patients characteristics and outcome of 518 arteriovenous fistulas for hemodialysis in a sub-saharan...
TRANSCRIPT
PresentedYaounde, Cam2010;21.
PresentedSwitzerland So10, 2010. Abs
1DepartmeCameroon.
2DepartmeCameroon.
3DepartmeCameroon.
4DepartmeCameroon.
674
Patients Characteristics and Outcome of 518Arteriovenous Fistulas for Hemodialysis ina Sub-Saharan African Setting
Marcus Fokou,1 Gloria Ashuntantang,2 Abel Teyang,1 Francois Kaze,2 Alain Chichom Mefire,3
Marie Patrice Halle,4 Fru Angwafo III,1 Samuel Takongmo,5 and Wilhelm Sandmann,6,7
Yaounde, Douala, and Limbe, Cameroon; D€usseldorf and Kamp-Lintfort, Germany
Background: To present the particular aspects of arteriovenous fistula (AVF) for hemodialysisin sub-Saharan Africa in terms of patients’ characteristics, patency and complication rates, aswell as factors influencing them.Methods: From November 2002 to November 2009, 518 fistulas were constructed on adults.Demographic data, patency, and complications were analyzed. The association between age,sex, and comorbidities (HIV, hypertension, diabetes) on one hand and complications as wellas AVF patency on the other was sought.Results: Males represented 73.7% of the patient population, and the mean age of the popula-tion was 45.3 years. As far as etiologies of end-stage renal disease (ESRD) and comorbiditiesare concerned, chronic glomerulonephritis was the leading cause of ESRD (134; 25.9%), fol-lowed by hypertension (22.3%), although prevalent in 83.2% of patients, and diabetes(20.1%), although prevalent in 22.2%. No cause for the ESRD could be identified in 89 patients(17.2%). Only 20.64% had AVF as the initial vascular access. The main types of AVF con-structed were radiocephalic (68%) and brachiocephalic (24.9%). The median follow-up periodwas 275 days. The cumulative patency rate at 1 year and 2 years was 76% and 51%, respec-tively. Altogether, 188 complications occurred in 16% of the AVFs. Aneurysms, failure to mature,and thrombosis were the most frequent complications occurring in 27.65%, 14.89%, and 10.63%of cases, respectively. The management options for the complications included the creation ofa new access for 63 complications (33.51%) and nonoperative management in 44.14% of thecases. We found no adverse effect of comorbid factors like diabetes mellitus (c2 ¼ 3.58, P >0.05) and HIV-positive status (c2 ¼ 0.64, P > 0.05) on the complications rate.Conclusion: According to our patients’ characteristics, there is a possibility of constructing AVFon nearly every hemodialysis patient with a good outcome.
at the 7th Congress of the Cameroon Cardiac Society,eroon, March 18, 2010. Abstract in Cardiovasc J Afr
at the Joint Congress of German, Austrian, andcieties of Vascular Surgery, Berlin, Germany, Septembertract in Ao3.3.1.4 Gef€aßchirurgie 2010;15:379.
nt of Surgery, Yaound�e General Hospital, Yaounde,
nt of Medicine, Yaound�e General Hospital, Yaounde,
nt of Medicine, Douala General Hospital, Douala,
nt of Surgery, Limbe Regional Hospital, Limbe,
5Department of Surgery, Yaound�e University Hospital Centre,Yaound�e, Cameroon.
6Centre for Vascular Surgery and Kidney Transplantation, Univer-sity of D€usseldorf, D€usseldorf, Germany.
7Vascular Centre, Kamp-Lintfort Hospital, Kamp-Lintfort,Germany.
Correspondence to: Marcus Fokou, MD, Department of Surgery, TheYaound�e General Hospital, BP 5408 Yaound�e, Cameroun; E-mail:[email protected]
Ann Vasc Surg 2012; 26: 674–679DOI: 10.1016/j.avsg.2011.07.019� Annals of Vascular Surgery Inc.Published online: January 27, 2012
Vol. 26, No. 5, July 2012 Aspects of AVFs for hemodialysis in Sub-Saharan Africa 675
INTRODUCTION
Since the advent of hemodialysis in 1944, and the
subsequent use of arteriovenous fistulae (AVFs) as
a long-term vascular access,1 there has been a drastic
increase in both the availability of hemodialysis and
long-term survival of patients with chronic renal
failure. This has resulted in permanent vascular
access procedures in dialysis (AVFs, prosthetic arte-
riovenous grafts, and autologous vein grafts)
becoming the most common operations performed
by vascular surgeons. Because of data illustrating
superiority of AVF in terms of patency rates, lower
complication rate, and lower costs, it has been rec-
ommended in countries such as United States that
AVF should be constructed in at least 50% of
patients on chronic hemodialysis.2 This goal has
even been raised to 65% in recent guidelines.3
In developing countries and particularly in sub-
Saharan Africa, there are very few reports concern-
ing vascular access surgery4,5 compared with North
Africa6,7, as hemodialysis is not yet widespread in
these areas. In Cameroon, since the year 2000,
hemodialysis is accessible to many. Therefore,
construction and maintenance of vascular accesses
has become a challenge.
In this report, we present the particularities of AVF
for hemodialysis in our setting. Specifically, we
sought to audit the patients’ characteristics, the
complications, patency rates, and todetermine factors
that affect these outcomes. To the best of our knowl-
edge, this is the largest published report concerning
vascular access surgery in sub-Saharan Africa.
METHODS
We retrospectively evaluated upper-limb AVFs per-
formed between November 2002 and 2009 at the
Yaound�e General Hospitalda university teaching
hospital with most modern facilities for vascular
surgery and hemodialysisdin Cameroon. The cases
included in the study were consecutive adult
patients with end-stage renal disease (ESRD)
referred for the creation of AVF (8 patients aged
<16 years were not considered for this study).
Primary (firstly created) and secondary (all subse-
quently constructed) AVFs were included for evalu-
ation. Outpatient and inpatient hospital records
were reviewed. All the patients underwent preoper-
ative general and local examination according to the
recommended reporting standards. The general
evaluation was aimed at identifying comorbid
conditions, such as heart failure, peripheral arterial
disease, stroke, diabetes mellitus, and high blood
pressure; medical history; and vascular access
history. The local evaluation was focused on the
potential access site. It included the following steps:
1) brachial pressures; 2) examination of the skin to
exclude lesions or scars on the site; 3) vein examina-
tion using a tourniquet or a blood pressure cuff to
ascertain the distensibility, continuity, and
compressibility; 4) palpation and Doppler explora-
tion of the radial, brachial, and ulna arteries. At
the end of this procedure, a decision on the
following was made: 1) the patient’s suitability for
surgery; 2) type of AVF to be performed: radioce-
phalic (RCAVF), brachiocephalic (BCAVF), brachio-
basilic, or cubitobasilic; and 3) the need for
a treatment (treatment of heart failure, sepsis,
severe anemia, or skin infection) before surgery. A
preoperative ultrasonography was mandatory only
in obese patients or in cases of a secondary fistula
with a poor venous visibility.
In our practice, we start first by placing a wrist
RCAVF on the nondominant arm, if anatomically
favorable. From there, we moved to a BCAVF. Our
next choices are a RCAVF and then a BCAVF on
the dominant arm. The brachiobasilic AVF and the
other possibilities are realized as last alternative.
All the operations were performed by the two
vascular surgeons of the team either as an ambula-
tory or a 1-day hospital stay procedure. Local anes-
thesia with 1% lidocaine was used for every patient.
Operative details, such as the quality of veins and
the perioperative complications, were noted by the
surgeon. The thrill and a hand-held Doppler were
used for perioperative quality control. The postoper-
ative management included analgesics and dressing
every 4 days. Patients were advised to avoid exerting
pressure or any compression on the upper limb. No
venous punctures or blood pressure measurements
with inflatable cuffs were permitted on the upper
limb concerned.
Follow-up visits with the vascular surgeon were
scheduled (two times during the first week and
then weekly during the first month postopera-
tively). Dressings, suture removal, and assessment
of fistula patency by physical examination were
done during these visits. The fistulas were allowed
to mature for a minimum of 4 weeks. If maturation
was not achieved 3 months after surgery, this was
considered as a failure to mature. The decision on
the first puncture of the access was jointly taken
by the attending surgeon and the nephrologists.
Postoperative complications were followed up by
the surgeon and the nephrologists. Functionality
was determined by the ability to access the fistula
for hemodialysis. Complications were graded
according to the time of occurrence as immediate
(within 48 hours after the surgery), early (before
676 Fokou et al. Annals of Vascular Surgery
first puncture or within the first 30 days after
surgery), and late (after 30 days). The seriousness
of the complications was also evaluated. Complica-
tions were considered to be severe when they led
to the loss of the access, to be moderate when
surgical or endovascular intervention was required
to save the fistula, and to be mild when no surgical
management was necessary.
The end points were the death of the patient, final
fistula failure, or failure of patient to turn up for
follow-ups. Statistical analysis was done using the
Epi-info 6 (Redmond, WA). Categorical data were
compared using c2 test, nominal data were
compared using Student t test, and the statistical
significance was set at a P < 0.05. Patency rate was
determined by KaplaneMeier curve.
RESULTS
Sociodemographic Data
During the 7-year study period, 518 upper-arm
vascular accesses were created on 492 limbs on
478 patients. Of these, 382weremales, representing
73.7% of the population. Two patients who relo-
cated overseas for kidney grafting were not included
in this analysis. The mean age of the patients at time
of surgery was 45.33 (range: 17e74) years.
As far as etiologies of ESRD and comorbidities are
concerned, chronic glomerulonephritis was the
leading cause of ESRD (134; 25.9%), followed by
hypertension (22.3%), although prevalent in
83.2% of patients, diabetes (20.1%), although prev-
alent in 22.2%, and HIV (5%), although prevalent
in 51 (9.84) of patients. No cause for the ESRD could
be identified in 89 patients (17.2%).
A history of previous permanent access (AVF,
graft) was found in 6.7% of the patients at presenta-
tion. At the time of AVF creation, 411 of patients
(79.34%) were undergoing dialysis through
a temporary venous catheter. In cases of subclavian
catheter, they were located on the side opposite to
the limb of the AVF in 70% of the cases. The dura-
tion of catheter carriage ranged from 20 to 214
days, with a mean of 79.83 days.
A corrective treatment was necessary in 6% of
the patients before surgery. This included mainly
blood transfusions and medical treatment of heart
failure. A preoperative vascular ultrasonography
was necessary to trace the veins in 2% of patients
(all of whom were obese).
Operative Data
Only native AVFs were constructed. No grafts were
placed. The distribution of AVFs showed 357
RCAVFs (68.91%), 129 BCAVFs (24.9%), and
approximately 6% for the other anatomical types
(Table I). Perioperative quality control demon-
strated a 97.8% success rate on first attempt. Perio-
perative exploration showed that 8% of veins were
not very suitable. In these cases, a dilatation with
heparinized saline as well as with dilators and selec-
tive obliteration of venous side branches were
done. However, in 3%, another vein segment was
chosen. Approximately 86% of AVFs were created
on the nondominant arm. Among those located
on the dominant arm, 5% were firstly constructed
AVFs.
Follow-up
The median follow-up period was 275 days, ranging
from 6 months to 7 years. Outcomes were catego-
rized as functioning, death, and complicated, as
defined in our Methods section.
Arteriovenous fistula patency rate was 76% and
51% at 1 and 2 years, respectively. Figure 1 repre-
sents the 3-year survival curve.
The time to first cannulation varied from 17 to
180 days, with a median of 31 ± 15 days.
There was no association between sociodemo-
graphic factors, comorbidities studied, type of AVF,
time to first cannulation, and catheter carriage on
the patency of AVF. HIV status did not also have
an adverse effect on the patency. On the contrary,
most of patients died with functioning fistulas.
During the follow-up period, among the AVFs
successfully constructed, 16% experienced one or
more complications. The total number of complica-
tions was 188, which are listed in Table II. Primary
failure (failure to mature) and aneurysmal forma-
tion were the most common complications. The
complication rate was not statistically significant
among the different types of AVF.
No adverse effect of the following patients’ char-
acteristics on the frequency of complications was
found: age, gender, time to first cannulation, and
length of catheter carriage. Furthermore, 46 of the
188 complications occurred in diabetic patients
(c2¼ 0.88, P> 0.05) and 15 in HIV-positive patients
(c2 ¼ 1.16, P > 0.05).
Complications were classified as immediate in 24
(12.82%), early in 44 (23.40%), and late in 74.10%
of cases. Failure to mature, hemorrhage, and aneu-
rysms were themost common complications seen in
the early, immediate, and late groups, respectively.
Severe complications were encountered in
35.89% of the cases. These weremainly thrombosis,
severe andmultifocal stenosis, multifocal or infected
aneurysms, and pseudoaneurysms. Minor and
Table I. Types of AVF realized
Variable Number %
RCAVF 357 68.91
BCAVF 129 24.90
BBAVF 26 5.01
CBAVF 6 1.15
AVF, arteriovenous fistula; BB, brachiobasilic; CB, cubitobasilic;
RC, radiocephalic; BC, brachiocephalic.
Fig. 1. KaplaneMeier patency curve of the arteriove-
nous fistula.
Table II. Different types of complications
Complication Number %
Aneurysm 52 27.65
Failure to mature 2 14.89
Hematoma/hemorrhage 24 12.27
Stenosis 20 10.63
Thrombosis 20 10.63
Ischemia 12 6.38
Collateral veins dilation 12 6.38
Infection 8 4.25
Cardiac insufficiency 8 4.25
Major edema 4 2.13
Total 188
Vol. 26, No. 5, July 2012 Aspects of AVFs for hemodialysis in Sub-Saharan Africa 677
moderate complications occurred in 43.58% and
20.5% of the cases, respectively.
The management options for our complications
were mainly conservative. The creation of a new
AVF as a definitive therapeutic decision was done
in 33.15% of cases.
DISCUSSION
The proportion of patients commencing dialysis
with an AVF was only 20.65%, whereas 79.35%
initiated dialysis with temporary central venous
catheter. The late presentation of patients with
ESRD for specialized care (owing to late referral
from nonspecialized centers, geographical distance,
ignorance, and low income) has been described
earlier in our setting.8 This late presentation at end
stage with uremic complications requires emer-
gency renal replacement therapy.8,9 Therefore,
central venous catheters are inevitable. In Nigeria,
Nwankwo et al. reported the use of catheters for
vascular access in most (91%) of the patients in
a cohort of 179 patients over a 5-year period.4 A
proportion of 92.8% and 86.3% catheter use was
also described in Ivory Coast10 andMorocco, respec-
tively.6 It is also interesting to mention that in most
sub-Saharan African countries, catheters will
constitute the only access modality owing to the
lack of vascular surgeon. On the contrary, in
western countries, there is a large variation in the
proportion of patients who commence hemodialysis
through an AVF versus catheters in favor of AVF in
Europe and catheter in the United States without
a sound scientific explanation.11e14
If we consider the goal of 65% native AVF for
patients entering into a hemodialysis program set
by the U.S. National Kidney Foundation guide-
lines,3 we still have room for improvement. This
improvement is, however, dependent on several
factors, including a sound prevention program for
chronic kidney disease, which will enable early
detection and appropriate care of persons with this
condition to avoid late diagnosis, late presentation,
and unprogrammed dialysis.
The duration of catheter carriage in our patients
at the time of surgery was 79.83 days, meaning
that the time from dialysis onset to fistula creation
is lengthy. The explanations for this lie in the long
period of nephrological care necessary to improve
on the state of these patients who present late.
Furthermore, most patients pay for the procedure
out of their pocket owing to the absence of a health
insurance policy. Finally, dialysis subsidies in the
country do not cover vascular access. Identical prob-
lems were also raised in other dialysis facilities in
sub-Saharan Africa.4e7,10
We succeeded in creating a fistula in nearly every
patient on chronic hemodialysis. Currently, the
prevalence of AVF versus graft in dialysis patients
is variable among countries. Our figures are largely
above the ultimate goal of ‘‘prevalent functional
AVF placement rate greater than 65% of patients’’
set up in guideline number 8.1.3 Apart from the
difficulties in affording prosthetic materials,
although they are available, and the high infectious
risk, the difference in patient characteristics could
explain the low need of prosthetic grafts in our
678 Fokou et al. Annals of Vascular Surgery
population. While the mean age of chronic dialysis
patients in Europe and United States is approxi-
mately 70 years, the mean age of our population
was <50 years. More so, although diabetes mellitus
with its inherent vascular complications accounts
for a major part of patients on renal replacement
therapy in the developed world, we still have
chronic glomerulonephritis prevalent in our popu-
lation.9 In western countries, because enough
health care is provided, there is also an aggressive
approach to indiscriminately draw blood and use
intravenous lines. These aforementioned factors
greatly affect the vascular integrity of patients and
make the creation of native fistulas difficult. The
samemay apply to other sub-Saharan African coun-
tries.4e7,10 Another specific hypothesis is that our
dialysis program was for <10 years, and within
this period, the vein sites for an AVF were not yet
exhausted for a patient.
The distribution of the type of AVF showed a great
proportion of RCAVF, which is the generally recom-
mended AVF and the most constructed in the
majority of current surgical works.11,14e16 On the
contrary, some authors, such as Mc Lafferty
et al.17 and Martin et al.,18 reported a high propor-
tion of BCAVF without a particular explanation.
The mean time to first cannulation in our study
group was 40 days. Actually, the best time for the
maturation of a fistula is not clearly defined and
differs among dialysis teams, from <28 days to 96
days.10 Hemodynamic studies addressing this issue,
such as those conducted by Le Blanc et al.14 and
other authors,18 led to the conclusion that after 4
weeks, a BCAVF was hemodynamically ready to
maintain a sustainable dialysis. In our experience,
4 weeks after the operation, most of the fistulas
are already well ‘‘arterialized.’’
The cumulative patency rate of 76% at 1 year and
51% at 2 years is not different from the figures in
many countries. A patency of 3 years is a reasonable
goal for a fistula.3 In our team, our aim was to
achieve an 80% patency at 1 year and >60%
patency at 2 years.
As far as complications are concerned, our overall
complication rate of 16%, including the frequency
and management of the different types, is difficult
to compare, as the issue is not addressed by most
of the reports dealing with vascular access.19,20 In
the U.S. National Kidney Foundation guidelines,
for example, no threshold was set that can be used
as a quality indicator, as setting a goal may
discourage native AVF construction in patients
with complex vascular anatomy.2,3 Primary
failure continues to be a major obstacle, with stud-
ies reporting values ranging from 25% to
43%.11,12,17,21 With regard to the different types of
complications, the paucity of the surgical literature
makes discussion difficult. Only a few reports have
addressed the management of aneurysms22,23 or
some specific complications, such as ischemia.24,25
Unlike some reports, we did not find a poorer fistula
patency rate or a higher complication rate in dia-
betic and elderly patients, as has been reported in
some studies,15,26,27 perhaps because of the rela-
tively young age of our patients and the relatively
lowprevalence of diabetes. EvenwithHIV, a relation
not yet analyzed by current surgical literature, an
adverse effect on the fistula outcome was not
noticed.
CONCLUSION
In this study, we reported our 7-year experience in
vascular access surgery in a setting where access to
prosthetic graft is seldom, and showed that in nearly
every patient with the same characteristics, it is
possible to construct an AVF in the course of dialysis
with a good outcome. There were no effects of HIV
status on the outcome of the fistula. Although not
specifically analyzed, we think there is a positive
effect of low income and insurance status in the
effort to always try to construct a native AVFda
policy that can be implemented in patients from
same socioeconomic conditions.
REFERENCES
1. Brescia MJ, Cimino EJ, Appel K, Hurwich BJ. Chronic
hemodialysis using venipuncture and a surgically created
arteriovenous fistula. N Engl J Med 1966;275:1089e92.
2. National Kidney Foundation, K/DOQI. Clinical practice
guidelines for vascular access 2000. Am J Kidney Dis
2001;37(Suppl. 1):137e81.
3. National Kidney Foundation, K/DOQI. Clinical practice
guidelines and clinical practice recommendations for 2006
updates: hemodialysis adequacy, peritonealdialysis adequacy,
vascular access. Am J Kidney Dis 2006;48(Suppl. 1):S1e322.
4. Nwankwo EA, Wudiri WW, Bassi A. Practice pattern of
hemodialysis vascular access in Maiduguri, Nigeria. Int J
Artif Organs 2006;29:956e60.
5. Yiltok SJ, Orkar KS, Agaba EI, et al. Arteriovenous fistula for
patients on long-term haemodialysis in Jos, Nigeria. Niger
Postgrad Med J 2005;12:6e9.6. Medkouri G, Aghai R, Anabi A, et al. Analysis of vascular
access in hemodialysis patients: a report from a dialysis
unit in Casablanca. Saudi J Kidney Dis Transpl 2006;17:
516e20.
7. Afiti A, Refaal H, Wahba AM, et al. Hemodialysis vascular
access among chronic renal failure patients in Egypt. J
Vasc Access 2002;3:164e8.
8. Halle MP, Kengne AP, Ashuntantang G. Referral of patients
with kidney impairment for specialist care in a developing
country of sub-Saharan Africa. Ren Fail 2009;31:341e8.
Vol. 26, No. 5, July 2012 Aspects of AVFs for hemodialysis in Sub-Saharan Africa 679
9. Roubicek C, Brunet P, Huiart L, et al. Timing of nephrology
referral: influence on mortality and morbidity. Am J Kidney
Dis 2000;36:35e41.
10. Ackoundou-Nguessan C, Gnionsahe A, Guie M, et al. High
failure rate for first arteriovenous fistula for patients starting
hemodialysis treatment: a report from Ivory Coast. Saudi J
Kidney Dis Transpl 2008;19:384e8.
11. Rayner HC, Pisoni RL, Gillespie BW, et al. Creation, cannu-
lation and survival of ant�erioveinous fistulae: data from the
Dialysis Outcomes and Practice Patterns Study (DOPPS).
Kidney Int 2003;63:323e30.
12. Patel ST, Hughes J, Mills JL Sr. Failure of arteriovenous
fistula maturation: an unintended consequence of
exceeding dialysis outcome quality initiative guidelines for
hemodialysis access. J Vasc Surg 2003;38:439e45.
13. Fan SL, Naqvi R, Ahmad R, et al. Haemodialysis access:
a single centre U K experience. J Vasc Access 2002;3:
101e7.
14. Le Blanc M, Saint Sauveur E, Pichette V. Native arteriove-
nous fistula for hemodialysis: what to expect early after
creation? J Vasc Access 2003;4:39e44.
15. Lazarides MK, Georgiadis GS, Antoniou GA, Staramos DN. A
meta-analysis of dialysis access outcome in elderly patients.
J Vasc Surg 2007;45:420e6.
16. Biuckians A, Scott EC, Meir GH, et al. The natural history of
autologous fistulas as first-time dialysis access in the KDOQI
era. J Vasc Surg 2008;47:415e21.17. McLafferty RB, Pryor RW 3rd, Johnson CM, et al. Outcome
of a comprehensive follow-up program to enhance matura-
tion of autogenous arteriovenous hemodialysis access. J
Vasc Surg 2007;45:981e5.
18. Martin AG, Grasty M, Lear PA. Hemodynamics of brachial
arterioveinous fistula development. J Vasc Access 2000;1:
54e9.
19. Hill SL, Donato AT. Complications of dialysis access: a six
year study. Am J Surg 1991;162:265e7.20. Hossny A. Brachiobasilic arteriovenous fistula: different
surgical techniques and their effects on fistula patency and
dialysis-related complications. J Vasc Surg 2003;37:821e6.
21. Plantinga LC, Jaar BG, Astor B, et al. Association of clinic
vascular access monitoring practices with clinical outcomes in
hemodialysis patients. Nephron Clin Pract 2006;104:151e5.
22. Allaria PM, Costantinie E, Lucatello A, et al. Aneurysm of
arteriovenous fistula in uremic patient: is endograft a viable
therapeutic approach? J Vasc Access 2002;3:85e8.
23. Zheng YH, Liu CW, Guan H, et al. Surgical management of
pseudoaneurysm complicating arteriovenous fistula for
hemodialysis. Chin Med Sci J 2007;22:196e8.
24. Schanzer H, Skladany M, Haimov M. Treatment of
angioaccess-induced ischemia by revascularisation. J Vasc
Surg 1992;16:861e4.25. Knox RC, Berman SS, Hughes JD, et al. Distal
Revascularisation-interval ligation: a durable and effective
treatment for ischemic steal syndrome after hemodialysis
access. J Vasc Surg 2002;36:250e6.26. Anel RL, Yevzlin AS, Ivanovich P. Vascular access and
patient outcomes in hemodialysis: questions answered in
recent literature. Artif Organs 2003;27:237e41.27. Huijbregts HJ, Bots ML, Moll FL, Blankestijn PJ. Hospital
specific aspects predominantly determine primary failure
of hemodialysis arteriovenous fistulas. J Vasc Surg
2007;45:962e7.