NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Bravata DM, McDonald KM, Gienger AL, et al. Comparative Effectiveness of Percutaneous Coronary Interventions and Coronary Artery Bypass Grafting for Coronary Artery Disease [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2007 Oct. (Comparative Effectiveness Reviews, No. 9.)

Cover of Comparative Effectiveness of Percutaneous Coronary Interventions and Coronary Artery Bypass Grafting for Coronary Artery Disease

Comparative Effectiveness of Percutaneous Coronary Interventions and Coronary Artery Bypass Grafting for Coronary Artery Disease [Internet].

Show details



Since atherosclerosis develops in a patchy, discontinuous fashion within the coronary artery, it is possible to address therapeutically the discrete areas of obstruction that most impede coronary blood flow to the myocardium. The mechanical approaches to coronary revascularization fall broadly into two categories: coronary artery bypass grafting surgery (CABG) and catheter-based percutaneous coronary interventions (PCI). Together, these coronary revascularization techniques are among the most common major medical procedures performed in North America and Europe. In 2004, there were 249,000 CABGs and 664,000 PCI procedures performed in the United States alone.1 There has been considerable interest in the effects of coronary revascularization procedures on patient outcomes, and especially upon the circumstances in which one procedure should be preferred over the other.

Coronary bypass surgery was first described in 1967 by Kolessov, who performed anastomoses of the left internal mammary artery to the left anterior descending artery in patients with a beating heart.2 Later that year, Favaloro at the Cleveland Clinic introduced the most common technique of CABG by constructing a bypass graft using a saphenous vein conduit.3 The initial technique consisted of the anastamosis of a segment of saphenous vein to the aorta and to the coronary artery beyond the most severely narrowed segment, thereby “bypassing” the obstruction. The techniques for coronary artery bypass graft (CABG) surgery have been progressively refined over the past four decades, based on incremental improvements in the operation itself, in cardiac anesthesia, in post-operative care, and in concomitant medical therapy. The most important improvement was the use of arterial grafts as conduits rather than saphenous veins whenever feasible - the left internal mammary artery (LIMA) to the left anterior descending (LAD) in particular is recognized as a superior approach.4 More recently, there has been great interest in returning to operative techniques that avoid the use of the cardiopulmonary bypass machine, or that use a smaller thoracotomy incision rather than a full sternum splitting operation, or both.5 These newer approaches to CABG, which aim to achieve the same level of coronary revascularization as the standard operation but with less trauma and fewer adverse effects, are under active development.

Coronary angioplasty was developed by Gruntzig in 1977 at the University of Zurich as an alternative approach to coronary bypass surgery.6, 7 The use of a balloon-tipped catheter to dilate localized areas of coronary obstruction was a revolutionary paradigm shift in cardiovascular medicine. The technique of angioplasty has also undergone progressive refinement over the past three decades as a result of better catheters, improved imaging, and new adjunctive medical therapy during and after the procedure. The major limitation of angioplasty has been restenosis - the tendency of the dilated coronary segment to constrict during healing, such that blood flow is again impeded several weeks to months after the initial procedure. Restenosis developed after 30% to 40% of balloon angioplasty procedures, typically within six months, leading to a repeat revascularization procedure. Many drugs were tried to prevent restenosis, but none was effective. The key improvement was the development of the coronary stent, an expandable metal mesh tube that buttresses the dilated segment and limits restenosis. Randomized clinical trials (RCTs) showed that bare metal coronary stents reduced significantly the rate of angiographic restenosis and of repeat revascularization procedures. However, these trials also showed that stents had no effect on the rate of death or of acute myocardial infarction compared with balloon angioplasty.8 Contemporary stents now coat the metal meshwork with drugs (e.g., sirolimus, paclitaxel) that reduce cellular proliferation in response to the injury of dilatation. Randomized trials comparing drug-coated stents with bare metal stents showed a further reduction in the rate of repeat revascularization by the drug-coated stent but no reduction in the rates of death or myocardial infarction.9 Reducing restenoses may not reduce death or myocardial infarctions, however, because restenosis is usually a slow process that rarely leads to acute obstruction and serious cardiac events.9 Recent evidence suggests that drug-coated stents may be associated with higher long-term rates of stent thrombosis and late myocardial infarction, especially after dual antiplatelet therapy is discontinued.10

Comparative Efficacy of CABG and PCI

In large clinical series, there are clear differences between the patients selected for PCI and for CABG. Coronary anatomy determines the technical feasibility of PCI11, 12 and CABG,13, 14 and PCI is often infeasible in patients with extensive disease or chronic total occlusions. In large clinical series, most patients with single-vessel disease are treated with PCI, whereas most patients with triple-vessel disease or left main disease are referred for CABG. The comparative effectiveness of bypass surgery and angioplasty is an open question primarily for those patients in which both procedures are technically feasible and whose coronary disease is neither too limited nor too extensive. This middle segment of the spectrum of CAD includes single-vessel disease of the proximal LAD, most forms of double-vessel CAD, and less extensive forms of triple-vessel CAD. This group of patients provides real treatment options for clinicians and the equipoise needed for randomization in clinical trials. Most comparative studies of angioplasty and surgery have been conducted in this segment of the patient population with CAD.

The effects of coronary revascularization can be assessed using many outcome measures. This comparative effectiveness review focuses on patient-centered outcomes rather than laboratory measures, since clinical outcomes evident to the patient are the most pertinent to the choice between angioplasty and surgery. The most important outcomes are the serious, irreversible clinical complications: death, myocardial infarction, and stroke. Other important outcomes are symptoms (especially angina), functional capacity, quality of life, employment, cost, and cognitive function. Finally, the durability of the revascularization procedure is important to patients - will a repeat revascularization procedure be needed, and if so, which one and when?

It is important to emphasize that these outcomes span several dimensions and the effect of a therapy need not be consistent across them. Coronary stents, as discussed above, reduce the need for repeat revascularization procedures, but have not been shown to reduce myocardial infarction or death. Thus, the comparative effectiveness of CABG and PCI must be assessed separately for the various clinical outcomes of interest.

Systematic overviews of trials comparing CABG and PCI have been conducted previously.1518 Additional data have become available since their publication, both new trials and additional reports with long-term outcomes from included trials. The previous systematic overviews generally focused on a limited set of outcomes and did not address efficacy in subgroups.

Variations in Effectiveness

One of the major issues in applying evidence from the medical literature to medical decisions is the question of generalizability: Namely, are the results of formal randomized studies a reliable guide to the likely results for a given patient treated in the local hospital? The reason this question is pertinent is that the outcomes of treating a disease may well vary according to characteristics of the patient, the details of the therapy itself, and the setting in which therapy is provided. The key questions we address in this report are predicated upon the hypothesis that the comparative effectiveness of bypass surgery and angioplasty may well vary according to these factors.

It is important to distinguish the different circumstances under which the effectiveness of therapy may vary according to patient characteristics. The first possibility is that the therapy is generally effective for certain well-defined groups of patients, but is ineffective or even harmful for other well-defined groups of patients. In the case of coronary revascularization, for instance, neither angioplasty nor surgery is likely to improve the outcomes of a patient with a single coronary occlusion that has led to complete myocardial infarction of the distal myocardium. Groups of patients who have no realistic expectation of benefit from treatment are generally not enrolled in clinical studies, as it would be unethical to expose them to a risky procedure. The more common question is whether, among patients who have some reasonable expectation of benefit from therapy, the benefit varies in a predictable fashion and with a magnitude great enough that the choice of therapy is affected. Here it is important to distinguish between variations in the “relative risk reduction” and “absolute risk differences” as measures of the magnitude of benefit from therapy. Most clinical studies report results as relative risk reductions (or similar measures such as odds ratios or hazard ratios). A therapy that reduces relative risk by 25 percent across all subgroups of patients is conventionally said to show a consistent therapeutic effect across subgroups.19 Some degree of variation in relative risk reduction is common in clinical trials, but statistically significant variation in relative risk reduction across subgroups (assessed using the treatment-by-covariate interaction test) is distinctly unusual. For instance, the time delay in administering fibrinolytic therapy for acute MI significantly affects relative risk reductions achieved (p for interaction = 0.002), but fibrinolytic therapy yields similar relative risk reductions in patients with or without diabetes, and in men and women.20 Nevertheless, patient subgroups may differ in the “absolute risk difference” from therapy despite equivalent “relative risk reductions.” While both measures have advantages, absolute risk differences are more pertinent to clinical decisions since they better reflect the increased chance of a better outcome from the choice of one therapy over the alternative therapy.

Throughout this report we present absolute risk differences whenever possible, since these risks are of the greatest interest to patients. (We did calculate both relative and absolute risk difference between the procedures and found no significant differences between these metrics).

Scope and Key Questions

The purpose of this report is to answer the following key questions:

Key Question 1a. In patients with ischemic heart disease, and angiographically-proven single or multi-vessel disease, what is the comparative effectiveness of PCI compared to CABG, in reducing the occurrence of adverse objective outcomes and improving subjective outcomes?

Key Question 1b. Over what period of time are the comparative benefits of PCI and CABG sustained?

For the purposes of answering these key questions, we provide the following definitions: PCI includes percutaneous coronary angioplasty (PTCA), with or without drug-eluting stents (DES) or bare metal stents (BMS). CABG includes traditional on-pump or off-pump bypass procedures; on-pump or off-pump minimally invasive procedures; and CABG with transmyocardial revascularization (TMR). Long-term and short-term objective outcomes refer to outcomes that impact patients' health, including, but not limited to, peri-procedural death or complications, non-fatal myocardial infarctions, congestive heart failure, stroke, nosocomial infections, respiratory failure or other pulmonary complications, acute or chronic renal failure, cardiac arrhythmias, and long-term survival and event-free survival (major adverse cardiac events). Long-term and short-term subjective outcomes refer to outcomes that impact patients' perceived quality of life, functional health status, or general health status. Subjective outcomes may include, but are not limited to, freedom from angina, functional angina classification, cognitive impairment, productivity, and functional capacity. Intermediate outcomes may also be considered if available evidence clearly links such outcomes to long-term or short-term outcomes. These include completeness of revascularization, target lesion revascularization, restenosis following PCI, CABG graft closure, the need for secondary revascularization procedures, readmission within 30 days, readmission within 6 months, and post-procedure discharge to rehabilitation facilities.

Key Question 2. Is there evidence that the comparative effectiveness of PCI and CABG varies based on:


Age, sex, race, or other demographic risk factors?


Coronary disease risk factors, diabetes, or other comorbid disease?


Angiographic-specific factors including, but not limited to, the number of diseased vessels amenable to bypass or stenting, vessel territory of stenoses (e.g., left main or anterior coronary arteries, right coronary artery, circumflex coronary artery), diffuse vs. focal stenoses, left ventricular function, or prior revascularization procedures?


CABG-specific factors including, but not limited to, cardiopulmonary bypass mode (normothermic vs. hypothermic), type of cardioplegia used (blood vs. crystalloid), or use of saphenous vein grafts, single or bilateral internal mammary artery grafts, or other types of bypass grafts?


Clinical presentation (e.g., stable angina or unstable angina, based on NYHA functional class I–IV, acute coronary syndrome, cardiogenic shock, acute myocardial infarction with or without ST elevation, or silent ischemia)?


Adjunctive medical therapies, such as short-term intravenous or oral antiplatelet drugs, or long-term use of oral antiplatelet drugs?


Process characteristics such as provider volume, hospital volume, and setting (e.g., academic vs. community)?


Prior PCI or CABG revascularization procedures?

PubReader format: click here to try


  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (1.2M)

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...