Journal of the American College of Cardiology Vol. 44, No. 9, by the American College of Cardiology Foundation ISSN /04/$30.

Transcription

1 Journal of the American College of Cardiology Vol. 44, No. 9, by the American College of Cardiology Foundation ISSN /04/$30.00 Published by Elsevier Inc. doi: /j.jacc Cost-Effectiveness and Cardiac Interventions Economic Evaluation of With Provisional Glycoprotein IIb/IIIa Inhibition Versus Heparin With Routine Glycoprotein IIb/IIIa Inhibition for Percutaneous Coronary Intervention Results From the REPLACE-2 Trial David J. Cohen, MD, MSC,* A. Michael Lincoff, MD, Tara A. Lavelle, BS, Huei-Ling Chen, PHD, Ameet Bakhai, MD,* Ronna H. Berezin, MPH, Daniel Jackman, MD, Ian J. Sarembock, MB, CHB, Eric J. Topol, MD, on behalf of the REPLACE-2 Investigators Boston and Brookline, Massachusetts; Tyler, Texas; Charlottesville, Virginia; and Cleveland, Ohio OBJECTIVES BACKGROUND METHODS RESULTS CONCLUSIONS The purpose of this study was to compare the cost of percutaneous coronary intervention (PCI) using bivalirudin with provisional platelet glycoprotein (GP) IIb/IIIa inhibition with that of heparin routine GP IIb/IIIa inhibition. Although GP IIb/IIIa inhibition has been shown to reduce ischemic complications in a broad range of patients undergoing PCI, many patients currently do not receive such therapy because of concerns about bleeding complications or cost. Recently, bivalirudin with provisional GP IIb/IIIa inhibition has been validated as an alternative to heparin routine GP IIb/IIIa inhibition for patients undergoing PCI. However, the cost-effectiveness of this novel strategy is unknown. In the Randomized Evaluation in PCI Linking Angiomax to Reduced Clinical Events (REPLACE)-2 trial, 4,651 U.S. patients undergoing non-emergent PCI were randomized to receive bivalirudin with provisional GP IIb/IIIa (n 2,319) versus heparin routine GP IIb/IIIa (n 2,332). Resource utilization data were collected prospectively through 30-day follow-up on all U.S. patients. Medical care costs were estimated using standard methods including bottom-up accounting (for procedural costs), the Medicare fee schedule (for physician services), hospital billing data (for 2,821 of 4,862 admissions), and regression-based approaches for the remaining hospitalizations. Among the bivalirudin group, 7.7% required provisional GP IIb/IIIa. Thirty-day ischemic outcomes including death or myocardial infarction were similar for the bivalirudin and GP IIb/IIIa groups, but bivalirudin resulted in lower rates of major bleeding (2.8% vs. 4.5%, p 0.002) and minor bleeding (15.1% vs. 28.1%, p 0.001). Compared with routine GP IIb/IIIa, in-hospital and 30-day costs were reduced by $405 (95% confidence interval [CI] $37 to $773) and $374 (95% CI $61 to $688) per patient with bivalirudin (p for both). Regression modeling demonstrated that, in addition to the costs of the anticoagulants themselves, hospital savings were due primarily to reductions in major bleeding (cost savings $107/patient), minor bleeding ($52/patient), and thrombocytopenia ($47/patient). Compared with heparin routine GP IIb/IIIa inhibition, bivalirudin provisional GP IIb/IIIa inhibition resulted in similar acute ischemic events and cost savings of $375 to $400/patient depending on the analytic perspective. (J Am Coll Cardiol 2004;44: ) 2004 by the American College of Cardiology Foundation Over the past decade, numerous clinical trials have demonstrated that the combination of an intravenous platelet glycoprotein (GP) IIb/IIIa inhibitor and low-dose unfractionated heparin leads to substantial reductions in periprocedural ischemic complications compared with unfractionated heparin alone in patients undergoing percutaneous From the *Cardiovascular Division, Beth Israel Deaconess Medical Center, Boston, Massachusetts; Harvard Clinical Research Institute, Brookline, Massachusetts; Mother Frances Hospital, Tyler, Texas; Cardiovascular Division and Cardiovascular Research Center, University of Virginia Health System, Charlottesville, Virginia; and the Cleveland Clinic Foundation, Cleveland, Ohio. Supported by a grant from The Medicines Company. Manuscript received February 3, 2004; revised manuscript received March 31, 2004, accepted May 11, coronary intervention (PCI) (1 3). Although the data are somewhat less definitive, several recent meta-analyses suggest that such therapy improves long-term survival as well See page 1809 (4,5). Nonetheless, many PCI patients both in the U.S. and worldwide do not receive a GP IIb/IIIa inhibitor, in part, owing to concerns about bleeding complications and cost. Recently, the direct thrombin inhibitor, bivalirudin, has been shown to result in rates of ischemic complications similar to those with heparin GP IIb/IIIa inhibition in a multicenter clinical trial of more than 6,000 PCI patients

2 JACC Vol. 44, No. 9, 2004 November 2, 2004: Cohen et al. Cost-Effectiveness of for PCI 1793 Abbreviations and Acronyms CABG coronary artery bypass graft surgery CI confidence interval CK-MB creatine kinase-mb fraction GP glycoprotein MI myocardial infarction PCI percutaneous coronary intervention REPLACE Randomized Evaluation in PCI Linking Angiomax to Reduced Clinical Events (6). In this trial, bivalirudin was associated with a nonsignificant increase in ischemic events (primarily non Qwave myocardial infarctions [MIs]) and a significant reduction in protocol-defined major and minor bleeding. The net impact of these two effects on overall medical care costs is unknown. Given the large number of angioplasty procedures performed in the U.S. (currently estimated at 1 million/year) (7), even modest cost savings on a per patient basis have the potential to result in substantial savings to the health care system. Therefore, we performed a prospective economic evaluation in conjunction with the Randomized Evaluation in PCI Linking Angiomax to Reduced Clinical Events (REPLACE)-2 trial. The objectives of the study were: 1) to compare the in-hospital and 30-day costs of PCI using bivalirudin provisional GP IIb/IIIa inhibition with those of PCI using the standard anticoagulation regimen of heparin routine GP IIb/IIIa inhibition, and 2) to explore the impact of both ischemic and bleeding complications on the cost of PCI in contemporary practice. METHODS Patient population and treatment protocol. Between October 2001 and August 2002, 6,010 patients undergoing non-emergent PCI were enrolled in the REPLACE-2 trial. As specified in the study protocol, the economic analysis was confined to those patients enrolled at U.S. treatment sites (n 4,651). Details of the study design have been described previously (6). Patients were eligible if they were undergoing PCI with an approved device. Key exclusion criteria included: ongoing acute MI; recent receipt of one of the study drugs or low-molecular-weight heparin; or need for concomitant warfarin therapy. Patients were also excluded if they had a platelet count 100,000, serum creatinine 4.0 mg/dl, or were at increased risk of bleeding complications. The study protocol was approved by the institutional review board at each site, and each patient provided informed consent before enrollment. Patients were randomized in a double-blind, tripledummy fashion to treatment with bivalirudin provisional GP IIb/IIIa inhibition or heparin routine GP IIb/IIIa inhibition. Randomization was stratified by study site and by the operator s intent to use either abciximab or eptifibatide as the GP IIb/IIIa inhibitor. was administered as a 0.75 mg/kg bolus followed by a 1.75 mg/kg/h infusion for the duration of the PCI procedure. Patients could receive up to 4 h of prolonged bivalirudin infusion at the completion of the PCI procedure at the operator s discretion. Heparin was administered as a 65 U/kg bolus. Additional boluses of bivalirudin or heparin were given as needed to maintain an activated clotting time 225 s. Abciximab and eptifibatide were administered using the Food and Drug Administration-approved doses for each with infusion durations of 12 and 18 h, respectively (2,3). Provisional GP IIb/IIIa inhibition (or matching placebo for patients already receiving such therapy) was given at the operator s discretion to treat potential thrombotic complications. All patients received aspirin before PCI, and pretreatment with clopidogrel (300 mg) was strongly encouraged. After PCI, all patients received aspirin clopidogrel (75 mg/day) for at least 30 days. Assessment of in-hospital outcomes and clinical follow-up. Baseline characteristics, procedural details, and clinical outcomes during the initial hospitalization and 30-day follow-up period were recorded on standardized case report forms. Myocardial infarction was defined as either development of new, pathologic Q waves; elevation of creatine kinase, MB fraction (CK-MB) to 3 the upper limit of normal within 48 h of PCI; or elevation of CK-MB to 2 the upper limit of normal at any other time during followup. Major bleeding was defined as any intracranial or retroperitoneal hemorrhage, clinically overt blood loss resulting in a 3 g/dl decrease in hemoglobin, any decrease in hemoglobin 4 g/dl, or transfusion of 2 U of packed red cells (6). Minor bleeding was defined as any other clinically overt blood loss that did not meet criteria for major bleeding. Determination of medical care costs. Medical care costs for the initial hospitalization and for the 30-day follow-up period were assessed from a societal perspective using a combination of bottom-up and top-down methods as previously described (8). CARDIAC CATHETERIZATION LABORATORY COSTS. Detailed resource utilization was recorded for each procedure, and the cost of each item was estimated on the basis of the mean hospital acquisition cost for the item in Costs for study drugs (bivalirudin, abciximab, eptifibatide) were based on the measured bolus and infusion volumes for each and current average wholesale prices, assuming that partially unused vials would be discarded. Costs of additional disposable equipment, overhead, and depreciation for the cardiac catheterization laboratory were estimated on the basis of the average cost per procedure at Beth Israel Deaconess Medical Center in 2002 and adjusted for actual procedure duration. OTHER HOSPITAL COSTS. All other hospital costs were determined using top-down accounting methods based on each hospital s Medicare cost report (8). For 2,500 randomly selected patients, itemized bills were obtained for the initial hospitalization and any subsequent cardiovascular

3 1794 Cohen et al. JACC Vol. 44, No. 9, 2004 Cost-Effectiveness of for PCI November 2, 2004: hospitalizations during the follow-up period. In addition, we obtained billing data on all patients who experienced a major in-hospital complication. Overall, billing data were collected for 2,821 of 4,862 hospital admissions during the 30-day study period. Hospital costs were determined by multiplying itemized hospital charges by the cost-center specific cost-to-charge ratio obtained from the hospital s Medicare cost report as previously described (9,10). For those admissions for which billing data were not collected (n 2,041), non-procedural hospital costs were imputed on the basis of a linear regression model developed using the hospital admissions for which complete billing information was available. Independent variables for this model included length of stay, intensive care unit length of stay, ischemic complications, bleeding complications, and revascularization procedures (model R ). OTHER COSTS. Physician s fees for inpatient services, major cardiac procedures, and surgical procedures were based on the 2002 Medicare Fee Schedule. Costs for outpatient medical services and medications were not assessed. Statistical analysis. The prespecified primary end point of the economic study was total 30-day costs. In addition, a secondary end point of index hospital costs was examined to consider the perspective of a typical hospital that is reimbursed for each episode of care. To reduce the impact of high cost outliers on group means, the protocol specified that 30-day costs greater than the 99th percentile for each treatment group were assigned costs equal to the 99th percentile for the group. Discrete data are reported as frequencies; continuous data are reported as mean SD. Cost data are reported as both mean and median values. Discrete variables were compared by Fisher exact test. Normally distributed continuous variables were compared by Student t test. Cost and length of stay data were compared by the Wilcoxon rank-sum test. All statistical analyses were performed according to the intention-to-treat principle. Multiple linear regression was performed to identify independent predictors of initial hospital costs. Candidate variables for this analysis included patient characteristics, ischemic complications, repeat procedures, and bleeding complications. Because the goal of this analysis was explanatory rather than predictive, length of stay variables were not considered. Untransformed cost was used as the dependent variable for these analyses for ease of interpretation and because model fit did not improve appreciably using logtransformed costs. Attributable costs were calculated by multiplying the independent cost of each event (derived from the regression model) by its frequency in the treatment group. The absolute cost savings associated with prevention of specific clinical events were estimated by multiplying the independent cost of each event by the difference in event frequency between the bivalirudin and heparin GP IIb/IIIa groups. Table 1. Baseline Clinical and Treatment Characteristics RESULTS (n 2,319) Heparin GP IIb/IIIa (n 2,332) Age (yrs) Gender (% male) Diabetes mellitus (%) Smoking in last 12 months (%) Previous myocardial infarction (%) Previous CABG (%) Previous PCI (%) Multivessel PCI (%) 17.3* 14.6 SVG PCI (%) Prespecified GP IIb/IIIa inhibitor Eptifibatide (%) Abciximab (%) Device selection Stent (%) Atherectomy (%) *p 0.05 versus heparin GP IIb/IIIa; all other comparisons p NS. CABG coronary artery bypass graft surgery; GP glycoprotein; PCI percutaneous coronary intervention; SVG saphenous vein graft. Patient population. Baseline characteristics of the two treatment groups were generally well-matched (Table 1). The mean age was years. Approximately 30% were diabetic, 20% had undergone previous coronary artery bypass graft surgery (CABG), and more than one-third had undergone previous PCI. There was a modest excess of patients who underwent multivessel PCI in the bivalirudin group (17.3% vs. 14.6%, p 0.01). Approximately 60% of patients in both groups were preselected for eptifibatide as the GP IIb/IIIa inhibitor, and 87% of patients underwent stent placement. Procedural resource utilization and cost. Table 2 summarizes resource utilization and cost for the initial PCI procedures. The major difference in resource utilization between the two groups was related to procedural anticoagulants. In the bivalirudin group, 26.9% of patients required 1 vial of bivalirudin, and mean use was 1.35 vials. In addition, 7.7% of patients assigned to bivalirudin received a GP IIb/IIIa inhibitor on a provisional basis. Among the heparin GP IIb/IIIa group, patients selected for abciximab required an average of 3.37 vials/patient, whereas those patients selected for eptifibatide required an average of 4.73 vials/patient. Overall, anticoagulant costs were reduced by approximately $400/patient for the bivalirudin group compared with the heparin GP IIb/IIIa group (95% confidence interval [CI] $373 to $431; p 0.001). This difference was substantially greater for those patients selected for abciximab (mean difference $848, 95% CI $803 to $894) versus those selected for eptifibatide (mean difference $106, 95% CI $85 to $128). There was a trend toward increased stent use in the bivalirudin group compared with the heparin GP IIb/IIIa group (1.36 vs. 1.31, p 0.13). This trend was explained by the imbalance in multivessel PCI procedures between the

4 JACC Vol. 44, No. 9, 2004 November 2, 2004: Cohen et al. Cost-Effectiveness of for PCI 1795 Table 2. Procedural Resource Utilization and Cost Heparin GP IIb/IIIa p Value Procedure duration (min) Contrast volume (ml) Balloons (n) Stents (n) Guide wires (n) Guiding catheters (n) use Number of vials vial used (%) 26.9% 0% GP IIb/IIIa inhibitor use Provisional GP IIb/IIIa (%) 7.7% 0% Abciximab vials* Eptifibatide vials Costs All anticoagulants $ $ $ $ Abciximab* $130 $432 $1, Eptifibatide $42 $156 $ Devices $2,075 1,399 $2,024 1, Supplies $ $ Room/overhead $1, $1, Non-physician personnel $ $ Total index procedure cost $4,606 1,916 [$4,141] $4,941 1,793 [$4,603] *Among patients preselected for abciximab (n 1,949). Among patients preselected for eptifibatide (n 2,792). Values in brackets are medians. GP glycoprotein. two groups, however, as there were no significant differences in mean stent use among patients undergoing single-vessel PCI (1.19 vs. 1.18, p 0.98) or multivessel PCI (2.23 vs. 2.21, p 0.55). There were no significant differences in any other resource measures between the two groups. Total costs for the index revascularization procedure were reduced by $335/patient in the bivalirudin group versus the GP IIb/IIIa group (95% CI $229 to $441, p 0.001). Initial hospital outcomes, resource utilization, and costs. Table 3 summarizes initial hospital outcomes for the two treatment groups. Consistent with the results of the overall trial, there was no significant difference in the combined Table 3. Initial Hospital Outcomes, Resource Utilization, and Cost Heparin GP IIb/IIIa p Value Death (%) Non-fatal MI (%) Any CK-MB CK-MB CK-MB Repeat revascularization (%) Any PCI CABG Bleeding complication (%) Major bleed Minor bleed Length of stay (days) [1] [1] 0.54 ICU length of stay (days) [0] [0] 0.45 Medical costs Initial procedure $4,606 1,916 [$4,141] $4,941 1,793 [$4,603] Repeat procedures $ [$0] $ [$0] 0.83 Hospital room/ancillary $3,655 5,295 [$2,263] $3,725 5,586 [$2,374] 0.39 Doctor fees $2,220 $810 [$2,042] $2, [$2,042] 0.41 Total $10,561 6,267 [$9,136]* $10,966 6,524 [$9,616]* Values in brackets are medians. CABG coronary artery bypass graft surgery; CK-MB creatine kinase-mb fraction; GP glycoprotein; ICU intensive care unit; MI myocardial infarction; PCI percutaneous coronary intervention.

5 1796 Cohen et al. JACC Vol. 44, No. 9, 2004 Cost-Effectiveness of for PCI November 2, 2004: Table 4. Follow-up Events, Resource Utilization, and Cost Heparin GP IIb/IIIa p Value Events between discharge and 30 days Death (%) Myocardial infarction (%) Repeat hospitalization (%) CABG (%) PCI (%) Diagnostic catheterization (%) Follow-up costs $488 2,829 [$0] $441 2,586 [$0] 0.62 Total costs (study entry to 30 days) $10,868 5,479 [$9,321]* $11,242 5,420 [$9,774]* Values in brackets are medians. *Total costs do not equal the sum of initial hospital costs and follow-up costs because 30-day costs were trimmed at the 99th percentile for each group. CABG coronary artery bypass surgery; GP glycoprotein; PCI percutaneous coronary intervention. incidence of death or MI between the bivalirudin and heparin GP IIb/IIIa groups (7.3% vs. 6.6%). However, bivalirudin provisional GP IIb/IIIa inhibition was associated with a significant and consistent reduction in bleeding complications compared with heparin routine GP IIb/IIIa inhibition. Protocol-defined major bleeding was reduced by 38% (2.8% vs. 4.5%, p 0.002), minor bleeding was reduced by 46% (15.1% vs. 28.1%, p 0.001), and thrombocytopenia (defined as any platelet count 100,000/ l) was reduced by 54% (0.7% vs. 1.5%, p 0.01). There was a modest trend toward lower hospital room and ancillary costs in the bivalirudin group compared with the heparin GP IIb/IIIa group ($3,655 vs. $3,725, p 0.38). There were no important differences in repeat procedure costs or physician fees between the two treatment groups, however. When combined with the lower costs for the initial revascularization procedures, initial hospital costs were reduced by an average of $405 per patient with bivalirudin compared with standard anticoagulation (95% CI $37 to $773, p 0.001). Follow-up resource utilization and costs. Between hospital discharge and 30-day follow-up, there were no significant differences in clinical outcomes or resource utilization between the two treatment groups (Table 4). As a result, follow-up medical care costs were generally similar for the two groups, and aggregate 30-day costs remained $374/ patient lower for the bivalirudin group compared with the heparin GP IIb/IIIa group (95% CI $61 to $688, p 0.001). Subgroup analyses. Prespecified subgroup analyses according to gender, age, and the presence of an acute coronary syndrome (defined as unstable angina within 48 h or MI within 7 days of PCI) failed to reveal any significant interactions between baseline patient characteristics and the Figure 1. Stratified analyses of aggregate 30-day costs by treatment group according to prespecified patient characteristics. The graph indicates the mean difference in costs between the bivalirudin provisional glycoprotein (GP) IIb/IIIa and heparin routine GP IIb/IIIa groups (black squares) along with the associated 95% confidence interval (bars). There was no evidence of heterogeneity of treatment effect across any of the subgroups (p value for interaction 0.05).

6 JACC Vol. 44, No. 9, 2004 November 2, 2004: Cohen et al. Cost-Effectiveness of for PCI 1797 Table 5. Multivariable Linear Regression Model of Initial Hospital Costs Factor Estimated Cost* Incidence in Heparin GP IIb/IIIa Incidence in Attributable cost in Heparin GP IIb/IIIa Net Cost Difference Associated With Events In-hospital CABG $29, % 0.86% $165 $87 Repeat PCI $8, % 0.82% $63 $4 Major bleed $6, % 2.76% $281 ($107) Thrombocytopenia $5, % 0.69% $88 ($47) MI (CK-MB 10 ) $4, % 1.64% $63 $4 Diagnostic catheterization $2, % 2.98% $64 $8 MI (CK-MB 5 10 ) $2, % 2.72% $39 $21 MI (CK-MB 3 5 ) $1, % 2.67% $33 ($2) Minor bleed $ % 15.05% $111 ($52) Patient characteristics Multivessel PCI $1, % 17.30% $206 $46 ACS $1, % 22.70% $354 ($6) History of CHF $1, % 7.30% $77 $8 Total for events $907 ($84) *Estimated cost of each complication derived from linear regression model of initial hospital costs (model R ). All coefficients were significant at the p 0.05 level. Difference in incidence between groups statistically significant (p 0.05). ACS acute coronary syndrome; CABG coronary artery bypass graft surgery; CHF congestive heart failure; CK-MB creatine kinase-mb fraction; GP glycoprotein; MI myocardial infarction. magnitude of cost savings seen with bivalirudin (Fig. 1). The extent of cost savings among patients selected for eptifibatide as the comparison GP IIb/IIIa inhibitor was somewhat less than among those patients selected for abciximab ($185 vs. $559/patient); however, there was no evidence of statistical heterogeneity of the treatment effect on cost between subgroups (p value for interaction 0.37). Among patients who underwent single-vessel PCI (n 3,906), bivalirudin reduced 30-day costs by $401/patient compared with $506/patient for those who underwent multivessel PCI (n 739). Determinants of hospital cost. Independent predictors of initial hospital cost (exclusive of study drug costs) are displayed in Table 5. Repeat revascularization procedures (in-hospital CABG, repeat PCI) were the strongest correlates of in-hospital cost. Among procedural complications, major bleeding, thrombocytopenia, and large periprocedural MI (CK-MB 10 upper limit of normal) had the greatest independent impact on cost, whereas small-to-moderate post-procedure MI and minor bleeding had lesser impacts. Several baseline patient characteristics, including the need for multivessel PCI, acute coronary syndrome presentation, and a history of congestive heart failure, were also associated with higher initial hospital costs. Attributable cost calculations indicated that adverse outcomes accounted for $907 of the initial hospital cost for the heparin GP IIb/IIIa group (9% of total hospital cost) (Table 5). The complications with the largest individual contributions to hospital cost included major bleeding, in-hospital CABG, minor bleeding, and thrombocytopenia. Comparison of event-related costs between the bivalirudin and heparin GP IIb/IIIa groups indicated that reductions in major bleeding accounted for $107/patient in hospital cost savings with bivalirudin, whereas reductions in thrombocytopenia and minor bleeding accounted for savings of $47 and $52 per patient, respectively. These savings were partially counterbalanced by increased costs of $87/patient due to bypass surgery and $21/patient due to the higher incidence of moderate-sized MIs in the bivalirudin group. Finally, $46/patient of excess cost for the bivalirudin group was attributable to the baseline imbalance in multivessel PCI procedures between the two groups. Overall, differences in hospital outcomes accounted for $84/patient in cost savings between the two groups (21% of the observed difference). When the analysis was restricted to those outcomes that differed significantly between the two groups (major bleeding, thrombocytopenia, moderate-sized MI, and minor bleeding), event-related cost savings with bivalirudin increased to $185/patient (in addition to differences in study drug costs). The impact of reduced complications on in-hospital costs was particularly important among the subgroup of patients for whom eptifibatide was selected as the GP IIb/IIIa inhibitor. Among this large subset (n 2,792), differences in in-hospital complications actually accounted for cost savings of $132/patient 71% of the observed cost savings in the subgroup. On the other hand, among those patients preselected for abciximab, differences in in-hospital complications only accounted for $26/patient in cost savings 5% of the observed reduction in 30-day costs for this subgroup. DISCUSSION In the REPLACE-2 trial, a strategy of bivalirudin with provisional GP IIb/IIIa inhibition led to similar in-hospital and 30-day outcomes among patients undergoing contemporary PCI procedures compared with the standard anticoagulation regimen of heparin with routine GP IIb/IIIa inhibition (6). Because these outcomes met prespecified

7 1798 Cohen et al. JACC Vol. 44, No. 9, 2004 Cost-Effectiveness of for PCI November 2, 2004: statistical criteria for non-inferiority, a prospective economic substudy was performed to determine which treatment strategy might be preferred on economic grounds. In our economic substudy of more than 4,600 U.S. patients, clinical outcomes were similar to those observed in the overall trial, whereas economic outcomes favored the bivalirudin provisional GP IIb/IIIa group. The absolute magnitude of cost savings ranged from $405/patient during the initial hospitalization period to $374/patient over the entire 30-day follow-up period and was consistent across a variety of prespecified patient subsets. The reduction in initial hospital costs with bivalirudin was driven by two main factors: the anticoagulants themselves and periprocedural complications. Approximately 80% of the reduction in in-hospital costs occurred during the index revascularization procedure owing to the lower acquisition cost of bivalirudin compared with parenteral GP IIb/IIIa inhibitors. Importantly, these savings were observed despite the fact that 7% of patients assigned to bivalirudin received a provisional GP IIb/IIIa inhibitor at the time of PCI. The remaining 20% of hospital cost savings associated with bivalirudin were related to differences in rates of ischemic and hemorrhagic complications. Although bivalirudin was associated with marginally higher rates of in-hospital bypass surgery and non-fatal MI, the excess costs of these events were more than offset by savings associated with lower rates of major bleeding, minor bleeding, and thrombocytopenia. The impact of reduced complications on in-hospital costs was particularly important among the subgroup of patients for whom eptifibatide was selected as the comparison GP IIb/IIIa inhibitor. For this subgroup, differences in in-hospital complications actually accounted for more than 70% of the observed cost savings with bivalirudin. In addition to quantifying the extent of cost savings associated with the use of bivalirudin, this study adds importantly to our understanding of those factors that determine the cost of contemporary PCI. On a per event basis, the most costly complications were the need for unplanned bypass surgery or repeat PCI before hospital discharge. On the other hand, when the frequency of complications was also considered, the most costly complications on a per patient basis were major bleeding (attributable cost $342/patient in the heparin GP IIb/IIIa group), minor bleeding ($151/patient), and in-hospital bypass surgery ($150/patient). Although the definition of major bleeding used in the REPLACE-2 trial was somewhat more liberal than that used in previous trials, the substantial cost of major bleeding on both a per event and per patient basis confirms that this definition has both clinical and economic relevance in contemporary practice. Previous studies have demonstrated the important impact of both ischemic and bleeding complications on hospital cost for PCI patients. In a single-center study of more than 1,200 patients, Ellis et al. (11) found that the strongest independent predictors of hospital cost were urgent bypass surgery, blood product transfusion, and acute renal failure. In the economic analysis of the Evaluation of c7e3 for the Prevention of Ischemic Complications (EPIC) trial, the incremental costs of repeat revascularization ranged from $4,973/event for non-urgent repeat PCI to $27,349/event for urgent CABG (12), whereas bleeding costs ranged from $1,327/event for minor bleeding to $5,896/event for major bleeding (using somewhat different definitions than in the current trial). Similar findings were reported in the Evaluation of PTCA to Improve Long-term Outcome by c7e3 GP IIb/IIIa receptor (abciximab) blockade (EPILOG) and Randomized Efficacy Study of Tirofiban for Outcome and Restenosis (RESTORE) trials as well (13,14). Although our findings are qualitatively similar to many previous studies, the current study is the first to directly examine hospital costs and their determinants in a large population of patients undergoing contemporary PCI with stent implantation. Compared with previous studies from the balloon angioplasty era, the contribution of ischemic complications to in-hospital costs of PCI has fallen dramatically. For example, in the EPIC trial, complications increased initial hospital costs by $2,000/patient in the control group, of which ischemic complications accounted for 80% (12). In contrast, in the REPLACE-2 trial, the aggregate cost of complications was $1,000/patient, of which ischemic events accounted for less than 40%. These findings reflect a shift in the predominant mode of ischemic complications from abrupt closure requiring repeat revascularization to non-fatal MI many of which are clinically silent. As a result, more than 60% of complication-related costs are now associated with hemorrhagic events and related outcomes (e.g., thrombocytopenia). Also noteworthy is the substantial cost associated with even minor bleeding complications. Although these events are associated with only a modest increase in costs on a per event basis, the fact that more than 25% of all patients treated with the heparin GP IIb/IIIa inhibition experienced a minor bleed led to substantial excess costs in the overall population. Practice implications. Our study has important implications for the selection of a periprocedural anticoagulation regimen in patients undergoing non-emergent PCI. Previous economic evaluations have demonstrated that heparin routine GP IIb/IIIa inhibition is reasonably cost-effective compared with heparin alone for patients undergoing PCI with stent placement, whether the GP IIb/IIIa inhibitor is abciximab or eptifibatide (15,16). Given the lower cost and similar clinical outcomes obtained with bivalirudin in the current study, there seems little question that bivalirudin would be reasonably cost-effective compared with heparin as well. Given its lower cost and similar outcomes compared with heparin routine GP IIb/IIIa inhibition, it is reasonable to question whether bivalirudin provisional GP IIb/IIIa inhibition should be considered an economically dominant anticoagulation regimen for contemporary PCI. However,

8 JACC Vol. 44, No. 9, 2004 November 2, 2004: Cohen et al. Cost-Effectiveness of for PCI 1799 in light of the design of the current study, some caution is warranted in interpreting these results. Although formal statistical analysis demonstrated non-inferiority based on the primary results of the REPLACE-2 trial, these calculations were based on a combination of regulatory considerations and clinical insight. Indeed, modest increases in overall costs (such as those observed with routine GP IIb/IIIa inhibition in this trial) might still be acceptable if long-term survival were, in fact, improved. At present, however, there is little evidence to support this assumption. In fact, both 30-day (0.2% vs. 0.4%, p 0.26) and one-year (1.9% vs. 2.6%, p 0.15) mortality rates from the REPLACE-2 trial tended to favor the bivalirudin group (17). Moreover, recent studies have suggested that for patients at very low risk of ischemic complications, a combination of high-dose heparin 600 mg of clopidogrel may yield similar rates of ischemic complications as low-dose heparin GP IIb/IIIa inhibition (18). In light of these findings, it is unclear whether bivalirudin provisional GP IIb/IIIa inhibition would offer economic or clinical advantages for these low-risk populations. Further studies with adequate statistical power to demonstrate modest differences in ischemic complication rates are an essential first step to resolve this issue. Study limitations. Several limitations of our study are worth noting. We did not collect primary cost data on all study participants. Given the size and scope of the trial, we felt that such an effort would have been relatively inefficient. Instead, we elected to collect primary cost data on a large subset of patients ( 2,500) selected at random, including all patients who experienced a major complication. As a result, our imputation model was highly predictive (R ) and stable over a broad range of alternative sampling scenarios. Second, the time frame encompassed by our analysis was relatively brief. Because six-month rates of target vessel revascularization did not differ between the two treatment groups in the current study (17), however, it is unlikely that longer-term follow-up would have improved our estimation of the relevant cost differences. Third, the heparin dose selected for our study (65 U/kg) was slightly higher than the dose previously studied in the Enhanced Suppression of the Platelet IIb/IIIa Receptor with Integrilin Therapy (ESPRIT) trial (60 U/kg) (3). Whether use of a lower heparin dose would have mitigated the excess bleeding events seen in the heparin GP IIb/IIIa arm of the REPLACE-2 trial with no loss of efficacy is purely speculative, however. Fourth, the results of our economic analysis apply only to those patients who were actually studied in the REPLACE-2 trial. Because patients with ongoing acute MI or high-risk acute coronary syndromes receiving GP IIb/IIIa inhibition before PCI were excluded by protocol, our findings should not be extrapolated to these specific populations. Finally, it is likely that we have underestimated the extent of cost savings that might ultimately be achieved by bivalirudin in contemporary PCI. In particular, it is possible that the design of the REPLACE-2 trial artificially increased hospital costs for the bivalirudin group. The mean infusion duration for bivalirudin was 44 min, compared with 12 h for abciximab and 18 h for eptifibatide. Nonetheless, as a double-blind trial, all patients in the bivalirudin group received 12 to 18 h of a placebo infusion, thus eliminating an important opportunity to streamline care and reduce cost. Outpatient PCI is rarely performed in U.S. today, in part because of the need for extended infusions of antiplatelet agents as well as lower reimbursement levels compared with inpatient procedures. Given its brief infusion duration and enhanced safety profile, bivalirudin with provisional GP IIb/IIIa inhibition might lead to substantial additional savings in the future by facilitating PCI in the outpatient setting. Reprint requests and correspondence: Dr. David J. Cohen, Cardiovascular Division, Beth Israel Deaconess Medical Center, 330 Brookline Avenue, Boston, Massachusetts dcohen@caregroup.harvard.edu. REFERENCES 1. The EPILOG Investigators. Platelet glycoprotein IIb/IIIa receptor blockade and low-dose heparin during percutaneous coronary revascularization. N Engl J Med 1997;336: The EPISTENT Investigators. Randomised placebo-controlled and balloon-angioplasty-controlled trial to assess safety of coronary stenting with use of platelet glycoprotein-iib/iiia blockade. Lancet 1998; 352: The ESPRIT Investigators. Novel dosing regimen of eptifibatide in planned coronary stent implantation (ESPRIT): a randomised, placebo-controlled trial. Lancet 2000;356: Anderson KM, Califf RM, Stone GW, et al. Long-term mortality benefit with abciximab in patients undergoing percutaneous coronary intervention. J Am Coll Cardiol 2001;37: Karvouni E, Katritsis DG, Ioannidis JP. Intravenous glycoprotein IIb/IIIa receptor antagonists reduce mortality after percutaneous coronary interventions. J Am Coll Cardiol 2003;41: Lincoff AM, Bittl JA, Harrington RA, et al. and provisional glycoprotein IIb/IIIa blockade compared with heparin and planned glycoprotein IIb/IIIa blockade during percutaneous coronary intervention: REPLACE-2 randomized trial. JAMA 2003;289: American Heart Association. Heart Disease and Stroke Statistics 2004 Update. Dallas, TX: American Heart Association, Cohen DJ, Krumholz HM, Sukin CA, et al. In-hospital and one-year economic outcomes after coronary stenting or balloon angioplasty: results from a randomized clinical trial. Circulation 1995;92: Shwartz M, Young DW, Siegrist R. The ratio of costs to charges: how good a basis for estimating hospital costs. Inquiry 1995;32: Taira DA, Seto TB, Siegrist R, Cosgrove R, Berezin R, Cohen DJ. Comparison of analytic approaches for the economic evaluation of new technologies alongside multicenter clinical trials. Am Heart J 2003; 145: Ellis SG, Miller DP, Brown KJ, et al. In-hospital cost of percutaneous coronary revascularization: critical determinants and implications. Circulation 1995;92: Mark D, Talley J, Topol E, et al. Economic assessment of platelet glycoprotein IIb/IIIa inhibition for prevention of ischemic complications of high-risk coronary angioplasty. Circulation 1996;94: Lincoff AM, Mark DB, Tcheng JE, et al. Economic assessment of platelet glycoprotein IIb/IIIa receptor blockade with abciximab and

9 1800 Cohen et al. JACC Vol. 44, No. 9, 2004 Cost-Effectiveness of for PCI November 2, 2004: low-dose heparin during percutaneous coronary revascularization: results from the EPILOG randomized trial. Circulation 2000;102: Weintraub WS, Culler S, Boccuzzi SJ, et al. Economic impact of GP IIb/IIIa blockade after high-risk angioplasty: results from the RESTORE trial. J Am Coll Cardiol 1999;34: Topol E, Mark D, Lincoff A, et al. Outcomes at 1 year and economic implications of platelet glycoprotein IIb/IIIa blockade in patients undergoing coronary stenting: results from a multicentre randomised trial (EPISTENT). Lancet 1999;354: Cohen DJ, Cosgrove RS, Berezin RH, et al. Cost-effectiveness of eptifibatide in patients undergoing planned coronary stenting: results from the ESPRIT trial (abstr). Circulation 2001;104: I Lincoff AM, Topol EJ. Long-term efficacy of bivalirudin and provisional GP IIb/IIIa blockade compared with heparin and planned GP IIb/IIIa blockade during percutaneous coronary intervention: 6-month and 1-year results of the REPLACE-2 trial (abstr). Circulation 2003;108:IV Kastrati A, Mehilli J, Schuhlen H, et al. A clinical trial of abciximab in elective percutaneous coronary intervention after pretreatment with clopidogrel. N Engl J Med 2004;350:232 8.