Disparities Between Patients Given Carotid Stenting, Surgery Deter True Comparisons

Although it seems that efforts are constantly being made to compare the efficacy of carotid artery stenting (CAS) and carotid endarterectomy (CEA), a new analysis of registry data published online January 11, 2011, ahead of print in Circulation: Cardiovascular Interventions suggests that extreme clinical disparities between patients referred for either procedure make valid comparisons between the 2 treatments using observational data difficult at best.

Ryan B. Longmore, DO, of Saint Luke's Mid America Heart Institute (Kansas City, MO), and colleagues tapped into the National Cardiovascular Data Registry-Carotid Artery Revascularization and Endarterectomy (CARE) registry to compare the clinical profiles of 12,701 patients treated for advanced carotid artery disease: 8,069 were referred for CAS and 4,632 were treated with CEA.

At baseline CAS patients were more likely to be non-Caucasian and to have multiple medical comorbidities including dialysis, previous neck irradiation or surgery, valvular disease, prior carotid revascularization, and history of neurological events. Multivariable analysis found 25 patient characteristics that were independent predictors of CAS referral. Further calculations—using a propensity score that reflected probability of CAS—showed persistent imbalance for 7 factors, all of which were more common for CAS patients:

• Prior CAS
• Prior CEA
• Prior MI
• Increased angina burden
• Ischemic heart disease
• CAS restenosis
• CEA restenosis

When patients were divided into quintiles based on propensity to undergo CAS, those least likely to be referred for the procedure had a 30-day mortality of 0.49%. Those in the highest 2 quintiles, indicating increased likelihood of CAS referral, had mortality rates of 1.51% and 1.01% (P for trend = 0.0006). CEA patients comprised only 14% of the upper 2 quintiles. Mortality did not appear to interact with treatment type and propensity quintile, suggesting that the “association between the probability of undergoing CAS and adverse outcome was independent of the method of revascularization, and more likely due to existing comorbidities,” the paper notes.

Apples to Oranges?

“Collectively, these findings call into question earlier reports comparing the outcomes of CAS and CEA and underscore the challenges of comparative effectiveness studies in the management of carotid disease,” Dr. Longmore and colleagues note.

In a telephone interview with TCTMD, William A. Gray, MD, of Columbia University Medical Center (New York, NY), framed the paper as an “important” rebuke to several retrospective analyses of carotid revascularization published in the past year or so. His advice to clinicians was clear: take retrospective, observational analyses “with a block of salt, not a grain.”

“I think it once and for all puts to rest the issue of whether it’s reasonable to do retrospective analyses on patients in this era, given the fundamental differences in both patient types and selection biases,” he continued. “There are reasons [the two patient groups are] different,” some having to do with regulatory and reimbursement issues, he added.

These baseline differences are not necessarily a sign of trouble, the investigators stress. “A great deal of the clinical characteristics of patients referred for CAS versus CEA probably can be explained by physician adherence to guidelines and recommendations regarding patient selection,” they note, pointing to a 2007 consensus document (Bates ER, et al. J Am Coll Cardiol. 2007;49:126-170).

Caution Urged for Researchers, Clinicians

In an e-mail communication, Dr. Longmore told TCTMD that he believed most investigators are keen on “using the most appropriate statistical methods and will benefit from this demonstration of the challenges inherent to comparing outcomes of these disparate groups.

“That being said, there is great pressure to compare these 2 procedures with observational data, and we know that many will conduct these analyses,” he continued, adding that many datasets lack the clinical richness of CARE and therefore make it hard for researchers to recognize disparities. “Our concern is that even if the best methods are used, such as propensity matching or instrumental variable analyses, that they will apply only to a small subset of the patients eligible for either procedure and not be generalizable to the entire spectrum of patients needing carotid revascularization.”

Dr. Gray similarly noted, “You can adjust for covariates that you know are confounders, but there may be covariates that are not known, ones that you’ve not measured, that you can’t adjust for.”

For clinicians, Dr. Longmore advised, it is “vital” to look closely at the methods of any given study comparing CEA and CAS before fully accepting their results. “In particular, studies utilizing simple regression modeling with a limited number of variables would be unlikely to adjust for the vast clinical differences that we have highlighted in our study.”

“[E]ven with the use of the most advanced statistical methods available, measured and unmeasured differences will be difficult to balance when attempting to compare outcomes in this population,” Dr. Longmore cautioned. More valid comparisons may be possible if reimbursement changes and relatively healthier patients begin receiving CAS, he noted.

 

---

Source:
Longmore RB, Yeh RW, Kennedy KF, et al. Clinical referral patterns for carotid artery stenting versus carotid endarterectomy: Results from the Carotid Artery Revascularization and Endarterectomy registry. Circ Cardiovasc Interv. 2011;Epub ahead of print.

 

 

Related Stories:

• Technical, Anatomic Factors Could Aid Patient Selection for Carotid Stenting
• Pooled Analysis Identifies Predictors of Poor Outcomes After Carotid Stenting
• Wide Variations Seen in National Carotid Revascularization Patterns