Author + information
- Received October 28, 1999
- Revision received February 22, 2000
- Accepted April 5, 2000
- Published online August 1, 2000.
- Edward L Hannan, PhD∗,* (, )
- Michael J Racz, MA∗,
- Djavad T Arani, MD, FACC†,
- Ben D McCallister, MD, FACC‡,
- Gary Walford, MD,FACC§ and
- Thomas J Ryan, MD, FACC#
- ↵*Reprint requests and correspondence: Dr. Edward L. Hannan, Professor and Chair, Department of Health Policy, Management, and Behavior, School of Public Health, State University of New York, University at Albany, One University Place, Rensselaer, New York 12144-3456
We sought to compare patient outcomes for coronary stent placement and balloon angioplasty.
Since 1994, the number of patients treated only with balloon angioplasty has decreased nationally, whereas the use of coronary stents as an alternative has grown tremendously. The objectives of this study were to compare short- and long-term survival and subsequent revascularization rates for patients undergoing single-vessel balloon angioplasty and coronary stent placement.
New York’s Coronary Angioplasty Registry was used to identify New York patients undergoing either balloon angioplasty or stent placement between July 1, 1994, and December 31, 1996. Statistical models were used to compare risk-adjusted short- and long-term survival and subsequent coronary artery bypass graft surgery (CABG) and percutaneous coronary interventions (PCIs).
No significant differences were found in adjusted in-patient mortality, but patients who had balloon angioplasty were, on average, 1.36 times more likely to have died at any time during the two-year period after the index procedure (p = 0.003). The adjusted in-patient CABG rate was significantly higher for balloon angioplasty (2.72% vs. 1.66%, p < 0.0001), and the adjusted two-year CABG rate was also significantly higher for balloon angioplasty (10.81% vs. 7.25%, p < 0.001). The adjusted two-year rate for subsequent PCIs was also significantly higher for balloon angioplasty (19.6% vs. 14.3%, p < 0.0001). Although measures were taken to eliminate or minimize the effect of selection bias, it should be noted that patients with stents were healthier at hospital admission than patients who had balloon angioplasty.
Stent placement is associated with significantly lower risk-adjusted long-term mortality, CABG and subsequent PCI rates, as compared with balloon angioplasty.
Since its introduction by Gruentzig in 1977, percutaneous transluminal coronary angioplasty (PTCA) has experienced extraordinary growth, with nearly 500,000 procedures performed in the United States in 1997 (1). Much of this growth is attributable to the technological advances in equipment and catheter design that have taken place during the past two decades to make the procedure both safe and successful, at least in the short term. The main limitation of the procedure has been restenosis, a renarrowing of the arterial lesion at the original site of the dilation. Restenosis occurs in ∼35% of patients undergoing PTCA within six months of the procedure. With the demonstration in 1994 that coronary stents could be used not only to “bail out” from the life-threatening complications after failed balloon angioplasty, but also to “back up” or improve suboptimal balloon angioplasty results, the stent has been enthusiastically embraced by the interventional community. By 1996, stents were employed in more than one-half of all percutaneous interventions in both the United States and Europe (2,3).
There is evidence that coronary stenting has been more successful than balloon angioplasty in improving short- and long-term angiographic outcomes, reducing the rate of restenosis, preventing abrupt vessel closure and decreasing the incidence of target vessel revascularization. However, no study to date has had a large enough group to adequately assess the separate impacts of stenting on mortality and the need for coronary artery bypass graft surgery (CABG) during the same hospital stay, as compared with balloon angioplasty alone. The purpose of this study was to use a very large observational data base to compare short- and long-term outcomes of patients treated with balloon angioplasty alone with the outcomes of those also receiving intracoronary stents.
The primary data base used for the study is New York State’s Coronary Angioplasty Reporting System (CARS), which was established in 1991 by the New York State Department of Health (the Department) and its Cardiac Advisory Committee (CAC). The cardiac catheterization laboratories in the 33 hospitals in which angioplasty is performed in New York are responsible for coding the CARS data form to capture the relevant information. To assure reporting accuracy, comprehensive audits of approximately one-half of the hospitals in the registry are conducted on behalf of the Department each year, and several hospitals have been asked to recode all or part of their data as a result of the audits.
For each patient undergoing a percutaneous coronary interventions (PCI) in New York State, CARS contains demographic information; clinical risk factors; patient, operator and hospital identifiers; complications; and discharge status from the hospital. CARS has been used to identify significant risk factors and volume–outcome relations for PCIs (4,5), generate public reports that provide risk-adjusted mortality rates for PCI in New York State hospitals in which the procedure is performed (6) and compare long-term outcomes for angioplasty and CABG (7).
Another data base needed for the study was New York’s vital statistics death file, which identifies all residents of the state who die each year. Because CARS and the death file contain patient social security numbers, deaths subsequent to discharge after the index hospital period were able to be tracked. Subsequent CABGs and PCIs for each patient in New York State were also tracked using CARS and a similar New York State registry for cardiac surgery—the Cardiac Surgery Reporting System (CSRS).
Study group and end points
The total number of patients undergoing balloon angioplasty with or without adjunctive stent placement from July 1, 1994 to December 31, 1996 in the 33 hospitals certified to perform the procedures was 38,468. This group was first limited by excluding non–New York State residents (n = 1,626). Further exclusions were patients who had a previous revascularization (n = 4,745), had the left main coronary artery attempted (n = 91) or had more than one lesion attempted (n = 12,149). Patients who underwent other PCIs (e.g., atherectomy, laser angioplasty) in addition to balloon angioplasty or stent placement (n = 5,752) were also excluded. Some patients were excluded for more than one reason. Patients with more than one lesion attempted were not considered because it was impossible to determine if patients who were coded as stent placement/balloon angioplasty had stent placements on all lesions attempted or if balloon angioplasty was the only intervention attempted on some lesions. This approach led to the identification of 19,792 patients, 12,595 of whom underwent balloon angioplasty and 7,197 of whom underwent stent placements.
Study end points include short-term (in-patient) mortality and CABG; and long-term (two-year) mortality, CABG and subsequent PCIs.
For each of the two procedures, the prevalence associated with each available determinant of mortality was calculated. These variables included the number of vessels with disease (with at least 70% stenosis) and lesion type attempted; patient age, gender, race and ethnicity; a variety of comorbidities; and measures of the patient’s hemodynamic state and ventricular function. All variables, including age and ejection fraction, were treated as categoric variables. Chi-square tests were used to identify significant differences in prevalence rates between the procedures.
Before the development of multivariate statistical models for short- and long-term mortality, subsequent CABG and subsequent PCI, treatment selection bias was controlled for in each model by obtaining a propensity score, which was the probability that a patient would undergo balloon angioplasty rather than stent placement. The propensity score for each patient was obtained by fitting a logistic regression model with a binary dependent variable representing balloon angioplasty. Independent variables consisted of all patient risk factors contained in CARS. The adequacy of the propensity score, ranging from 0 to 1, was confirmed by subdividing it into quintiles and testing for differences in the frequency of model covariates between patients who underwent balloon angioplasty and those who underwent stent placement. The propensity score was then used as an independent variable to adjust for treatment selection bias in subsequent statistical models with patient outcomes as dependent variables (8–10).
The difference in the observed in-patient mortality rates for the two procedures was tested using the chi-square test. To compare risk-adjusted in-patient mortality rates for the two interventions, a stepwise logistic regression model was developed using the LOGISTIC procedure in SAS, version 6.12 (SAS Institute, Cary, North Carolina). Discharge status from the hospital after the procedure, with in-hospital mortality coded as “1,” was used as the binary dependent variable. The propensity score was used in the model to control for selection bias, and candidates for other independent variables included all the demographic and clinical variables available in CARS. After the logistic regression model was developed, risk-adjusted mortality rates (observed/expected mortality rate ratios multiplied by the statewide mortality rate) for the two interventions were tested for significant differences. The same analysis strategy was then repeated using subsequent in-patient CABG as the dependent variables.
To check the results, a second approach to the propensity score method was used. The propensity score was subdivided into five equal ranges, and differences in outcome rates for patients who had balloon angioplasty or stent placement were tested using the chi-square test. In both cases, the same conclusion was reached.
Two-year survival was examined for each intervention while controlling for differences in patient severity of illness, using the stepwise Cox proportional hazards model and the SAS procedure, PHREG (version 6.12). The intervention type was coded as a binary independent variable. Candidate variables in the model included age, gender, ethnicity, race, ejection fraction, previous myocardial infarction (MI), number of vessels with disease, lesion type attempted and numerous comorbidities (Table 1). Ninety-five percent confidence intervals for the hazard ratios were calculated to test for significant differences in survival between balloon angioplasty and stent placement. This was done for all patients and for eight selected subsets of patients.
To measure survival differences between balloon angioplasty and stent placement in percentages rather than in relative terms for each anatomic subgroup, a Cox proportional hazards model for the subgroup was used to construct adjusted Kaplan-Meier survival curves for balloon angioplasty and stent placement, where treatment was used as a stratification factor instead of a model covariate. A chi-square test was used to test for differences in two-year survival among patients undergoing the two procedures.
The long-term analyses just described were then repeated with subsequent CABG and subsequent PCI used as the dependent variables in lieu of mortality, with deaths during the two-year period being treated as censored data.
Table 1 presents the frequency with which each available risk factor was present among patients undergoing balloon angioplasty and stent placement. As noted in Table 1, there were statistically significant differences in the prevalences of all patient risk factors, except for Hispanic ethnicity, malignant ventricular arrhythmia, chronic obstructive pulmonary disease (COPD), renal failure and left main coronary artery disease between the balloon angioplasty cohort and stent placement cohort. In general, stents were deployed in younger, healthier patients. In fact, for 17 of the 21 risk factors that were examined, patients who had balloon angioplasty had significantly higher prevalence rates.
Table 2 presents the unadjusted and adjusted in-patient mortality rates and the same admission CABG rates for patients undergoing balloon angioplasty and stent placement. Although the unadjusted (observed) in-patient mortality rates (1.01% for balloon angioplasty and 0.49% for stent placement) were significantly different (p < 0.001), the respective risk-adjusted rates (0.85% and 0.71%) were not statistically different (p = 0.50).
With respect to CABG in the same admission after the index angioplasty procedure, the unadjusted rates (2.95% for patients undergoing balloon angioplasty and 1.42% for patients undergoing stent placement) were significantly different (p < 0.001). After risk adjustment, the respective rates of 2.72% and 1.66% remained statistically significant (p < 0.001).
Figure 1 presents the logarithms of the adjusted mortality hazard ratios for the entire two-year period for balloon angioplasty/stent placement for all patients in the study and for eight subgroups of patients. The logarithm of the hazard ratio for balloon angioplasty/stent placement is 0.307, which translates (by exponentiation) into a hazard ratio of 1.36 (p = 0.003). Thus, after adjusting for differences in patient risk factors, patients undergoing balloon angioplasty were 1.36 times more likely to be dead at any point in time, as compared with patients undergoing stent placement.
As demonstrated in Figure 1, three of the eight subgroups that were examined (non–left anterior descending coronary artery [LAD] attempted, diabetes and MI within 24 h before the procedure) had a statistically significant hazard ratio. However, patients undergoing balloon angioplasty were significantly more likely to be dead at a fixed time for all eight groups.
Figure 2 contains the two-year adjusted mortality curve for balloon angioplasty and stent placement for all patients. As indicated, the survival advantage of stent placement was present throughout the two-year period. The cumulative adjusted mortality rates for stent placement for half-year increments during the course of two years were 1.7%, 2.5%, 3.0% and 3.8%. The comparable adjusted mortality rates for balloon angioplasty were 2.4%, 3.2%, 4.0% and 4.8%. The adjusted mortality difference at two years was statistically significant (p < 0.001), and differences at other times were at least marginally significant (with respective p values of 0.02, 0.08 and 0.01). Recalling that the adjusted in-patient mortality rate was 0.85% for balloon angioplasty, we can see that all six-month percent increases were 0.8%, except for the period between hospital discharge and six months. During that period, the cumulative mortality rate for balloon angioplasty rose from 0.85% to 2.4%.
Figure 3 presents the logarithms of the adjusted balloon angioplasty/stent placement hazard ratios for subsequent CABG. As depicted in Figure 3, the logarithm of the CABG hazard ratio for balloon angioplasty/stent placement is 0.515, which when exponentiated, translates into a statistically significant 1.67 hazard ratio.
Of the same eight patient subgroups investigated in Figure 1, six (single-vessel disease, two- or three-vessel disease, all patients who did not have their LAD attempted, all patients with no or single-vessel disease who did not have their LAD attempted, all patients who had their proximal LAD attempted and all patients with two- or three-vessel disease who had their proximal LAD attempted) had significant hazard ratios for subsequent CABG. All eight subgroups had hazard ratios >1.
In addition to mortality rates, Figure 2 demonstrates the two-year curves that represent CABG for patients with index procedures of balloon angioplasty and stent placement. As indicated, patients undergoing stent placement were less likely to undergo subsequent CABG throughout the two-year period. The cumulative adjusted CABG rates for patients undergoing stent placement for half-year increments during the course of two years were 4.3%, 5.8%, 6.4% and 7.2%. The comparable adjusted CABG rates for patients undergoing balloon angioplasty were 7.3%, 9.1%, 10.0% and 10.8%. The CABG rates at two years were statistically different (p < 0.001). An examination of the two curves indicates that the two rates were quite different shortly after the procedure was performed, and that this differential remained about the same throughout the two-year period. This suggests that the benefit of stenting is a result of its preventive effect on acute or subacute vessel closure.
Figure 4 presents the logarithms of the adjusted balloon angioplasty/stent placement hazard ratios for subsequent PCI. The logarithm of the PCI hazard ratio is 0.391, which yields a statistically significant 1.48 hazard ratio (p < 0.0001). Also, five of the eight patient subgroups had significant hazard ratios for subsequent PCI.
Figure 2 also contains the two-year curves that represent subsequent PCI for patients with index procedures of balloon angioplasty and stent placement. Patients undergoing stent placement were less likely to undergo subsequent PCI throughout the two-year period. The subsequent PCI rates at two years (14.3% for stent placement and 19.6% for balloon angioplasty) were statistically different (p < 0.001). An inspection of the two curves indicates that the gap between the two procedures widened until about six months subsequent to the performance of the procedure, after which time the gap remained constant. This suggests that higher periprocedural restenosis rates among patients who had balloon angioplasty is the primary determinant of the difference.
Several studies have demonstrated the superiority of coronary stenting as compared with balloon angioplasty. For example, Fischman et al. (11) showed that patients who underwent stenting had significantly larger lumen diameters immediately after the procedure and six months later, a significantly lower rate of restenosis after six months (31.6% vs. 42.1% for balloon angioplasty) and a significantly lower rate of revascularization of the target lesion within six months (10.2% vs. 15.4%).
Serruys et al. (12) of the BElgian NEtherlands STENT (BENESTENT) study group found that patients undergoing stenting had a significantly reduced need for a second coronary angioplasty during the seven months subsequent to the index procedure (relative risk 0.58, p = 0.005) and a significantly lower rate of restenosis (22% vs. 32%, p = 0.02), but more peripheral vascular complications (13.5% vs. 3.1%, p < 0.001). In another randomized clinical trial conducted by the BENESTENT Study Group, Macaya et al. (13) found that the need for repeat angioplasty was significantly lower in the stent group than in the balloon angioplasty group (10% vs. 21%, p = 0.001).
Several other studies reported the superiority of outcomes in patients undergoing stent placement rather than balloon angioplasty for selected groups of patients, including those with small coronary arteries (14), acute MI (15), obstructed coronary artery bypass grafts (16) and restenosis after balloon angioplasty (17). Other studies devoted entirely to stent implantation for selected groups of patients reported encouraging outcomes for patients with an acute MI (18), for multiple stent implantation in single coronary arteries (19), for stenoses of the left main coronary artery in patients who cannot undergo CABG (20), for stenoses in saphenous vein grafts (21,22) and for total occlusions that have been recanalized with PTCA or laser angioplasty (23,24).
Summary of results
This study found that risk-adjusted in-patient mortality rates for balloon angioplasty and stent placement were not statistically different. However, the risk-adjusted in-patient CABG rate for patients undergoing balloon angioplasty was significantly higher (2.72% vs. 1.66%, p < 0.001). Also, patients undergoing balloon angioplasty were, on average, significantly more likely (relative risk 1.36, p = 0.003) to die at any point during the two-year period after the index procedure than were patients undergoing stent placement, after adjusting for differences in preprocedural severity of illness.
The adjusted rate of subsequent CABG after two years was significantly lower for patients undergoing stent placement (7.2%) than for patients undergoing balloon angioplasty (10.8%), and the adjusted rate of subsequent PCI after two years was also lower after stent placement (14.3% vs. 19.6%).
The results reported here should be of considerable interest, although they are not derived from a clinical trial. First, except for deliberate exclusions, the data represent consecutive patients undergoing balloon angioplasty and stent placement in New York for a two-year period, and as such, are more indicative of the full range of treatment in the “real world” where these two procedures are employed. Also, because the volume of each type of procedure was so large, the findings enable us to judge the generalizability of the randomized clinical trials that were conducted earlier and to test for significant differences in mortality and subsequent CABG separately with adequate statistical power.
There are some caveats that are important to note. First, the study is not a randomized clinical trial. This means that there could have been a selection bias such that we did not adequately control for some variable that increased the probability of undergoing balloon angioplasty and the probability of short- and long-term adverse outcomes. In fact, it appears as if stent placements were used cautiously from 1994 through 1996, in keeping with a “learning period” approach to their use. Of 21 of the variables used as candidates in the risk-adjustment process, 17 (81%) had significant prevalence differences that increased the risk of balloon angioplasty intervention versus stent intervention. Consequently, the two patient groups were quite different with respect to preprocedural risk.
We used a propensity score to control for treatment selection bias, in addition to using multivariate statistical models to control for differences in the prevalence of significant risk factors. The propensity score method controls for selection bias more extensively than any method previously used to compare outcomes for stent placement and balloon angioplasty. Nevertheless, the propensity score is ideally comprised of all factors that are predictive of treatment selection, and we did not have access to important determinants such as vessel size. Also, the study took place during a time when the use of coronary stenting changed dramatically, such that there was no steady state in the use of stents. Consequently, the propensity to use stents in early 1994 was much different than the propensity to use them in late 1996, even among patients who had identical risk factors and propensity factors in the two periods.
Another caveat is that the study was confined to patients who had a single lesion attempted, because it was impossible to determine whether patients with multiple lesions attempted underwent a combination of balloon angioplasty and stent placement. However, 71% of the patients who underwent one of the two procedures did have a single lesion attempted, so the results should be at least representative of the majority of patients.
In addition, there is a possibility that the results could be biased as a result of incomplete follow-up of postdischarge deaths. The national death file was unusable because patient names were not available, so New York State vital statistics data were used to identify patients who died after discharge. Because of this, the study was limited to patients residing in New York at the time the procedure was performed. We were not able to capture deaths or subsequent revascularizations for patients who moved to another state subsequent to discharge, but results from a similar study comparing outcomes of CABG and PCI in New York indicate that very few deaths are missed because of patients moving out of state (7).
We look forward to new studies that compare balloon angioplasty and stent placement for large patient populations and to studies that compare stent placement with CABG as a function of the number of vessels with disease and the location of disease.
The authors thank Kenneth Shine, MD, Chair of New York State’s Cardiac Advisory Committee (CAC), and the remainder of the CAC for their encouragement and support of this study, as well as Donna Doran, Rhonda O’Brien, MA, Barbara Lubowski and the cardiac catheterization laboratories of the participating hospitals for their tireless efforts to ensure the timeliness, completeness and accuracy of the registry data.
☆ This work was funded in part by the New York State Department of Health.
- coronary artery bypass graft surgery
- Cardiac Advisory Committee
- Coronary Angioplasty Reporting System
- chronic obstructive pulmonary disease
- left anterior descending coronary artery
- Cardiac Surgery Reporting System
- myocardial infarction
- percutaneous coronary intervention
- percutaneous transluminal coronary angioplasty
- Received October 28, 1999.
- Revision received February 22, 2000.
- Accepted April 5, 2000.
- American College of Cardiology
- ↵1999 Heart and Stroke Statistical Update. Dallas: American Heart Association, 1998.
- Carrozza J.P.,
- Kuntz R.E.,
- Levine M.J.,
- et al.
- Laham R.J.,
- Carrozza J.P.,
- Berger C.,
- et al.
- New York State Department of Health. Angioplasty in New York State: 1995. New York State Department of Health, November 1997.
- Hannan E.L.,
- Racz M.J.,
- McCallister B.D.,
- Ryan T.J.,
- et al.
- Rosenbaum P.R.,
- Rubin D.B.
- Macaya C.,
- Serruys P.W.,
- Ruygrok P.,
- et al.
- Savage M.P.,
- Fischman D.L.,
- Rake R.,
- et al.
- Suryapranata H.,
- van’t Hof A.W.J.,
- Hoorntje J.C.A.,
- et al.
- Itoh A.,
- Colombo A.,
- Hall P.,
- et al.
- Piana R.N.,
- Moscucci M.,
- Cohen D.J.,
- et al.
- Wong S.C.,
- Baim D.S.,
- Schatz R.S.,
- et al.
- Goldberg S.L.,
- Colombo A.,
- Maiello L.,
- Borrione M.,
- Finci L.,
- Almagor Y.
- Serruys P.W.,
- Leon M.,
- Hamburger J.N.,
- et al.