Professional Documents
Culture Documents
Dakouraridi 2017
Dakouraridi 2017
ABSTRACT
Objective: The aim of this study was to analyze the rates, reasons, and risk factors of 30-day readmission, both planned
and unplanned, after carotid revascularization as well as to evaluate major outcomes associated with those readmissions.
Methods: Using the Premier Healthcare database, we retrospectively identified patients undergoing carotid endarter-
ectomy (CEA) and carotid artery stenting (CAS) between 2009 and 2015. The primary outcome was 30-day all-cause
readmission. Secondary outcomes included mortality and overall cost associated with readmissions. Univariate and
multivariate analyses were used and further validated using coarsened exact matching on baseline differences between
CEA and CAS patients.
Results: A total of 95,687 patients underwent carotid revascularization, 13.5% of whom underwent CAS. Crude 30-day
readmission rates were 6.5% after CEA vs 6.1% after CAS (P ¼ .10). Stroke, bleeding, pneumonia, and respiratory failure
were the most common reasons for readmission after both CEA and CAS (6.7% vs 8.3%, 6.9% vs 5.3%, 3.4% vs 2.4%, and
4.4% vs 3.9%; all P > .05). Myocardial infarction and wound complications were more likely to be an indication for
readmission after CEA (4.1% vs 2.5% and 4.1% vs 1.5%, respectively; P < .05). On the other hand, readmissions due to
vascular or stent-related complications were more likely after CAS compared with CEA (5.8% vs 3.8%; P ¼ .003). On
multivariate analysis, CEA was found to be associated with 41% higher odds of readmission than CAS (adjusted odds ratio,
1.41; 95% confidence interval, 1.29-1.54; P < .001). Age, female gender, emergency/urgent procedures, concomitant cardiac
procedures, rural hospitals, and Midwest region were significantly associated with 30-day readmission. Other risk factors
included major preoperative comorbidities (diabetes, congestive heart failure, renal disease, chronic obstructive pul-
monary disease, peripheral vascular disease, and history of cancer) as well as the occurrence of postoperative stroke and
renal complications during the index admission and nonhome discharge. Coarsened exact matching between CEA and
CAS patients also yielded higher adjusted rates of readmission after CEA (6.2% vs 4.9%; P < .001). On the other hand,
patients readmitted after CAS had a longer length of hospital stay (5 days vs 4 days; P ¼ .001), increased readmission
mortality (6.2% vs 2.8%; P < .001), and higher rehospitalization costs ($8903 vs $7629; P ¼ .01) compared with those
readmitted after CEA.
Conclusions: Our results show that CAS is associated with lower 30-day readmission rates compared with CEA. However,
CAS readmissions are more complex and are associated with higher mortality and costs. We have also identified patients
who are at high risk of readmissions, which can help focus attention on interventions that can improve the management
of these patients and reduce readmission rates. (J Vasc Surg 2017;-:1-13.)
Carotid revascularization has been shown to effectively endarterectomy (CEA) and carotid artery stenting
reduce the risk of ischemic stroke in selected patients (CAS).2-8 However, reducing readmission has become an
with severe carotid artery stenosis.1 Prior studies have important measurable quality outcome and an area of
focused on the safety and efficacy of carotid active research because of the increased pressure placed
on hospitals and the planned monetary penalties to be
From the Johns Hopkins Bayview Medical Center. imposed on centers with higher than expected readmis-
Author conflict of interest: none. sion rates. This has been established through the Hospital
Presented as a poster at the 2017 Vascular Annual Meeting of the Society for Readmissions Reduction Program and the Patient
Vascular Surgery, San Diego, Calif, May 31-June 3, 2017. Protection and Affordable Care Act.1,9-15 Between 2009
Additional material for this article may be found online at www.jvascsurg.org.
and 2011, nearly 10% of Medicare patients who underwent
Correspondence: Mahmoud B. Malas, MD, MHS, FACS, Johns Hopkins Bayview
Medical Center, 4940 Eastern Ave, Bldg A/5, Ste 547, Baltimore, MD 21224 carotid revascularization returned to the hospital within
(e-mail: bmalas1@jhmi.edu). 30 days of discharge.1 CEA has been considered the surgi-
The editors and reviewers of this article have no relevant financial relationships to cal procedure with the third highest readmission rate.16
disclose per the JVS policy that requires reviewers to decline review of any However, a study published in the Journal of the
manuscript for which they may have a conflict of interest.
American College of Cardiology found that CAS was asso-
0741-5214
Copyright Ó 2017 by the Society for Vascular Surgery. Published by Elsevier Inc. ciated with higher 30-day readmission rates compared
http://dx.doi.org/10.1016/j.jvs.2017.05.097 with CEA.
1
2 Dakour Aridi et al Journal of Vascular Surgery
--- 2017
Race <.001 CHF, Congestive heart failure; COPD, chronic obstructive pulmonary
disease; MI, myocardial infarction; SD, standard deviation.
White 68,408 (82.6) 10,367 (80.6) Values are reported as number (%) unless otherwise indicated.
Black 3450 (4.2) 604 (4.7)
Others 10,959 (13.2) 1899 (14.7)
Insurance type <.001 between CEA and CAS patients, including age, gender,
Medicare 64,679 (78.5) 9893 (77.6) race, insurance type, admission type (elective, emer-
gency, and urgent), concomitant cardiac procedures,
Medicaid 2359 (2.9) 509 (4.0)
teaching status, region (rural vs urban), provider area,
Private insurance 13,406 (16.3) 2031 (15.9)
symptomatic status, medical comorbidities (old
Other 1978 (2.4) 315 (2.5)
myocardial infarction [MI], congestive heart failure [CHF],
Admission type <.001
renal disease, connective tissue disease, paraplegia/
Elective 66,504 (80.3) 8707 (67.7) hemiplegia, peripheral vascular disease, and cancer), and
Emergency 5870 (7.1) 1773 (13.8) comorbidity index. Some of the advantages of CEM over
Urgent 10,008 (12.1) 2317 (18.0) standard matching approaches, such as propensity score
Others 435 (0.5) 73 (0.6) matching, are that it requires fewer postestimation as-
Concomitant cardiac 18,948 (22.9) 3548 (27.6) <.001 sumptions about how to define a match, automatically
procedures balances treatment and control populations, and has
Teaching hospitals 34,966 (42.5) 7675 (59.8) <.001 superior computational properties for large data sets.
Region <.001 Analysis was performed using Stata version 14.1 (Stata-
Rural 11,128 (13.5) 892 (7.0) Corp, College Station, Tex). Significance tests were two
Urban 71,189 (86.5) 11,948 (93.1) sided with a .05 significance level.
Provider area <.001 RESULTS
Midwest 15,202 (18.5) 2631 (20.5) A total of 95,687 patients underwent carotid revascular-
North-East 10,511 (12.8) 2173 (16.9) ization between 2009 and 2015: 82,817 CEA (86.5%) and
South 44,878 (54.5) 6478 (50.5) 12,870 (13.5%) CAS. Of those, 6118 (6.4%) were readmitted
West 11,726 (14.2) 1558 (12.1) within 30 days of discharge (CEA, 6.5%; CAS, 6.1%; P ¼ .10).
Comorbidities Comparison of baseline characteristics between the two
Symptomatic 10,591 (12.9) 2334 (18.5) <.001 groups is shown in Table I. Occurrence of more than one
Diabetes 28,658 (34.6) 4427 (34.4) .65 readmission was more common after CAS than after
Old MI 10,573 (12.8) 1723 (13.4) .05
CEA (31% vs 25%; P < .001). Compared with patients
not readmitted within 30 days, those readmitted were
CHF 7047 (8.5) 1701 (13.2) <.001
older (73 6 8.6 years vs 71.8 6 8.3 years; P < .001), were
Renal disease 9488 (11.5) 1727 (13.4) <.001
more often women (43.4% vs 41.2%; P < .001) and African
COPD 20,134 (24.3) 3120 (24.2) .87
Americans (5.6% vs 4.1%; P < .001), and were mostly
Connective tissue 1770 (2.1) 186 (1.5) <.001 covered by Medicare (81.7% vs 78.1%; P < .001; Table II).
disease
Moreover, they were more likely to be readmitted after
Peptic ulcer disease 531 (0.6) 72 (0.6) .28
emergency or urgent index procedures (36.2% vs 19.9%;
Paraplegia/hemiplegia 1593 (1.9) 755 (5.9) <.001 P < .001) or with concomitant cardiac procedures
Liver disease 694 (0.8) 104 (0.8) .73 (34.2% vs 22.8%; P < .001).
Peripheral vascular 20,790 (25.1) 4067 (31.6) <.001 Readmitted patients were more likely to be symptom-
disease atic on initial presentation (21.4% vs 13.1%; P < .001) and
Cancer 1471 (1.8) 315 (2.5) <.001 had a higher prevalence of comorbid conditions, such as
(Continued) diabetes (40.9% vs 34.2%), history of MI (15.5% vs 12.7%),
4 Dakour Aridi et al Journal of Vascular Surgery
--- 2017
Rural 11,218 (12.6) 802 (13.2) .18 Home 77,265 (86.6) 3720 (61.2) <.001
Urban 77,855 (87.4) 5282 (86.8) Home under home 6792 (7.6) 687 (11.3)
health care
Provider area
General hospital 462 (0.52) 52 (0.9)
Midwest 16,569 (18.6) 1264 (20.8) <.001
Skilled nursing facility 3041 (3.4) 585 (9.6)
North-East 11,860 (13.3) 824 (13.5)
Other rehabilitation or 1672 (1.9) 1036 (17.0)
South 48,167 (54.1) 3189 (52.4) hospice
West 12,477 (14.0) 807 (13.3)
AKI, Acute kidney injury; CAS, carotid artery stenting; CEA, carotid
Comorbidities endarterectomy; CHF, congestive heart failure; COPD, chronic
obstructive pulmonary disease; DVT, deep venous thrombosis; IQR,
Symptomatic 11,632 (13.1) 1293 (21.4) <.001
interquartile range; MI, myocardial infarction; PE, pulmonary embo-
Diabetes 30,585 (34.2) 2500 (40.9) <.001 lism; SD, standard deviation; UTI, urinary tract infection.
Values are reported as number (%) unless otherwise indicated.
Old MI 11,346 (12.7) 950 (15.5) <.001
CHF 7703 (8.6) 1045 (17.1) <.001
Renal disease 9991 (11.2) 1224 (20.0) <.001 CHF (17.1% vs 8.6%), chronic obstructive pulmonary
COPD 21,453 (24.0) 1801 (29.4) <.001 disease (COPD; 29.4% vs 24.0%), and peripheral vascular
Connective tissue 1805 (2.0) 151 (2.5) .02 disease (29.6% vs 25.7%), compared with patients who
disease were not readmitted within 30 days (all P < .001). Read-
Peptic ulcer disease 550 (0.6) 53 (0.9) .02 mitted patients experienced more stroke (5.3% vs 1.2%),
Paraplegia/hemiplegia 1903 (2.1) 445 (7.3) <.001 hemorrhage/bleeding (11.4% vs 7.3%), cardiac complica-
Liver disease 735 (0.8) 63 (1.0) .08 tions (including acute MIs, CHF, and arrhythmias; 10.1%
(Continued) vs 5.8%), and more respiratory (4.7% vs 1.5%) and renal
Journal of Vascular Surgery Dakour Aridi et al 5
Volume -, Number -
35
30
25
CEA CAS
Percentage
20
15
10
Fig. Most common readmission indications after carotid endarterectomy (CEA) and carotid artery stenting
(CAS). GI, Gastrointestinal.
(6.9% vs 2.6%) complications during the index admission likely to be an indication for readmission after CEA
compared with patients not readmitted (all P < .001). Me- compared with CAS (4.1% vs 2.5% [P ¼ .02] and 4.1% vs
dian index length of stay was significantly higher in read- 1.5% [P < .001], respectively; Table III). On the other hand,
mitted patients (median [interquartile range], 3 [1-8] vs 1 readmissions due to vascular or stent-related complica-
[1-3] days; P < .001). A significant portion of readmitted pa- tions were more likely after CAS than after CEA (5.8% vs
tients were not discharged home initially, with 10.5% 3.8%; P ¼ .003).
transferred to a general hospital or a skilled nursing facility
Predictors of 30-day readmission. After multivariate
and 11.3% discharged home under a home health service
adjustment, CEA was found to be associated with 41%
organization.
higher odds of readmission compared with CAS
Causes of readmission. A total of 7153 readmissions (adjusted odds ratio [aOR], 1.41; 95% confidence interval
occurred, mostly 1 week after discharge from the hospi- [CI], 1.29-1.54). Advanced age (aOR, 1.01; 95% CI,
tal. Other than nonspecific complaints such as fatigue, 1.00-1.01), female gender (aOR, 1.09; 95% CI, 1.03-1.16),
fever, headache, syncope, and admission for physical emergency and urgent procedures (aOR, 1.43 [95% CI,
therapy, most common indications of these readmis- 1.32-1.55] and 1.14 [95% CI, 1.03-1.26], respectively), and
sions (n ¼ 7153) were related to cardiac, neurologic, infec- concomitant cardiac procedures (aOR, 1.18; 95% CI,
tious, and respiratory complications (Fig). Stroke was one 1.11-1.26) were significantly associated with higher odds of
of the most common reasons for readmission after both readmission (P < .05). Other risk factors included hospi-
CAS and CEA (8.3% vs 6.7%; P ¼ .07), followed by hem- tals’ location (rural and Midwest areas had higher odds of
orrhage/shock/bleeding (5.3% vs 6.9%; P ¼ .06), pneu- readmission); major comorbidities, such as diabetes,
monia (2.4 vs 3.4%; P ¼ .12), and respiratory failure (3.9% vs CHF, renal disease, COPD, peripheral arterial disease, and
4.4%; P ¼ .5). MI and wound complications were more a history of cancer; and nonhome discharge. The model
6 Dakour Aridi et al Journal of Vascular Surgery
--- 2017
was able to predict only 8.8% of variability in 30-day penalize readmissions after vascular procedures in the
readmission (R2 ¼ 8.8%; C statistic ¼ 69.5%). Further future is high; thus, reducing unplanned readmissions
including in-hospital complications did not significantly has become of prominent importance to improve qual-
improve predictability (R2 ¼ 8.9%; C statistic ¼ 69.6%). ity of care and to reduce significant expenses both to
Among in-hospital complications, stroke increased the the health care system and to the patient.19
odds of 30-day readmission by 40% (aOR, 1.40; 95% CI, The overall 30-day readmission rate in our study was
1.20-1.64; P < .001) and renal adverse events by 14% (aOR, around 6.4%, which is comparable to that reported using
1.14; 95% CI, 1.01-1.30; P ¼ .04; Table IV). other large population databases.20-22 Al-Damluji et al1
reported a much higher unadjusted 30-day readmission
Subgroup analysis: CEM. Exact one-to-one matching
rate (9.6%). For CEA, overall reported readmission rates
was performed on the basis of baseline characteristics
are around 6.5%15,20,22,23 but can be as high as 9.4%.24
that were shown to be significantly different between
On the other hand, reported readmission rates after
CEA and CAS patients, including age, gender, race, in-
CAS range between 9.7%22 and 11.11%,25 which is higher
surance coverage, admission type, presence of concom-
compared with our study (6.1%).
itant cardiac procedures, region (urban vs rural), provider
Our study shows an association between CEA and
area, hospital teaching status, and medical comorbidities
increased all-cause readmission rates after risk adjust-
that make up the Charlson Comorbidity Index (history of
ment (aOR, 1.41; 95% CI, 1.29-1.54; P < .001) as well as after
MI, CHF, connective tissue disease, paraplegia/hemiple-
exact matching (6.2% vs 4.9%; P < .001). This is in contrast
gia, peripheral arterial disease, and renal disease). This
to two previous studies that demonstrated higher
yielded two comparison groups each with 8966 patients
adjusted risk of readmission after CAS compared with
(Supplementary Table II, online only). Comparison of the
CEA.1,25 Galinanes et al25 used the Centers for Medicare
two groups showed higher adjusted 30-day readmission
and Medicaid Services Medicare Provider Analysis and
rates after CEA compared with CAS (6.2% vs 4.9%; P <
Review file between 2005 and 2009 and showed
.001). Secondary outcomes that are significantly different
increased odds of readmission in patients undergoing
between CEA and CAS were in-hospital stroke (1.8% vs
CAS compared with those undergoing CEA within
1.2%; P < .01), arrhythmia (4.9% vs 6.7%; P < .001), hem-
30 days (aOR, 1.21; 95% CI, 1.15-1.26; P < .0001), 60 days
orrhage/bleeding (8.6% vs 7.4%; P < .01), acute kidney
(aOR, 1.34; 95% CI, 1.29-1.39; P < .0001), and 90 days after
injury (1.8% vs 1.5%; P ¼ .05), and discharge disposition
discharge (aOR, 1.36; 95% CI, 1.31-1.40), with coronary ar-
(Table V). We further controlled for in-hospital post-
tery disease being the most common reason for read-
operative complications, and readmission was still 35%
mission. In addition, Al-Damluji et al1 showed 13%
higher after CEA compared with CAS (aOR, 1.35; 95% CI,
higher adjusted risk of readmission after CAS (aOR, 1.13,
1.18-1.54; P < .001).
95% CI, 1.08-1.18; P < .001) using Medicare beneficiaries
Outcomes of readmitted patients. Patients readmitted from 2009 to 2011 with the same relationship holding
after CAS had a longer length of hospital stay (5 days vs true in a propensity-matched cohort (OR, 1.18; 95% CI,
4 days; P ¼ .001) and significantly higher readmission 1.07-1.23) regardless of symptomatic status, age, sex, and
mortality (6.2% vs 2.8%; P < .001) compared with those race. An important finding by the authors was that the
readmitted after CEA. Adjusted readmission mortality variation in proportional use of CAS was not associated
was two times higher in patients readmitted after CAS with differences in hospital risk-standardized readmis-
vs CEA (aOR, 2.0; 95% CI, 1.5-2.8; P < .001). Throughout sion rates. Thus, hospitals with more frequent CAS use
the readmission, patients were also subject to various will not necessarily be disadvantaged when hospitals
adverse events complicating the course of their rehospi- with higher than expected readmissions are penalized.
talization after CEA and CAS. These included bleeding However, an editorial comment on that study cautioned
(4.8% after CEA vs 3.1% after CAS; P ¼ .02), cardiac (6.7% that the poorer performance of CAS might be due to se-
vs 6.1%; P ¼ .53), respiratory (3.5% vs 5.0%; P ¼ .03), and lection bias, which could not be adjusted for, and that
renal (5.3% vs 6.7%; P ¼ .10) complications. The median the small difference in readmission rates between the
total, fixed, and variable costs of readmission were higher two procedures might not be meaningful as it could
if the readmission occurred after CAS compared with be a potential artifact of the study’s large sample size.26
CEA ($8903, $4313, and $4390 vs $7629, $3762, and The lower readmission rates after CAS in our study
$3651, respectively). Adjusted rehospitalization costs after might be explained by several factors, such as the use
CAS were higher by around U.S. $700 compared with of different databases and different time frames as well
those after CEA (Table VI). as the continuous decline in 30-day major adverse events
after CAS due to the refinement in CAS techniques,
DISCUSSION development of better stents and protection devices,
Carotid revascularization is a high-volume procedure better selection of patients for each procedure, and
with significant readmission rates. The likelihood that increased expertise of operators over time. Unlike the
the Centers for Medicare and Medicaid Services will cited studies that examined Medicare data between
Journal of Vascular Surgery Dakour Aridi et al 7
Volume -, Number -
Table III. Most common indications for 30-day readmission after carotid endarterectomy (CEA) and carotid artery stenting
(CAS)
CEA CAS P value Total
No. of readmissions 6205 (86.7) 948 (13.3) 7153
More than one readmission 1552 (25.0) 293 (30.9) <.001 1845 (25.8)
Time to readmission, days (IQR) 6 (1-16) 7 (1-17) .89 6 (1-16)
Hemorrhage/shock/bleeding 428 (6.9) 50 (5.3) .06 478 (6.7)
Vascular or stent complication 235 (3.8) 55 (5.8) .003 290 (4.1)
Cardiac
MI 255 (4.1) 24 (2.5) .02 279 (3.9)
Heart failure 212 (3.4) 33 (3.5) .92 245 (3.4)
Chest pain 221 (3.6) 34 (3.6) .97 255 (3.6)
Arrhythmia 240 (3.9) 35 (3.7) .79 275 (3.8)
Other cardiac 16 (0.3) 1 (0.1) .37 17 (0.2)
Respiratory
Pneumonia 211 (3.4) 23 (2.4) .12 234 (3.3)
Respiratory failure 272 (4.4) 37 (3.9) .50 309 (4.3)
Other 433 (7.0) 56 (5.9) .22 489 (6.8)
Renal or urologic
Acute kidney injury 151 (2.4) 21 (2.2) .68 172 (2.4)
Others 180 (2.9) 21 (2.2) .23 201 (2.8)
Neurologic
Stroke 417 (6.7) 79 (8.3) .07 496 (6.9)
Paraplegia 5 (0.1) 0 .38 5 (0.1)
Other neurologic 485 (7.8) 92 (9.7) .05 577 (8.1)
Infectious
Wound complication 254 (4.1) 14 (1.5) <.001 268 (3.7)
Sepsis 184 (3.0) 36 (3.8) .17 220 (3.1)
UTI/pyelonephritis 31 (0.5) 5 (0.5) .91 36 (0.5)
Other infection 173 (2.8) 12 (1.3) .01 185 (2.6)
Gastrointestinal
Bowel obstruction 16 (0.3) 3 (0.3) .73 19 (0.3)
Bowel ischemia 5 (0.1) 0 1.0 5 (0.1)
Venous thromboembolic
Deep venous thrombosis 13 (0.2) 3 (0.3) .46 16 (0.2)
Pulmonary embolism 44 (0.7) 5 (0.5) .67 49 (0.7)
Other venous 5 (0.1) 0 1.0 5 (0.1)
Reoperation
Carotid revascularization 17 (0.3) 3 (0.3) .74 20 (0.3)
Other vascular reoperation 74 (1.2) 16 (1.7) .20 90 (1.3)
Other abdominal 183 (3.0) 33 (3.5) .37 216 (3.0)
Cancer 58 (0.9) 5 (0.5) .21 63 (0.9)
Nonspecific/others
Headache 78 (1.3) 9 (1.0) .42 87 (1.2)
Nausea/vomiting/dysphagia 128 (2.1) 12 (1.3) .10 140 (2.0)
Syncope/collapse 183 (3.0) 23 (2.4) .37 206 (2.9)
Fever 79 (1.3) 11 (1.2) .77 90 (1.3)
Fatigue 120 (1.9) 14 (1.5) .33 134 (1.9)
Physical therapy 1065 (17.2) 199 (21.0) .004 1264 (17.7)
Unknown 1360 (21.9) 213 (22.5) .70 1573 (22.0)
IQR, Interquartile range; MI, myocardial infarction; UTI, urinary tract infection.
Values are reported as number (%) unless otherwise indicated.
8 Dakour Aridi et al Journal of Vascular Surgery
--- 2017
Table IV. Bivariate and multivariate logistic regression models of the predictors of 30-day readmission after carotid
revascularization
Unadjusted OR (95% CI) P value Adjusted OR (95% CI) P value
Procedure
CAS 1.0 (reference) 1.0 (reference)
CEA 1.07 (0.99-1.15) .097 1.41 (1.29-1.54) <.001
Age 1.02 (1.01-1.02) <.001 1.01 (1.00-1.01) <.001
Gender
Male 1.0 (reference) 1.0 (reference)
Female 1.09 (1.04-1.15) <.01 1.09 (1.03-1.16) <.01
Race
White 1.0 (reference) 1.0 (reference)
Black 1.37 (1.22-1.54) <.001 1.09 (0.97-1.24) .16
Others 1.02 (0.95-1.10) .60 0.98 (0.90-1.06) .60
Insurance type
Medicare 1.0 (reference) 1.0 (reference)
Medicaid 1.06 (0.92-1.23) .43 1.06 (0.90-1.24) .51
Private insurance 0.73 (0.67-0.79) <.001 0.94 (0.85-1.02) .15
Other 0.95 (0.80-1.13) .57 1.08 (0.90-1.31) .40
Admission type
Elective 1.0 (reference) 1.0 (reference)
Emergency 2.78 (2.62-2.96) <.001 1.43 (1.32-1.55) <.001
Urgent 1.56 (1.43-1.71) <.001 1.14 (1.03-1.26) .01
Others 1.53 (1.11-2.13) .01 1.26 (0.88-1.79) .20
Concomitant cardiac procedures 1.76 (1.67-1.86) <.001 1.18 (1.11-1.26) <.001
Teaching hospitals 0.99 (0.94-1.04) .66 d d
Region
Rural 1.0 (reference) 1.0 (reference)
Urban 0.95 (0.88-1.02) .18 0.88 (0.81-0.96) <.01
Provider area
Midwest 1.0 (reference) 1.0 (reference)
North-East 0.91 (0.83-0.99) .04 0.83 (0.76-0.92) <.001
South 0.87 (0.81-0.93) <.001 0.92 (0.85-0.98) .02
West 0.85 (0.77-0.93) <.001 0.94 (0.85-1.03) .20
Comorbidities
Symptomatic 1.81 (1.70-1.93) <.001 1.06 (0.98-1.14) .17
Diabetes 1.33 (1.26-1.40) <.001 1.15 (1.09-1.22) <.001
Old MI 1.27 (1.18-1.36) <.001 1.06 (0.98-1.14) .16
CHF 2.19 (2.04-2.35) <.001 1.47 (1.36-1.60) <.001
Renal disease 1.99 (1.86-2.13) <.001 1.39 (1.29-1.50) <.001
COPD 1.32 (1.25-1.40) <.001 1.17 (1.10-1.25) <.001
Connective tissue disease 1.23 (1.04-1.46) .02 1.10 (0.92-1.31) .31
Peptic ulcer disease 1.41 (1.07-1.88) .02 1.07 (0.79-1.45) .66
Paraplegia/hemiplegia 3.61 (3.25-4.02) <.001 1.02 (0.89-1.17) .75
Liver disease 1.26 (0.97-1.63) .08 1.06 (0.80-1.40) .69
Peripheral arterial disease 1.21 (1.15-1.29) <.001 1.11 (1.04-1.18) <.01
Cancer 1.79 (1.54-2.08) <.001 1.52 (1.29-1.79) <.001
Discharge destination
Home 1.0 (reference) 1.0 (reference)
Home under home health care 2.10 (1.93-2.29) <.001 1.59 (1.45-1.74) <.001
(Continued on next page)
Journal of Vascular Surgery Dakour Aridi et al 9
Volume -, Number -
2005 and 2009 and between 2009 and 2011, our results significantly higher adjusted readmission costs for CAS
represent more recent real-world outcomes and include patients by around $700 (Table VI).
private insurance data in addition to Medicare data. Identifying preventable causes of readmission after ca-
The most common indications of readmission in our rotid revascularization is a complex task, especially given
cohort were stroke, cardiac, infectious, and respiratory that vascular surgery patients are usually older and have
complications, which are more likely to develop in CEA multiple comorbidities compared with patients in other
patients after hospital discharge compared with the im- surgical specialties. Readmission in this population of pa-
mediate complications of CAS.20,27,28 In the Carotid tients is thus highly related to their medical illnesses and
Revascularization Endarterectomy versus Stenting Trial comorbidities rather than to the index procedure. On the
(CREST), the majority of CAS strokes occurred on day 0 af- other hand, many readmissions in vascular surgery are
ter the procedure, whereas CEA strokes were distributed planned.25,29 A significant portion of patients in our study
evenly over 30 days postoperatively. This could explain were readmitted for nonspecific or nonrelated causes,
increased readmission after CEA due to stroke despite such as cancer/chemotherapy, abdominal surgeries,
lower overall stroke rates compared with CAS.27 A recent and nonspecific complaints, such as fatigue, fever, head-
study by our group using the American College of Sur- ache, syncope, and physical therapy (Table III).
geons National Surgical Quality Improvement Program The key to reducing preventable readmissions is better
database and another by Greenleaf et al using the Penn- perioperative planning and comorbidity management in
sylvania Health Care Cost Containment Council database high-risk patients. Older patients and those with major
showed that the risk of 30-day readmission was not pro- preoperative comorbidities (diabetes, renal disease,
cedure related but rather determined by symptomatic COPD, peripheral arterial disease, and cancer) are re-
status and comorbidities.22,24 In our analyses, readmis- ported to have increased risk of adverse events leading
sion remained significantly higher after CEA than after to higher hospital readmissions. These may require a
CAS even after adjusting for patients’ comorbidities and more intense preoperative maximization of the patient
symptomatic status. These findings add to the benefits and close postoperative follow-up.25,30 Another promi-
and cost-effectiveness of the less invasive CAS in high- nent risk factor associated with increased readmission
risk patients who are not candidates for surgery. Howev- was nonhome discharge, mainly to skilled nursing facil-
er, the lower readmission rates of CAS are opposed by ities, rehabilitation centers, and hospice care. Among pa-
the high mortality associated with these readmissions. tients who were readmitted from those facilities, 16.8%
As evident in Table VI, patients readmitted after CAS had prior CAS compared with 11.2% who were readmit-
have a longer hospital stay compared with patients read- ted from home (P < .001). Moreover, they had significant
mitted after CEA (5 days vs 4 days; P ¼ .001), higher crude comorbidities, such as paraplegia/hemiplegia (21.6% vs
mortality (6.2% vs 2.8%; P < .001), and increased respira- 1.3%), CHF (21.4% vs 13.6%), renal disease (23.3% vs
tory complications (5.0 vs 3.5%; P ¼ .03), reflecting a more 17.0%), diabetes (43.8% vs 39.1%), and dementia (2.4%
difficult rehospitalization course in the high-risk CAS vs 0.9%), and more frequent complications during their
patients compared with CEA patients. This leads to index admission, including stroke (16.0% vs 0.8%) and
10 Dakour Aridi et al Journal of Vascular Surgery
--- 2017
Table V. Perioperative outcomes of carotid endarterectomy (CEA) and carotid artery stenting (CAS) patients before and
after coarsened exact matching (CEM)
Before CEM, No. (%) After CEM, No. (%)
CEA (n ¼ 82,817) CAS (n ¼ 12,870) P value CEA (n ¼ 8966) CAS (n ¼ 8966) P value
Readmission
30 days 5338 (6.5) 780 (6.1) .1 560 (6.2) 438 (4.9) <.001
90 days 8512 (10.3) 1286 (10.0) .32 881 (9.8) 749 (8.3) <.01
In-hospital complications 300 (0.4) 72 (0.6)
Stroke 1082 (1.3) 321 (2.5) <.001 111 (1.2) 161 (1.8) <.01
Cardiac
MI 553 (0.7) 99 (0.8) .19 49 (0.5) 42 (0.5) .46
Heart failure 344 (0.4) 60 (0.5) .41 24 (0.27) 22 (0.24) .77
Arrhythmia 4202 (5.1) 907 (7.1) <.001 439 (4.9) 598 (6.7) <.001
Any cardiac 4794 (5.8) 999 (7.8) <.001 489 (5.4) 637 (7.1) <.001
Hemorrhage/shock/bleeding 6113 (7.4) 1149 (8.9) <.001 666 (7.4) 772 (8.6) <.01
Vascular or stent complication 79 (0.1) 87 (0.7) <.001 6 (0.07) 46 (0.5) <.001
Respiratory
Pneumonia 500 (0.6) 112 (0.9) <.001 49 (0.5) 37 (0.4) .20
Respiratory failure 1015 (1.2) 204 (1.6) .001 113 (1.3) 88 (1.0) .08
Any respiratory 1386 (1.7) 283 (2.2) <.001 148 (1.7) 110 (1.2) .02
Renal or urologic
Acute kidney injury 1817 (2.2) 314 (2.4) .08 164 (1.8) 130 (1.5) .05
Urinary tract infection 560 (0.7) 173 (1.3) <.001 43 (0.5) 73 (0.8) .01
Any renal 2264 (2.7) 464 (3.6) <.001 203 (2.3) 192 (2.1) .58
Infectious 167 (0.2) 37 (0.3) .05 17 (0.2) 15 (0.2) .72
Ileus 112 (0.1) 16 (0.1) .75 10 (0.1) 7 (90.08) .47
Venous thromboembolic event 58 (0.07) 22 (0.2) <.001 7 (0.08) 5 (0.06) .56
Discharge destination <.001 <.001
Home 70,447 (85.1) 10,538 (81.9) 7746 (86.4) 7948 (88.7)
Home under home health care 6655 (8.0) 824 (6.4) 677 (7.6) 433 (4.8)
General hospital 402 (0.5) 112 (0.9) 38 (0.4) 51 (0.6)
Skilled nursing facility 3046 (3.7) 580 (4.5) 288 (3.2) 244 (2.7)
Others or unknown 1964 (2.4) 744 (5.8) 216 (2.4) 282 (3.2)
MI, Myocardial infarction.
acute renal failure (10.7% vs 2.7%). This subpopulation of after including in-hospital complications, the predictabil-
patients might benefit from more discharge planning ity did not significantly improve (R2 ¼ 8.9; C statistic ¼
and improved transitional care and, in certain cases, 69.6). This is not uncommon because the creation of
deferring carotid intervention, especially for asymptom- models predicting preventable readmissions using inpa-
atic high-risk patients. Similarly, patients who experience tient hospital data and information present at discharge
stroke and renal complications during their index admis- has been shown to be challenging in medical as well as
sion have 40% and 14% increased odds of 30-day read- in surgical patients, with C statistics similar to that re-
mission (aOR, 1.4 [95% CI, 1.20-1.64] and 1.14 [95% CI, ported in our study.19,31-34 Moreover, readmissions in pa-
1.01-1.30], respectively) and should be followed up closely. tients undergoing vascular surgery are mainly driven by
Because concomitant cardiac procedures were also postoperative complications that are identified after
associated with an increase in the odds of 30-day read- discharge.34,35 Al-Damluji et al1 found that almost one-
mission (aOR, 1.18; 95% CI, 1.11-1.26; P < .001), staging of third of readmission diagnoses were potentially due to
those procedures, especially in high-risk patients, could procedural complications. Similarly, Lawson et al35 also
prove effective in reducing readmission (Table IV). identified postoperative complications occurring after
Despite the granularity of our clinical data, the model discharge as the single most predictive factor in surgical
was able to predict only around 9% of the variability in readmissions and a high-yield area for improvement of
30-day readmission (R2 ¼ 8.8; C statistic ¼ 69.5). Even patient care.
Journal of Vascular Surgery Dakour Aridi et al 11
Volume -, Number -
Readmission significantly constitutes a large burden to pruned from the matched cohort. Thus, the matched
patients and health care systems.36-38 Prior studies have cohort is likely not representative of the whole popula-
suggested several interventions that focus on minimizing tion of patients but rather provides risk-adjusted esti-
risk factors and addressing known and preventable rea- mates of treatment effects. Another important limitation
sons for readmission, such as wound infections, to is that readmissions to a different hospital, other than the
improve health care quality and resource use.26 Such in- index hospital, are not tracked in the database, which
terventions include improving outpatient management might underestimate readmission rates. We cannot
and transitional care,39,40 closer primary care follow-up provide exact estimates on the number of readmissions
after discharge,41 and avoidance of premature discharge to hospitals other than the index hospital, but this should
in high-risk patients to limit early and short readmissions. not compromise the validity of our findings because our
The majority of readmissions in our study occurred readmissions are limited to 30 days after discharge, in
within the first week of discharge (median, 6 which most patients are readmitted usually to the index
[interquartile range, 1-16] days), with some patients hav- hospital where their vascular surgeon is available, espe-
ing more than one readmission within 30 days. This cially if the readmissions are procedure related. In addi-
was also reported in other published studies.1,42 However, tion, administrative claims data may not be suitable for
the success of isolated interventions has been ques- identifying staged or planned revascularization proced-
tioned, especially that most have been studied in a retro- ures. Planned or staged readmissions are significant in
spective manner.43 Thus, prospective studies on overall discussing peripheral revascularization procedures rather
cost-effectiveness of patient-centered bundled interven- than carotid revascularizations.
tions are warranted.19 Until then, it is reasonable to recon-
sider penalizing hospitals with high risk-standardized
CONCLUSIONS
rates of readmissions.
In this study, overall 30-day readmission after carotid
Limitations. This retrospective analysis offers an over- revascularization was 6.5%. After risk adjustment, CAS
view of readmission after carotid revascularization using had lower readmission rates than CEA. However, the mor-
a large nationwide database, which can provide a plat- tality and overall costs of readmissions were higher after
form for further investigations and targets for interven- CAS. Reducing readmissions requires identifying high-
tion. The multivariate analysis is further validated using risk patients and managing their comorbidities and risk
CEM to ensure the validity and help reduce selection factors. Further prospective studies are needed to eval-
bias. However, risk adjustment and exact matching uate the cost-effectiveness of strategies aiming to reduce
cannot control for all confounding variables or eliminate readmission rates after carotid revascularization.
inherent selection bias of patients undergoing CAS vs
those undergoing CEA. Furthermore, patients in both The authors thank Nasr Ghajar, MS, for his contribution
groups that could not be matched were discarded or in reviewing the literature.
12 Dakour Aridi et al Journal of Vascular Surgery
--- 2017
34. Glebova NO, Bronsert M, Hammermeister KE, Nehler MR, 40. Rümenapf G, Geiger S, Schneider B, Amendt K, Wilhelm N,
Gibula DR, Malas MB, et al. Drivers of readmissions in Morbach S, et al. Readmissions of patients with diabetes
vascular surgery patients. J Vasc Surg 2016;64:194.e3. mellitus and foot ulcers after infra-popliteal bypass
35. Lawson EH, Hall BL, Louie R, Ettner SL, Zingmond DS, surgerydattacking the problem by an integrated case
Han L, et al. Association between occurrence of a post- management model. Vasa 2013;42:56-67.
operative complication and readmission: implications for 41. Brooke BS, Stone DH, Cronenwett JL, Nolan B,
quality improvement and cost savings. Ann Surg 2013;258: DeMartino RR, MacKenzie TA, et al. Early primary care pro-
10-8. vider follow-up and readmission after high-risk surgery.
36. Hockenberry JM, Burgess JF Jr, Glasgow J, Vaughan- JAMA Surg 2014;149:821-8.
Sarrazin M, Kaboli PJ. Cost of readmission: can the Veterans 42. Dawes AJ, Sacks GD, Russell MM, Lin AY, Maggard-
Health Administration (VHA) experience inform national Gibbons M, Winograd D, et al. Preventable readmissions to
payment policy? Med Care 2013;51:13-9. surgical services: lessons learned and targets for improve-
37. Jencks SF, Williams MV, Coleman EA. Rehospitalizations ment. J Am Coll Surg 2014;219:382-9.
among patients in the Medicare fee-for-service program. 43. Hansen LO, Young RS, Hinami K, Leung A, Williams MV. In-
N Engl J Med 2009;360:1418-28. terventions to reduce 30-day rehospitalization: a systematic
38. Orszag PR, Emanuel EJ. Health care reform and cost control. review. Ann Intern Med 2011;155:520-8.
N Engl J Med 2010;363:601-3.
39. Brooke BS, Kraiss LW, Stone DH, Nolan B, De Martino RR,
Reiber GE, et al. Improving outcomes for diabetic patients Submitted Mar 6, 2017; accepted May 5, 2017.
undergoing revascularization for critical limb ischemia: does
the quality of outpatient diabetic care matter? Ann Vasc Additional material for this article may be found online
Surg 2014;28:1719-28. at www.jvascsurg.org.
13.e1 Dakour Aridi et al Journal of Vascular Surgery
--- 2017
Supplementary Table I (online only). International Classification of Diseases, Ninth Revision diagnosis and procedure
codes used to identify readmission diagnoses and certain major adverse events
Diagnoses
Hemorrhage, shock, or bleeding
285.1 Acute posthemorrhagic anemia
99.04 Transfusion of packed cells
458.29 Other iatrogenic hypotension
998 Other complications of procedures not elsewhere classified
998.1 Hemorrhage or hematoma complicating a procedure not elsewhere classified
99.06 Transfusion of coagulation factors
99.07 Transfusion of other serum
998.01 Postoperative shock, cardiogenic
998.11 Hemorrhage complicating a procedure
998.12 Hematoma complicating a procedure
39.41 Control of hemorrhage following vascular surgery
39.98 Control of hemorrhage, not otherwise specified
41.50 Total splenectomy
Vascular or graft related
996.1 Mechanical complication of other vascular device, implant, and graft
996.6 Infection and inflammatory reaction due to internal prosthetic device implant and graft
996.62 Infection and inflammatory reaction due to other vascular device, implant, and graft
996.74 Other complications due to other vascular device, implant, and graft
38.00 Incision of vessels, unspecified
38.06 Incision of vessels, abdominal arteries
38.08 Incision of vessels, lower limb arteries
38.80 Other surgical occlusion of vessels, unspecified
38.86 Other surgical occlusion of vessels, abdominal arteries
39.49 Other revision of vascular procedure
39.5 Other repair of vessels
39.56 Repair of blood vessel with tissue patch graft
39.57 Repair of blood vessel with synthetic patch graft
39.58 Repair of blood vessel with unspecified type of patch graft
39.59 Other repair of vessel
Respiratory
Pneumonia
465.9 Acute upper respiratory infections of unspecified site
466 Acute bronchitis and bronchiolitis
481 Pneumococcal pneumonia (Streptococcus pneumoniae pneumonia)
482 Other bacterial pneumonia
482.1 Pneumonia due to Pseudomonas
482.2 Pneumonia due to Haemophilus influenzae (H. influenzae)
482.3 Pneumonia due to streptococcus
482.31 Pneumonia due to streptococcus, group A
482.32 Pneumonia due to streptococcus, group B
482.39 Pneumonia due to other streptococcus
482.4 Pneumonia due to staphylococcus
482.41 Methicillin-susceptible pneumonia due to Staphylococcus aureus
482.49 Other staphylococcus pneumonia
482.81 Pneumonia due to anaerobes
482.82 Pneumonia due to Escherichia coli (E. coli)
482.83 Pneumonia due to other gram-negative bacteria
(Continued on next page)
Journal of Vascular Surgery Dakour Aridi et al 13.e2
Volume -, Number -
Supplementary Table II (online only). Baseline charac- Supplementary Table II (online only). Continued.
teristics of the study groups after coarsened exact
No. (%) No. (%)
matching (CEM)
Charlson Comorbidity Index
After CEM
I 3904 (43.4) 3904 (43.4) 1.0
CEA CAS P value
II 2955 (32.9) 2955 (32.9)
No. of patients 8966 8966
III 1262 (14.0) 1262 (14.0)
Age, years, 72.4 (8.2) 72.4 (8.2) .63
IV 566 (6.3) 566 (6.3)
mean (SD)
V 199 (2.2) 199 (2.2)
No. (%) No. (%)
VI 110 (1.2) 110 (1.2)
Gender
CAS, Carotid artery stenting; CEA, carotid endarterectomy; CHF,
Female 3408 (37.9) 3408 (37.9) 1.0 congestive heart failure; COPD, chronic obstructive pulmonary disease;
MI, myocardial infarction; SD, standard deviation.
Male 5588 (62.1) 5588 (62.1)
Race
White 7814 (86.9) 7814 (86.9) 1.0
Black 230 (2.6) 230 (2.6)
Others 952 (10.6) 952 (10.6)
Insurance type
Medicare 7371 (82.2) 7371 (82.2) 1.0
Medicaid 157 (1.8) 157 (1.8)
Private insurance 1313 (14.7) 1313 (14.7)
Other 122 (1.4) 122 (1.4)
Admission type
Elective 7154 (79.5) 7154 (79.5) 1.0
Emergency 868 (9.7) 868 (9.7)
Urgent 954 (10.6) 954 (10.6)
Others 20 (0.22) 20 (0.22)
Concomitant cardiac 1894 (21.1) 1894 (21.1) 1.0
procedures
Teaching hospitals 5206 (57.9) 5206 (57.9) 1.0
Region
Rural 528 (5.9) 528 (5.9) 1.0
Urban 8457 (94.1) 8457 (94.1)
Provider area
Midwest 1721 (19.2) 1721 (19.2) 1.0
North-East 1365 (15.2) 1365 (15.2)
South 4846 (53.9) 4846 (53.9)
West 1053 (11.7) 1053 (11.7)
Comorbidities
Symptomatic 993 (11.1) 993 (11.1) 1.0
Diabetes 2810 (31.2) 2710 (30.1) .11
Old MI 844 (9.4) 844 (9.4) 1.0
CHF 654 (7.3) 654 (7.3) 1.0
Renal disease 751 (8.4) 751 (8.4) 1.0
COPD 1899 (21.1) 1940 (21.6) .46
Connective tissue 47 (0.5) 47 (0.5) 1.0
disease
Peptic ulcer disease 43 (0.5) 41 (0.5) .83
Paraplegia/hemiplegia 150 (1.7) 150 (1.7) 1.0
Liver disease 60 (0.67) 48 (0.53) .25
Peripheral vascular 2452 (27.3) 2452 (27.3) 1.0
disease
Cancer 96 (1.1) 123 (1.4) .07
(Continued)