Author + information
- Received December 26, 1996
- Revision received May 14, 1997
- Accepted July 1, 1997
- Published online October 1, 1997.
- Lloyd W Klein, MD, FACCA,* (, )
- Gary L Schaer, MD, FACCA,
- James E Calvin, MD, FACCA,
- Brian Palvas, BSA,
- Jill Allen, MSA,
- Joshua Loew, MDA,
- Eugene Uretz, MSA and
- Joseph E Parrillo, MD, FACCA
- ↵*Dr. Lloyd W. Klein, Jelke Pavilion, Suite 1035, Cardiology Section, Rush-Presbyterian-St. Luke’s Medical Center, 1653 West Congress Parkway, Chicago, Illinois 60612.
Objectives. To assess the relation between individual operator coronary interventional volume and incidence of complications, the in-hospital outcome at a single, moderate volume urban academic center was prospectively collected over a 3-year period.
Background. A minimum of 75 coronary interventions/operator per year may be required in the future to obtain formal certification. However, few data exist regarding individual operator volumes and procedural outcome.
Methods. Between January 1993 and December 1995, 1,389 consecutive procedures were performed or supervised by nine geographic full-time operators: 171 (12.3%) utilized various devices, and 350 (25.2%) involved multivessel coronary intervention. Left ventricular ejection fraction was 59 ± 15% (mean ± SD), and there were 1.7 ± 0.7 vessels diseased (with ≥70% stenosis). Clinical indications included stable angina in 22.5% of cases, unstable angina in 31.9%, acute myocardial infarction (MI) in 2.9%, post MI in 20.6%, shock or acute heart failure in 3.0% and restenosis in 19.1%. In the last consecutive 857 lesions in 655 cases, 20.7% type A, 55.5% type B and 23.8% type C lesions were categorized before coronary intervention.
Results. Average yearly operator volume ranged from 26 to 83 cases (mean 51 ± 26). Each operator has performed a total of 590 ± 268 coronary interventions, with 10.0 ± 4.3 years of coronary interventional experience. The mean angioplasty volume rating for the nine operators was 180 ± 37 (>170 considered adequate). The in-hospital major complication rate was 1.4% (95% confidence interval 0.7% to 1.893%) for all coronary interventions, including death in 3 patients, bypass surgery in 13, arrhythmia in 3 and Q wave MI in 2.
To ascertain how these outcomes compared with standard measures of coronary interventional outcome, four previously published registries were reanalyzed in a similar manner. The rate of complications in the present study was found to be significantly lower than that of the 1992–1993 Society for Cardiac Angiography and Intervention registry (1.9%, n = 19,594, p < 0.05 [excludes ventricular arrhythmias]), the 1994 American College of Cardiology database (3.9%, n = 38,963, p = 0.001), the Mid-America Heart Institute outcome in 1988 (2.3%, n = 5,413, p = 0.02) and the 1985–1986 National Heart, Lung, and Blood Institute Registry (7.2%, n = 1,801, p = 0.001). Odds ratios and 95% confidence intervals showed the outcome in the current study to be at least comparable to the standard registries.
Conclusions. Despite individual operator volumes below those currently being considered for credentialing, the overall institutional outcome was excellent in a diverse and complex patient population.
The cardiology community is currently engaged in resolving the issue of the safety and efficacy of coronary intervention when performed by low volume operators [1–3]. Physician organizations and third-party payers are actively considering whether a minimal number of cases, such as 75 cardiac interventions/year, should be required in the future to obtain formal coronary intervention certification, maintain privileges and qualify for contracts.
This issue stems in part from reports showing that individual operator volumes for coronary artery bypass graft surgery (CABG) are significantly and inversely related to in-hospital mortality and morbidity [4, 5]. In addition, there is a higher rate of angioplasty-related complications in cardiac catheterization laboratories performing a low volume of procedures [2, 6, 7]. The relation between low procedural volume and an increased incidence of complications and mortality is independent of differences in the mix of cases treated at higher volume, compared with lower volume, laboratories . Ryan suggested that it may not be so much a question of procedural skill as a lack of experienced judgment in selecting cases that determines whether low volume operators have a higher complication rate. Several studies have shown [9, 10]the importance of case selection in determining clinical outcome. Additionally, they often do not reflect the benefit of improved catheter and guide wire technology or the benefit of new devices available only in the past 3 or 4 years.
As health care consumers and third-party payers have become aware of these findings, they are increasingly cognizant of operator-specific complication rates and are choosing providers with well defined data that demonstrate superior outcomes. Physician organizations are responding by recommending a minimal operator volume to obtain formal certification and to maintain active privileges. However, the credentialing requirements currently under consideration do not take into account the total volume of procedures performed in a given institution, which may be important because the total experience of support personnel and overall institutional care are also important determinants of outcome.
The present study was undertaken to assess whether multiple, relatively low volume, individual coronary interventional operators would necessarily result in an increased incidence of complications at a single, moderate volume urban academic center over a 3-year period. The overall outcome at our center was compared with those obtained in angioplasty registries acknowledged as the standard.
1.1 Patients and database.
Between January 1, 1993 and December 31, 1995, 1,389 consecutive coronary interventional procedures were performed at Rush-Presbyterian-St. Luke’s Medical Center. The clinical and coronary angiographic findings for all procedures were prospectively entered into a computerized database designed at this institution and based on a substantial modification of the American College of Cardiology (ACC) database . Over 800 data fields were collected for each patient. No patient undergoing an interventional procedure was excluded. All clinical, angiographic and morphologic characteristics, as well as tests of disease severity and extent, were prospectively entered as part of the database. Specific morphologic characterization of each stenosis was collected for the last 655 consecutive cases. Additional angiographic and clinical details were obtained from a previously described catheterization database . The database platform was Paradox 4.5 for Windows, which allows querying of the database on a routine basis; it was maintained by three full-time computer specialists. Each patient form was completed by the assisting interventional cardiology fellow and checked on a routine basis by the attending physician responsible for that patient. The accuracy of the database was assured because the official procedure report was produced from the data entered. Random checks were performed. In addition, the final check was a formal review of all cases with complications at a monthly morbidity and mortality conference. To test the validity of the outcomes recorded in the database, an independent chart review involving 1% of the total patients, selected as a known high risk subgroup (thrombus and acute coronary syndrome), was undertaken. The database was accurate in all but one patient who underwent CABG >1 week after the index interventional procedure, but before hospital discharge; the physician considered the CABG to be unrelated to the procedure, which was clinically successful.
1.2 Variables and definitions.
The following preprocedural, periprocedural and postprocedural variables were collected in the relational database described in the previous section. Baseline demographic variables included age, gender, clinical indication, congestive heart failure class, previous CABG, previous percutaneous transluminal coronary angioplasty and previous myocardial infarction (MI). Baseline angiographic and procedural variables included number of lesions attempted, ACC/American Heart Association (AHA) lesion morphology, devices used, vessels treated and percent stenosis (measured by the caliper method) before and after coronary angioplasty. Outcome variables were occurrence of a complication (see later), in-hospital procedural success (final stenosis <50% and no death, MI or CABG), angiographic success and absence of extensive dissection. Left ventricular ejection fraction and number of diseased vessels (≥70% stenosis) were obtained on the basis of diagnostic catheterization.
Coronary angiography was performed in multiple views using standard catheters. Percent diameter stenosis was measured by the caliper technique. Left ventricular ejection fraction was measured by a modification of the Dodge method for right anterior oblique left ventriculography. Stenosis morphology was characterized prospectively using previously described definitions based on the ACC/AHA Task Force criteria . Non-Q MI was not routinely sought and so was not analyzed as an end point, which is similar to the standard registries.
1.3 Operator volume.
During the time frame of the present study, 22 physicians with privileges performed at least one coronary intervention at this institution. However, the 13 lowest (<15 interventional procedures/year at this institution) volume operators, most of whom have active privileges elsewhere, performed these procedures under the supervision of the 9 highest volume operators. For purposes of analysis, these 13 lowest volume operators and their results were not included in reporting individual primary operator outcomes; however, all cases (including the 4 complications incurred in the 169 affected cases) were included in the complete institution-based analysis and comparison with the standard registries.
Angioplasty volume rating (AVR) was calculated according to the formula of Ryan : AVR = n + 0.2y + 0.3z, where n = number of cases as primary operator/year; y = number of cases/year at the institution; z = number of cases as primary operator for the previous 3 years. For purposes of the present study, the values used were n = z/3 and y = 1,389/3, or 463.
1.4 Outcomes measured.
The clinical adverse outcome variables collected for all patients included 1) death (occurring at anytime during the hospital period; 2) acute Q wave MI (diagnosed by the development of new Q waves on the electrocardiogram (ECG) within 24 h of the procedure); 3) CABG (at any time during the hospital period); 4) sustained ventricular arrhythmias during the procedure; 5) cardiogenic shock or prolonged hypotension within 24 h; and 6) abrupt vessel closure. These outcome variables were selected because these are the end points in standard registries [2, 11, 17, 18](see later). Adverse events and complications were defined as closely to these other registries as possible, consistent with ACC/AHA definitions . ECGs were routinely recorded within 6 to 12 h after the angioplasty procedure; further testing was done if chest pain occurred, if persistent ECG ST segment changes were noted compared with baseline levels or on the cardiac monitor, or if an adverse angiographic event was noted. For the purpose of comparative analysis with the previous studies, the first four of these adverse outcomes were considered “major” complications, and the composite variable was also tested. Patients with more than one complication were counted only once in the analysis of major complications, according the previous hierarchic classification.
Angiographic successwas defined as a final result with <50% diameter narrowing and Thrombolysis in Myocardial Infarction (TIMI) grade 3 flow. If a dissection after balloon angioplasty led to a successful rescue stent procedure, and no other clinical complications occurred, the procedure was considered successful. Abrupt closuresthat did not occur in the catheterization laboratory were identified by the presence of chest pain or ECG changes, or both, resulting in the patient’s return to the catheterization laboratory but were counted only when no other major complication occurred (i.e., not an inclusive event). Similarly, an in-laboratory abrupt closurewas defined as a documented episode of absence of flow in which no other major complication occurred.
1.5 Standard interventional registries.
To ascertain how the overall outcomes in the current study compare with accepted standard interventional outcomes, four well regarded registries were analyzed. These included the 1992–1993 Society for Cardiac Angiography and Intervention (SCAI) registry , the 1994 ACC database , the 1988 report of the Mid-America Heart Institute (MAHI) and the 1988 report of the 1985–1986 National Heart, Lung, and Blood Institute (NHLBI) registry . An analysis of total and individual event complications, as presented in each of the original manuscripts, was performed taking into account how patients with several complications were reported. The SCAI and NHLBI registries did not include ventricular arrhythmias as a major complication, necessitating a separate analysis for comparison. The MAHI database did not break down the overall complication rate by individual complications. These registries were selected because they are widely accepted as representative of expected clinical outcomes; the two analyzing outcomes in the late 1980s remain widely quoted, and the others are of a generation similar to the current study. They are considered the standard by which all laboratories and operators are judged. All yield broadly comparable outcomes overall, although the patients included in each of them differ in some minor respects. The MAHI registry also provides a basis of comparison with a small group of very large volume operators, whereas the other three registries include operators, presumably, of a wide spectrum of volumes. The ACC database includes devices, but the others do not. However, these devices were available to operators in the SCAI series, but the outcomes were not collected.
1.6 Statistical analysis.
Results are presented as mean value ± SD. Chi-square and Fisher exact tests were used to assess univariate significance of categoric variables. Continuous variables not normally distributed were analyzed nonparametrically using Kruskal-Wallis analysis of variance. Odds ratios and 95% confidence intervals were calculated for each adverse outcome variable in each of the four reference trials versus the current study. Least-square regression analysis (Fig. 1) included calculation of Spearman coefficients; obtaining the confidence interval for the slope was done using nonparametric methods for linear regression. A full-time statistician (J.A.) with experience in meta-analysis and clinical trials supervised the study design and performed the analysis.
2.1 Clinical and angiographic outcomes.
The overall clinical and angiographic success rate for all 1,389 coronary interventional procedures over the 3-year period of the present study was 97.1%. The in-hospital complication rate (calculated per patient) was 1.4%; an additional 1.5% failed to meet the angiographic criteria for success, but no complications occurred. Two patients died; 13 patients underwent emergency CABG; 3 had ventricular arrhythmias requiring cardioversion; and 2 had a Q wave MI. Other significant adverse events occurred in an additional 0.6% of patients, including abrupt closure outside the catheterization laboratory in two patients, in-laboratory abrupt closures in 2 and prolonged hypotension in 4 (Table 1).
Balloon angioplasty alone was utilized in the majority of cases, but 171 (12.3%) utilized various new devices and techniques, including directional atherectomy, excimer laser, coronary stenting, transluminal extraction catheter and rotablation. In addition, 25.2% of the procedures involved multivessel coronary intervention. A mean of 1.2 ± 0.41 vessels were dilated per procedure, and the number of stenotic coronary segments treated per procedure ranged from one to six (mean 1.44 ± 0.70). Mean preprocedural and postprocedural stenosis rates were 85.9% and 19.2%, respectively.
2.2 Operator background and results.
During the 3-year period of the present study, all procedures were performed or supervised by nine geographic full-time operators. The physicians in this group have been practicing interventional cardiology for 2 to 14 years (median 12). The average yearly operator volume of these nine physicians ranged from 26 to 83 cases (mean 51 ± 26, including all primary operator and supervised cases). The mean career interventional volume performed by these operators is 590 ± 268 coronary interventions (range 150 to 1,000). As summarized in Table 2the angioplasty volume rating for the nine operators over the 3-year period covered in the present study was 180 ± 37, where >170 is considered adequate and 150 to 170 borderline; five of the nine operators are in the 150 to 170 range. Fig. 1illustrates the lack of a significant linear operator volume–complication relation (r = 0.56, p = 0.12). However, the very small number and volume of operators and outcomes does not entirely exclude the possibility that a more complex relation could exist; nevertheless, such a relation was not discerned at this institution.
2.3 Comparison with four standard registries.
To ascertain whether these results were consistent with the outcomes obtained in large registries, four standard databases were selected for comparison: the 1994 ACC database, the 1992–1993 SCAI registry, the 1988 MAHI registry and the 1985–1986 NHLBI registry. Tables 3 and 4⇓compare the outcomes from the four standard registries with those of the current study. As noted, the overall end point of “any major complication” was consistently and statistically significantly better in the current study than in any of the four standard registries (Table 3). Many of the individual complications also occurred less frequently in the present trial than in the others (Table 4).
Fig. 2shows the odds ratios for risk of complications in the current study versus the other registries and the corresponding 95% confidence intervals. For three of the four registries, there is evidence that the current outcomes are actually statistically better than those of the standard registries: odds ratio (OR) 0.36 (95% confidence interval [CI] 0.23 to 0.56) for the ACC registry, OR 0.56 (95% CI 0.34 to 0.92) for the MAHI registry and OR 0.15 (95% CI 0.09 to 0.26) for the NHLBI registry. For the SCAI registry, a clear trend can be seen: OR 0.61 (95% CI 0.37 to 1.0). There is no evidence that the current outcome was statistically worse than that for the standard registries.
2.4 Patient comparisons.
To determine whether these excellent outcomes were due to differential patient selection, numerous prognostic variables were tested. Mean left ventricular ejection fraction, determined by left ventriculography in 890 patients, was 59 ± 15%. The mean number of diseased vessels was 1.7 ± 0.70, mean age was 63.3 ± 11.7 years, and there was a male preponderance (973 [70.1%] of 1,389 patients). These values are comparable to patients in the three standard registries for which these data are available (Table 5). Additionally, the percentage of multivessel procedures and extent of coronary disease was similar to these registries.
The clinical indications for intervention included stable angina (22.5%), unstable angina (31.9%), acute myocardial infarction (direct or rescue angioplasty [2.9%]), recent (<2 weeks) MI (20.6%), shock (0.7%), acute heart failure (2.3%) and restenosis (19.1%). Stenosis morphology was prospectively collected in the last consecutive subset of 889 lesions in 655 cases. 20.7% were type A, 55.5% were type B, and 23.8% were type C lesions. These values are similar to those of the SCAI and ACC databases (Table 5).
At this single center, the overall institutional outcome of coronary interventional procedures over a 3-year period was at least equivalent, if not superior, to that of several published registries regarded as the standard. These outstanding results are due to many factors, including patient selection, operator skill and long-term experience, open communication and cooperation among colleagues and outstanding nursing and fellowship support. As a consequence, despite individual operator volumes below those currently being considered for interventional credentials, excellent outcomes were obtained in a diverse and complex patient population. Relatively low volume operators can have acceptable outcomes in monitored settings with moderate to high overall volume.
The most likely explanation for these results is that appropriate clinical judgment was used to select out patients at especially high risk, as Ryan has suggested. Each operator may have managed some patients with CABG and managed others medically who would have been treated with a complex interventional procedure by a very high volume operator. However, known clinical and angiographic factors associated with increased procedural risk were analyzed and failed to disclose a systematic selection process compared with that of the standard registries [2, 11, 17, 18]. Although a complete analysis could not be performed because each registry reported patient characteristics variably, comparison with the current study group revealed no differences, when presented. In particular, the incidence of such high risk variables as unstable angina, cardiogenic shock and type C lesions were at least as prevalent in the current study as in the standard registries [1, 9–11]. Patients in the present study had an average of 1.7 diseased vessels and a calculated left ventricular ejection fraction (when measured) of 59%, which is comparable to the registries surveyed. Although the proportion of patients with various risk factors is similar among studies, this is not as rigorous as adjusting for differences in case mix on a patient-level basis, which would have been ideal from a statistical viewpoint. However, on a practical level, the conclusions of the present study are most likely valid and are applicable in a wide variety of clinical settings.
3.1 Applicability to community practice.
The present results should not be interpreted to imply that any physician could expect to have such outcomes in any and all situations. Hartzler discussed the dangers of attempting to practice coronary intervention less than full time, and such concerns have increased dramatically in the 10 years since his warning, due to both the increasing complexity of patients and new device procedures. At our institution, two or three physicians are selected to learn and introduce new techniques and to gain experience before privileges are offered to all operators. The proper relationship among interventionists in the catheterization laboratory must exist so that physicians with less experience can ask for supervision or direct assistance, or both, in the performance of a specific technique.
The adjunctive but separate issue of the relation between individual operator volume and rate of individual complications is briefly addressed. Although no relation was observed, the data cannot be conclusively interpreted because of the low number of operator complications and volume, as reflected in the confidence interval. Additionally, the absence of a linear operator volume–complication relation in the present study could also reflect the absence of larger volume operators or sufficient numbers of low to moderate volume operators. It is also possible that a more complex mathematical relation may exist.
3.2 Implications for credentialing.
The present study raises important questions for physician organizations that are currently considering board examinations and minimal volumes for certification in the subspecialty of coronary intervention. The formula of Ryan et al. for rating experience is an excellent first step because it takes into account both total institutional volume and operator volume over a 3-year period. However, it does not include a factor for the number of years of individual experience, the results that the operator has produced or the capability to consult colleagues. Low volume operators who practice in settings without these advantages should be carefully monitored. The development of an index to partially take these factors into consideration would be optimal, but designating values as cut points that rigidly allow or prevent a physician from performing a procedure that has been successfully accomplished over a number of years is both arbitrary and unlikely to accurately predict future success.
Although the practice of basing credentialing on the performance of a minimal number of procedures is widely used in medicine because it is convenient to measure, it creates certain problems as well. Individuals learn, and experienced professionals maintain their skills, at different rates; consequently, the number of procedures performed in any given year is not a guarantee of proficiency. The requisite number of coronary interventions has never been established by study and is arbitrary. In general, the establishment of such standards is rarely based on a population-based analysis. A final concern is that once a minimal number is agreed to, that number becomes the standard of quality, rather than what is intended (a minimum), as has been the experience with other procedures .
Previous attempts to address the question of low operator volume have produced unclear results, especially with regard to its importance relative to clinical and angiographic predictors of adverse events. Jacob et al. demonstrated a decade ago that very low volume operators under the supervision of more experienced interventionists can have acceptable outcomes. Ellis et al. attempted to construct a model to assess outcomes using volume as one variable but found that quality control, data manipulation and low adverse event frequency prevented a single group of variables from being predictive of all adverse outcomes in all cases. Kimmel et al. created an index of four categories of risk, concluding that risk stratification was possible, but did not consider individual operator volume as a variable; nevertheless, this index had a good predictive overall ability for individual cases.
3.3 “Quality” versus “quantity.”
An important aspect of this problem is that few medium volume interventionists (e.g., many of those who compose the standard registries) discuss their exact procedural volumes publicly for fear of unflattering comparisons with the highest volume operators nationally. “Quality” and “quantity” have become inextricably linked in some minds, preventing a rational discussion of this issue; clearly, there is a linkage, but it is an inferential one. Most existing published reports acknowledge that volume minimums are not a guarantee of quality but then proceed to use an arbitrary value as a surrogate for competence . The outcomes and patient characteristics described in the present study are probably similar to the experiences of many, perhaps most, interventional cardiologists.
Hannan et al. utilized the New York State angioplasty database from 1991 to 1994 to evaluate interventional outcome by volume. Both hospital angioplasty volume and cardiologist volume were inversely related to in-hospital mortality and postprocedural CABG rates. Between 130 and 163 operators in 30 to 31 institutions were evaluated over this time period. Hospitals with ≤400 angioplasties performed annually had higher CABG and mortality rates than those with >600 cases/year, with 400 to 600 being intermediate. Additionally, patients undergoing angioplasty by cardiologists with annual volumes of ≥75 and >250 had higher mortality rates than intermediate operators; those with ≤175 cases/year had higher rates of CABG than those with more. However, the confidence intervals overlapped substantially in many of the comparisons; although statistically significant, only <0.5% to 1.0% absolute differences were seen between groups. The current study exhibits both overall and individual operator outcomes that would place this institution’s results among the highest volume hospitals and operators in that study, suggesting that arbitrary cut points may be useful generally but not as a measure of specific operators or interventional programs. Hannan et al. note that “numerous exceptions” exist; about 50% of the operators they evaluated performed ≥75 cases/year, so these “exceptions” may be common. It is important to analyze them appropriately for the sake of patients, hospitals and physicians alike. Any clinical registry will demonstrate scatter around a relation, often reflecting a wide range; thus, the results from this study are consistent with Hannan et al. .
3.4 Clinical implications.
The implications of these data for establishing operator certification are important. The results suggest that yearly operator volume, although very important, should not be the only factor used in setting standards for interventional cardiology certification or as the full measure of quality. Such standards should ideally include measurements of clinical outcomes, total operator experience and overall results at a particular institution, in addition to individual procedural volume.
☆ To discuss this article on-line, visit the ACC Home Page at http://www.acc.org/membersand click on the JACC Forum.
- American College of Cardiology
- American Heart Association
- coronary artery bypass graft surgery
- confidence interval
- electrocardiogram, electrocardiographic
- Mid-America Heart Institute
- myocardial infarction
- National Heart, Lung, and Blood Institute
- odds ratio
- Society for Cardiac Angiography and Intervention
- Received December 26, 1996.
- Revision received May 14, 1997.
- Accepted July 1, 1997.
- The American College of Cardiology
- Ellis SG,
- Omoigui N,
- Bittl JA,
- et al.
- Ritchie JL,
- Phillips KA,
- Luft HS
- for the Registry Committee of the Society for Cardiac Angiography and Interventions,
- Kimmel SE,
- Berlin JA,
- Strom BL,
- Laskey WK
- ↵Preliminary tabulations from the angioplasty volume. Bethesda (MD): ACC National Database Library, December 1, 1994.
- Uretz E,
- Murray T,
- Millar R,
- et al.
- Ryan TJ,
- Faxon DP,
- Gunnar RM,
- et al.
- Klein LW,
- Kramer BL,
- Howard E,
- Lesch M
- Park DD,
- Laramee LA,
- Telrstein P,
- et al.
- Holmes DR Jr.,
- Holubkov R,
- Vlietstra RE,
- et al.
- Hannan EL,
- Racz M,
- Ryan TJ,
- et al.