Annual procedure rates were calculated by dividing the number of unique patients undergoing the specified carotid revascularization procedure in each year by the corresponding number of beneficiary-years of fee-for-service Medicare enrollment for all beneficiaries in that year within subgroups for age, sex, and race. The shaded area represents the 95% CI. The lines were smoothed using the LOESS method (local regression).
A spatial mixed model with a Poisson link function and adjustment for age, sex, and race was fit for each period to calculate and map annualized county-specific risk-standardized rates for carotid endarterectomy and carotid artery stenting. The US counties are shaded from green (lowest rate) to red (highest rate) according to the carotid revascularization rate per 100 000 beneficiary-years. The scales are specific to the procedure and period, and they reflect 6 equal quantiles based on the data. The carotid endarterectomy rates ranged from 77 to 735 per 100 000 beneficiary-years for 1999-2000 and from 41 to 351 per 100 000 beneficiary-years for 2013-2014. The carotid artery stenting rates ranged from 11 to 263 per 100 000 beneficiary-years in 1999-2000 and from 15 to 209 per 100 000 beneficiary-years for 2013-2014. Areas shaded white indicate insufficient data that precluded rate calculations (not calculable). Data from Puerto Rico were used to estimate the national county-level carotid revascularization rates, but they are not included in the maps. For Puerto Rico, the median risk-standardized annual procedure rate for carotid endarterectomy was 195 (interquartile range [IQR], 170-240) per 100 000 beneficiary-years in 1999-2000 and 121 (IQR, 110-132) per 100 000 beneficiary-years in 2013-2014 vs 38 (IQR, 33-49) per 100 000 beneficiary-years in 1999-2000 and 42 (IQR, 40-46) per 100 000 beneficiary-years in 2013-2014 for carotid artery stenting.
Mixed models with hospital-specific random intercepts, an interval time variable, and adjustment for demographic characteristics, comorbid conditions, and symptomatic status were used to calculate adjusted odds ratios (for in-hospital mortality and 30-day all-cause mortality) and hazard ratios (for 30-day ischemic stroke or death, 30-day ischemic stroke, myocardial infarction, or death, and 1-year ischemic stroke) representing annual change in outcomes after carotid endarterectomy and carotid artery stenting. The time variable was transformed to reflect the percentage annual reduction in outcome by subtracting the odds ratio or hazard ratio from the null value of 1.0; a positive number indicates a decline in the adverse outcome. The forest plots summarize these annual percentage reductions and corresponding 95% CIs for the overall carotid endarterectomy and stenting procedure populations and for subgroups defined by sex, race, and age (the data marker and 95% CI values appear in eTable 5 in the Supplement).
eTable 1. Patient Comorbid Conditions and Medical History Variables Included in Risk-Adjustment Models for Outcomes After Carotid Endarterectomy, 1999-2014
eTable 2. Patient Comorbid Conditions and Medical History Variables Included in Risk-Adjustment Models for Outcomes After Carotid Artery Stenting, 1999-2014
eTable 3. National Carotid Endarterectomy Rates by Sex, Race, and Age, 1999-2014
eTable 4. National Carotid Artery Stenting Rates by Sex, Race, and Age, 1999-2014
eTable 5. Adjusted Annual Percentage Reductions and 95% Confidence Intervals for Outcomes After Carotid Revascularization: Overall and by Patient Demographic Subgroups, 1999-2014
eTable 6. Observed Outcomes After Carotid Endarterectomy by Symptomatic Status, 1999-2014
eTable 7. Observed Outcomes After Carotid Artery Stenting by Symptomatic Status, 1999-2014
eTable 8. Carotid Artery Stenting Outcomes for Patients Identified by International Classification of Diseases, Ninth Revision, Clinical Modification Procedure Codes 00.61, 00.63, and 00.64, 2005-2014
eFigure 1. Top Five International Classification of Diseases, Ninth Revision, Clinical Modification Principal Discharge Diagnosis Codes for Patients Undergoing Carotid Revascularization in 1999 and 2014
eFigure 2. National Carotid Revascularization Rates per 100,000 Beneficiary-Years by Sex, Race, and Age, 1999-2014
eFigure 3. Adjusted Annual Percentage Reductions and 95% Confidence Intervals for Outcomes After Carotid Revascularization by Symptomatic Status, 1999-2014
Customize your JAMA Network experience by selecting one or more topics from the list below.
Lichtman JH, Jones MR, Leifheit EC, et al. Carotid Endarterectomy and Carotid Artery Stenting in the US Medicare Population, 1999-2014. JAMA. 2017;318(11):1035–1046. doi:10.1001/jama.2017.12882
Have there been changes in the performance and outcomes of carotid endarterectomy and carotid artery stenting between 1999 and 2014?
In this serial cross-sectional analysis of fee-for-service Medicare beneficiaries, the national annualized rate of carotid endarterectomy per 100 000 beneficiary-years decreased from 298 in 1999-2000 to 128 in 2013-2014, whereas the rate of carotid artery stenting increased initially from 40 in 1999-2000 to 75 in 2005-2006 followed by a decline to 38 in 2013-2014. Periprocedural and 1-year outcomes improved from 1999 to 2014.
Performance of carotid revascularization procedures among fee-for-service Medicare patients decreased from 1999 to 2014.
Carotid endarterectomy and carotid artery stenting are the leading approaches to revascularization for carotid stenosis, yet contemporary data on trends in rates and outcomes are limited.
To describe US national trends in performance and outcomes of carotid endarterectomy and stenting among Medicare beneficiaries from 1999 to 2014.
Design, Setting, and Participants
Serial cross-sectional analysis of Medicare fee-for-service beneficiaries aged 65 years or older from 1999 to 2014 using the Medicare Inpatient and Denominator files. Spatial mixed models adjusted for age, sex, and race were fit to calculate county-specific risk-standardized revascularization rates. Mixed models were fit to assess trends in outcomes after adjustment for demographics, comorbidities, and symptomatic status.
Carotid endarterectomy and carotid artery stenting.
Main Outcomes and Measures
Revascularization rates per 100 000 beneficiary-years of fee-for-service enrollment, in-hospital mortality, 30-day stroke or death, 30-day stroke, myocardial infarction, or death, 30-day all-cause mortality, and 1-year stroke.
During the study, 937 111 unique patients underwent carotid endarterectomy (mean age, 75.8 years; 43% women) and 231 077 underwent carotid artery stenting (mean age, 75.4 years; 49% women). There were 81 306 patients who underwent endarterectomy in 1999 and 36 325 in 2014; national rates per 100 000 beneficiary-years decreased from 298 in 1999-2000 to 128 in 2013-2014 (P < .001). The number of patients who underwent stenting ranged from 10 416 in 1999 to 22 865 in 2006 (an increase per 100 000 beneficiary-years from 40 in 1999-2000 to 75 in 2005-2006; P < .001); by 2014, there were 10 208 patients who underwent stenting and the rate decreased to 38 per 100 000 beneficiary-years (P < .001). Outcomes improved over time despite increases in vascular risk factors (eg, hypertension prevalence increased from 67% to 81% among patients who underwent endarterectomy and from 61% to 70% among patients who underwent stenting) and the proportion of symptomatic patients (all P < .001). There were adjusted annual decreases in 30-day ischemic stroke or death of 2.90% (95% CI, 2.63% to 3.18%) among patients who underwent endarterectomy and 1.13% (95% CI, 0.71% to 1.54%) among patients who underwent stenting; an absolute decrease from 1999 to 2014 was observed for endarterectomy (1.4%; 95% CI, 1.2% to 1.5%) but not stenting (−0.1%; 95% CI, −0.5% to 0.4%). Rates for 1-year ischemic stroke decreased after endarterectomy (absolute decrease, 3.5% [95% CI, 3.2% to 3.7%]; adjusted annual decrease, 2.17% [95% CI, 2.00% to 2.34%]) and stenting (absolute decrease, 1.6% [95% CI, 1.2% to 2.1%]; adjusted annual decrease, 1.86% [95% CI, 1.45%-2.26%]). Additional improvements were noted for in-hospital mortality, 30-day stroke, myocardial infarction, or death, and 30-day all-cause mortality as well as within demographic subgroups.
Conclusions and Relevance
Among fee-for-service Medicare beneficiaries, the performance of carotid endarterectomy declined from 1999 to 2014, whereas the performance of carotid artery stenting increased until 2006 and then declined from 2007 to 2014. Outcomes improved despite increases in vascular risk factors.
Carotid endarterectomy and carotid artery stenting are the most frequently performed revascularization procedures to prevent stroke.1 Randomized clinical trials have demonstrated the benefits of carotid revascularization among patients with recent minor stroke or transient ischemic attack ipsilateral to the stenosis2,3 and for select patients with no recent attributable symptoms.4-6 In addition, new tools for the medical management of vascular risk factors have emerged that may have lessened the underlying burden of atherosclerotic disease.7-10 Contemporary information is limited for patients not enrolled in clinical trials regarding national trends in the performance and outcomes of carotid revascularization in the United States, particularly among demographic subgroups.11-14
Accordingly, for the 16-year period from 1999 to 2014, trends in carotid endarterectomy and carotid artery stenting and in-hospital, 30-day, and 1-year outcomes for Medicare beneficiaries were assessed. Because of the paucity of information regarding postdischarge outcomes for demographic subgroups, outcome trends were also evaluated for age, sex, and race subgroups. In addition, geographic patterns of carotid revascularization were examined.
In this serial cross-sectional analysis, US Centers for Medicare & Medicaid Services (CMS) Medicare Denominator data were used to identify beneficiaries aged 65 years or older enrolled in fee-for-service Medicare for 1 month or longer between January 1999 and December 2014. We determined the number of beneficiaries each year and calculated beneficiary-years of fee-for-service enrollment to account for new enrollment, disenrollment, or death. Data for beneficiary-years were linked to Medicare Inpatient claims to identify patients who underwent carotid endarterectomy (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] code 38.12) or carotid artery stenting (ICD-9-CM code 00.61, 00.63, or 00.64; or ICD-9-CM noncoronary stenting code 39.50 or 39.90 with co-occurring cerebrovascular disease code 433.10, 433.11, 433.1, 433.30, 433.31, or 435.9)15 in US acute care hospitals.
Among patients with multiple carotid procedures during the study period, we selected the first procedure as the index admission. Patients could only contribute data for the calculation of procedure rates and outcomes for this first procedure. We further quantified the number of patients who underwent a second carotid revascularization procedure within 1 year.
Patients were excluded if they underwent both carotid endarterectomy and carotid artery stenting during the index hospitalization or received any other concomitant major interventions (eg, coronary artery bypass grafting) during the index admission. Institutional review boards at Rutgers University and Yale University approved the study and waived informed consent for these analyses.
Patient demographic information (age, sex, and race/ethnicity) was obtained from Medicare Denominator data, which are drawn from the Medicare enrollment database. Race/ethnicity was analyzed because prior studies have suggested racial/ethnic status may modify the benefits and risks of carotid procedures.16,17 Clinical information and comorbidities (a complete list appears in eTables 1-2 in the Supplement) were extracted from the Medicare Inpatient data. Specifically, the list of comorbidities is based on the secondary diagnoses at the index admission and the diagnosis and procedure codes from the 12-month preindex claims, according to Hierarchical Condition Category coding.18 Patients were considered symptomatic if they had an ICD-9-CM principal discharge diagnosis code indicating occlusion or stenosis of the precerebral or cerebral arteries with cerebral infarction (433.11, 433.31, 434.01, 434.11, or 434.91) or a secondary diagnosis code indicating prior stroke (342.xx or 438.xx), transient ischemic attack (435.x or 781.4), or amaurosis fugax (362.34 or 368.12).15
Annual procedure rates were calculated by dividing the number of unique patients who underwent carotid endarterectomy or carotid artery stenting each year by the corresponding number of beneficiary-years that year within the subgroups of age, sex, and race.
Clinical outcomes included in-hospital mortality, 30-day all-cause mortality, 30-day ischemic stroke or death, 30-day ischemic stroke, myocardial infarction, or death, and 1-year ischemic stroke.19,20 We also assessed 1-year all-cause mortality and a composite measure of readmission for ischemic stroke, transient ischemic attack, or a second carotid revascularization procedure (either carotid endarterectomy or carotid artery stenting). Death and time to second procedure were determined from the hospital admission date for the index procedure (for 30-day all-cause mortality and the 30-day and 1-year composites), and ischemic stroke (ICD-9-CM principal discharge diagnosis code 433.xx, 434.xx, or 436.xx), myocardial infarction (ICD-9-CM code 410.xx except 410.x2), and transient ischemic attack (ICD-9-CM code 435.x) events were determined from the discharge date.
Process of care outcomes included discharge disposition, Medicare payment for the index hospitalization (adjusted for inflation using the consumer price index21 with 2014 as the reference year), and length of hospital stay.
A Poisson link function and county-specific random intercepts were used to model the number of carotid endarterectomy and carotid artery stenting procedures as a function of patients’ age, sex, and race and accounting for geographic differences between counties. A spherical covariate structure was included in the models to account for spatial autocorrelation. County-level geographic differences were considered because factors such as lifestyle, access to care, and local care practices vary across counties and may affect receipt of procedures. Data were combined for 1999-2000 and 2013-2014 to increase the sample sizes at the county level. Using these models, the annualized risk-standardized procedure rates for each county for the 2 periods were mapped and counties were shaded with a gradient from green to red (lowest rates to highest). To further assess geographic variation over time, the weighted Pearson correlation coefficient was calculated for county-specific procedure rates in 1999-2000 and 2013-2014. The Mantel-Haenszel χ2 test was used to assess temporal changes in patient characteristics and observed outcomes.
We fit mixed models with a logit link function and hospital-specific random intercepts to evaluate temporal trends in in-hospital and 30-day all-cause mortality. To assess change in rates of 30-day ischemic stroke or death, 30-day ischemic stroke, myocardial infarction, or death, and 1-year ischemic stroke, Cox proportional hazards models were constructed with censoring for death. The proportional hazards assumption for the Cox models was verified.
All models were adjusted for patient demographics, comorbidities, and symptomatic status and included an interval time variable corresponding to each study year (1999 [time = 0] through 2014 [time = 15]) to represent the annual change in outcome. Odds ratios were used for annual change in in-hospital and 30-day mortality. Hazard ratios were used for 30-day ischemic stroke or death, 30-day ischemic stroke, myocardial infarction, or death, and 1-year ischemic stroke. Estimates for the time variable were transformed to reflect the percentage annual reduction in outcome by subtracting the odds ratio or hazard ratio from the null value of 1.0. To permit complete follow-up, we restricted the 1-year analyses to discharges through 2013 and the 30-day analyses to discharges through November 30, 2014.
For the primary analyses, we fit separate models for each procedure and repeated the models for prespecified subgroups defined by age, sex, and race. Sensitivity analyses were conducted to assess trends in outcomes by symptomatic status and among patients who underwent carotid artery stenting with an ICD-9-CM procedure code of 00.61, 00.63, or 00.64 (limited to 2005-2014 due to introduction of the codes and CMS expansion of carotid artery stenting reimbursement).
Patient characteristics and outcomes are reported in 2-year intervals. Analyses were conducted using SAS version 9.4 (SAS Institute Inc). Statistical tests used a 2-sided α of .05. Corrections were not made for multiplicity of outcomes to address the potential for type I error; therefore, the analyses should be considered exploratory.
There were 937 111 unique patients who underwent carotid endarterectomy (mean age, 75.8 years; 43% women) and 231 077 unique patients who underwent carotid artery stenting (mean age, 75.4 years; 49% women) during the 16-year study. The number of patients who underwent carotid endarterectomy ranged from 81 306 in 1999 to 36 325 in 2014. The number of patients who underwent carotid artery stenting ranged from 10 416 in 1999 to 22 865 in 2006 to 10 208 in 2014 (Table 1, Table 2, and Figure 1). For both procedures, there was an increase in the proportion of symptomatic patients over time (P < .001; Tables 1-2). For carotid artery stenting, the percentage of symptomatic patients varied from 14.4% in 1999-2000 to 25.9% in 2013-2014. The prevalence of several comorbidities and surgical risk factors (eg, hypertension, kidney failure, depression, and diabetes) increased from 1999 to 2014 (all P < .001; Tables 1-2).
The most frequent principal discharge diagnosis for patients who underwent carotid endarterectomy was occlusion and stenosis of the precerebral arteries (ICD-9-CM code 433.xx), comprising 90.8% of diagnoses in 1999 and 93.6% in 2014 (eFigure 1A in the Supplement). For patients who underwent carotid artery stenting, occlusion and stenosis of the precerebral arteries and atherosclerosis (ICD-9-CM codes 433.xx and 440.xx) accounted for 37.7% of the diagnoses in 1999 and 58.1% in 2014 (eFigure 1B in the Supplement).
The national carotid endarterectomy rate decreased from 298 (95% CI, 297-300) per 100 000 beneficiary-years in 1999-2000 to 128 (95% CI, 127-129) per 100 000 beneficiary-years in 2013-2014 (P < .001; Table 1 and Figure 1). Although rates varied across demographic subgroups, patterns were consistent over time (eTable 3 and eFigure 2 in the Supplement). Carotid endarterectomy was performed in 2527 hospitals in 1999 and in 1914 in 2014. The annual volume per hospital ranged from a median of 18 (interquartile range [IQR], 6-45) to 14 (IQR, 5-29) procedures over the 16-year study.
Procedure rates varied across the United States. In 1999-2000, the annualized county-specific carotid endarterectomy rates adjusted for age, sex, and race ranged from 77 to 735 per 100 000 beneficiary-years (Figure 2A). Southern and central US regions had the highest adjusted rates. Regional variation persisted in 2013-2014, with rates from 41 to 351 per 100 000 beneficiary-years. The weighted Pearson correlation between rates in 1999-2000 and 2013-2014 was 0.44 (P < .001), and there was no change in rank over time for 52% of counties in the lowest 2 rate quantiles in 1999-2000 and 53% of counties in the highest 2 rate quantiles.
National carotid artery stenting rates increased between 1999-2000 and 2005-2006 from 40 to 75 per 100 000 beneficiary-years (P < .001) but later decreased annually to 38 per 100 000 beneficiary-years in 2013-2014 (P < .001; Table 2 and Figure 1). There was variation among demographic subgroups, but temporal patterns were generally consistent (eTable 4 and eFigure 2 in the Supplement). Carotid artery stenting was performed in 1738 hospitals in 1999, 1821 in 2006, and 1608 in 2014. There was a median of 4 (IQR, 1-8) carotid artery stenting procedures performed per hospital in 1999, 7 (IQR, 3-19) in 2006, and 4 (IQR, 2-11) in 2014.
Annualized county-specific rates for carotid artery stenting adjusted for age, sex, and race varied geographically from 11 to 263 per 100 000 beneficiary-years in 1999-2000 to 15 to 209 per 100 000 beneficiary-years in 2013-2014 (Figure 2B). Rates in 1999-2000 were highest in the southern and central regions, with additional peaks in California and in parts of the southwest. In 2013-2014, rates remained highest for the southern and central regions. The weighted Pearson correlation between rates in 1999-2000 and 2013-2014 was 0.20 (P < .001), and there was no change in rank over time for 43% of counties in the lowest 2 quantiles in 1999-2000 and 42% of counties in the highest 2 quantiles.
During the 16-year study period, 69 390 patients (7.4%) underwent carotid endarterectomy as the index procedure and 14 004 (6.1%) underwent carotid artery stenting as the index procedure and then underwent a second carotid revascularization procedure within 1 year of their index procedure. The median time to the second procedure was 46 (IQR, 28-75) days after carotid endarterectomy and 53 (IQR, 27-121) days after carotid artery stenting. A total of 817 patients (1.2%) who underwent carotid endarterectomy and 406 (2.9%) who underwent carotid artery stenting were hospitalized for ischemic stroke or transient ischemic attack between the first and second procedures.
Thirty-day outcomes after carotid endarterectomy improved during the study years. The composite of ischemic stroke or death decreased from 4.4% in 1999-2000 to 3.1% in 2013-2014 (absolute decrease, 1.4% [95% CI, 1.2%-1.5%]) and all-cause mortality decreased from 1.6% to 1.1% (absolute decrease, 0.5% [95% CI, 0.4%-0.6%]; Table 1). These improvements were significant in adjusted analyses, with annual reductions of 2.90% (95% CI, 2.63%-3.18%) for 30-day ischemic stroke or death and 2.97% (95% CI, 2.56%-3.37%) for 30-day all-cause mortality (Figure 3A and eTable 5 in the Supplement). Patterns were similar across prespecified demographic subgroups, with the exception of all-cause mortality for those with race/ethnicity of other.
There were improvements between 1999-2000 and 2013-2014 in in-hospital mortality (absolute decrease, 0.5% [95% CI, 0.4%-0.5%]; adjusted annual reduction, 5.24% [95% CI, 4.69%-5.79%]), 30-day ischemic stroke, myocardial infarction, or death (absolute decrease, 1.4% [95% CI, 1.2%-1.6%]; adjusted annual reduction, 2.73% [95% CI, 2.46%-2.99%]), and 1-year ischemic stroke (absolute decrease, 3.5% [95% CI, 3.2%-3.7%]; adjusted annual reduction, 2.17% [95% CI, 2.00%-2.34%]; Table 1, Figure 3A, and eTable 5 in the Supplement). These reductions were statistically significant across demographic subgroups.
There were statistically significant adjusted annual reductions in all outcomes among asymptomatic patients, whereas among symptomatic patients, there were reductions for everything except 30-day all-cause mortality. The adjusted annual reductions were larger among asymptomatic patients compared with symptomatic patients (eTable 6 and eFigure 3 in the Supplement).
For carotid artery stenting, outcome patterns were more complex (Table 2). Rates of 30-day ischemic stroke or death decreased from 7.0% to 4.8% between 1999-2000 and 2005-2006 and then increased to 7.0% by 2013-2014. The absolute decrease in the rates of 30-day ischemic stroke or death between 1999-2000 and 2013-2014 (−0.1%; 95% CI, −0.5% to 0.4%) was not statistically significant. Similar results were observed for in-hospital mortality (absolute decrease, 0.5%; 95% CI, 0.2% to 0.8%), 30-day all-cause mortality (absolute decrease, −0.1%; 95% CI, −0.5% to 0.3%), and 30-day ischemic stroke, myocardial infarction, or death (absolute decrease, −0.1%; 95% CI, −0.6% to 0.4%). One-year ischemic stroke decreased during the study years from 8.1% to 6.3% (absolute decrease, 1.6%; 95% CI, 1.2%-2.1%).
With adjustment, there were statistically significant annual reductions of 2.78% (95% CI, 2.12%-3.43%) for in-hospital mortality, 1.13% (95% CI, 0.71%-1.54%) for 30-day ischemic stroke or death, 1.06% (95% CI, 0.66%-1.46%) for 30-day ischemic stroke, myocardial infarction, or death, 1.43% (95% CI, 0.92%-1.93%) for 30-day all-cause mortality, and 1.86% (95% CI, 1.45%-2.26%) for 1-year ischemic stroke (Figure 3B and eTable 5 in the Supplement). These reductions remained significant across demographic subgroups with a few exceptions. There were no statistically significant reductions in the 30-day outcomes that included ischemic stroke among black patients and those aged 85 years or older, 30-day all-cause mortality among those aged 75 to 84 years, or 1-year ischemic stroke among those aged 85 years or older.
When stratified by symptomatic status, there were statistically significant adjusted annual reductions, ranging from 1.58% to 2.66%, in all outcomes among asymptomatic patients, but reductions only reached statistical significance for in-hospital mortality and 1-year ischemic stroke among symptomatic patients (eTable 7 and eFigure 3 in the Supplement). In sensitivity analyses of patients who underwent carotid artery stenting with an ICD-9-CM procedure code of 00.61, 00.63, or 00.64, the outcomes of observed ischemic stroke, myocardial infarction, and mortality followed patterns similar to the larger carotid artery stenting cohort; however, annual trends only reached statistical significance for 1-year ischemic stroke (eTable 8 in the Supplement).
Between 1999-2000 and 2013-2014, the median hospital length of stay for the index hospitalization decreased from 2 days to 1 day among patients who underwent carotid endarterectomy (Table 1). Among patients who underwent carotid artery stenting, the median length of stay decreased from 2 days to 1 day between 1999-2000 and 2011-2012 but then returned to 2 days by 2013-2014 (Table 2). The inflation-adjusted median Medicare payment for the index hospitalization decreased during the study years from $8278 to $6779 for carotid endarterectomy (Table 1) but increased from $12 963 to $14 796 for carotid artery stenting (Table 2). For both procedures, the rates of discharge to home decreased, whereas discharge to home care and skilled nursing or intermediate care facilities increased (Tables 1-2). Results stratified by symptomatic status appear in eTables 6-7 in the Supplement.
Quiz Ref IDIn this nationwide 16-year study of fee-for-service Medicare beneficiaries, there was a reduction in the performance of carotid revascularization and improvement in postprocedure outcomes. Carotid endarterectomy rates decreased annually from 1999 to 2014, whereas carotid artery stenting rates increased from 1999 to 2006 and then decreased from 2007 to 2014. Despite lower hospital procedure volume over time, there were annual improvements in mortality and ischemic stroke outcomes for both procedures after accounting for demographic characteristics, comorbidities, and symptomatic status. Reductions in adverse outcomes were observed for both sexes and within most age and race subgroups.
The reasons for the temporal changes in the performance of carotid endarterectomy and carotid artery stenting merit consideration. Carotid artery stenting availability may have contributed to the reduction in carotid endarterectomy during early years of the study. Consistent with prior work, carotid artery stenting rates increased through 2006,13,22 presumably because of the expansion of Medicare reimbursement and the publication of clinical trial results.14,23 A study from Ontario, Canada, found that overall use of carotid revascularization from 2002 to 2014 among patients aged 40 years or older decreased from 6.0 to 4.3 procedures per 100 000 population, largely driven by the decrease in carotid endarterectomy from 5.6 to 3.6 procedures per 100 000; carotid artery stenting increased from 0.39 to 0.67 per 100 000.14
In the current study, carotid artery stenting rates increased from 1999 to 2006 but subsequently declined, falling to baseline levels by 2014. The overall reduction in revascularization suggests that the availability of carotid artery stenting is not solely responsible for the decline in carotid endarterectomy. Likewise, the decline of 1.6% in the proportion of asymptomatic patients during the study period is not a feasible explanation for the decline of 57% in the rate of carotid endarterectomy (from 298 to 128 per 100 000 beneficiary-years).
Quiz Ref IDBetter clinical management of vascular risk factors could have affected the performance of carotid revascularization via biological and medical practice mechanisms. Even though hypertension prevalence among adults has remained unchanged from 1999 to 2008, US data show age-adjusted relative increases of 16% for hypertension awareness, 21% for reported treatment, and 53% for blood pressure control (<140 mm Hg for systolic and <90 mm Hg for diastolic blood pressure).10 Increased treatment and control of dyslipidemia, particularly with statins, also has been reported.7,9 The estimated number of adults reporting prescription statin use increased from an average of 17.6 million annually in 2000-2001 to 40.8 million in 2010-2011, according to the nationally representative Medicare Expenditure Panel Survey.9Quiz Ref ID In addition, health behaviors have improved, with increases in exercise frequency and decreases in smoking.7,24 These advances may have reduced the population-level prevalence of carotid stenosis that typically warrants revascularization. A decline in the atherosclerotic burden is also supported by the 28% decline in the age-adjusted US stroke rate from 54.6 per 100 000 in 2003 to 39.1 per 100 000 in 2010.25
Quiz Ref IDBecause of advances in medical therapy, physicians may less frequently recommend revascularization for patients with asymptomatic carotid disease. Asymptomatic patients represent the majority of those undergoing revascularization. If their stroke risk is low, the perceived need to intervene by referring physicians could have been affected. Recent observational studies suggested an annual stroke risk of 1% or less for asymptomatic patients,26,27 which is about half the risk rate reported in clinical trials initiated in the 1980s and 1990s.4,6
Changes in surgical reimbursement policies over time may have also contributed to the reported patterns.23,28 As carotid endarterectomy rates declined from 1999 to 2014, CMS reimbursement also decreased. Performance of carotid artery stenting increased from 1999 to 2006 and then decreased annually through 2014. Prior to 2005, the CMS did not reimburse carotid artery stenting when patients were treated outside approved clinical trials and postapproval studies.23 In March 2005, the CMS determined that carotid artery stenting coverage could be provided for symptomatic patients with 70% or greater stenosis who were at high risk to undergo carotid endarterectomy.28 Thereafter, the CMS has not reimbursed the costs of carotid artery stenting for symptomatic patients meeting the conventional risk profile nor asymptomatic patients with conventional or high risk treated outside an approved clinical trial or postapproval study. These restrictions for asymptomatic patients may have played a role in the decreasing proportion of asymptomatic patients undergoing carotid artery stenting, and the higher proportion of symptomatic patients could have attenuated outcome improvements.
From 2008 to 2010, reports from 4 large randomized trials comparing carotid endarterectomy vs carotid artery stenting were published.20,29-31 These trials showed a safety advantage for carotid endarterectomy with regard to periprocedural 30-day stroke risk, and subsequent treatment guidelines have favored carotid endarterectomy for most symptomatic and asymptomatic patients.32
Annual improvements in carotid endarterectomy and carotid artery stenting outcomes within demographic subgroups suggest improved safety for older individuals (aged ≥65 years). The adjusted annual percentage reductions in 30-day outcomes after carotid endarterectomy were similar across age groups. Annual reductions were comparable across sexes and race/ethnicity for both procedures, with the exception of black patients who underwent carotid artery stenting and had no significant annual reduction in the 30-day composite outcomes that included stroke. These findings extend prior research that compared overall outcomes by demographic characteristics16,17,33,34 by assessing whether improvements were seen within subgroups over time.
Clinical trial results completed during an overlapping period confirm steady declines in revascularization complications but provide limited information on patterns by demographic characteristics.20,29-31,35,36 A systematic review of carotid endarterectomy perioperative safety, which included trials and community registries, found an overall decrease in postoperative stroke or death for asymptomatic patients over a 30-year period but did not assess patterns within demographic subgroups.8 Analyses from administrative data sets confirm an overall decline for in-hospital, 30-day, and 1-year mortality, but these analyses do not report data stratified by demographic characteristics.1,11,15 Assessing safety and outcomes in the oldest old (aged ≥85 years) is important given the aging demographics of individuals in the United States, which will increase the population at risk for atherosclerotic disease.37
The mechanisms mediating improved outcomes following revascularization are not clear. Criteria for case selection may have changed, favoring procedures for individuals at lower risk for adverse outcomes (ie, those who are younger, asymptomatic, and have a lower comorbidity burden). However, there was an overall increase in the proportion of patients with symptomatic disease, hypertension, diabetes, and kidney failure. Higher case volume has been associated with improved postprocedural outcomes,38,39 but there was a decline in the average number of procedures performed at hospitals during the study period.
Quiz Ref IDFor carotid endarterectomy, improvements in procedural technique, perioperative care, implementation of quality-control audits, and surgical and anesthesia credentialing may have contributed to improved outcomes.8 In contrast, carotid artery stenting was an evolving procedure during the study, and improved outcomes could reflect new stent designs, acceptance of embolic protection devices, and technological improvements.40 Hospital length of stay decreased and a greater proportion of patients were discharged to nursing facilities or home health care during the study, potentially confirming a greater emphasis on postdischarge care. In addition, patients may have been at lower postsurgical risk in more recent years, given improved risk factor management.7,9,10
This study has several limitations. First, analyses of Medicare data are subject to coding errors, changes in coding patterns over time, and lack of detailed clinical information on the procedure, anatomical details (eg, stroke etiology, ipsilateral vs contralateral location), intraprocedural practices, chronology of events during the hospitalization, and patient preferences. Second, available comorbidity data may not adequately characterize the medical complexity of patients; however, the performance of models using administrative databases was comparable with that of models using medical records for short-term outcomes in prior work.18 Third, compared with other patients who underwent carotid revascularization and met CMS criteria, patients who underwent carotid artery stenting while enrolled in industry-sponsored registries during the study may have had more extensive postprocedure assessment of outcomes. Fourth, outpatient and observation stay revascularization procedures, which may have increased during the latter study years, could not be assessed. Fifth, analyses were limited to fee-for-service beneficiaries aged 65 years or older, and results may not generalize to those with Medicare Advantage or to younger patients. However, fee-for-service Medicare provides the largest national database with rates of hospitalization and postdischarge outcomes. Sixth, our analyses included a large number of comparisons, and corrections were not made for multiplicity to address the potential for type I error; accordingly, the analyses should be considered exploratory.
Among fee-for-service Medicare beneficiaries, the performance of carotid endarterectomy declined from 1999 to 2014, whereas the performance of carotid artery stenting increased until 2006 and then declined from 2007 to 2014. Outcomes improved despite increases in vascular risk factors.
Corresponding Author: Judith H. Lichtman, PhD, Yale School of Public Health, PO Box 208034, New Haven, CT 06520 (email@example.com).
Accepted for Publication: August 9, 2017.
Author Contributions: Dr Wang had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Lichtman, Jones, Leifheit, Sheffet, G. Howard, V. Howard, Wang, Curtis, Brott.
Acquisition, analysis, or interpretation of data: Lichtman, Jones, Leifheit, Sheffet, G. Howard, Lal, V. Howard, Wang, Curtis, Brott.
Drafting of the manuscript: Lichtman, Leifheit, Brott.
Critical revision of the manuscript for important intellectual content: Lichtman, Jones, Leifheit, Sheffet, G. Howard, Lal, V. Howard, Wang, Curtis, Brott.
Statistical analysis: Leifheit, G. Howard, Wang.
Obtained funding: Lichtman, Sheffet, Curtis, Brott.
Administrative, technical, or material support: Lichtman, Jones, Sheffet, Curtis, Brott.
Supervision: Lichtman, Brott.
Conflict of Interest Disclosures: The authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr G. Howard reported being a principal investigator of the CREST-2 trial. Dr V. Howard reported receiving grants from the National Institute for Neurological Disorders and Stroke for the CREST trial. Dr Curtis reported receiving salary support under contracts with the American College of Cardiology and the Centers for Medicare & Medicaid Services; and holding equity interest in Medtronic. No other disclosures were reported.
Funding/Support: This study was funded by grant U01 NS038384 from the National Institute for Neurological Disorders and Stroke and grant U01 HL105270 from the National Heart, Lung, and Blood Institute.
Role of the Funder/Sponsor The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Meeting Presentation: This work was previously presented as an abstract at the American Heart Association/American Stroke Association International Stroke Conference; February 17, 2016; Los Angeles, California.
Create a personal account or sign in to: