Author + information
- Received July 31, 2009
- Revision received October 13, 2009
- Accepted October 15, 2009
- Published online December 1, 2009.
- Eric L. Eisenstein, DBA⁎,⁎ (, )
- William Wijns, MD, PhD†,
- Jean Fajadet, MD‡,
- Laura Mauri, MD, MSc§,
- Rex Edwards, BA⁎,
- Patricia A. Cowper, PhD⁎,
- David F. Kong, MD, AM⁎ and
- Kevin J. Anstrom, PhD⁎
- ↵⁎Reprint requests and correspondence:
Eric L. Eisenstein, Assistant Professor in Medicine, Duke Clinical Research Institute, 2400 Pratt Street, Terrace Level 0311, Durham, North Carolina, 27705
Objectives This study was designed to evaluate long-term clinical and economic outcomes for subjects receiving Endeavor drug-eluting versus Driver bare-metal stents (both Medtronic CardioVascular, Santa Rosa, California).
Background Early studies found that the drug-eluting stent (DES) was a clinically and economically attractive alternative to the bare-metal stent; however, associations between DES and very late stent thrombosis suggest that longer follow-up is required.
Methods We used clinical, resource use and follow-up data from 1,197 subjects randomized to receive Endeavor (n = 598) versus Driver (n = 599) stents in ENDEAVOR II (Randomized Controlled Trial to Evaluate the Safety and Efficacy of the Medtronic AVE ABT-578 Eluting Driver Coronary Stent in De Novo Native Coronary Artery Lesions) study with Medicare cost weights and quality of life adjustments applied from secondary sources. We compared differences through 4-year follow-up (1,440 days).
Results Patients in both treatment groups had similar baseline characteristics. The use of Endeavor versus Driver reduced 4-year target vessel revascularization rates per 100 subjects (10.4 vs. 21.5; difference: −11.1; 95% confidence interval [CI]: −16.0 to −6.1; p < 0.001), with no difference in the rates per 100 subjects of death (5.0 vs. 5.2; difference: −0.2; 95% CI: −2.7 to 2.4; p = 0.90) or nonfatal myocardial infarction (3.2 vs. 4.4; difference: −1.2; 95% CI: −3.4 to 1.0; p = 0.29). After discounting at a 3% annual rate, there were no differences in quality-adjusted survival days (1,093 vs. 1,090; difference: 3; 95% CI: −13 to 19; p = 0.69) and total medical costs ($21,483 vs. $21,680; difference: −$198; 95% CI: −$1,608 to $1,207; p = 0.78).
Conclusions The use of Endeavor versus Driver was associated with a significant reduction in target vessel revascularization through 4-year follow-up with no difference in death, nonfatal myocardial infarction, quality-adjusted survival, or total medical costs. These results are comparable to those for other studies evaluating drug-eluting versus bare-metal stents. (Randomized Controlled Trial to Evaluate the Safety and Efficacy of the Medtronic AVE ABT-578 Eluting Driver Coronary Stent in De Novo Native Coronary Artery Lesions [ENDEAVOR II]; NCT00614848)
The introduction of bare-metal stents (BMS) in the mid-1990s successfully reduced short-term percutaneous coronary intervention (PCI) repeat revascularization rates at an acceptable cost to society (1,2). Drug-eluting stents (DES) were designed to further reduce repeat revascularization rates by delivering medications that suppress neointimal hyperplasia, the principal cause of restenosis (3). Early clinical trials of DES versus BMS reported significant reductions in restenosis rates to 1 year, and DES appeared to be a clinically and economically attractive alternative to BMS (4–7). However, an association between DES use and very late stent thrombosis (>1 year) was undetected in early assessments of short-term restenosis rates, suggesting that longer follow-up periods may be required (8,9).
The ENDEAVOR II (Randomized Controlled Trial to Evaluate the Safety and Efficacy of the Medtronic AVE ABT-578 Eluting Driver Coronary Stent in De Novo Native Coronary Artery Lesions) study was a randomized, controlled, clinical trial that compared the Endeavor zotarolimus-eluting stent (ZES) to the Driver bare cobalt-chromium metal alloy stent (BMS) (both Medtronic CardioVascular, Santa Rosa, California) (10). We evaluated long-term clinical and economic outcomes associated with the use of ZES versus BMS using 4-year follow-up information from the ENDEAVOR II clinical trial. Our objectives were to compare differences in: 1) clinical event rates and survival duration; 2) medical resource use and medical costs; and 3) disease-specific cost-effectiveness (incremental medical costs per target vessel revascularization [TVR] avoided) and comprehensive cost-effectiveness (incremental medical costs per quality-adjusted life-year [QALY] saved) for subjects receiving ZES versus BMS.
Patient population and treatment protocol
Between July 14, 2003, and January 13, 2004, the ENDEAVOR II clinical trial randomly assigned 1,197 subjects to receive ZES (n = 598) or BMS (n = 599) of whom 1,167 (ZES = 583, BMS = 584) were followed for 4 years (10–12). Subjects were included if they had clinical evidence of ischemia or an abnormal functional study and were undergoing intracoronary stenting in a single, untreated, native coronary artery. The primary study end point was target vessel failure (TVF) at 9-month follow-up—a composite of TVR, recurrent nonfatal myocardial infarction (MI) (both Q- and non–Q-wave), or cardiac death not attributable to a nontarget vessel. At the 270-day follow-up, fewer subjects receiving ZES versus BMS experienced TVF (7.9% vs. 15.1%; −7.1% difference; p < 0.001), with TVR being the primary TVF component contributing to this difference (5.6% vs. 12.5%; −6.9% difference; p < 0.001).
Long-term clinical and economic analysis overview
Although the ENDEAVOR II trial showed a clear reduction in TVF and TVR for subjects randomized to ZES versus BMS, the long-term clinical and economic implications of these results were uncertain. Our analysis included unpublished 4-year follow-up results from the ENDEAVOR II clinical trial. The analysis had 2 components: 1) an intention-to-treat analysis to evaluate clinical and economic outcomes for ZES versus BMS; and 2) a cost-effectiveness analysis to extend the intention-to-treat results by combining clinical and economic outcomes to estimate medical cost versus health benefit trade-offs (disease specific and comprehensive) for ZES versus BMS in the study population. The ENDEAVOR II protocol was approved by each site's institutional review board and the long-term clinical and economic analysis protocol was approved by Duke University Medical Center's Institutional Review Board. All study subjects consented to participate in the ENDEAVOR II clinical trial.
Research coordinators completed case report forms documenting each subject's baseline clinical characteristics, index procedure resource use, clinical events, serious adverse events (SAEs), and adverse events. Major adverse cardiac events (MACE) were adjudicated from ENDEAVOR II clinical data by an independent clinical events committee whose members were blinded to treatment assignment. We identified clinical and economic events from MACE, SAE, and adverse events data using a hierarchy-of-evidence approach. The MACE events were considered the highest level of evidence. The SAE records were used as a secondary evidence source to define other types of cardiac and noncardiac hospital stays not included in MACE records. The SAE hospitalization records were matched with adverse events records, which were the source for event start and stop dates, organ system (to differentiate cardiac from noncardiac hospital stays), and Medical Dictionary for Regulatory Activities's preferred terms (used to describe cardiac hospital stays for medical cost assignment).
We developed an automated procedure whereby MACE and hospital stays were collapsed into unique episodes of care (1 index procedure per subject and a variable number of follow-up episodes). Except for deaths outside a hospital (outpatient deaths), all episodes of care were associated with an inpatient stay. Diagnosis-related groups (DRGs) were assigned to episodes of care using the logic of the CMS MS-DRG Grouper (Centers for Medicare and Medicaid Services, Baltimore, Maryland) and audited by a medical records professional (13). Medical costs for all episodes of care were estimated using 2008 Medicare national average payment amounts (calculated using an average hospital Medicare base rate of $4,893) (14,15). Medical costs and arithmetic mean lengths of stay for cardiovascular episodes of care were estimated by DRG; whereas, estimates for noncardiovascular episodes of care were assigned using a national average relative weight. Costs for physician services were estimated using published sources to determine DRG-specific amounts and were adjusted to 2008 values using the medical care component of the consumer price index (16). As ZES and BMS stent costs are not specifically included in Medicare's payment amounts for DES DRGs, the average 2008 unit costs for the ZES ($2,100) and BMS ($900) were added to Medicare's reimbursement amounts for balloon angioplasty procedures (data on file, Medtronic CardioVascular). Because the ENDEAVOR II case report forms did not stipulate type of repeat PCI procedure, we assumed a distribution of 13% balloon angioplasty, 19% BMS, and 68% DES, based on the distribution of repeat PCI procedures at Duke University Medical Center during the early DES era (17).
We used previously defined methods to assign quality-of-life estimates based upon clinical events occurring during the study period (3). All study subjects received a 0.79 QALY adjustment for their index procedure year (4,5). A QALY of 0.85, corresponding to having coronary artery disease (CAD) was assigned to all years without a revascularization; whereas, years with a revascularization were assigned a 0.79 QALY value. Subjects experiencing a nonfatal MI received a permanent 0.88 QALY adjustment, and those with hospital stays received a QAL day decrement (0.0028 QALY) for each day's length of stay (18).
Dual antiplatelet therapy use (aspirin, clopidogrel, ticlopidine, and aspirin plus clopidogrel or ticlopidine) was evaluated at discharge; 30 days; and 6, 9, 12, 24, 36, and 48 months. Clinical outcomes included the following events: death, nonfatal MI, stroke, TVR (PCI, coronary artery bypass graft [CABG] and total), non-TVR (PCI, CABG, and total), total revascularizations (PCI, CABG, and total), hospital stays (revascularization, other cardiac, noncardiac, and total), survival days (with and without quality-of-life adjustment, and with and without discounting at a 3% annual rate), and follow-up period total length of stay (19). The composite outcomes of death or MI and death or MI or TVR also were evaluated. Medical costs were reported by time interval (first year, second year, third year, and fourth year), and cumulative (with and without discounting at a 3% annual rate).
Index procedure clinical characteristics and resource use results are presented as percentages for discrete variables and as medians, 25th and 75th percentiles for continuous variables. Differences between dichotomous variables are assessed using the chi-square statistic or Fisher exact test, and differences between continuous variables are assessed using the Kruskal-Wallis test. Follow-up outcomes, hospital stays, survival days, and medical costs are presented as cumulative values by treatment with differences, 95% confidence intervals (CI), and p values. Medical costs also are presented within time periods of interest (first year, second year, third year, fourth year, and cumulative). These analyses are performed on partitioned data using generalized linear models with adjustment for censoring and empirical standard errors (20,21). All analyses were performed with SAS software (version 8.2 or higher, SAS Institute, Cary, North Carolina).
Two types of cost-effectiveness analyses were planned. However, the presentation of these analyses as results was conditioned upon the existence of significant differences in the numerators and denominators of these ratios. If these differences did not exist, we would report their numerators and denominators as separate values. First, because the primary clinical benefit of DES versus BMS is a reduction in TVR procedures, we sought to perform a disease-specific cost-effectiveness analysis to evaluate the incremental medical costs per TVR avoided by dividing the difference in 4-year medical costs for subjects receiving the ZES versus BMS stents by the difference in TVR procedures over the same interval. A second, comprehensive cost-effectiveness analysis sought to estimate the cumulative 4-year incremental total medical costs per QALY saved with the use of ZES versus BMS stent by dividing the difference in 4-year discounted medical costs by the difference in discounted quality-adjusted survival. The variability of this estimate would be assessed using the nonparametric bootstrap procedure, with results presented as mean values for the ratio (22,23). Incremental medical costs, incremental quality-adjusted survival, and comprehensive cost-effectiveness would be calculated with and without a 3% annual discount rate.
Patients receiving ZES and BMS stents were well-matched with regard to age, sex, and race (Table 1). Most subjects underwent revascularization for angina or MI and nearly one-third had multivessel CAD. During their index procedure, most subjects received pre-treatment with balloon angioplasty and had a single stent implanted. There were no instances of emergency bypass surgery and the typical patient had 1-day post-procedural length of stay.
Dual antiplatelet therapy
There were no treatment-related differences in the use of dual antiplatelet therapy throughout the 4-year follow-up period (Table 2). At discharge, more than 95% of subjects receiving ZES and BMS had received dual antiplatelet therapy. This declined to <50% at 9 months and <10% in the third and fourth follow-up years.
Over a 4-year follow-up period, the use of ZES versus BMS led to a significant reduction in TVR procedures that was driven largely by reductions in TVR with PCI (Table 3,Fig. 1). Treatment-related differences in TVR per 100 subjects continued to increase during the second through fourth follow-up years (3.3 ZES vs. 6.4 BMS; difference: −3.1; 95% CI: −5.9 to 0.4; p = 0.027). There were no treatment-related differences in death, nonfatal MI, stroke, non-TVR, or the composite of death or MI. However there was a difference in the composite of death or MI or TVR.
4-year other hospital stays
Subjects receiving ZES versus BMS stents tended to experience fewer revascularization and total hospital stays during the 4-year follow-up period with no difference in other cardiac and noncardiac hospital stays (Table 4). These differences in hospitalization rates were associated with a nonsignificant difference in total length of stay.
4-year quality-adjusted survival and medical costs
At 4-year follow-up, subjects receiving ZES versus BMS had greater survival and quality-adjusted survival and lower total medical costs; however, none of these differences were statistically significant (Table 5). Whereas treatment-related quality-adjusted survival curves did not diverge throughout the 4-year study period, medical costs converged between the index procedure and the end of the 4-year follow-up (Figs. 2 and 3).⇓⇓
Cost-effectiveness ratios are not presented as there were no significant differences in key components of these ratios. In the disease-specific cost-effectiveness analysis, a ZES versus BMS treatment strategy reduced TVR with no difference in cumulative medical costs. In the comprehensive cost-effectiveness analysis, a ZES versus BMS treatment strategy was associated with nonsignificant differences in discounted quality-adjusted survival and total medical costs. Figure 4 is a plot of 200 bootstrap samples for incremental discounted medical cost versus incremental discounted quality-adjusted survival for a ZES versus BMS treatment strategy.
The use of ZES versus BMS was associated with a significant reduction in TVR through 4-year follow-up with no difference in death, nonfatal MI, quality-adjusted survival, or total medical costs. These results are comparable to other studies of DES versus BMS. Subjects randomized to receive ZES versus BMS had higher index procedure costs with similar lengths of stay. However, subjects receiving BMS underwent TVR procedures more frequently—ultimately eliminating total medical cost differences at 4-year follow-up. This suggests a continuing benefit from ZES versus BMS that extends well beyond the first follow-up year and encourages clinical and economic analyses of DES versus BMS with a long-term perspective.
Comparisons with other DES economic analyses
Two pivotal clinical trials—TAXUS IV (4) and SIRIUS (Sirolimus-Eluting Stent in Coronary Lesion) (5)—evaluated DES versus BMS and published cost-effectiveness analyses. At 1-year follow-up, our study and the other 2 reported higher index procedure costs and lower follow-up period costs for subjects receiving DES versus BMS with cumulative 1-year costs being higher for subjects receiving DES. In each of these studies, higher costs for DES versus BMS were the primary factor contributing to differences in index procedure costs; whereas, lower repeat revascularization rates were the primary contributor to lower follow-up period costs for study subjects receiving DES versus BMS. Differences in repeat revascularizations drove between-trial differences in disease-specific cost-effectiveness estimates; whereas, differences in total medical costs and estimated quality of life drove between-trial differences in comprehensive cost-effectiveness estimates.
During the past 6 years, costs for BMS have remained relatively constant; whereas, the costs for DES have decreased. The net effect of these changes is that the $2,000 and $1,900 cost differentials for DES versus BMS reported in the SIRIUS and TAXUS IV economic analyses were reduced to $1,200 by the time of the ENDEAVOR II economic analysis. This creates a $700 to $800 cost advantage for the ZES versus BMS comparison in the present study that needs to be considered when assessing these results versus those from previous DES versus BMS economic analyses.
Previous researchers have suggested that differences in the use of protocol-mandated coronary angiography may have contributed to between-trial differences in repeat revascularization rates (4,9). Protocol-driven angiography rates were significantly higher for SIRIUS (66% of subjects) than for TAXUS IV (43%) or ENDEAVOR II (44%) (6,10,24). Of note, the 9-month TVF rate for DES versus BMS was 7.9% versus 15.1% in ENDEAVOR II, 7.6% versus 14.4% in TAXUS IV, and 8.6% versus 21.0% in SIRIUS. Thus, it appears that between-trial differences in TVF results for DES versus BMS are driven by differences in BMS event rates.
The absence of treatment-related survival differences in the 3 trials of DES versus BMS magnified small differences in total medical costs as quality-of-life adjustments were applied in disease-specific and comprehensive cost-effectiveness ratios. One-year total medical cost increments for DES versus BMS were $309 in SIRIUS, $572 in TAXUS IV, and $781 in ENDEAVOR II. By 4-year follow-up, the choice of ZES versus BMS was associated with a $198 cost savings in ENDEAVOR II. Although it would be difficult to demonstrate a statistically significant difference between these values, comprehensive cost-effectiveness ratios magnified them to obtain values of $27,540 per QALY saved in SIRIUS and $47,798 in TAXUS IV. The 4-year comprehensive cost-effectiveness ratio for ENDEAVOR II was dominant (lower total medical costs and better quality-adjusted survival). However, all of these ratios rely upon point estimates and do not account for the substantial variability inherent in medical cost and QALY estimates.
Regulator economic evaluation
A recent evaluation by the United Kingdom's National Institute for Health and Clinical Excellence concluded that DES are recommended for the treatment of CAD only if, “the target artery to be treated has less than a 3-mm caliber or the lesion is longer than 15 mm, and the price difference between DES and BMS is no more than 300 pounds.” The 300-pound-price-differential recommendation was based upon the results of an economic model that assumed the greatest benefit from TVR reduction occurred in the first follow-up year and by design did not include follow-up after this point (25). Our analysis beyond the 1-year follow-up indicated that a significant difference in TVR rates occurs during the 2- to 4-year follow-up and demonstrated that the use of DES versus BMS is cost neutral with a $1,200 price differential.
Long-term economic attractiveness
The present study extends the follow-up period for DES versus BMS economic analysis from 1 year to 4 years. Other studies examined a 1-year time frame because most target lesion-related clinical events occur during the first year in BMS clinical trials; clinical events after the first year are more likely related to progression of CAD and are not target lesion-related (26). However, our results demonstrate that the difference in TVR rates between DES and BMS in ENDEAVOR II continued to evolve beyond the first year. This is particularly important as very late stent thrombosis has attracted increased scrutiny (3,27). In this regard, clinical and economic comparisons between DES and BMS may resemble those between CABG and PCI in that their respective cumulative cost and benefit curves may converge as repeat revascularizations occur more frequently in the less expensive treatment strategy (28). It is therefore important to allow sufficient time for longer-term events to occur when evaluating technologies with different biological effects.
This study has several limitations that are largely related to study data collection methods. First, as economic and quality-of-life data were not collected prospectively in ENDEAVOR II, we needed to rely upon secondary sources for our medical cost and quality-of-life weights. This limitation both muted the precision of our analyses and prevented certain cost elements (outpatient and medication costs) from being included in our work. Nonetheless, we believe that our results accurately reflect important cost elements and cost drivers (stent costs and repeat revascularization rates). The omission of dual antiplatelet therapy costs will have a minimal effect upon this study's results as subjects in both study arms were prescribed aspirin indefinitely and clopidogrel for 12 weeks. However, guidelines for the use of these therapies have changed since the time of this trial and may influence both the medical costs and health benefits associated with the use of ZES versus BMS. Second, our study ended at 4 years and did not include lifetime estimates of medical costs and quality-adjusted life expectancy. However, survival differences at 4 years were negligible, and we do not believe that a lifetime analysis would have added substantial information to our results. Lastly, our study is limited by the population of subjects enrolled in the ENDEAVOR II clinical trial. Although we believe that our results accurately reflect long-term clinical and economic outcomes within this population, they may not be transferable to patients in other populations.
Percutaneous coronary intervention with ZES versus BMS is associated with significant reductions in TVR through 4-year follow-up. The resulting medical cost savings are sufficient to fully cover the additional costs of DES. These results are comparable to those for other studies evaluating DES versus BMS.
The authors would like to thank Allyn Meredith, MA, for her expert editorial assistance.
The ENDEAVOR II clinical trial and the 4-year clinical and economic analysis were funded by Medtronic CardioVascular, Santa Rosa, California. Eric L. Eisenstein has received research or grant support from Medtronic, Inc. and Eli Lilly and Company. Dr. Wijns incurs grants, research contracts, fees, and honoraria on behalf of the Cardiovascular Research Center, Aalst, Belgium. Dr. Mauri has served as a consultant to Abbott Vascular, Boston Scientific, Cordis, and Medtronic, Inc. Rex Edwards has received research support from Medtronic, Inc. Dr. Cowper has received research support from Medtronic, Inc. Dr. Kong has received research support from Medtronic, Inc. Dr. Anstrom has received research support from AstraZeneca, Bristol-Myers Squibb, Eli Lilly and Co., and Medtronic, Inc., and has served as a consultant for Johnson & Johnson and Pfizer.
- Abbreviations and Acronyms
- bare-metal stent(s)
- coronary artery bypass graft
- coronary artery disease
- drug-eluting stent(s)
- diagnosis-related group
- major adverse cardiac events
- myocardial Infarction
- percutaneous coronary intervention
- quality-adjusted life-year
- serious adverse events
- target vessel failure
- target vessel revascularization
- zotarolimus-eluting stent(s)
- Received July 31, 2009.
- Revision received October 13, 2009.
- Accepted October 15, 2009.
- American College of Cardiology Foundation
- Williams D.O.,
- Holubkov R.,
- Yeh W.,
- et al.
- Peterson E.D.,
- Cowper P.A.,
- DeLong E.R.,
- Zidar J.P.,
- Stack R.S.,
- Mark D.B.
- Garg P.,
- Cohen D.J.,
- Gaziano T.,
- Mauri L.
- Bakhai A.,
- Stone G.W.,
- Mahoney E.,
- et al.,
- TAXUS IV Investigators
- Cohen D.J.,
- Bakhai A.,
- Shi C.,
- et al.
- Fajadet J.,
- Wijns W.,
- Laarman G.J.,
- et al.
- Medtronic, Inc. FDA executive summary memorandum: Endeavor zotarolimus drug-eluting coronary stent system & Over-the-Wire (OTW), Rapid Exchange (RX), and Multi-Exchange II (MX2) stent delivery systems. P060033. Prepared for the meeting of the Circulatory System Devices Advisory Panel; Gaithersburg, MD; October 10, 2007.
- ENDEAVOR II Investigators. The ENDEAVOR II study: clinical study report (index procedure through 1,080 days). Appendix 2: Endeavor II 36-month report to the FDA. October 29, 2008; Boston, MA. 87–463.
- ↵(2008) The Trees, Version 25.0: CMS MS-DRG Grouper (Ingenix, Eden Prairie, MN).
- ↵(2008) DRG Expert: A Comprehensive Guidebook to the DRG Classification System (Ingenix, Salt Lake City, UT), 24th edition.
- (2007) DRG Plus! 2008: A Comprehensive Guide to Medicare Severity Diagnosis Related Groups (Practice Management Information Corporation, Los Angeles, CA).
- ↵Mitchel J, Burge R, Lee A, et al. Per case prospective payment for episodes of hospital care. PB95226023. Report presented to: U.S. Department of Commerce National Technical Information Service, 1995. Springfield, VA.
- Tsevat J.,
- Goldman L.,
- Soukup J.R.,
- et al.
- Gold M.R.,
- Siegel J.,
- Russell L.,
- Weinstein M.
- Bang H.,
- Tsiatis A.A.
- Thompson S.G.,
- Barber J.A.
- Stone G.W.,
- Ellis S.G.,
- Cox D.A.,
- et al.
- National Institute for Health and Clinical Excellence
- Cutlip D.E.,
- Chhabra A.G.,
- Baim D.S.,
- et al.
- Cutlip D.E.,
- Windecker S.,
- Mehran R.,
- et al.