Abstract 1292: Comparative Effectiveness of Carotid Arterial Stents versus Endarterectomy: An Instrumental Variables Approach
Background: It is uncertain whether carotid arterial stents (CAS) have comparable outcomes to carotid endarterectomy (CEA). Medicare’s coverage policy for CAS introduced substantial geographic variation in CAS use, thus outcomes comparisons across geographic areas can provide insight into the comparative effectiveness of CAS in non-experimental settings.
Methods: Patients are selected for CAS non-randomly, thus observational comparisons of CAS vs. CEA may be biased by selection on unobserved factors (e.g., disease severity, comorbidity) that affect outcomes. To address this, we used a nonparametric instrumental variable approach by which regional variation in CAS utilization was assumed (and confirmed) to be uncorrelated with patient-level factors. We first determined the % of 2005–2006 carotid revascularizations that were CAS in each of the 306 U.S. Hospital Referral Regions (HRRs). We then matched HRRs with similar demographic, geographic, and carotid revascularization patient characteristics, but different CAS usage rates. We next matched Medicare carotid revascularization patients based on multiple demographic and clinical characteristics across each matched pair of HRRs. Finally, we estimated an instrumental variable probit regression model predicting clinical outcomes, in which each HRR’s CAS uptake rate “instrumented” for the receipt of CAS versus CEA.
Results: We matched 2,265 pairs of carotid revascularization patients across HRRs with high (mean proportion of carotid procedures that were CAS=19%) versus low (mean CAS proportion=7%) CAS adoption. Instrumental variable analyses indicated CAS were associated with no difference in 90-day mortality (absolute risk difference [ARD]= −0.3%, 95% confidence interval[CI] -1.0% to 0.3%) or 270-day mortality (ARD=0.2%, 95% CI −0.7% to 1.1%), but with increased 90-day combined adverse events of peri-procedural myocardial infarction, stroke, or death (ARD=1.1%, 95% CI 0.0% to 2.3%) and 270-day combined events (ARD=2.3%, 95% CI 1.0% to 3.7%).
Conclusions: In this observational study comparing CAS and CEA outcomes using an instrumental variable approach, CAS were associated with no mortality difference, but higher combined rates of major adverse clinical events, compared to CEA.