Stent vs. Bypass: The Gold Standard Study?

The New England Journal of Medicine recently published the results of the SYNTAX trial (here), a multi-center, multi-country randomized clinical trial comparing percutaneous coronary intervention (PCI) with drug-eluting stents vs. coronary-artery bypass grafting (CABG). The study seems to suggest that more patients should be undergoing CABG since there was more revascularization in the PCI group (death rates at 12 months were the same, although stroke in the CABG group was higher). Let's take a closer look using the causal model of heart attacks discussed in an earlier post (here).

Consider first the problems of designing a true randomized clinical trial comparing a procedure (PCI) with an invasive surgery (CABG). You simply cannot randomly assign patients to either treatment condition. Many patients would prefer to avoid a seriously invasive procedure such as CABG and a few more patients would prefer to avoid any procedure at all. And, that's exactly what happened. From the hierarchy diagram above (click on the diagram to see an enlarged view) 1262 patients were found ineligible, the majority because they had a treatment preference (probably PCI). Another 1275 were eligible only to be placed in a parallel, nested registry where 1077 enrolled in CABG and 198 enrolled in PCI. In other words 2537 patients were not randomized compared to 1800 that were randomized.

As a practical matter, self-selection cannot be avoided. As a consequence, study results really are only applicable to people that meet criteria (three-vessel or left main coronary artery disease) and had no treatment preference. It would be nice to know the rates of revascularization, mortality and stroke in the groups that were not randomized.

The study has a number of other limitations reported by the authors but no mention is made of the 12-month histories of cardiac rehabilitation in any of the groups. If there was no cardiac rehabilitation (e.g., substantially increased aerobic exercise), the rate of revascularization might be explained by that alone. If, for example, all 1800 patients were sedentary in the first 12 months, CABG may produce better outcomes. What cardiac rehabilitation, weight control or smoking cessation programs the PCI group went on (if any) is unclear. The issue isn't discussed in either the paper or the supplementary material, although my guess is that the standards of cardiac rehabilitation are very different in the countries that participated in the study.

On the basis of this study (and the other existing clinical trials reviewed in the paper), should treatment protocols change?