Reasons Why Data from the Nationwide Inpatient Sample Can Be Misleading for Carotid Endarterectomy and Carotid Stenting

doi:10.1053/j.semvascsurg.2012.02.003

Seminars in Vascular Surgery

Volume 25, Issue 1, March 2012, Pages 13-17

https://doi.org/10.1053/j.semvascsurg.2012.02.003 Get rights and content

The Nationwide Inpatient Sample (NIS) is often used for population-based research comparing the safety of carotid artery stenting (CAS) to that of carotid endarterectomy (CEA) in the United States. At least two findings from the NIS dataset seem questionable, however. First, several NIS studies indicate that >90% of CEAs and CAS procedures are currently being performed for asymptomatic carotid stenosis, which considerably exceeds the prevalence of asymptomatic patients reported elsewhere. Second, these studies also suggest that periprocedural stroke rates for CEA and CAS are collectively lower at hundreds of community hospitals contributing data to the NIS than they were in the Carotid Revascularization Endarterectomy Versus Stenting Trial (CREST), even though the participating surgeons and interventionalists in CREST were vetted on the basis of their previous experience and results. In addition, some unexpectedly low stroke to death ratios are present in NIS studies, implying that not all iatrogenic strokes have been entered into the NIS dataset. These issues might be related to inadequate documentation of preprocedural symptoms and periprocedural strokes in the medical records, leading to subsequent coding errors in the hospital discharge abstracts from which NIS data are extracted. The clinical limitations of the NIS and other administrative datasets have been pointed out in the past, but they appear to be particularly relevant to carotid interventions and must be recognized.

Section snippets

Neurologic Indications for CEA and CAS

Table 1 shows the prevalence of asymptomatic stenosis as an indication for carotid intervention in the six NIS studies selected for review, for three statewide surveys that also have been based on claims data within the past decade,12, 13, 14 and for two prospective registries maintained by the Society for Vascular Surgery (SVS)¹⁵ and the Vascular Study Group of Northern New England¹⁶ during the same period of time. The prevalence of asymptomatic patients in the NIS studies ranges from 92% to

Stroke Rates for CEA and CAS

The NIS dataset does not include strokes that occur after hospital discharge but within the 30-day periprocedural period used in other studies.4, 6, 8 Stroke rates during this missing interval are rarely reported elsewhere, but data from the SVS registry suggest that 29% of the 30-day strokes after CEA and 43% of those after CAS happen after patients have left the hospital.¹⁵ In contrast, however, the Vascular Study Group of Northern New England registry indicates that just 10% of the 30-day

Clinical Pitfalls of Administrative Datasets

According to McPhee et al,⁸ the limitations of miscoded or absent data in administrative datasets like the NIS are well known. This may be correct for investigators who use the NIS, but it might not be so obvious to clinicians who have no personal experience with administrative data. Although the authors of the six NIS studies on CEA and CAS in the present review do encourage caution in the interpretation of their findings,3, 4, 5, 6, 7, 8 these disclaimers are not conventionally disclosed

Conclusions

The mortality rates of CEA and CAS are undoubtedly accurate in the NIS dataset because deaths are invariably noted in medical records by clinicians and are unmistakable to coding specialists who prepare discharge abstracts. Information about preprocedural symptoms and periprocedural strokes is a different matter, however. Its accuracy depends entirely on adequate documentation by physicians (or their assistants) and uniform ICD-9-CM coding throughout the United States. It might be unrealistic

References (20)

T.F. Kresowik et al.
Multistate utilization, processes, and outcomes of carotid endarterectomy
J Vasc Surg
(2001)
M.H. Eslami et al.
National trends in utilization and postprocedure outcomes for carotid artery revascularization 2005 to 2007
J Vasc Surg
(2011)
C.H. Timaran et al.
Intracranial hemorrhage after carotid endarterectomy and carotid stenting in the United States in 2005
J Vasc Surg
(2009)
T.R. Vogel et al.
Outcomes of carotid artery stenting and endarterectomy in the United States
J Vasc Surg
(2009)
T.R. Vogel et al.
Carotid artery stenting: impact of practitioner specialty and volume on outcomes and resource utilization
J Vasc Surg
(2009)
A.N. Sidawy et al.
Risk-adjusted 30-day outcomes of carotid stenting and endarterectomy: results from the SVS Vascular Registry
J Vasc Surg
(2009)
P.P. Goodney et al.
Factors associated with stroke or death after carotid endarterectomy in Northern New England
J Vasc Surg
(2008)
E.A. Halm et al.
Risk factors for perioperative death and stroke after carotid endarterectomy: results of the New York Carotid Artery Surgery Study
Stroke
(2009)
J.T. McPhee et al.
Carotid artery stenting has increased rates of postprocedure stroke, death, and resource utilization than does carotid endarterectomy in the United States, 2005
J Vasc Surg
(2008)
C.B. Rockman et al.
Outcome of carotid artery interventions among female patients, 2004 to 2005
J Vasc Surg
(2011)

There are more references available in the full text version of this article.

Cited by (22)

Carotid artery disease-What we still don’t know
2021, Vascular Disease in Women: An Overview of the Literature and Treatment Recommendations
Stroke is the second leading cause of death worldwide and a major cause of disability. Eighty-seven percent of strokes are ischemic in etiology, and 10%–15% is due to carotid artery disease. Women present at a later age than men do with carotid disease. Among patients with symptomatic carotid disease, randomized clinical trials indicate that women derive less benefit than men from carotid endarterectomy (CEA). Likewise, randomized clinical trials have questioned the benefit of CEA among women with asymptomatic carotid disease, although real-world experience provides conflicting data. Reported evidence also indicates that the outcomes of transfemoral carotid artery stent procedures may be better in men than women. This chapter will examine the appropriate role for CEA and carotid artery stenting among women based on reported experience and explore possible explanations for sex discrepancies associated with carotid revascularization.
Carotid Revascularization Procedures and Perioperative Outcomes: A Multistate Analysis, 2007-2014
2019, Journal of Cardiothoracic and Vascular Anesthesia
Citation Excerpt :
The reported in-hospital mortality or stroke rates are not equivalent to 30-day periprocedural or 1-year event rates. The authors acknowledge that such a methodology may underestimate the rate of total adverse outcomes after CEA and CAS.40–43 Patient baseline demographics differed between the CEA and CAS cohorts (Table 2).
To compare in-hospital mortality, postoperative stroke, and combined stroke/mortality in carotid artery stenting (CAS) patients and carotid endarterectomy (CEA) patients.
Retrospective observational study using data from the State Inpatient Database, Healthcare Cost and Utilization Project, and Agency for Healthcare Research and Quality.
All coded CAS or CEA hospitalizations from 2007 to 2014 in California, Florida, New York, Kentucky, and Maryland.
A total of 198,120 patients, 18 years of age or older, undergoing CAS or CEA.
None.
Primary outcomes of the study were unadjusted rates and adjusted odds of in-hospital mortality, postoperative stroke, postoperative cardiovascular complications, and combined stroke/mortality, before and after correcting for confounders, following either CEA or CAS. In multivariate logistic regression analyses, in each successive individual year, CAS was associated with higher odds of in-hospital mortality (odds ratio [OR] ≥1.5 and p < 0.05), postoperative stroke (OR ≥1.4 and p < 0.05), and combined stroke/mortality (OR ≥1.5 and p < 0.05). Similar significant results were obtained when multivariate logistic regression was stratified by symptomatology. Carotid artery stenting was associated with higher odds of cardiovascular complications in 2012 (OR = 1.5, p < 0.05) and lower odds in 2009 (OR = 0.8, p < 0.05).
This study associated carotid stenting, as compared to endarterectomy, with an increased risk of dying and/or stroke. These associations persisted after statistical adjustment for patient demographics, comorbidities, and symptomatology, as well as after post-stratification by patient symptomatology. Despite this study's large, representative sample and well-defined a priori statistical methods, further research into real-world revascularization outcomes with longer-term follow-up is needed to formulate treatment guidelines.
Long-term comparative effectiveness of carotid stenting versus carotid endarterectomy in a large tertiary care vascular surgery practice
2018, Journal of Vascular Surgery
Citation Excerpt :
Second, because this study used administrative data based on medical coding, important clinical events may have been missed or coded incorrectly, thus introducing errors into the data set. Recent publications by other authors have already identified potential pitfalls with use of the National Surgical Quality Improvement Program data,17 National Inpatient Sample data,18 and even data from state-specific databases that contain the “present on admission” indicator in the study of outcomes of carotid interventions.19 However, we believe that the limitations of administrative data sets would apply equally to both endarterectomy and stent patients in our study, not lending bias toward one vs the other in our study.
Carotid interventional trials have strict inclusion and exclusion criteria that make translation of their results to the real-world population challenging. Furthermore, the specialty of the operating surgeon and the role of clinical decision-making are not well studied. This study compares the effectiveness of carotid endarterectomy (CEA) vs carotid artery stenting (CAS) in a real-world setting when the procedure is performed by fellowship-trained vascular surgeons.
A retrospective study was conducted of all consecutive patients undergoing CEA and CAS performed by vascular surgeons in a large rural tertiary health care system from 2004 to 2014. Postoperative outcomes of stroke, acute myocardial infarction (AMI), and death were analyzed at 30 days and during the long term (median follow-up of 5.5 years for CEA and 4.8 years for CAS). Standard statistical analysis was performed. Differences in long-term outcomes were expressed as cumulative incidence functions for nondeath outcomes (stroke and AMI), which account for the high death rate in this population of vascular patients, and as Kaplan-Meier curves for death itself.
From January 1, 2004, through December 31, 2014, there were 2331 carotid interventions performed (CEA, 1853; CAS, 478), all by fellowship-trained vascular surgeons. The average age of the patients was 71 years, and 63% were male, with more men in the CAS group (61.5% vs 67.8%; P = .011). Preoperatively, 30% of patients were symptomatic, and 77% of patients had high-grade stenosis in the 70% to 99% range. CEA patients were more likely to have preoperative hypertension (89.7% vs 86.2%; P = .029) and were less likely to have a history of cardiovascular disease (53.4% vs 59.4%; P = .018). There were no significant differences in 30-day outcomes between CEA and CAS (stroke, 1.1% vs 1.3% [P = .743]; AMI, 2.2% vs 1.7% [P = .474]; death, 0.7% vs 0.6% [P = .859]) or long-term outcomes (stroke, 6.8% vs 7.7% [P = .321]; AMI, 22.7% vs 21.0% [P = .886]; death, 28.4% vs 28.2% [P = .122]).
The short- and long-term outcomes after CEA vs CAS are similar when the procedure is performed in a real-world setting by fellowship-trained vascular surgeons.
Carotid artery stenting is associated with a higher incidence of major adverse clinical events than carotid endarterectomy in female patients
2017, Journal of Vascular Surgery
The optimal approach to carotid revascularization in female patients with carotid artery stenosis is widely debated. Information available is largely derived from clinical trials that include only highly selected patients. The goal of this study was to compare the early clinical outcomes in women who undergo carotid artery stenting (CAS) vs carotid endarterectomy (CEA).
Female patients undergoing CAS or CEA between January 1, 2012 and December 31, 2015, and who were included in the Procedure Targeted American College of Surgeons National Surgical Quality Improvement Program were assessed for their incidence of early postoperative complications. The primary outcome measure was 30-day incidence of a major adverse clinical event (MACE; defined as death, stroke, transient ischemic attack, or myocardial infarction/arrhythmia). Univariable analyses were used to compare results between female patients undergoing CEA and those undergoing CAS. Propensity score matching techniques were used to create a cohort of 125 CAS and CEA patients who were well matched for all known patient-, disease-, and procedure-related factors. Analysis of comparative outcomes between the propensity-matched groups was then performed.
The overall study population consisted of 5620 female CEA patients and 131 female CAS patients. Of these patients, 290 (5.2%) from the CEA group and 16 (12.2%) from the CAS group sustained a MACE in the first 30 days after their procedures. Within the propensity-matched cohort, the 30-day incidence of postoperative MACE in the CAS group of this cohort was 11.2% (14 patients) compared with 4.0% (5 patients; odds ratio, 1.01 [95% confidence interval, 1.01-7.77]; P = .04) in the CEA group.
Our analysis of a “real-world” clinical registry suggests that CAS may be inferior to CEA in female patients who require carotid artery revascularization.
The effect of carotid stenting on endarterectomy practice - A single institution experience
2016, Surgeon
The number of operations performed per surgeon is thought to determine the quality of carotid endarterectomy (CEA) surgery. The advent of carotid artery stenting (CAS) threatens to reduce the volume of CEA. This paper assesses CEA and the effects of the introduction of CAS service on outcomes.
Retrospective cohort study.
Clinical data and results of CEA were reviewed retrospectively for the treatment of carotid stenosis, between January 1988 and December 2010. CEA patients were grouped into those treated before and after the introduction of CAS to our hospital in 2001.
757 patients underwent a CEA between 1988 and 2010. The perioperative stroke rate prior to the introduction of CAS was 4.9%, and 3.3% after stent introduction in 2001. In this latter period, 85.5% had symptomatic stenosis which suggests that the patients were not low risk. The major adverse event rate (inclusive of death and myocardial infarction) post introduction of CAS from 2001 to 2010 was 4.1%. There was no correlation between post-operative stroke/MAE and procedure volume, despite the trend of decreasing CEA numbers over time.
The introduction of carotid artery stenting has led to a decrease in carotid endarterectomy volume. However, outcomes in our high risk patient population are acceptable. Therefore, CEA remains the procedure of choice for carotid artery revascularization.
Reimbursement Policies for Carotid Duplex Ultrasound that are Based on International Classification of Diseases Codes May Discourage Testing in High-Yield Groups
2016, Annals of Vascular Surgery
To curb increasing volumes of diagnostic imaging and costs, reimbursement for carotid duplex ultrasound (CDU) is dependent on “appropriate” indications as documented by International Classification of Diseases (ICD) codes entered by ordering physicians. Historically, asymptomatic indications for CDU yield lower rates of abnormal results than symptomatic indications, and consensus documents agree that most asymptomatic indications for CDU are inappropriate. In our vascular laboratory, we perceived an increased rate of incorrect or inappropriate ICD codes. We therefore sought to determine if ICD codes were useful in predicting the frequency of abnormal CDU. We hypothesized that asymptomatic or nonspecific ICD codes would yield a lower rate of abnormal CDU than symptomatic codes, validating efforts to limit reimbursement in asymptomatic, low-yield groups.
We reviewed all outpatient CDU done in 2011 at our institution. ICD codes were recorded, and each medical record was then reviewed by a vascular surgeon to determine if the assigned ICD code appropriately reflected the clinical scenario. CDU findings categorized as abnormal (>50% stenosis) or normal (<50% stenosis) were recorded. Each individual ICD code and group 1 (asymptomatic), group 2 (nonhemispheric symptoms), group 3 (hemispheric symptoms), group 4 (preoperative cardiovascular examination), and group 5 (nonspecific) ICD codes were analyzed for correlation with CDU results.
Nine hundred ninety-four patients had 74 primary ICD codes listed as indications for CDU. Of assigned ICD codes, 17.4% were deemed inaccurate. Overall, 14.8% of CDU were abnormal. Of the 13 highest frequency ICD codes, only 433.10, an asymptomatic code, was associated with abnormal CDU. Four symptomatic codes were associated with normal CDU; none of the other high frequency codes were associated with CDU result. Patients in group 1 (asymptomatic) were significantly more likely to have an abnormal CDU compared to each of the other groups (P < 0.001, P < 0.001, P = 0.020, P = 0.002) and to all other groups combined (P < 0.001).
Asymptomatic indications by ICD codes yielded higher rates of abnormal CDU than symptomatic indications. This finding is inconsistent with clinical experience and historical data, and we suggest that inaccurate coding may play a role. Limiting reimbursement for CDU in low-yield groups is reasonable. However, reimbursement policies based on ICD coding, for example, limiting payment for asymptomatic ICD codes, may impede use of CDU in high-yield patient groups.

View all citing articles on Scopus

View full text

Reasons Why Data from the Nationwide Inpatient Sample Can Be Misleading for Carotid Endarterectomy and Carotid Stenting

Section snippets

Neurologic Indications for CEA and CAS

Stroke Rates for CEA and CAS

Clinical Pitfalls of Administrative Datasets

Conclusions

J Vasc Surg

J Vasc Surg

J Vasc Surg

J Vasc Surg

J Vasc Surg

J Vasc Surg

J Vasc Surg

Risk factors for perioperative death and stroke after carotid endarterectomy: results of the New York Carotid Artery Surgery Study

Stroke

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

J Vasc Surg

Outcome of carotid artery interventions among female patients, 2004 to 2005

J Vasc Surg