Wide Variability in CAS Outcomes Across Hospitals Not Tied to Procedural Volume

Risk-standardized rates of in-hospital stroke or death after carotid artery stenting (CAS) vary up to 4-fold among US hospitals, according to a study published in the May 2015 issue of JACC: Cardiovascular Interventions. Notably, hospital and operator volumes did not have an effect on outcomes. 

Next Step:   Wide Variability in CAS Outcomes Across Hospitals Not Tied to Procedural Volume

Robert W. Yeh, MD, MSc, of Massachusetts General Hospital (Boston, MA), and colleagues analyzed hospital-level data from 188 hospitals that contributed at least 5 CAS procedures to the National Cardiovascular Data Registry (NCDR) Carotid Artery Endarterectomy and Revascularization (CARE) Registry (now the PVI Registry) from 2005 through 2013. In all, 19,381 CAS procedures were examined.

Overall incidence of in-hospital stroke/death was 2.4%, with rates ranging from 0% to 18.8% among hospitals. When hospitals were divided into tertiles based on unadjusted outcomes, mean rates were 0%, 1.6%, and 4.5% for institutions with low, average, and high risk, respectively.

After adjustment for case mix using a validated CAS risk model, mean in-hospital stroke/death rates ranged from 1.2% to 4.7%. Moreover, more than one-third of hospitals (n = 63) shifted tertile positions; twenty-five moved from the low to the average tertile, 6 from the average to the high tertile, 25 from the average to the low tertile, and 7 from the high to the average tertile.

The researchers calculated that if the same patient were to be treated at 2 random hospitals in the registry, there would be an average 50% difference in the odds of experiencing an in-hospital death/stroke (OR 1.51; 95% CI 1.28-1.71).

Results Contradict Earlier Evidence

Contrary to previous Medicare and database studies, the current study found that adjusted risk of in-hospital stroke/death was not negatively impacted by low operator volume (P = .15), although low hospital volume tended to predict worse outcomes (P = .09).

Because the NCDR registry “contains much richer clinical and periprocedural information than the Medicare administrative database… it is a potentially better resource for addressing the relationship between volumes as an independent predictor of outcomes,” Jay Giri, MD, MPH, of the Hospital of the University of Pennsylvania (Philadelphia, PA), told TCTMD in an email.

The study authors suggest that advances, such as increased use of embolic protection and improved patient selection, in the last several years may have mitigated the impact of low procedural volume. Also, hospitals participating in the CARE Registry’s quality assessment and improvement program may have more comprehensive patient-safety measures or higher credentialing standards for operators, they say.

Many Sources of Variability

Dr. Giri pointed out that the variations in unadjusted event rates and in adjusted outcomes have different explanations.

When dealing with the unadjusted rates, “these cases were elective and most were performed in asymptomatic patients,” he stated, adding that in such instances, harm was generally predictable. “Hence, the study reveals that there is as an opportunity to improve patient selection for CAS through [broader] formalized preprocedure risk assessment.” 

Variability after adjustment, meanwhile, “implies that there are additional characteristics not identified in the original risk model that place patients at risk for adverse events following CAS,” Dr. Giri said. Possibilities include inconsistencies in periprocedural medical management and “ongoing wide differences in patient selection,” he explained, adding that factors such as rejection of the patient for CEA and the operator’s specialty or level of training could increase the risk of CAS in individual cases.

How to Engender Consistency

Dr. Yeh and colleagues encourage clinicians to use the metric utilized for risk-standardizing outcomes in this study to compare their hospital’s results with those of others. “[T]hrough the identification of best practices, [it] has the potential to engender policies aimed at quality improvement,” they suggest. Uniform CAS-operator credentialing requirements or “additional procedural metrics, such as embolic protection device use, procedural times, and medication use (eg, thienopyridine preloading [and] statins)” could prove helpful, they say.

Dr. Giri added that “[s]tandardized preprocedural risk and benefit assessments might help,” as might “evaluation of procedural and postprocedural care best practices at the hospital level.” Another possible solution—though one less likely to be adopted— is preferential performance of CAS at high-performing “centers of excellence,” he said.

A broader question is whether, given more widespread use of modern medical therapies, “80% of domestic carotid revascularizations should be performed in asymptomatic patients, as is the case currently,” Dr. Giri concluded.


Hawkins BM, Kennedy KF, Aronow HD, et al. Hospital variation in carotid stenting outcomes. J Am Coll Cardiol Intv. 2015;8:858-863.


  • The study was supported by the American College of Cardiology Foundation’s NCDR. 
  • Dr. Yeh reports serving as an investigator at the Harvard Clinical Research Institute. 
  • Dr. Giri reports no relevant conflicts of interest. 

Related Stories:

We Recommend