Cost-Effectiveness of Coronary Stenting and Abciximab for Patients With Acute Myocardial Infarction
Results From the CADILLAC (Controlled Abciximab and Device Investigation to Lower Late Angioplasty Complications) Trial
Background— Both stenting and the glycoprotein IIb/IIIa inhibitor abciximab improve outcomes for patients undergoing primary angioplasty for acute myocardial infarction (AMI). However, the cost-effectiveness of these strategies is unknown.
Methods and Results— We performed a prospective cost-utility analysis among US participants in the CADILLAC trial. Patients with AMI (n=1703) were randomized to stenting versus balloon angioplasty (PTCA) and abciximab versus no abciximab according to a 2-by-2 factorial design. Total 1-year costs and lifetime incremental cost-effectiveness ratios, measured as cost per quality-adjusted year of life (QALY) gained, were calculated. Compared with PTCA, stenting increased procedural costs by $1148 and initial hospital costs by $1384 (both P<0.001). By 1-year, stenting led to fewer repeat revascularization procedures and reduced follow-up medical care costs by $1215, such that aggregate costs were similar for the PTCA and stent groups ($18 690 versus $18 859, P=0.75). The cost-effectiveness ratio for stenting versus PTCA was favorable at $11 237/QALY gained and remained <$20 000/QALY in sensitivity analyses. Compared with standard anticoagulation, abciximab increased initial procedural costs by $1122 (P<0.001). By facilitating accelerated hospital discharge, abciximab reduced length of stay by ≈0.6 days, offsetting most of the drug costs. These cost offsets were not maintained, however; aggregate 1-year costs for the abciximab group were $1244 greater than for standard therapy ($19 389 versus $18 145, P=0.02). Abciximab was reasonably cost-effective (cost-effectiveness ratio $21 305/QALY) only if nonsignificant differences in 1-year mortality (3.7% versus 4.3%, P=0.62) were incorporated in the analysis.
Conclusions— Primary stenting is a highly cost-effective treatment for AMI. The cost-effectiveness of abciximab in this setting is uncertain and depends primarily on whether long-term survival is enhanced.
Received August 8, 2003; revision received September 23, 2003; accepted September 25, 2003.
Primary PTCA can improve early and late outcomes compared with thrombolytic therapy in patients with acute myocardial infarction (AMI).1,2 The results of balloon angioplasty, however, may be compromised by dissection, reocclusion, and restenosis. Recently, both stenting and glycoprotein (GP) IIb/IIIa inhibition have been shown to reduce the incidence of these complications.3–6
See p 2831
Whether the clinical benefits of these treatments justify their additional costs is unknown. In the Stent-Primary Angioplasty in acute Myocardial Infarction (PAMI) trial, stenting reduced restenosis and the need for repeat revascularization among patients undergoing primary PTCA for AMI.4 These benefits were offset by a trend toward increased mortality, however, which led to uncertainty as to both the clinical and economic value of stent implantation in AMI.7 Moreover, this trial was performed before widespread use of GP IIb/IIIa inhibition.
The Controlled Abciximab and Device Investigation to Lower Late Angioplasty Complications (CADILLAC) trial demonstrated that among primary PTCA patients, stenting improved late outcomes compared with balloon angioplasty alone (by reducing restenosis), whereas abciximab improved short-term outcomes by reducing acute thrombosis and the need for urgent target-vessel revascularization.8 To determine the cost-effectiveness of these adjunctive therapies, both alone and in combination, we performed a prospective economic evaluation in conjunction with the CADILLAC trial.
Patient Population and Protocol
Between November 1997 and September 1999, 2082 patients were enrolled in the CADILLAC trial, a randomized study to compare the clinical and angiographic outcomes of PTCA versus stenting, with or without abciximab, for patients undergoing primary percutaneous coronary intervention for AMI. The economic substudy was performed in parallel with the randomized clinical trial and included all 1703 patients enrolled at the 60 US centers. Details of the study protocol and the 6-month clinical and angiographic results have been published.8 Briefly, patients presenting within 12 hours of a suspected AMI were referred for emergent coronary angiography. Patients with suitable anatomy were then randomly assigned to 1 of 4 reperfusion strategies: PTCA alone, PTCA plus abciximab, stenting alone, or stenting plus abciximab. Abciximab (Centocor) was administered as a 0.25-mg/kg bolus, followed by a 12-hour infusion of 0.125 μg · kg−1 · min−1 (maximum 10 μg/min).
For PTCA patients, crossover to stenting was allowed if necessary to treat a severe dissection or an unacceptable PTCA result. For nonabciximab patients, crossover to abciximab was allowed if flow could not be reestablished in the absence of a mechanical obstruction (“no reflow”) or if there was persistent angiographic thrombus. Patients randomized to abciximab were encouraged to be discharged at ≈2 days (for low-risk patients) or ≈3 days (for high-risk patients) if clinically stable. Patients not receiving abciximab were discharged at the discretion of the treating physician.
Assessment of In-Hospital Outcomes and Clinical Follow-Up
Detailed case report forms that included baseline patient characteristics, procedural details, and clinical events during the initial hospitalization and at 6-month and 1-year follow-up were completed by research coordinators at each site and submitted to the data coordinating center. Follow-up angiography at 7 months was prespecified for a subgroup of 900 patients. An independent clinical events committee, blinded to treatment assignment, adjudicated all major end points.
Determination of Medical Care Costs
For patients in the economic substudy, UB-92 forms (summary bills) and itemized bills were collected beginning with each patient’s index hospitalization and continuing for 1 year.
Cardiac Catheterization Laboratory Costs
The costs of each cardiac catheterization laboratory procedure were determined by standard “bottom-up” cost-accounting methods.7 Detailed resource utilization was recorded for each procedure, and the costs of each item were estimated on the basis of the hospital acquisition cost for the item in 2001. The cost of each stent was set at $1200 on the basis of the typical price of stents during that period, and the cost of abciximab was estimated on the basis of the number of vials opened at a cost of $350/vial. Overhead and depreciation for the cardiac catheterization laboratory and labor costs were determined on the basis of measured procedure duration and the unit cost of these resources at Beth Israel Deaconess Medical Center in 2001.
Other Hospital and Physician Costs
All other hospital costs were determined by “top-down” accounting methods based on each hospital’s annual Medicare Cost Report.7 Hospital room and nursing costs were based on the average per diem cost for the specific patient care unit, multiplied by the length of stay on that unit. Ancillary costs were determined by multiplying hospital charges by the cost-center specific cost-to-charge ratio obtained from the hospital’s Medicare Cost Report. All costs were adjusted to 2001 levels with the medical care component of the Consumer Price Index. To reduce the impact of high-cost outliers, hospitalization costs greater than the 97.5% margin for the overall distribution were assigned costs equivalent to that margin. Physician fees were estimated with the 2001 Medicare Fee Schedule.
For those patients with hospitalizations for which billing data were unavailable (n=186, 11%), hospital room and ancillary costs were imputed on the basis of a linear regression model of nonprocedural costs from those patients with available billing data (model R=0.64). A clinical events committee, blinded to treatment assignment, evaluated all repeat revascularization procedures to determine whether they were “clinically driven.” Any procedures and their associated hospitalizations not judged to be clinically driven were excluded from our economic analysis.
Analytic Approach and Statistical Methods
Continuous, normally distributed data are described as mean±1 SD and were compared with Student’s t test. Nonnormally distributed data were compared with the Wilcoxon rank-sum test. Cost data are described as both mean and median values and were compared by Student’s t test. Although costs were not generally normally distributed, the question addressed by the t test (ie, “Are the mean costs significantly different?”) most closely approximates the perspective of a typical third-party payer or healthcare system. All comparisons were made on an intention-to-treat basis, and a 2-tailed P<0.05 was considered statistically significant. All analyses were performed with STATA version 7.0 (Stata Corporation).
Because stenting and abciximab increased overall 1-year costs, we performed formal cost-effectiveness analyses to compare the additional cost of each therapy with the clinical benefits achieved. For each comparison, we assessed an incremental cost-effectiveness ratio in terms of cost per quality-adjusted year of life (QALY) gained. Quality-adjusted survival for each treatment group over the first year of follow-up was estimated on the basis of observed rates of repeat revascularization with empiric health-state-specific utility weights derived from 771 patients enrolled in the Stent-PAMI trial (1 year without repeat revascularization=0.86±0.18 QALYs; 1 year with repeat revascularization=0.80±0.19 QALYs; P=0.003).7 Because there were no significant mortality differences over the 1-year follow-up period, we excluded the effects of mortality from the primary analysis but performed a secondary analysis that incorporated these effects. Quality-adjusted life expectancy for each 1-year survivor was estimated at 11.99 QALYs (discounted at 3%), and follow-up medical care costs beyond year 1 were estimated at $3801/year on the basis of projections for a typical 60-year-old myocardial infarction patient from the Cholesterol And Recurrent Events (CARE) trial population.9 To estimate the uncertainty surrounding the cost-effectiveness ratios, we calculated bias-corrected confidence intervals for each ratio by the bootstrap method, using 1000 repeat samplings of the study population.10
Baseline and angiographic characteristics of the study population are summarized in Table 1. All 4 treatment groups were well matched for sociodemographic characteristics, clinical characteristics, and risk factors.
Procedural Resource Utilization and Cost
Table 2 summarizes selected resource utilization measures and costs for the initial revascularization procedures. Stenting was performed in 98% of patients in the stent group, whereas 18% of patients in the PTCA group required bailout stenting. Mean procedural costs were $1148/patient higher for the stent group than for the PTCA group ($4652±1283 versus $3504±1417, P<0.001). Randomization to abciximab (rather than standard anticoagulation) was associated with modest reductions in procedure duration and number of angioplasty balloons. After the acquisition cost of abciximab was included, mean procedural costs were increased by $1122/patient in the abciximab group ($4629±1270 versus $3507±1442, P<0.001).
Initial Hospital Outcomes, Resource Utilization, and Costs
Initial hospital outcomes and resource utilization were similar for the stent and PTCA groups (Table 3). In particular, there were no significant differences for in-hospital mortality, the rate of repeat revascularization, postprocedure length of stay, or postprocedure costs between the 2 groups. Thus, mean medical care costs for the index hospitalization were $1384/patient higher for patients assigned to stenting rather than PTCA ($13 905±5506 [median $12 340] versus $12 521±5742 [median $10 982], P<0.001).
In contrast, abciximab was associated with modest clinical benefits over standard anticoagulation during the index hospitalization, including lower rates of recurrent myocardial infarction (0% versus 0.5%, P=0.06) and subacute thrombosis (0.4% versus 1.4%, P<0.001). There were trends toward reduced in-hospital mortality and need for bypass surgery as well. As a result of these improved clinical outcomes and the protocol-recommended early-discharge protocol, use of abciximab reduced mean postprocedure length of stay by more than half a day (5.4 versus 4.8 days, P<0.001) and offset nearly $1000 of the drug-acquisition cost for abciximab. Thus, initial hospital costs for the abciximab group were only $413/patient higher than for the standard-therapy group ($13 413±5309 [median $12 045] versus $13 000±6006 [median $11 359], P=0.13).
Follow-Up Medical Resource Utilization and Costs
There were no statistically significant differences in death, recurrent myocardial infarction, or stroke between the treatment groups over the 1-year follow-up period (Table 4). However, patients assigned to stenting were 16% less likely to be rehospitalized and required 35% fewer revascularization procedures. As a result, mean follow-up medical care costs were $1215/patient lower for stenting than for PTCA ($4954±8106 versus $6169±9539, P=0.005), with significant reductions in all cost categories such that aggregate 1-year medical care costs were only $169 higher for the stent group than for the PTCA group. Of note, virtually all of the follow-up cost savings were realized during the first 6 months of follow-up, before the performance of protocol-mandated angiographic follow-up in a subset of patients (Figure 1).
Compared with standard anticoagulation, randomization to abciximab was associated with more frequent rehospitalization during follow-up (39.3% versus 32.9%, P=0.006). There were trends toward increased repeat revascularization with abciximab treatment as well (22% versus 19%, P=0.13). These differences resulted in higher repeat procedural costs and physician fees, such that total follow-up costs were $831 higher for the abciximab group ($5976±9479 versus $5145±8194, P=0.05), which offset much of the savings achieved during the index hospitalization (Figure 2). Aggregate 1-year costs were thus $1244 higher for the abciximab group than for the standard anticoagulation group ($19 389±11 345 [median $16 041] versus $18 145±$10 441 [median $14 718], P=0.02).
Because there were no significant differences in 1-year mortality between any of the treatment groups, our primary cost-effectiveness analysis excluded any long-term mortality effects and assumed that there would be no further differences in annual costs, utilities, or mortality rates beyond the 1-year follow-up period. Under these assumptions, stenting was estimated to improve quality-adjusted life expectancy by 0.015 years compared with PTCA, with an overall cost-effectiveness ratio of $11 237/QALY gained (Table 5). The cost-effectiveness ratio for stenting compared with PTCA remained <$50 000/QALY in 86% of bootstrap simulations. In this scenario, the benefits of stenting were determined entirely by the observed reductions in clinically driven repeat revascularization and the associated short-term quality-of-life benefits. Under the same assumptions, standard (no abciximab) anticoagulation was an economically dominant strategy (with lower long-term costs and greater overall quality-adjusted life expectancy) than routine abciximab use in more than 75% of simulations.
If the observed differences in 1-year mortality associated with either stenting (0.2% absolute risk reduction) or abciximab treatment (0.6% absolute risk reduction) were included in our analysis, the cost-effectiveness ratio for stenting versus PTCA remained highly favorable at $7067/QALY, but the variability increased. In contrast, the cost-effectiveness of abciximab was highly sensitive to the inclusion of 1-year mortality differences. When mortality effects were included in our analysis, abciximab was projected to improve overall quality-adjusted life expectancy by 0.058 years and had a cost-effectiveness ratio of $25 136/QALY gained, which remained <$50 000/QALY in 64.0% of bootstrap simulations.
This prospective economic substudy addresses several critical questions about the value of stenting and adjunctive GP IIb/IIIa inhibition for patients undergoing primary percutaneous coronary intervention for AMI. Although stenting increased initial hospital costs by ≈$1400/patient, by 1 year, costs were increased only minimally because of substantial reductions in rehospitalization and repeat revascularization. Given the improved short-term outcomes, minimal incremental cost, and absence of any adverse effects on death or reinfarction, these findings suggest that stenting as performed in the CADILLAC trial is an economically attractive treatment strategy for patients with AMI, with cost-effectiveness ratios ranging from $5000 to $15 000 per QALY gained- values that compare favorably with other accepted medical interventions.
These findings corroborate and extend previous observations from the Stent-PAMI trial. In our economic evaluation of Stent-PAMI, we speculated that the availability of longer stents and high-pressure delivery systems would reduce the 1-year cost difference from more than $900/patient to ≈$300/patient.7 In CADILLAC, the 1-year cost difference was even less, most likely owing to continued reductions in stent prices. Indeed, with recent reductions of bare-metal stent prices to ≈$1000 US (compared with $1200 during CADILLAC), stenting may now be cost saving compared with PTCA.
Unlike Stent-PAMI, the CADILLAC trial did not find evidence of increased mortality with stenting compared with PTCA alone. Consequently, the cost-effectiveness ratio for stenting was highly attractive, regardless of whether 1-year mortality differences were included in the present analysis. The differential mortality effects between these studies may relate to differences in stent design, implantation techniques, or random chance. It does not appear that abciximab was critical to this benefit, however, because 1-year mortality in CADILLAC tended to be lower in the stent-alone group than in the stent-with-abciximab group (3.4% versus 4.4%, P=0.42).
The results of the present analyses comparing abciximab and standard anticoagulation are more complex and depend in part on the perspective of the evaluation. During the index hospitalization, treatment with abciximab was associated with cost offsets of nearly $1000/patient, such that the net cost of abciximab use was only ≈$400/patient. These cost offsets are directly relevant to most US hospitals that currently operate under noncapitated payment systems and suggest that at least in the short term, the potential benefits of abciximab therapy might be achieved at only a modest incremental cost. These cost offsets were related to 2 main factors: reduced complexity of the index revascularization procedures (which accounted for ≈20% of the offset) and reduced in-hospital complications and briefer lengths of stay for the abciximab group (which accounted for ≈80% of the offset). Because the study was nonblinded and the protocol encouraged very early discharge for the abciximab group, it is unknown whether similar lengths of stay could have been achieved in the absence of abciximab therapy. It does not appear that the protocol artificially increased length of stay for the nonabciximab group, however, because mean length of stay in other contemporary studies was identical to that observed in the no-abciximab group in CADILLAC.7
During the 1-year follow-up period, randomization to abciximab was associated with an increase in follow-up costs of >$800/patient. As a result, net 1-year costs were ≈$1200/patient higher for the abciximab group than for standard anticoagulation. The precise mechanism for these higher follow-up costs is uncertain. Although there were modest increases in rehospitalization and repeat revascularization procedures, detailed angiographic analysis in a subset of CADILLAC patients did not demonstrate differences in measures of coronary restenosis.8 It is possible that by reducing the extent of distal embolization, abciximab improved myocardial salvage, thus leading to more severe anginal symptoms for any degree of subsequent angiographic restenosis.11 Indeed, a similar finding of increased repeat revascularization among abciximab-treated patients was noted in the EPILOG trial.12 An alternative hypothesis is that the accelerated-discharge pathway recommended for the abciximab group resulted in increased readmission rates during follow-up. There was no evidence of increased complications in the abciximab group, however. In fact, readmission rates within the first 7 days actually tended to be lower in patients randomized to abciximab versus no abciximab (3.7% versus 5.6%, P=0.08).
Regardless of the mechanism, most of the initial cost offsets with abciximab were no longer apparent at 1-year. As a result, the cost-effectiveness of abciximab in the AMI setting was highly dependent on its effect on long-term survival. In our primary analysis, in which we assumed that there was no late survival benefit with abciximab treatment, conventional anticoagulation was an economically dominant strategy. In a sensitivity analysis, however, life expectancy projections based on the observed 0.6% absolute reduction in 1-year mortality translated into a reasonable cost-effectiveness ratio of ≈$25 000/QALY gained. Although the CADILLAC trial by itself was not adequately powered to detect mortality differences of this magnitude, a recent meta-analysis of all abciximab trials (including CADILLAC) found that abciximab treatment was associated with a 29% reduction in all-cause mortality and an absolute mortality difference of 0.9% at 1 year.13 Because AMI patients represented only ≈20% of the pooled population, however, the level of evidence for mortality reduction in this specific subgroup is relatively modest. Nonetheless, these findings suggest that the mortality reduction observed in CADILLAC may be real and justify its incorporation in the sensitivity analysis.
Our study has several limitations. Because we did not collect primary utility data in CADILLAC, we used health-state-specific utility weights from the Stent-PAMI trial to estimate patient-specific utility weights for the CADILLAC population. We believe that this approach is reasonable because both were studies of AMI patients and had virtually identical inclusion and exclusion criteria. Moreover, the present study had limited statistical power to detect clinically relevant mortality differences. Finally, neither outcomes nor cost data were available beyond the first year. If the observed trends in follow-up costs were to persist for several more years, it is possible that stenting would have been cost-saving in the long run, and the cost-effectiveness of abciximab would be less favorable than we have projected.
Given the results of the CADILLAC trial, primary stenting for AMI appears to be an economically attractive strategy compared with other accepted medical interventions. The cost-effectiveness of abciximab in this setting is less certain and depends primarily on whether long-term survival is improved. Further studies with longer follow-up are therefore needed to define more clearly the benefits of adjunctive GP IIb/IIIa inhibition in the AMI setting, particularly in conjunction with stent placement.
The CADILLAC trial was supported by grants from Guidant (Santa Clara, Calif) and Lilly Research Laboratories (Indianapolis, Ind). Dr Bakhai was supported by a grant from the Royal Brompton Hospital (London, UK). A complete list of the study organization, contributing centers, and investigators may be found in Reference 8.
Dr Grines is a recipient of research grants from Berlex; Pfizer (REVERSAL); GLAXO Smith Kline (PPAR); Aventis (SYNERGY and LV Thrombus); Guidant, Eli Lilly, SCIMED, Johnson & Johnson, and Amersham Health (Senior PAMI); Otsuka (CREST); Esperion Therapeutics (Esperion, etc); and Innercool Therapies (ICE-IT). She has served as a consultant to and/or member of the advisory boards of Innercool Therapies, Aventis, Guidant, Pfizer, and The Medicines Company. Dr Stone has served as a consultant to Guidant and has received research support from Guidant and Eli Lilly. Dr Bakhai has participated in research that has grant support from Eli Lilly, Guidant, and Boston Scientific.
Suryapranata H, van’t Hof AW, Hoorntje JC, et al. Randomized comparison of coronary stenting with balloon angioplasty in selected patients with acute myocardial infarction. Circulation. 1998; 97: 2502–2505.
Lefkovits J, Ivanhoe RJ, Califf RM, et al. Effects of platelet glycoprotein IIb/IIIa receptor blockade by a chimeric monoclonal antibody (abciximab) on acute and six-month outcomes after percutaneous transluminal coronary angioplasty for acute myocardial infarction. Am J Cardiol. 1996; 77: 1045–1051.
Cohen DJ, Taira DA, Berezin R, et al. Cost-effectiveness of coronary stenting in acute myocardial infarction: results from the Stent-PAMI trial. Circulation. 2001; 104: 3039–3045.
Hunink MG, Bult JR, de Vries J, et al. Uncertainty in decision models analyzing cost-effectiveness: the joint distribution of incremental costs and effectiveness evaluated with a nonparametric bootstrap method. Med Decis Making. 1998; 18: 337–346.
Lincoff AM, Tcheng JE, Califf RM, et al. Sustained suppression of ischemic complications of coronary intervention by platelet GP IIb/IIIa blockade with abciximab: one-year outcome in the EPILOG trial. Circulation. 1999; 99: 1951–1958.