Sensitivity and Specificity of Ultrasonography in the Diagnosis of Upper Extremity Deep Vein Thrombosis A Systematic Review

REVIEW ARTICLE
Sensitivity and Specificity of Ultrasonography
in the Diagnosis of Upper Extremity
Deep Vein Thrombosis
A Systematic Review
Bisher O. Mustafa, MD; Suman W. Rathbun, MD; Thomas L. Whitsett, MD; Gary E. Raskob, PhD
Objectives: To determine the sensitivity and specific-
Data on sensitivity and specificity and the associated 95%
ity of ultrasonography in the diagnosis of upper extrem-
confidence intervals were recorded when available.
ity deep vein thrombosis and to determine the safety of
withholding anticoagulant therapy in patients with nega-
Data Synthesis: Only one study met all of the pre-
tive ultrasonographic results.
defined criteria for adequately evaluating sensitivity and
specificity. The sensitivity of duplex ultrasonography
Data Sources: The MEDLINE database was searched
ranged from 56% to 100%, and the specificity ranged from
for literature published from January 1, 1980, to Decem-
94% to 100%. No study evaluated the safety of withhold-
ber 31, 2000, that evaluated ultrasonography for the di-
ing anticoagulant therapy without additional testing in
agnosis of upper extremity deep vein thrombosis. Bibli-
patients with negative ultrasonographic results.
ographies of the retrieved articles were cross-checked to
identify additional studies.
Conclusion: The safety of withholding anticoagulant
treatment in a patient with suspected upper extremity deep
Study Selection: All prospective English-language stud-
vein thrombosis and negative ultrasonographic results
ies were selected. Retrospective studies, review articles,
is uncertain.
and case reports were excluded.
Data Extraction: Two of us (B.O.M. and S.W.R.) used
predefined criteria to independently assess each study.
Arch Intern Med. 2002;162:401-404
UPPEREXTREMITYdeepvein sonographyforthediagnosisofupper
thrombosis is an increas-
extremity deep vein thrombosis. For ex-
ingly common clinical
ample, the ability to image and compress the
problem.1 It may cause
middle of the subclavian vein is hindered
pulmonary embolisms,
by the overlying segment of the clavicle.
including fatal embolisms.2-6 Therefore, ac-
Moreover, which vessels are imaged and the
curate diagnosis is essential.
diagnostic criteria for the presence or ab-
The clinical diagnosis of upper extrem-
sence of deep vein thrombosis vary. Most
ity deep vein thrombosis is nonspecific.6 Its
commonly, the subclavian and axillary veins
prevalence is less than 50% among symp-
are imaged, but ultrasonographic testing for
tomatic patients,6 necessitating objective
upper extremity deep vein thrombosis may
testing to confirm or exclude the diagno-
also include imaging of the internal jugu-
sis. While venography remains the diag-
lar, innominate, brachial, and basilic veins.7,8
nostic reference standard, it is invasive, has
Real-time ultrasonography is used to as-
an associated risk of thrombophlebitis, and
sess the presence or absence of vein com-
may be unavailable or impractical. Ultra-
pressibility and the echogenicity within the
From the Departments of
sonography is the most frequently used ob-
vein lumen. Doppler ultrasonography evalu-
Medicine (Drs Mustafa,
jective test. However, in contrast to pa-
ates the characteristics of venous flow, in-
Rathbun, Whitsett, and Raskob)
tients with suspected deep vein thrombosis
cluding phasicity, pulsatility, and varia-
and Biostatistics and
of the legs, the validity of ultrasonographic
tion with physiologic maneuvers. Duplex
Epidemiology (Dr Raskob),
testing for suspected upper extremity deep
ultrasonography uses combined real-time
The University of Oklahoma
vein thrombosis is uncertain. Anatomical
imaging and Doppler ultrasonographic as-
Health Sciences Center, and
Veterans Administration
differences between the upper and lower ex-
sessment. Color flow Doppler imaging en-
Medical Center,
tremity deep venous system may influence
ables the assessment of the presence and di-
Oklahoma City.
the performance characteristics of ultra-
rection of venous flow.8
(REPRINTED) ARCH INTERN MED/ VOL 162, FEB 25, 2002
WWW.ARCHINTERNMED.COM
401
on April 29, 2011
www.archinternmed.com
Downloaded from
©2002 American Medical Association. All rights reserved.

---

Because of the uncertain valid-
thrombosis and negative results by
vein thrombosis was identified. The
ity of ultrasonography for the diag-
ultrasonography.
Medical Subject Headings’ terms up-
nosis of upper extremity deep vein
per extremity and thrombosis were
thrombosis, we conducted a system-
MATERIALS AND METHODS
used in separate searches, and stud-
atic review of the literature. Our re-
ies found during each search were
view had 2 objectives: (1) to deter-
LITERATURE SEARCH
combined. Limits were set for hu-
mine the sensitivity and specificity
AND DATA SOURCES
man only and English language only.
of ultrasonography in the diagno-
We supplemented this reference list
sis of upper extremity deep vein
The MEDLINE database was searched
by cross-checking bibliographies of
thrombosis and (2) to determine the
for literature published from Janu-
retrieved articles to identify addi-
safety of withholding anticoagu-
ary 1, 1980, through December 31,
tional studies.
lant therapy without further objec-
2000. Before 1980, no study that used
tive testing in patients with sus-
noninvasive vascular imaging for the
STUDY SELECTION
pected upper extremity deep vein
diagnosis of upper extremity deep
AND DATA EXTRACTION
Before performing the literature re-
Table 1. Criteria Used to Review the Studies
view, we defined criteria for the in-
clusion of studies and for assessing
Criteria
the validity of these studies.9,10 We
Questions Used to Assess Studies That Evaluated the Sensitivity and Specificity
decided a priori to include all pro-
of Ultrasonography in the Diagnosis of Upper Extremity Deep Vein Thrombosis
spective studies identified by the
Does the study include consecutive patients with suspected upper extremity deep vein thrombosis?
literature search, including ab-
What methods of ultrasonography were used and which vessels were imaged?
stracts, that included ultrasono-
Do all patients undergo ultrasonography and the reference standard, venography?
graphic assessment of the upper ex-
Was the venogram interpreted without knowledge of the ultrasonographic results and vice versa?
Does the study include a broad spectrum of patients (patients with and without upper extremity
tremity veins by one or more of the
deep vein thrombosis) and a broad spectrum of patient characteristics (eg, age, sex, and
following techniques: real-time ul-
comorbid conditions)?
trasonographic imaging, standard
What are the reported sensitivity and specificity?
Doppler evaluation, or color flow
What are the 95% confidence intervals for the sensitivity and specificity?
Doppler imaging. Retrospective
Questions Used to Assess Studies That Evaluated the Safety of Withholding
studies, review articles, and case re-
Anticoagulant Therapy Based on Negative Ultrasonographic Results
ports were excluded. Two of us
Did all patients with negative ultrasonographic results have their anticoagulation therapy withdrawn
(B.O.M. and S.W.R.) reviewed each
or withheld?
article or abstract independently us-
What is the follow-up period (0,
3, 3-6, or
6 mo)?
ing the criteria given in Table 1.
What were the outcomes of the follow-up (death, a fatal or symptomatic pulmonary embolism, or
symptomatic upper extremity deep vein thrombosis)?
These criteria were established a
priori before the articles were re-
Table 2. Prospective Studies Evaluating the Use of Ultrasonography in the Diagnosis of Suspected Upper Extremity Deep Vein
Patients
Those Who
Included
Patients With
Underwent
Consecutive
Ultrasonographic
Blind
Broad
Catheter-Related
Source, y
Total
Venography
Patients
Technique
Veins Imaged
Interpretation
Spectrum
Thrombosis
Weissleder et al,15
2
2
No
Duplex
Intrajugular and subcla-
NI
No
0
1987
vian
Falk and Smith,16
22
14
No
Duplex
Intrajugular, innominate,
Yes
NI
9
1987
subclavian, and axillary
Hubsch et al,17
26
9
NI
Duplex plus color
Intrajugular and subcla-
NI
NI
26
1988
flow
vian
Haire et al,18 1991
43
43
No
Duplex
Subclavian
Yes
NI
39
Baxter et al,19
19
19
No
Duplex
Intrajugular, innominate,
Yes
No
14
1991
subclavian, and axillary
Duplex plus color
flow
Prandoni et al,6
58
58
Yes
Duplex
Intrajugular, innominate,
Yes
Yes
8
1997
subclavian, axillary, and
brachial
Duplex plus color
flow
Doppler alone
*NI indicates data not indicated; NA, data not applicable.
†Data are given as percentage (number/total) [95% confidence interval].
(REPRINTED) ARCH INTERN MED/ VOL 162, FEB 25, 2002
WWW.ARCHINTERNMED.COM
402
on April 29, 2011
www.archinternmed.com
Downloaded from
©2002 American Medical Association. All rights reserved.

---

viewed according to established
tiveseriesofallpatientswithsuspected
First, only 1 study6 met all of the
methodologic standards for the
upper extremity deep vein thrombo-
predefined methodologic criteria11 for
evaluation of diagnostic tests.11 A
sis was examined. All 6 studies6,15-19
adequately evaluating the sensitivity
third independent reviewer (T.L.W.)
included information on the vessels
and specificity of a diagnostic test.
adjudicated disagreements.
that were imaged and the ultrasono-
However, even in this study, the 95%
graphic methods used. In 4 stud-
confidence intervals for sensitivity and
ies,6,16,18,19 the results of ultrasonogra-
specificity were broad, indicating that
RESULTS
phy and venography were interpreted
the true sensitivity of duplex ultra-
LITERATURE SEARCH
independently.Only1study6included
sonography could be as low as 82%
AND DATA SOURCES
enoughinformationtoassessthespec-
and that the true specificity could be
trum of patients examined. All stud-
as low as 69% (Table 2). The remain-
The literature search identified 18 ar-
ies reported the number of patients in
ing studies15-19 lacked details regard-
ticles, 6 case reports, and 1 abstract.
whomthrombosiswasassociatedwith
ing the characteristics of the patients
Of the 18 articles, 12 were excluded:
a venous catheter. The sample sizes
and an independent interpretation of
9 were retrospective studies (a list is
of these studies ranged from 2 to 58
ultrasonography and venography.
available from the authors), 1 in-
patients (Table 2).
Second, there was a clinically im-
cluded only patients with neck swell-
portant variation (56%-100%) in the
ing or palpable masses and jugular
SENSITIVITY
sensitivity of ultrasonography for up-
vein thrombosis,12 1 examined only
AND SPECIFICITY
per extremity deep vein thrombosis
asymptomatic patients,13 and 1 did
(Table 2). This variation is likely due
not use ultrasonography in the ini-
Five studies reported the sensitiv-
to differences in patient selection and
tial diagnosis.14 The 6 case reports
ity and 3 studies reported the speci-
the ultrasonographic technique used
were excluded (a list is available from
ficity or provided enough informa-
and the relatively few patients exam-
the authors). The one abstract was ex-
tion to calculate these values. The
ined in all studies. The main limita-
cluded because of insufficient infor-
sensitivity of ultrasonography ranged
tion of real-time imaging of the up-
mation to assess study validity. Thus,
from 56% to 100%, and the speci-
per extremity is imaging the portion
6 original prospective studies6,15-19
ficity ranged from 77% to 100%. The
of the subclavian vein that passes be-
were assessed using the criteria given
sensitivity and specificity varied with
neath the clavicle. Haire et al18 found
in Table 1 and were included in the
the ultrasonographic method used
that this was the major cause of poor
systematic review.
(Table 2). Only 1 study6 reported the
sensitivity and that it occurred most
95% confidence intervals for the sen-
often on the left side. The specificity
STUDY APPRAISAL
sitivity and specificity. The 95% con-
of duplex ultrasonography was high
fidence intervals for sensitivity and
(94%-100%) (Table 2).
Table 2 summarizes the results of
specificity were calculated for the re-
The sensitivity and specificity of
ourappraisalofthe6prospectivestud-
maining studies where possible.
ultrasonography varied with the tech-
ies.Only1study6statedthataconsecu-
nique used. The results indicate that
SAFETY OF WITHHOLDING
Doppler evaluation alone is less sen-
ANTICOAGULANT THERAPY
sitive and less specific than real-time
Thrombosis*
imaging or duplex ultrasonography
No prospective study was identified
for upper extremity deep vein throm-
in which anticoagulant therapy was
bosis.6 The available data do not sup-
withheld without additional testing in
port an important improvement in
patients with suspected upper ex-
sensitivity and specificity by the ad-
Sensitivity†
Specificity†
Follow-up
tremity deep vein thrombosis and
dition of color flow Doppler imag-
negative results on ultrasonographic
NA
NA
None
ing over real-time imaging alone.
testing. One study6 followed up pa-
Third, no study has evaluated
88 (7/8)
100 (6/6)
None
tients who had undergone ultraso-
the safety of withholding anticoagu-
[47-100]
[61-100]
nography and venography; in this
lant therapy without additional test-
100 (9/9)
NI
None
study, the decision to give or with-
ing in patients with suspected upper
[72-100]
hold anticoagulant therapy was based
56 (14/25)
100 (18/18)
None
extremity deep vein thrombosis and
[35-76]
[85-100]
on the venographic results.
negative ultrasonographic results.
100 (8/8)
NI
None
We conclude that ultrasonogra-
[69-100]
COMMENT
phy for clinically suspected upper ex-
88 (7/8)
NI
tremity deep vein thrombosis has not
[47-100]
We sought to determine the sensitiv-
been adequately evaluated. The safety
96 (26/27)
94 (29/31)
6 mo
[82-100]
[69-99]
ity and specificity of ultrasonogra-
of withholding anticoagulant treat-
100 (19/19)
93 (14/15)
phy for the diagnosis of upper ex-
ment in a patient with suspected up-
[82-100]
[68-100]
tremity deep vein thrombosis and to
per extremity deep vein thrombosis
81 (17/21)
77 (20/26)
determine the safety of withholding
and negative ultrasonographic re-
[58-95]
[56-91]
anticoagulant therapy in patients with
sults is uncertain. There is a need for
negative ultrasonographic results. Our
prospective studies in more patients
results support 3 inferences.
using the design features outlined in
(REPRINTED) ARCH INTERN MED/ VOL 162, FEB 25, 2002
WWW.ARCHINTERNMED.COM
403
on April 29, 2011
www.archinternmed.com
Downloaded from
©2002 American Medical Association. All rights reserved.

---

Table 1 to definitively evaluate the
treatment. Arch Intern Med. 1991;151:1934-
12. Albertyn LE, Alcock MK. Diagnosis of internal jugu-
sensitivity, specificity, and safety of
1943.
lar vein thrombosis. Radiology. 1987;162:505-
3. Monreal M, Lafoz E, Ruiz J, Valls R, Alastrue A.
508.
duplex ultrasonography for the diag-
Upper-extremity deep venous thrombosis and pul-
13. Haire WD, Lynch TG, Lieberman RP, Lund GB, Ed-
nosis of upper extremity deep vein
monary embolism: a prospective study. Chest.
ney JA. Utility of duplex ultrasound in the diag-
thrombosis.
1991;99:280-283.
nosis of asymptomatic catheter-induced subcla-
4. Black MD, French GJ, Rasuli P, Bouchard AC. Up-
vian vein thrombosis. J Ultrasound Med. 1991;
Accepted for publication July 31, 2001.
per extremity deep venous thrombosis: underdi-
10:493-496.
agnosed and potentially lethal. Chest. 1993;103:
14. Grassi CJ, Polak JF. Axillary and subclavian ve-
This study was supported by a re-
1887-1890.
nous thrombosis: follow-up evaluation with color
search grant from the Presbyterian
5. Monreal M, Raventos A, Lerma R, et al. Pulmo-
Doppler flow US and venography. Radiology. 1990;
Health Foundation, Oklahoma City,
nary embolism in patients with upper extremity DVT
175:651-654.
Okla. Dr Raskob is a Samuel Roberts
associated to venous central lines: a prospective
15. Weissleder R, Elizondo G, Stark DD. Sono-
study. Thromb Haemost. 1994;72:548-550.
graphic diagnosis of subclavian and internal jugu-
Noble Foundation Presidential Pro-
6. Prandoni P, Polistena P, Bernardi E, et al. Upper-
lar vein thrombosis. J Ultrasound Med. 1987;6:
fessor.
extremity deep vein thrombosis. Arch Intern Med.
577-587.
Corresponding author and
1997;157:57-62.
16. Falk RL, Smith DF. Thrombosis of upper extrem-
reprints: Suman W. Rathbun, MD,
7. Longley DG, Finlay DE, Letourneau JG. Sonogra-
ity thoracic inlet veins: diagnosis with duplex Dop-
Department of Medicine, Room WP
phy of the upper extremity and jugular veins. AJR
pler sonography. AJR Am J Roentgenol. 1987;
Am J Roentgenol. 1993;160:957-962.
149:677-682.
3120, The University of Oklahoma
8. Oliver MA. Doppler color flow imaging. In: Tal-
17. Hubsch PJ, Stiglbauer RL, Schwaighofer BA, Kain-
Health Sciences Center, 920 Stanton
bot SR, Oliver MA, eds. Techniques of Venous Im-
berger FM, Barton PP. Internal jugular and sub-
L. Young Blvd, Oklahoma City, OK
aging. Pasadena, Calif: Appleton Davies Inc; 1992:
clavian vein thrombosis caused by central ve-
73104 (e-mail: suman-rathbun@ouhsc
119-123, 129-130.
nous catheters: evaluation using Doppler blood
9. Meade MO, Richardson WS. Selecting and ap-
flow imaging. J Ultrasound Med. 1988;7:629-
.edu).
praising studies for a systematic review. Ann In-
636.
tern Med. 1997;127:531-537.
18. Haire WD, Lynch TG, Lund GB, Lieberman RP, Ed-
REFERENCES
10. McAlister FA, Clark HD, van Walraven C, et al.
ney JA. Limitations of magnetic resonance imag-
The medical review article revisited: has the sci-
ing and ultrasound-directed (duplex) scanning in
ence improved? Ann Intern Med. 1999;131:
the diagnosis of subclavian vein thrombosis.
1. Horattas MC, Wright DJ, Fenton AH, et al. Chang-
947-951.
J Vasc Surg. 1991;13:391-397.
ing concepts of deep venous thrombosis of the
11. Jaeschke R, Guyatt G, Sackett DL, for the Evidence-
19. Baxter GM, Kincaid W, Jeffrey RF, Millar GM, Por-
upper extremity: report of a series and review of
Based Medicine Working Group. Users’ guides to
teous C, Morley P. Comparison of colour Dop-
the literature. Surgery. 1988;104:561-567.
the medical literature, III: how to use an article
pler ultrasound with venography in the diagno-
2. Becker DM, Philbrick JT, Walker FB. Axillary and
about a diagnositc test: A: are the results of the
sis of axillary and subclavian vein thrombosis. Br
subclavian venous thrombosis: prognosis and
study valid? JAMA. 1994;271:389-391.
J Radiol. 1991;64:777-778.
(REPRINTED) ARCH INTERN MED/ VOL 162, FEB 25, 2002
WWW.ARCHINTERNMED.COM
404
on April 29, 2011
www.archinternmed.com
Downloaded from
©2002 American Medical Association. All rights reserved.

---