Principal Findings:

A total of 18,867 patients were enrolled at 60 centers, 14,149 to PCI at sites without on-site cardiac surgery, and 4,718 to sites with on-site cardiac surgery (72 were unassigned). The median procedure volume at these institutions was 150. Baseline characteristics were similar between the two arms. About 80% of the patients were white, 39% had diabetes mellitus, 43% had a prior history of myocardial infarction, and 31% had a history of prior PCI. Of these patients, 64% were undergoing PCI for acute coronary syndrome; 23% of the procedures were either urgent or emergent. There was an even distribution between patients with one-, two-, and three-vessel disease (roughly 28-34%); left main PCI was undertaken in 3% of the patients and saphenous vein graft PCI in about 9%. Per study design, there were more lab visits for the index PCI in the patients randomized to PCI at sites with on-site cardiac surgery (1.73 vs. 1.28, p < 0.0001). PCI failure was higher at sites without on-site cardiac surgery, both at a patient level (3.4% vs. 2.5%, p = 0.007), and at a lesion level (6.6% vs. 5.9%, p = 0.04).

The primary endpoint of all-cause mortality at 6 weeks was noninferior between sites without and with on-site cardiac surgery (0.9% vs. 1.0%, p for noninferiority p = 0.004). At 9 months, rates of major adverse cardiac events (MACE) (death, myocardial infarction, target vessel revascularization [TVR]) were noninferior (12.1% vs. 11.2%, p for noninferiority = 0.05). When coronary artery bypass grafting (CABG) was not included in the definition of MACE, sites without on-site cardiac surgery had inferior MACE rates, as compared with sites with on-site cardiac surgery (11.9% vs. 10.5%, p < 0.0001). Similar inferior results were noted on per-protocol analysis at 9 months (12.0% vs. 10.4%, p < 0.001). Individual endpoints at 9 months such as death (3.2% vs. 3.2%) and myocardial infarction (3.1% vs. 3.1%) were similar; rates of TVR (6.5% vs. 5.4%, p = 0.001) were higher at sites without on-site cardiac surgery.  

Other endpoints such as bleeding (3.4% vs. 3.2%), vascular repair (0.4% vs. 0.4%), stroke (0.3% vs. 0.2%), and renal failure (0.5% vs. 0.4%) were similar (p > 0.05 for all). Rates of all CABG (0.6% vs. 1.5%, p < 0.0001) and emergency CABG (0.1% vs. 0.2%, p = 0.05) were lower at sites without on-site cardiac surgery, while rates of unplanned diagnostic catheterization (4.3% vs. 3.2%, p < 0.001) and unplanned emergent PCI (0.2% vs. 0.1%) were higher at sites without on-site cardiac surgery.

Cost-effectiveness: Estimated medical costs per patient were similar between sites without and with on-site cardiac surgery during index hospitalization ($23,343 vs. $20,372, p = 0.16), but higher during the follow-up period ($5,794 vs. $5,040, p = 0.0004). The latter was especially true for low-volume hospitals ($6,126 vs. $4,918, p = 0.0002), but not at high-volume hospitals ($5,529 vs. $5,211, p = 0.18). Cumulative costs were similar at 9 months between sites without and with on-site cardiac surgery ($29,136 vs. $25,412, p = 0.09), with no differences between high- and low-volume hospitals.