Author + information
- Received August 13, 2001
- Revision received January 3, 2002
- Accepted January 9, 2002
- Published online April 3, 2002.
- H.Vernon Anderson, MD, FACC*,* (, )
- Richard E Shaw, PhD, FACC*,
- Ralph G Brindis, MD, FACC*,
- Kathleen Hewitt, MS, RN*,
- Ronald J Krone, MD, FACC*,
- Peter C Block, MD, FACC*,
- Charles R McKay, MD, FACC*,
- William S Weintraub, MD, FACC*,
- ↵*Reprint requests and correspondence:
Dr. H. Vernon Anderson, Texas Heart Institute and the University of Texas, 6431 Fannin, Suite 1246, Houston, Texas 77030, USA.
Objectives The American College of Cardiology (ACC) established the National Cardiovascular Data Registry (ACC–NCDR) to provide a uniform and comprehensive database for analysis of cardiovascular procedures across the country. The initial focus has been the high-volume, high-profile procedures of diagnostic cardiac catheterization and percutaneous coronary intervention (PCI).
Background Several large-scale multicenter efforts have evaluated diagnostic catheterization and PCI, but these have been limited by lack of standard definitions and relatively nonuniform data collection and reporting methods.
Methods Both clinical and procedural data, and adverse events occurring up to hospital discharge, were collected and reported according to uniform guidelines using a standard set of 143 data elements. Datasets were transmitted quarterly to a central facility for quality-control screening, storage and analysis. This report is based on PCI data collected from January 1, 1998, through September 30, 2000.
Results A total of 139 hospitals submitted data on 146,907 PCI procedures. Of these, 32% (46,615 procedures) were excluded because data did not pass quality-control screening. The remaining 100,292 procedures (68%) were included in the analysis set. Average age was 64 ± 12 years; 34% were women, 26% had diabetes mellitus, 29% had histories of prior myocardial infarction (MI), 32% had prior PCI and 19% had prior coronary bypass surgery. In 10% the indication for PCI was acute MI ≤6 h from onset, while in 52% it was class II to IV or unstable angina. Only 5% of procedures did not have a class I indication by ACC criteria, but this varied by hospital from a low of 0 to a high of 38%. A coronary stent was placed in 77% of procedures, but this varied by hospital from a low of 0 to a high of 97%. The frequencies of in-hospital Q-wave MI, coronary artery bypass graft surgery and death were 0.4%, 1.9% and 1.4%, respectively. Mortality varied by hospital from a low of 0 to a high of 4.2%.
Conclusions This report presents the first data collected and analyzed by the ACC–NCDR. It portrays a contemporary overview of coronary interventional practices and outcomes, using uniform data collection and reporting standards. These data reconfirm overall acceptable results that are consistent with other reported data, but also confirm large variations between individual institutions.
Large-scale multicenter data collection efforts have been used to evaluate cardiovascular procedures (1–6). These have examined in-hospital and short-term success along with adverse outcome events, identified variables associated with higher risks and ultimately have been used to develop risk-adjusted predictive multivariate models. This work has been limited somewhat either geographically or by institutional selection. Furthermore, the data elements collected and their definitions have varied.
To help overcome these limitations, the American College of Cardiology (ACC) established the National Cardiovascular Data Registry (ACC–NCDR) to provide a uniform and geographically comprehensive view of cardiovascular procedures across the country (7). The initial focus has been on the high-volume, high-profile procedures of diagnostic cardiac catheterizations and percutaneous coronary interventions (PCI). This required a parsimonious set of core data elements with standard definitions, thus providing a common language for all institutions. A central data repository was established at Heart House in Bethesda, Maryland. Pilot efforts conducted at several institutions helped launch the effort by collecting data using the new uniform dataset, identifying and correcting ambiguities and errors and transmitting data in a standard file format to the ACC–NCDR. Afterwards, formal data collection began in earnest, and efforts to recruit hospitals into this voluntary effort were initiated. This report represents some of the first data collected and analyzed by the ACC–NCDR. It portrays a contemporary descriptive overview of coronary interventional practices, concentrating on in-hospital outcomes.
Hospitals (institutions) performing diagnostic cardiac catheterizations and PCI participated voluntarily in the ACC–NCDR. Each participating hospital identified an on-site data manager who supervised data collection. Patient demographics, clinical history and procedural information and adverse events that occurred up to hospital discharge were collected using a standard set of 143 data elements (7). The elements had written definitions and were gathered using common specifications for all participants. Data were stored at individual hospitals using commercially available software programs from various vendors precertified by the ACC–NCDR. These datasets were transmitted quarterly to the central facility.
Quarterly datasets received at the ACC–NCDR passed through quality-control checks, identifying missing or improperly coded items. To be acceptable, each quarterly submission had to meet quality measures for accuracy and completeness listed in Table 1. Rejected quarterly datasets could be resubmitted by an institution after correction and could then requalify for inclusion if they passed.
The current report is based on PCI data collected from January 1, 1998, through September 30, 2000, representing 11 quarters of dataset submissions. Demographic, clinical, procedural and outcome variables were analyzed. Angiographic success for a treated coronary lesion was defined as ≥20%-point decrease in diameter stenosis to a postprocedure value <50%, with Thrombolysis In Myocardial Infarction Trial flow grade 3. Clinical success was defined as angiographic success at all treated lesions without death, myocardial infarction (MI) or re-MI or coronary artery bypass graft surgery (CABG). Patients with no adverse events were those discharged without death, MI or CABG, regardless of angiographic success. Urgent or emergency reintervention was captured as a separate event, and this did not constitute an adverse outcome unless it resulted in death, MI or CABG.
Continuously distributed variables are shown as mean ± 1 SD unless otherwise specified. Categorical variables are shown by their frequencies in percent. Descriptive variables for acceptable procedures (included group) were compared to those not included (excluded group). Remaining analyses were done using only included procedures. Several subgroups of special interest were identified, and demographic, clinical, procedural and outcome variables were compared among these subgroups. For all comparisons, continuously distributed variables were tested for differences using unadjusted analysis of variance or ttests, while categorical variables were tested using chi-square tests.
There were 146,907 procedures from 139 hospitals submitted during the collection period. One-third of these (46,615 procedures, 32%) were excluded from analysis because datasets did not pass quality-control screening for the quarter they were submitted. The remaining two-thirds (100,292 procedures, 68%) that passed screening criteria were included. To understand better the effects of the screening process that excluded what appeared to be a large proportion of procedures, demographic and clinical characteristics of included and excluded procedures were compared (Tables 2 and 3). ⇓⇓Outcomes for the two groups are shown in Table 4. There were small differences between included and excluded procedures for a number of variables, which because of the large numbers resulted in low p values according to the statistical tests employed. However, the major outcome variable of in-hospital mortality was equivalent in the two groups (1.4% vs. 1.3%, p = 0.2008, Table 4).
Class I indications for PCI were examined and compared to published ACC criteria (8,9). Results are shown in Table 5. The most common indication for PCI was the presence of class II to IV angina, either with single or multivessel coronary disease. Very few patients had only mild angina. In 10% of procedures, the indication for PCI was acute MI <6 h from onset. Conversely, class I indications were not identified for 5% of the procedures. This varied among individual institutions, from a low of 0 to a high of 38% without class I indications (Fig. 1).
Clinical features and procedure settings
For procedures included in the Registry, the frequencies of single, double, triple and left main disease overall were 41%, 31%, 23% and 5%, respectively (Table 3). At individual hospitals, patients with multivessel disease (MVD) undergoing PCI ranged from a low of 4% of procedures to a high of 76% across the 139 institutions. Patients with treatment of a single lesion ranged only from a low of 37% of procedures to a high of 94%. Hospital outcomes by clinical presentation (Stable Angina, Unstable Angina, Acute MI) and procedure setting (Elective, Urgent, Emergency) are shown in Table 6. Both acute MI presentation and categorization as an emergency procedure were associated with lower success rates.
Success and complication rates by types of devices used are shown in Table 7. The majority of lesions were treated with a stent (71% of lesions, 77% of procedures). However, stent usage ranged from a low of none to a high of 97% of procedures across the 139 hospitals (Fig. 2).
Three subgroups considered to be of special clinical interest were identified (Table 8). These were women compared to men, single-vessel disease patients compared to MVD, and diabetics compared to nondiabetics. Angiographic success rates were slightly higher for women than for men, whereas in-hospital mortality for women was 50% higher than for men (2.0% vs. 1.2%, p < 0.0001). Across the 139 included hospitals, mortality rates in women varied from 0 to 8.0% (Fig. 3A). Patients with MVD had lower angiographic and clinical success rates than single-vessel disease patients. Mortality in MVD patients overall was 1.8%, but varied from 0 to 9.1% across hospitals (Fig. 3B). Patients with diabetes had lower success rates than nondiabetics. Mortality in diabetics overall was 1.8%, but varied from 0 to 9% across hospitals (Fig. 3C). However, the overall very high angiographic and clinical success rates and low complication rates in these three subgroups are noteworthy.
Percutaneous coronary interventions have undergone tremendous growth over the past two decades, driven in part by numerous technological improvements and much better adjunctive drug regimens. Assessing the utility and proper application of these innovations typically is performed in academic and investigative centers under the guidance of the most experienced interventional cardiologists. Procedural success and complication rates after adoption in more widespread clinical practice have been less well studied. Controversies have existed in the past over the appropriate use (both underuse as well as overuse) of PCI in general in the management of coronary disease (10); more specifically, questions have arisen among the cardiology community about clinical outcomes in certain settings like acute MI, women, diabetic patients and the elderly. The high-profile visibility of PCI, its resource demands and its potential for morbidity and mortality clearly identify interventional cardiology as a logical place to begin national efforts within the cardiology community to examine performance measures and outcomes.
Additionally, it has been noted that there are wide regional variations both in utilization rates and clinical outcomes with PCI, both in general and for populations defined by gender, age or insurance status (11). As an example, it is interesting to note that widespread variation existed in the proportion of patients at each hospital who fulfilled ACC class I criteria for undergoing PCI. The ACC leadership recognizes that one major challenge is understanding the sources and reasons for such regional variations (12). These may represent clinical differences in populations, differing availability of resources, personal preferences on the part of patients, personal practice styles of physicians or other factors. Our analysis here was not designed to test hypotheses or draw conclusions about this observation, and much more research will be required to understand these differences fully.
Maintaining high data-quality standards is vital to the success of any national registry. Unacceptable datasets received at the ACC–NCDR are returned to institutions for review, allowing the opportunity for correction and resubmission. Of the procedures submitted during the study period of 1998 to 2000, one-third (32%) were excluded from analysis for quality-control reasons. Comparing data from included and excluded procedures suggests that this likely did not bias results.
The clinical and procedural characteristics of patients reported here, as well as the in-hospital outcomes, appear very similar to other large-scale reports covering approximately the same time frame. The Society for Cardiac Angiography and Interventions (SCAI) (3), the National Cardiovascular Network (NCN) (4), the New Approaches in Coronary Interventions (NACI) Registry (5), and the National Heart, Lung and Blood Institute (NHLBI) Dynamic Registry (6)noted similar age ranges for patients, and similar frequencies of women, diabetics, presence of MVD and intervention for acute MI or unstable angina. The frequency of in-hospital death was 1.4% in our series, and 0.5%, 1.3%, 1.6% and 1.9% in the SCAI, NCN, NACI and NHLBI series, respectively. Likewise, the frequency of same-admission CABG was 1.9% in our series, and 0.5%, 1.7%, 3.4% and 1.5% in the other four series, respectively. Stents were the most common coronary device used in PCI, with a frequency of 71%. Stent usage was 63% and 64% in the SCAI and NHLBI registries, respectively, but was somewhat lower at 38% in the NCN registry. These reports do not indicate the variation in stent usage among participating institutions. The Dartmouth Atlas of Cardiovascular Health Care(11)reported that stents were used overall in 58% of PCI procedures in 1997, with a fourfold variation (from ≅20% to ≅80%) across the geographic regions of the U.S. We noted that stent usage varied widely between institutions submitting data. The reasons for this are unknown but clearly deserve further attention in the future.
Substantial variations existed between hospitals for several important clinical subgroups. The unadjusted mortality rates by institution varied from none to 4.2%, with an overall average of 1.4%. Unadjusted mortality in certain subgroups of interest (women, MVD, diabetics) also varied greatly among hospitals. Our descriptive study was not designed to test hypotheses about this. Risk-adjusted mortality is the subject of a companion study. Future research will concentrate on understanding the sources of these mortality variations more fully.
This report presents the initial results of a large-scale national effort to bring more uniform and balanced analysis to interventional cardiology. The data reported here represent a contemporary overview of PCIs. The ACC has been increasing its attention and resources on assessing the implementation of clinical guidelines into clinical practice, along with taking the lead in the cardiology community for developing performance measures and outcome standards in cardiovascular diseases. The ACC–NCDR is a superb tool that allows hospitals and clinicians to examine clinical practices and outcomes in PCI voluntarily. We have reconfirmed, using uniform definitions and standardized reporting, that overall success rates are high, but wide variations continue to exist in many clinical features of PCI practice as well as outcomes. We anticipate continued contributions from the ACC–NCDR in understanding these differences, and ultimately further improving the quality of cardiovascular care for our patients.
The authors and the ACC-NCDR Oversight and Planning Task Force thank the members of the Publications and Development Subcommittee: Ben D. McCallister, MD; Charles R. McKay, MD; David O. Williams, MD; H. Vernon Anderson, MD; John F. Williams, Jr., MD; Leslee J. Shaw, PhD; Lloyd W. Klein, MD; Martha J. Radford, MD; Michael A. Kutcher, MD; Michael J. Wolk, MD; Peter C. Block, MD; Ralph G. Brindis, MD, MPH; Raymond J. Gibbons, MD; Richard E. Shaw, PhD; Ronald J. Krone, MD; Ronald N. Riner, MD; Ross A. Davies, MD; William S. Weintraub, MD.
The authors acknowledge the technical assistance provided by Jim Emlet, MS in the preparation of the data analyzed.
- American College of Cardiology–National Cardiovascular Data Registry
- coronary artery bypass graft surgery
- length of stay in hospital
- myocardial infarction
- multivessel disease
- New Approaches in Coronary Interventions
- National Cardiovascular Network
- National Heart, Lung and Blood Institute
- percutaneous coronary intervention
- Society for Cardiac Angiography and Interventions
- Received August 13, 2001.
- Revision received January 3, 2002.
- Accepted January 9, 2002.
- American College of Cardiology Foundation
- Hannan E.L,
- Racz M,
- Ryan T.J,
- et al.
- Malenka D.J,
- McGrath P.D,
- Wennberg D.E,
- et al.
- Williams D.O,
- Holubkov R,
- Yeh W,
- et al.
- Ryan T.J,
- Bauman W.B,
- Kennedy J.W,
- et al.
- Smith S.C Jr.,
- Dove J.T,
- Jacobs A.K,
- et al.
- Birkmeyer N.J.O,
- Lucas F.L,
- Malenka D.J,
- et al.
- Beller G.A