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

Recent advances in catheter technology and antithrombotic therapy have led to a continuous improvement in outcomes of percutaneous coronary intervention (PCI). These improved outcomes have been associated with broadening of the indications for PCI, with an exponential growth in number of procedures performed, but they have also been paralleled by incremental procedure costs. The estimated costs of PCI currently range from $8,000–$13,000.¹ With over 800,000 cases performed each year in the United States (US) alone, this represents over $10 billion annually for the US Healthcare System.² Roughly half of these costs are incurred by the Center for Medicare and Medicaid Services (CMS, formerly known as the Health Care Financing Administration).3 Total costs of PCI include disposable equipment used during the procedure (balloons, catheters, stents, etc.), cardiac catheterization laboratory overhead and depreciation, nursing and pharmacy costs, laboratory costs and physician services. In addition, factors that have been found to be associated with increased PCI costs include the use of special devices such as atherectomy or vascular closure devices, the use of multiple stents, the use of platelet glycoprotein (GP) IIb/IIIa inhibitors, and the presence of certain patient demographic characteristics including advanced age, gender and other comorbidities.^1,4,5 Finally, complications related to the procedure have been identified in several studies as the single most significant contributor to increased costs of PC.^5–7 Methods to reduce the cost of PCI include re-use of balloon catheters,⁸ percutaneous revascularization performed at the same time as diagnostic catheterization,⁹ reduced anticoagulation, the use of new devices or pharmacological interventions to reduce restenosis and complications, and the use of competitive bidding for cardiac cath lab supplies.¹⁰ For example, the evolution of anticoagulation therapy in stented patients from a regime of post-procedural heparin and warfarin to one of thienopyridines and aspirin,¹¹ and the subsequent reduction of length of stay from 4 days in 1995 to 2 days in 2000, have helped keep total procedure costs down.¹² In addition, a reduction in complication rates appears to be a key target for cost reduction efforts. In support of this statement, in the economic assessment of the Evaluation of 7E3 for the Prevention of Ischemic Complications (EPIC) trial in high-risk patients, Mark et al. identified bleeding complications, urgent and non-urgent coronary artery bypass graft surgery (CABG), and urgent and non-urgent percutaneous transluminal coronary angioplasty (PTCA) as important correlates of incremental costs.⁷ Unfortunately, standard aggressive antithrombotic therapy aimed toward a reduction of ischemic complications is often associated with an increase in bleeding complications. In the analysis of the EPIC trial, the benefits of abciximab in decreasing procedure costs through a reduction of ischemic complications were offset by drug acquisition costs and by an increase in bleeding complications.⁷ Thus, with ischemic complications becoming more rare as a result of improvement in PCI technology and more aggressive antithrombotic therapy, bleeding has become a rather common and costly complication of PCI, with a blood transfusion estimated to add up to $8,000 to the cost of care for the PCI patient.¹³ Based on these premises, it appears that the next challenge in the care of PCI patients will be to determine how to continue to prevent ischemic complications without increasing the risk of bleeding. This paper examines the frequency of PCI complications in both recent clinical trials and actual practice, discusses the costs of complications, and explores improvements in patient management and particularly changes in anticoagulation therapy that might impact total costs of PCI. Complication rates in clinical trials Ischemic complications in clinical trials. Despite advances in PCI technology and adjunctive pharmacotherapy, data from clinical trials indicate that ischemic complications still occur in 5–15% of patients.^14–19 Typically, clinical trials define ischemic complications as a combination of death, myocardial infarction (MI; both Q-wave and non-Q wave) and either urgent or any target vessel revascularization (TVR). Different definitions of MI or revascularization can make comparisons across trials difficult. However, comparisons may still be possible through the application of strict meta-analysis methodology. A recent meta-analysis combined data from 6 double-blind PCI trials conducted predominantly in North America between 1993 and 1998.20 A total of 16,546 patients were enrolled in these trials (Table 1). Protocols and case report forms for trials included in the analysis were compared to ensure reasonable consistency of study methods, patient management, data reporting and data structure. Integration of the databases from the trials enabled a direct comparison of key event rates at 7 days, using standard classifications and criteria for severity. The meta-analysis showed that the use of high-dose heparin (175 U/kg) was associated with significantly less frequent clinical ischemic events (8.1%) than lower doses of heparin (100 U/kg; 10.3%). In this same meta-analysis, event rates in patients treated with low-dose heparin (70 U/kg) plus a GP IIb/IIIa inhibitor was 6.5%.²⁰ Although not included in this meta-analysis, it is worth noting that the incidence of death, MI and revascularization in the ESPRIT trial was 9.3% in patients treated with low-dose heparin alone (60 U/kg).²¹Bleeding complications in clinical trials. In clinical trials of antiplatelet and anti-thrombotic therapy in PCI, bleeding complications are generally defined using either thrombolysis in myocardial infarction (TIMI)²² or global utilization of streptokinase or tPA outcomes (GUSTO)²³ criteria (Table 2). Rates of major bleeding in clinical trials using these criteria are generally less than 2% (Table 3).^{14–19,21,24,25} However, these restrictive definitions may not capture all clinically significant bleeding. For example, neither the TIMI nor the GUSTO major bleeding definition includes the need for a blood transfusion as part of the criteria. Thus, a broader measure of bleeding using a combination of both major and minor bleeding defined by TIMI or GUSTO criteria appears more likely to be representative of bleeding rates in clinical practice. In the meta-analysis of contemporary PCI trials, TIMI criteria were used to classify hemorrhagic events, permitting direct comparisons between trials. In the high-dose heparin group, the combination of TIMI major and minor bleeding occurred in 10.5% of patients compared with a rate of 10.7% in the low-dose heparin group, while the bleeding rate was 14.3% in patients receiving a combination of GP IIb/IIIa inhibitors and low-dose heparin. As shown in Table 3, when both TIMI major and minor bleeding are combined in contemporary PCI trials, bleeding complications average 4–14%, depending on patient characteristics and the drug regime used. In addition, when transfusions are included in the definition, the frequency of bleeding complications increases substantially. For example, in NICE-3, bleeding complications were 10.5% when transfusions were included in the criteria, but only 2% of the patients experienced TIMI major bleeding.²⁶ Notably, the only adjunctive anti-thrombotic agent shown to reduce both ischemic and bleeding complications in PCI is bivalirudin. In the Bivalirudin Angioplasty Trial,²⁷ the risk of bleeding was decreased 62% in the bivalirudin group compared with high-dose heparin. The combined rate of TIMI major and TIMI minor bleeding in bivalirudin patients (n = 2,161) was found to be 4.3% in the meta-analysis of contemporary PCI trials with a corresponding ischemic event rate of 6.6%.²⁰ Complications in practice Ischemic complications in practice. Rates of ischemic complications in clinical practice are difficult to determine. Although several investigators have published data from multicenter databases, these data tend to be 3–5 years old by the time manuscripts are in print. Since trends in the published literature do show continued reduction in PCI complications over time, the frequency of complications noted in these publications may overestimate the actual rate of complications in clinical practice today. In addition, rates of complications can vary widely across institutions due to differences in practice patterns, definitions, operator skills and resource utilization. For example, in the Society for Cardiac Angiography and Interventions (SCA&I) registry, stent use among laboratories varied from 29–95%.²⁸ Others have found lower complication rates in patients whose procedure was performed by a high-volume operator or in a high-volume institution.²⁹ We identified 6 published reports of PCI complications in clinical practice reporting a variety of ischemic outcomes.^1,28–31 Saucedo et al. prospectively collected data on 900 patients undergoing successful elective stent placement in native coronary arteries between January 1994 and December 1995.³⁰ The purpose of this study was to evaluate the incidence and long-term clinical consequences of patients with creatine kinase (CK) myocardial isoenzyme band (CK-MB) elevations after stenting. By design, all patients in this observational study had a successful procedure defined as an increase of > 20% in luminal diameter with final percent diameter stenosis of 5 times ULN. In total, 3.9% of patients required a repeat diagnostic catheterization for recurrent ischemia and 1.2% required urgent target vessel revascularization. In this study, patients requiring the use of GP IIb/IIIa inhibitors were excluded. The Northern New England group (NNE) collected data on 14,498 patients undergoing PCI between 1994 and 1996.²⁹ In this study, outcomes included the in-hospital occurrence of death; emergency CABG (eCABG) or non-eCABG; or new MI (defined as chest pain, diaphoresis, dyspnea or hypotension associated with the development of new Q-waves or ST-T wave changes and a rise in CK to at least twice normal with a positive CK-MB). Overall, death occurred in 1.2% of patients, CABG in 2.6% (0.8% eCABG and 1.8% non-eCABG), and MI in 2%. Stents were used in 22% of patients enrolled in this registry. In the National Cardiovascular Network database (NCN), Batchelor et al. reported complications of PCI in 109,708 patients who underwent PCI between 1994 and 1997.³¹ In this observational study, in-hospital mortality was defined as the occurrence of death after the procedure, MI was defined as the appearance of new Q-waves in 2 contiguous leads on a 12-lead electrocardiogram (ECG) for up to 30 days post-PCI, and repeat revascularization was defined as the need for CABG or additional PCI prior to discharge. In this study, death occurred in 1.3% of patients, Q-wave MI in 1.4% and repeat revascularization in 4.5%. Half of the patients underwent stenting in this study. Notably, this database did not record myocardial enzymes or the use of GP IIb/IIIa inhibitors. Aronow and colleagues observed outcomes in a cohort of consecutive registry patients undergoing coronary stent placement between 1995 and 1997.³² A total of 373 patients underwent PCI during this time period, with death occurring in 9 patients (2.4%), CABG in 3 (0.8%) and MI in 19 (5.1%, including both QWMI and NQWMI). Repeat diagnostic catheterization was performed in 3.2% of patients and repeat PCI in 0.8%. The SCA&I registry evaluated outcomes in 16,811 patients undergoing either balloon angioplasty (n = 6,121) or stenting (n = 10,690) between July 1996 and December 1998.²⁸ In this observational analysis, 12.9% of patients received a GP IIb/IIIa inhibitor, 87% of patients enrolled in the database underwent PCI between 1997 and 1998, and 60% of the stent patients were enrolled in 1998. Outcomes reported included in-hospital death (occurring at any time during the hospitalization) and eCABG, defined as CABG occurring immediately after PCI. Death occurred in 0.4% of patients and eCABG in 0.5%. Finally, Cohen and others recorded in-laboratory complications in 26,421 patients at 70 different centers undergoing PCI in 1998.1 In-laboratory complications were rare, with death occurring in 0.17%, cardiac arrest in 0.32%, stroke in 0.03%, ventricular fibrillation or tachycardia in 0.94%, abrupt closure in 0.71%, and eCABG in 0.53%. Overall, 72% of patients received stents and 20% received GP IIb/IIIa inhibitors. In addition to published reports of PCI complications, data from unpublished sources can be used to determine outcomes in a more contemporary cohort of patients undergoing PCI.³³ The MQ-Profile (MQ-Pro) Database [Cardinal Information Corporation (CIC), Marlborough, Massachusetts] is maintained by CIC, which sells and distributes software to US acute-care hospitals for the collection of detailed clinical and administrative data. Data from 5,373 PCI procedures performed between July 1, 1998 and June 30, 1999 were obtained from the database using International Classification of Diseases 9th Edition (ICD-9) procedure codes for PCI (36.01, 36.02, 36.05). Demographic, clinical and economic data were collected on each patient using a combination of database retrieval and chart review. In this analysis, death was defined as discharge disposition of “deceased”, MI as the presence of ECG changes consistent with MI (new Q-waves or ST-segment changes) or an increase in CK-MB of at least 2 times the testing facility’s ULN. CABG was identified by the presence of ICD-9 procedure code 36.1 and repeat PCI by either code 36.01, 36.02, or 36.05. Failed PCI was defined by the term “failed PTCA” in chart notes (for patients without a previous history of PCI) and recurrent ischemia documented by ECG changes. Death occurred in 2.0% of patients, MI in 3.1%, CABG in 1.3% and repeat PCI in 5.5%. Translated into a combined endpoint similar to those used in clinical trials, the rate of death/MI/revascularization was 11.9%. Data from these published and unpublished observations of contemporary PCI practice indicate that while in-laboratory ischemic complications are exceedingly rare, in-hospital ischemic complications still occur in a substantial number of patients. Using an approximation of outcomes from these published and unpublished reports, mortality averages 1%, Q-wave MI occurs in 2% of patients, NQWMI in 6%, CABG in 2% and repeat PCI occurs in 3–5% of patients. It is important to underscore that although most deaths following PCI are due to underlying comorbidities (i.e., acute MI, cardiogenic shock, etc.) rather than to the procedure itself, few deaths still occur as a complication of the procedure.^34,35 Extrapolated to the estimated PCI population of 800,000 cases per year, then 8,000 people will die and 64,000 will experience an MI. In addition, approximately 16,000 will require CABG and as many as 40,000 will need a repeat PCI before hospital discharge. Continued on next page