Level 1: Likely reliable evidence

A systematic review of 23 randomized trials compared coronary artery bypass grafting (CABG) and percutaneous coronary intervention (PCI) with or without stenting in 9,963 patients (Ann Intern Med. 2007;147:703-716). PCI included balloon angioplasty (used mostly in trials from 1987-1993) and stents (used mostly in trials from 1994-2002); only one small trial used drug-eluting stents. Exclusion criteria in most trials included left main coronary artery disease, single-vessel disease (other than proximal left anterior descending coronary artery), prior CABG, and poor left ventricular function. Follow-up ranged from six months to 13 years.

Twenty-one trials were considered high quality, with clear randomization methods, intention-to-treat analyses, and low dropout rates. Study patients had a mean age of 61 years (few trials enrolled patients older than 75 years). They included 27% women and 20% with diabetes; 50% had hypertension, 50% had hyperlipidemia, and 40% had previous MI.

Comparing CABG vs. PCI, PCI was associated with lower rate of stroke within 30 days of the procedure (1.2% vs. 0.6%). There were no differences in overall survival at one or five years. At five years, the rates of overall survival were 90.7% with CABG vs. 89.7% with PCI in analysis of 12 trials; results varied with year of trial. Five-year survival favored CABG by 2.1% in balloon-era trials (P=.04), but there was no significant difference in stent-era trials. There was no difference in overall survival in subgroup analysis of patients with diabetes, based on seven trials. More CABG patients were without angina at one year (84% vs. 75%) and five years (84% vs. 79%). CABG was also associated with greater freedom from repeat vascularization at one year (96.2% vs. 73.5%) and five years (90.2% vs. 53.9% for balloon trials [59.9% for stent trials]).