CAS vs CEA Outcomes in Medicare Population Align With Clinical Trial Data

One of the major talking points in the debate over endarterectomy (CEA) vs stenting (CAS) in patients with carotid stenosis has been the striking differences between outcomes in large US clinical trials and real-world observation. But a new study linking Medicare data to two large registries suggests that accounting for disparate patient factors such as age, disease severity, and comorbidities, as well as procedural and operator characteristics, results in a leveling of the playing field between the two interventions in the Medicare population.  

The Take Home. CAS vs CEA Outcomes in Medicare Population Align with Clinical Trial Data

“It does basically show that both procedures are equivalent when performed by qualified providers under the national coverage determination,” said lead author Jessica J. Jalbert, PhD (Brigham and Women’s Hospital, Boston, MA), in an interview with TCTMD.

Despite considerable data from clinical trials and registries, as well as an FDA advisory panel vote in 2011 in support of expansion of CAS coverage to include those at standard surgical risk—in addition to the previously approved indication in high-surgical-risk patients—no changes have been made to the national coverage determination (NCD) issued by the Centers for Medicare & Medicaid Services (CMS) in many years.

The pivotal randomized CREST trial and ACT I both have shown that CAS holds up well over the long term against CEA in symptomatic and asymptomatic patients. But CREST had safety trade-offs due to more minor strokes in the CAS group and more MIs and cranial nerve injuries in the CEA group.

Revelations in the Details

For the new study, published April 26, 2016, ahead of print in Circulation: Cardiovascular Quality and Outcomes, Jalbert and colleagues looked at Medicare patients treated with CAS or CEA whose outcomes were tracked in the Vascular Quality Initiative Registry (n = 5,254) and the National Cardiovascular Data Registry's defunct CARE registry (n = 4,055). Authors also mined the American Hospital Association's Annual Survey Database for information on hospital organizational structure and size, and the American Medical Association’s Physician Masterfile for information on physician demographics and practice data.

Linking all of the various registries allowed the researchers to combine detailed sociodemographic and disease severity information, procedural characteristics, comorbidities, preprocedural medication use, healthcare resource use, operator characteristics, and hospital characteristics.

Compared with those undergoing CEA, patients in the CAS group were sicker and at higher risk, as would be expected under the NCD.

Unadjusted hazard ratios for the composite endpoint (mortality, stroke or TIA, and perioprocedural MI) and all of its individual components, with the exception of periprocedural MI, favored CEA over CAS. After adjustment for patient-level characteristics, only mortality remained elevated. However, adjustment for both patient-level and provider-level characteristics removed the mortality difference.

Table. CAS vs CEA Outcomes in Medicare Population Align with Clinical Trial Data

The study also showed two important trends among subgroups, suggesting a better outcome with CEA in patients age 80 years and over as well as in those who were symptomatic. Although not statistically significant, the age-related signal is one that also was seen in the CREST trial.

“Clinical trials are the gold standard for evaluating treatment interventions, but the patients they enroll tend to be younger and healthier than the average Medicare patient,” Jalbert said. “What we can conclude from our study is that the conclusions from the landmark trials also seem to apply to the real-world Medicare population when the procedures are performed by qualified providers.”

A major problem when it comes to evaluating Medicare data is the lack of detailed information available in the administrative database, noted Jay S. Giri, MD, MPH (Hospital of the University of Pennsylvania, Philadelphia, PA), in an interview with TCTMD. By linking to the registries and conducting the adjustments, he said, the investigators gained access to over 100 very specific data elements that normally would not be known.

“So now, you have a much richer picture of who these patients are,” Giri commented. “You still are not going to get to the point where it is as good as flipping a coin in a randomized trial, but it’s still so much better than simply using the administrative dataset alone.”

To TCTMD, Giri called the paper “rigorously methodological” and “unbelievably outstanding work.” He added that while differences in operator characteristics account for some of the differences in outcomes, the study strengthens the assertion that individual patient characteristics are of paramount importance.

Where Do We Go From Here?

Giri said the study corrects some of the flaws in past observational studies and is “a big step in providing information,” specifically the type of information that CMS has said it wants to see before it is willing to expand coverage beyond the current restrictions on the NCD.

“It shows in the Medicare population that these procedures are not so much competitive procedures but complementary procedures and that we need to tailor our decision making to the patient’s unique circumstances,” he observed. “It provides some of the information that [CMS] has asked for, but it doesn’t address all of their concerns.” Chief among those concerns is whether either procedure is really needed or whether best medical therapy is sufficient, a point echoed by Jalbert.

“Medical management is the elephant in the room,” she said, adding that future studies should compare the effectiveness of CAS, CEA, and medical management. Indeed, the much anticipated CREST 2 study, currently enrolling patients, is actually two parallel multicenter, randomized controlled trials of carotid revascularization—one with stenting and the other endarterectomy—and intensive medical management versus medical management alone in patients with asymptomatic high-grade carotid stenosis.



  • The study was funded by the Agency for Healthcare Research and Quality and CMS. 
  • Jalbert reports no relevant conflicts of interest. 
  • Giri reports grants to his institution from St. Jude Medical. 

Related Stories:

  • Jalbert JJ, Nguyen LL, Gerhard-Herman MD, et al. Comparative effectiveness of carotid artery stenting versus carotid endarterectomy among Medicare beneficiaries. Circ Cardiovasc Qual Outcomes. 2016;Epub ahead of print.