Annualized percentage change (APC) in the period 2007-2012 was 6.0% (95% CI, 1.7% to 10.6%; P < .01), and APC from 2012 to 2014, −5.6% (95% CI, −14.4% to 4.0%; P = .20).
Annualized percentage change (APC) shown in women (A) and men (B).
eFigure 1. Flowchart of Study Inclusions and Exclusions
eFigure 2. Trends in Proportion of Carotid Artery Stenting Utilization Among All Adult Patients Requiring Carotid Revascularization in the United States From 2007-2014 According to Age Group
eTable. Baseline Characteristics of Elderly Patients Undergoing Carotid Revascularization in the United States From 2007-2014
eTable 2. Association of Carotid Artery Stenting With Sociodemographic, Clinical and Hospitalization Factors in Hospitalized Carotid Stenosis Patients in the United States Between 2007-2014
Customize your JAMA Network experience by selecting one or more topics from the list below.
Otite FO, Khandelwal P, Malik AM, Chaturvedi S. National Patterns of Carotid Revascularization Before and After the Carotid Revascularization Endarterectomy vs Stenting Trial (CREST). JAMA Neurol. 2018;75(1):51–57. doi:10.1001/jamaneurol.2017.3496
Given that the Carotid Revascularization Endarterectomy vs Stenting Trial showed slightly better outcome with carotid endarterectomy compared with carotid artery stenting in adults older than 70 years, has there been any decline in carotid artery stenting utilization in patients older than 70 years in the United States after publication of the Carotid Revascularization Endarterectomy vs Stenting Trial in 2010?
In this population-based analysis of the 2007-2014 Nationwide Inpatient Sample, carotid artery stenting utilization in 494 733 weighted admissions of patients older than 70 years increased steadily at a mean of 6.0% per year, from 10.6% in 2007 to 14.5% in 2012. After multivariable adjustment for patient and hospitalization characteristics, the odds of carotid artery stenting as the method of carotid revascularization in patients older than 70 years increased by 13% when comparing times before (2007-2010) with times after (2011-2014) publication of the Carotid Revascularization Endarterectomy vs Stenting Trial.
Contrary to expectations, there has not been any significant decline in carotid artery stenting utilization in adults older than 70 years after publication of the Carotid Revascularization Endarterectomy vs Stenting Trial.
The Carotid Revascularization Endarterectomy vs Stenting Trial (CREST) showed greater safety of carotid artery stenting (CAS) in patients younger than 70 years and carotid endarterectomy (CEA) in those older than 70 years. It is unknown how the result of CREST has influenced carotid revascularization choices in the United States.
To evaluate national patterns in CAS performance in patients older than 70 years in the post-CREST (2011-2014) compared with the pre-CREST (2007-2010) era.
Design, Setting, and Participants
All adults older than 70 years undergoing carotid revascularization in the United States from 2007 to 2014 were retrospectively identified from the 2007-2014 National Inpatient Sample using International Classification of Disease, Ninth Revision procedural codes. From 61 324 882 unweighted hospitalizations contained in the 2007-2014 National Inpatient Sample, 494 733 weighted carotid revascularization admissions in adults older than 70 years were identified using International Classification of Disease, Ninth Revision procedural codes.
Main Outcomes and Measures
The proportion of CAS performed in all age groups over time was estimated and multivariable-adjusted models were used to compare the odds of receiving CAS in the pre-CREST with those in the post-CREST era in adults older than 70 years.
A total of 41.8% of all patients were women, and mean (SE) age at presentation was 78.1 (0.03) years. A total of 16.3% of CAS and 10.1% of CEA procedures were performed in patients with symptomatic stenosis. The proportion of patients older than 70 years receiving CAS increased from 11.9% in the pre-CREST to 13.8% in the post-CREST era (P = .005). In multivariable models, the odds of receiving CAS increased by 13% in all patients older than 70 years in the post-CREST compared with the pre-CREST period (odds ratio [OR], 1.13, 95% CI, 1.00-1.28, P = .04), including symptomatic women (OR, 1.31, 1.05-1.65, P = .02). Symptomatic stenosis (OR 1.39; 95% CI, 1.27-1.52; P < .001), congestive heart failure (OR, 1.48; 95% CI, 1.35-1.63; P < .001), and peripheral vascular disease (OR, 1.35; 95% CI, 1.27-1.43; P < .001) were associated with higher odds of CAS; comorbid hypertension (OR, 0.70; 95% CI, 0.66-0.74; P < .001), smoking (OR, 0.84; 95% CI, 0.78-0.91; P < .001), and weekend admission (OR, 0.77; 95% CI, 0.68-0.88; P < .001) were negatively associated with the odds of CAS.
Conclusions and Relevance
Despite concerns for higher periprocedural complications with CAS in elderly patients, the odds of CAS increased in the post-CREST compared with pre-CREST era in patients older than 70 years, including symptomatic women.
Extracranial carotid artery stenosis accounts for 7% to 10% of all ischemic strokes.1 Based on the results of multiple randomized clinical trials,2-4 carotid endarterectomy (CEA) combined with medical therapy is the established standard treatment for high-grade asymptomatic and symptomatic carotid artery stenosis. Advances in endovascular techniques over the past 2 decades have resulted in the emergence of carotid artery stenting (CAS) as an effective revascularization alternative to CEA in selected patients, and numerous studies have assessed the relative efficacy of CAS compared with CEA.5
The Carotid Revascularization Endarterectomy vs Stenting Trial (CREST) evaluated patients with severe carotid artery stenosis and showed no significant difference between CAS and CEA for the primary end point of stroke, myocardial infarction, or death in the periprocedural period or any ipsilateral stroke within 4 years of randomizationQuiz Ref ID, but a significant interaction between age and treatment effect was noted.6 Patients older than 70 years fared better with CEA, and those younger than 70 years had a slightly better outcome with CAS.6,7 There was no difference in the primary end point of CREST by sex, but on subgroup analysis, the risk of periprocedural complications was higher following CAS compared with CEA in women.8 Among all women undergoing CAS, the periprocedural risk of stroke was 5.5% compared with 2.2% in women undergoing CEA (P = .01). In the subgroup of symptomatic women, the periprocedural stroke risk was 7.5% following CAS compared with 2.7% following CEA.8 These differences were not noted in men.
With the above results from CREST, we hypothesize that CAS use in patients older than 70 years, including symptomatic women in the post-CREST period in the United States, will be on the decline, but data on carotid revascularization trends in the United States after CREST are sparse. Few studies have evaluated carotid revascularization patterns and these studies have been limited by short post-CREST follow-up period, insufficient information to evaluate trends in the subgroup of elderly patients, or lack of generalizability to the entire US population.9,10
The primary aim of this study was to evaluate national trends in CAS performance in patients older than 70 years in the pre-CREST (2007-2010) and post-CREST (2011-2014) era to see how the trial has affected clinical practice. We also evaluated the multivariable association of CAS with other clinical and hospital characteristics in patients older than 70 years.
The 2007-2014 Nationwide Inpatient Sample (NIS) was used for this study. The NIS is maintained as part of the Health Care Utilization Project of the Agency for Healthcare Research and Quality. It is the largest all-payer inpatient care database in the United States and comprises a 20% stratified random sample of all US hospital discharges. Further details on the NIS design are available at http://www.hcup-us.ahrq.gov.
We received the exemption status to conduct this study from the University of Miami Institutional Review Board. The need for informed consent was waived.
From 61 324 882 unweighted hospitalizations contained in the NIS from 2007 to 2014, all adults undergoing carotid revascularization were identified using all primary and secondary International Classification of Diseases, Ninth Revision (ICD-9) procedural codes corresponding to CAS (code 00.63) and CEA (code 38.12). These codes have been validated previously and shown to be highly specific for CAS and CEA.11 Study inclusions are summarized in eFigure 1 in the Supplement.
Comorbidities associated with revascularization hospitalizations, including congestive heart failure, chronic lung disease, chronic renal failure, and peripheral vascular disease, were identified by Agency for Healthcare Research and Quality comorbidity measures or ICD-9 diagnostic codesQuiz Ref ID. Patients with symptomatic carotid artery stenosis were identified by searching all primary and secondary ICD-9 codes for those corresponding to amaurosis fugax (code 362.34), any transient ischemic attack (codes 435.xx), and anterior circulation strokes (433.x1 and 434.x1, excluding 433.01 and 433.21).12 Information on hospital characteristics was extracted. Total annual revascularization admissions of hospitals were counted and hospitals were grouped annually into tertiles based on their yearly volumes.
Baseline characteristics were summarized using descriptive statistics. Sampling weights provided in the NIS allow for calculation of national estimates from the 20% of all admissions in the NIS.13 We used the provided weights to compute the annual national weighted proportions of CAS in various age groups. Joinpoint regression models with permutation model selection method and autocorrelation errors were fitted to identify points of change in CAS trend over time and to quantify annualized percentage change (APC) in CAS utilization over time. In joinpoint regression, a series of Monte Carlo permutation-based tests are used to identify points of change in trends (joinpoints) in a data set. The program starts with the minimum number of joinpoint (eg, 0, which is a straight line) and tests whether more joinpoints are statistically significant and must be added to the model. The APC is then computed for each of the identified trends by fitting a regression line to the natural logarithm of the rates using calendar year as a regressor variable.14
Multivariable-adjusted logistic regression models adjusted for patient demographic, clinical, and hospital characteristics were used to compare the odds of receiving CAS in the pre-CREST and post-CREST eras. All models were adjusted for age, sex, race, insurance, median household income, hospital region, hospital yearly revascularization volume, hospital bed size, hospital location/teaching status, coronary artery disease, congestive heart failure, peripheral vascular disease, valvular heart disease, chronic renal failure, atrial fibrillation, chronic lung disease, pulmonary circulation disorder, obesity, dyslipidemia, alcohol abuse, smoking, hypertension, diabetes, liver disease, baseline coagulopathy, electrolyte disorders, weekend admission, and elective admission. Additional models comparing both periods were constructed in the subgroup of women with symptomatic stenosis. A 2-tailed α level of <.05 was required for statistical significance. All analyses were performed using Stata, version 14 (StataCorp LP). We took into account the weighting, clustering, and stratification needed in the complex NIS survey design in all analysis by use of Stata’s SVY suite of commands with use of the hospital as the primary sampling unit and applying relevant probability sampling weights for robust variance estimation to all models.
Data for most variables were missing in less than 1% of the patients. Race data were missing in 13% of the patients. Missing data for race were handled using multiple imputation as recommended by the Health Care Utilization Project.15
We conducted additional sensitivity analysis stratifying patients with symptomatic stenosis into those with transient ischemic attack, amaurosis fugax, and ischemic stroke to evaluate their individual association with CAS utilization. The Charlson Comorbidity Index16 was calculated for all patients, divided into tertiles, and incorporated in this multivariable-adjusted logistic regression to adjust for potential residual confounding.
Across the period 2007 to 2014 in the United States, 912 438 weighted carotid revascularizations were performed in adults aged 18 years or older. Of these procedures, 494 733 (54.2%) were performed in individuals older than 70 years, with 87.2% undergoing CEA and 12.8% undergoing CAS. The number of revascularizations per million hospitalizations declined from 1898 in 2007 to 1505 in 2014 (P < .001) (Table 1). The proportion of revascularization procedures done in symptomatic patients increased by 55% from 9.0% in 2007 to 13.9% in 2014 (P < .001) (Table 1). A total of 16.3% of CAS and 10.1% of CEA procedures were performed in hospitalizations for symptomatic stenosis.
Baseline characteristics of patients according to the period of hospitalization are reported in eTable 1 in the Supplement. A total of 41.8% of the revascularizations were done in women; mean (SE) age at presentation was 78.1 (0.03) years; and 88.5% and 88.2% of procedures were done in white patients in 2007-2010 and 2011-2014, respectively. More than 90% of all patients had Medicare insurance. The proportion of patients with dyslipidemia, chronic renal failure, and congestive heart failure increased marginally in the 2011-2014 period compared with the 2007-2010 period.
The proportion of patients older than 70 years receiving CAS continued to increase after publication of CREST in 2010. In 2009, 12.1% of all revascularizations were CAS; this proportion increased to 12.6% in 2010. The CAS proportion increased further to 13.3% the following year and 14.4% in 2012. Comparing the pre-CREST (2007-2010) with the post-CREST (2011-2014) period, the proportion of all revascularizations that were CAS increased from 11.9% in the pre-CREST to 13.8% in the post-CREST era (P = .005).
After joinpoint regression, CAS utilization in patients older than 70 years increased at an annual rate of 6.0% per year from 2007 and continued to increase at this steady pace until 2012 (APC, 6.0%; 95% CI, 1.7% to 10.6%). However, rates declined by 5.6% from 2012 to 2014 (APC, −5.6%; 95% CI, −14.4% to 4.0%; P = .20) (Figure 1), albeit nonsignificantly. A similar decrease in CAS proportion was also noted in patients 70 years or younger (eFigure 1 in the Supplement) and in the subset of symptomatic women (Figure 2A; eFigure 2 in the Supplement) in 2013-2014. However, on joinpoint regression, CAS use in symptomatic women increased at an annual rate of 5.6% per year (APC, 5.6%; 95% CI, 0.9% to 10.5%) across the entire study period (Figure 2A).
Among all patients older than 70 years, the odds of receiving CAS rather than CEA increased marginally in 2011-2014 compared with 2007-2010 after multivariable adjustment (odds ratio [OR], 1.13; 95% CI, 1.00-1.28; P = .04) (Table 2), but in the subset of women with symptomatic stenosis, there was an even greater increase in the odds of CAS when comparing 2011-2014 with 2007-2010 (OR, 1.31; 95% CI, 1.05-1.65) (Table 3).
The presence of any symptomatic carotid artery stenosis increased the odds of CAS utilization by approximately 40% (OR, 1.39; 95% CI, 1.27-1.52) (Table 2), but in sensitivity analysis, only anterior circulation strokes and transient ischemic attacks were associated with an increased odds of CAS (eTable 2 in the Supplement). Amaurosis fugax was not associated with an increased odds of CAS (OR, 1.14; 95% CI, 0.95-1.37).
Quiz Ref IDOther important clinical factors associated with increased odds of CAS use include comorbid coronary artery disease, congestive heart failure, and peripheral vascular disease; comorbid hypertension, diabetes, liver disease, or smoking were associated with reduced odds of CAS (Table 2). No significant difference in the odds of CAS by hospital region was noted, but patients admitted to urban teaching hospitals were 49% more likely to receive CAS compared with patients treated at urban nonteaching hospitals (OR, 1.49; 95% CI, 1.27-1.75). Patients admitted on weekends were also less likely to receive CAS compared with those admitted on weekdays (OR, 0.77; 95% CI, 0.68-0.88).
Quiz Ref IDIn this contemporary analysis of a nationwide database, we found that total annual carotid revascularizations in patients older than 70 years declined over 2007-2014, but the proportion of procedures done in patients with symptomatic stenosis increased by more than 50% over time. The odds of receiving CAS as the method of revascularization increased from the pre-CREST to the post-CREST period, admission for symptomatic carotid artery stenosis was associated with 40% greater odds of CAS compared with CEA, and among symptomatic women, the odds of CAS increased by more than 30% from the pre-CREST to post-CREST period. Based on the results of CREST, these findings are contrary to expectations.
Potential reasons for the increased odds of CAS in the post-CREST period are multiple. First, the results of CREST have been subjected to a variety of interpretations. Some physicians view CREST simply as an indication of efficacy equivalence between CAS and CEA17 and give no strong consideration to the subtle differential treatment effects by age, sex, or possibly symptomatic status. It is also possible that many US physicians still rely on evidence from the stent-favorable Stenting and Angioplasty With Protection in Patients at High Risk for Endarterectomy (SAPPHIRE) trial.18 The SAPPHIRE trial compared CEA with CAS in patients older than 80 years who were considered high risk for surgery and found CAS to be noninferior to CEA. CAS utilization in the United States in 2005 nearly doubled after SAPPHIRE was published in 2004 but decreased by 22% immediately after publication of the CEA-favorable Endarterectomy vs Angioplasty in Patients with Symptomatic Severe Carotid Stenosis (EVA-3S) and Stent-Protected Angioplasty vs Carotid Endarterectomy (SPACE) trials in 2007.19 In the present study, we found no change in CAS utilization in the first 2 years after CREST. A recent study evaluating carotid revascularization in the province of Ontario, Canada, from 2002 to 2014 also noted an increase in CAS performance after publication of SAPPHIRE, but no change in CAS rate after CREST was released in 2010. A similar trend may exist in the United States.20
Furthermore, the differential efficacy of CAS compared with CEA by age in CREST was primarily due to stroke events, but CAS technology has improved over the past decade. Altered configuration of the aortic arch, vessel tortuosity, increased plaque length, and plaque instability associated with age are believed to be possible factors contributing to increased stroke risk with CAS in patients older than 70 years.7,21 In CREST, distal emboli protection devices were utilized, but newer stents with improved deliverability and using proximal emboli protection devices that may allow for easier navigation of tortuous extracranial vessels and lower likelihood of distal embolization are now available.22 It is arguable that stroke risk associated with newer devices and stenting techniques may approximate the stroke risk associated with CEA23 and hence positively influence CAS utilization.
Existing referral systems for carotid revascularizations and subspecialty of physicians performing these procedures may also be influencing revascularization patterns. In this study, although the absolute number of CEAs declined over time, we found a fairly steady volume of CAS revascularizations during the study period. Cardiologists and interventionalists may prefer CAS24 while neurologists and surgeons may be more cautious about this procedure.25 A study of elderly Medicare beneficiaries from 2005 to 2007 found that more than 50% of CAS procedures in this population were done by cardiologists for asymptomatic disease and hospital referral regions where cardiologists performed most revascularizations had higher population-based CAS utilization rates.26Quiz Ref ID Another study of Medicare beneficiaries from 2002 to 2010 also noted that cardiologists performed 49% of all CAS in 2010 and that regional utilization of carotid revascularization procedures increased with greater regional density of cardiologists.24 Cardiologists care for multiple comorbidities that are increasingly prevalent in elderly patients and are more likely to also manage their atherosclerotic diseases.
Although a survey of patients with carotid artery stenosis showed that patients older than 70 years requiring revascularization prefer CEA compared with CAS when presented with information on both procedures,27 it is possible that changing patient preferences or alterations in how information on both procedures is presented to patients over time may also account for some of the observed changes. CAS is less invasive compared with CEA and can be done under local anesthesia. Increased utilization of mechanical thrombectomy to treat acute stroke in patients with symptomatic stenosis may also potentially lead to increased CAS use over time as both procedures may be done concurrently. In this study, the proportion of symptomatic patients undergoing mechanical thrombectomy during the same hospitalization increased from 4.2% in the pre-CREST period to 7.9% in the post-CREST period (eTable 1 in the Supplement).
It is not surprising that CAS use in patients younger than 70 years continued to increase after CREST, but we also noted a decline in CAS proportion in this population from 2013 to 2014. Although this decrease may appear surprising at first glance, this period coincides with the time of closure of some of the postmarketing surveillance registries for stents. The Centers for Medicare & Medicaid Services in the United States will not reimburse for CAS except in patients classified as high risk or as part of a postmarketing surveillance registry.28 The decreased CAS proportion in this age group combined with the decrease in patients older than 70 years from 2013 to 2014 indicates that reimbursement policies may also be playing a role in revascularization patterns. Recent changes to the American Heart Association guideline recommendations that occurred in 2014 may also have contributed to the declining CAS utilization in patients older than 70 years from 2013 to 2014.29
In this retrospective analysis of an ICD-9–based database, we are unable to exclude inaccuracies due to coding errors. However, such errors are likely to be random and bias effect estimates toward the null. Our results should be viewed with caution since many of the patients included in this study may not have been eligible for enrollment in CREST. We are unable to provide information on congestive heart failure, ejection fraction, degree of peripheral vascular disease, contralateral carotid occlusion, degree of stenosis, and other plaque characteristics that may influence revascularization choice. We cannot provide information on carotid plaque pathophysiology, including atherosclerotic vs radiation or dissection related, owing to inherent limitations in our database. We are also unable to provide information on the timing of revascularization in patients with symptomatic carotid artery stenosis and on the specific approach adopted by surgeons performing CAS (transradial vs transfemoral). Although we controlled for an extensive list of covariates in our multivariable analysis, we cannot exclude residual confounding due to uncontrolled factors. Although to our knowledge this is the longest post-CREST study to date, declining use of CAS from 2013 to 2014 indicates that an even more extended post-CREST follow-up will be needed to understand future projections and directions of revascularization patterns. We are unable to accurately identify the subspecialty of physicians performing each revascularization procedure.
Among elderly adults (>70 years) in the United States, the odds of utilization of CAS as the method of carotid revascularization increased in 2011-2014 compared with 2007-2010. These findings are inconsistent with the results of CREST and suggest possible slow and incomplete incorporation of the trial results into clinical practice owing to interplay of a variety of factors.
Accepted for Publication: September 8, 2017.
Corresponding Author: Fadar Oliver Otite, MD, ScM, Department of Neurology, C215, University of Miami Miller School of Medicine, 1120 NW 14th St, Miami, FL 33136 (firstname.lastname@example.org).
Published Online: December 4, 2017. doi:10.1001/jamaneurol.2017.3496
Author Contributions: Dr Otite had full access to all of the study data and takes responsibility for the integrity and accuracy of the data analysis.
Study concept and design: Otite, Malik, Chaturvedi.
Acquisition, analysis, or interpretation of data: Otite, Khandelwal, Chaturvedi.
Drafting of the manuscript: Otite.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Otite.
Administrative, technical, or material support: Khandelwal, Malik.
Study supervision: Khandelwal, Malik, Chaturvedi.
Conflict of Interest Disclosures: Dr Chaturvedi serves on the executive committee of the Carotid Revascularization Endarterectomy vs Stenting Trial 2 and Asymptomatic Carotid Trial 1 studies. No other disclosures are reported.