March 27, 2012

March 27, 2012 (Updated March 29, 2012 ) (Chicago, Illinois) — Adjusted four-year all-cause mortality was about 20% lower after CABG than after PCI in an observational study of about 190 000 patients in the US who underwent nonemergent revascularization for two- or three-vessel coronary disease [1]. Mortality at one year was similar for the two procedures.

Dr William S Weintraub

Based on patients >65 years of age taken from two huge society-sponsored databases, the analysis used several methods to control for risk levels and other influences on outcomes. It found the long-term advantage for CABG was independent of age, sex, diabetes, renal function, and lung disease and was evident "even among patients whose propensity scores were most consistent with selection for PCI," write the authors, led by Dr William S Weintraub (Christiana Care Health System, Newark, DE), in a report published today in the New England Journal of Medicine.

"That was one of the major surprises in the study. We really thought we'd find subsets of patients that would benefit from surgery and other subsets that would benefit from percutaneous intervention," surgeon Dr Fred H Edwards (University of Florida, Jacksonville), co–principal investigator of the trial with Weintraub, told heartwire . "And much to our surprise, all of the subsets showed a survival advantage for surgery."

The study, called the ACCF and STS Database Collaboration on the Comparative Effectiveness of Revascularization Strategies (ASCERT), combined data from the American College of Cardiology Foundation (ACCF) CathPCI registry, the Society of Thoracic Surgeons (STS) CABG database, and Medicare claims. It was published today to coincide with Weintraub's presentation of the study at the American College of Cardiology 2012 Scientific Sessions.

It was presented in preliminary form in January at the Society of Thoracic Surgeons (STS) 2012 Annual Meeting, as covered by heartwire at the time.

About a dozen randomized trials plus observational studies and other analyses have compared CABG with whatever form of coronary angioplasty there were adequate data for at the time (mainly balloon angioplasty and stenting with either bare-metal or drug-eluting stents, depending on the era), the authors observe. The current study is the largest and national in scope, they write.

You tried multiple ways to eliminate residual confounding. . . . But I think it's fair to say that, at least in this data set, it's probably impossible to do that.

On the other hand, Weintraub et al acknowledge--and an accompanying editorial [2] emphasizes, and virtually everyone connected to or speaking on ASCERT at the ACC sessions observed--that the study couldn't adjust for all potential confounders and so has limitations.

In fact, that was the focus of the accompanying editorial. "The validity of these findings rests largely on a determination of whether adequate control for confounding was possible," writes interventionalist Dr Laura Mauri (Brigham and Women's Hospital, Boston, MA). "Even with the findings adjusted for propensity score, the authors state their conclusions cautiously."

Dr Alice Jacobs

And the subject pervaded the panel discussion following Weintraub's presentation. "Your analysis has certainly set the standard . . . for conducting these analyses, as you tried multiple ways to eliminate residual confounding and selection bias that might be present. But I think it's fair to say that, at least in this data set, it's probably impossible to do that," said panelist Dr Alice Jacobs (Boston University School of Medicine, Boston, MA), an interventionalist.

That's largely because much of what goes into clinical decision making can't easily be measured: "Factors that a clinician, interventional cardiologist, or a surgeon integrates at the bedside," Jacobs said. She cited as examples variables such as patient preferences, the angiogram, bleeding risk, and compliance with medical therapy, "which were not collected in this data base."

According to ASCERT coauthor and interventionalist Dr Kirk N Garratt (Lenox Hill Heart and Vascular Institute, New York, NY), such unmeasured confounders "could easily explain this differential we saw [between PCI and CABG] on mortality. But we have no way of knowing."

Nuances and Complex Reasons

If you can get me out of the hospital tomorrow and I can go ahead and take my cruise to the Bahamas with my family next week, I'm in.

Speaking on behalf of the Society for Cardiac Angiography and Interventions (SCAI), Garratt observed for heartwire that "we choose patients for one treatment over another for complex reasons, and the reasons aren't always the same." Data are collected on some things that "seem to make sense," like age, sex, stroke history, and renal function. But, agreeing with Jacobs, Garratt said, "The things that really drive a clinician to recommend surgery are often far more nuanced."

For example, he said, an older person who decides on PCI might say, "Living forever is not my agenda. I want the time left to be about quality. If you can get me out of the hospital tomorrow and I can go ahead and take my cruise to the Bahamas with my family next week, I'm in."

What the Analysis Found

Of those getting PCI in the ASCERT analysis, 78% received drug-eluting stents and 16% bare-metal stents; another 6% had the procedure without stenting; their follow-up time averaged 2.63 years, compared with 2.82 years for those undergoing CABG.

Adjusted* Mortality Hazard Ratios, CABG vs PCI, by Follow-up Time

Interval HR (95% CI)
30 d 1.72 (1.52–1.89)
1 y 0.95 (0.90–1.00)
2 y 0.79 (0.76–0.83)
4 y 0.79 (0.76–0.82)

*Using inverse probability weighting based on propensity scores

"Here we have a national confirmation of observational studies and several randomized studies showing that long-term survival for surgery is better than that for percutaneous intervention" in patients with stable multivessel coronary disease, Edwards said to heartwire , speaking on behalf of the STS. Similar results with CABG and PCI in trials have led to a frequent preference for the less invasive option. "I think we are seeing the waning of a controversy there; this is as confirmatory as it can be."

But Edwards, Garratt, Weintraub, and others at the sessions seemed to agree that ASCERT isn't primarily about whether CABG or PCI is better than the other in patients with multivessel CAD.

I think we are seeing the waning of a controversy there, this is as confirmatory as it can be.

"One of the major messages to take away from the study is that cardiologists and surgeons need to be talking to each other more," according to Edwards. If CABG provides better long-term survival than PCI, "that information should be presented to the patient and be used in a dialog between the surgeon and the cardiologist."

The team approach brings "universal acclaim," he said. "But how much it's practiced in the real world I think remains to be seen. Hopefully, ASCERT will prompt more than just lip service."

Observation vs Randomization

"Observational registries" have their strengths, Mauri writes in the editorial, "but there is no substitute for randomized trials to eliminate selection bias between treatments. The two approaches are thus complementary." Priority must be given to randomized trials, however, "on the most salient questions regarding treatment strategy."

At his presentation of ASCERT, Weintraub said, "I do believe that for comparative effectiveness to really reach its potential, it will require both randomized and observational studies. They both do have critical roles to play."

But his leanings were apparent. "ASCERT really pushes the boundaries of comparative-effectiveness analysis. Observation studies can provide real-world outcomes with greater generalizability than randomized trials."

He added during the panel discussion, "The problem of generalizability of randomized trials is not a small one. People complain as much about the results of randomized trials as they do about observational studies. We've got to look at all of these together, and we're never going to overcome all the problems. It just doesn't work that way."

The study is supported by a grant from the National Heart, Lung, and Blood Institute . Weintraub reports no conflicts of interest. Disclosures for the coauthors are listed in the paper. Mauri reports consulting for Medtronic and Cordis; providing expert testimony to Johnson & Johnson; and receiving institutional grants from Abbot Vascular, Boston Scientific, Medtronic Vascular, Cordis, and Eli Lilly/Daiichi Sankyo.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.