Author + information
- Received November 9, 2012
- Revision received January 23, 2013
- Accepted February 26, 2013
- Published online July 30, 2013.
- Bimal R. Shah, MD, MBA∗,†∗ (, )
- Lisa A. McCoy, MS†,
- Jerome J. Federspiel, AB†,‡,§,
- Daniel Mudrick, MD, MPH∗⋮,
- Patricia A. Cowper, PhD†,
- Frederick A. Masoudi, MD, MSPH¶,
- Barbara L. Lytle, MS†,
- Cynthia L. Green, PhD† and
- Pamela S. Douglas, MD∗,†
- ∗Duke University Medical Center, Durham, North Carolina
- †Duke Clinical Research Institute, Durham, North Carolina
- ‡University of North Carolina School of Medicine, Chapel Hill, North Carolina
- §University of North Carolina Gillings School of Global Public Health, Chapel Hill, North Carolina
- ⋮McConnell Heart Health Center, Columbus, Ohio
- ¶University of Colorado, Denver, Colorado
- ↵∗Reprint requests and correspondence:
Dr. Bimal R. Shah, Duke Clinical Research Institute, 2400 Pratt Street, Durham, North Carolina 27705.
Objectives The aim of this study was to determine diagnostic testing patterns after percutaneous coronary intervention (PCI).
Background Little is known about patterns of diagnostic testing after PCI in the United States or the relationship of these patterns to clinical outcomes.
Methods Centers for Medicare and Medicaid Services inpatient and outpatient claims were linked to National Cardiovascular Data Registry CathPCI Registry data from 2005 to 2007. Hospital quartiles of the cumulative incidence of diagnostic testing use within 12 and 24 months after PCI were compared for patient characteristics, repeat revascularization, acute myocardial infarction, and death.
Results A total of 247,052 patients underwent PCI at 656 institutions. Patient and site characteristics were similar across quartiles of testing use. There was a 9% and 20% higher adjusted risk for repeat revascularization in quartiles 3 and 4 (highest testing rate), respectively, compared with quartile 1 (lowest testing rate) (p = 0.020 and p < 0.0001, respectively). The adjusted risk for death or acute myocardial infarction did not differ among quartiles.
Conclusions Although patient characteristics were largely independent of rates of post-PCI testing, higher testing rates were not associated with lower risk for myocardial infarction or death, but repeat revascularization was significantly higher at these sites. Additional studies should examine whether increased testing is a marker for improved quality of post-PCI care or simply increased health care utilization.
Despite the utility of cardiac imaging technologies for guiding clinical decision making, increased utilization of testing in recent decades has raised concerns regarding possible overuse, particularly in light of rising overall U.S. health care costs (1). This increase in cardiac imaging has occurred as the incidence of coronary disease has remained stable, if not decreased slightly (2,3). Cardiac stress testing and diagnostic angiography are among the most commonly performed diagnostic tests and are often used after revascularization, particularly percutaneous coronary intervention (PCI).
Previous studies have demonstrated significant regional variation in cardiac imaging and cardiac catheterization use (4–7) that does not appear to be due entirely to differences in patient characteristics. Cross-sectional and observational studies examining variation in testing have demonstrated that greater diagnostic testing results in increased invasive procedures and interventions (8,9). Nevertheless, it is unclear if these patterns represent increased use or overuse from routine surveillance testing (5). Together, these concerns prompted the American College of Cardiology Foundation to develop appropriate use criteria for imaging stress testing in the hope of decreasing variation and to target scenarios providing the most clinical benefit (10,11).
Prior studies have not examined the institutional variations in testing or the associated clinical outcomes after PCI. To address these gaps, we assessed the relationship between rates of diagnostic testing for coronary artery disease after PCI and the incidence of repeat revascularization, acute myocardial infarction (AMI), and death in a national patient cohort.
The Duke University Medical Center Institutional Review Board granted a waiver of the requirement for informed consent and authorization for this study, and the Duke Clinical Research Institute conducted all analyses.
The study population included all patients undergoing PCI with stenting who were at least 65 years of age, were admitted and discharged between January 2004 and December 2008, and were enrolled in the National Cardiovascular Data Registry CathPCI Registry. The CathPCI Registry is a large, national, clinical registry of patients undergoing cardiac catheterization or PCI. For each patient, the first PCI with stent procedure captured in the CathPCI Registry was considered to be the index event and was treated as the initial unit of analysis. There was a total of 672,617 eligible index events, of which 67% were linked to a single record in the Centers for Medicare and Medicaid Services (CMS) inpatient claims data for patients >65 years of age using an established probabilistic matching method (12,13). Index PCI events were matched to CMS claims data using 3 variables: sex, age, and date of service for the PCI procedure. Matching to the CMS data allowed the identification of subsequent inpatient and outpatient claims. The patient was treated as the unit of analysis.
The linked cohort was restricted to patients receiving only stent type (either bare-metal or drug-eluting stents), allowing comparisons by stent type. Multiple revascularization procedures within a single encounter were considered a single revascularization event. Patients were excluded if they did not have both Part A (inpatient hospital) and Part B (physician services and outpatient hospital) Medicare coverage during their index admissions. Also, patients were excluded if they did not have physician claims for their index interventions or had primary payers besides Medicare (Fig. 1A).
Because Version 3 of the CathPCI Registry data collection form was implemented nationwide on January 1, 2005, we restricted our analysis to patients treated on or after this date; similarly, because we had follow-up data through December 31, 2008, only patients treated before December 31, 2007, were included to ensure minimal 1-year follow-up (Fig. 1B). Patients who did not have fee-for-service Medicare coverage for the entire follow-up period were censored once they ceased to have such coverage.
A 60-day blackout period after PCI was defined for each patient because diagnostic tests during this early period may be performed for cardiac rehabilitation, staging of procedures, or functional capacity assessments. Patients were excluded if they did not survive this blackout period or did not retain both Part A and Part B Medicare coverage. Additionally, patients undergoing stress positron emission tomography, coronary computed tomographic angiography, or stress magnetic resonance imaging after the 60-day blackout period were excluded, because these studies were rarely performed in this cohort. Finally, we limited the sample to those hospitals that performed at least 50 PCI procedures during analysis period to eliminate hospitals with low procedural volumes.
The use of cardiac stress testing with and without imaging after coronary stent implantation was assessed by examining testing patterns after the initial revascularization procedure overall and stratified by type of test, which were identified by Current Procedural Terminology codes (see the Online Appendix for codes). The types, numbers, and dates of any cardiac testing or imaging procedures 60 days after the index coronary stent placement were used to stratify patients on the basis of type of first diagnostic test after stenting. Stress electrocardiography and imaging procedures performed within 1 day of each other were considered a single stress event.
The numbers and dates of repeat catheterizations and coronary revascularization procedures (e.g., PCI or coronary artery bypass grafting [CABG]) after the first diagnostic test were identified using Current Procedural Terminology and International Classification of Diseases-Ninth Revision, Clinical Modification codes. Last, AMI was defined according to International Classification of Diseases-Ninth Revision, Clinical Modification coding (see the Online Appendix for codes).
Patient and hospital baseline characteristics of patients undergoing index revascularization stent procedures were provided overall and by hospital quartiles of cumulative incidence rate stress testing at 12 months after PCI using descriptive statistics (number of observations, mean, standard deviation, median, 25th and 75th percentiles, minimum, and maximum) for numerical (or continuous) variables and with frequencies and percents for categorical variables. Bivariate tests of association were based on either Pearson chi-square tests for categorical variables or Kruskal-Wallis tests for continuous or ordinal variables.
Time to first stress test occurring at least 60 days after the index revascularization episode was calculated using cumulative incidence functions that accounted for administrative censoring; death, AMI, and catheterization were considered competing risks. Cumulative incidence rates at 12 months were calculated separately for each institution. Institutions were categorized on the basis of quartiles of cumulative incidence of stress testing. To examine trends of institutional cumulative incidence rates at 12 months over calendar time, we estimated an intraclass correlation from a linear random-effects model with institution as a random effect.
Cause-specific Cox proportional hazards models were used to estimate hazard ratios associated with risk for death first, AMI first, and revascularization first, with adjustment for baseline variables that were selected a priori on the basis of clinical expertise. A robust variance estimator was used to account for the possible within-cluster correlation (14). We adjusted for age, race, sex, body mass index, acute coronary syndrome at the time of index PCI, peripheral vascular disease, history of congestive heart failure, diabetes, hypertension, dyslipidemia, current smoking status, cerebrovascular disease, family history of coronary artery disease before 55 years of age, previous myocardial infarction more than 7 days before index PCI, chronic lung disease, drug-eluting stent versus bare-metal stent, glomerular filtration rate <30 ml/min or receiving dialysis, previous PCI, previous CABG, current congestive heart failure status, year of index PCI, hospital type (government, private and teaching, or private and nonteaching vs. university), mean annual PCI volume, number of CMS-certified beds, and region. In each model, patients were censored at the end of follow-up or at 365 or 720 days. To describe whether clinical events rates (AMI, repeat revascularization, and death) differed by quartile of testing, we calculated 12-month and 720-day institutional event rates per 100 person-years and calculated the median of the institutional event rates separately for each quartile. To examine outcomes up to 365 and 720 days, we plotted the cumulative incidence using the Kaplan-Meier method and tested for equality of survivor functions using the log-rank test.
All statistical analyses were conducted using SAS version 9.2 or higher (SAS Institute Inc., Cary, North Carolina) and Stata Statistical Software: Release 11 (StataCorp LP, College Station, Texas). All statistical tests were 2 sided, with a significance level of 0.05.
The study population included 247,052 patients who underwent PCI with stenting between 2005 and 2007 at 656 hospitals (Figs. 1A and 1B). The mean duration of follow-up was 756 days (25th and 75th percentiles: 512 and 1,047 days). Overall, 79,741 patients (32.3%) underwent stress electrocardiography, stress echocardiography, or stress nuclear imaging as the first test in the interval from 60 to 365 days after coronary stenting, 18,455 (7.5%) underwent invasive diagnostic coronary angiography as the first test, and 148,856 (60.3%) underwent no testing in the 60-day to 365-day interval after coronary stenting.
Institutional quartiles of testing incidence
We examined the site-level 12-month cumulative incidence rate of stress testing used as the first test within 60 to 365 days after PCI, with either stress testing or cardiac catheterization used as the first test after PCI. The institutional 12-month cumulative incidence rates of any stress testing varied from 8.6% to 66.0% (median 31.6%, 25th and 75th percentiles 24.7% and 38.9%). The intraclass correlation for the cumulative incidence of noninvasive testing between sites over the study years was 0.702 (95% confidence interval: 0.668 to 0.735).
Patient and hospital characteristics
After sites were stratified into quartiles on the basis of their 12-month testing cumulative incidence during the study period, we compared baseline characteristics and demographics of patients at the time of index PCI. This comparison was done first for quartiles of hospitals on the basis of rates of patients undergoing stress testing as the first test after PCI (Table 1).
Stress test first
Patient characteristics were largely similar across hospital quartiles for the cumulative incidence of testing. Compared with the lowest quartiles of testing rate, patients undergoing PCI at the highest testing rate were less likely to be Caucasian and to have a body mass index ≥30 kg/m2, previous congestive heart failure, or acute coronary syndromes at index presentation. There also was some variability of site characteristics across the quartiles in the use of drug-eluting stents, annual PCI volumes, and geography, but there were no clear trends in these differences across quartiles.
Clinical outcomes at least 60 days after index PCI and up to 12 months after index PCI were examined overall and by hospital testing quartile (Fig. 2). Overall, the incidence rates of repeat revascularization (CABG or PCI), AMI, or death within 12 months were 4.1, 2.3, and 5.2, respectively, per 100 person-years.
Adjusted outcomes for patients after PCI
Among hospital quartiles of stress testing use as the first test after PCI, no statistical difference in the cumulative incidence of death (p = 0.187) or hospitalization for AMI (p = 0.51) at 12 months after PCI was found when the higher 3 quartiles were jointly compared with the lowest quartile of testing (quartile 1). In contrast, there was a highly significant difference in the incidence of repeat revascularization at 12 months after PCI (p < 0.001). In pairwise comparisons, there were statistically significant 9% and 20% increases in repeat revascularization among sites in quartiles 3 and 4, respectively, compared with Quartile 1 (p = 0.020 and p < 0.001, respectively) (Fig. 3).
When examining outcomes at least 60 days and up to 720 days after PCI, we found that the cumulative incidence rate of death for the lowest tested quartile (quartile 1) compared with the other 3 quartiles was similar, although statistically significant (10.5%, 10.8%, 10.1%, and 10.1%, respectively, p = 0.011). No significant clinical differences at 720 days were found when comparing quartile 1 with the other 3 quartiles for the cumulative incidence of hospitalization for AMI (4.5%, 4.8%, 4.6%, and 4.3% for quartiles 1 to 4, respectively, p = 0.041). Finally, a significant difference in the rates of repeat revascularization was found across the 4 groups with increasing rates of revascularization as testing quartile increased (13.8%, 13.8%, 14.7%, and 15.4% for quartiles 1 to 4, respectively, p < 0.001). When examined by repeat revascularization type (PCI or CABG), there was no difference in the use of CABG across the 4 testing quartiles (1.93%, 1.92%, 2.04%, and 1.94% for quartiles 1 to 4, respectively, p = 0.567), whereas increased for PCI as testing quartile increased (12.3%, 12.3%, 13.2%, and 14.0% for quartiles 1 to 4, respectively, p < 0.001) at 720 days.
In this national cohort of patients undergoing PCI, there was marked variation in the use of subsequent cardiovascular testing that is not fully explained by differences in patient and hospital characteristics. Furthermore, increased use of testing after PCI is associated with a clear increase in repeat revascularization; however, the variation of testing and downstream revascularization did not result in decreased AMI or mortality.
In the present study, we found significant variation in noninvasive stress testing after PCI across Medicare fee-for-service sites, with a range of 17% to 73% and a mean of 40% at 12 months. These rates parallel previous reports of a cumulative incidence of 36% for either stress echocardiography or stress nuclear testing at 12 months in a non-Medicare population (5). Relative to previous reports in other PCI analyses, these rates are high, where <15% patients develop symptoms requiring reevaluation within 12 months after PCI (15,16).
In our study, there were no significant differences in patient characteristics among sites stratified by quartile of testing use, demonstrating that increased patient risk was not associated with increased testing use and vice versa. Specifically, patient characteristics, such as diabetes or a history of silent ischemia (i.e., no chest pain before index PCI) or bare-metal stent use (vs. drug-eluting stent use), were not more common at sites with more frequent testing. This finding is similar to our prior observations in non-Medicare patients (5), suggesting that significant opportunities exist to improve guidance on appropriate indications for testing.
We used the naturally occurring variation in testing rates across institutions to explore whether variation in the specified outcomes might be associated with different testing intensities. In particular, the variation in testing rates may be interpreted as a surrogate for the different post-PCI management strategies of ischemic symptom-driven testing (lower rates of use) versus surveillance testing (higher rates of use). The lack of association between stress testing use and clinical outcomes of death or AMI suggests that the more intensive testing use (implying a surveillance testing strategy) did not prevent or reduce post-PCI events in the short to medium term. This lack of difference in death related to testing use after PCI parallels findings from prior studies examining the association of testing intensity and outcomes after AMI when controlling for patient characteristics (17). Nevertheless, our data only imply the presence of different diagnostic strategies and the associated outcomes.
In contrast to AMI and death, when examining the physician-guided outcomes of repeat revascularization, a significantly higher utilization (up to 37% more) was found among sites that tested more frequently after PCI. Additionally, we found that the use of repeat revascularization with CABG was constant across testing quartiles, but repeat PCI increased with increased site testing. This clinical cascade mirrors previous studies demonstrating that increased imaging use leads to increased invasive procedures and interventions (8,9). To our knowledge, our study is the first to use observational data to demonstrate that increased testing after PCI is not associated with lower rates of death or AMI, and our findings are consistent with those of multiple randomized clinical trials that have found no reduction in events after PCI (18–20).
Despite the large number of sites and patient data available, our study had several limitations. First, data were limited to fee-for-service Medicare and CathPCI Registry patients. As a result, data on symptoms, clinical presentation, and findings at the time of retesting were unavailable and may differ among groups. It is unknown if higher rates of testing and revascularization result in better symptomatic outcomes, which are also important indicators to patients. Second, we had limited follow-up for patients and could not observe longer term (>24 months) outcomes. Despite similar patients across the utilization quartiles, we could not exclude clinicians who may have correctly identified higher clinical risk and may have acted appropriately on abnormal test results by repeat coronary intervention. However, given the very low rates of subsequent repeat revascularization in this population, it is unlikely that repeat revascularization would be a significant driver of AMI reduction or death in our population. Also, follow-up events following PCI relied on International Classification of Diseases-Ninth Revision, codes as opposed to clinically adjudicated events, which could understate the occurrence of events in our analysis cohort. Finally, we did not have information regarding other valid reasons to test, such as patient or provider preferences or need for patient reassurance.
Our findings suggest that linking clinically rich patient registries with administrative data can create a platform to examine the use of testing after PCI. It is unknown whether increased testing after PCI with increased subsequent repeat revascularization is a marker of higher quality post-PCI care or simply an indication of increased health care utilization. Future studies should examine the indications for and results of post-PCI testing to further assess their use for patients to better understand the associated clinical outcomes.
The authors thank Erin LoFrese, MS, for her editorial contributions to this manuscript. Ms. LoFrese did not receive compensation for her assistance, apart from her employment at the institution at which this study was conducted.
This project was sponsored by the Agency for Healthcare Research and Quality, U.S. Department of Health and Human Services (Rockville, Maryland), as part of the Cardiovascular Consortium and funded under project ID 24-DKE-3 and work assignment number HHSA290-2005-0032-I-TO4-WA3 as part of the Developing Evidence to Inform Decisions About Effectiveness program. The authors of this report are responsible for its content. Statements in the report should not be construed as endorsement by the Agency for Healthcare Research and Quality or the Department of Health and Human Services. Additional support was obtained from the National Cardiovascular Data Registry, American College of Cardiology (Washington, DC). Dr. Shah is a consultant to Castlight Health, LLC. Dr. Masoudi has a contract with the American College of Cardiology as the senior medical officer of the National Cardiovascular Data Registry. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- acute myocardial infarction
- coronary artery bypass grafting
- Centers for Medicare and Medicaid Services
- drug-eluting stent(s)
- percutaneous coronary intervention
- Received November 9, 2012.
- Revision received January 23, 2013.
- Accepted February 26, 2013.
- American College of Cardiology Foundation
- ↵U.S. Government Accountability Office. Medicare Part B imaging services: rapid spending growth and shift to physician offices indicate need for CMS to consider additional management practices. Available at: http://www.gao.gov/products/GAO-08-452. Accessed February 19, 2012.
- Lloyd-Jones D.,
- Adams R.J.,
- Brown T.M.,
- et al.
- Lucas F.,
- DeLorenzo M.A.,
- Siewers A.E.,
- Wennberg D.E.
- Shah B.R.,
- Cowper P.A.,
- O'Brien S.M.,
- et al.
- Hendel R.C.,
- Berman D.S.,
- Di Carli M.F.,
- et al.
- Douglas P.S.,
- Garcia M.J.,
- Haines D.E.,
- et al.
- Brennan J.M.,
- Peterson E.D.,
- Messenger J.C.,
- et al.
- Venkitachalam L.,
- Kip K.E.,
- Mulukutla S.R.,
- et al.
- Strauss W.E.,
- Fortin T.,
- Hartigan P.,
- Folland E.D.,
- Parisi A.F.,
- for the Veterans Affairs Study of Angioplasty Compared to Medical Therapy Investigators
- Pocock S.J.,
- Henderson R.A.,
- Clayton T.,
- Lyman G.H.,
- Chamberlain D.A.