From the Peripheral Vascular Surgery Society
Association between minor and major surgical
complications after carotid endarterectomy:
Results of the New York Carotid Artery
Alexander J. Greenstein, MD,aMark R. Chassin, MD, MPP, MPH,b,dJason Wang, PhD,b
Caron B. Rockman, MD,cThomas S. Riles, MD,cStanley Tuhrim, MD,eand Ethan A. Halm, MD,
MPH,b,dNew York, NY
Objective: Most studies on outcomes of carotid endarterectomy (CEA) have focused on the major complications of death
and stroke. Less is known about minor but more common surgical complications such as hematoma, cranial nerve palsy,
and wound infection. This study used data from a large, population-based cohort study to describe the incidence of minor
surgical complications after CEA and examine associations between minor and major complications.
Methods: The New York Carotid Artery Surgery (NYCAS) study examined all Medicare beneficiaries who underwent CEA
from January 1998 to June 1999 in NY State. Detailed clinical information on preoperative characteristics and complications
<30 days of surgery was abstracted from hospital charts. Associations between minor (cranial nerve palsies, hematoma, and
wound infection) and major complications (death/stroke) were examined with ?2tests and multivariate logistic regression.
Results: The NYCAS study had data on 9308 CEAs performed by 482 surgeons in 167 hospitals. Overall, 10% of patients
had a minor surgical complication (cranial nerve (CN) palsy, 5.5%; hematoma, 5.0%; and wound infection, 0.2%). Cardiac
complications occurred in 3.9% (myocardial 1.1%, unstable angina 0.9%, pulmonary edema 2.1%, and ventricular
tachycardia 0.8%). In both unadjusted and adjusted analyses, the occurrence of any minor surgical complication, CN palsy
alone, or hematoma alone was associated with 3 to 4-fold greater odds of perioperative stroke or combined risk of death
and nonfatal stroke (P < 0.0001). Patients with cardiac complications had 4 to 5-fold increased odds of stroke or
combined risk of death and stroke.
Conclusion: Minor surgical complications are common after CEA and are associated with much higher risk of death and
stroke. Patient factors, process factors, and direct causality are involved in this relationship, but future work will be
needed to better understand their relative contributions. (J Vasc Surg 2007;46:1138-46.)
Randomized controlled trials (RCTs) and national sub-
specialty practice guidelines for carotid endarterectomy
(CEA) have focused on balancing the short-term risks of
death or stroke due to the procedure with its long-term
and risk factors for the major complications of perioperative
death or stroke.1-7In the RCTs and large population-based
studies, the rates of death or stroke among symptomatic
patients average about 6% among symptomatic patients8-11
and approximately 3% among asymptomatic ones.12,13
Less is known about rates of minor complications such as
cranial nerve (CN) palsy, hematoma, bleeding, and wound
infection. These minor complications are believed to be con-
siderably more common than the major adverse events of
death or stroke, but the incidence of such events has been less
well studied. The reported rates of CN palsies are 2% to
27%,2,14-23and wound hematomas have been reported in
1.2% to 12% of cases.2,15,24-27Prior studies of these minor
complications have been limited by their focus on one type of
surgical complication, single institutions,14,15,17-19,21,25-27
small sample sizes, or highly selected surgeons and patients
who participated in RCTs.16,19
Very little is also known about the prognostic signifi-
cance of minor complications after CEA. Although large
hematomas can cause airway compromise, tracheal devia-
tion, and airway compromise, they are typically considered
to be minor and of no clinical consequence.2Cranial nerve
or detract from quality of life, for example, tracheal ob-
the vocal cords.2,14,17The importance of CN injuries has
been questioned because most resolve over time.14,17,18
This study used data from a large, population-based
cohort study to (1) describe the incidence of minor surgical
complications after CEA, including hematoma, CN palsy,
wound infection and cardiac complications and (2) exam-
ine the association between minor surgical complications
From the Departments of Surgery,aHealth Policy,band Medicine,dMount
Sinai School of Medicine; the Department of Surgery,cNew York Uni-
versity School of Medicine and the Department of Neurology,eMount
Sinai School of Medicine.
Supported by the Agency for Healthcare Research and Quality (RO1
HS09754-01), Center for Medicare Services, and the Robert Wood
Johnson Foundation (#020803).
Competition of interest: none.
Presented at the Spring meeting of the Peripheral Vascular Surgery Society,
Baltimore, Md, Jun 6-9, 2007.
Correspondence: Ethan A. Halm, MD, MPH, Division of General Internal
Medicine, Box 1087, Mount Sinai School of Medicine, One Gustave L.
Levy Pl, New York, NY 10029 (e-mail: firstname.lastname@example.org).
Copyright © 2007 by The Society for Vascular Surgery.
and the major outcomes of perioperative death or stroke.
We hypothesized that minor surgical complications, al-
though often regarded as clinically insignificant outcomes,
would be associated with a higher incidence of periopera-
tive death and stroke even after adjusting for other factors
known to influence the risk of major complications.
gery (NYCAS) study examined all Medicare beneficiaries
who underwent CEA between January 1, 1998, and June
30, 1999, in New York (NY) State. Details about the
NYCAS study have been published previously.28Briefly,
eligible cases (International Classification of Diseases, 9th Re-
vision [ICD-9] code 38.12) with Medicare fee-for-service
insurance were identified using Medicare Part A hospital
claims. Medicare managed-care patients who had a CEA
were identified with an algorithm that used the NY State
hospital discharge database, age ?65 years, and the Medi-
care eligibility files. The study was approved by the Mount
Sinai Institutional Review Board.
We obtained the medical charts of 10,817 of 11,406
(94.8%) potentially eligible patients. We excluded 628 that
were same-side operations for restenosis (n ? 308), surgery
combined with a major procedure in addition to coronary
artery bypass graft surgery (CABG, n ? 210), or no CEA
was performed (n ? 110). For the analyses presented here,
we excluded an additional 280 patients who underwent
simultaneous CEA and CABG surgery and 601 who were
missing key clinical data. The results reported are based on
9308 patients, representing 93.9% of all eligible cases.
Data collection and measurement. Detailed clinical
information was abstracted from hospital charts by trained
nurse abstractors, including sociodemographics, admission
source; neurologic, medical, and surgical history; admission
tions, and diagnostic imaging test results. We collected data
on individual comorbid conditions and calculated several co-
ified Rankin score of 4 or 5) was defined as bedridden or
unable to walk/attend bodily needs without assistance.31
There were several different indications for surgery.
The presence of a neurologic event in the 12 months
before surgery was identified as a stroke or transient isch-
emic attack (TIA). Patients with crescendo TIAs or stroke-
in-evolution were grouped as “acute syndromes.” Those
without neurologic symptoms referable to a carotid artery
distribution in the 12 months before surgery were consid-
We also abstracted the percentage of stenosis of the
internal carotid arteries that were and were not operated on
according to imaging tests. Carotid angiography was con-
sidered to be the most accurate test, followed by Doppler
ultrasonography and magnetic resonance imaging. If no
imaging test was available, we used stenosis information
from preoperative notes.
The abstractors were required to pass a series of quality
assurance and inter-rater reliability tests. All of the data
elements reported here had substantial to nearly perfect
agreement (? ? 0.60 to 1.0).
Minor surgical complications were defined as CN palsy,
hematoma, or wound infection. Cardiac complications
were defined as myocardial infarction, angina, pulmonary
edema, and ventricular tachycardia. Information about
these complications was abstracted from the review of the
inpatient medical record of the index admission and all
readmissions ?30 days of surgery anywhere in NY State.
Trained, experienced research nurses abstracted data on
these minor complications from a review of admission
notes, the daily hospital notes, and discharge summaries,
including notes from surgeons, neurologists, and medical
categories: (1) tongue deviation toward operative side, (2)
vocal cord paralysis, (3) lip or facial droop on the operative
side, (4) ear numbness on the operative side, and (5) other.
In the event that the abstractor could not determine
whether CN palsy or a stroke had occurred, a study physi-
cian investigator reviewed the chart and made a determina-
Hematoma was defined as documentation of a wound
adjectives such as, “significant,” “moderate,” “severe,” or
“large.” Descriptions of “slight” or “small amount of” hema-
Wound infections were coded when the term “wound
infection” was used in the chart.
Cardiac complications included myocardial infarction,
dia. Myocardial infarction was defined as subendocardial,
transmural, Q-wave, and non-Q-wave myocardial infarc-
tions. Unstable angina was coded if the word “unstable
angina” or “class IV angina” was seen. Ventricular tachy-
cardia was defined as more than five continuous premature
ventricular contractions at a rate ?100 beats/min.
Pulmonary edema included those cases of congestive
heart failure with associated respiratory symptoms or signs,
left-sided heart failure, pulmonary congestion, lung edema
or paroxysmal nocturnal dyspnea.
Major complications. Information about periopera-
tive deaths, strokes, and TIAs was abstracted from the
inpatient medical record of the index admission and all
readmissions ?30 days of surgery in a similar fashion as
outlined and in our previous reports.28In addition, the
records of all deaths, strokes, and TIAs were independently
reviewed by two study physicians, including a neurologist.
Initial agreement was 95%, and disagreements were re-
solved by consensus or a third reviewer, if necessary.
Analysis plan. Descriptive statistics were used to dis-
play the characteristics of the study patients. The main
analytic task was assessing association between minor sur-
gical complications (CN palsy, hematoma, or wound infec-
JOURNAL OF VASCULAR SURGERY
Volume 46, Number 6
Greenstein et al 1139
tion) and major adverse events (death and stroke). We
examined two main dependent variables in these analyses:
(1) combined rate of death or nonfatal stroke ?30 days of
surgery (the main composite outcome in the RCT and
practice guidelines of CEA), and (2) any fatal or nonfatal
strokes ?30 days of surgery.
Univariate relationships between each category of mi-
nor (CN palsy, hematoma, wound infection, cardiac) and
major complication were examined with ?2tests. In addi-
tion, we calculated median and mean length of stay (LOS)
for each individual complication and performed univariate
tests to assess for significant differences from the group
without the complication in question.
We then used multivariable logistic regression to assess
if minor complications were associated with higher rates of
combined death or stroke (or any stroke) after adjusting for
other risk factors known to influence major outcomes. The
risk-adjustment model contained other key preoperative
patient characteristics associated with death or stroke, in-
cluding age, sex, race/ethnicity, age, indication for CEA,
admission source, admission from the emergency depart-
ment, degree of stenosis of the nonoperative side, ulcer
depth on the operative side, degree of disability, and Re-
vised Cardiac Risk Index. When calculating the significance
level for LOS for each individual complication, we added
the occurrence of death/stroke to the risk adjustment
model. We used generalized estimating equations to ac-
count for the hierarchic clustering of patients by surgeons
and surgeons by hospitals. All analyses considered two-
sided P ? .05 as statistically significant and were performed
using SAS 8.2 software (SAS Institute, Cary, NC).
During the study, 9308 CEAs were performed in NY
State. Characteristics of the patients are summarized in
Table I. The mean age of the patients was 74.6 ? 6.8 years
(range, 40 to 98), and 55.7% were men. Overall, 71.5%
were operated on for asymptomatic carotid stenosis, 18.9%
for carotid TIA, 9.3% for stroke, and 0.3% for acute syn-
dromes. Nearly all patients (95.4%) were operated on for
high-grade (70% to 99%) carotid stenosis. The degree of
in the NYCAS were performed by 482 surgeons in 167
hospitals. The median hospital length of stay was 2 days
(interquartile range, 1 to 4 days).
Rates of complications. Within 30 days of surgery,
106 patients died (1.14%) and 305 strokes (3.28%) occurred,
of which 265 (2.85%) were nonfatal and 40 were fatal. Given
the overlap of 40 cases, there were thus 371 cases of periop-
(10%) had one or more minor surgical complications. Cranial
nerve palsies occurred in 514 (5.5%), hematoma in 461
(5.0%), and wound infection in 22 (0.2%). More than one
minor surgical complication occurred in 65 patients (0.7%):
60 (0.6%) had hematoma and CN palsy, three (0.03%) had a
hematoma and a wound infection, and two (0.02%) had CN
palsy and a wound infection.
Table II summarizes additional details about the CN
palsies. The two most common deficits were described as
tongue deviation towards the operative side (33%) and
lip/facial droop on the side of surgery (24%). More than
one nerve was affected in 117 patients (23%) who had a CN
palsy, and all of these had one or more cutaneous nerve
injury as well as a single motor cranial nerve injury. The
distribution of motor nerve injuries was as follows: 47
hypoglossal (40%), 43 facial or branch or both (37%), 14
branch of Vagus (12%), and 13 glossopharyngeal (11%).
Table I. Characteristics of 9308 patients undergoing
Patient risk factor
(%)Patient risk factor
55.7 Neurologic indication
Revised Cardiac Risk
Operated artery on
Valvular heart disease
Admitted from ED
Stenosis of operated
Diabetes on insulin
TIA, Transient ischemic attack.
*Carotid TIA and stroke were determined if one or more of these events
occurred within the last year.
†Acute syndrome was defined as crescendo TIA (?3 TIAs within 3 days of
surgery) or stroke in evolution (stroke with progressing or fluctuating
neurologic defects over 1 to 2 days).
JOURNAL OF VASCULAR SURGERY
1140 Greenstein et al
Reoperation was required in 146 of the 461 patients (32%)
who developed postoperative bleeding or a hematoma,
representing 64% of all patients who were taken back to the
A total of 451 cardiac complications occurred in 365
cases (3.9%). The perioperative myocardial infarction rate
was 1.1% (n ? 104), unstable angina rate was 0.87% (n ?
81), pulmonary edema rate was 2.1% (n ? 194), and
ventricular tachycardia rate was 0.77% (n ? 72).
Univariate associations between minor and major
surgical complications. Univariate associations between
minor surgical complications and death/stroke are summa-
rized in Table III. Patients with any minor surgical compli-
cation had approximately three times the rate of death or
stroke (9.7% vs 2.6%) or death (2.9% vs 0.9%, P ? .001 for
all). Those who had a CN palsy had three times the risk of
death/stroke or any stroke (P ? .001), with a trend to-
wards higher rates of death as well (P ? .08). Hematoma
combined death/stroke, any stroke, or death (P ? .001).
Wound infection had a similar pattern, but because there
were very few of these outcomes, the trends were not
statistically significant (P ? .05). In addition, each group
had a significantly longer LOS (all P ? .001 except wound
infection, for whom P ? .03).
The timing of the minor complications in relation to
when a stroke was recorded was as follows: 23 minor
complications occurred before the stroke, 38 were noted
on the same day as a stroke, and 29 were mentioned after
the day of the operation. Secondary analyses that ex-
cluded these 29 minor complications that occurred after
the index CEA produced similar results to those summa-
rized in Table II.
Univariate associations between cardiac and major
surgical complications. Univariate associations between
cardiac complications and death/stroke are also in Table
III. Patients with cardiac complication had approximately a
five times the rate of death or stroke (18.9% vs 3.4%) and
stroke alone (12.9% vs 2.9%) and approximately 12 times
the rate of death (9.9% vs 0.8%, P ? .001 for all), and had
a significantly longer LOS (P ? .001).
Multivariable associations between minor and ma-
jor surgical complications. Table IV shows that minor
complications remained strongly correlated with major ad-
verse events even after controlling for other key prognostic
triple the odds of combined death or stroke (odds ratio
[OR], 3.39, 95% confidence interval [CI], 2.65 to 4.33) or
death alone (OR, 2.89; 95% CI, 1.84 to 4.54) and nearly
quadruple the odds of any stroke (OR, 3.83; 95% CI, 2.95
to 4.97). Hematoma approximately quadrupled the odds
alone (OR, 4.30; 95% CI, 2.61 to 7.08), and stroke alone
(OR, 3.89; 95% CI, 2.82 to 5.38). Patients who required
operations for hematoma had higher odds of death/stroke
than patients with hematomas who were observed (adjust
0.0001 for both).
A multivariate subanalysis for death/stroke of these
patients showed a relative risk of 5.5 (95% CI, 3.5 to 8.5;
P ? .0001) vs a relative risk of 2.6 (95% CI, 1.7 to 3.9; P ?
.0001) for those patients with a hematoma that was ob-
served. Cranial nerve palsy tripled the odds of combined
death/stroke (OR, 2.92; 95% CI, 2.13 to 4.00) and any
stroke (OR, 3.34; 95% CI, 2.41 to 4.63), although the
relationship with death was borderline (OR, 1.71; 95% CI,
0.88 to 3.34; P ? .11).
Amongst the CN palsies, post hoc subgroup analyses
revealed increased odds of major complications for most
but not all of the individual CN palsies. For example,
death and stroke rates were higher in those with hypo-
glossal (XII) palsy (OR, 2.0; 95% CI, 1.2 to 3.3; P ?
0.01), facial or branch (VII) palsy (OR, 4.0; 95% CI, 2.6
to 6.2; P ? .0001), or both, and vagus nerve (X) palsy
(OR, 6.5; 95% CI, 2.3 to 18.8; P ? .001), but not for
glossopharyngeal (IX) palsy (OR, 1.6; 95% CI, 0.6 to
4.3; P ? .30). No statistically significant relationship was
found between wound infection and major complica-
tions. On multivariate analysis controlling for death/
stroke, each individual complication was significantly
associated with an increase in LOS.
Multivariable associations between cardiac and ma-
jor surgical complications. Table IV also shows that car-
adverse events, even after controlling for other key prog-
nostic factors. Patients who had a cardiac complication had
four to five times the odds of combined death or stroke
(OR, 5.42; 95% CI, 4.01 to 7.34) or stroke alone (OR,
4.14; 95% CI, 2.93 to 5.84) and approximately ten times
the odds of death (OR, 10.4; 95% CI, 6.68 to 16.2).
Finally, on multivariate analysis controlling for death/
stroke, cardiac complications were significantly associated
with an increase in LOS.
In this statewide study of all 9308 CEAs performed by
palsy, hematoma, or wound infection). Cranial nerve palsies
occurred in 5.5% of patients and hematomas in 5.0%. Al-
though minor surgical complications have traditionally been
considered outcomes of little or no consequence, we found
Table II. Description of cranial nerve palsies in 9308
Cranial NerveNo. CN palsies, %
Facial and/or branch (VII)
Branch of Vagus (X)
Branches of cervical plexus
?1 nerve group involved
JOURNAL OF VASCULAR SURGERY
Volume 46, Number 6
Greenstein et al 1141
that patients with a minor surgical complication had an in-
creased LOS and a threefold to fourfold greater odds of
perioperative death or stroke, even after risk adjustment.
This study confirms and extends the literature on
minor complications after CEA in the following ways.
Prior studies of minor complications after CEA focused
on single types of complications, were based at single
institutions, had small sample sizes, or were based on the
highly selected samples enrolled in RCTs. To the best of
our knowledge, the incidence of CN palsy, hematoma,
and wound infection we report here represents the larg-
est, most generalizable estimate of these complications
in widespread community practice.14-19,21,25-27
Although incidence of CN palsies after CEA reported by
others have varied widely, from 2% to 27%.2,14-23the rate of
5.5% we found in NYCAS is higher than that reported in a
lower than rates of 6.3% and 8.6% reported in the European
Carotid Surgery Trial and North American Symptomatic Ca-
rate of CN palsies in these RCTs may be partly because all
CEAs were among symptomatic patients and our NYCAS
study were three-quarters asymptomatic. Alternatively, the
differences can also be attributed to the different protocols of
the studies. In the RCTs, there was routine prospective eval-
a retrospective chart review based primarily on clinically sig-
nificant findings reported in everyday practice. The current
study reports on all CN palsies (both transient and perma-
nent), whereas others have focused exclusively on permanent
of the range reported by others (1.2% to 12%)2,15,24-27and
ison with other studies in the case of hematoma is more
challenging and should be made with caution because of the
Table IV. Adjusted* odds ratios for minor surgical and cardiac complications in 9308 patients
OR 95% CI
OR 95% CI
OR 95% CI
Cranial nerve palsy
CI, Confidence inteval; LOS, length of stay; OR, odds ratio.
admit from emergency department, admit source home, non-op stenosis ?50%, deep ulcer, severe disability, and Revised Cardiac Risk Index (comorbidity).
†Cranial nerve palsy or hematoma or wound infection.
‡Myocardial infarction, unstable angina, pulmonary edema, or ventricular tachyarrhythmia.
§The variables included in the adjusted model for LOS were death/stroke, male, white, age ?80, indication for carotid endarterectomy (transient ischemic
attack, stroke, acute syndromes) admit from emergency department, admit source home, non-op stenosis ?50%, deep ulcer, severe disability, and Revised
Cardiac Risk Index (comorbidity).
Table III. Prevalence of minor surgical and cardiac complications, association with death and stroke outcomes in 9308
Cranial nerve palsy
LOS, Length of stay.
*Cranial nerve palsy or hematoma or wound infection: The numeric result is not equal to the sum of the three components because some patients had more
than one minor surgical complication.
†Includes myocardial infarction, unstable angina, pulmonary edema, or ventricular tachyarrhythmia.
JOURNAL OF VASCULAR SURGERY
1142 Greenstein et al
substantial differences in definitions used in other investiga-
tions. For example, one series that classified a hematoma as a
radiographically evident reduction in airway cross section re-
ported an incidence as high as 26%,32whereas another series
that reported only hematomas that led to wound revision
published a rate of 1.4%.15
Finally, the wound infection rate in this study of 0.2%
was low and consistent with other studies of CEA, as well as
clean head and neck cases in general,15,33-35and the rate of
2.1% reported in NASCET.2
To our knowledge, this is the first study to demonstrate
a multivariable relationship between minor and major sur-
gical complications of CEA or any other major procedure.
There are several potential explanations for this association.
First, both types of outcomes could be correlated be-
other underlying patient factors. For example, if a case were
high internal carotid artery lesion, there might be a greater
chance of a CN injury or bleeding, or both, on that basis,
and a similarly increased chance of death or stroke due to
the more difficult repair. From the NYCAS study nurses
review of the operative notes, we obtained information on
key surgical processes of care but were not able to know the
fine details regarding the difficulty of repair. Such informa-
tion may be incompletely documented in operative notes
and this was a level of detail that was not part of the chart
review protocol in the parent study. It is also possible that
this association reflects the impact of subtle differences in
patients’ underlying severity of carotid disease, comorbid
illness burden, or hemostasis/bleeding risk of the patient.
Our multivariable models adjusted for several markers of
carotid disease, comorbidity, and death/stroke risk after
CEA, so we think this is a less likely explanation.
Second, both minor and major complications might be
related to the overall skill and quality of the surgical repair
or perioperative management, or both. Extremely skilled
and proficient surgeons and surgical teams would be ex-
pected to possess superior technical and cognitive skills and
resources that might decrease the chance of CN injuries,
bleeding, and infection as well as translate into lower rates
of death or stroke. As such, the rate of minor complications
might be seen as a surrogate marker of overall quality of
surgical care at either the surgeon or hospital level. Con-
versely, a less-experienced surgeon might have higher rates
of inadvertent CN injuries and perform a less skillful arterial
reconstruction, which might be manifested in higher rates
of postoperative stroke. That we found a significant associ-
ation between CN palsy and stroke (and combined death/
stroke) but not with death alone may provide additional
support for this hypothesis. Although post-CEA stroke is
most commonly due to thromboembolic phenomena
stemming from the endarterectomy site, many post-CEA
deaths are due to cardiac events. Strokes are thus more
likely than deaths to be related to the same process of care
factors that affect minor surgical complications (like CN
palsies), explaining their stronger association.
Third, there might be occasions where there is a direct
causal link; that is, the minor complication causes stroke or
death. In rare cases, a very large hematoma could cause
airway compromise and hypoxia, increasing the risk of both
death and stroke. Alternatively, in the presence of a serious
postoperative wound hematoma, perioperative aspirin or
other antiplatelet/anticoagulation therapy might be with-
held, which might then increase the risk of perioperative
stroke. It is most likely that a combination of these different
pathways could contribute to the overall relationship be-
tween minor and major complications. Better understand-
ing of these relationships will require further investigation.
Fourth, it is conceivable that a major complication such
as stroke could make the documentation of a minor com-
plication more likely. Amongst other factors, the occur-
with additional physicians being involved in the care of the
patient, thereby increasing the chance of more detailed
assessment and documentation.
on a few limitations. First, this was a retrospective observational
of CN palsies, hematomas, wound infections, and strokes in the
medical charts. Because we evaluated outcomes of CEA in real-
world practice, no standardized approach was used to assess or
mild CN deficits or small hematomas were not documented in
represent events that were serious enough to merit documenta-
tion in real-world practice. Because the degree of postoperative
Second, because we only reviewed inpatient medical
records (and the median LOS was 2 days), we do not have
any data on the extent to which the CN deficits persisted
over time or resolved, as many commonly do.
Third, our results may not be generalizable to other
states or patient populations. However, NY State accounts
for 14% of the Medicare population,368% of all CEAs
performed on Medicare beneficiaries.38
We found that minor surgical complications are com-
mon after CEA and are associated with a much higher risk
of death and stroke. Future work is needed to better
understand the reason for these associations, as well as to
identify potential interventions to reduce their occurrence.
We acknowledge the assistance of the Island Peer Re-
view Organization (IPRO) and the Centers for Medicare
and Medicaid Services (CMS) in providing the data that
made this research possible. The conclusions presented are
solely those of the authors and do not represent those of
JOURNAL OF VASCULAR SURGERY
Volume 46, Number 6
Greenstein et al 1143
IPRO or CMS. We would acknowledge the assistance of
Patricia Formisano, MPH, R. Edward Park, PhD, Hugh
lan, Anna Arreglado, Camille Cohen, RN, Larry Hollier,
MD, Chiaki Nakazono, Ying Qiu, Diane Thomas, MD,
Kathleen Burger, MD, and Matthew Press, MD.
Conception and design: AG, EH, MC, ST
Analysis and interpretation: AG, EH, MC, JW, CR, TR, ST
Data collection: EH, AG, JW, ST
Writing the article: AG, EH, MC, CR, TR
Critical revision of the article: EH, MC, CR, TR
Statistical analysis: JW, AG, EH, MC
Obtained funding: EH, MC
Overall responsibility: AG, EH, MC
1. Halm EA, Hannan EL, Rojas M, Tuhrim S, Riles TS, Rockman CB, et
al. Clinical and operative predictors of outcomes of carotid endarterec-
tomy. J Vasc Surg 2005;42:420-8.
2. Ferguson GG, Eliasziw M, Barr HW, Clagett GP, Barnes RW, Wallace
MC, et al. The North American Symptomatic Carotid Endarterectomy
Trial: surgical results in 1415 patients. Stroke 1999;30:1751-8.
3. Kresowik TF, Bratzler D, Karp HR, Hemann RA, Hendel ME, Grund
SL, et al. Multistate utilization, processes, and outcomes of carotid
endarterectomy. J Vasc Surg 2001;33:227-34; discussion 234-5.
An analysis of perioperative surgical mortality and morbidity in the asymp-
tomatic carotid atherosclerosis study. ACAS Investigators. Asymptomatic
Carotid Artheriosclerosis Study. Stroke 1996;27:2216-24.
5. Sundt TM, Sandok BA, Whisnant JP. Carotid endarterectomy. Com-
plications and preoperative assessment of risk. Mayo Clin Proc 1975;
of preoperative risk factors for endarterectomy for asymptomatic carotid
artery stenosis. Stroke 1998;29:750-3.
7. Hertzer NR, O’Hara PJ, Mascha EJ, Krajewski LP, Sullivan TM, Beven
EG. Early outcome assessment for 2228 consecutive carotid endarter-
ectomy procedures: the Cleveland Clinic experience from 1989 to
1995. J Vasc Surg 1997;26:1-10.
al. Benefit of carotid endarterectomy in patients with symptomatic moder-
ate or severe stenosis. North American Symptomatic Carotid Endarterec-
tomy Trial Collaborators. N Engl J Med 1998;339:1415-25.
9. Beneficial effect of carotid endarterectomy in symptomatic patients with
high-grade carotid stenosis. North American Symptomatic Carotid
Endarterectomy Trial Collaborators. N Engl J Med 1991;325:445-53.
10. Biller J, Feinberg WM, Castaldo JE, Whittemore AD, Harbaugh RE,
Dempsey RJ, et al. Guidelines for carotid endarterectomy: a statement
for healthcare professionals from a special writing group of the Stroke
Council, American Heart Association. Stroke 1998;29:554-62.
11. Moore WS, Barnett HJ, Beebe HG, Bernstein EF, Brener BJ, Brott T,
et al. Guidelines for carotid endarterectomy. A multidisciplinary con-
sensus statement from the ad hoc Committee, American Heart Associ-
ation. Stroke 1995;26:188-201.
12. Endarterectomy for asymptomatic carotid artery stenosis. Executive
Committee for the Asymptomatic Carotid Atherosclerosis Study.
13. Rothwell PM, Slattery J, Warlow CP. A systematic comparison of the
and asymptomatic stenosis. Stroke 1996;27:266-9.
14. Ballotta E, Da Giau G, Renon L, Narne S, Saladini M, Abbruzzese E, et
al. Cranial and cervical nerve injuries after carotid endarterectomy: a
prospective study. Surgery 1999;125:85-91.
15. Chiesa R, Melissano G, Castellano R, Frigerio S, Catenaccio B. Carotid
endarterectomy: experience in 5425 cases. Ann Vasc Surg 2004;18:527-34.
16. Cunningham EJ, Bond R, Mayberg MR, Warlow CP, Rothwell PM.
Risk of persistent cranial nerve injury after carotid endarterectomy.
J Neurosurg 2004;101:445-8.
et al. Role of surgical techniques and operative findings in cranial and
cervical nerve injuries during carotid endarterectomy. Eur J Vasc Endo-
vasc Surg 1998;15:528-31.
18. Schauber MD, Fontenelle LJ, Solomon JW, Hanson TL. Cranial/
cervical nerve dysfunction after carotid endarterectomy. J Vasc Surg
19. Forssell C, Takolander R, Bergqvist D, Bergentz SE, Gramming P,
Kitzing P. Cranial nerve injuries associated with carotid endarterec-
tomy. A prospective study. Acta Chir Scand 1985;151:595-8.
20. Maniglia AJ, Han DP. Cranial nerve injuries following carotid endar-
terectomy: an analysis of 336 procedures. Head Neck 1991;13:121-4.
21. Knight FW, Yeager RM, Morris DM. Cranial nerve injuries during
carotid endarterectomy. Am J Surg 1987;154:529-32.
22. Dehn TC, Taylor GW. Cranial and cervical nerve damage associated
with carotid endarterectomy. Br J Surg 1983;70:365-8.
23. Massey EW, Heyman A, Utley C, Haynes C, Fuchs J. Cranial nerve
paralysis following carotid endarterectomy. Stroke 1984;15:157-9.
24. Clagett GP, Krupski WC. Antithrombotic therapy in peripheral arterial
occlusive disease. Chest 1995;108:431S-43S.
25. Self DD, Bryson GL, Sullivan PJ. Risk factors for post-carotid endarter-
ectomy hematoma formation. Can J Anaesth 1999;46:635-40.
26. Kunkel JM, Gomez ER, Spebar MJ, Delgado RJ, Jarstfer BS, Collins
GJ. Wound hematomas after carotid endarterectomy. Am J Surg 1984;
27. Welling RE, Ramadas HS, Gansmuller KJ. Cervical wound hematoma
after carotid endarterectomy. Ann Vasc Surg 1989;3:229-31.
28. Halm EA, Tuhrim S, Wang JJ, Rojas M, Hannan EL, Chassin MR. Has
evidence changed practice?: appropriateness of carotid endarterectomy
after the clinical trials. Neurology 2007;68:187-94.
29. Lee TH, Marcantonio ER, Mangione CM, Thomas EJ, Polanczyk CA,
Cook EF, et al. Derivation and prospective validation of a simple index
for prediction of cardiac risk of major noncardiac surgery. Circulation
30. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of
classifying prognostic comorbidity in longitudinal studies: development
and validation. J Chronic Dis 1987;40:373-83.
31. Rankin J. Cerebral vascular accidents in patients over the age of 60. II.
Prognosis. Scott Med J 1957;2:200-15.
32. Carmichael FJ, McGuire GP, Wong DT, Crofts S, Sharma S, Montan-
era W. Computed tomographic analysis of airway dimensions after
carotid endarterectomy. Anesth Analg 1996;83:12-7.
33. Rosato L, Avenia N, Bernante P, De Palma M, Gulino G, Nasi PG, et al.
Complications of thyroid surgery: analysis of a multicentric study on
14,934 patients operated on in Italy over 5 years. World J Surg
34. Olson M, O’Connor M, Schwartz ML. Surgical wound infections. A
5-year prospective study of 20,193 wounds at the Minneapolis VA
Medical Center. Ann Surg 1984;199:253-9.
35. Culver DH, Horan TC, Gaynes RP, Martone WJ, Jarvis WR, Emori
TG, et al. Surgical wound infection rates by wound class, operative
procedure, and patient risk index. National Nosocomial Infections
Surveillance System. Am J Med 1991;91:152S-7S.
36. Kaiser Family Foundation. Medicare Fact Sheets. 2007. Available at:
York State: 1995-1997. Albany, NY: New York State Department of
Revisiting the appropriateness of carotid endarterectomy. Stroke 2003;
Submitted Jun 1, 2007; accepted Aug 17, 2007.
JOURNAL OF VASCULAR SURGERY
1144 Greenstein et al
Dr Chris Kwolek (Boston, Mass). I would like to congratu-
late Dr Greenstein and his coauthors on a very nice presentation
and thank them for providing me with a copy of their manuscript
well in advance of this meeting. The authors described the results
of a contemporary series, although retrospective, of carotid endar-
terectomy and a large cohort of patients performed by 482 sur-
geons in over 167 hospitals within the state of New York with a
very reasonable death and stroke rate for symptomatic and asymp-
tomatic patients. In this era of increased concern about the role of
carotid endarterectomy versus carotid angioplasty and stenting,
and the management of the patients with carotid stenosis, I think
this paper will be increasingly important. However, I am a little bit
intrigued by the methodology and I will get to my question in a
We have reported a 10% incidence of minor complications,
including an incidence of 5.5% of cranial nerve injuries, 5% incidence
approximately 1.5% to 2% required a surgical re-exploration, and the
author very interestingly notes that the existence of one of these
potentially death, at least for the cranial nerve injuries both stroke
alone and for hematoma, stroke, and death in a threefold to fourfold
manner. You alluded to it in your presentation, but one wonders,
what is the etiology of this relationship? It is that these are complex
patients? I think very appropriately, patients who have combined
carotid CABGs or redo procedures were factored out of the evalua-
tion initially, but it begs the question. Is this merely a marker perhaps
for operator experience?
So my first question is, do you have any data about the results
of individual surgeons since you allude to it in your final slide and
how did that play out with respect to results? And then the second
question, we also know there has been some interest on a national
level both by the American College of Surgeons and the NSQIP
[National Surgical Quality Improvement Program] as to hospital
dependent rates as well as surgeon independent rates. I would like
you to comment on the higher incidence of significant complica-
tions in a very contemporary series albeit it across the large state-
wide registry. When we compare your results to such retrospective
reviews that Bruce Perler has presented from the state of Maryland
and the large retrospective reviews from similar databases in the
state of California with over 40,000 patients, was it just that New
York has more problems, or is it the methodology where much like
the NSQIP, these are independent nurse reviewers going back and
looking specifically at charts?
My final question, I think this lends itself to future investiga-
tion and, not being familiar with the New York statewide project,
do the investigators or the Agency for Healthcare Research and
Quality (AHRQ) have plans to perform future studies perhaps
looking at carotid angioplasty and stenting and comparing that to
the results of carotid endarterectomy in a contemporary series
within the state of New York? I thank the society for the privilege
of the floor.
Dr Alexander J. Greenstein (New York, NY). There are
several possible explanations for the relationship that we detected
types of complications might be more common in sicker patients.
We think our CEA-specific risk adjustment model should have
accounted for most of this variation. Second, the presence of
anatomic variants or other technically difficult aspects of a case
might lead to an increase of both complications. These data are not
readily reported in the charts or easily extracted, so we were not
able to discount this as a possibility. Third, a minor complication
(eg, a large hematoma) could itself lead to death or stroke. This
wouldn’t explain the sequence for the vast majority of major
complications. Fourth, the risk of minor and major complications
might reflect the technical quality and experience of the surgeon,
operative team, and/or hospital. We think this is the most likely
explanation, though this is difficult to prove with observational
We did observe significant variations in the rates of minor and
major complications at both the individual surgeon and hospital
level. In analyses done with the surgeon (or hospital) as the unit of
analyses (not the patient as was done in this paper), surgeons (and
hospitals) with higher rates of minor complications also had higher
rates of major complications. These provider-level comparisons are
tricky because those with very low volume will have unstable point
estimates of adverse events based on chance alone. Excluding
low-volume providers solves this problem, but then limits the
generalizability of the results.
Dr Kwolek is correct that the methodology we used to
ascertain adverse events was direct clinical chart review, much
like the NSQIP. We had trained, independent research nurses
carefully review the full inpatient medical chart, including ad-
mission, discharge, daily progress, and operative notes, and
diagnostic imaging reports, and recorded information on minor
and major complications. We also did chart review on any
subsequent hospitalizations in the 30 days after surgery to
detect late strokes or deaths. Studies that rely solely on admin-
istrative billing data or just the initial hospitalization tend to
report lower rates of adverse outcomes.
Finally, the NYCAS Study reflects practice in 1998 and 1999,
so we have no data on outcomes of carotid angioplasty and
Dr Vik Kashyap (Cleveland, Ohio). Can you explain the
methodology again to me? There were 9500 charts. How much
of the information was precoded or collected prospectively at
the time of the operations versus how many variables did you
have to go through and cull through these close to 10,000
charts to get?
Dr Greenstein. As we indicate in the article, all the data in the
NYCAS Study were based on information documented in the
hospital chart as part of usual care. Our trained research nurses
retrospectively abstracted all clinical data from these hospital charts
several years after the actual care occurred. There was no prospec-
tive coding of the data for this study as might be done in a
randomized clinical trial.
Dr Kashyap. And then, in terms of the very specifics that you
have on cranial nerve policies, these were confirmed by physicians
or neurologists or were these all abstracted from the texts of the
data that was built?
Dr. Greenstein. It was a combination of the research nurse
looking at the medical chart and seeing a specific diagnosis of
cranial nerve palsy or documentation of signs or symptoms sugges-
tive of a nerve injury. If there is any question, the nurse abstractors
discussed the details with the investigative team. Cases in which a
stroke, TIA, or death was recorded by the nurses were reviewed by
two physician reviewers (including a neurologist).
Dr Kwolek. I apologize, there was one more important point
that did not come out of the presentation, but is in the manuscript,
which will be important as we start comparing this, because this
paper will be cited by the carotid angioplasty stenting advocates,
since there is a higher incidence of cranial nerve injury. One of the
subsets that needs to be broken out is temporary numbness around
the ear or earlobe or facial numbness. After an endarterectomy, I
would argue that larger number of our patients have that deficit
and whether that is coded as significant or not, needs to be
Dr Greenstein. I have to look in to that. Thank you.
Unidentified speaker. I think my question is along the same
lines. There is a lot of difference between having a deviation of
tongue versus hoarseness of voice versus not being able to
breathe versus losing a high pitch in your voice. So, it seems to
me to be not as meaningful to have a big bucket of patients
having cranial nerve deficiency, because it really is not going to
JOURNAL OF VASCULAR SURGERY
Volume 46, Number 6
Greenstein et al 1145
get home the message of the intensity of the consequence or the Download full-text
significance. Did you segregate and then pool it together or did
you just add it as a collective cranial nerve deficiency? Because
there are studies that say that as high as 16% incidence of
hypoglossal nerve palsy, but we do not take that as seriously as
hoarseness of voice or inability to swallow and losing the
Dr Greenstein. Again, this is one of the limitations of a retro-
spective study of this nature. We have large numbers, but we do not
have the richness of clinical detail about the severity or persistence of
cranial nerve injuries that could answer this question.
Dr George Levinson (Bethesda, Md). Was there any cor-
relation between the use of clopidogrel preoperatively and the
incidence of early postoperative stroke due to thrombosis and
bleeding? In other words, with Plavix [Sanofi-Aventis, Bridge-
water, NJ], was there more bleeding and fewer strokes, and if
one did not have Plavix immediately preoperatively, were there
more strokes and fewer bleeding?
Dr Greenstein. The NYCAS database has data on the use of
antiplatelet agents and anticoagulants, but their potential impact
on risk of thrombotic or bleeding complications has not yet been
JOURNAL OF VASCULAR SURGERY
1146 Greenstein et al