Journal of Vascular Surgery
Volume 48, Issue 6 , Pages 1442-1450.e1, December 2008

Carotid artery stenting has increased rates of postprocedure stroke, death, and resource utilization than does carotid endarterectomy in the United States, 2005

Presented at the Annual Meeting of the Society for Vascular Surgery, San Diego, Calif, June 5-8, 2008.

  • James T. McPhee, MD

      Affiliations

    • University of Massachusetts Medical School, Department of Surgery, Worcester, Mass
    • ,
  • Andres Schanzer, MD

      Affiliations

    • University of Massachusetts Medical School, Division of Vascular Surgery, Worcester, Mass
    • ,
  • Louis M. Messina, MD

      Affiliations

    • University of Massachusetts Medical School, Division of Vascular Surgery, Worcester, Mass
    • ,
  • Mohammad H. Eslami, MD

      Affiliations

    • University of Massachusetts Medical School, Division of Vascular Surgery, Worcester, Mass
    • Corresponding Author InformationReprint requests: Mohammad H. Eslami, MD, University of Massachusetts Medical School, Division of Vascular Surgery, 55 Lake Avenue North, Worcester, MA 01655

Received 2 June 2008; accepted 10 July 2008. published online 02 October 2008.

Article Outline

Objective

Carotid endarterectomy (CEA) remains the procedure of choice for treatment of patients with severe carotid artery stenosis. The role of carotid artery stenting (CAS) in this patient group is still being defined. Prior single and multicenter studies have demonstrated economic savings associated with CEA compared with CAS. The purpose of this study was to compare surgical outcomes and resource utilization associated with these two procedures at the national level in 2005, the first year in which a specific ICD-9 procedure code for CAS was available.

Methods

All patient discharges for carotid revascularization for the year 2005 were identified in the Nationwide Inpatient Sample based on ICD9-CM procedure codes for CEA (38.12) and CAS (00.63). The primary outcome measures of interest were in-hospital mortality and postoperative stroke; secondary outcome measures included total hospital charges and length of stay (LOS). All statistical analyses were performed using SAS version 9.1 (Cary, NC), and data are weighted according to the Nationwide Inpatient Sample (NIS) design to draw national estimates. Univariate analyses of categorical variables were performed using Rao-Scott χ2, and continuous variables were analyzed by survey weighted analysis of variance (ANOVA). Multivariate logistic regression was performed to evaluate independent predictors of postoperative stroke and mortality.

Results

During 2005, an estimated 135,701 patients underwent either CEA or CAS nationally. Overall, 91% of patients underwent CEA. The mean age overall was 71 years. Postoperative stroke rates were increased for CAS compared with CEA (1.8% vs 1.1%, P < .05), odds ratio (OR) 1.7; (95% confidence interval [CI] 1.2-2.3). Overall, mortality rates were higher for CAS compared with CEA (1.1% vs 0.57%, P < .05) this difference was substantially increased in regard to patients with symptomatic disease (4.6% vs 1.4%, P < .05). By logistic regression, CAS trended toward increased mortality, OR 1.5; (95% CI .96-2.5). Overall, the median total hospital charges for patients that underwent CAS were significantly greater than those that underwent CEA ($30,396 vs $17,658 P < .05).

Conclusions

Based on a large representative sample during the year 2005, CEA was performed with significantly lower in-hospital mortality, postoperative stroke rates, and lower median total hospital charges than CAS in US hospitals. As the role for CAS becomes defined for the management of patients with carotid artery stenosis, clinical as well as economic outcomes must be continually evaluated.

 

Carotid artery stenting (CAS) has emerged as a technically feasible operation with obvious perceptible benefits, most notably the potentially less invasive percutaneous vascular access required to perform the procedure. Experience with other minimally invasive access techniques such as iliac and aortic stent grafts has translated into equivalent or improved operative morbidity1 and mortality2 with concomitant decrease in hospital length of stay and resource utilization. The benefits of CAS, unlike other minimally invasive vascular procedures, have not consistently translated into demonstrably improved clinical outcomes compared with carotid endarterectomy (CEA). In light of disparate results of clinical3, 4, 5, 6 as well as institutional7 and population based trials,8 the indications for CAS remain controversial.

For two procedures to achieve equipoise, in addition to comparable risk-adjusted clinical outcomes, the overall cost should not be significantly different in order to allow equal access across all patient demographics. Kilaru et al, in a decision analysis model, concluded that CAS could become cost effective compared with CEA only if stroke and mortality rates became equivalent.9 Similarly, previous studies have demonstrated that CAS may be associated with increased cost compared directly with CEA at the institutional level.10, 11, 12

Prior to October 2004, no specific International Classification of Diseases Ninth Revision, Clinical Modification (ICD9-CM) procedural code existed for CAS. Therefore, previous works from administrative datasets relied on diagnostic codes and less specific procedural codes to evaluate this procedure, leading to potential inaccuracies.2, 8 Therefore, a contemporary comparative analysis of CAS with CEA in terms of operative mortality and stroke rates limited to 2005, the most recent available data, is lacking. Likewise, a comparison of resource utilization between the two procedures at the national level is also lacking.

Our objective was to perform a national population-based observational study exclusively in a 1-year period (2005) during which a specific ICD9-CM code existed for CAS. With results based on reliably coded data, the postoperative stroke and mortality data as well as periprocedural length of stay, hospital charges, and discharge disposition of symptomatic and asymptomatic patients undergoing CAS and CEA may be helpful to physicians in determining the best treatment for their patients with carotid artery occlusive disease.

Back to Article Outline

Methods 

In order to evaluate surgical outcomes for patients undergoing carotid artery revascularization in the United States, we used the Nationwide Inpatient Sample (NIS) for the calendar year 2005, the most recent available data. The NIS is the largest database of its kind and includes all-payer discharge information from a national survey of 20% of all non-federal hospitals in the United States. A complete overview and description of the NIS is available on their website.13 The studied cohort was comprised by linking the ICD9-CM procedure codes for all patient-discharges that occurred for either CAS (00.63)14 or CEA (38.12) with the corresponding diagnostic code for carotid artery stenosis with (433.11) or without (433.10) stroke.15 If a patient's discharge diagnosis (diagnostic fields 1-15) was “carotid artery stenosis without mention of stroke” with no accompanying secondary diagnoses for transient ischemic attack (TIA), they were classified as “asymptomatic”. If a patient's discharge diagnosis was either “carotid artery stenosis with stroke” or, if there was no mention of stroke, but a secondary diagnosis code included that for TIA, patients were classified as “symptomatic” (Appendix, online only). A patient was excluded from the final dataset if they had procedural codes listed for both CAS and CEA (<1% of total) during the index admission in the interest of keeping the cohorts as homogenous as possible to facilitate comparison between the two procedure types. Hospital level data were included in this analysis based on available information provided by the American Hospital Association's annual survey of hospitals, a direct link to which is available through the NIS dataset.13

The primary outcome measures for this retrospective study were postoperative in-hospital stroke and death. “Postoperative stroke” has a specific ICD9-CM code (‘997.02')15 therefore any patient undergoing CAS or CEA that had this code under one of their secondary ICD9-CM diagnostic codes (up to 15) was classified as having had an iatrogenic stroke. Secondary outcome measures included resource utilization, namely hospital length of stay, total hospital charges, and discharge disposition. Detailed specific procedure cost information is unavailable in the NIS therefore total hospital charges are used as a surrogate for cost information. For those patients surviving to hospital discharge, their disposition is available. For the purposes of analysis, this has been condensed to a dichotomous variable, home vs other (rehabilitation facility, skilled nursing facility, nursing home, etc.).

Statistical analyses were performed using SAS version 9.1 (Cary, NC). A univariate analysis of categorical variables was performed using Rao-Scott χ2 and continuous variables were analyzed by survey-weighted analysis of variance (ANOVA), with a P < .05 considered statistically significant. The Elixhauser comorbidity software designed for use with administrative datasets was utilized to identify patient comorbidities for the purposes of univariate and multivariate analyses.16 This software allows for accurate identification of pre-existing medical comorbidities, however, does not allow for comprehensive risk stratification as the data of greatest clinical interest in terms of risk stratification such as the degree of carotid stenosis, previous neck surgery, previous neck radiation, high carotid bifurcation, routine shunting/patch usage for CEA, embolic protection device usage for CAS as well as information on the degree of medical comorbidity such as echocardiogram and persantine thallium results as well as the degree of neurologic disability from previous stroke, are not available in this database.

Multivariate logistic regression was performed to determine which factors were independently predictive of postoperative stroke or death. Covariates included in the logistic regression were patient age, gender, insurance type, comorbid conditions (hypertension, diabetes mellitus [DM], chronic lung disease, coronary artery disease [CAD], congestive heart failure [CHF], valvular heart disease, obesity, renal failure), symptom status (symptomatic vs asymptomatic), procedure type performed (CAS vs CEA) and hospital teaching status. Patient age was evaluated as a categorical variable in the logistic regression.

The NIS provides a weighting strategy in order to draw estimates at the national level based on a 20% annual survey of hospitals. The statistical analyses are performed based on these weighted numbers and therefore the data provided in the results section are in the weighted format. This utilization of survey weights to make observations regarding surgical procedures at the national level is a method that has been previously described.17, 18

Back to Article Outline

Results 

During the calendar year 2005, an estimated 135,701 patient discharges occurred for either CEA or CAS in the United States. The majority of patients were treated for asymptomatic disease (92.1%). Overall, a greater percentage of patients underwent CEA (90.5%) than CAS (9.5%). Table I contains overall demographic data for those undergoing CEA or CAS (Table I).

Table I. Carotid revascularization patient characteristics, 2005
FactorOverall no. (%)
No. patients135,701(100)
Age (y)
Mean [SEM]71[0.11]
Median [range]72[23-98]
Age groups
<6017,450(12.9)
60-6938,558(28.4)
≥7079,693(58.7)
Sex
Men78,464(57.9)
Women57,147(42.1)
Race
Non-white9677(9.7)
White89,633(90.3)
Presentation type
Asymptomatic122,986(92.1)
Symptomatic10,495(7.9)
Transient ischemic attack5656(4.2)
Stroke4975(3.7)
Procedure type
CEA122,786(90.5)
CAS12,914(9.5)
Hospital type
Non-teaching77,625(57.2)
Teaching58,076(42.8)
Hospital bed size
Small8126(6.0)
Medium29,686(21.9)
Large97,888(72.1)
Payer
Private/Medicare128,596(94.9)
Medicaid/self-pay6947(5.1)

CEA, Carotid endarterectomy; CAS, carotid artery stenting.

Overall patient characteristics by procedure 

The CEA patients had increased average age compared with the CAS patients (71 vs 70.9, P < .0001), although the median age was 72 for both groups. The distribution of age groups was similar for both CEA and CAS patients (P =.41). Likewise the two groups were similar in terms of certain comorbid medical conditions such as diabetes mellitus (P = .34), coronary artery disease (P = .88), and valvular heart disease (P = .05). The two groups also had similar percentages of patients with private insurance (P = .27) and were treated at similarly sized hospitals (P = .1). The patients undergoing CEA had a higher percentage of patients with hypertension, chronic lung disease, and obesity (P < .05 for all), and were more likely to be treated at a non-teaching hospital (P <.0001). In contrast, the CAS patients were more likely to be male (P = .0002), and have a history of CHF (P <.0001), or renal failure (P = .02).

Univariate analysis of overall postoperative morbidity and mortality 

Patients that underwent CEA in 2005 had a significantly lower iatrogenic stroke rate (1.1%) than did those that underwent CAS (1.8%), P = .0004. Likewise patients that underwent CEA had a significantly lower in-hospital mortality rate (.57%) than did the CAS patients (1.1%), P = .004.

Univariate analysis of overall patient resource utilization 

The patients that underwent CEA had a longer median hospital length of stay (2 days) than did those that underwent CAS (1 day), P < .0001. Overall, the patients that underwent CEA were more likely to be discharged to home after their procedure than the CAS patients (94.1% vs 91.9%, P = .008). The total median hospital charges were greater for patients that were included in the CAS group ($30,396) than those included in the CEA group ($17,658), P <.0001.

Univariate analysis of asymptomatic patient characteristics 

The CEA and CAS patient groups for asymptomatic disease were similar in terms of the percentage of patients greater than 70 years of age (P = .19), hospital bed size where they were treated (P = .15), as well as some comorbid medical conditions, specifically diabetes mellitus, coronary artery disease and obesity (all P values >.05). Compared with the CEA patients, the CAS patients had a higher mean age (P < .0001), a higher percentage of male patients (P < .0001), as well as a higher percentage of patients with congestive heart failure (P < .0001), valvular heart disease (P = .01), and renal failure (P = .002). In contrast, the CEA patients had a higher prevalence of hypertension (P = .0008) and were much more likely to be treated at a non-teaching hospital (59.7% vs 35.1%, P < .0001) than the CAS patients (Table II).

Table II. Univariate analysis of patient characteristics based on procedure type for asymptomatic and symptomatic disease, 2005
FactorOverallAsymptomaticSymptomatic
CEACASPCEACASPCEACASP
Overall number (%)122,78612,914n/a111,68411,302n/a93801116n/a
Mean age [SEM]71[.11]70.9[.28]<.000171.1[.11]71.6[.23]<.000169.5[.28]68.9[.81]<.0001
Age groups .41 .19 .81
% <6012.813.5 12.311.5 18.119.5
% 60-6928.527.2 28.727.3 28.426.7
% ≥ 7058.759.3 59.061.2 53.553.8
Gender
% Men57.462.3.000257.262.9<.000158.160.9.42
Comorbid conditions
% Hypertension71.164.2<.000171.965.7.000867.562.9.11
% Diabetes mellitus26.325.3.3426.626.5.9424.121.3.37
% Chronic lung disease21.217.6.00420.817.9.0322.718.1.13
% CAD/MI10.811.0.881111.8.389.56.6.17
% CHF6.811.4<.00016.611.1<.00017.717.0<.0001
% Valvular disease6.57.5.056.47.7.017.56.8.69
% Renal failure3.03.9.022.83.9.0023.32.6.53
% Obesity4.83.6.024.83.6.025.24.1.49
Hospital type <.0001 <.0001 <.0001
% Teaching40.465.8 40.364.5 39.571.5
% Non-teaching59.634.2 59.735.5 60.528.5
Hospital bed size .1 .15 .002
% Small6.33.2 6.33.4 5.71.4
% Medium22.416.8 22.317.1 25.112.3
% Large71.380 71.479.5 69.286.3
Insurance type .27 .66 .44
% Private/Medicare94.994.2 95.295.0 92.190.1
% Medicaid/self-pay5.15.8 4.85.0 7.99.9

CAS, Carotid artery stenting; CEA, carotid endarterectomy.

Univariate analysis of asymptomatic patient postoperative morbidity and mortality 

Asymptomatic patients that underwent CEA for carotid artery stenosis had a significantly lower postoperative stroke rate (.88%) than did those that underwent CAS (1.6%), P = .001. The unadjusted in-hospital mortality rate for CEA (.38%) was similar to that of CAS (.57%), P = .18 (Fig 1, A and B).

  • View full-size image.
  • Fig 1. 

    This series of bar graphs demonstrates the outcomes for asymptomatic patients undergoing carotid revascularization, 2005. A, In-hospital mortality. B, Post-operative stroke rate.

Univariate analysis of asymptomatic patient resource utilization 

Asymptomatic patients that underwent CEA and CAS both had median hospital lengths of stay of 1 day. The percentage of patients that were discharged to a facility other than home was similar for CEA (4.8%) and CAS (6.0%), P = .1. Patients that underwent CEA had significantly lower median ($16,956) total hospital charges compared with those patients that underwent CAS ($28,853), P < .0001 (Table III).

Table III. Postoperative outcomes and resource utilization by presentation type CEA and CAS, 2005
FactorOverallAsymptomaticSymptomatic
CEACASPCEACASPCEACASP
In-hospital mortality (%).571.1.004.38.57.181.44.6.0002
Postoperative stroke (%)1.11.8.0004.881.6.0012.54.1.15
Length of stay (d) <.0001 <.0001 <.0001
Median [range]2 [0-117]1 [0-66] 1 [0-110]1 [0-53] 4 [1-67]5 [0-66]
Total hospital charges ($) <.0001 <.0001 <.0001
Median [range]17,658 [29-772,000]30,396 [649-508,734] 16,956 [29-632,341]28,853 [649-508,734] 29,894 [482-568,181]49,535 [4176-473,639]
Discharge disposition .008 .1 <.0001
% Home94.191.9 95.294.0 84.973.9
% Rehab, SNF, other5.98.1 4.86.0 15.126.1

CAS, Carotid artery stenting; CEA, carotid endarterectomy.

Univariate analysis of symptomatic patient characteristics 

Symptomatic patients that underwent either CEA or CAS for carotid artery stenosis were similar in terms of patients that were older than 70 years (P = .81). Likewise, the two groups were similar in the percentage of male patients (P = .42), as well as the prevalence of certain comorbid medical conditions, specifically hypertension, diabetes mellitus, chronic lung disease, coronary artery disease, renal failure, and obesity (all P values >.05). Symptomatic patients that underwent CEA had a higher mean age (P < .0001) and were more likely to have their procedure performed at a non-teaching hospital than the CAS patients (P < .0001). In contrast, a higher percentage of patients that underwent CAS had congestive heart failure than did the CEA patients (P < .0001) and were more likely to be treated at a large hospital (P = .002) (Table II).

Univariate analysis of symptomatic patient postoperative morbidity and mortality 

The symptomatic patients with carotid artery stenosis that underwent CEA had a similar postoperative stroke rate (2.5%) to those that underwent CAS (4.1%), P = .15. The unadjusted in-hospital mortality rate was significantly lower for symptomatic patients that underwent CEA (1.4%) than those that underwent CAS (4.6%), P = .0002. (Fig 2, A and B).

  • View full-size image.
  • Fig 2. 

    This series of bar graphs demonstrates the outcomes for symptomatic patients undergoing carotid revascularization, 2005. A, In-hospital mortality. B, Post-operative stroke rate.

Univariate analysis of symptomatic patient resource utilization 

Symptomatic patients that underwent CEA had a significantly shorter median (4 days) length of hospital stay compared with those that underwent CAS (5 days), P < .0001. For CEA, a significantly greater percentage of patients were discharged to home (84.9%) compared with the CAS patients (73.9%), P <.0001. In terms of total hospital charges, those patients included in the CEA group had significantly lower median ($29,894) total charges compared with the CAS group ($49,535), P < .0001 (Table III).

Multivariate analysis of postoperative stroke 

By multivariate logistic regression, CAS was independently predictive of postoperative stroke compared with CEA (odds ratio [OR] 1.7; 95% confidence interval [CI] 1.2-2.3, P = .003). Another factor that independently predicted postoperative stroke was presentation with symptomatic disease (OR 2.8; 95% CI 2.1-3.8) (Table IV).

Table IV. Multivariate analyses of in-hospital mortality and post-procedure stroke for carotid revascularization, 2005
FactorIn-hospital mortalityPost-procedure stroke
OR95% CIPOR95% CIP
Sex .22 .24
Men (vs women)1.3.87-1.9 .86.68-1.1
Age group .36 .94
≥70 (vs <60)1.5.74-2.9 1.1.74-1.6
≥70 (vs 60-69)1.3.84-2.1 1.0.76-1.4
Insurance type .8 .31
Private (vs Medicaid/self-pay)1.1.44-2.9 .78.48-1.3
Comorbid conditions
Hypertension (vs none).53.37-.75.0003.82.64-1.0.11
Diabetes (vs none)1.2.81-1.7.38.92.7-1.2.57
Chronic lung disease (vs none)2.01.3-3.0.0011.1.78-1.4.75
Coronary artery disease/MI (vs none).28.11-.71.0071.1.74-1.6.7
Congestive heart failure (vs none)2.81.7-4.5<.00011.4.90-2.0.14
Valvular heart disease (vs none).88.44-1.7.70.98.64-1.5.94
Obesity (vs none).51.19-1.3.17.93.54-1.6.78
Renal failure (vs none)2.71.6-4.6.0004.93.44-1.9.84
Presentation type <.0001 <.0001
Symptomatic (vs asymptomatic)3.52.4-5.3 2.82.1-3.8
Procedure type .08 .003
CAS (vs CEA)1.5.96-2.5 1.71.2-2.3
Hospital teaching status .79 .35
Non-teaching (vs teaching)1.1.72-1.6 .88.66-1.2
Hospital bed size .45 .06
Large (vs small)1.8.70-4.8 .74.47-1.2
Medium (vs small)1.6.60-4.5 1.1.63-1.8

OR, Odds ratio; CI, confidence intervals; MI, myocardial infarction; CAS, carotid artery stenting; CEA, carotid endarterectomy.

Multivariate analysis of in-hospital mortality 

CAS trended toward increased in-hospital mortality compared with CEA but was not statistically significant (OR 1.5; 95% CI .96-2.5, P = .08) after adjustment for multiple covariates including symptom type as well as comorbid medical conditions. Factors that were independently predictive of increased in-hospital mortality included presentation with symptomatic disease (OR 3.5; 95% CI 2.4-5.3, P <.0001), the presence of chronic lung disease (OR 2.0; 95% CI 1.3-3.0), congestive heart failure (OR 2.8; 95% CI 1.7-4.5, P < .0001), and renal failure (OR 2.7; 95% CI 1.6-4.6) (Table IV).

Back to Article Outline

Discussion 

In this cross-sectional observational study of the most recently available national data (2005), we have found that overall, patients undergoing CAS had significantly higher overall rates of postoperative stroke (1.8% vs 1.1%, P = .0004), and in-hospital mortality (1.1% vs .57%, P = .004) than those undergoing CEA. By separate analysis, for patients with asymptomatic carotid artery stenosis, CAS was associated with a nearly twofold higher postoperative stroke rate than CEA (P = .001). For symptomatic patients with carotid artery stenosis CAS was associated with a nearly fourfold higher mortality rate than CEA (P = .0002). By multivariate analysis after adjusting for presentation type, age, gender, and multiple comorbid medical conditions, patients that underwent CAS had higher odds of postoperative stroke (1.7) than those that underwent CEA. Those patients that underwent CAS trended toward but did not achieve statistical significance in regard to multivariate analysis of in-hospital mortality compared with those that underwent CEA. In terms of overall resource utilization, the median total hospital charges were significantly greater for those patients included in the CAS group (P < .0001) and a greater percentage of CAS patients treated for symptomatic disease were discharged someplace other than home (P = .0006). For the symptomatic patient group, CAS patients were more likely to be discharged to a facility other than home (26.1% vs 15.1%, P < .0001).

As many providers anticipate the definitive results of randomized controlled trials such as The Carotid Revascularization Endarterectomy vs Stenting Trial (CREST),19 conflicting results have been published regarding the outcomes of CEA compared with CAS. The clinical observations in the current work, that CAS is associated with higher unadjusted in-hospital stroke and mortality rates, are similar to those of previous population-based studies2, 8 as well as large randomized multi-institutional trials.3, 4 Nowygrod et al in their population-based work incorporating the years 1998 through 2003 found that for CAS, procedural stroke (2.13%) and mortality (2.3%) rates were greater than that for CEA with respective stroke and mortality rates of 1.28% and .5% nationally.2 The current work with similar respective overall stroke and mortality rates nearly twofold higher for CAS compared with CEA focused specifically on 2005 in which a dedicated ICD9 CM code for CAS existed. Previous similar works, including our own, relied on a combination of specific diagnostic codes for carotid artery stenosis and less specific peripheral stenting procedural codes.2, 8

Similar to our observations, collaborators for the Endarterectomy versus Stenting in Patients with Symptomatic Severe Carotid Stenosis3 in their large (N = 527) randomized multicenter trial determined that the 30-day risk of any stroke or death of CAS was more than twofold greater than that for CEA (relative risk 2.5) leading to early termination of the trial. Likewise, a recent meta-analysis found CAS to be associated with increased 30-day rates of stroke and mortality compared with CEA.20 These results are in contrast to the immediate and long-term follow-up reports of several industry-sponsored reports6, 21, 22, 23 that have determined no significant difference is found in terms of stroke and mortality rates between CAS and CEA, concluding that CAS is a noninferior treatment modality to CEA.6, 21, 22, 23 Because the definition of what constitutes a “high-risk” surgical candidate as well as the proportion of symptomatic to asymptomatic patients included in these trials is variable, it is difficult to directly compare clinical outcomes. Of note, the methodology of some of these industry sponsored registries, namely the heterogeneous case-mix, as well as the use of a historical control as the surgical arm in a noninferiority trial, has been previously critiqued.24, 25

It is important to note that our group previously performed a population-based clinical outcomes analysis of CAS and CEA at the national level for the years 2003 and 2004, prior to the release of the dedicated ICD9-CM code for CAS.8 The findings in our previous work demonstrated significantly greater stroke and mortality rates overall, as well as for asymptomatic and symptomatic groups separately than reported here. The current work demonstrates that in comparison, the gap between the outcomes for CEA and CAS is possibly narrowing over time. The difference between CEA and CAS for overall stroke (1.2%, P < .0001) and mortality (.91%, P < .0001) observed in the previous study has narrowed such that in the current work the difference in observed overall stroke (.7%, P = .001) and mortality (.53%, P = .004) rates between CEA and CAS was substantially less. More prominently, the stroke rate gap between symptomatic patients undergoing CAS or CEA substantially narrowed. In the 2003 and 2004 study, the difference in overall stroke rates was 3.1% in favor of CEA, P < .0001. In the current work, the difference in observed stroke rates between the two groups, although still higher for CAS, was not statistically significant (1.6%, P = .15). Likewise for mortality rates, the previous study found an in-hospital survival advantage for symptomatic patients of 6.5% (P = .0001) in favor of CEA. In the current work, the observed difference was decreased to 3.2%. While the methodology of the two studies is somewhat different due to the change in ICD9-CM coding that occurred in 2005, the design and endpoints of the two studies are the same. Therefore, the finding that by 2005 the substantial observed differences between CEA and CAS in 2003 and 2004 may be decreasing, is a noteworthy finding. Most notably in the current work by multivariate logistic regression, CAS was not an independent risk factor for in-hospital mortality (P =.08). This change over time, to some degree, may reflect an improvement in patient selection as more knowledge is gained regarding which lesions are less likely to be amenable to CAS.26 Likewise, over time there have likely been technical improvements in overcoming the “learning curve” for CAS as has been demonstrated in other complex surgical procedures.27 While not measurable in this type of study, the decreasing stroke rates for CAS may also in part be attributable to the universal application of embolic protection devices, which are currently mandated by the Centers for Medicare and Medicaid Services for reimbursement.

Due to the disparity in reported clinical results of CAS compared with CEA, the forthcoming definitive results of randomized controlled trials are eagerly anticipated. In the interim, other measures of adequacy of performance between the two procedures have been evaluated such as economic outcomes including length of stay data and cost information. The current work found that for patients undergoing either CEA or CAS, the difference in hospital length of stay overall was largely negligible. The exception to this observation was that for patients with symptomatic disease, which showed that CAS patients had a lengthier median (5 vs 4 days) hospital stay than the CEA patients. It is not possible to determine if this increased length of stay was related to the procedure itself or rather the other factors related to evaluation and disposition of a stroke patient (ie, speech and swallow evaluations, physical therapy clearance etc.). While no detailed specific procedure-related cost information is available in this dataset, total hospital charges, which have been previously used to compare outcomes following complex surgical procedures,10, 28 is available, and did demonstrate a marked difference between the two procedures. Overall, the patients that were included in the CAS group had significantly greater median total hospital charges than those included in the CEA group. This difference increased dramatically for the symptomatic patient cohort where CAS was associated with greatly increased median total hospital charges compared with CEA. It is important to recognize that this increase may likely be due to factors unrelated to the procedure itself. To our knowledge, this is the first study to evaluate total charges for CEA and CAS at the national level. Previous economic comparisons have been made between CEA and CAS, the majority of which are confirmatory, demonstrating significantly increased costs associated with CAS compared with CEA.9, 10, 11, 12 Park et al, in a single institution cost comparison of CAS vs CEA, found that in a cohort of 94 patients with equivalent clinical outcomes, CAS was associated with higher total costs ($17,402 vs $12,112, P = .029) than CEA.10 They attributed the majority of the cost difference to direct procedure-related materials as noted by the difference between surgical vs angiography suite supplies for CAS ($15,407) compared with CEA ($1953), P = .001.10 Similarly Kilaru et al, in a comprehensive Markov decision analysis model, concluded that CAS costs more and is less effective than CEA.9 While it is not possible to directly compare cost results from single institutional and decision analysis models with the total hospital charges reported in a large national database, the trends observed in three studies with three distinct methodologies are very similar. It is important to note that the total charges in this study are not direct procedure-related costs but rather the total charges associated with the index hospital stay. In contrast to the above studies as well as our findings, one study was encountered that found CEA to be a costlier procedure than CAS. Gray et al, in their single institutional review of 272 carotid artery procedures, found that CEA was associated with higher cost than CAS. Of note, their group found no difference in direct procedure costs for CAS and CEA (P = .1), and report 0% stroke and mortality rates in the CAS arm, findings that make their results difficult to generalize to other populations.29

While causality cannot be proven in administrative datasets, we offer that one possible explanation for the large discrepancy in hospital charges between CAS and CEA beyond the cost differential associated with the procedure itself, may be the increased postoperative mortality and stroke rate overall as well as the increased length of stay required by the symptomatic patients that underwent CAS. This is intuitive as those patients that suffered a postoperative stroke or ultimately died or were unable to be discharged to home likely had a more complicated hospital course with more intensive multidisciplinary care, diagnostic work-ups and secondary interventions all of which would be included in the nonspecific total hospital charge category.

The limitations of studies based on administrative datasets such as the NIS have been described previously.30 The main limitations specific to this study pertain to unknown clinical factors that may represent an inherent selection bias in favor of CEA. We have no knowledge of anatomical factors namely degree of carotid artery stenosis, previous neck surgery or radiation or whether a patient was deemed medically too high risk for CEA and therefore underwent CAS. We have attempted to account for certain patient comorbid medical conditions using validated software, however, this is limited information and does not speak to the severity of a patients' disease state, which would represent a selection bias whereby the sickest and highest risk patients likely underwent CAS. Similarly, total hospital charges as a marker of economic impact of a surgical procedure provides only crude information and, therefore, in this study is used as a surrogate for cost information and conclusions should be drawn with this fact in mind. It is unable to be shown which portion of the charges is attributable to the procedure itself. Patient factors unrelated to the procedure including diagnostic evaluations for comorbid conditions, consultations, and imaging costs are all included in this charge information and these unmeasured factors may account for the majority of the hospital charge discrepancy. This may again represent a selection bias in that the CAS patients were likely high risk for open surgery and therefore would likely incur greater hospital cost related to their comorbidity which would contribute to the discrepancy observed in terms of total hospital charges. Previous works have utilized this total hospital charge information,11, 28 and the fact that the observations in the current study are congruent with the findings of other published data on the subject tempers this limitation to some degree. In this study, the percentage of asymptomatic patients is higher than expected (92%). Patients that were miscoded under diagnostic codes other than carotid stenosis may have been inappropriately classified as asymptomatic. To minimize this effect all secondary diagnoses (up to 15) of TIA or stroke were included as a means of further identifying the symptomatic group. However, because these data are collected at the time of patient discharge the code for carotid stenosis with or without the presence of symptoms was used to initially stratify the patient groups by symptoms. Coding inaccuracies in terms of patient case-mix, such as prevalence of comorbidities, or complications such as postoperative stroke are possible. To minimize these inaccuracies, we used established comorbidity software to attempt to appropriately characterize patients with pre-existing comorbid conditions.16 Likewise, to identify iatrogenic stroke, we used an established ICD9-CM code. If patients were miscoded under other less-specific stroke codes, they may have been missed in this postoperative stroke analysis, however, there is no reason to suspect that this miscoding would bias one procedure type over another. Additionally, in-hospital outcomes are a suboptimal measure of overall success when evaluating surgical interventions. Ideally long-term information such as 30-day and 1-year stroke and mortality rates would be used because we are evaluating procedures whose goal is to limit long-term stroke risk. Likewise, we may be underestimating postoperative stroke and mortality rates due to the lack of 30-day data. Unfortunately, due to the arduous patient de-identification process employed by the NIS, follow-up information is unavailable. It is important to recognize that the observations made in this work regarding mortality, postoperative stroke and resource utilization related to CEA and CAS are likely multifactorial. We are unable to determine to what extent a patient's mortality, morbidity, LOS or hospital charge is directly attributable to the procedure itself or whether it was related to a pre-existing comorbidity or other mitigating factor. To attempt to temper this limitation we included as many cofactors as possible in the multivariate analysis.

CAS has gained wide popularity, and annual volume seems to be increasing nationwide.31 Despite this technical feasibility and wide procedural application, controversy continues to surround patient selection for this procedure, specifically in terms of application in the asymptomatic population. This was denoted by the Center for Medicare and Medicaid services when they reversed their initial proposal to widely expand coverage for carotid artery stenting in May, 2007. Of interest, it seems that over time, the outcomes for CAS may be improving, and the current work shows that by 2005 the previously large gap between CEA and CAS may be narrowing. However, if stenting is to gain universal application, its clinical outcomes in terms of stroke and mortality rates should be equivalent to CEA and the associated costs must not be prohibitive to allow application to all patient groups, which the current and previous data do not yet support.

Back to Article Outline

Conclusion 

During the year 2005, the most recently available national data, within the limitations of a large administrative dataset, which include a lack of comprehensive risk stratification by procedure type, we have found that patients that underwent CAS had an overall increased postoperative stroke and mortality rate compared with those that underwent CEA. Further, those patients that underwent CAS had an increase in crude total hospital charges associated with their hospitalization compared with those that underwent CEA. These observations indicate that the anticipated forthcoming results of randomized controlled trials will be of critical importance in determining the future applicability of CAS in patients with carotid artery stenosis.

Back to Article Outline

Author contributions 


Conception and design: JM, ME

Analysis and interpretation: JM, ME

Data collection: JM

Writing the article: JM

Critical revision of the article: LM, AS

Final approval of the article: ME

Statistical analysis: JM, ME

Obtained funding: Not applicable

Overall responsibility: ME

Back to Article Outline

Appendix, online only 

Appendix (online only). ICD-9CM diagnostic and procedural codes
DiagnosisDiagnostic codeProcedureProcedural code
Carotid artery occlusion and stenosis without mention of cerebral infarction433.10Carotid endarterectomy38.12
Carotid artery occlusion and stenosis with cerebral infarction433.11Percutaneous insertion of carotid artery stent00.63
Multiple and bilateral carotid artery occlusion and stenosis without mention of cerebral infarction433.30
Multiple and bilateral carotid artery occlusion and stenosis with cerebral infarction433.31
Transient cerebral ischemia435.9
Amaurosis fugax362.34
Iatrogenic cerebrovascular infarction or hemorrhage; Postoperative stroke997.02

Back to Article Outline

References 

  1. Chaer RA, Barbato JE, Lin SC, Zenati M, Kent KC, McKinsey JF. Isolated iliac artery aneurysms: a contemporary comparison of endovascular and open repair. J Vasc Surg. 2008;47:708–713
  2. Nowygrod R, Egorova N, Giampaolo G, Anderson P, Gelijns A, Moskowitz A, et al. Trends, complications, and mortality in peripheral vascular surgery. J Vasc Surg. 2006;43:205–216
  3. Mas JL, Chatellier G, Beyssen B, Branchereau A, Moulin T, Becquemin JP. Endarterectomy versus stenting in patients with symptomatic severe carotid stenosis. N Engl J Med. 2006;355:1660–1671
  4. The SPACE Collaborative Group. Thirty-day results from the SPACE trial of stent-protected angioplasty versus carotid endarterectomy in symptomatic patients: a randomized noninferiority trial. Lancet. 2006;368:1239–1247
  5. Yadav JS, Wholey MH, Kuntz RE, Fayad R, Katzen BT, Mishkel GJ, et al. Protected carotid-artery stenting versus endarterectomy in high-risk patients. N Engl J Med. 2004;351:1493–1501
  6. Gray WA, Hopkins LN, Yadav S, Davis T, Wholey M, Atkinson R, et al. Protected carotid stenting in high-surgical-risk patients: The ARCHeR results. J Vasc Surg. 2006;44:258–269
  7. Cao P, De Rango P, Verzini F, Maselli A, Norgiolini L, Giordano G. Outcome of carotid stenting versus endarterectomy (A case-control study). Stroke. 2006;37:1221–1226
  8. McPhee JT, Hill JS, Ciocca RG, Messina LM, Eslami MH. Carotid endarterectomy was performed with lower stroke and death rates than carotid artery stenting in the United States in 2003 and 2004. J Vasc Surg. 2007;46:1112–1118
  9. Kilaru S, Korn P, Kasirajan K, Lee TY, Beavers FP, Lyon RT. Is carotid angioplasty and stenting more cost effective than carotid endarterectomy?. J Vasc Surg. 2003;37:331–339
  10. Park B, Mavanur A, Dahn M, Menzoian J. Clinical outcomes and cost comparison of carotid artery angioplasty with stenting versus carotid endarterectomy. J Vasc Surg. 2006;44:270–276
  11. Jordan WD, Roye GD, Fisher WS, Redden D, McDowell HA. A cost comparison of balloon angioplasty and stenting versus endarterectomy for the treatment of carotid artery stenosis. J Vasc Surg. 1998;27:16–22
  12. Pawaskar M, Satiani B, Balkrishnan R, Starr JE. Economic evaluation of carotid artery stenting versus carotid endarterectomy for the treatment of carotid artery stenosis. J Am Coll Surg. 2007;205:413–419
  13. Nationwide Inpatient Sample (NIS). http://www.hcup-us.ahrq.gov/nisoverview.jspAccessed April 27, 2008
  14. Centers for Medicare and Medicaid Services. Updates and Revisions to ICD9-CM Procedure Codes (Addendum). Accessed April 27, 2008 http://www.cms.hhs.gov/ICD9ProviderDiagnosticCodes/Downloads/icd9addendafy05.pdf
  15. US Department of Health and Human Services, International classification of diseases, ninth revision, clinical modification, US Government Printing Office, Washington (DC) (1998).
  16. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36:3–5
  17. Kotwall CA, Maxwell GJ, Brinker CC, Koch GG, Covington DL. National estimates of mortality rates for radical pancreaticoduodenectomy in 25,000 patients. Ann Surg Oncol. 2002;9:847–854
  18. Timaran CH, Veith FJ, Rosero EB, Modrall JG, Arko GR, Clagett GP, et al. Endovascular aortic aneurysm repair in patients with the highest risk and in-hospital mortality in the United States. Arch Surg. 2007;142:52–54
  19. Hobson RW, Howard VJ, Roubin GS, Brott TG, Ferguson RD, Popma JJ. Carotid artery stenting is associated with increased complications in octogenarians: 30-day stroke and death rates in the CREST lead-in phase. J Vasc Surg. 2004;40
  20. Brahmanandam S, Ding EL, Conte MS, Belkin M, Nguyen LL. Clinical results of carotid artery stenting compared with carotid endarterectomy. J Vasc Surg. 2008;47:343–349
  21. Iyer SS, White CJ, Hopkins LN, Katzen BT, Safian R, Wholey MH, et al. Carotid artery revascularization in high-surgical-risk patients using the carotid WALLSTENT and FilterWire EX/EZ 1-year outcomes in the BEACH pivotal group. J Am Coll Card. 2008;51:427–434
  22. Gurm HS, Yadav JS, Fayad P, Katzen BT, Mishkel GJ, Bajwa TK, et al. Long-term results of carotid stenting versus endarterectomy in high-risk patients. N Eng J Med. 2008;358:1572–1579
  23. Hopkins LN, Myla S, Grube E, Wehman JC, Levy EI, Bersin RM, et al. Carotid artery revascularization in high surgical risk patients with the NexStent and the Filterwire EX/EZ 1-year results in the CABERNET trial. DOI 10.1002/ccd.21564.
  24. Naylor RA. Regarding “Protected carotid stenting in high-surgical-risk patients: the ARCHeR results”. J Vasc Surg. 2007;45:222–223
  25. Mackey WC. Regarding protected carotid stenting. J Vasc Surg. 2006;44:258–268
  26. Sayeed S, Stanziale SF, Wholey MH, Makaroun MS. Angiographic lesion characteristics can predict adverse outcomes after carotid artery stenting. J Vasc Surg. 2008;47:81–87
  27. Tseng JF, Pisters PW, Lee JE, Wang H, Gomez HF, Sun CS, et al. The learning curve in pancreatic surgery. Surgery. 2007;141:456–463
  28. Dimick JB, Cattaneo SM, Lipsett PA, Pronovost PJ, Heitmiller RF. Hospital volume is related to clinical and economic outcomes of esophageal resection in Maryland. Ann Thorac Surg. 2001;72:334–341
  29. Gray W, White H, Barrett D, Chandran G, Turner R, Reisman M. Carotid stenting and endarterectomy: a clinical and cost comparison of revascularization strategies. Stroke. 2002;33:1063–1070
  30. Iezzoni LI, Foley SM, Daley J, Hughes J, Fisher ES, Heeren T. Comorbidities, complications, and coding bias (Does the number of diagnosis codes matter in predicting in-hospital mortality?). JAMA. 1992;267:2197–2203
  31. Goodney PP, Powell RJ. Carotid artery stenting: what have we learned from the clinical trials and registries and where do we go from here?. Ann Vasc Surg. 2008;22:148–158

 Competition of interest: none.

 Additional material for this article may be found online at www.jvascsurg.org.

PII: S0741-5214(08)01174-9

doi:10.1016/j.jvs.2008.07.017

Journal of Vascular Surgery
Volume 48, Issue 6 , Pages 1442-1450.e1, December 2008

Article Tools

* Image Usage & Resolution