Thrombosis Research 156 (2017) 5459

Contents lists available at ScienceDirect

Thrombosis Research

journal homepage: www.elsevier.com/locate/thromres

Review Article

The diagnostic management of upper extremity deep vein thrombosis: A

review of the literature
Nomie Kraaijpoel a, Nick van Es a, Ettore Porreca b, Harry R. Bller a, Marcello Di Nisio a,c,
a
Department of Vascular Medicine, Academic Medical Center, Amsterdam, The Netherlands
b
Department of Medical, Oral and Biotechnological Sciences, Gabriele D'Annunzio University, Chieti, Italy
c
Department of Medicine and Ageing Sciences, Gabriele D'Annunzio University, Chieti, Italy

a r t i c l e i n f o a b s t r a c t

Article history: Upper extremity deep vein thrombosis (UEDVT) accounts for 4% to 10% of all cases of deep vein thrombosis.
Received 28 April 2017 UEDVT may present with localized pain, erythema, and swelling of the arm, but may also be detected incidentally
Received in revised form 21 May 2017 by diagnostic imaging tests performed for other reasons. Prompt and accurate diagnosis is crucial to prevent pul-
Accepted 31 May 2017
monary embolism and long-term complications as the post-thrombotic syndrome of the arm. Unlike the diag-
Available online 01 June 2017
nostic management of deep vein thrombosis (DVT) of the lower extremities, which is well established, the
Keywords:
work-up of patients with clinically suspected UEDVT remains uncertain with limited evidence from studies of
Upper extremity deep vein thrombosis small size and poor methodological quality. Currently, only one prospective study evaluated the use of an algo-
Diagnosis rithm, similar to the one used for DVT of the lower extremities, for the diagnostic workup of clinically suspected
D-dimer UEDVT. The algorithm combined clinical probability assessment, D-dimer testing and ultrasonography and ap-
Ultrasonography peared to safely and effectively exclude UEDVT. However, before recommending its use in routine clinical prac-
Diagnostic algorithm tice, external validation of this strategy and improvements of the efciency are needed, especially in high-risk
subgroups in whom the performance of the algorithm appeared to be suboptimal, such as hospitalized or cancer
patients.
In this review, we critically assess the accuracy and efcacy of current diagnostic tools and provide clinical guid-
ance for the diagnostic management of clinically suspected UEDVT.
2017 Published by Elsevier Ltd.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
2. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3. Epidemiology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4. Clinical probability assessment and D-dimer testing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.1. Clinical probability assessment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.2. D-dimer testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5. Imaging for UEDVT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.1. Ultrasonography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.2. Magnetic resonance imaging. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
6. Diagnostic work-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
7. Recommendations and future perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Conicts of interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Author contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

Abbreviations: AUROC, area under the receiver operating characteristic curve; CI, condence interval; CVC, central venous catheter; DVT, deep vein thrombosis; PE, pulmonary
embolism; UEDVT, upper extremity deep vein thrombosis.
Corresponding author at: Department of Medicine and Ageing Sciences, Gabriele D'Annunzio University, 66100 Chieti, Italy.
E-mail address: mdinisio@unich.it (M. Di Nisio).

http://dx.doi.org/10.1016/j.thromres.2017.05.035
0049-3848/ 2017 Published by Elsevier Ltd.
 N. Kraaijpoel et al. / Thrombosis Research 156 (2017) 5459
1. Introduction
The diagnostic work-up of suspected deep vein thrombosis (DVT) of
the lower extremity is well established and includes the sequential ap-
plication of a clinical decision rule and D-dimer testing, followed by
compression ultrasonography in patients with a high clinical probability
or abnormal D-dimer levels [1]. In contrast, the management of clinical-
ly suspected upper extremity deep vein thrombosis (UEDVT) remains
unclear and largely unexplored. Patients with UEDVT should be identi-
ed promptly to prevent short-term complications, such as pulmonary
embolism (PE) and loss of venous access, as well as long-term sequelae
including the post-thrombotic syndrome. At the same time, similar to
suspected DVT of the lower extremities, the diagnosis of UEDVT is con-
rmed in only 10% to 45% of patients in whom the disease is suspected
[25]. Obviously, unnecessary diagnostic tests and anticoagulant thera-
py should be avoided as much as possible in those without the disease.
The aim of this review is to critically assess the accuracy and efcacy of
current diagnostic tools and provide clinical guidance for the diagnostic
management of suspected UEDVT.
2. Methods
MEDLINE and Embase databases as well as the conference proceed-
ings of the International Society on Thrombosis and Haemostasis and
the American Society of Hematology were searched from January 1st,
2009 up to February 16th, 2017, combining terms for upper extremity
deep vein thrombosis and diagnosis. We screened the reference list
of a systematic review on the accuracy of diagnostic tests for clinically
suspected UEDVT to identify eligible studies published until 2010 [6]. Ti-
tles, abstracts, and subsequently full text articles were screened by one
of the authors (NK) for studies evaluating at least one diagnostic test or
strategy in patients with suspected UEDVT. For each diagnostic test, we
reported the sensitivity, specicity and, where available, the area under
the receiver operating characteristic curve (AUROC) as an overall mea-
sure of discriminative performance. Failure rate was dened as the pro-
portion of patients with either symptomatic DVT, fatal PE, or non-fatal
PE during 3-month follow-up, in those in whom UEDVT was considered
excluded based on an initial negative diagnostic work-up.
3. Epidemiology
UEDVT represents 4% to 10% of all cases of DVTs, which translates to
an estimated annual incidence of 3.6 per 100,000 persons [7]. UEDVT
most often involves the axillary or subclavian veins, although more dis-
tal arm veins (i.e. radial, ulnar, or brachial) as well as the deep veins of
the neck (i.e. internal jugular or brachiocephalic) can also be affected
[3,813]. Supercial vein thrombosis affects the supercial veins and re-
quires a different diagnostic and therapeutic approach, which is beyond
the scope of this review.
Primary UEDVT accounts for approximately 20% to 50% of all
UEDVTs, and includes genuine idiopathic cases and the Paget-Schroetter
syndrome, which is also referred to as effort-related thrombosis, since it
is provoked by repetitive or strenuous arm movements causing com-
pression of the subclavian vein [7,11,12,1416]. Venous thoracic outlet
syndrome may play a substantial role in the pathogenesis of the
Paget-Schroetter syndrome [17].
Secondary UEDVT is provoked by transient or persistent risk factors.
Weak risk factors are a family history of venous thromboembolism, im-
mobilization, oral contraceptives, thrombophilia, and pacemakers [14,
1821]. Strong risk factors include the presence of a central venous
catheter (CVC), cancer, and surgery of the upper extremity.
Up to 70% of secondary UEDVT events are CVC-related [10,15]. Be-
sides local activation of the intrinsic coagulation pathway by the cathe-
ter surface, insertion-related damage of the vessel wall, infusion of
medications that trigger coagulation (e.g. chemotherapy), as well as
the type (e.g. multiple-lumen catheters) and site of insertion (e.g.
55
peripherally inserted catheters versus implanted ports, subclavian
vein versus jugular-vein insertion) of the CVC may predispose to throm-
bus formation [20,2224].
Approximately 30% to 40% of all UEDVTs occur in patients with ac-
tive cancer, who have a considerably higher risk compared to patients
without cancer (odds ratio [OR] 18; 95% condence interval [CI], 9.4 to
35) [7,15]. The risk of UEDVT in cancer patients is approximately dou-
bled by the presence of a concomitant CVC. Consequently, the presence
of a CVC in cancer patients increases the risk exponentially up to 44-fold
(OR 44; 95% CI, 24 to 75), when compared to those without cancer and a
CVC [14].
4. Clinical probability assessment and D-dimer testing
Patients with UEDVT usually present with a red, swollen, and painful
arm, and may complain of weakness, paresthesia, visible collateral
veins, low-grade fever, and heaviness. The differential diagnosis in-
cludes supercial vein thrombosis, hematomas, muscle injury, skin in-
fections, and lymphedema. The diagnosis of UEDVT on the basis of
signs and symptoms alone is unreliable because of the poor specicity
of the clinical manifestations. Imaging is therefore essential to conrm
or refute the diagnosis. To reduce the burden of imaging in all patients
with a clinical suspicion, stratication of UEDVT risk by clinical probabil-
ity assessment and D-dimer testing may be valuable to select low risk
patients in whom imaging can be withheld.
4.1. Clinical probability assessment
Clinical decision rules are widely used in patients with DVT of the
lower extremities to guide decisions about further diagnostic tests. For
example, the commonly used Wells rule classies patients with clinical-
ly suspected DVT of the lower extremities as DVT likely or DVT un-
likely, to guide decisions about performing D-dimer testing and
compression ultrasonography [1]. Constans and colleagues developed
a four-item score to assess clinical pre-test probability in patients with
suspected UEDVT (Table 1) [2]. The score was derived in a retrospective
single-center cohort of 140 hospitalized patients with suspected
UEDVT. Four risk factors, including the presence of a CVC or pacemaker
thread, localized pain, unilateral edema, and another diagnosis being at
least as plausible as UEDVT, were retained in the nal score. One point
was assigned to each of the rst three items and 1 point was subtracted
for the fourth item. Patients with a score of 1 or 0 points were classi-
ed as low clinical probability, those with 1 point as intermediate
probability, and those with 2 points or more as high probability.
The score was internally validated in 103 patients from the same hospi-
tal as the derivation cohort and externally validated in a retrospective
group of 214 in and outpatients. In the derivation and validation co-
horts, UEDVT prevalence was 36%, 46% and 30%, respectively. The prev-
alence ranged between 9% and 13% among patients with a low clinical
probability score, 20% and 38% among those with an intermediate clin-
ical probability, and 64% and 70% in patients with a high clinical
Table 1
Constans score.
Item Points
Central venous catheter or pacemaker thread +1
Localized pain +1
Unilateral edema +1
Other diagnosis at least as plausible as UEDVT 1
Classication
Three-level score [2]
Low clinical probability 0
Intermediate clinical probability 1
High clinical probability 2
Two-level score [3]
UEDVT unlikely 1
UEDVT likely 2
56 N. Kraaijpoel et al. / Thrombosis Research 156 (2017) 5459
probability. The score's discriminatory performance appeared to be con-
sistent in the derivation and validation cohorts with an AUROC ranging
from 0.68 (95% CI, 0.61 to 0.76) to 0.76 (95% CI, 0.68 to 0.83). In the ex-
ternal validation cohort the dichotomized Constans score at a positivity
threshold of 2 points had a sensitivity of 78% (95% CI, 67 to 88) and a
specicity of 64% (95% CI, 56 to 72) (Table 2). The corresponding nega-
tive predictive value of 87% is not high enough to use the score as a
stand-alone test.
4.2. D-dimer testing
Two studies have evaluated the diagnostic value of D-dimer for pa-
tients with a clinical suspicion of UEDVT. Merminod and colleagues
studied 52 consecutive inpatients and outpatients, including 23 cancer
patients (44%) and 18 (35%) with a CVC [25]. UEDVT was conrmed in
15 patients (29%) by either Doppler ultrasonography or computed to-
mography venography. Quantitative D-dimer testing with the VIDAS
D-dimer assay (bioMrieux, Marcy-l'Etoile, France) was performed in
all patients. At the conventional cut-off of 500 g/L, this test had a sen-
sitivity of 100% (95% CI, 78 to 100), but a specicity of only 14% (95%
CI, 4 to 29) (Table 2).
In a much larger prospective cohort, Sartori and colleagues evaluat-
ed 239 outpatients with suspected UEDVT of whom 39 (16%) had active
cancer and 14 (6%) carried a CVC [4]. D-dimer levels were measured
with the quantitative STA Liatest assay (Diagnostica Stago, Asnires,
France) in all patients followed by Doppler ultrasonography of the
upper extremity. If the initial ultrasonography was inconclusive, the
test was repeated after 5 to 7 days, and if the second exam was still in-
determinate, a computed tomography venography was performed.
UEDVT was conrmed in 23 patients (9.6%) at baseline and in one addi-
tional case during the 3-month follow-up period, resulting in a UEDVT
prevalence of 10%. D-dimer testing, at the conventional cut-off, had a
sensitivity of 92% (95% CI, 73 to 99) and a specicity of 60% (95% CI,
52 to 67), while overall discrimination was good (AUROC 0.81; 95% CI,
0.74 to 0.89) (Table 2). Although the specicity of the D-dimer assay
was remarkably higher in this cohort (60%) than in the study by
Merminod and colleagues (10%), it still appeared to be suboptimal to
use D-dimer as single-test approach to rule out UEDVT. Similar gures
have been observed for D-dimer testing in patients with suspected
DVT of the lower extremities with sensitivities between 78% and 97%,
and specicities between 42% and 100% [26].
5. Imaging for UEDVT
While previously considered the diagnostic gold standard, venogra-
phy is nowadays rarely performed since it is cumbersome and time con-
suming, it exposes patients to ionizing radiation, and requires the use of
intravenous contrast with the potential for renal complications and al-
lergic reactions. Due to these disadvantages, ultrasonography has re-
placed venography in clinical practice as the preferred imaging
technique [27]. UEDVT diagnosis is conrmed by the presence of
Table 2
Sensitivity, specicity and overall accuracy of diagnostic tests for suspected upper extrem-
ity deep vein thrombosis.
Diagnostic strategy Sensitivity (%) Specicity (%)
Clinical decision rule [2] 78 64
D-dimer [4,25] 92100 1460
Ultrasonography
Doppler [6] 84 94
Compression [6] 97 96
Doppler with compression [5,6] 9198 8193
Magnetic resonance imaging
Time-of-ight [29] 71 89
Gadolinium-enhanced [29] 50 80
intraluminal thrombi, venous ow abnormalities, or non-compressibil-
ity of a venous segment.
5.1. Ultrasonography
Several ultrasonography approaches have been evaluated in
suspected UEDVT, including compression ultrasonography, Doppler,
and Doppler compression ultrasonography [6]. Overall, the discrimina-
tory performance of ultrasonography is high with an accuracy that
seems to be best when compression is used alone or in combination
with Doppler. In a systematic review of eleven studies that compared
various ultrasonography methods against venography, the pooled sen-
sitivity and specicity were 84% (95% CI, 72 to 97) and 94% (95% CI, 86
to 100) for Doppler ultrasonography without compression, 97% (95%
CI, 90 to 100) and 96% (95% CI, 87 to 100) for compression ultrasonog-
raphy, and 91% (95% CI, 85 to 97) and 93% (95% CI, 80 to 100) for Dopp-
ler compression ultrasonography, respectively (Table 2). These results
should be interpreted with caution, since the included studies were
small and of poor methodological quality, limiting the validity and gen-
eralizability of the ndings. In addition, these comparisons across ultra-
sonography methods may have limited relevance in clinical practice
where both Doppler and compression are often used simultaneously
during the examination.
In a recent large, single-center, prospective cohort study, the diag-
nostic accuracy of whole-arm Doppler compression ultrasonography
was evaluated in 483 consecutive outpatients with suspected UEDVT
of whom 82 (17%) had active cancer and 66 (14%) a CVC or pacemaker
[5]. According to a standardized protocol, ultrasonography was per-
formed in all patients and repeated after one week in case of an incon-
clusive rst examination. UEDVT was conrmed in 62 (13%) patients
at baseline and in 2 (10%) of the 21 patients who underwent repeated
ultrasonography. One patient with a negative ultrasonography at base-
line, developed UEDVT during 3-month follow-up, resulting in a failure
rate of 0.3% (95% CI, 0.05 to 1.7). The overall diagnostic accuracy of ultra-
sonography was high, with a sensitivity and specicity of 98% and 81%,
respectively (Table 2). Some aspects of this study merit consideration.
The study had a single-center design, which may limit the generalizabil-
ity of the ndings to other clinical settings. In addition, ve patients
were lost to follow-up, which might have led to an underestimation of
the failure rate. However, the overall data are reassuring and, together
with previous evidence, support the safety and accuracy of ultrasonog-
raphy as rst-line imaging test in patients with clinically suspected
UEDVT. An important limitation of ultrasonography is that visualization
and compression of the subclavian and brachiocephalic veins are ham-
pered by the clavicle, which limits the accuracy of ultrasonography in
these segments. Computed tomography venography could be consid-
ered when ultrasonography remains inconclusive on serial examina-
tions [28].
5.2. Magnetic resonance imaging
Magnetic resonance imaging may be an attractive diagnostic alter-
native, since it does not expose the patient to radiations and allows for
detailed visualization of the veins, including the subclavian and
brachiocephalic segments under the clavicle. The feasibility of magnetic
resonance venography was evaluated in a single-center prospective
study that included 31 patients with suspected UEDVT in whom con-
AUROC
trast venography conrmed the presence of UEDVT in 11 (32%) [29]. A
0.76 magnetic resonance venography was planned within the next 48 h
0.81
using either a non-contrast (time-of-ight) technique or a contrast
(gadolinium)-enhanced technique. Only 21 patients (48%) were able
to undergo the exam, which was inconclusive in three due to subopti-
mal quality. Based on the 18 patients with evaluable images, the report-
ed sensitivity and specicity were 71% and 89% for time-of-ight, and
50% and 80% for gadolinium-enhanced magnetic resonance venogra-
phy, respectively (Table 2). Magnetic resonance venography is
 N. Kraaijpoel et al. / Thrombosis Research 156 (2017) 5459
expensive and time consuming and, based on the current limited evi-
dence, cannot currently be recommended for UEDVT. Magnetic reso-
nance direct thrombus imaging is being evaluated in patients with
clinically suspected recurrent DVT of the leg (https://clinicaltrials.gov;
NCT02262052) and owns a potential as a future alternative tool in the
diagnostic management of UEDVT.
6. Diagnostic work-up
In patients with suspected DVT of the lower extremities or pulmo-
nary embolism, the diagnostic approach is well established and consists
of the sequential application of a clinical decision rule, D-dimer testing,
and imaging [1]. A similar algorithm was prospectively evaluated in the
multinational ARMOUR study which assessed the safety and efciency
of the sequential use of the Constans rule, D-dimer testing, and Doppler
compression ultrasonography in a large cohort of hospitalized and am-
bulatory patients with suspected UEDVT [3]. The clinical probability of
UEDVT was assessed in all patients by the Constans score, which was di-
chotomized by combining the low- and intermediate-clinical probabili-
ty categories. UEDVT was considered unlikely in patients with a score
of 1 or less and likely in those with a score of 2 or more (Table 1).
In patients classied as UEDVT unlikely, D-dimer was measured
and, if normal, UEDVT was considered ruled out. Patients with abnormal
D-dimer levels were referred for compression ultrasonography, which
was repeated after 3 to 5 days in case of inconclusive imaging.
Patients with a UEDVT likely score underwent ultrasonography di-
rectly, and if normal, subsequent D-dimer testing was performed. Nor-
mal D-dimer levels were considered to exclude UEDVT, whereas
abnormal values or an inconclusive ultrasonography were an indication
for a repeated ultrasonography after 3 to 5 days. Venography or com-
puted tomography venography were mandatory for those with incon-
clusive serial ultrasonography. All patients in whom UEDVT was
excluded by the diagnostic algorithm had 3-month follow-up for symp-
tomatic venous thromboembolic events.
Fig. 1. Suggested diagnostic algorithm for suspected upper extremity deep vein thrombosis. *High risk patients include patients with a central venous catheter or pacemaker, active cancer,
inpatients, and patients aged older than 75 years. In case of an inconclusive exam, ultrasonography should be repeated after 3 to 5 days and if still inconclusive, a computed tomography
venography should be performed.
57
Overall, 406 patients were included of whom 137 (34%) had active
cancer and 92 (23%) carried a CVC or pacemaker. UEDVT was diagnosed
in 103 patients (prevalence 25%). The diagnostic algorithm ruled out
UEDVT based on a UEDVT unlikely Constans score and a normal D-
dimer level in 87 patients (21%) in whom anticoagulant treatment
was withheld without further imaging. None of these patients devel-
oped venous thromboembolic events during 3-month follow-up.
Among all patients in whom the strategy had excluded UEDVT, one de-
veloped UEDVT during 3-month follow-up, corresponding to a failure
rate of 0.4% (95% CI, 0 to 2.2). This upper limit of the 95% condence in-
terval was below the predened safety threshold of 3%, indicating that
this diagnostic algorithm can safely rule out UEDVT without additional
imaging. Consequently, unnecessary imaging was avoided in one fth
of patients as compared to a strategy that uses ultrasonography in all
patients without stratication by clinical probability and D-dimer [5].
As D-dimer levels naturally increase with age, the specicity of D-
dimer testing is lower in elderly patients. An age-adjusted D-dimer,
which uses a progressively higher D-dimer threshold with increasing
age in patients older than 50 years (patients age in years 10 g/L),
may safely increase the specicity of D-dimer testing. In a large study
of patients with suspected PE, the age-adjusted threshold safely in-
creased the proportion of patients in whom imaging could be withheld
when compared to the conventional threshold of 500 g/L [30]. A simi-
lar strategy was evaluated in a post-hoc analysis of the ARMOUR study,
in which the age-adjusted D-dimer threshold appeared to safely in-
crease the proportion of patients with suspected UEDVT in whom imag-
ing could be withheld from 21% to 25% (absolute difference 3.7%; 95% CI,
2.3 to 6.0) [31]. None of the patients with a UEDVT unlikely Constans
score and a D-dimer level between the conventional 500 g/L and age-
adjusted threshold were diagnosed with UEDVT during 3-month fol-
low-up, for a failure rate of 0% (95% CI, 0 to 3.6).
The performance of the ARMOUR algorithm was evaluated in several
high risk subgroups including patients with a CVC or pacemaker, active
cancer, inpatients, and patients aged older than 75 years [32]. Overall,
58 N. Kraaijpoel et al. / Thrombosis Research 156 (2017) 5459
fewer patients were classied as UEDVT unlikely and more patients
had abnormal D-dimer levels. Accordingly, the proportion of patients
in whom imaging could be withheld was lower in all high-risk sub-
groups (4.5% to 12.5%) than in the remaining patients (16.8% to
26.4%), reducing the clinical benet of the algorithm in these groups.
7. Recommendations and future perspectives
The evidence on the diagnostic management of UEDVT is scarce,
with most studies addressing this topic being of low quality and small
size. The interpretation and generalizability of the results is further
hampered by the broad variation in UEDVT prevalence, duration of fol-
low-up, reference tests used, patient populations, and clinical settings.
While the Constans clinical decision score and D-dimer testing
should not be used alone, performing ultrasonography as a rst-line
test in every patient with suspected UEDVT may also not be desirable,
since it is time consuming and may not be cost-effective. As for
suspected DVT of the lower extremities, the use of an algorithm
consisting of a clinical probability assessment and D-dimer testing can
help to select patients in whom imaging can be safely withheld with po-
tential lowering of health care costs (Fig. 1) [3]. Given the lower efcien-
cy of the algorithm in high-risk patients, such as those with active
cancer, an indwelling CVC or pacemaker, hospitalized patients, and pa-
tients older than 75 years, clinicians may consider to proceed to ultraso-
nography directly without clinical probability assessment nor D-dimer
measurement (Fig. 1). Serial ultrasonography adds complexity to the di-
agnostic work-up, however, it identies UEDVT in approximately 10% of
patients with initially inconclusive imaging [3,5]. External validation of
the ARMOUR algorithm is needed before this strategy can be adopted
in routine clinical practice. Future studies should aim at improving the
work-up in high-risk subgroups where the algorithm is less efcient
and explore the use of age-adjusted or clinical probability-adjusted D-
dimer further.
Conicts of interest
None of the authors have conicts of interest to declare.
Author contributions
Study conception and design: N. Kraaijpoel, N. van Es, E. Porreca, H.R.
Bller, M. Di Nisio.
Data acquisition: N. Kraaijpoel, M. Di Nisio.
Interpretation of the data: N. Kraaijpoel, N. van Es, E. Porreca, H.R.
Bller, M. Di Nisio.
Drafting of the manuscript: N. Kraaijpoel, N. van Es, M. Di Nisio.
Critical revision of the manuscript for important intellectual content:
N. Kraaijpoel, N. van Es, E. Porreca, H.R. Bller, M. Di Nisio.
Final approval of the manuscript: N. Kraaijpoel, N. van Es, E. Porreca,
H.R. Bller, M. Di Nisio.
References
[1] M. Di Nisio, N. van Es, H.R. Bller, Deep vein thrombosis and pulmonary embolism,
Lancet 388 (2016) 30603073.
[2] J. Constans, L.-R. Salmi, M.-A. Sevestre-Pietri, S. Perusat, M. Nguon, M. Degeilh, J.
Labarere, O. Gattolliat, C. Boulon, J.-P. Laroche, P. Le Roux, O. Pichot, I. Qur, C.
Conri, J.-L. Bosson, A clinical prediction score for upper extremity deep venous
thrombosis, Thromb. Haemost. 99 (2008) 202207.
[3] A. Kleinjan, M. Di Nisio, J. Beyer-Westendorf, G. Camporese, B. Cosmi, A. Ghirarduzzi,
P.W. Kamphuisen, H.-M. Otten, E. Porreca, A. Aggarwal, M. Brodmann, M.D.
Guglielmi, M. Iotti, K. Kaasjager, V. Kamvissi, T. Lerede, P. Marschang, K. Meijer, G.
Palareti, F.R. Rickles, M. Righini, A.W.S. Rutjes, C. Tonello, P. Verhamme, S. Werth,
S. van Wissen, H.R. Bller, Safety and feasibility of a diagnostic algorithm combining
clinical probability, D-dimer testing, and ultrasonography for suspected upper ex-
tremity deep venous thrombosis: a prospective management study, Ann. Intern.
Med. 160 (2014) 451457.
[4] M. Sartori, L. Migliaccio, E. Favaretto, M. Cini, C. Legnani, G. Palareti, B. Cosmi, D-
dimer for the diagnosis of upper extremity deep and supercial venous thrombosis,
Thromb. Res. 135 (2015) 673678.
[5] M. Sartori, L. Migliaccio, E. Favaretto, C. Brusi, E. Conti, G. Rodorigo, B. Cosmi, Whole-
arm ultrasound to rule out suspected upper-extremity deep venous thrombosis in
outpatients, JAMA Intern. Med. 175 (2015) 12261227.
[6] M. Di Nisio, G.L. Van Sluis, P.M.M. Bossuyt, H.R. Bller, E. Porreca, A.W.S. Rutjes, Ac-
curacy of diagnostic tests for clinically suspected upper extremity deep vein throm-
bosis: a systematic review, J. Thromb. Haemost. 8 (2010) 684692.
[7] N. Isma, P.J. Svensson, A. Gottster, B. Lindblad, Upper extremity deep venous
thrombosis in the population-based Malm thrombophilia study (MATS). Epidemi-
ology, risk factors, recurrence risk, and mortality, Thromb Res. 125 (2010)
e335e338.
[8] P. Prandoni, E. Bernardi, Upper extremity deep vein thrombosis, Curr. Opin. Pulm.
Med. 5 (1999) 222226.
[9] A. Hingorani, E. Ascher, E. Lorenson, P. DePippo, S. Salles-Cunha, M. Scheinman, W.
Yorkovich, J. Hanson, Upper extremity deep venous thrombosis and its impact on
morbidity and mortality rates in a hospital-based population, J. Vasc. Surg. 26
(1997) 853860.
[10] M.A. Marinella, S.K. Kathula, R.J. Markert, Spectrum of upper-extremity deep venous
thrombosis in a community teaching hospital, Heart Lung 29 (2000) 113117.
[11] B. Linnemann, F. Meister, J. Schwonberg, M. Schindewolf, D. Zgouras, E. Lindhoff-
Last, MAISTHRO registry, Hereditary and acquired thrombophilia in patients with
upper extremity deep-vein thrombosis. Results from the MAISTHRO registry,
Thromb. Haemost. 100 (2008) 440446.
[12] P. Prandoni, P. Polistena, E. Bernardi, A. Cogo, D. Casara, F. Verlato, F. Angelini, P.
Simioni, G.P. Signorini, L. Benedetti, A. Girolami, Upper-extremity deep vein throm-
bosis. Risk factors, diagnosis, and complications, Arch. Intern. Med. 157 (1997)
5762.
[13] A. Hingorani, E. Ascher, J. Hanson, M. Scheinman, W. Yorkovich, E. Lorenson, P.
DePippo, S. Salles-Cunha, Upper extremity versus lower extremity deep venous
thrombosis, Am. J. Surg. 174 (1997) 214217.
[14] J.W. Blom, C.J.M. Doggen, S. Osanto, F.R. Rosendaal, Old and new risk factors for
upper extremity deep venous thrombosis, J. Thromb. Haemost. 3 (2005)
24712478.
[15] S.M. Bleker, N. van Es, A. Kleinjan, H.R. Bller, P.W. Kamphuisen, A. Aggarwal, J.
Beyer-Westendorf, G. Camporese, B. Cosmi, T. Gary, A. Ghirarduzzi, K. Kaasjager, T.
Lerede, P. Marschang, K. Meijer, H.-M. Otten, E. Porreca, M. Righini, P. Verhamme,
S. van Wissen, M. Di Nisio, Current management strategies and long-term clinical
outcomes of upper extremity venous thrombosis, J. Thromb. Haemost. 14 (2016)
973981.
[16] B. Lindblad, L. Tengborn, D. Bergqvist, Deep vein thrombosis of the axillary-subcla-
vian veins: epidemiologic data, effects of different types of treatment and late se-
quelae, Eur. J. Vasc. Surg. 2 (1988) 161165.
[17] K.A. Illig, A.J. Doyle, A comprehensive review of Paget-Schroetter syndrome, J. Vasc.
Surg. 51 (2010) 15381547.
[18] I. Martinelli, M. Cattaneo, D. Panzeri, E. Taioli, P.M. Mannucci, Risk factors for deep
venous thrombosis of the upper extremities, Ann. Intern. Med. 126 (1997) 707711.
[19] I. Martinelli, T. Battaglioli, P. Bucciarelli, S.M. Passamonti, P.M. Mannucci, Risk factors
and recurrence rate of primary deep vein thrombosis of the upper extremities, Cir-
culation 110 (2004) 566570.
[20] R.S. Evans, J.H. Sharp, L.H. Linford, J.F. Lloyd, J.S. Tripp, J.P. Jones, S.C. Woller, S.M.
Stevens, C.G. Elliott, L.K. Weaver, Risk of symptomatic DVT associated with periph-
erally inserted central catheters, Chest 138 (2010) 803810.
[21] C.J. van Rooden, S.G. Molhoek, F.R. Rosendaal, M.J. Schalij, A.E. Meinders, M.V.
Huisman, Incidence and risk factors of early venous thrombosis associated with per-
manent pacemaker leads, J. Cardiovasc. Electrophysiol. 15 (2004) 12581262.
[22] J.-J. Parienti, N. Mongardon, B. Mgarbane, J.-P. Mira, P. Kalfon, A. Gros, S. Marqu, M.
Thuong, V. Pottier, M. Ramakers, B. Savary, A. Seguin, X. Valette, N. Terzi, B. Sauneuf,
V. Cattoir, L.A. Mermel, D. du Cheyron, 3SITES Study Group, Intravascular complica-
tions of central venous catheterization by insertion site, N. Engl. J. Med. 373 (2015)
12201229.
[23] V. Chopra, D. Ratz, L. Kuhn, T. Lopus, A. Lee, S. Krein, Peripherally inserted central
catheter-related deep vein thrombosis: contemporary patterns and predictors, J.
Thromb. Haemost. 12 (2014) 847854.
[24] G.S. Patel, K. Jain, R. Kumar, A.H. Strickland, L. Pellegrini, J. Slavotinek, M. Eaton, W.
McLeay, T. Price, M. Ly, S. Ullah, B. Koczwara, G. Kichenadasse, C.S. Karapetis, Com-
parison of peripherally inserted central venous catheters (PICC) versus subcutane-
ously implanted port-chamber catheters by complication and cost for patients
receiving chemotherapy for non-haematological malignancies, Support Care Cancer
22 (2014) 121128.
[25] T. Merminod, S. Pellicciotta, H. Bounameaux, Limited usefulness of D-dimer in
suspected deep vein thrombosis of the upper extremities, Blood Coagul. Fibrinolysis
17 (2006) 225226.
[26] M. Di Nisio, A. Squizzato, A.W.S. Rutjes, H.R. Bller, A.H. Zwinderman, P.M.M.
Bossuyt, Diagnostic accuracy of D-dimer test for exclusion of venous thromboembo-
lism: a systematic review, J. Thromb. Haemost. 5 (2007) 296304.
[27] L. Mazzolai, V. Aboyans, W. Ageno, G. Agnelli, A. Alatri, R. Bauersachs, M.P.A.
Brekelmans, H.R. Bller, A. Elias, D. Farge, S. Konstantinides, G. Palareti, P.
Prandoni, M. Righini, A. Torbicki, C. Vlachopoulos, M. Brodmann, Diagnosis and
management of acute deep vein thrombosis: a joint consensus document from
the European society of cardiology working groups of aorta and peripheral circula-
tion and pulmonary circulation and right ventricular function, Eur. Heart J. (2017).
[28] H.-C. Kim, J.W. Chung, J.H. Park, Y.H. Yin, S.H. Park, C.J. Yoon, Y.H. Choi, Role of CT ve-
nography in the diagnosis and treatment of benign thoracic central venous obstruc-
tion, Korean J. Radiol. 4 (2003) 146152.
 N. Kraaijpoel et al. / Thrombosis Research 156 (2017) 5459
[29] H.J. Baarslag, E.J.R. Van Beek, J.A. Reekers, Magnetic resonance venography in con-
secutive patients with suspected deep vein thrombosis of the upper extremity: ini-
tial experience, Acta Radiol. 45 (2004) 3843.
[30] M. Righini, J. Van Es, P.L. Den Exter, P.-M. Roy, F. Verschuren, A. Ghuysen, O.T.
Rutschmann, O. Sanchez, M. Jaffrelot, A. Trinh-Duc, C. Le Gall, F. Moustafa, A.
Principe, A.A. Van Houten, M. Ten Wolde, R.A. Douma, G. Hazelaar, P.M.G. Erkens,
K.W. Van Kralingen, M.J.J.H. Grootenboers, M.F. Durian, Y.W. Cheung, G. Meyer, H.
Bounameaux, M.V. Huisman, P.W. Kamphuisen, G. Le Gal, Age-adjusted D-dimer
cutoff levels to rule out pulmonary embolism: the ADJUST-PE study, JAMA 311
(2014) 11171124.
59
[31] N. van Es, S.M. Bleker, M. Di Nisio, A. Kleinjan, J. Beyer-Westendorf, G. Camporese, A.
Aggarwal, P. Verhamme, M. Righini, H.R. Bller, P.M. Bossuyt, Improving the diag-
nostic management of upper extremity deep vein thrombosis, J. Thromb. Haemost.
15 (2017) 6673.
[32] N. van Es, S.M. Bleker, M. Di Nisio, A. Kleinjan, J. Beyer-Westendorf, G. Camporese,
P.W. Kamphuisen, H.R. Bller, P.M. Bossuyt, ARMOUR study investigators, a clinical
decision rule and D-dimer testing to rule out upper extremity deep vein thrombosis
in high-risk patients, Thromb. Res. 148 (2016) 5962.