- Email: [email protected]
High Prevalence and Mortality Associated with Upper Extremity Deep Venous Thrombosis in Hospitalized Patients at a Tertiary Care Center
Journal Pre-proof High Prevalence and Mortality Associated with Upper Extremity Deep Venous Thrombosis in Hospitalized Patients at a Tertiary Care Center Rae S. Rokosh, MD, Neel Ranganath, MD, Patricia Yau, MD, Caron Rockman, MD, Mikel Sadek, MD, Todd Berland, MD, Glenn Jacobowitz, MD, Jeff Berger, MD, Thomas S. Maldonado, MD PII:
S0890-5096(19)30902-1
DOI:
https://doi.org/10.1016/j.avsg.2019.10.055
Reference:
AVSG 4717
To appear in:
Annals of Vascular Surgery
Received Date: 10 September 2019 Revised Date:
11 October 2019
Accepted Date: 14 October 2019
Please cite this article as: Rokosh RS, Ranganath N, Yau P, Rockman C, Sadek M, Berland T, Jacobowitz G, Berger J, Maldonado TS, High Prevalence and Mortality Associated with Upper Extremity Deep Venous Thrombosis in Hospitalized Patients at a Tertiary Care Center Annals of Vascular Surgery (2019), doi: https://doi.org/10.1016/j.avsg.2019.10.055. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier Inc.
1 1
High Prevalence and Mortality Associated with Upper Extremity Deep Venous Thrombosis
2
in Hospitalized Patients at a Tertiary Care Center
3 4
Author Names:
5
Rae S Rokosh MD1, Neel Ranganath MD1, Patricia Yau MD1, Caron Rockman MD1, Mikel
6
Sadek MD1, Todd Berland MD1, Glenn Jacobowitz MD1, Jeff Berger MD1, Thomas S
7
Maldonado MD1
8
1
Division of Vascular Surgery, Department of Surgery, NYU Langone Health New York, NY
9 10
Corresponding Author:
11
Thomas S Maldonado, MD
12
Department of Surgery
13
NYU Langone Health
14
530 First Ave, Sixth Floor
15
New York, NY 10016
16
Telephone: 212-263-7311
17
Email: [email protected]
18 19
Presentation Information: This study was presented at the Society for Clinical Vascular
20
Surgery; Las Vegas, NV; March 17-21, 2018
21 22 23
2 24
ABSTRACT
25
Objective: Upper extremity deep venous thrombosis (UEDVT) and its associated complications
26
are increasing in incidence, but management strategies are largely derived from experience
27
treating lower extremity deep venous thrombosis (LEDVT). The purpose of this study was to
28
examine our single institution’s experience with in-hospital venous thromboembolism (VTE),
29
specifically the characteristics and outcomes of the UEDVT population as it compares to
30
LEDVT.*
31
Material and Methods: This is a single tertiary care center retrospective cohort study of all
32
consecutive inpatients diagnosed with acute VTE from June 2015 to December 2015. During this
33
period, 4495 patients underwent venous duplex examination (622 UE and 3873 LE), identifying
34
83 inpatient DVTs. Chronic DVT as well as those diagnosed in the outpatient population were
35
excluded. DVTs were classified as either provoked or unprovoked. Provoked DVT were defined
36
as the presence of any of the following factors within 30 days prior to diagnosis: major surgery,
37
immobilization (greater than 3 days of bedrest), trauma, infection requiring antibiotics, central
38
venous access, pregnancy, and/or hormonal medication use. Inpatient pulmonary embolisms
39
(PE) detected on chest computed tomography (CT) were also evaluated during this time frame.
40
Patient data were collected including age, gender, race, lifestyle factors, comorbidities, VTE risk
41
factors, symptomatology at presentation, management including anticoagulation choice and filter
42
placement if applicable, as well as discharge disposition. Statistical analysis was performed using
43
GraphPad Prism 8.0 (GraphPad Software, San Diego, California, USA), and a threshold p-value
44
of < 0.05 set for significance.
*
Upper extremity deep venous thrombosis (UEDVT), lower extremity deep venous thrombosis (LEDVT). venous thromboembolism (VTE), pulmonary embolism (PE), inferior vena cava (IVC), computed tomography (CT)
3 45
Results: During the study period, 83 DVTs (48 LEDVT, 35 UEDVT) and 24 PE were identified
46
in 96 inpatients. Of these DVTs, 77.1% of these were defined as provoked. Eleven patients had
47
simultaneous DVT and PE, and thirteen patients had PE with presumed occult pelvic or LEDVT.
48
UEDVT patients had a higher proportion of comorbidities than LEDVT patients: coronary artery
49
disease (25.7% vs. 13.1%, p=0.16), congestive heart failure (20% vs. 6.6%, p=0.09), as well as a
50
trend toward higher incidence of malignancy (60% vs. 42.6%, p=0.13). Of provoked VTE,
51
UEDVT correlated more significantly with central venous catheters (88.4% vs. 12.5%,
52
p=<0.0001), but was less commonly associated with prolonged bed rest (19.2% vs. 39.5%,
53
p=0.11). PE was diagnosed in 24/96 (25%) of the study population. Patients with LEDVT were
54
found to have a significantly higher incidence of PE compared to those with UEDVT (34.4% vs.
55
8.6%, p=0.006). Same-admission mortality for patients with VTE was 13/96 (13.5%). Of these,
56
patients with UEDVT had significantly higher all-cause mortality than patients with LEDVT
57
(28.5% vs. 4.9%, p=0.004). When catheter-related UEDVT were excluded, there remained a
58
significant difference in mortality between non-catheter related UEDVT and LEDVT (33.3% vs.
59
4.9% p=0.0119).
60
Conclusions: This study demonstrates a high prevalence of UEDVT in hospitalized patients who
61
experience VTE. Despite a lower incidence of synchronous PE, patients with UEDVT had a
62
higher prevalence of significant medical comorbidities and higher all-cause mortality on the
63
index hospital admission.
64
Keywords: upper extremity deep venous thrombosis
65
Declarations of Interest: None
66
Funding: This research did not receive any specific grant from funding agencies in the public,
67
commercial, or not-for-profit sectors.
4 68 69
INTRODUCTION Venous thromboembolism (VTE), which includes deep venous thrombosis (DVT) and
70
pulmonary embolism (PE), is the third most common cardiovascular disorder with an annual
71
incidence of 0.1%, affecting approximately 5% of the population.1 Upper extremity DVT
72
(UEDVT) has been thought to account for 4-10% of all cases of DVT and may involve the
73
radial, ulnar, brachial, axillary, subclavian, brachiocephalic or internal jugular veins.2
74
Historically, UEDVT has been considered a relatively benign event.3 However as the incidence
75
of UEDVT increases, so do its subsequent complications including PE, venous access
76
difficulties, superior vena cava syndrome, post-thrombotic syndrome and bleeding on therapeutic
77
anticoagulation therapy, suggesting an insufficient understanding and management of UEDVT.
78
Management guidelines, with the exception of those for thoracic outlet syndrome, are largely
79
extrapolated from lower extremity DVT (LEDVT) and PE management.4
80
The objective of this study was to examine and characterize our single institution
81
experience with in-hospital VTE, specifically the characteristics and outcomes of the UEDVT
82
population as it compares to LEDVT.
83
MATERIAL AND METHODS
84
An IRB-approved retrospective review was performed of prospectively collected data on all
85
consecutive inpatients diagnosed with acute VTE at our tertiary care center from June 2015 to
86
December 2015. During this period, 4495 patients underwent venous duplex examination: 622
87
UE and 3873 LE. Each upper extremity venous duplex examination included evaluation of the
88
subclavian, axillary, and brachial veins; each lower extremity duplex examination included
89
evaluation of femoral, popliteal, tibial, gastrocnemius, soleal, and peroneal veins. Research staff
90
was notified via email of acute DVT on duplex and/or PE on chest computed tomography (CT).
5 91
Acute DVT was defined as a hypoechoic smooth thrombus in a distended thin-walled vein with
92
incomplete compressibility and abnormal flow in the absence of collateral veins or evidence of
93
recanalization. This entity was distinguished from chronic DVT, which was defined as an
94
echogenic rigid thrombus in a contracted non-compressible vein with evidence of recanalization
95
or the formation of collateral veins.5 Chronic DVT were excluded from analysis. Other exclusion
96
criteria included outpatient diagnosis of VTE; as a result, this study does not include patients
97
with venous thoracic outlet syndrome.
98
Once the diagnosis of acute VTE was confirmed, patient data was prospectively collected
99
including age, gender, race, lifestyle factors, comorbidities, VTE risk factors, symptomatology at
100
presentation, management including anticoagulation choice and filter placement if applicable, as
101
well as mortality. DVT were classified as provoked or unprovoked. Provoked DVT were defined
102
as the presence of any of the following factors within 30 days prior to diagnosis: major surgery,
103
immobilization (greater than 3 days of bedrest), trauma, infection requiring antibiotics, central
104
venous access, pregnancy, and/or hormonal medication use. DVT were considered catheter-
105
related if the thrombus occurred at the site of a prior catheter, pacemaker or mediport even if that
106
device had been discontinued at the time of DVT diagnosis, as long as the device had been in
107
place within the past 30 days. Patients who presented with concomitant PE were categorized for
108
analysis by DVT location, and those patients with synchronous UE and LEDVT were
109
categorized for analysis as UEDVT. Statistical analysis was performed with GraphPad Prism 8.0
110
(GraphPad Software, San Diego, California, USA) using the student’s t-test for continuous
111
variables and the Fisher exact test for categorical variables, and a threshold p-value of < 0.05 set
112
for significance. New York University School of Medicine provided Institutional Review Board
113
approval and requirement for informed consent was waived.
6 114
RESULTS
115
Patient Demographics and Etiology of Provoked VTE
116
Ninety-six patients with a new diagnosis of VTE during the study period were reviewed.
117
Population demographics include: mean age of 67.6 years (±16.9), 47.9% female, 69.8%
118
Caucasian, and mean BMI 27.6 (Table I). Over 40% of patients were current or former smokers
119
and 12.5% had a personal history of prior VTE. Notably, malignancy was present in 50%; the
120
most common types were genitourinary, gastrointestinal and dermatologic. Provoked VTE were
121
detected in 74 of 96 patients (UE 26, LE 48; 77.1%). The majority of provoked events were
122
attributable to recent major surgery (66.2%), indwelling venous catheters (35.1%), and
123
prolonged immobilization (32.4%). Orthopedic and neurosurgical procedures were the surgical
124
procedures most commonly associated with postoperative VTE (Table II). The remaining 22.9%
125
of VTE cases diagnosed in the inpatient setting were classified as unprovoked.
126
VTE Anatomic Distribution
127
Of the 4495 duplex examinations performed during the 7-month time period, 83 detected an
128
acute DVT in the inpatient population: 35 UE and 48 LE, for an overall 1.8% incidence of
129
inpatient acute DVT. Of these, seven patients (7.3%) were found to have both UE and LEDVT;
130
these were analyzed as UEDVT for the purposes of this study. Thirteen (13.5%) of the 96
131
patients diagnosed with VTE had a PE without an extremity DVT documented on duplex
132
ultrasound; these patients were presumed to have occult pelvic or LEDVT and were analyzed as
133
part of the LEDVT group. Eleven of 96 (11.5%) patients had simultaneous DVT and PE (2 UE,
134
8 LE, 1 mixed). Of the 48 patients diagnosed with LEDVT, 29 had DVT in the infrageniculate
135
veins, 15 in the suprageniculate (iliofemoral/popliteal) veins, and 4 in both distributions. Of the
7 136
35 patients diagnosed with UEDVT, 10 were isolated to the arm (axillary/brachial), 8 were
137
central (internal jugular/subclavian), and 17 were mixed (Table III).
138
Comparison of UEDVT vs. LEDVT Groups
139
Patients with UEDVT had a higher prevalence of medical comorbidities when compared to
140
patients with LEDVT, including coronary artery disease (25.7% vs. 13.1%, p=0.16) and
141
congestive heart failure (20% vs. 6.6%, p=0.09). There was also a trend toward a higher
142
incidence of concomitant malignancy in the UEDVT population (60% vs. 42.6%, p=0.13) (Table
143
IV). Of patients with provoked DVT, those with UEDVT were more likely to have central
144
venous catheters (88.4% vs. 12.5%, p=<0.0001), whereas patients with LEDVT were more likely
145
to have had prolonged bed rest (19.2% vs. 39.5%, p=0.11) (Table V). All catheter-related
146
UEDVT in this study were attributable to peripherally inserted central catheters or non-tunneled
147
central venous catheters; none were attributed to mediports or pacing wires. Indications for
148
catheter placement included resuscitation, long-term antibiotic administration, parenteral
149
nutrition, or hemodialysis. There were no documented central-line associated infections in this
150
cohort.
151
Symptomatology of UEDVT vs. LEDVT Groups
152
The indications for duplex examination included swelling, extremity pain, shortness of
153
breath, fevers, and chest pain. Irrespective of anatomic location, swelling was the most common
154
presenting symptom of DVT (51.4% vs. 35.4%, p=NS) (Table VI); asymptomatic DVT was
155
incidentally diagnosed in 39/83 (46.9%; 42.9% vs. 50%, p=NS).
156
Incidence of PE
157 158
PE was diagnosed in 24/96 (25%) of the study population. Of patients diagnosed with LEDVT, 8/48 (16.7%) had an associated PE. However, if the thirteen occult presumed
8 159
pelvic/LEDVT are included, the incidence of PE in the LEDVT group was 21/61 (34.4%). In
160
comparison, 3/35 (8.6%) of patients with diagnosed UEDVT were found to have associated PE
161
(8.6% vs. 34.4%; p=0.006). Overall, the incidence of PE is significantly lower in UEDVT than
162
LEDVT.
163
Management of UEDVT vs. LEDVT
164
There was no difference in use of inpatient therapeutic intravenous heparin therapy between
165
UE and LEDVT (48.6% vs. 31.1%, p=0.12). Seventy-six (79.1%) patients within this cohort
166
were discharged from the hospital to home or a rehabilitation facility, 21/35 (60%) with UEDVT
167
and 55/61 (90.2%) with LEDVT. The remaining 20/96 (20.8%) of patients were same-admission
168
mortalities (13/96, 13.5%) or made hospice care (7/96, 7.3%). While the majority of each cohort
169
was discharged on anticoagulation, 3/21 UEDVT and 9/55 LEDVT patients did not receive
170
anticoagulation at discharge due to preclusive medical comorbidities (14.3% vs. 16.4%, p=NS).
171
Of these 12 patients discharged without anticoagulation, eight (66.7%) had an IVC filter placed
172
during admission (33.3% vs. 77.8%, p=0.23) and four (33.3%) were on dual antiplatelet therapy
173
at discharge (50% vs. 50%, p=NS).
174
There were no significant differences between the UEDVT and LEDVT cohorts with respect
175
to choice of anticoagulant: apixaban (5.7% vs. 13.1%, p=NS), enoxaparin (42.9% vs. 42.6%,
176
p=NS) with or without bridge to warfarin (22.9% vs. 32.8%, p=NS), dabigatran (0% vs. 3.3%,
177
p=NS), except for rivaroxaban (0% vs. 13.1%, p=0.03) (Table VII). Overall, IVC filters were
178
placed in 29 patients (6 UE, 23 LE, 30.2%), with the most common indication being a transient
179
or indefinite contraindication to anticoagulation (82.8%) at the time of VTE diagnosis (66.7% vs.
180
87%, p=NS). Of these 29 patients, six (20.7%) were same-admission mortalities or discharged to
181
hospice and therefore did not receive anticoagulation; however, of the remaining 23 (79.3%) that
9 182
were discharged from the hospital, 15 (65.2%) were on anticoagulation at discharge (50% vs.
183
52.2%, p=NS) (Table VIII).
184
Disposition
185
Between the UE and LEDVT groups, there were no significant differences in discharge to
186
home (34.3% vs. 39.3%, p=NS), acute/sub-acute rehabilitation facility (22.9% vs. 39.3%, p=NS),
187
long-term care facility (2.9% vs. 11.3%, p=NS), or hospice (11.4% vs. 4.9%, p=NS). Overall
188
same-admission mortality for patients with VTE was 13/96 (13.5%). Of these, patients with
189
UEDVT had significantly higher all-cause mortality than patients with other VTE (28.5% vs.
190
4.9%, p=0.004) (Table IX).
191
Catheter vs. Non-Catheter-Related UEDVT
192
Two subgroup analyses were performed comparing catheter-related vs. non-catheter-
193
related UEDVT, as well as non-catheter-related UEDVT and LEDVT. The first subset analysis
194
comparing catheter-related UEDVT to non-catheter-related UEDVT demonstrated no significant
195
differences in patient demographics and no significant difference in comorbidities with the
196
exception of hyperlipidemia (47.8% vs. 0%) (Table X). Concomitant PE was diagnosed in a
197
relatively small percentage of each group (4.3% vs. 16.7%, p=NS). Catheter-related UEDVT
198
were more likely to present with swelling (65.2% vs. 25%, p=0.0354). There was no significant
199
difference in IVC filter placement (13% vs. 25%, p=NS) or choice of anticoagulation regimen;
200
however, patients with catheter-related UEDVT were more likely to be discharged on enoxaparin
201
(52.2% vs. 25%, p=0.16) whereas patients with non-catheter-related UEDVT were more often
202
discharged on warfarin (30.4% vs. 66.7%, p=0.071), despite no significant difference in
203
malignancy between the two groups (56.5% vs. 66.7%, p=NS). There was also no significant
10 204
difference in disposition, with a similar proportion of discharges to hospice and all-cause same
205
admission mortalities between the two groups (39.1% vs. 41.6%, p=NS).
206
Non-Catheter Related UEDVT vs. LEDVT
207
In comparing non-catheter-related UEDVT to LEDVT in general, there remained no
208
significant differences between patient demographic factors with the exception of hyperlipidemia
209
(0% vs. 42.6%) (Table XI). LEDVT more often presented post-operatively (33% vs. 72.9%,
210
p=0.0169) with pain as the presenting symptom in comparison to non-catheter related UEDVT
211
(0% vs. 18.8%, p=0.18), which are more commonly asymptomatic (66.7% vs. 50%, p=NS).
212
Concomitant PE was more often diagnosed in patients with LEDVT (16.7% vs. 34.4%, p=NS)
213
and a higher proportion of these LEDVT patients had IVC filters placed (25% vs. 37.7%, p=NS)
214
with no significant difference in anticoagulation choice at discharge. Importantly, as previously
215
stated, there was no difference in mortality between catheter related and non-catheter related
216
UEDVT; furthermore, when catheter-related UEDVT are excluded, there remains a significant
217
difference in mortality between non-catheter related UEDVT and LEDVT (33.3% vs. 4.9%,
218
p=0.0119).
219
DISCUSSION
220
Approximately 900,000 people are diagnosed with VTE per year in the United States, one-
221
third of these in the inpatient setting, with an estimated 30% mortality within 30 days of
222
diagnosis.6 As a result, the Agency for Healthcare Research and Quality has deemed VTE the
223
most common cause of preventable death in hospitalized patients, with prevention being the
224
primary priority to improve safety in hospitals.7 The characteristics of our patient population
225
underscore many of the risk factors applicable to vascular disease in general: older age, smoking,
226
malignancy, and both acute and chronic inflammatory states. While LEDVT is a well-studied
11 227
sequela of these risks, we found in our study that UEDVT were more common in the inpatient
228
population than may be expected, representing over one third of all inpatient VTE.
229
Cote et al. recently compared the characteristics of 2,272 patients with UEDVT vs. 35,094
230
patients with LEDVT in the RIETE registry, consisting of a heterogeneous population of both
231
inpatient and outpatients.8 This study had similar results to ours in showing that UEDVT is less
232
often associated with PE than LEDVT (9.8% vs. 25%). This group also found that patients with
233
catheter-related UEDVT have a higher incidence of medical comorbidities (coronary artery
234
disease, congestive heart failure, and diabetes) compared to those with non-catheter-related
235
UEDVT, as in our cohort. Similar to our study, concomitant PE was more common in non-
236
catheter related rather than catheter-related UEDVT. Similarly, they also found that when
237
compared to LEDVT, despite comparable comorbidities, non-catheter-related UEDVT were less
238
often associated with PE (13% vs. 24%).
239
An important difference between our results and those obtained by Cote et al. was our
240
finding of significant increased mortality in patients with UEDVT. While Cote et al.
241
demonstrated a trend toward increased mortality in patients with UEDVT compared to LEDVT
242
(10% vs. 8.2%), this was found to be insignificant on multivariate analysis.16 Our finding of
243
increased mortality in UEDVT patients may reflect our exclusive focus on inpatient VTE. The
244
underlying reasons for why inpatients with VTE have higher UEDVT-associated mortality are
245
worth exploring in detail.
246
Mortality associated with VTE is often presumed to be a result of PE. Review of the
247
literature demonstrates incidence of acute PE in patients diagnosed with UEDVT, primary or
248
secondary, to be approximately 7-9%.8, 13-15 This is consistent with our findings, in which three
249
(8.6%) of the UEDVT patients were diagnosed with PE. Of these patients, one had a left internal
12 250
jugular, one had a left axillary, and one had a right internal jugular/subclavian and concomitant
251
bilateral LEDVTs. All were provoked postoperative DVTs, two status post orthopedic surgery
252
and one status post abdominal surgery. Two had a history of malignancy, endometrial and colon
253
cancer, respectively. Given current management standards for LEDVT, of those patients
254
diagnosed with both UEDVT and PE, an inferior vena cava (IVC) filter was placed in all three
255
patients during their admission secondary to poor reserve following PE complicated by right
256
heart strain. However, whereas IVC filter placement has been well established as standard of
257
care for certain patients with LEDVT, a lack of evidence documenting significant risk of PE
258
from UEDVT and the absence of data supporting the safety and efficacy of superior vena cava
259
filters makes their use in UEDVT controversial.16 No SVC filters were placed in this cohort.
260
Therefore the rationale for IVC filter placement in these patients relies on the assumption of
261
either occult LE/pelvic DVT, or, as we discuss below, a more global
262
thrombophilic/inflammatory state that may predispose patients to second-hit and lethal VTE.
263
Although UEDVT are rarely associated with PE compared to LEDVT as demonstrated in the
264
literature1,15-16, they are associated with significantly increased mortality. Overall we felt this
265
difference most likely reflected patients’ more severe underlying comorbidities and should not be
266
considered a failure of prophylaxis, nor should it be deemed a marker of poor quality of care.
267
Importantly, we found that this significant difference in mortality persisted with the exclusion of
268
patients with catheter-related UEDVT, the cohort with overall more severe underlying
269
comorbidities (Table X). This suggests that perhaps patients with UEDVT have an overall
270
increased thrombotic risk, which may be associated with the increased prevalence of malignancy
271
in these patients irrespective of catheter placement. Interestingly, ALKindi et al. recently
272
evaluated 200 consecutive cases of malignancy-associated UEDVT and demonstrated that, for
13 273
those patients with recurrent catheter-related UEDVT, the majority of recurrences were not in the
274
ipsilateral limb. This supports the notion of an overall increased thrombotic risk in UEDVT
275
patients, particularly with underlying malignancy, rather than simply local endothelial damage or
276
thrombus propagation secondary to catheter presence.17
277
This finding is particularly relevant to our patient population, as in our study, 50% of the
278
global VTE cohort were associated with a pre-existing diagnosis of malignancy. This proportion
279
of patients with cancer was higher than that previously documented in literature, which was
280
closer to 40%.18 It is well established that malignancy is an independent risk factor for VTE: a
281
recent meta-analysis including 45 studies comprising 4580 patients demonstrated that cancer
282
patients have a 2 to 3-fold higher risk of recurrent VTE, and an 8-fold increase in mortality.19
283
However, special considerations for management of patients with malignancy-associated
284
UEDVT may be warranted. Notably 60% of our UEDVT population suffered from malignancy.
285
It is important to address why the 13 patients in this study with PE without associated
286
documented DVT were presumed to have occult pelvic or LEDVT for analysis rather than being
287
treated as a separate entity. Radiographically isolated PE is not a unique clinical situation: a large
288
5,039-autopsy series in Sweden over 30 years diagnosed 1,500 PE with the absence of extremity
289
DVT in 28% of patients.9 Similarly, Girard et al. examined patients diagnosed with PE on axial
290
imaging with concomitant lower extremity duplexes at time of diagnosis and found that 18% had
291
no evidence of LEDVT; given an estimated UEDVT prevalence of 4% in the literature, this
292
leaves approximately 14% of PE with an undefined source.10 There are several established
293
hypotheses regarding the source of PE in the absence of detected extremity DVT: 1) pelvic
294
venous thrombosis, particularly after major orthopedic surgery or post-partum, which would
295
otherwise be undetected on standard lower extremity duplex ultrasound; 2) embolization to
14 296
pulmonary vasculature after complete dislodgement of pre-existing peripheral thrombus; 3)
297
clinically asymptomatic, small infrageniculate LEDVT for which duplex ultrasound has lower
298
sensitivity11; and 4) de novo pulmonary artery thrombosis in the setting of endothelial injury or
299
inflammation.12 De novo pulmonary artery thrombosis, while an interesting hypothesis, is
300
difficult to prove. In light of this, the decision was made to classify this sub-population as
301
LEDVT given that three-quarters of the aforementioned hypotheses pertain to undetected pelvic
302
or LEDVT and, that of the 13 patients in our study with PE without DVT, two (15.4%) did not
303
have extremity duplexes at time of diagnosis to evaluate for DVT. For those eleven patients that
304
did have a negative lower extremity duplex at time of PE diagnosis, eight (72.7%) studies
305
incompletely evaluated the calf veins, with no available serial imaging. We acknowledge, similar
306
to Schwartz et al., that an area of interest for future study to further elucidate this population of
307
PE without documented DVT on conventional lower extremity duplex ultrasound would be the
308
combination of CTA for PE as well as lower extremity CT venography to evaluate
309
simultaneously for both pelvic and LEDVT.
310
The mainstay of therapy for UEDVT, as with LEDVT, is anticoagulation with the aim of
311
alleviating symptoms as well as preventing propagation of thrombus, development of post-
312
thrombotic syndrome, and progression to PE. The 2016 CHEST guidelines treat UEDVT patients
313
as a homogenous group, recommending at least three months of anticoagulation regardless of
314
etiology.4 To date there have been no randomized control trials comparing anticoagulation agents
315
in UEDVT. Four observational studies with a total of 209 UEDVT treated predominately with
316
low molecular weight heparin had a recurrence rate of 1.9% and no incidence of PE.20-24 In our
317
cohort, with the exception of rivaroxaban, which was not used in UEDVT, there were no
318
differences in choice of agent for long-term anticoagulation. The overall agnostic approach
15 319
characterized here highlights a paucity of data assessing the use of novel anticoagulants in
320
subpopulations of acute VTE and warrants future investigation.
321
A recent study by Newton et al. demonstrated the challenges of managing anticoagulation in
322
this highly comorbid UEDVT population.25 Their group evaluated 1100 patients with non-
323
catheter-related UEDVT in the RIETE registry. This cohort had a slightly lower incidence of PE
324
(1.2%) than ours but similarly had a high incidence of malignancy (29%) and associated
325
mortality (15%). As in our study, there was no single specific long-term anticoagulation agent
326
chosen with 59% discharged on warfarin, 35.8% on low molecular weight heparin, and 1.7% on
327
novel oral agents. Long-term follow-up demonstrated association of malignancy with VTE
328
recurrence, suggesting a possible advantage to aggressive anticoagulation in this cohort.
329
However malignancy, in addition to age, was also strongly associated with hemorrhagic
330
complications of anticoagulation. Therefore the benefits of aggressive anticoagulation in the
331
UEDVT cohort, which are most commonly elderly patients with malignancy, must be carefully
332
weighed against the risk of hemorrhage.
333
The unexpected higher prevalence of UEDVT in our study is likely multi-factorial and
334
reflective of broader healthcare trends. The increased placement (peripheral or otherwise) of
335
central venous catheters, pacemakers, and mediports in an aging, sicker population contributes to
336
increasing incidence of UEDVT, as reflected in our inpatient cohort for whom 88.4% of
337
inpatients with UEDVT had catheter placement for resuscitation, long-term antibiotic
338
administration, parenteral nutrition, and/or hemodialysis. In addition, more liberal use and
339
availability of low-cost sonographic imaging results in detection of otherwise subclinical DVT,
340
suggesting that the previously reported incidence of UEDVT in the literature may not be an
341
accurate representation of true incidence.
16 342
In comparison to LEDVT, UEDVT was also associated with higher all-cause mortality in our
343
study, even with exclusion of catheter-related UEDVT, which were associated with a more
344
comorbid patient population. Absent a more significant incidence of PE, this does not illustrate a
345
direct causal relationship between UEDVT and mortality, and the precise etiology of the higher
346
all-cause mortality is unknown. However, the correlations identified in this study suggest two
347
possibilities: first, that UEDVT is a surrogate marker for a highly morbid population without
348
itself having direct effect on outcomes. Second, that UEDVT is the result of an overall increased
349
thrombotic risk in this population. Either of these alternatives have far-reaching implications for
350
future management. As our study was not designed to assess mid and long-term outcomes related
351
to choice or duration of anticoagulant, it is challenging to definitively conclude whether or not
352
applying LEDVT management strategies in the UEDVT population is an adequate approach.
353
However, the possibility that UEDVT patients are at a higher thrombotic risk than their LEDVT
354
counterparts suggests that developing a distinct approach to UEDVT management warrants
355
future investigation.
356
This study has several limitations. First, it is limited to a single center and retrospective in
357
design with a small overall sample size. In addition, standard of care at our institution is
358
sequential compression device boots and prophylactic anticoagulation with subcutaneous heparin
359
for all hospitalized patients without contraindications; however, details on specific patients’ VTE
360
prophylaxis regimens were not recorded for the purposes of this study and we are unable to
361
correlate VTE with prophylaxis failure. Furthermore, long-term follow-up data was unavailable
362
to capture recurrence or chronic sequelae. Lastly, there was insufficient data regarding
363
comprehensive thrombophilia workups in these patients, though this may be less relevant given
364
that Gabriel et al. recently demonstrated on review of the RIETE registry data that thrombophilic
17 365
defects were not risk factors for UEDVT or UEDVT and PE.26 Also important to note is that the
366
reported inpatient incidence of VTE in our present study was 1.8%, which is lower than the
367
reported hospital related VTE in 282 per 10,000 inpatients per Heit et al.27 While this could be
368
considered a study shortcoming, there are several factors that can account for the low reported
369
incidence in this study including the short-term study duration, possible occult infrageniculate
370
LEDVT on lower extremity duplex, exclusion of chronic and age indeterminate DVT from
371
analysis, as well as inclusion of only inpatient VTE rather than those diagnosed in the outpatient
372
population.
373
CONCLUSION
374
UEDVT is less common than LEDVT, and as a result there is little data directly comparing
375
the two entities. To the best of our knowledge, this study is novel in its assessment of the
376
characteristics and outcomes of the UEDVT population as it compares to LEDVT specifically in
377
an inpatient population. We found the prevalence of UEDVT to be higher than expected in
378
hospitalized patients, that patients with UEDVT tend to be more comorbid, and present more
379
commonly with catheter-associated VTE. We also found that despite a lower incidence of PE,
380
patients with UEDVT had higher all-cause mortality (28.5%), greater than the 11% UEDVT
381
three-month mortality previously published by the RIETE registry, which includes inpatient and
382
outpatient VTEs.15 Importantly, a significant difference in mortality persisted with the exclusion
383
of patients with catheter-related UEDVT. These data do not suggest UEDVT is itself a direct
384
cause of higher mortality seen in this population. Rather, increased mortality in patients with
385
UEDVT may suggest an overall increased thrombotic risk, or, further, a more severe global
386
inflammatory state in these patients. Alternatively UEDVT may be a surrogate marker for a
387
sicker patient population with higher mortality rate. Larger studies investigating these findings
18 388
could improve identification of factors associated with increased risk for development of
389
UEDVT as well as identify patients that may benefit from more aggressive treatment.
390
REFERENCES
391
1.
392 393
for development of venous thromboembolism. Clin Chest Med 2010;31:611-28. 2.
394 395
Stein PD, Matta F. Epidemiology and incidence: the scope of the problem and risk factors
Mai C, Hunt D. Upper-extremity deep venous thrombosis: a review. Am J Med 2011;124(5):402–7.
3.
Hingorani A, Ascher E, Lorenson E, et al. Upper extremity deep venous thrombosis and
396
its impact on morbidity and mortality rates in a hospital-based population. J Vasc Surg
397
1997;26(5):853-860.
398
4.
399 400
guideline and expert panel report. Chest 2016;149(2):315-352. 5.
401 402
6.
Beckman MG, Hooper WC, Critchley SE, et al. Venous thromboembolism: a public health concern. Am J Prev Med 2010;38(4):S495-S501.
7.
405 406
Barleben A, Bandyk D. Interpretation of peripheral venous duplex testing. Semin Vasc Surg 2013;26:111-119.
403 404
Kearon C, Akl EA, Ornelas J, et al. Antithrombotic therapy for VTE disease: CHEST
Shojania KG, Duncan BW, McDonald KM, et al. Making health care safer: a critical analysis of patient safety practices. Evid Rep Technol Assess 2001;(43):1-668.
8.
Cote LP, Greenberg S, Caprini JA, et al. Comparisons between upper and lower
407
extremity deep vein thrombosis: a review of the RIETE registry. Clin Appl Thromb
408
Hemost 2017;23(7):748-754.
409 410
9.
Diebold J, Lohrs U. Venous thrombosis and pulmonary embolism: a study of 5039 autopsies. Pathol Res Pract 1991;187:260-266.
19 411
10.
412 413
Girard P, Musset D, Parent F, et al. High prevalence of detectable deep venous thrombosis in patients with acute pulmonary embolism. Chest 1999;116:903-908.
11.
Kassai B, Boissel JP, Cucherate M, et al. A systematic review of the accuracy of
414
ultrasound in the diagnosis of deep venous thrombosis in asymptomatic patients. Thromb
415
Haemost 2004;91:655-666.
416
12.
417 418
thrombosis. Ann Vasc Surg 2012;26(7):973-976. 13.
419 420
14.
Becker DM, Philbrick JT, Walker IV FB. Axillary and subclavian venous thrombosis: prognosis and treatment. Arch Intern Med 1991;151(10):1934-1943.
15.
423 424
Monreal M, Lafoz E, Ruiz J, et al. Upper-extremity deep venous thrombosis and pulmonary embolism: a prospective study. Chest 1991;99(2):280-283.
421 422
Schwartz T, Hingorani A, Ascher E, et al. Pulmonary embolism without deep venous
Munoz FJ, Mismetti P, Poggio R, et al. Clinical outcome of patients with upper-extremity deep vein thrombosis: results from the RIETE registry. Chest 2008;133(1):143-148.
16.
Owens CA, Bui JT, Knuttinen MG, et al. Pulmonary embolism from upper extremity
425
deep vein thrombosis and the role of superior vena cava filters: a review of the literature.
426
J Vasc Interv Radiol 2010;21(6):779-87.
427
17.
ALKindi S, Chai-Adisaksopha C, Cheah M, et al. Management of cancer-associated
428
upper extremity deep vein thrombosis with and without venous catheters at a tertiary care
429
center. Thrombosis Research 2018;166:92-95.
430 431
18.
Bernardi E, Pesavento R, Prandoni P. Upper extremity deep venous thrombosis. Semin Thromb Hemost 2006;32(7):729–36.
20 432
19.
Bleker SM, van Es N, van Gils L, et al. Clinical course of upper extremity deep vein
433
thrombosis in patients with or without cancer: a systematic review. Thromb Res
434
2016;140(1):S81-88.
435
20.
Kovacs MJ, Kahn SR, Rodger M, et al. A pilot study of central venous catheter survival
436
in cancer patients using low-molecular-weight heparin (dalteparin) and warfarin without
437
catheter removal for the treatment of upper extremity deep vein thrombosis (The Catheter
438
Study). J Thromb Haemost 2007;5:1650-1653.
439
21.
440 441
treatment characteristics. J Int Med Res 200432:429-435. 22.
442 443
Karabay O, Yetkin U, Onol H. Upper extremity deep vein thrombosis: clinical and
Prandoni P, Bernardi E, Marchiori A, et al. The long term clinical course of acute deep vein thrombosis of the arm: prospective cohort study. BMJ 2004;329:484-485.
23.
Savage KJ, Wells PS, Schulz V, et al. Outpatient use of low molecular weight heparin
444
(dalteparin) for the treatment of deep vein thrombosis of the upper extremity. Thromb
445
Haemost 1999;82:1008-1010.
446
24.
447 448
Kucher N. Deep-vein thrombosis of the upper extremities. N Engl J Med 2011;364:861869.
25.
Newton DH, Monreal Bosch M, Amendola M, et al. Analysis of noncatheter-associated
449
upper extremity deep venous thrombosis from the RIETE registry. J Vasc Surg Venous
450
Lymphat Disord 2017;5(1):18-24.
451
26.
Gabriel F, Portoles O, Labios M, et al. Usefulness of thrombophilia testing in venous
452
thromboembolic disease: findings from the RIETE registry. Clin Appl Thromb Hemost
453
2013;19(1):42-47.
21 454
27.
Heit JA, Crusan DJ, Ashrani AA, et al. Effect of a near-universal hospitalization-based
455
prophylaxis regimen on annual number of venous thromboembolism events in the US.
456
Blood 2017;130(2):109-114.
457
Table I. Baseline Characteristics of Patients Diagnosed with VTE (n=96)* Age (years) 67.6 ±16.9 Gender (F, %)
47.9
Race (%) Caucasian
69.8
Black
13.5
Asian
4.2
Other
12.5
Smoking (%)
41.7
Prior VTE (%)
12.5
Myocardial Infarction (%)
3.1
PTCA/Stent (%)
7.2
Atrial Fibrillation (%)
17.7
Diabetes (%)
16.7
COPD (%)
5.2
Hypertension (%)
61.5
Hyperlipidemia (%)
38.5
Malignancy (n =48, %)
50
Urologic
20.8
Gastrointestinal
18.8
Dermatologic
14.6
Gynecologic
10.4
* Venous thromboembolism (VTE), percutaneous transluminal coronary angioplasty (PTCA), chronic obstructive pulmonary disease (COPD)
Hematologic
14.6
Neurologic
8.3
Lung
6.3
Breast/Thyroid
4.2
Thrombophilia (%)
1
Table II. Breakdown of Provoked VTE (n = 74) Surgery (%) 66.2 Orthopedic
28.6
Neurosurgical
26.5
Abdominal
20.4
Cardiothoracic
16.3
Gynecologic
4.1
Vascular
2
Central Venous Access (%)
35.1
Prolonged Immobilization (%)
32.4
Infection Requiring Antibiotics (%)
29.7
Trauma (%)
4.1
Pregnancy (%)
1.4
Hormonal Medication (%)
0
Table III. Anatomic Distribution of VTE (n=96)† Upper Extremity
35
Isolated Arm (axillary/brachial)
10
Central (IJ/subclavian)
8
Mixed
17
Lower Extremity
48
Proximal (iliofemoral/popliteal)
29
Distal (infrageniculate)
15
Mixed
4
Occult Pelvic/LEDVT (PE only)
13
Concurrent UEDVT and LEDVT
7
† Internal jugular (IJ), lower extremity deep venous thrombosis (LEDVT), upper extremity deep venous thrombosis (UEDVT), pulmonary embolism (PE)
Table IV. Baseline Characteristics of Patients Diagnosed with UE vs. LEDVT (n=96) UE (n=35) LE (n=61) p-value Age (years) 66.2 (±19.5) 68.4 (±15.4) NS Gender (F)
15 (42.9%)
31 (50.8%)
NS
Smoking
11 (31.4%)
29 (47.5%)
NS
Prior VTE
4 (11.4%)
8 (13.1%)
NS
Coronary Artery Disease
9 (25.7%)
8 (13.1%)
0.16
Heart Failure
7 (20%)
4 (6.6%)
0.09
Cerebrovascular Accident
3 (8.6%)
7 (11.5%)
NS
Diabetes
8 (22.9%)
8 (13.1%)
NS
COPD
1 (2.9%)
4 (6.6%)
NS
Hypertension
21 (60%)
38 (62.2%)
NS
11 (31.4%)
26 (42.6%)
NS
Malignancy
21 (60%)
26 (42.6%)
0.13
Pulmonary Embolism
3 (8.6%)
21 (34.4%)
0.006
Hyperlipidemia
Table V. Breakdown of Provoked UE vs. LEDVT (n = 74) UE (n=26) LE (n=48) Surgery 14 (53.8%) 35 (72.9%)
p-value 0.12
Prolonged Immobilization
5 (19.2%)
19 (39.5%)
0.11
Central Venous Access
23 (88.4%)
6 (12.5%)
<0.0001
Infection Requiring Antibiotics
14 (53.8%)
10 (20.8%)
0.0083
1 (3.8%)
2 (4.2%)
NS
0
1 (2.1%)
NS
Trauma Pregnancy
Table VI. Presenting Symptoms of UE and LEDVT (n=83) UE (n=35) LE (n=48) Pain 2 (5.7%) 9 (18.8%) Swelling Chest Pain Shortness of Breath Asymptomatic
p-value 0.11
18 (51.4%)
17 (35.4%)
NS
0
2 (4.2%)
NS
1 (2.9%)
7 (14.6%)
0.13
15 (42.9%)
24 (50%)
NS
Table VII. Therapeutic Management of UE vs. LEDVT (n=96) UE (n=35) LE (n=61) Intravenous Heparin 17 (48.6%) 19 (31.1%) Warfarin
p-value 0.12
8 (22.9%)
20 (32.8%)
0.35
0
8 (13.1%)
0.03
Apixaban
2 (5.7%)
8 (13.1%)
0.32
Dabigatran
0
2 (3.3%)
NS
Enoxaparin
15 (42.9%)
26 (42.6%)
NS
Rivaroxaban
Table VIII. IVC Filter Placement in UE vs. LEDVT (n=29)‡ UE (n=6) LE (n=23) Indications for Filter Placement
p-value
AC contraindicated at VTE diagnosis
4 (66.7%)
20 (87%)
0.26
Poor reserve after PE
2 (33.3%)
3 (13%)
0.26
3 (50%)
12 (52.2%)
NS
Anticoagulation at Discharge
‡ Anticoagulation (AC), inferior vena cava (IVC)
Table IX. Disposition of UE vs. LE DVT (n=96) UE (n=35) Home (%) 12 (34.3%)
LE (n=61) 24 (39.3%)
p-value NS
Acute/Sub-acute Facility (%)
8 (22.9%)
24 (39.3%)
0.12
Long Term Care Facility (%)
1 (2.9%)
7 (11.3%)
0.25
Hospice (%)
4 (11.4%)
3 (4.9%)
0.25
Expired (%)
10 (28.5%)
3 (4.9%)
0.004
Table X. Comparison of Catheter-Related vs. Non-Catheter Related UEDVT (n=35) Non Catheter Catheter Related Related UEDVT UEDVT (n =23) (n=12) Age (years) 64.5 (±21.1) 69.1 (±16.5)
p-value NS
Gender (F)
11 (47.8%)
4 (33.3%)
NS
Smoking
8 (34.8%)
3 (25%)
NS
Prior VTE
2 (8.7%)
2 (16.7%)
NS
Coronary Artery Disease
7 (30.4%)
2 (16.7%)
NS
Heart Failure
5 (21.7%)
2 (16.7%)
NS
Cerebrovascular Accident
2 (8.7%)
1 (8.3%)
NS
Diabetes
6 (26.1%)
2 (16.7%)
NS
0 (0%)
1 (8.3%)
NS
Hypertension
14 (60.9%)
7 (58.3%)
NS
Hyperlipidemia
11 (47.8%)
0 (0%)
0.0055
Malignancy
13 (56.5%)
8 (66.7%)
NS
IVC filter
3 (13%)
3 (25%)
NS
Pulmonary Embolism
1 (4.3%)
2 (16.7%)
NS
10 (43.5%)
4 (33.3%)
NS
2 (8.7%)
3 (25%)
NS
Central Venous Access
23 (100%)
0 (0%)
<0.0001
Infection Requiring Antibiotics
11 (47.8%)
3 (25%)
0.28
Trauma
0 (0%)
1 (8.3%)
NS
Pregnancy
0 (0%)
0 (0%)
NS
COPD
Surgery Prolonged Immobilization
Pain
2 (8.7%)
0 (0%)
NS
15 (65.2%)
3 (25%)
0.0354
Chest Pain
0 (0%)
0 (0%)
NS
Shortness of Breath
0 (0%)
1 (8.3%)
NS
Asymptomatic
7 (30.4%)
8 (66.7%)
0.071
Intravenous Heparin
11 (47.8%)
6 (50%)
NS
Warfarin
3 (13%)
5 (41.7%)
0.09
Rivaroxaban
0 (0%)
0 (0%)
NS
Apixaban
1 (4.3%)
1 (8.3%)
NS
Dabigatran
0 (0%)
0 (0%)
NS
Enoxaparin
12 (52.2%)
3 (25%)
0.16
Home (%)
9 (39.1%)
3 (25%)
NS
Acute/Sub-acute Facility (%)
4 (17.4%)
4 (33.3%)
NS
Long Term Care Facility (%)
1 (4.3%)
0 (0%)
NS
Hospice (%)
3 (13%)
1 (8.3%)
NS
Expired (%)
6 (26.1%)
4 (33.3%)
NS
12/14 (85.7%)
6/7 (85.7%)
NS
Swelling
AC at discharge
Table XI. Comparison of Non-catheter related UEDVT vs. LEDVT Non-Catheter Related UEDVT (n=12) Age (years) 69.1 (±16.5)
LEDVT (n=varies) 68.4 (±15.4)
p-value NS
4 (33.3%)
31/61 (50.8%)
NS
Smoking
3 (25%)
29/61 (47.5%)
NS
Prior VTE
2 (16.7%)
8/61 (13.1%)
NS
Coronary Artery Disease
2 (16.7%)
8/61 (13.1%)
0.66
Heart Failure
2 (16.7%)
4/61 (6.6%)
0.25
Cerebrovascular Accident
1 (8.3%)
7/61 (11.5%)
NS
Diabetes
2 (16.7%)
8/61 (13.1%)
NS
COPD
1 (8.3%)
4/61 (6.6%)
NS
Hypertension
7 (58.3%)
38/61 (62.2%)
NS
0 (0%)
26/61 (42.6%)
0.0059
8 (66.7%)
26/61 (42.6%)
0.21
3 (25%)
23/61 (37.7%)
NS
Pulmonary Embolism
2 (16.7%)
21/61 (34.4%)
NS
Surgery
4 (33.3%)
35/48 (72.9%)
0.0169
Prolonged Immobilization
3 (25%)
19/48 (39.5%)
NS
Central Venous Access
0 (0%)
6/48 (12.5%)
NS
Infection Requiring Antibiotics
3 (25%)
10/48 (20.8%)
NS
Trauma
1 (8.3%)
2/48 (4.2%)
NS
Pregnancy
0 (0%)
1/48 (2.1%)
NS
Pain
0 (0%)
9/48 (18.8%)
0.18
Gender (F)
Hyperlipidemia Malignancy IVC filter
Swelling
3 (25%)
17/48 (35.4%)
NS
Chest Pain
0 (0%)
2/48 (4.2%)
NS
Shortness of Breath
1 (8.3%)
7/48 (14.6%)
NS
Asymptomatic
8 (66.7%)
24/48 (50%)
NS
6 (50%)
19/61 (31.1%)
NS
5 (41.7%)
20/61 (32.8%)
NS
0 (0%)
8/61 (13.1%)
NS
Apixaban
1 (8.3%)
8/61 (13.1%)
NS
Dabigatran
0 (0%)
2/61 (3.3%)
NS
Enoxaparin
3 (25%)
26/61 (42.6%)
NS
Home (%)
3 (25%)
24/61 (39.3%)
NS
Acute/Sub-acute Facility (%)
4 (33.3%)
24/61 (39.3%)
NS
Long Term Care Facility (%)
0 (0%)
7/61 (11.3%)
NS
Hospice (%)
1 (8.3%)
3/61 (4.9%)
NS
Expired (%)
4 (33.3%)
3/61 (4.9%)
0.0119
6/7 (85.7%)
46/55 (83.6%)
NS
Intravenous Heparin Warfarin Rivaroxaban
AC at discharge
AVS Highlights: •
High inpatient prevalence upper extremity deep venous thrombosis (UEDVT)
•
UEDVT compared to LEDVT more comorbid and significantly lower likelihood of PE
•
UEDVT higher associated all-cause mortality even excluding catheter-related UEDVT
•
UEDVT suggests increased thrombotic risk or surrogate marker for sick population
S0890-5096(19)30902-1
DOI:
https://doi.org/10.1016/j.avsg.2019.10.055
Reference:
AVSG 4717
To appear in:
Annals of Vascular Surgery
Received Date: 10 September 2019 Revised Date:
11 October 2019
Accepted Date: 14 October 2019
Please cite this article as: Rokosh RS, Ranganath N, Yau P, Rockman C, Sadek M, Berland T, Jacobowitz G, Berger J, Maldonado TS, High Prevalence and Mortality Associated with Upper Extremity Deep Venous Thrombosis in Hospitalized Patients at a Tertiary Care Center Annals of Vascular Surgery (2019), doi: https://doi.org/10.1016/j.avsg.2019.10.055. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier Inc.
1 1
High Prevalence and Mortality Associated with Upper Extremity Deep Venous Thrombosis
2
in Hospitalized Patients at a Tertiary Care Center
3 4
Author Names:
5
Rae S Rokosh MD1, Neel Ranganath MD1, Patricia Yau MD1, Caron Rockman MD1, Mikel
6
Sadek MD1, Todd Berland MD1, Glenn Jacobowitz MD1, Jeff Berger MD1, Thomas S
7
Maldonado MD1
8
1
Division of Vascular Surgery, Department of Surgery, NYU Langone Health New York, NY
9 10
Corresponding Author:
11
Thomas S Maldonado, MD
12
Department of Surgery
13
NYU Langone Health
14
530 First Ave, Sixth Floor
15
New York, NY 10016
16
Telephone: 212-263-7311
17
Email: [email protected]
18 19
Presentation Information: This study was presented at the Society for Clinical Vascular
20
Surgery; Las Vegas, NV; March 17-21, 2018
21 22 23
2 24
ABSTRACT
25
Objective: Upper extremity deep venous thrombosis (UEDVT) and its associated complications
26
are increasing in incidence, but management strategies are largely derived from experience
27
treating lower extremity deep venous thrombosis (LEDVT). The purpose of this study was to
28
examine our single institution’s experience with in-hospital venous thromboembolism (VTE),
29
specifically the characteristics and outcomes of the UEDVT population as it compares to
30
LEDVT.*
31
Material and Methods: This is a single tertiary care center retrospective cohort study of all
32
consecutive inpatients diagnosed with acute VTE from June 2015 to December 2015. During this
33
period, 4495 patients underwent venous duplex examination (622 UE and 3873 LE), identifying
34
83 inpatient DVTs. Chronic DVT as well as those diagnosed in the outpatient population were
35
excluded. DVTs were classified as either provoked or unprovoked. Provoked DVT were defined
36
as the presence of any of the following factors within 30 days prior to diagnosis: major surgery,
37
immobilization (greater than 3 days of bedrest), trauma, infection requiring antibiotics, central
38
venous access, pregnancy, and/or hormonal medication use. Inpatient pulmonary embolisms
39
(PE) detected on chest computed tomography (CT) were also evaluated during this time frame.
40
Patient data were collected including age, gender, race, lifestyle factors, comorbidities, VTE risk
41
factors, symptomatology at presentation, management including anticoagulation choice and filter
42
placement if applicable, as well as discharge disposition. Statistical analysis was performed using
43
GraphPad Prism 8.0 (GraphPad Software, San Diego, California, USA), and a threshold p-value
44
of < 0.05 set for significance.
*
Upper extremity deep venous thrombosis (UEDVT), lower extremity deep venous thrombosis (LEDVT). venous thromboembolism (VTE), pulmonary embolism (PE), inferior vena cava (IVC), computed tomography (CT)
3 45
Results: During the study period, 83 DVTs (48 LEDVT, 35 UEDVT) and 24 PE were identified
46
in 96 inpatients. Of these DVTs, 77.1% of these were defined as provoked. Eleven patients had
47
simultaneous DVT and PE, and thirteen patients had PE with presumed occult pelvic or LEDVT.
48
UEDVT patients had a higher proportion of comorbidities than LEDVT patients: coronary artery
49
disease (25.7% vs. 13.1%, p=0.16), congestive heart failure (20% vs. 6.6%, p=0.09), as well as a
50
trend toward higher incidence of malignancy (60% vs. 42.6%, p=0.13). Of provoked VTE,
51
UEDVT correlated more significantly with central venous catheters (88.4% vs. 12.5%,
52
p=<0.0001), but was less commonly associated with prolonged bed rest (19.2% vs. 39.5%,
53
p=0.11). PE was diagnosed in 24/96 (25%) of the study population. Patients with LEDVT were
54
found to have a significantly higher incidence of PE compared to those with UEDVT (34.4% vs.
55
8.6%, p=0.006). Same-admission mortality for patients with VTE was 13/96 (13.5%). Of these,
56
patients with UEDVT had significantly higher all-cause mortality than patients with LEDVT
57
(28.5% vs. 4.9%, p=0.004). When catheter-related UEDVT were excluded, there remained a
58
significant difference in mortality between non-catheter related UEDVT and LEDVT (33.3% vs.
59
4.9% p=0.0119).
60
Conclusions: This study demonstrates a high prevalence of UEDVT in hospitalized patients who
61
experience VTE. Despite a lower incidence of synchronous PE, patients with UEDVT had a
62
higher prevalence of significant medical comorbidities and higher all-cause mortality on the
63
index hospital admission.
64
Keywords: upper extremity deep venous thrombosis
65
Declarations of Interest: None
66
Funding: This research did not receive any specific grant from funding agencies in the public,
67
commercial, or not-for-profit sectors.
4 68 69
INTRODUCTION Venous thromboembolism (VTE), which includes deep venous thrombosis (DVT) and
70
pulmonary embolism (PE), is the third most common cardiovascular disorder with an annual
71
incidence of 0.1%, affecting approximately 5% of the population.1 Upper extremity DVT
72
(UEDVT) has been thought to account for 4-10% of all cases of DVT and may involve the
73
radial, ulnar, brachial, axillary, subclavian, brachiocephalic or internal jugular veins.2
74
Historically, UEDVT has been considered a relatively benign event.3 However as the incidence
75
of UEDVT increases, so do its subsequent complications including PE, venous access
76
difficulties, superior vena cava syndrome, post-thrombotic syndrome and bleeding on therapeutic
77
anticoagulation therapy, suggesting an insufficient understanding and management of UEDVT.
78
Management guidelines, with the exception of those for thoracic outlet syndrome, are largely
79
extrapolated from lower extremity DVT (LEDVT) and PE management.4
80
The objective of this study was to examine and characterize our single institution
81
experience with in-hospital VTE, specifically the characteristics and outcomes of the UEDVT
82
population as it compares to LEDVT.
83
MATERIAL AND METHODS
84
An IRB-approved retrospective review was performed of prospectively collected data on all
85
consecutive inpatients diagnosed with acute VTE at our tertiary care center from June 2015 to
86
December 2015. During this period, 4495 patients underwent venous duplex examination: 622
87
UE and 3873 LE. Each upper extremity venous duplex examination included evaluation of the
88
subclavian, axillary, and brachial veins; each lower extremity duplex examination included
89
evaluation of femoral, popliteal, tibial, gastrocnemius, soleal, and peroneal veins. Research staff
90
was notified via email of acute DVT on duplex and/or PE on chest computed tomography (CT).
5 91
Acute DVT was defined as a hypoechoic smooth thrombus in a distended thin-walled vein with
92
incomplete compressibility and abnormal flow in the absence of collateral veins or evidence of
93
recanalization. This entity was distinguished from chronic DVT, which was defined as an
94
echogenic rigid thrombus in a contracted non-compressible vein with evidence of recanalization
95
or the formation of collateral veins.5 Chronic DVT were excluded from analysis. Other exclusion
96
criteria included outpatient diagnosis of VTE; as a result, this study does not include patients
97
with venous thoracic outlet syndrome.
98
Once the diagnosis of acute VTE was confirmed, patient data was prospectively collected
99
including age, gender, race, lifestyle factors, comorbidities, VTE risk factors, symptomatology at
100
presentation, management including anticoagulation choice and filter placement if applicable, as
101
well as mortality. DVT were classified as provoked or unprovoked. Provoked DVT were defined
102
as the presence of any of the following factors within 30 days prior to diagnosis: major surgery,
103
immobilization (greater than 3 days of bedrest), trauma, infection requiring antibiotics, central
104
venous access, pregnancy, and/or hormonal medication use. DVT were considered catheter-
105
related if the thrombus occurred at the site of a prior catheter, pacemaker or mediport even if that
106
device had been discontinued at the time of DVT diagnosis, as long as the device had been in
107
place within the past 30 days. Patients who presented with concomitant PE were categorized for
108
analysis by DVT location, and those patients with synchronous UE and LEDVT were
109
categorized for analysis as UEDVT. Statistical analysis was performed with GraphPad Prism 8.0
110
(GraphPad Software, San Diego, California, USA) using the student’s t-test for continuous
111
variables and the Fisher exact test for categorical variables, and a threshold p-value of < 0.05 set
112
for significance. New York University School of Medicine provided Institutional Review Board
113
approval and requirement for informed consent was waived.
6 114
RESULTS
115
Patient Demographics and Etiology of Provoked VTE
116
Ninety-six patients with a new diagnosis of VTE during the study period were reviewed.
117
Population demographics include: mean age of 67.6 years (±16.9), 47.9% female, 69.8%
118
Caucasian, and mean BMI 27.6 (Table I). Over 40% of patients were current or former smokers
119
and 12.5% had a personal history of prior VTE. Notably, malignancy was present in 50%; the
120
most common types were genitourinary, gastrointestinal and dermatologic. Provoked VTE were
121
detected in 74 of 96 patients (UE 26, LE 48; 77.1%). The majority of provoked events were
122
attributable to recent major surgery (66.2%), indwelling venous catheters (35.1%), and
123
prolonged immobilization (32.4%). Orthopedic and neurosurgical procedures were the surgical
124
procedures most commonly associated with postoperative VTE (Table II). The remaining 22.9%
125
of VTE cases diagnosed in the inpatient setting were classified as unprovoked.
126
VTE Anatomic Distribution
127
Of the 4495 duplex examinations performed during the 7-month time period, 83 detected an
128
acute DVT in the inpatient population: 35 UE and 48 LE, for an overall 1.8% incidence of
129
inpatient acute DVT. Of these, seven patients (7.3%) were found to have both UE and LEDVT;
130
these were analyzed as UEDVT for the purposes of this study. Thirteen (13.5%) of the 96
131
patients diagnosed with VTE had a PE without an extremity DVT documented on duplex
132
ultrasound; these patients were presumed to have occult pelvic or LEDVT and were analyzed as
133
part of the LEDVT group. Eleven of 96 (11.5%) patients had simultaneous DVT and PE (2 UE,
134
8 LE, 1 mixed). Of the 48 patients diagnosed with LEDVT, 29 had DVT in the infrageniculate
135
veins, 15 in the suprageniculate (iliofemoral/popliteal) veins, and 4 in both distributions. Of the
7 136
35 patients diagnosed with UEDVT, 10 were isolated to the arm (axillary/brachial), 8 were
137
central (internal jugular/subclavian), and 17 were mixed (Table III).
138
Comparison of UEDVT vs. LEDVT Groups
139
Patients with UEDVT had a higher prevalence of medical comorbidities when compared to
140
patients with LEDVT, including coronary artery disease (25.7% vs. 13.1%, p=0.16) and
141
congestive heart failure (20% vs. 6.6%, p=0.09). There was also a trend toward a higher
142
incidence of concomitant malignancy in the UEDVT population (60% vs. 42.6%, p=0.13) (Table
143
IV). Of patients with provoked DVT, those with UEDVT were more likely to have central
144
venous catheters (88.4% vs. 12.5%, p=<0.0001), whereas patients with LEDVT were more likely
145
to have had prolonged bed rest (19.2% vs. 39.5%, p=0.11) (Table V). All catheter-related
146
UEDVT in this study were attributable to peripherally inserted central catheters or non-tunneled
147
central venous catheters; none were attributed to mediports or pacing wires. Indications for
148
catheter placement included resuscitation, long-term antibiotic administration, parenteral
149
nutrition, or hemodialysis. There were no documented central-line associated infections in this
150
cohort.
151
Symptomatology of UEDVT vs. LEDVT Groups
152
The indications for duplex examination included swelling, extremity pain, shortness of
153
breath, fevers, and chest pain. Irrespective of anatomic location, swelling was the most common
154
presenting symptom of DVT (51.4% vs. 35.4%, p=NS) (Table VI); asymptomatic DVT was
155
incidentally diagnosed in 39/83 (46.9%; 42.9% vs. 50%, p=NS).
156
Incidence of PE
157 158
PE was diagnosed in 24/96 (25%) of the study population. Of patients diagnosed with LEDVT, 8/48 (16.7%) had an associated PE. However, if the thirteen occult presumed
8 159
pelvic/LEDVT are included, the incidence of PE in the LEDVT group was 21/61 (34.4%). In
160
comparison, 3/35 (8.6%) of patients with diagnosed UEDVT were found to have associated PE
161
(8.6% vs. 34.4%; p=0.006). Overall, the incidence of PE is significantly lower in UEDVT than
162
LEDVT.
163
Management of UEDVT vs. LEDVT
164
There was no difference in use of inpatient therapeutic intravenous heparin therapy between
165
UE and LEDVT (48.6% vs. 31.1%, p=0.12). Seventy-six (79.1%) patients within this cohort
166
were discharged from the hospital to home or a rehabilitation facility, 21/35 (60%) with UEDVT
167
and 55/61 (90.2%) with LEDVT. The remaining 20/96 (20.8%) of patients were same-admission
168
mortalities (13/96, 13.5%) or made hospice care (7/96, 7.3%). While the majority of each cohort
169
was discharged on anticoagulation, 3/21 UEDVT and 9/55 LEDVT patients did not receive
170
anticoagulation at discharge due to preclusive medical comorbidities (14.3% vs. 16.4%, p=NS).
171
Of these 12 patients discharged without anticoagulation, eight (66.7%) had an IVC filter placed
172
during admission (33.3% vs. 77.8%, p=0.23) and four (33.3%) were on dual antiplatelet therapy
173
at discharge (50% vs. 50%, p=NS).
174
There were no significant differences between the UEDVT and LEDVT cohorts with respect
175
to choice of anticoagulant: apixaban (5.7% vs. 13.1%, p=NS), enoxaparin (42.9% vs. 42.6%,
176
p=NS) with or without bridge to warfarin (22.9% vs. 32.8%, p=NS), dabigatran (0% vs. 3.3%,
177
p=NS), except for rivaroxaban (0% vs. 13.1%, p=0.03) (Table VII). Overall, IVC filters were
178
placed in 29 patients (6 UE, 23 LE, 30.2%), with the most common indication being a transient
179
or indefinite contraindication to anticoagulation (82.8%) at the time of VTE diagnosis (66.7% vs.
180
87%, p=NS). Of these 29 patients, six (20.7%) were same-admission mortalities or discharged to
181
hospice and therefore did not receive anticoagulation; however, of the remaining 23 (79.3%) that
9 182
were discharged from the hospital, 15 (65.2%) were on anticoagulation at discharge (50% vs.
183
52.2%, p=NS) (Table VIII).
184
Disposition
185
Between the UE and LEDVT groups, there were no significant differences in discharge to
186
home (34.3% vs. 39.3%, p=NS), acute/sub-acute rehabilitation facility (22.9% vs. 39.3%, p=NS),
187
long-term care facility (2.9% vs. 11.3%, p=NS), or hospice (11.4% vs. 4.9%, p=NS). Overall
188
same-admission mortality for patients with VTE was 13/96 (13.5%). Of these, patients with
189
UEDVT had significantly higher all-cause mortality than patients with other VTE (28.5% vs.
190
4.9%, p=0.004) (Table IX).
191
Catheter vs. Non-Catheter-Related UEDVT
192
Two subgroup analyses were performed comparing catheter-related vs. non-catheter-
193
related UEDVT, as well as non-catheter-related UEDVT and LEDVT. The first subset analysis
194
comparing catheter-related UEDVT to non-catheter-related UEDVT demonstrated no significant
195
differences in patient demographics and no significant difference in comorbidities with the
196
exception of hyperlipidemia (47.8% vs. 0%) (Table X). Concomitant PE was diagnosed in a
197
relatively small percentage of each group (4.3% vs. 16.7%, p=NS). Catheter-related UEDVT
198
were more likely to present with swelling (65.2% vs. 25%, p=0.0354). There was no significant
199
difference in IVC filter placement (13% vs. 25%, p=NS) or choice of anticoagulation regimen;
200
however, patients with catheter-related UEDVT were more likely to be discharged on enoxaparin
201
(52.2% vs. 25%, p=0.16) whereas patients with non-catheter-related UEDVT were more often
202
discharged on warfarin (30.4% vs. 66.7%, p=0.071), despite no significant difference in
203
malignancy between the two groups (56.5% vs. 66.7%, p=NS). There was also no significant
10 204
difference in disposition, with a similar proportion of discharges to hospice and all-cause same
205
admission mortalities between the two groups (39.1% vs. 41.6%, p=NS).
206
Non-Catheter Related UEDVT vs. LEDVT
207
In comparing non-catheter-related UEDVT to LEDVT in general, there remained no
208
significant differences between patient demographic factors with the exception of hyperlipidemia
209
(0% vs. 42.6%) (Table XI). LEDVT more often presented post-operatively (33% vs. 72.9%,
210
p=0.0169) with pain as the presenting symptom in comparison to non-catheter related UEDVT
211
(0% vs. 18.8%, p=0.18), which are more commonly asymptomatic (66.7% vs. 50%, p=NS).
212
Concomitant PE was more often diagnosed in patients with LEDVT (16.7% vs. 34.4%, p=NS)
213
and a higher proportion of these LEDVT patients had IVC filters placed (25% vs. 37.7%, p=NS)
214
with no significant difference in anticoagulation choice at discharge. Importantly, as previously
215
stated, there was no difference in mortality between catheter related and non-catheter related
216
UEDVT; furthermore, when catheter-related UEDVT are excluded, there remains a significant
217
difference in mortality between non-catheter related UEDVT and LEDVT (33.3% vs. 4.9%,
218
p=0.0119).
219
DISCUSSION
220
Approximately 900,000 people are diagnosed with VTE per year in the United States, one-
221
third of these in the inpatient setting, with an estimated 30% mortality within 30 days of
222
diagnosis.6 As a result, the Agency for Healthcare Research and Quality has deemed VTE the
223
most common cause of preventable death in hospitalized patients, with prevention being the
224
primary priority to improve safety in hospitals.7 The characteristics of our patient population
225
underscore many of the risk factors applicable to vascular disease in general: older age, smoking,
226
malignancy, and both acute and chronic inflammatory states. While LEDVT is a well-studied
11 227
sequela of these risks, we found in our study that UEDVT were more common in the inpatient
228
population than may be expected, representing over one third of all inpatient VTE.
229
Cote et al. recently compared the characteristics of 2,272 patients with UEDVT vs. 35,094
230
patients with LEDVT in the RIETE registry, consisting of a heterogeneous population of both
231
inpatient and outpatients.8 This study had similar results to ours in showing that UEDVT is less
232
often associated with PE than LEDVT (9.8% vs. 25%). This group also found that patients with
233
catheter-related UEDVT have a higher incidence of medical comorbidities (coronary artery
234
disease, congestive heart failure, and diabetes) compared to those with non-catheter-related
235
UEDVT, as in our cohort. Similar to our study, concomitant PE was more common in non-
236
catheter related rather than catheter-related UEDVT. Similarly, they also found that when
237
compared to LEDVT, despite comparable comorbidities, non-catheter-related UEDVT were less
238
often associated with PE (13% vs. 24%).
239
An important difference between our results and those obtained by Cote et al. was our
240
finding of significant increased mortality in patients with UEDVT. While Cote et al.
241
demonstrated a trend toward increased mortality in patients with UEDVT compared to LEDVT
242
(10% vs. 8.2%), this was found to be insignificant on multivariate analysis.16 Our finding of
243
increased mortality in UEDVT patients may reflect our exclusive focus on inpatient VTE. The
244
underlying reasons for why inpatients with VTE have higher UEDVT-associated mortality are
245
worth exploring in detail.
246
Mortality associated with VTE is often presumed to be a result of PE. Review of the
247
literature demonstrates incidence of acute PE in patients diagnosed with UEDVT, primary or
248
secondary, to be approximately 7-9%.8, 13-15 This is consistent with our findings, in which three
249
(8.6%) of the UEDVT patients were diagnosed with PE. Of these patients, one had a left internal
12 250
jugular, one had a left axillary, and one had a right internal jugular/subclavian and concomitant
251
bilateral LEDVTs. All were provoked postoperative DVTs, two status post orthopedic surgery
252
and one status post abdominal surgery. Two had a history of malignancy, endometrial and colon
253
cancer, respectively. Given current management standards for LEDVT, of those patients
254
diagnosed with both UEDVT and PE, an inferior vena cava (IVC) filter was placed in all three
255
patients during their admission secondary to poor reserve following PE complicated by right
256
heart strain. However, whereas IVC filter placement has been well established as standard of
257
care for certain patients with LEDVT, a lack of evidence documenting significant risk of PE
258
from UEDVT and the absence of data supporting the safety and efficacy of superior vena cava
259
filters makes their use in UEDVT controversial.16 No SVC filters were placed in this cohort.
260
Therefore the rationale for IVC filter placement in these patients relies on the assumption of
261
either occult LE/pelvic DVT, or, as we discuss below, a more global
262
thrombophilic/inflammatory state that may predispose patients to second-hit and lethal VTE.
263
Although UEDVT are rarely associated with PE compared to LEDVT as demonstrated in the
264
literature1,15-16, they are associated with significantly increased mortality. Overall we felt this
265
difference most likely reflected patients’ more severe underlying comorbidities and should not be
266
considered a failure of prophylaxis, nor should it be deemed a marker of poor quality of care.
267
Importantly, we found that this significant difference in mortality persisted with the exclusion of
268
patients with catheter-related UEDVT, the cohort with overall more severe underlying
269
comorbidities (Table X). This suggests that perhaps patients with UEDVT have an overall
270
increased thrombotic risk, which may be associated with the increased prevalence of malignancy
271
in these patients irrespective of catheter placement. Interestingly, ALKindi et al. recently
272
evaluated 200 consecutive cases of malignancy-associated UEDVT and demonstrated that, for
13 273
those patients with recurrent catheter-related UEDVT, the majority of recurrences were not in the
274
ipsilateral limb. This supports the notion of an overall increased thrombotic risk in UEDVT
275
patients, particularly with underlying malignancy, rather than simply local endothelial damage or
276
thrombus propagation secondary to catheter presence.17
277
This finding is particularly relevant to our patient population, as in our study, 50% of the
278
global VTE cohort were associated with a pre-existing diagnosis of malignancy. This proportion
279
of patients with cancer was higher than that previously documented in literature, which was
280
closer to 40%.18 It is well established that malignancy is an independent risk factor for VTE: a
281
recent meta-analysis including 45 studies comprising 4580 patients demonstrated that cancer
282
patients have a 2 to 3-fold higher risk of recurrent VTE, and an 8-fold increase in mortality.19
283
However, special considerations for management of patients with malignancy-associated
284
UEDVT may be warranted. Notably 60% of our UEDVT population suffered from malignancy.
285
It is important to address why the 13 patients in this study with PE without associated
286
documented DVT were presumed to have occult pelvic or LEDVT for analysis rather than being
287
treated as a separate entity. Radiographically isolated PE is not a unique clinical situation: a large
288
5,039-autopsy series in Sweden over 30 years diagnosed 1,500 PE with the absence of extremity
289
DVT in 28% of patients.9 Similarly, Girard et al. examined patients diagnosed with PE on axial
290
imaging with concomitant lower extremity duplexes at time of diagnosis and found that 18% had
291
no evidence of LEDVT; given an estimated UEDVT prevalence of 4% in the literature, this
292
leaves approximately 14% of PE with an undefined source.10 There are several established
293
hypotheses regarding the source of PE in the absence of detected extremity DVT: 1) pelvic
294
venous thrombosis, particularly after major orthopedic surgery or post-partum, which would
295
otherwise be undetected on standard lower extremity duplex ultrasound; 2) embolization to
14 296
pulmonary vasculature after complete dislodgement of pre-existing peripheral thrombus; 3)
297
clinically asymptomatic, small infrageniculate LEDVT for which duplex ultrasound has lower
298
sensitivity11; and 4) de novo pulmonary artery thrombosis in the setting of endothelial injury or
299
inflammation.12 De novo pulmonary artery thrombosis, while an interesting hypothesis, is
300
difficult to prove. In light of this, the decision was made to classify this sub-population as
301
LEDVT given that three-quarters of the aforementioned hypotheses pertain to undetected pelvic
302
or LEDVT and, that of the 13 patients in our study with PE without DVT, two (15.4%) did not
303
have extremity duplexes at time of diagnosis to evaluate for DVT. For those eleven patients that
304
did have a negative lower extremity duplex at time of PE diagnosis, eight (72.7%) studies
305
incompletely evaluated the calf veins, with no available serial imaging. We acknowledge, similar
306
to Schwartz et al., that an area of interest for future study to further elucidate this population of
307
PE without documented DVT on conventional lower extremity duplex ultrasound would be the
308
combination of CTA for PE as well as lower extremity CT venography to evaluate
309
simultaneously for both pelvic and LEDVT.
310
The mainstay of therapy for UEDVT, as with LEDVT, is anticoagulation with the aim of
311
alleviating symptoms as well as preventing propagation of thrombus, development of post-
312
thrombotic syndrome, and progression to PE. The 2016 CHEST guidelines treat UEDVT patients
313
as a homogenous group, recommending at least three months of anticoagulation regardless of
314
etiology.4 To date there have been no randomized control trials comparing anticoagulation agents
315
in UEDVT. Four observational studies with a total of 209 UEDVT treated predominately with
316
low molecular weight heparin had a recurrence rate of 1.9% and no incidence of PE.20-24 In our
317
cohort, with the exception of rivaroxaban, which was not used in UEDVT, there were no
318
differences in choice of agent for long-term anticoagulation. The overall agnostic approach
15 319
characterized here highlights a paucity of data assessing the use of novel anticoagulants in
320
subpopulations of acute VTE and warrants future investigation.
321
A recent study by Newton et al. demonstrated the challenges of managing anticoagulation in
322
this highly comorbid UEDVT population.25 Their group evaluated 1100 patients with non-
323
catheter-related UEDVT in the RIETE registry. This cohort had a slightly lower incidence of PE
324
(1.2%) than ours but similarly had a high incidence of malignancy (29%) and associated
325
mortality (15%). As in our study, there was no single specific long-term anticoagulation agent
326
chosen with 59% discharged on warfarin, 35.8% on low molecular weight heparin, and 1.7% on
327
novel oral agents. Long-term follow-up demonstrated association of malignancy with VTE
328
recurrence, suggesting a possible advantage to aggressive anticoagulation in this cohort.
329
However malignancy, in addition to age, was also strongly associated with hemorrhagic
330
complications of anticoagulation. Therefore the benefits of aggressive anticoagulation in the
331
UEDVT cohort, which are most commonly elderly patients with malignancy, must be carefully
332
weighed against the risk of hemorrhage.
333
The unexpected higher prevalence of UEDVT in our study is likely multi-factorial and
334
reflective of broader healthcare trends. The increased placement (peripheral or otherwise) of
335
central venous catheters, pacemakers, and mediports in an aging, sicker population contributes to
336
increasing incidence of UEDVT, as reflected in our inpatient cohort for whom 88.4% of
337
inpatients with UEDVT had catheter placement for resuscitation, long-term antibiotic
338
administration, parenteral nutrition, and/or hemodialysis. In addition, more liberal use and
339
availability of low-cost sonographic imaging results in detection of otherwise subclinical DVT,
340
suggesting that the previously reported incidence of UEDVT in the literature may not be an
341
accurate representation of true incidence.
16 342
In comparison to LEDVT, UEDVT was also associated with higher all-cause mortality in our
343
study, even with exclusion of catheter-related UEDVT, which were associated with a more
344
comorbid patient population. Absent a more significant incidence of PE, this does not illustrate a
345
direct causal relationship between UEDVT and mortality, and the precise etiology of the higher
346
all-cause mortality is unknown. However, the correlations identified in this study suggest two
347
possibilities: first, that UEDVT is a surrogate marker for a highly morbid population without
348
itself having direct effect on outcomes. Second, that UEDVT is the result of an overall increased
349
thrombotic risk in this population. Either of these alternatives have far-reaching implications for
350
future management. As our study was not designed to assess mid and long-term outcomes related
351
to choice or duration of anticoagulant, it is challenging to definitively conclude whether or not
352
applying LEDVT management strategies in the UEDVT population is an adequate approach.
353
However, the possibility that UEDVT patients are at a higher thrombotic risk than their LEDVT
354
counterparts suggests that developing a distinct approach to UEDVT management warrants
355
future investigation.
356
This study has several limitations. First, it is limited to a single center and retrospective in
357
design with a small overall sample size. In addition, standard of care at our institution is
358
sequential compression device boots and prophylactic anticoagulation with subcutaneous heparin
359
for all hospitalized patients without contraindications; however, details on specific patients’ VTE
360
prophylaxis regimens were not recorded for the purposes of this study and we are unable to
361
correlate VTE with prophylaxis failure. Furthermore, long-term follow-up data was unavailable
362
to capture recurrence or chronic sequelae. Lastly, there was insufficient data regarding
363
comprehensive thrombophilia workups in these patients, though this may be less relevant given
364
that Gabriel et al. recently demonstrated on review of the RIETE registry data that thrombophilic
17 365
defects were not risk factors for UEDVT or UEDVT and PE.26 Also important to note is that the
366
reported inpatient incidence of VTE in our present study was 1.8%, which is lower than the
367
reported hospital related VTE in 282 per 10,000 inpatients per Heit et al.27 While this could be
368
considered a study shortcoming, there are several factors that can account for the low reported
369
incidence in this study including the short-term study duration, possible occult infrageniculate
370
LEDVT on lower extremity duplex, exclusion of chronic and age indeterminate DVT from
371
analysis, as well as inclusion of only inpatient VTE rather than those diagnosed in the outpatient
372
population.
373
CONCLUSION
374
UEDVT is less common than LEDVT, and as a result there is little data directly comparing
375
the two entities. To the best of our knowledge, this study is novel in its assessment of the
376
characteristics and outcomes of the UEDVT population as it compares to LEDVT specifically in
377
an inpatient population. We found the prevalence of UEDVT to be higher than expected in
378
hospitalized patients, that patients with UEDVT tend to be more comorbid, and present more
379
commonly with catheter-associated VTE. We also found that despite a lower incidence of PE,
380
patients with UEDVT had higher all-cause mortality (28.5%), greater than the 11% UEDVT
381
three-month mortality previously published by the RIETE registry, which includes inpatient and
382
outpatient VTEs.15 Importantly, a significant difference in mortality persisted with the exclusion
383
of patients with catheter-related UEDVT. These data do not suggest UEDVT is itself a direct
384
cause of higher mortality seen in this population. Rather, increased mortality in patients with
385
UEDVT may suggest an overall increased thrombotic risk, or, further, a more severe global
386
inflammatory state in these patients. Alternatively UEDVT may be a surrogate marker for a
387
sicker patient population with higher mortality rate. Larger studies investigating these findings
18 388
could improve identification of factors associated with increased risk for development of
389
UEDVT as well as identify patients that may benefit from more aggressive treatment.
390
REFERENCES
391
1.
392 393
for development of venous thromboembolism. Clin Chest Med 2010;31:611-28. 2.
394 395
Stein PD, Matta F. Epidemiology and incidence: the scope of the problem and risk factors
Mai C, Hunt D. Upper-extremity deep venous thrombosis: a review. Am J Med 2011;124(5):402–7.
3.
Hingorani A, Ascher E, Lorenson E, et al. Upper extremity deep venous thrombosis and
396
its impact on morbidity and mortality rates in a hospital-based population. J Vasc Surg
397
1997;26(5):853-860.
398
4.
399 400
guideline and expert panel report. Chest 2016;149(2):315-352. 5.
401 402
6.
Beckman MG, Hooper WC, Critchley SE, et al. Venous thromboembolism: a public health concern. Am J Prev Med 2010;38(4):S495-S501.
7.
405 406
Barleben A, Bandyk D. Interpretation of peripheral venous duplex testing. Semin Vasc Surg 2013;26:111-119.
403 404
Kearon C, Akl EA, Ornelas J, et al. Antithrombotic therapy for VTE disease: CHEST
Shojania KG, Duncan BW, McDonald KM, et al. Making health care safer: a critical analysis of patient safety practices. Evid Rep Technol Assess 2001;(43):1-668.
8.
Cote LP, Greenberg S, Caprini JA, et al. Comparisons between upper and lower
407
extremity deep vein thrombosis: a review of the RIETE registry. Clin Appl Thromb
408
Hemost 2017;23(7):748-754.
409 410
9.
Diebold J, Lohrs U. Venous thrombosis and pulmonary embolism: a study of 5039 autopsies. Pathol Res Pract 1991;187:260-266.
19 411
10.
412 413
Girard P, Musset D, Parent F, et al. High prevalence of detectable deep venous thrombosis in patients with acute pulmonary embolism. Chest 1999;116:903-908.
11.
Kassai B, Boissel JP, Cucherate M, et al. A systematic review of the accuracy of
414
ultrasound in the diagnosis of deep venous thrombosis in asymptomatic patients. Thromb
415
Haemost 2004;91:655-666.
416
12.
417 418
thrombosis. Ann Vasc Surg 2012;26(7):973-976. 13.
419 420
14.
Becker DM, Philbrick JT, Walker IV FB. Axillary and subclavian venous thrombosis: prognosis and treatment. Arch Intern Med 1991;151(10):1934-1943.
15.
423 424
Monreal M, Lafoz E, Ruiz J, et al. Upper-extremity deep venous thrombosis and pulmonary embolism: a prospective study. Chest 1991;99(2):280-283.
421 422
Schwartz T, Hingorani A, Ascher E, et al. Pulmonary embolism without deep venous
Munoz FJ, Mismetti P, Poggio R, et al. Clinical outcome of patients with upper-extremity deep vein thrombosis: results from the RIETE registry. Chest 2008;133(1):143-148.
16.
Owens CA, Bui JT, Knuttinen MG, et al. Pulmonary embolism from upper extremity
425
deep vein thrombosis and the role of superior vena cava filters: a review of the literature.
426
J Vasc Interv Radiol 2010;21(6):779-87.
427
17.
ALKindi S, Chai-Adisaksopha C, Cheah M, et al. Management of cancer-associated
428
upper extremity deep vein thrombosis with and without venous catheters at a tertiary care
429
center. Thrombosis Research 2018;166:92-95.
430 431
18.
Bernardi E, Pesavento R, Prandoni P. Upper extremity deep venous thrombosis. Semin Thromb Hemost 2006;32(7):729–36.
20 432
19.
Bleker SM, van Es N, van Gils L, et al. Clinical course of upper extremity deep vein
433
thrombosis in patients with or without cancer: a systematic review. Thromb Res
434
2016;140(1):S81-88.
435
20.
Kovacs MJ, Kahn SR, Rodger M, et al. A pilot study of central venous catheter survival
436
in cancer patients using low-molecular-weight heparin (dalteparin) and warfarin without
437
catheter removal for the treatment of upper extremity deep vein thrombosis (The Catheter
438
Study). J Thromb Haemost 2007;5:1650-1653.
439
21.
440 441
treatment characteristics. J Int Med Res 200432:429-435. 22.
442 443
Karabay O, Yetkin U, Onol H. Upper extremity deep vein thrombosis: clinical and
Prandoni P, Bernardi E, Marchiori A, et al. The long term clinical course of acute deep vein thrombosis of the arm: prospective cohort study. BMJ 2004;329:484-485.
23.
Savage KJ, Wells PS, Schulz V, et al. Outpatient use of low molecular weight heparin
444
(dalteparin) for the treatment of deep vein thrombosis of the upper extremity. Thromb
445
Haemost 1999;82:1008-1010.
446
24.
447 448
Kucher N. Deep-vein thrombosis of the upper extremities. N Engl J Med 2011;364:861869.
25.
Newton DH, Monreal Bosch M, Amendola M, et al. Analysis of noncatheter-associated
449
upper extremity deep venous thrombosis from the RIETE registry. J Vasc Surg Venous
450
Lymphat Disord 2017;5(1):18-24.
451
26.
Gabriel F, Portoles O, Labios M, et al. Usefulness of thrombophilia testing in venous
452
thromboembolic disease: findings from the RIETE registry. Clin Appl Thromb Hemost
453
2013;19(1):42-47.
21 454
27.
Heit JA, Crusan DJ, Ashrani AA, et al. Effect of a near-universal hospitalization-based
455
prophylaxis regimen on annual number of venous thromboembolism events in the US.
456
Blood 2017;130(2):109-114.
457
Table I. Baseline Characteristics of Patients Diagnosed with VTE (n=96)* Age (years) 67.6 ±16.9 Gender (F, %)
47.9
Race (%) Caucasian
69.8
Black
13.5
Asian
4.2
Other
12.5
Smoking (%)
41.7
Prior VTE (%)
12.5
Myocardial Infarction (%)
3.1
PTCA/Stent (%)
7.2
Atrial Fibrillation (%)
17.7
Diabetes (%)
16.7
COPD (%)
5.2
Hypertension (%)
61.5
Hyperlipidemia (%)
38.5
Malignancy (n =48, %)
50
Urologic
20.8
Gastrointestinal
18.8
Dermatologic
14.6
Gynecologic
10.4
* Venous thromboembolism (VTE), percutaneous transluminal coronary angioplasty (PTCA), chronic obstructive pulmonary disease (COPD)
Hematologic
14.6
Neurologic
8.3
Lung
6.3
Breast/Thyroid
4.2
Thrombophilia (%)
1
Table II. Breakdown of Provoked VTE (n = 74) Surgery (%) 66.2 Orthopedic
28.6
Neurosurgical
26.5
Abdominal
20.4
Cardiothoracic
16.3
Gynecologic
4.1
Vascular
2
Central Venous Access (%)
35.1
Prolonged Immobilization (%)
32.4
Infection Requiring Antibiotics (%)
29.7
Trauma (%)
4.1
Pregnancy (%)
1.4
Hormonal Medication (%)
0
Table III. Anatomic Distribution of VTE (n=96)† Upper Extremity
35
Isolated Arm (axillary/brachial)
10
Central (IJ/subclavian)
8
Mixed
17
Lower Extremity
48
Proximal (iliofemoral/popliteal)
29
Distal (infrageniculate)
15
Mixed
4
Occult Pelvic/LEDVT (PE only)
13
Concurrent UEDVT and LEDVT
7
† Internal jugular (IJ), lower extremity deep venous thrombosis (LEDVT), upper extremity deep venous thrombosis (UEDVT), pulmonary embolism (PE)
Table IV. Baseline Characteristics of Patients Diagnosed with UE vs. LEDVT (n=96) UE (n=35) LE (n=61) p-value Age (years) 66.2 (±19.5) 68.4 (±15.4) NS Gender (F)
15 (42.9%)
31 (50.8%)
NS
Smoking
11 (31.4%)
29 (47.5%)
NS
Prior VTE
4 (11.4%)
8 (13.1%)
NS
Coronary Artery Disease
9 (25.7%)
8 (13.1%)
0.16
Heart Failure
7 (20%)
4 (6.6%)
0.09
Cerebrovascular Accident
3 (8.6%)
7 (11.5%)
NS
Diabetes
8 (22.9%)
8 (13.1%)
NS
COPD
1 (2.9%)
4 (6.6%)
NS
Hypertension
21 (60%)
38 (62.2%)
NS
11 (31.4%)
26 (42.6%)
NS
Malignancy
21 (60%)
26 (42.6%)
0.13
Pulmonary Embolism
3 (8.6%)
21 (34.4%)
0.006
Hyperlipidemia
Table V. Breakdown of Provoked UE vs. LEDVT (n = 74) UE (n=26) LE (n=48) Surgery 14 (53.8%) 35 (72.9%)
p-value 0.12
Prolonged Immobilization
5 (19.2%)
19 (39.5%)
0.11
Central Venous Access
23 (88.4%)
6 (12.5%)
<0.0001
Infection Requiring Antibiotics
14 (53.8%)
10 (20.8%)
0.0083
1 (3.8%)
2 (4.2%)
NS
0
1 (2.1%)
NS
Trauma Pregnancy
Table VI. Presenting Symptoms of UE and LEDVT (n=83) UE (n=35) LE (n=48) Pain 2 (5.7%) 9 (18.8%) Swelling Chest Pain Shortness of Breath Asymptomatic
p-value 0.11
18 (51.4%)
17 (35.4%)
NS
0
2 (4.2%)
NS
1 (2.9%)
7 (14.6%)
0.13
15 (42.9%)
24 (50%)
NS
Table VII. Therapeutic Management of UE vs. LEDVT (n=96) UE (n=35) LE (n=61) Intravenous Heparin 17 (48.6%) 19 (31.1%) Warfarin
p-value 0.12
8 (22.9%)
20 (32.8%)
0.35
0
8 (13.1%)
0.03
Apixaban
2 (5.7%)
8 (13.1%)
0.32
Dabigatran
0
2 (3.3%)
NS
Enoxaparin
15 (42.9%)
26 (42.6%)
NS
Rivaroxaban
Table VIII. IVC Filter Placement in UE vs. LEDVT (n=29)‡ UE (n=6) LE (n=23) Indications for Filter Placement
p-value
AC contraindicated at VTE diagnosis
4 (66.7%)
20 (87%)
0.26
Poor reserve after PE
2 (33.3%)
3 (13%)
0.26
3 (50%)
12 (52.2%)
NS
Anticoagulation at Discharge
‡ Anticoagulation (AC), inferior vena cava (IVC)
Table IX. Disposition of UE vs. LE DVT (n=96) UE (n=35) Home (%) 12 (34.3%)
LE (n=61) 24 (39.3%)
p-value NS
Acute/Sub-acute Facility (%)
8 (22.9%)
24 (39.3%)
0.12
Long Term Care Facility (%)
1 (2.9%)
7 (11.3%)
0.25
Hospice (%)
4 (11.4%)
3 (4.9%)
0.25
Expired (%)
10 (28.5%)
3 (4.9%)
0.004
Table X. Comparison of Catheter-Related vs. Non-Catheter Related UEDVT (n=35) Non Catheter Catheter Related Related UEDVT UEDVT (n =23) (n=12) Age (years) 64.5 (±21.1) 69.1 (±16.5)
p-value NS
Gender (F)
11 (47.8%)
4 (33.3%)
NS
Smoking
8 (34.8%)
3 (25%)
NS
Prior VTE
2 (8.7%)
2 (16.7%)
NS
Coronary Artery Disease
7 (30.4%)
2 (16.7%)
NS
Heart Failure
5 (21.7%)
2 (16.7%)
NS
Cerebrovascular Accident
2 (8.7%)
1 (8.3%)
NS
Diabetes
6 (26.1%)
2 (16.7%)
NS
0 (0%)
1 (8.3%)
NS
Hypertension
14 (60.9%)
7 (58.3%)
NS
Hyperlipidemia
11 (47.8%)
0 (0%)
0.0055
Malignancy
13 (56.5%)
8 (66.7%)
NS
IVC filter
3 (13%)
3 (25%)
NS
Pulmonary Embolism
1 (4.3%)
2 (16.7%)
NS
10 (43.5%)
4 (33.3%)
NS
2 (8.7%)
3 (25%)
NS
Central Venous Access
23 (100%)
0 (0%)
<0.0001
Infection Requiring Antibiotics
11 (47.8%)
3 (25%)
0.28
Trauma
0 (0%)
1 (8.3%)
NS
Pregnancy
0 (0%)
0 (0%)
NS
COPD
Surgery Prolonged Immobilization
Pain
2 (8.7%)
0 (0%)
NS
15 (65.2%)
3 (25%)
0.0354
Chest Pain
0 (0%)
0 (0%)
NS
Shortness of Breath
0 (0%)
1 (8.3%)
NS
Asymptomatic
7 (30.4%)
8 (66.7%)
0.071
Intravenous Heparin
11 (47.8%)
6 (50%)
NS
Warfarin
3 (13%)
5 (41.7%)
0.09
Rivaroxaban
0 (0%)
0 (0%)
NS
Apixaban
1 (4.3%)
1 (8.3%)
NS
Dabigatran
0 (0%)
0 (0%)
NS
Enoxaparin
12 (52.2%)
3 (25%)
0.16
Home (%)
9 (39.1%)
3 (25%)
NS
Acute/Sub-acute Facility (%)
4 (17.4%)
4 (33.3%)
NS
Long Term Care Facility (%)
1 (4.3%)
0 (0%)
NS
Hospice (%)
3 (13%)
1 (8.3%)
NS
Expired (%)
6 (26.1%)
4 (33.3%)
NS
12/14 (85.7%)
6/7 (85.7%)
NS
Swelling
AC at discharge
Table XI. Comparison of Non-catheter related UEDVT vs. LEDVT Non-Catheter Related UEDVT (n=12) Age (years) 69.1 (±16.5)
LEDVT (n=varies) 68.4 (±15.4)
p-value NS
4 (33.3%)
31/61 (50.8%)
NS
Smoking
3 (25%)
29/61 (47.5%)
NS
Prior VTE
2 (16.7%)
8/61 (13.1%)
NS
Coronary Artery Disease
2 (16.7%)
8/61 (13.1%)
0.66
Heart Failure
2 (16.7%)
4/61 (6.6%)
0.25
Cerebrovascular Accident
1 (8.3%)
7/61 (11.5%)
NS
Diabetes
2 (16.7%)
8/61 (13.1%)
NS
COPD
1 (8.3%)
4/61 (6.6%)
NS
Hypertension
7 (58.3%)
38/61 (62.2%)
NS
0 (0%)
26/61 (42.6%)
0.0059
8 (66.7%)
26/61 (42.6%)
0.21
3 (25%)
23/61 (37.7%)
NS
Pulmonary Embolism
2 (16.7%)
21/61 (34.4%)
NS
Surgery
4 (33.3%)
35/48 (72.9%)
0.0169
Prolonged Immobilization
3 (25%)
19/48 (39.5%)
NS
Central Venous Access
0 (0%)
6/48 (12.5%)
NS
Infection Requiring Antibiotics
3 (25%)
10/48 (20.8%)
NS
Trauma
1 (8.3%)
2/48 (4.2%)
NS
Pregnancy
0 (0%)
1/48 (2.1%)
NS
Pain
0 (0%)
9/48 (18.8%)
0.18
Gender (F)
Hyperlipidemia Malignancy IVC filter
Swelling
3 (25%)
17/48 (35.4%)
NS
Chest Pain
0 (0%)
2/48 (4.2%)
NS
Shortness of Breath
1 (8.3%)
7/48 (14.6%)
NS
Asymptomatic
8 (66.7%)
24/48 (50%)
NS
6 (50%)
19/61 (31.1%)
NS
5 (41.7%)
20/61 (32.8%)
NS
0 (0%)
8/61 (13.1%)
NS
Apixaban
1 (8.3%)
8/61 (13.1%)
NS
Dabigatran
0 (0%)
2/61 (3.3%)
NS
Enoxaparin
3 (25%)
26/61 (42.6%)
NS
Home (%)
3 (25%)
24/61 (39.3%)
NS
Acute/Sub-acute Facility (%)
4 (33.3%)
24/61 (39.3%)
NS
Long Term Care Facility (%)
0 (0%)
7/61 (11.3%)
NS
Hospice (%)
1 (8.3%)
3/61 (4.9%)
NS
Expired (%)
4 (33.3%)
3/61 (4.9%)
0.0119
6/7 (85.7%)
46/55 (83.6%)
NS
Intravenous Heparin Warfarin Rivaroxaban
AC at discharge
AVS Highlights: •
High inpatient prevalence upper extremity deep venous thrombosis (UEDVT)
•
UEDVT compared to LEDVT more comorbid and significantly lower likelihood of PE
•
UEDVT higher associated all-cause mortality even excluding catheter-related UEDVT
•
UEDVT suggests increased thrombotic risk or surrogate marker for sick population