Risk-adjusted 30-day outcomes of carotid stenting and endarterectomy: Results from the SVS Vascular Registry
Article Outline
Objective
As the first operational societal registry of carotid procedures, the Outcomes Committee of the Society for Vascular Surgery (SVS) developed the Vascular Registry (VR) in response to the Centers for Medicare and Medicaid Services' (CMS) National Coverage Decision on carotid artery stenting (CAS). Although CMS requires data submission only on CAS, the VR collects similar data on carotid endarterectomy (CEA) to allow comparison of outcomes, as well as potential for expansion to other procedures.
Methods
SVS-VR on-line provider-reported data include baseline through follow-up visits to better understand long-term risks and benefits associated with CAS and CEA. The primary outcomes are combined death, stroke, and myocardial infarction (MI). An independent data coordinating center maintains the database, which is Health Insurance Portability and Accountability Act (HIPAA)-compliant and auditable.
Results
As of December 26, 2007, 6403 procedures with discharge data were entered by 287 providers at 56 centers on 2763 CAS patients (1450 with 30-day outcomes, 52.5%) and 3259 CEA patients (1368 with 30-day outcomes, 42%). Of the total cohort, 98% of CEA and 70.7% of CAS (P < .001) were performed for atherosclerotic disease. Restenosis accounted for 22.3% and post-radiation induced stenosis in 4.5% of CAS patients. Preprocedure lateralizing neurologic symptoms were present in a greater proportion of CAS patients (49.2%) than CEA patients (42.4%, P < .001). CAS patients also had higher preprocedure prevalence of coronary artery disease (CAD), MI, congestive heart failure (CHF), chronic obstructive pulmonary disease (COPD), and cardiac arrhythmia. For CAS, death/stroke/MI at 30 days was 7.13% for symptomatic patients and 4.60% for asymptomatic patients (P = .04). For CEA, death/stroke/MI at 30 days was 3.75% in symptomatic patients and 1.97% in asymptomatic patients (P = .05). After risk-adjustment for age, history of stroke, diabetes, and American Society of Anesthesiologists (ASA) grade (ie, factors found to be significant confounders in outcomes using backwards elimination), logistic regression analysis suggested better outcomes following CEA. There were no statistically significant differences when examining CAS outcomes based on center volume. CAS in atherosclerotic disease had significantly worse outcomes than in nonatherosclerotic stenosis. When CAS and CEA were compared in the treatment of atherosclerotic disease only, the difference in outcomes between the two procedures was more pronounced, with death/stroke/MI 6.42% after CAS vs 2.62% following CEA, P < .0001.
Conclusion
Following best possible risk adjustment of these unmatched groups, symptomatic and asymptomatic CAS patients had significantly higher 30-day postprocedure incidence of death/stroke/MI when compared with CEA patients. The initial 1.5 years of data collection provide proof of concept that a specialty society based VR can succeed in meeting regulatory and scientific goals. With continued enrollment and follow-up, analysis of SVS-VR will supplement randomized trials by providing real-world comparisons of CAS and CEA with sufficient numbers to serve as an outcome assessment tool of important patient subsets and across the spectrum of peripheral vascular procedures.
The Society for Vascular Surgery (SVS) Outcomes Committee developed the Vascular Registry (VR) for Carotid Procedures in response to the 2004 Centers for Medicare and Medicaid Services (CMS) National Coverage Decision (NCD) on carotid artery stenting (CAS)1 for symptomatic high-risk patients with >70% stenosis, and other patients enrolled in IDE trials, which was based upon the Acculink for Revascularization of Carotids in High-Risk (ARCHeR)2 (Guidant Corporation, Santa Clara, Calif) and Stenting and Angioplasty with Protection in Patients at High Risk for Endarterectomy (SAPPHiRE)3 (Cordis Corporation, Warren, NJ) clinical trials approved by the FDA. Although CMS requires data submission for facility certification/recertification only on CAS, SVS designed VR to collect similar data on carotid endarterectomy (CEA) to allow comparison of outcomes, with potential for expansion to procedures involving other vascular beds.
As the first societal registry to enroll CAS and CEA patients, the SVS-VR is available to all clinical facilities and individual providers in the United States wishing to participate. The VR does not have inclusion or exclusion criteria for patient eligibility and is reliant on site entry of patients in whom CAS or CEA is performed. Since the VR is designed to capture real-world practices, it does not have predefined visit intervals and relies upon each facility's standards of care practice. However, for purposes of reporting, data are allocated to 30-day, 6-month, 1-year, and annual subsequent time points. The VR Audit Program has two purposes: (1) to ensure all cases of CAS/CEA are being reported; and, (2) to verify data accuracy and completeness.
The VR allows each facility real-time access to downloadable datasets of all carotid procedures entered in the registry for that facility with the ability to assess correlations among practitioners, procedures, comorbidities, and outcomes within the facility. In addition, although CMS only requires in-hospital data for certification, longer follow-up data were included in the VR since perioperative mortality and complications may take place after discharge, and longer follow-up would provide information on recurrent disease. Furthermore, the VR provides reports that allow comparison of outcomes among various facilities in an anonymous manner.
The purpose of this publication is to report the feasibility of the VR and to provide the baseline demographics and risk-adjusted 30-day outcomes of CAS and CEA.
Methods
VR data are reported by providers through web-based electronic data capture. The measurement schedule includes baseline (preoperative) demographics, medical history, carotid symptom status, and preprocedural diagnostic imaging and laboratory; procedural (CAS or CEA) information including clinical utility, and intraoperative and predischarge complications; and follow-up information such as postoperative mortality, stroke, myocardial infarction and other morbidity. All data entered into the VR are fully compliant with the Health Insurance Portability and Accountability Act (HIPAA) regulations and are auditable. All data reports and analyses performed include only de-identified and aggregated data.
The New England Research Institutes, Inc (NERI, Watertown, Mass) maintains the on-line database. Funding for the administration and database management of the VR has been provided by the Society for VR.
Outcomes
In order to provide information required by the CMS NCD, the primary outcome measure is combined death, stroke, and myocardial infarction (MI). Analysis of 30-day outcomes was based on only those patients who had at least a 30-day postprocedure visit (window 15 to 59 days) or who experienced an endpoint (death, stroke, or MI) within 30 days of treatment. Stroke is defined as any nonconvulsive, focal neurological deficit of abrupt onset persisting more than 24 hours. The ischemic event must correspond to a vascular territory. An MI is classified as either Q wave MI in which one of the following criteria is required: (1) chest pain or other acute symptoms consistent with myocardial ischemia and new pathological Q waves in two or more contiguous ECG leads; or (2) new pathologic Q waves in two or more contiguous ECG leads and elevation of cardiac enzymes; or (3) non-Q wave MI, defined as CK ratio >2, and CK-MB >1 in the absence of new, pathological Q waves.
Procedural success data were also collected. A procedure, either CAS or CEA, is deemed successful when all of its components are completed without the need of conversion (CAS to CEA or vice versa) or its abandonment prior to completion.
Statistical methods
Tests of statistical significance were conducted with χ2 or Fisher's exact tests for categorical variables and analysis of variance (ANOVA) for continuous variables. Descriptive statistics are listed as mean ± standard deviation for continuous variables and percent (frequency) for categorical variables. Subset analyses were performed using the two-tailed t test for continuous variables and the χ2 or Fisher exact, as necessary, for discrete/categorical data. Unadjusted and adjusted odds ratios were used to compare the primary outcomes across treatment groups. Odds ratios were adjusted for age and any significant baseline factors that were kept after using backwards elimination methods. Differences were considered significant if P < .05. All statistical analyses were performed by NERI using SAS Statistical Software (Cary, NC).
Results
For the purpose of this study, data collected in the VR from beginning of electronic data entry on July 11, 2005 to December 26, 2007 were analyzed for death/stroke/MI and compared between CAS and CEA at 30 days postprocedure. Since CMS requires in-hospital peri-CAS results for recertification, a subgroup analysis was performed to analyze intraoperative and predischarge data. A total of 6403 procedures with discharge data were entered by 287 providers from six specialties at 56 centers on 2763 CAS patients (1450 with 30-day outcomes, 52.5%) and 3259 CEA patients (1368 with 30-day outcomes, 42%). The specialties represented in the VR are vascular surgery, interventional cardiology, interventional radiology, neurosurgery, neuroradiology, and cardiothoracic surgery. Baseline demographics for the entire cohort by treatment arm are presented in Table I. The average age was 71 years, approximately 59% were male, and 93% were white for both treatment arms.
Table I. Baseline demographics and past medical history for all patients (entire cohort) by treatment arm
| CAS n = 2763 | CEA n = 3259 | P valuea | |
|---|---|---|---|
| Age (y, mean) | 70.74 ±9.86(range36-96) | 71.06±9.52(range24-99) | .208 |
| Gender (male, %) | 59.8%(1651/2763) | 58.5%(1907/3259) | .330 |
| White (%) | 92.7%(2560/2763) | 93.2%(3036/3259) | .447 |
| Hispanic | 4.2%(115/2763) | 2.3%(76/3259) | <.001 |
| Etiology | |||
| Atherosclerosis | 70.7%(1954/2763) | 98.0%(3193/3259) | <.001 |
| Dissection | 0.9%(25/2763) | 0.1%(4/3259) | |
| Fibromuscular dysplasia | 0.1%(4/2763) | 0.1%(4/3259) | |
| Radiation | 4.5%(123/2763) | 0.2%(5/3259) | |
| Trauma | 0.3%(7/2763) | 0.1%(2/3259) | |
| Restenosis | 22.3%(615/2763) | 1.2%(38/3259) | |
| Other | 1.3%(35/2763) | 0.4%(13/3259) | |
| Symptomatology (% symptomatic) | 49.2%(1359/2763) | 42.4%(1382/3259) | <.001 |
| Coronary artery disease | 60.8%(1679/2763) | 48.1%(1569/3259) | <.001 |
| Myocardial infarction | 22.1%(610/2763) | 16.5%(537/3259) | <.001 |
| Valvular heart disease | 7.3%(203/2763) | 7.3%(238/3259) | .948 |
| Cardiac arrhythmia | 13.2%(365/2763) | 12.3%(402/3259) | .310 |
| Congestive heart failure | 13.5%(373/2763) | 6.9%(226/3259) | <.001 |
| Hypertension | 80.2%(2215/2763) | 80.9%(2637/3259) | .465 |
| Diabetes | 33.0%(912/2763) | 28.8%(940/3259) | <.001 |
| Stroke (CVA) | 25.4%(702/2763) | 19.2%(626/3259) | <.001 |
| Transient ischemic attack | 21.2%(587/2763) | 18.8%(612/3259) | .017 |
| TMB/amaurosis fugax | 7.1%(196/2763) | 6.5%(213/3259) | .391 |
| COPD | 17.9%(494/2763) | 15.8%(516/3259) | .034 |
| Renal failure (Cr ≥ 3mg/dL) | 3.3%(91/2763) | 3.1%(100/3259) | .619 |
| Peripheral vascular disease | 37.4%(1033/2763) | 34.8%(1133/3259) | .035 |
| GI ulcer/bleeding | 3.5%(97/2763) | 2.5%(82/3259) | .024 |
| Current or past smoker | 58.2%(1609/2763) | 55.8%(1819/3259) | .059 |
| Cancer | 16.8%(465/2763) | 11.5%(376/3259) | <.001 |
| Coagulopathy | 1.0%(27/2763) | 1.3%(43/3259) | .217 |
| ASA grade | |||
| ≤3 | 91.5%(2529/2763) | 92.9%(3029/3259) | .041 |
| >3 | 8.5%(234/2763) | 7.1%(230/3259) | |
| NYHA scale | |||
| ≤2 | 87.4%(2415/2763) | 93.2%(3038/3259) | <.001 |
| >2 | 12.6%(348/2763) | 6.8%(221/3259) |
aP values are all found using χ2 tests. |
There was a statistically significant difference in symptomatology, with 49.2% of CAS and 42.4% of CEA being performed in patients presenting with lateralizing neurologic symptoms (P < .001). CAS patients had higher prevalence of preprocedure stroke (25% vs 19% CEA, P < .001), transient ischemic attack (TIA) (21% vs 19% CEA, P = .02), CAD (61% vs 48% CEA, P < .001), MI (22% vs 17% CEA, P < .001), CHF (13% vs 7% CEA, P < .001), chronic obstructive pulmonary disease (COPD) (18% vs 16% CEA, P = .035), peripheral vascular disease (37% vs 35% CEA, P = .035), and cancer (17% vs 12% CEA, P < .001). In addition, the CAS arm had a statistically significant higher incidence of American Society of Anesthesiologists (ASA) grade >3 (9% vs 7% CEA, P = .04) and New York Heart Association (NYHA) class for congestive heart failure >2 (13% vs 7% CEA, P < .001).
There was no significant difference in procedural success rates between CEA (99.9%) and CAS (99.1%). The CAS procedural failures include conversions to CEA. A total of 38 CAS patients (1.4%) had a second ipsilateral procedure also captured in the VR; while two had a third ipsilateral procedure. Of these secondary procedures, six were considered conversions to CEA (defined as a CEA performed on the same side within 30 days of the initial CAS procedure). Of the CEA patients, 14 (0.4%) had a second ipsilateral procedure and none had a third ipsilateral procedure.
In order to report on 30-day outcomes, an assessment of follow-up visits was made. At the time of the data freeze, there were a total of 56 sites participating in the VR. Of those, 37 sites had entered follow-up information on at least 20% of their enrolled patients. In order to capture all the visits, wide visit windows were used for each interval. For example, for 30-day complications, any complication that was reported to have happened within 30 days of the procedure and was reported by the patient during a visit that occurred 15 to 59 days postprocedure was considered for the 30-day interval. A sub-dataset of patients with any follow-up that accurately determine 30-day complication rates was created. Of the entire cohort (6022 patients), 2818 (44%) patients had a follow-up visit or died within 30 days (CAS = 1450, 52.2% of total CAS patients; CEA = 1368, 42% of total CEA patients).
In order to ensure adequate comparison of baseline risk factors, Table II shows the baseline demographic and medical history of the subset of patient with follow-up information. Similar to the entire cohort, the subset of patients with follow-up demonstrated a statistically significant difference in symptomatology, with 45% of CAS and 37% of CEA being performed for symptomatic disease (P < .001). CAS patients had higher prevalence of preprocedure stroke (25% vs 21% CEA, P = .004), transient monocular blindness (TMB)/amaurosis fugax (8% vs 5% CEA, P = .002), CAD (61% vs 46% CEA, P < .001), MI (24% vs 16% CEA, P < .001), cardiac arrhythmia (14% vs 11% CEA, P = .03), CHF (15% vs 7% CEA, P < .001), diabetes (33% vs 26% CEA, P < .001), COPD (18% vs 12% CEA, P < .001), gastrointestinal (GI) ulcer/bleeding (4% vs 2% CEA, P = .02), smoking (59% vs 56% CEA, P = .05), and cancer (19% vs 10% CEA, P < .001). In patients with follow-up, there was a higher incidence of peripheral vascular disease in the CEA arm (46% vs 38% CAS, P < .001). In addition, the CAS arm had a statistically significant higher incidence of ASA grade >3 (6% vs 5% CEA, P = .04) and NYHA class >2 (13% vs 10% CEA, P = .04).
Table II. Baseline demographics and past medical history by treatment arm in 30-day follow-up cohort
| CAS n = 1450 | CEA n = 1368 | P valuea | |
|---|---|---|---|
| Age (y, mean) | 70.78±10.03(range37-94) | 71.17±9.39(range35-95) | .280 |
| Gender (male, %) | 59.5%(863/1450) | 59.7%(817/1368) | .912 |
| White (%) | 93.6%(1357/1450) | 94.9%(1298/1368) | .141 |
| Hispanic | 3.3%(48/1450) | 1.4%(19/1368) | <.001 |
| Etiology | |||
| Atherosclerosis | 67.7%(982/1450) | 97.8%(1338/1368) | <.001* |
| Dissection | 0.8%(12/1450) | 0.3%(4/1368) | |
| Fibromuscular dysplasia | 0.1%(2/1450) | 0.2%(3/1368) | |
| Radiation | 5.4%(79/1450) | 0.3%(4/1368) | |
| Trauma | 0.2%(3/1450) | 0.1%(2/1368) | |
| Restenosis | 24.3%(353/1450) | 1.1%(15/1368) | |
| Other | 1.3%(19/1450) | 0.1%(2/1368) | |
| Carotid symptomatology (% symptomatic) | 44.5%(645/1450) | 37.0%(506/1368) | <.001 |
| Coronary artery disease | 61.4%(890/1450) | 45.7%(625/1368) | <.001 |
| Myocardial infarction | 23.7%(343/1450) | 15.6%(214/1368) | <.001 |
| Valvular heart disease | 7.7%(112/1450) | 6.6%(90/1368) | .239 |
| Cardiac arrhythmia | 13.8%(200/1450) | 11.1%(152/1368) | .031 |
| Congestive heart failure | 14.7%(213/1450) | 6.9%(95/1368) | <.001 |
| Hypertension | 81.6%(1183/1450) | 78.8%(1078/1368) | .064 |
| Diabetes | 33.0%(478/1450) | 26.1%(357/1368) | <.001 |
| Stroke (CVA) | 25.1%(364/1450) | 20.5%(281/1368) | .004 |
| Transient ischemic attack | 21.7%(314/1450) | 18.9%(259/1368) | .073 |
| TMB/amaurosis fugax | 7.9%(114/1450) | 5.0%(68/1368) | .002 |
| COPD | 18.1%(262/1450) | 12.4%(170/1368) | <.001 |
| Renal failure | 2.9%(42/1450) | 2.3%(31/1368) | .292 |
| Peripheral vascular disease | 38.3%(556/1450) | 46.1%(631/1368) | <.001 |
| GI ulcer/bleeding | 4.0%(58/1450) | 2.5%(34/1368) | .024 |
| Current or past smoker | 59.3%(860/1450) | 55.6%(761/1368) | .048 |
| Cancer | 19.5%(283/1450) | 9.8%(134/1368) | <.001 |
| Coagulopathy | 1.3%(19/1450) | 1.5%(21/1368) | .614 |
| ASA grade | |||
| ≤3 | 93.7%(1359/1450) | 95.5%(1306/1368) | .041 |
| >3 | 6.3%(91/1450) | 4.5%(62/1368) | |
| NYHA scale | |||
| ≤2 | 86.8%(1258/1450) | 90.2%(1234/1368) | .004 |
| >2 | 13.2%(192/1450) | 9.8%(134/1368) |
aP values are all found using χ2 tests except for etiology which uses Fisher exact test (denoted with an asterisk *). |
Procedurally, out of 1450 CAS patients, 1376 (94.9%) had embolic protection and 74 (5.1%) did not. Out of 1368 CEA patients, 952 (69.6%) had patch and 416 (30.4%) did not. For the subset of patients with 30-day follow-up, there were unadjusted statistically significant differences among 30-day event rates between treatment arms, as seen in Table III. The CAS arm had higher 30-day event rates compared with CEA. This significant difference was maintained even after the data were risk-adjusted. Table IV provides the unadjusted and adjusted odds ratios for several pertinent outcomes using logistic regression. Age, race/ethnicity, symptomatology, history of atherosclerosis, CAD, MI, arrhythmia, CHF, diabetes, TIA, TMB/amaurosis fugax, stroke, COPD, peripheral vascular disease (PVD), GI bleed, cancer, ASA grade, and NYHA class were all considered as potential confounders on the outcomes of death, stroke, and MI. Mean-centered age, diabetes, stroke, and ASA grade (≤3, >3) were significantly different between the CAS and CEA arms at baseline and had a significant effect on the outcomes of death, stroke, and MI using backwards elimination techniques. Therefore, a parsimonious model of mean-centered age, diabetes, stroke, and ASA grade (≤3, >3) was used to adjust the estimates, which are presented in Table IV. Although preoperative stroke was included in the risk adjustment model, TIA and TMB/amaurosis fugax were not because they were not statistically significant factors on backward elimination.
Table III. Outcomes at 30-days (includes intraop and predischarge) by treatment arm in 30-day follow-up cohort
| Outcome | CAS n = 1450 % (m) | CEA n = 1368 % (m) | P value |
|---|---|---|---|
| Combined death/stroke/MI | 5.72%(83/1450) | 2.63%(36/1368) | <.001 |
| Death | 2.07%(30/1450) | 0.73%(10/1368) | .004 |
| Stroke | 3.52%(51/1450) | 1.68%(23/1368) | .003 |
| MI | 1.17%(17/1450) | 0.58%(8/1368) | .110 |
| TIA | 1.59%(23/1450) | 0.80%(11/1368) | .060 |
| TMB/amaurosis fugax | 0.21%(3/1450) | 0.00%(0/1368) | .250 |
Table IV. Unadjusted and adjusted odds ratios for CAS vs CEA in 30-day follow-up cohort
| 30-day outcomes | CAS (vs CEA) - unadjusted | P value | CAS (vs CEA) - adjusteda | P value |
|---|---|---|---|---|
| Death/stroke/MI | 2.247 | <.001 | 1.965 | <.001 |
| Death/stroke | 2.331 | <.001 | 2.016 | .002 |
| Death | 2.865 | .004 | 2.513 | .013 |
| Stroke | 2.132 | .003 | 1.812 | .021 |
| MI | 2.016 | .103 | 1.789 | .181 |
aAdjusted for age and any statistically significant baseline factors and then used backwards elimination to get a parsimonious model consisting of age, diabetes, stroke, and ASA grade (≤3, >3). |
In order to analyze CAS outcomes based on the experience of the medical center performing this procedure, centers were divided into those performing less than 25 procedures, 25 to 50 procedures, and over 50 procedures (Table V). The data show no statistically significant difference in outcomes based on the number of CAS procedures performed.
Table V. 30-day endpoints by site by CAS volume
| 30-day adverse event | CAS patient volume | P valuea | ||
|---|---|---|---|---|
| <25 patients 26 sites % (m) | 25-50 patients 11 sites % (m) | >50 patients 13 sites % (m) | ||
| Mortality | 1.67%(2) | 2.79%(5) | 2.00%(23) | .6602 |
| Stroke | 3.33%(4) | 4.47%(8) | 3.39%(39) | .7226 |
| MI | 0.83%(1) | 1.12%(2) | 1.22%(14) | 1.0000 |
| TIA | 3.33%(4) | 1.12%(2) | 1.48%(17) | .2522 |
| TMB/amaurosis fugax | 0.83%(1) | 0.00%(0) | 0.17%(2) | .2665 |
| Death, Stroke, or MI | 5.83%(7) | 7.26%(13) | 5.47%(63) | .5956 |
aP values were based on Fisher exact test. |
There was a statistically significant difference in carotid artery disease etiology, with 98% of CEA compared with 70.7% of CAS (P < .001) being performed for atherosclerotic disease. In the CAS arm, 22.3% and 4.5% were performed for recurrent stenosis and radiation-induced stenosis, respectively. Within the CAS arm, when we compared the results based on the etiology of the stenotic lesion, there was a statistically significant higher incidence of 30-day postprocedural stroke (P = .049) when the atherosclerosis group (41/982, 4.1%) was compared with the nonatherosclerosis group (10/468, 2.1%), Table VI. When CAS and CEA were compared in the treatment of atherosclerotic disease only, the difference in outcomes between the two procedures was more pronounced, with CAS faring worse in death/stroke/MI (6.42% vs 2.62% CEA, P < .0001, respectively), death (2.04% vs 0.75% CEA, P = .0085, respectively), and stroke (4.18% vs 1.64% CEA, P = .0003; Table VII).
Table VI. Outcomes at 30-days (includes intra-op and predischarge) in CAS based on the etiology of carotid disease
| 30-day adverse event | CAS | P valuea | |
|---|---|---|---|
| Atherosclerotic n = 982 % (m/n) | Nonatherosclerotic n = 468 % (m/n) | ||
| Death, stroke, or MI | 6.42%(63/982) | 4.27%(20/468) | .1159 |
| Mortality | 2.04%(20/982) | 2.14%(10/468) | 1.0000 |
| Stroke | 4.18%(41/982) | 2.14%(10/468) | .0486 |
| MI | 1.43%(14/982) | 0.64%(3/468) | .2962 |
| TIA | 1.73%(17/982) | 1.28%(6/468) | .6552 |
| TMB/amaurosis fugax | 0.10%(1/982) | 0.43%(2/468) | .2451 |
aP values were based on Fisher exact test. |
Table VII. Outcomes at 30 days (includes intra-op and predischarge) by treatment arm in atherosclerotic disease only
| 30-day adverse event | CAS atherosclerotic n = 982 % (m/n) | CEA atherosclerotic n = 1338 % (m/n) | P valuea |
|---|---|---|---|
| Death, stroke, or MI | 6.42%(63/982) | 2.62%(35/1338) | <.0001 |
| Mortality | 2.04%(20/982) | 0.75%(10/1338) | .0085 |
| Stroke | 4.18%(41/982) | 1.64%(22/1338) | .0003 |
| MI | 1.43%(14/982) | 0.60%(8/1338) | .0509 |
| TIA | 1.73%(17/982) | 0.82%(11/1338) | .0548 |
| TMB/amaurosis fugax | 0.10%(1/982) | 0.00%(0/1338) | .4233 |
aP values were based on Fisher exact test. |
In addition to examining the adjusted odds ratios in Table IV, a subset analysis of symptomatic vs asymptomatic patients was performed. Table VIII demonstrates that symptomatic patients fared worse than asymptomatic patients. The CAS arm had significantly more symptomatic patients 44.5% vs 37.0% in the CEA arm. The P values in Table VIII indicate that although symptomatology may confound the relationship between procedure type and key outcomes, it does not modify the relationship (ie, symptomatic patients are more likely to have adverse events, regardless of the procedure type).
Table VIII. Outcomes at 30-days (including intraop and pre-discharge) by treatment arm by symptomatology in 30-day follow-up cohort
| 30-day outcomes | CAS | CEA | ||||
|---|---|---|---|---|---|---|
| SYMPT n = 645 % (m) | ASYMPT n = 805 % (m) | P value (CAS)a | SYMPT n = 506 % (m) | ASYMPT n = 862 % (m) | P value (CEA)a | |
| Combined Death/stroke/MI | 7.13%(46/645) | 4.60%(37/805) | .0410 | 3.75%(19/506) | 1.97%(17/862) | .0544 |
| Death | 2.17%(14/645) | 1.99%(16/805) | .8539 | 0.79%(4/506) | 0.70%(6/862) | 1.0000 |
| Stroke | 5.27%(34/645) | 2.11%(17/805) | .0014 | 2.37%(12/506) | 1.28%(11/862) | .1333 |
| MI | 0.93%(6/645) | 1.37%(11/805) | .4747 | 0.59%(3/506) | 0.58%(5/862) | 1.0000 |
| TIA | 2.02%(13/645) | 1.24%(10/805) | .2918 | 1.38%(7/506) | 0.46%(4/862) | .1119 |
| TMB/amaurosis fugax | 0.16%(1/645) | 0.25%(2/805) | 1.0000 | 0.00%(0/506) | 0.00%(0/862) | N/A |
aP values were based on Fisher exact test. |
For CAS, combined death/stroke/MI at 30 days was 7.13% for symptomatic patients and 4.60% for asymptomatic patients. For CEA, death/stroke/MI at 30 days was 3.75% in symptomatic patients and 1.97% in asymptomatic patients. After risk-adjustment for age, history of stroke, diabetes, and ASA grade (ie, factors found to be significant confounders in outcomes using backwards elimination); logistic regression analysis demonstrated better outcomes following CEA.
There were no statistically significant differences in either the CAS or CEA arm in 30-day outcomes when performing gender analysis, as seen in Table IX, A, Table IX, B. In the CAS arm, there was a higher rate of death/stroke/MI in females (7.0%) compared with males (4.9%), P = .106. In the CEA arm, males had a higher death/stroke/MI rate (3.2%) compared with females (1.8%), P = 0.167.
Table IX, A. Adverse events through 30-days in CAS arm by gender
| Adverse event | CAS | P value | |
|---|---|---|---|
| Male n = 863 % (m/n) | Female n = 587 % (m/n) | ||
| Combined death/stroke/MI | 4.87%(42/863) | 6.98%(41/587) | .106 |
| Death | 1.85%(16/863) | 2.39%(14/587) | .574 |
| Stroke | 2.78%(24/863) | 4.60%(27/587) | .081 |
| MI | 0.93%(8/863) | 1.53%(9/587) | .326 |
| TIA | 1.74%(15/863) | 1.36%(8/587) | .671 |
| TMB/amaurosis fugax | 0.35%(3/863) | 0.00%(0/587) | .277 |
Table IX, B. Adverse events through 30-days in CEA arm by gender
| Adverse event | CEA | P value | |
|---|---|---|---|
| Male n = 817 % (m/n) | Female n = 551 % (m/n) | ||
| Combined death/stroke/MI | 3.18%(26/817) | 1.81%(10/551) | .167 |
| Death | 1.10%(9/817) | 0.18%(1/551) | .057 |
| Stroke | 1.96%(16/817) | 1.27%(7/551) | .396 |
| MI | 0.61%(5/817) | 0.54%(3/551) | 1.000 |
| TIA | 0.86%(7/817) | 0.73%(4/551) | 1.000 |
| TMB/amaurosis fugax | 0.00%(0/817) | 0.00%(0/551) | N/A |
Since CMS requires only in-hospital, predischarge results to be reported for CAS facility recertification, we performed a subgroup analysis for intraprocedure and predischarge data (Table X). CAS data indicated that an additional 1.79% of combined death/stroke/MI take place after hospital discharge but within 30 days of procedure (Fig) compared with only 0.58% for CEA.
Table X. Intraoperative and predischarge events by treatment arm in 30-day follow-up cohort
| Outcome | CAS n = 1450 | CEA n = 1368 | P valuea |
|---|---|---|---|
| Intraoperative only | |||
| Stroke | 0.897%(13/1450) | 0.219%(3/1368) | .022 |
| TIA | 0.552%(8/1450) | 0.073%(1/1368) | .039 |
| TMB/amaurosis fugax | 0.069%(1/1450) | N/A | N/A |
| Predischarge only | |||
| Combined death/stroke/MI | 3.379%(49/1450) | 1.827%(25/1368) | .013 |
| Death | 1.448%(21/1450) | 0.512%(7/1368) | .013 |
| Stroke | 2.000%(29/1450) | 1.170%(16/1368) | .098 |
| MI | 0.828%(12/1450) | 0.512%(7/1368) | .362 |
| TIA | 0.552%(8/1450) | 0.219%(3/1368) | .228 |
| TMB/amaurosis fugax | 0.138%(2/1450) | 0.000%(0/1368) | .500 |
| Combined intraop and predischarge | |||
| Combined death/stroke/MI | 3.931%(57/1450) | 2.047%(28/1368) | .004 |
| Death | 1.448%(21/1450) | 0.512%(7/1368) | .013 |
| Stroke | 2.690%(39/1450) | 1.389%(19/1368) | .017 |
| MI | 0.828%(12/1450) | 0.512%(7/1368) | .362 |
| TIA | 1.034%(15/1450) | 0.292%(4/1368) | .020 |
| TMB/amaurosis fugax | 0.207%(3/1450) | 0.000%(0/1368) | .250 |
aP values found using Fisher exact test. |
Fig.
30-day outcomes of CAS during in-hospital stay and post-discharge up to 30 days.
Discussion
The SVS has developed vascular registry expertise over the last decade with the initial emphasis being on compiling the data from US FDA approved clinical trials of abdominal aortic endografts compared with control surgical procedures. This registry has provided valuable data regarding the efficacy of endovascular repair of AAA, particularly in high-risk patients,4 and has produced a surgical control data analysis that will be available for comparison for future endograft trials.5 This registry's experience and the recent CMS NCD decision that called for outcomes reporting as a condition for CAS facility recertification stimulated the SVS and its Outcomes Committee to create an on-line registry that prospectively collects data on carotid artery procedures. Although the NCD required data only on CAS, SVS decided to include data collection on CEA as well to allow for further comparisons between the two procedures. This is particularly important for not only determining the natural history of CAS compared with CEA but also for providing broad-based clinical outcomes and practice pattern data that will provide quality assessment measurements and standards developed using contemporary data. In addition, while the CMS NCD requires only in-hospital predischarge data on CAS procedures, the SVS-VR collects long-term data well beyond the immediate postprocedure period to allow for better understanding of the long-term impact of CAS and comparison between CAS and CEA. The Audit Program, which began in September 2006 when the first site was audited, was designed to (1) verify that the data entered into the VR were accurate and complete, and (2) ensure all cases of CAS and/or CEA are entered into the VR. For the purpose of this review, we analyzed data collected during the initial 18 months of registry activity to provide and compare 30-day outcomes of CAS to those of CEA. Of 6022 patients who were entered in the VR, 2818 patients (1450 CAS patients and 1368 CEA patients) have had 30-day outcomes entered; therefore, this study was based on data derived from this sub-group. As the VR matures, and if CMS mandates longer term outcomes assessment, the percentage of patients with extended follow-up should improve. In addition, since the CMS NCD regarding CAS calls for reporting only on in-hospital data, we analyzed and compared intraprocedure and predischarge data for CAS and CEA to 30-day outcomes to find out whether complications of these procedures do occur and are reported beyond the in-hospital, predischarge period. We focused our outcome comparisons on the combined death/ stroke/MI since many other reported series used this end-point and that the CAS NCD was based on this composite endpoint.
In comparison to other reported studies, the overall 30-day outcome of death/stroke/MI in the CAS arm (5.7%) is consistent with the outcomes observed in Boston Scientific EPI-A Carotid Stenting Trial for High Risk Surgical Patients (BEACH)6 (5.8%) and Medtronic AVE Self-Expanding Carotid Stent System with distal protection in the treatment of Carotid stenosis (MAVErIC)7 (5.9%), but higher than Carotid Revascularization using Endarterectomy or Stenting Systems (CaRESS)8 (2.1%), Carotid Artery Revascularization Using The Boston Scientific FilterWire EX/EZ and the EndoTex NexStent (CABernNET)9 (3.9%), Stenting and Angioplasty with Protection in Patients at High Risk for Endarterectomy (SAPPHIRE)3 (4.8%), and Carotid Artery Stenting with Emboli protection Surveillance study (CASES-PMS)10 (5.0%). The 30-day event rate observed in the VR in the CAS arm was lower than CREATE11 (6.2%), ARCHeR2 (8.3%), SECuRITY12 (8.5%), and CAPTURE13 (6.3%). However, a 5.7% composite outcome for CAS in our study was significantly higher than that for CEA (2.63%, P < .001). When these outcomes were risk-adjusted for age, diabetes, stroke, and ASA grade, CAS continued to have a significantly higher composite outcome than CEA with adjusted odds ratio (AOR) of 1.96, P < .001. In addition, except for MI, each of the components of combined outcome has shown a significantly higher AOR for CAS compared with CEA (Table IV). Although the AOR for MI did not reach statistical significance, the AOR for MI in CAS was 1.78 compared with CEA. Results of the symptomatic VR subset may also be compared with those of the recently published EVA-3S14 randomized controlled trial of CAS vs CEA in 527 symptomatic patients. The 30-day stroke or death rate following CEA was 3.9% in EVA-3S, while the combined end point of stroke, death, or MI was slightly less (3.75%) in symptomatic VR patients. Likewise, the 30-day stroke or death rate following CAS in EVA-3S was 9.6% compared with 7.13% for the combined death/stroke/MI end point in symptomatic VR patients. While the reported event rates in both studies favor CEA in symptomatic patients, the VR CAS results were less morbid than those of EVA-3S.
In this series, patients in the CAS group had significantly higher proportion of preprocedure lateralizing symptoms (49.2% vs 42.4% for CEA, P < .001). This is expected since the NCD coverage is limited to high surgical risk symptomatic patients with greater than 70% carotid artery stenosis, and asymptomatic patients may be covered by Medicare only when enrolled in IDE trials. In both CAS and CEA patients, asymptomatic patients fared better on the composite outcome than those with symptoms; however, the results reached statistical significance in the CAS group only (Table VIII).
It is of interest to note that the etiology of the carotid stenosis in the CAS and CEA groups was significantly different. Although over 97% of CEA procedures were performed for stenoses caused by atherosclerotic disease, only 67.7% of CAS procedures were performed for this etiology. About one third of CAS procedures in our study were performed for stenoses caused by etiology other than atherosclerosis the most common of which were recurrent stenosis and history of prior neck irradiation. When outcomes in CAS group were analyzed based on etiology, only postoperative stroke rate was found to be significantly higher when CAS was performed for stenoses caused by atherosclerotic lesions (4.1% vs 2.1% for nonatherosclerotic etiology, P = .049). This confirms other reported series that suggested that CAS performed for recurrent carotid artery stenosis has more favorable results.15
It is important to note that this study's results suggest that in-hospital outcomes are not adequate to capture the full spectrum of carotid procedures impact especially in CAS. We performed a subgroup analysis for intraprocedure and predischarge data and compared them with outcome data from full 30-day follow-up, which was documented in almost half of the patients in this study for both CAS and CEA (Table X). The results indicated that 31% of 30-day combined complications appear after hospital discharge in CAS patients (Fig); therefore, reporting in-hospital data only as required for CAS facility recertification do not reflect the full picture of CAS outcomes.
The main weakness of this study is its reliance on self-reporting with its biases inherent to any registry-based study. To minimize such bias, the Outcomes Committees provided strict definitions of complications for conformity. In addition, the Committee put together an auditing process of the data. Furthermore, some facilities entered either CAS or CEA data; some institutions do not perform CAS and elected to participate in the VR and enter only CEA data. Others entered only CAS data. One criticism usually leveled against studies comparing CAS with CEA is that the centers performing CAS usually perform small numbers of this procedure; therefore, its outcomes are worse than CEA.16 To test this argument, we analyzed our data based on the number of CAS procedures performed by various centers. We found no statistically significant difference in CAS outcomes among centers performing less than 25 procedures, 25 to 50 procedures, or over 50 CAS procedures (Table V). In addition, the majority of CAS procedures were performed in centers performing more than 25 procedures. Therefore, the reason why CAS outcomes do not match those of CEA in this report is not low CAS volume experience in participating centers.
The debate about the interpretation of the results of this study as well as results of other CAS studies will continue until randomized trials such as International Carotid Stenting Study (ICSS)17 in Europe and CREST15 in North America are reported. However, independent of the potential impact that randomized trials may have on defining the role of CAS and CEA, concurrent cohort entry of all patients treated for extracranial carotid disease in independent and verifiable registries will provide information about current practice patterns and efficacy and will provide an important component of quality assessment and establishment of practice standards.
The SVS Outcomes Committee wish to acknowledge Rebecca Shackelton ScM and Christopher Kenwood, MS, at New England Research Institutes, Inc, for their contributions with statistical analysis on this manuscript.
Author contributions
References
- . http://www.cms.hhs.gov/mcd/viewdecisionmemo.asp?id=157Last accessed 07/Apr/2008
- Protected carotid stenting in high-surgical-risk patients: the ARCHeR results. J Vasc Surg. 2006;44:258–269
- Protected carotid-artery stenting versus endarterectomy in high-risk patients. N Engl J Med. 2004;351:1493–1501
- . Endovascular abdominal aortic aneurysm repair: long-term outcome measures in patients at high-risk for open surgery. J Vasc Surg. 2006;44:229–236
- . Lifeline registry of endovascular aneurysm repair: open repair surgical controls in clinical trials. J Vasc Surg. 2008;48:511–518
- . Carotid stenting with distal protection in high surgical risk patients: the BEACH trial 30 day results. Cathet Cardiovasc Intervent. 2006;67:503–512
- Multicenter evaluation of a self-expanding carotid stent system with distal protection in the treatment of carotid stenosis. Am J Neuroradiol. 2006;27:759–765
- . Carotid revascularization using endarterectomy or stenting systems CaRESS phase 1 clinical trial: 1-year results. J Vasc Surg. 2005;42:213–219
- NexStent Carotid Stent and Delivery System (Summary of Safety and Effectiveness Data P050025). October 27 2006;
- Carotid artery stenting with emboli protection surveillance study: 30-day results of the CASES-PMS study. Cathet Cardiovasc Intervent. 2007;70:316–323
- Protected carotid stenting in high-risk patients with severe carotid artery stenosis. J Am Coll Cardiol. 2006;47:2384–2389
- Xact Carotid Stent System (Summary of Safety and Effectiveness Data P040038). September 6 2005;
- The CAPTURE Registry: results of carotid stenting with embolic protection in the post approval setting. Cathet Cardiovasc Intervent. 2007;69:341–348
- Endarterectomy versus stenting in patients with symptomatic severe carotid stenosis. N Engl J Med. 2006;355:1660–1671
- In-stent restenosis after carotid angioplasty-stenting: incidence and management. J Vasc Surg. 2001;33:220–226
- Appropriateness of learning curve for carotid artery stenting: analysis of periprocedural complications. J Vasc Surg. 2006;44:1205–1212
- . International carotid stenting study: protocol for a randomized clinical trial comparing carotid stenting with endarterectomy in symptomatic carotid artery stenosis. Cerebrovasc Dis. 2004;18:69–74
The analysis of the SVS Vascular Registry carotid dataset was supported exclusively by funds from the Society for Vascular Surgery (SVS).
Competition of interest: none.
PII: S0741-5214(08)01391-8
doi:10.1016/j.jvs.2008.08.039
© 2009 The Society for Vascular Surgery. All rights reserved.
