Author + information
- Received July 27, 2007
- Revision received September 19, 2007
- Accepted October 1, 2007
- Published online February 19, 2008.
- Dennis T. Ko, MD, MSc⁎,†,1,⁎ (, )
- Yongfei Wang, MS‡,
- David A. Alter, MD, PhD†,2,
- Jeptha P. Curtis, MD‡,
- Saif S. Rathore, MPH‡,
- Therese A. Stukel, PhD†,
- Fredrick A. Masoudi, MD, MSPH, FACC§,
- Joseph S. Ross, MD, MHS‖,3,
- JoAnne M. Foody, MD, FACC‡ and
- Harlan M. Krumholz, MD, SM, FACC‡,¶
- ↵⁎Reprint requests and correspondence:
Dr. Dennis Ko, Institute for Clinical Evaluative Sciences G1 06, 2075 Bayview Ave, Toronto, Ontario, Canada M4N 3M5.
Objectives We evaluated whether appropriateness and baseline risk of cardiac catheterization varied according to regional intensity of invasive therapy after acute myocardial infarction (AMI), and whether AMI mortality varied according to invasive intensity regions.
Background Marked regional variations exist in cardiac invasive procedure use after AMI within the U.S.
Methods We performed an analysis of 44,639 Medicare fee-for-service beneficiaries hospitalized with AMI between 1998 and 2001. Invasive procedure intensity was determined based on overall cardiac catheterization rates for Medicare enrollees. Cardiac catheterization appropriateness was determined by the American College of Cardiology/American Heart Association classification and baseline risk was estimated using the GRACE (Global Registry of Acute Coronary Events) risk score. The primary outcomes of the study were cardiac catheterization use within 60 days and 3-year mortality after hospital admission.
Results Higher invasive intensity regions were more likely to perform cardiac catheterizations on class I patients (appropriate) (RR 1.38, 95% confidence interval [CI] 1.27 to 1.48), class II patients (equivocal) (RR 1.42, 95% CI 1.31 to 1.53), and class III patients (inappropriate) (RR 1.29, 95% 0.97 to 1.67) compared with low-intensity regions after adjusting for patient and physician characteristics. The overall cardiac catheterization use was 5.2% lower for each increase in GRACE risk decile, and this relationship was observed similarly in all regions. Risk-standardized mortality rates of AMI patients at 3 years were not substantially different between regions.
Conclusions Although higher-risk patients and those with more appropriate indications may have the most to benefit from an invasive strategy after AMI, we found that higher-invasive regions do not differentiate procedure selection based on the patients’ appropriateness or their baseline risks.
Although published recommendations on cardiac interventional procedure indications and effectiveness are widely available (1–3), marked regional variations exist in the use of cardiovascular procedures (4,5). Indeed, previous evidence has demonstrated that there exists a 2-fold regional difference in the use of cardiac catheterization after acute myocardial infarction (AMI) within the U.S. (5,6). These regional differences persist after adjusting for patient, physician, and hospital factors (4,6–8). However, the extent to which regional procedure variations may be attributed to performing cardiac catheterizations in AMI patients with different levels of procedure appropriateness or different baseline risks is unknown. Specifically, are regions of higher invasive intensity more likely to perform cardiac catheterizations among more appropriate and higher-risk patients where benefits are the largest? Alternatively, are regions of lower invasive intensity more selective in performing cardiac catheterization in more appropriate or higher-risk patients due to resource constraints?
Data evaluating the appropriateness of cardiac catheterization use are mixed, with some studies reporting that procedures are used more appropriately in higher-invasive regions and others finding no difference in their use (9–13). In contrast, studies evaluating the relationship between patients’ baseline risk and cardiac catheterization have consistently shown less frequent procedure use among higher-risk patients (14,15), a subgroup that has been demonstrated to benefit the most from an invasive strategy (3,16–18). However, none of these studies has investigated whether this pattern varies by procedures performed per capita within a particular region.
Accordingly, we sought to determine whether appropriateness of cardiac catheterization use varied according to regional intensity of invasive therapy. Second, we examined whether baseline risks of those who received cardiac catheterizations varied in different invasive regions. Third, we examined whether AMI patients hospitalized in higher invasive intensity regions had better long-term mortality than those in lower-intensity regions. To address these questions, we used data from the National Heart Care (NHC) project, a Centers for Medicare and Medicaid Services initiative designed to improve the quality of care for Medicare beneficiaries hospitalized with AMI in the U.S.
The NHC project
The NHC project included a random sample of Medicare fee-for-service beneficiaries hospitalized in March 1998 to April 1999 and March 2000 to April 2001 with a principal discharge diagnosis of AMI (International Classification of Diseases-Ninth Revision-Clinical Modification [ICD-9-CM] code 410) from each state and the District of Columbia (19).
The NHC study sample included patients ≥65 years of age who had a confirmed AMI, defined as elevation in troponin I or T level, creatine kinase–muscle-brain isoenzyme fraction >0.05, lactate dehydrogenase (LDH) level of more than 1.5 times normal with LDH1 levels greater than those of LDH2, or 2 of the following 3 criteria: chest pain, a 2-fold elevation of the creatine kinase level, or a new AMI documented on the official electrocardiogram report (19). We excluded patients who had repeat AMI hospitalizations during the study period to have each patient represented in the database with 1 index hospitalization. We did not investigate the relationship of cardiac catheterization use and reinfarction. Patients who were transfered in from another hospital were excluded, because we did not have the medical records that described the hospitalization at the first hospital. We also excluded 1,998 patients because of the inability to determine cardiac catheterization utilization after hospital discharge. The demographics and clinical characteristics of these patients did not differ substantially from patients included in the study.
Appropriateness of cardiac catheterization
We evaluated cardiac catheterization appropriateness based on the American College of Cardiology/American Heart Association (ACC/AHA) guidelines and appropriateness criteria previously reported (Online Appendix) (10,20,21). The strong-indication group consisted of patients in whom cardiac catheterization was generally recognized as “beneficial, useful, and effective” (ACC/AHA class I). We did not include “reinfarction during hospitalization” as a class I indicator, because this data field was not captured in our dataset. The equivocal group (ACC/AHA class II) included patients who had conditions for which the effectiveness of cardiac catheterization was unclear. The weak indication group (ACC/AHA class III) consisted of patients for whom cardiac catheterization was considered unlikely to be effective (Online Appendix) (10,20,21).
Baseline risk of AMI patients
We determined baseline risk of hospitalized AMI patients using the GRACE (Global Registry of Acute Coronary Events) prediction risk score (22). The GRACE mortality model was developed using an international registry of acute coronary syndrome patients and includes prediction variables such as age, admission characteristics (heart rate, Killip class, systolic blood pressure, cardiac arrest during presentation, ST-segment deviation), and laboratory values (serum creatinine, positive cardiac markers) (22). Although the GRACE risk score was developed using patients with acute coronary syndrome, it has been validated in an AMI population (23).
The primary outcome was the use of cardiac catheterization within 60 days of AMI admission. Cardiac catheterization use was determined by linking the hospital medical record and Medicare Part A billing records for ICD-9-CM codes associated with the procedure (37.22, 37.33, 88.53 to 88.57). The secondary outcome was 3-year mortality after AMI admission, as determined by the Medicare enrollment database and billing records.
Regional intensity of cardiac invasive procedures
Hospital referral regions (HRRs) are discrete geographic regions defined on the basis of travel distance to tertiary care centers among Medicare beneficiaries (24). In each HRR, we first determined the number of cardiac catheterizations performed per Medicare enrollee residing in that area. We then arranged HRRs into 3 groups to represent low, intermediate, and high invasive intensity regions based on the number of cardiac catheterizations performed per 1,000 Medicare enrollee in each region (for example, low invasive intensive regions had the lowest tertile of cardiac catheterizations use). These data were obtained by linking our medical hospitalization AMI records with the 1999 Aggregated Surgical Discharge Data File using residential or hospital zip codes.
We first compared the demographic and clinical characteristics of AMI patients hospitalized in regions of different invasive intensity. Categoric variables were compared among the 3 invasive regions using chi-square tests, and continuous variables were compared using F tests from analysis of variance.
We then developed multivariate regression models to examine the effects of regional invasive intensity on cardiac catheterization use, adjusting for patient characteristics (age, Killip class, systolic blood pressure, heart rate, serum creatinine, cardiac resuscitation, ST-segment elevation, cardiac enzymes, chronic obstructive pulmonary disease, dementia, mobility, urinary continence), and physician characteristics (age, gender, race, specialty). We also examined the effects of regional intensity with mortality, adjusting for patient, physician, and hospital characteristics (mean AMI volume, cardiac care facility level, teaching status). All analyses incorporated probability weights based on the inverse sampling fraction for the population size of each state, and all models were adjusted for clustering. Odds ratios were converted to estimated relative risks (RRs) (25). Physician characteristics were obtained using the American Medical Association Masterfile, and hospital data were obtained from the 1998 and 2000 American Hospital Association Survey of Hospitals.
The relationship between cardiac catheterization use and baseline risk in the 3 invasive intensity regions was first examined using scatterplots of cardiac catheterization rates and GRACE risk score deciles. We modeled the relationship using multiple linear regression models with catheterization use as the dependent variable and risk deciles as the independent variables. A series of secondary analyses were then undertaken to examine the robustness of our results. First, we examined the relationship based on different appropriate cardiac catheterization indications. Second, we examined the relationship based on whether AMI patients had ST-segment elevation or not. Finally, we modified the GRACE risk score to include other comorbidities. In all of the above additional analyses, our overall results did not change significantly.
We conducted the statistical analyses using SAS software version 9.1 (SAS Institute, Cary, North Carolina). All statistical tests were 2 tailed. Use of the NHC project database was approved by the Yale University School of Medicine Human Investigation Committee.
Baseline characteristics, procedure indications, and regional intensity
The study cohort consisted of 44,639 Medicare beneficiaries who were hospitalized with AMI. The mean age of the cohort was 78.4 years; 51.2% were female; 27.9% had ST-segment elevation (Table 1). Demographic and clinical characteristics of AMI patients hospitalized in different invasive-intensity regions were not substantially different (Table 1). The AMI patients in high-invasive regions were most often admitted to hospitals that had advanced cardiac invasive facilities (32.5% in low invasive intensity regions, 40.1% in intermediate invasive intensity regions, and 50.1% in high invasive intensity regions; p < 0.001) (Table 1).
The proportion of AMI patients who had ACC/AHA class I indication for cardiac catheterization was slightly higher among those hospitalized in lower invasive intensity regions (45.6% vs. 43.6% in intermediate invasive intensity regions vs. 41.9% in high invasive intensity regions; p < 0.001) (Fig. 1). Conversely, the proportion of AMI patients who had class II indications was slightly higher among those hospitalized in higher-intensity regions. The proportion of patients who had class III indication for cardiac catheterization was not substantially different among the regions (Fig. 1).
Cardiac catheterization, procedure appropriateness, and regional intensity
Overall, 39% of the patients underwent cardiac catheterization within 60 days of AMI, and there was significant geographic variation across the U.S. (Fig. 2). Cardiac catheterization rates were 45.1% among patients who had ST-segment elevation AMI and 36.6% among patients who had non–ST-segment elevation AMI. Cardiac catheterization rates were 41.7% among patients who had class I indications, 39.1% among patients who had class II indications, and 19.8% among patients who had class III indications. High invasive intensity regions had the highest rates of unadjusted and adjusted use of cardiac catheterization among AMI patients in each of the ACC/AHA procedure classes (Figs. 3A and 3B). High invasive intensity regions also had the unadjusted highest rates of coronary revascularization within 60 days of an AMI (21.3% in low-intensive regions, 24.3% in intermediate-intensive regions, 28.2% in high-intensive regions; p < 0.001).
Higher invasive intensity regions had a 38% increased RR (1.38, 95% confidence interval [CI] 1.27 to 1.48) of performing cardiac catheterization among class I patients, 42% increased RR (1.42, 95% CI 1.31 to 1.53) among class II patients, and 19% increased RR (1.29, 95% CI 0.97 to 1.67) among class III patients compared with low invasive intensity regions after adjusting for clinical and physician characteristics (Table 2). Similarly, intermediate invasive intensity regions had increased RRs of performing cardiac catheterization compared with low-invasive regions among patients who had class I (RR 1.19, 95% CI 1.09 to 1.29) and class II indications (RR 1.18, 95% CI 1.08 to 1.28) but not among class III patients (RR 0.95, 95% CI 0.70 to 1.26) (Table 2). We did not observe significant interaction between regional invasive intensity and procedure appropriateness level in predicting the overall use of cardiac catheterization.
Cardiac catheterization, baseline risk, and regional intensity
Figures 4 and 5⇓⇓ illustrate the relationship between utilization of cardiac catheterization and baseline risk of AMI patients estimated by GRACE risk score in each invasive region. In all invasive regions, lower utilization of cardiac catheterization was observed with progressively higher baseline risk. On average, the use of cardiac catheterization was 5.2% lower for each decile increase in the GRACE risk score. This relationship existed across the full spectrum of baseline risk among all AMI patients (Fig. 4) and among patients who had class I indications (Fig. 5). Furthermore, rather than preferentially selecting higher risk patients, higher-intensive regions performed more procedures across the entire risk spectrum.
Mortality by regional invasive intensity
Risk-standardized mortality rates were not significantly different at 3 years after AMI hospitalization (46.7% [95% CI 45.5% to 47.9%] in low-intensity regions vs. 47.5% [95% CI 46.3% to 48.7%] in intermediate-intensity regions vs. 48.6% [95% CI 47.3% to 50.0%] in high-intensity regions; p = 0.22).
In this large cohort of hospitalized AMI patients, we found that the highest invasive intensity regions performed more cardiac catheterizations among patients who had class I indications, that is, procedures that were considered useful and effective or appropriate. To a lesser extent, we also found that the highest invasive intensity regions performed more cardiac catheterizations among patients with class III indications, procedures that were considered to be ineffective or inappropriate. Furthermore, none of the regions selected higher baseline risk patients for cardiac catheterizations after AMI. As a result, cardiac catheterization is used most frequently in the lowest-risk patients and least frequently in the highest-risk patients. Finally, hospitalization of AMI patients in higher invasive intensity regions was not associated with better long-term mortality compared with hospitalization in lower-intensity regions.
Previous studies examining whether appropriateness of cardiac invasive procedures relates to invasive intensity have been inconsistent (9–12). In a study comparing the appropriateness of coronary angiography and coronary artery bypass surgery between New York State and Canada, McGlynn et al. (9) demonstrated that the proportion of appropriate and inappropriate patients who underwent these procedures was similar. In contrast, other data have shown that AMI patients with class I indications are less likely to receive cardiac catheterization in the managed care or the Veterans Affairs health care system, both of which have less procedural capacity compared with fee-for-service systems (10,11). Our findings confirmed that higher invasive intensity regions performed more cardiac catheterizations among class I AMI patients. Unlike many other studies, we also found that higher invasive intensity regions performed more procedures among class III patients, a subgroup in which the procedure is considered inappropriate.
Another major finding was the discordant utilization pattern of cardiac catheterization compared with guideline recommendations (2,3). In this study, cardiac catheterization use among class I AMI patients was only 42%, and the utilization rate was 20% among class III patients. National societies such as the ACC and the AHA have invested substantial efforts to develop practice guidelines to assist physicians in AMI management (2,3). These efforts are likely responsible in part for the dramatic improvements in the use of evidence-based medical therapy. Yet, the significant discordant treatment patterns we observed raise concerns about the uptake and acceptance of these guideline recommendations in cardiac interventional management, which are more difficult to implement than prescribing medical therapy in clinical practice. Ongoing efforts to develop appropriateness criteria for cardiac catheterization and coronary revascularization may provide a better framework that can be better applied in clinical practice.
Although the relative and absolute survival benefits associated with cardiac catheterization after AMI have been shown to positively correlate with baseline risk (26), no study to our knowledge has evaluated whether higher-invasive regions optimize the use of coronary angiography through selection of higher-risk patients. We found that higher-invasive regions performed more cardiac catheterizations across the entire risk spectrum without taking into consideration the potential added benefit among higher-risk patients. We also found a pattern of risk avoidance in which cardiac catheterization rates progressively declined as baseline risk increased. The practice pattern of diminishing utilization of effective therapy in higher-risk patients has been shown previously and termed the “treatment-risk paradox” (6,27). Because the majority of procedures are concentrated among lower-risk patients, where only small mortality benefits are expected, our observation may provide an explanation as to why previous studies have not been able to demonstrate survival benefits of an invasive therapy in clinical practice (5,6,28–30). There remains a need to understand how to optimize the effectiveness of an invasive therapy in clinical practice.
Several limitations of our study merit consideration. First, we were unable to identify the reasons underlying the discordant treatment-risk relationship. However, it has been suggested that in treating higher-risk patients with multiple comorbidities, many factors often lead to a “hands-off” approach in which interventional procedures are often withheld from higher-risk patients (31). Second, our findings are specific to the patient population and the time period studied, but there is no information to suggest that these patterns have changed since then. Finally, we used practice guidelines developed by the 1996 ACC/AHA guidelines (20) to classify cardiac catheterization indications to reflect practice recommendations when this study cohort was assembled and to promote comparisons of the findings with previous studies. One of the major changes of contemporary guidelines relates to the endorsement of routine cardiac catheterization among non–ST-segment elevation AMI. Therefore, we performed additional analyses combining AMI patients who had class I and class II procedure indications and found similar results.
Our study shows that variation in the use of cardiac catheterization occurs without regard for the strength of indication for the procedure. Higher-utilization regions perform more indicated, nonindicated, and discretionary procedures than lower-utilization regions. The higher rates are not differentially the result of performing more procedures in the patients with the strongest indications. This pattern may account, in part, for the lack of a strong association between the use of procedures and better outcomes.
For a table of the ACC/AHA appropriateness criteria for cardiac catheterization, please see the online version of this article.
↵1 Dr. Ko is supported by a Clinician Scientist Award from the Heart and Stroke Foundation of Ontario.
↵2 Dr. Alter is supported by a Career Investigator Award from the Heart and Stroke Foundation of Ontario.
↵3 Dr. Ross is supported by Department of Veterans Affairs Health Services Research and Development Service project no. TRP-02-149.
The analyses upon which this publication is based were performed under contract no. 500-02-CO-01 entitled “Utilization and Quality Control Peer Review Organization for the State of Colorado,” sponsored by the Centers for Medicare and Medicaid Services (CMS, formerly the Health Care Financing Administration), Department of Health and Human Services. The content of the publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. government. The authors assume full responsibility for the accuracy and completeness of the ideas presented. This article is a direct result of the Health Care Quality Improvement Program initiated by CMS, which has encouraged identification of quality improvement projects from analysis of patterns of care and therefore required no special funding on the part of this contractor; tracking no. HIQIOSCHEARTCARE-001-OK-0506. Ideas and contributions to the author concerning experiences in engaging with issues presented are welcomed. The CMS did not play a role in the design and conduct of the study nor in the analysis and interpretation of the data.
- Abbreviations and Acronyms
- American College of Cardiology/American Heart Association
- acute myocardial infarction
- lactate dehydrogenase
- National Heart Care
- Received July 27, 2007.
- Revision received September 19, 2007.
- Accepted October 1, 2007.
- American College of Cardiology Foundation
- Smith S.C. Jr.,
- Feldman T.E.,
- Hirshfeld J.W. Jr..,
- et al.
- Antman E.M.,
- Anbe D.T.,
- Armstrong P.W.,
- et al.
- Braunwald E.,
- Antman E.M.,
- Beasley J.W.,
- et al.
- Ko D.T.,
- Krumholz H.M.,
- Wang Y.,
- et al.
- Alter D.A.,
- Tu J.V.,
- Austin P.C.,
- Naylor C.D.
- Ross J.S.,
- Ho V.,
- Wang Y.,
- et al.
- Fox K.A.,
- Anderson F.A.,
- Dabbous O.H.
- Ryan T.J.,
- Anderson J.L.,
- Antman E.M.,
- et al.
- Bradshaw P.J.,
- Ko D.T.,
- Newman A.M.,
- Donovan L.R.,
- Tu J.V.
- Wennberg D.E.,
- Dartmouth Atlas of Cardiovascular Health Care Working Group
- Van de Werf F.,
- Gore J.M.,
- Avezum A.,
- et al.