Frequency and Costs of Ischemic and Bleeding Complications After Percutaneous Coronary Interventions: Rationale for New Antithro

Continued from previous page Bleeding complications in practice. Demographics and definitions. As mentioned previously, data from multi- and single-center registries of PCI report much higher bleeding complication rates than those seen in clinical trials. Differing patient demographics in clinical practice may partly explain this discrepancy. For example, in the ESPRIT trial, patients with a baseline hemoglobin of less than 11 g/dl were excluded from the trial, as were patients with moderate renal dysfunction.²¹ In addition, patients enrolled in clinical trials are younger and more likely to be male compared with patients in practice. Multivariate analysis has shown that advanced age, female gender and renal insufficiency are risk factors for bleeding complications.³⁶ Secondly, definitions of bleeding complications used in registries tend to be more inclusive than criteria for major bleeding used in clinical trials. For example, bleeding that results in a drop in hemoglobin >= 3 g/dl, receipt of a blood transfusion or other blood products, or any bleeding complication that requires modification of care or results in a prolonged length of stay may be included as a bleeding complication in registry data. This definition of bleeding is more similar to a combination of TIMI major, TIMI minor plus transfusions. When defined in this manner, rates of bleeding complications in clinical trials are actually similar to rates reported by various national datasets and registries (Table 3). Transfusion rates and vascular complications in practice. One way of tracking bleeding complications in large databases is the use of blood transfusions. In the NCN database, an observational study of over 100,000 patients undergoing PCI, 4.0% of patients required a transfusion.31 Of note, in 7,472 patients over 80 years of age enrolled in this database, the transfusion rate was 9.9%. Aronow et al. found that 6.7% of patients required a blood transfusion in his analysis of 373 patients undergoing stent placement.³² In a different single-center analysis, Moscucci et al. found that blood transfusions were administered in 8.9% of patients undergoing PCI.³⁷ In the PRICE study of 320 consecutive patients undergoing elective stent placement with either abciximab (n = 163) or eptifibatide (n = 157), the frequency of serious bleeding complications (intracranial, retroperitoneal, surgical repair of the vascular access site, blood or platelet transfusion) was 6.2% in patients receiving abciximab and 3.2% in patients receiving eptifibatide.³⁸ More recently, a transfusion rate of 5.6% was reported from a quality-controlled regional multicenter PCI registry.⁴⁰ Finally, in the unpublished MQ-Pro database, transfusion not associated with CABG occurred in 6.0% of patients undergoing PCI.³³ The frequency of local vascular complications in patients undergoing PCI has also been reported in several registry studies.^40,41 For example, in the NCN registry, vascular complications defined as entry-site vessel occlusion, large groin hematoma prolonging the hospital stay, documented pseudoaneurysm formation, entry-site bleeding requiring transfusion or surgical repair, dissection of the aorta, iliac or femoral arteries, or acute limb ischemia occurred in 3.5% of patients overall and in 6.7% of patients > 80 year old.³¹ Aronow et al. defined vascular complications as pseudoaneurysm, arteriovenous fistula, femoral neuropathy, retroperitoneal hematoma, vascular access site hematoma or other vascular access site complications requiring transfusion or causing a drop in hemoglobin of > 3 g/dl. In that database, vascular complications occurred in 4.0% of patients.32 In addition, Saucedo et al. noted vascular complications in 5.7% of patients undergoing “successful” coronary stenting, although specific criteria for this endpoint were not provided.³⁰ Again extrapolating to the PCI population as a whole, blood transfusions or other serious bleeding complications occur in approximately 6.0% of patients undergoing PCI, resulting in approximately 48,000 bleeding complications each year. Economics in healthcare As the cost of providing healthcare continues to rise, economic analyses have become increasingly important to healthcare decision makers. In the PCI realm, economic analyses have focused on adjunctive pharmacotherapy and coronary stenting, two strategies shown to increase the cost of PCI.^7,42 When a more expensive technology improves outcomes compared to usual care, cost-effectiveness analysis determines if the new therapy is worth the added expenditure. This type of economic analysis is performed from the societal perspective and is concerned with long-term outcomes such as mortality (years of life saved) and quality of life (quality-adjusted years of life saved).⁴³ Typically, a therapy is said to be “cost-effective” if it costs less than $50,000 per year of life saved.⁴³ Notably, cost-effective therapies by definition do not decrease costs. In the rare event a new therapy improves outcomes to the extent that it offsets its own cost (i.e., it is cost-saving), that therapy is said to be “dominant”.⁴³ Dominant therapies need not be subjected to further analysis and should be adopted since they both improve outcomes and decrease costs. Overview of reimbursement issues. Economic analyses such as cost-effectiveness do not take into account the realities of reimbursement in today’s healthcare market. A minority of markets in the US function under a global capitation system, which rewards healthcare systems for decreasing the cost of care in the long-term (i.e., providing treatments that are “cost-effective”). In this type of market, both hospital and follow-up costs are important since reimbursement is not tied to a particular incident of care, but to the overall cost of cardiovascular services incurred by the covered population.⁴⁴ In addition, other private payers reimburse for inpatient services on a “usual, customary and reasonable” (UCR) basis, discounted charges, per diem, or per case rate. In this financial environment, increased utilization of resources is typically reimbursed if it meets the criteria of UCR.⁴⁴ However, the majority of patient hospital expenses for PCI are paid using the Diagnosis-Related Groups (DRG) prospective payment system, which pays a fixed rate for each admission regardless of resources consumed.⁴⁴ This is the payment system used for reimbursement in Medicare patients, who receive about half of all PCI procedures. In this reimbursement structure, institutions do not have a financial incentive to improve long-term outcomes using interventions that increase total hospital cost unless those interventions are reimbursed at a higher rate. In 1999, the CMS, recognizing the long-term benefits of coronary stent placement toward decreasing the need for repeat PCI, created a new DRG classification for stenting that was reimbursed an average of $1,500 higher than balloon angioplasty.⁴⁵ In 2002, DRG reimbursement codes and rates for PCI changed again, increasing reimbursement for patients with MI undergoing intervention, but decreasing reimbursement an average of 8% in patients undergoing routine stent placement or angioplasty (Table 4).⁴⁶ In this type of financial environment, economically dominant therapies would be most attractive by improving quality of care while decreasing costs. Cost of complications To illustrate the importance of the cost of complications on the total cost of PCI, one series found that complications related to the procedure accounted for 28% of in-hospital costs for PCI.⁶ Increased length of stay, death, eCABG, post-procedural MI and blood transfusions were the most costly items in this analysis. In this section, we will review the data on costs of complications as determined in both clinical trials and in practice. Cost of complications in clinical trials. Variable cost estimates of ischemic and hemorrhagic complications of PCI are available in several studies.^{13,33,47–51} A summary of these results is presented in Table 5. In the EPILOG study, patients were randomized to heparin, or abciximab with regular or low-dose heparin.¹⁵ For the economic substudy, hospital bills were collected and charges converted to costs using department-specific cost to charge ratios.⁵² In order to differentiate between the effects of decreased ischemic endpoints and increased bleeding complications with abciximab therapy, multivariate analysis was conducted to determine individual components of the cost of hospitalization. Sources of cost differences between heparin and abciximab plus low-dose heparin groups during the index hospitalization were determined using multivariate linear regression. The incremental cost of eCABG was estimated at $10,380, repeat urgent PCI at $2,467 and MI at $1,180. In addition, a TIMI major bleed cost $3,295, whereas a TIMI minor bleed cost $343. A similar methodology was used in the RESTORE economic analysis.⁴⁸ This trial randomized patients to receive heparin or tirofiban plus heparin. Multivariate analysis was used to determine cost of ischemic complications, but costs of bleeding complications were not reported. In this analysis, CABG cost $14,461, repeat PCI $4,925, and MI $1,605. In the BARI trial, patients with multivessel disease were randomized to PCI or CABG.⁴⁹ An economic and health-related quality of life study was performed in approximately half of the participating centers. Baseline costs of PCI were higher in the BARI trial, but incremental costs due to complications were similar to RESTORE.⁴⁹ For example, cost of CABG was $15,339, repeat PCI was $5,762, QWMI was $3,419 and NQWMI was $1,049. Vascular complications in this trial resulted in an incremental cost of $11,767. Notably, none of these trials routinely used stenting. Although some trials have shown a decrease in hospital mortality and eCABG with routine or bail-out stenting, the primary benefit of stenting is decreased repeat PCI in the long-term.⁵³ Since in-hospital mortality and eCABG are infrequent and the cost of stents relatively high, stenting has not been shown to decrease hospital costs.^54–56 Cost of complications in practice. Single-center registries and national databases have estimated the incremental cost of complications from PCI in practice. Lauer et al. used hospital claims data to estimate the cost of ischemic and bleeding complications in 37,088 patients undergoing PCI between 1995 and 1997.¹³ A multivariate regression model was used to predict baseline hospital costs and additional costs attributable to ischemic and bleeding complications. Same-admission CABG and repeat PCI were estimated to result in additional costs of $13,800 and $2,400, respectively. Incremental cost of medically managed acute MI was estimated to be $2,700. The incremental cost of bleeding, defined as a transfusion of one or more units of blood, was estimated to be $10,200. This dataset did not allow the authors to determine a temporal relationship between PCI and the complications of interest. Thus, some patients who were assumed to have complications as a result of PCI may in fact have experienced the events (e.g., MI) prior to the procedure. Using similar multivariate techniques, Cohen et al. estimated the incremental costs of complications in 3,241 patients at 134 different sites.50 Incremental cost of CABG during the index admission was $16,847, cost of repeat PCI was $5,780, cost of QWMI was $6,925, cost of NQWMI (CK-MB > 5 times ULN) was $4,561 and cost of a blood transfusion was $8,098. Similarly, in the MQ-Pro database, incremental cost of CABG during the index admission was $22,982, repeat PCI was $7,932, non-fatal QWMI was $6,744 and death was $10,122. Cost of transfusion in the MQ-Pro database averaged $8,096.³³ Also, investigators from Owen Healthcare used cost-accounting methods to determine incremental costs of complications in PCI.⁵¹ In 187 patients undergoing either elective or urgent PCI, repeat PCI cost $8,920, QWMI cost $3,625, death cost $11,181 and a blood transfusion cost $7,874. Finally, in a single-center registry, Moscucci et al. evaluated the relationship between patient demographics, procedure variables, complications, length of hospital stay and total PCI costs. In this analysis, complications including eCABG, contrast nephropathy and blood transfusion emerged as important independent predictors of higher hospital costs.⁵ Based on these estimates of complication frequency and incremental costs, complications represent a substantial and burdensome component of the $10 billion annual PCI expenditure in the US. For example, in the 800,000 PCI procedures performed each year, the estimated cost of MI is $216 million, the cost of repeat PCI is $300 million and the estimated cost of blood transfusions is $384 million. Strategies for reducing complications In an effort to reduce complications and improve outcomes in patients undergoing PCI, the American College of Cardiology suggests limiting angioplasty services to hospitals that perform more than 200–400 interventions annually and to physicians who perform more than 50–75 interventions per year.⁵⁷ Using models based on outcomes data from Medicare administrative datasets, Ellis et al. estimated that in order to have a substantial impact on mortality, catheterization laboratories performing less than 400–900 total cases per year would need to be closed.⁵⁸ However, such laboratory closures would require transfer to hospitals > 50 miles distant in 6–38% of cases, and in as many as 18–94% in low-density states. These results suggest that in order to achieve a sizable reduction in PCI-related mortality, large numbers of laboratories would need to be closed, which could adversely affect patient access to care. However, the authors found that substantial cost savings could be realized by only closing laboratories performing PCI in 58 Another strategy that could be used to decrease complications and cost of PCI is the use of anti-thrombotic regimes with improved safety and efficacy compared to usual care. As mentioned previously, bivalirudin is the only anti-thrombotic agent shown to reduce ischemic complications while decreasing the risk of bleeding complications compared to usual care.²⁷ However, the drug acquisition cost of bivalirudin is approximately $335 per patient,59 and it is > 300 times more expensive than unfractionated heparin. As discussed previously, under the DRG reimbursement system, hospitals do not have a financial incentive to adopt more expensive therapies unless they are reimbursed at a higher rate. However, if bivalirudin demonstrated a dominant economic profile (i.e., it reduced the cost of PCI complications enough to offset its own cost), then hospitals would have an incentive to use this therapy. In addition, if bivalirudin was substituted for a more expensive antithrombotic agent (e.g., GP IIb/IIIa inhibitors) without loss of efficacy, it would be economically attractive to institutions performing PCI. Lauer et al. conducted a cost analysis of the BAT trial using cost data from three published trials.⁶⁰ The low- and high-end estimates of cost from these three studies were used to determine a range of costs for PCI complications. These cost ranges were then applied to the event rates from BAT to determine potential cost savings with bivalirudin. The per patient cost of complications for heparin ranged from $1,310–$2,107. For bivalirudin, the range was $719–$1,264, resulting in potential cost savings per patient of $592–$843. These figures do not include the cost of drug therapy. If the cost of bivalirudin therapy averages $335 per patient, this would result in bivalirudin being economically dominant, with an average in-hospital cost-savings of $250–$500 per patient. Another scenario that would demonstrate an economically dominant position for bivalirudin is if bivalirudin use resulted in a decreased use of the more expensive anti-thrombotic agents abciximab and eptifibatide. Figures 1 and 2 illustrate the economic impact of bivalirudin use in a theoretical 1,000 PCI patients where GP IIb/IIIa inhibitors are used 80% of the time. The “break-even” point for bivalirudin in the abciximab scenario occurs when abciximab use is reduced by 20%. Likewise, reduction of eptifibatide use by 50% would completely offset the cost of bivalirudin in all patients. In addition, more cost savings are likely to be realized by the reduction in bleeding complications using bivalirudin alone compared with heparin plus GP IIb/IIIa inhibitors.²⁰ Conclusions Despite advances in technology, the risks of ischemic and bleeding complications among patients undergoing PCI remain significant. The development of these events is associated not only with significant morbidity and mortality, but also with substantially higher costs. Reduction in hospital complications associated with PCI should thus lead to a corresponding reduction in hospitalization cost. Higher-than-standard doses of heparin and use of more potent antiplatelet agents reduces ischemic complications of angioplasty, but is associated with increased ble eding complications among PCI patients, appear to be a very attractive economic strategy in certain groups of patients undergoing PCI.