obstructive coronary disease may be subject to reporter bias and lacks a target for quality improvement in patient selection. Alternatively, an emphasis on procedural indication reflected by symptoms, ischemic risk, optimal medical therapy, and the potential implications of angiographic findings may support strategies to ensure that patients selected for angiography are anticipated to benefit from the procedure.
Our study should be considered in light of the following limitations. First, the decision to proceed to coronary angiography often incorporates an understanding of the patient’s global coronary risk, noninvasive study results, and symptoms. This approach is the basis for recently published appropriate use criteria for diagnostic coronary angiography.2 In our approach to ascertaining patient selection for coronary angiography, we were unable to use these appropriate use criteria because of high rates of missingness for data elements necessary to estimate Framingham risk34 (99% missing) or noninvasive study results (38.2% not performed and 33.4% missing results). Because patient symptoms were uniformly collected, we emphasized this preprocedural data element as our primary mode of ascertaining patient selection for the procedure. Because the decision to proceed with coronary angiography is not a mandate to proceed to PCI, these findings are not a tautology of the appropriate use criteria. Second, our study does not address the clinical outcomes associated with a strategy of performing PCI in asymptomatic patients; although the clinical benefit of coronary angiography in asymptomatic patients is unknown, this does not equate to an inappropriate procedure. Application of the appropriate use criteria to coronary angiography as a more inclusive measure of patient selection in diagnostic coronary angiography is an area for future research. Third, we are unable to ensure that hospitals included in the analysis reported all coronary angiograms to the CathPCI Registry because reporting of coronary angiography is voluntary in the program. However, sensitivity analysis suggests that our findings were not influenced by the inclusion of hospitals with incomplete reporting of angiography data. Fourth, the CathPCI Registry data elements do not include all possible indications for coronary angiography, with a notable example being that angiography is performed before consideration of valve surgery. However, failure to exclude these patients would likely bias our association toward the null because evidence of obstructive coronary disease in this population would likely lead to coronary bypass at the time of valve surgery rather than PCI. Fifth, the generalizability of our findings may be influenced by the limitation of our analysis to hospitals that participated in the CathPCI Registry and performed at least 50 elective PCIs annually. However, evidence has shown that the appropriateness of coronary procedures performed in non–CathPCI Registry hospitals is comparable to that observed within the registry.16 Sixth, we cannot exclude potential misclassification of patient symptoms. However, we are reassured by findings congruent with more aggressive use of coronary procedures at hospitals with more asymptomatic patients, including less use of antianginal medications and lower symptom severity among those patients with angina. Seventh, although we explored facility-level factors associated with rates of angiography in asymptomatic patients, our analysis does not define hospital factors that are correlated with rates of inappropriate PCI. Eighth, as in prior investigations, the lack of preprocedural stress testing limited our ability to map all PCIs to the appropriate use criteria. However, in sensitivity analyses that evaluated the best-case and worst-case scenarios for PCI with missing stress test data, we continued to observe an association between hospital use of diagnostic coronary angiography in asymptomatic patients and inappropriate PCI. Ninth, current appropriate use criteria consider the implications of fractional flow reserve in assessing the physiologic significance of CAD in a limited number of PCI scenarios.1,35,36 However, recent data from the CathPCI Registry demonstrate that fractional flow reserve is performed in only 6% of patients with intermediate coronary lesions (40%-70% stenosis).37 Tenth, we are unable to comment on the potential for the underuse of invasive coronary procedures because we lacked data on patients who did not undergo the procedure.
In conclusion, at 544 US hospitals participating in the CathPCI Registry, approximately 1 in 4 patients was asymptomatic at the time of elective coronary angiography. We observed wide hospital-level variation in the rate of asymptomatic patients at angiography, and higher hospital rates were associated with higher rates of inappropriate PCI and lower rates of appropriate PCI. Current emphasis on proper patient selection for PCI alone fails to address the dramatic variation in the use of upstream diagnostic coronary angiography. Furthermore, strategies to improve patient selection for coronary angiography before proceeding to the cardiac catheterization laboratory may concurrently reduce inappropriate use of PCI and barriers to appropriate use of ad hoc PCI.
Accepted for Publication: May 8, 2014.
Corresponding Author: Steven M. Bradley, MD, MPH, Veterans Affairs Eastern Colorado Health Care System, 1055 Clermont St (Mail Stop 111B), Denver, CO 80220-3808 (email@example.com).
Published Online: August 25, 2014. doi:10.1001/jamainternmed.2014.3904.
Author Contributions: Dr Bradley had full access to all data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Bradley, Spertus, Chan, Patel, Rumsfeld.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Bradley.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Kennedy.
Study supervision: Bradley, Rumsfeld.
Conflict of Interest Disclosures: Dr Spertus reported serving as principal investigator on a contract with the American College of Cardiology Foundation to conduct analyses of the National Cardiovascular Data Registry. No other disclosures were reported.
Funding/Support: Dr Bradley is supported by Veterans Affairs Health Services Research and Development Career Development Award HSR&D-CDA2 10-199. This study was supported by the American College of Cardiology Foundation’s National Cardiovascular Data Registry.
Role of the Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: The views expressed in this article represent those of the authors and do not necessarily represent the official views, position, or policy of the National Cardiovascular Data Registry, its associated professional societies identified at www.ncdr.com, the Department of Veterans Affairs, or the United States Government.
Additional Information: Representatives of the CathPCI Registry research and publications committee approved the final manuscript.