Abstract
The objective of this study was to investigate the risk of acute internal jugular, subclavian, and axillary deep venous thrombosis (upper torso DVT [UTDVT]) and pulmonary embolism (PE) and the role of anticoagulation in a cohort of hospitalized patients.
A 2-year retrospective review of hospitalized patients who underwent upper torso vein duplex scanning was performed. Patient demographics, underlying comorbidities, indication for scanning, diagnostic tests, intensive care unit stay, length of stay, presence of a central line (current or within the last 2 weeks), malignancy (current or former), hypercoaguable condition, postoperative state, renal failure, mortality, and use of anticoagulation were recorded. Univariate and multivariate analyses were performed to investigate significant risk factors for acute UTDVT. The impact of an acute UTDVT and use of anticoagulation on hospital length of stay, survival to 30 days and 1 year, and PE rate were calculated.
One hundred eighty-nine patients were scanned. Sixty-three patients (33%) were found to have an acute UTDVT. The internal jugular vein was the most common site of thrombosis. The presence of a central venous catheter was the only factor found to be a significant risk factor for an acute UTDVT (p = .03). Five patients (7.9%) with an UTDVT had a PE documented by computed tomographic angiography-pulmonary arteriography, and all had an internal jugular thrombosis (four isolated and one combined with an axillary-subclavian thrombosis). No PE was fatal. Thirty-eight (60%) patients with an acute UTDVT were treated with therapeutic anticoagulation; the remainder were observed. All patients with a PE received anticoagulation. Hospital length of stay, 30-day mortality, and 12-month survival were no different for patients with and without an UTDVT (p = .7). The use of anticoagulation had no observable effect on survival in patients with UTDVT (p = .1).
An acute internal jugular, subclavian, or axillary DVT is a relatively common finding in the hospitalized patient. Patients with a central line (current or within the previous 14 days) were at greatest risk, with an internal jugular vein thrombosis being the most common source. The inconsistent use of anticoagulation therapy for UTDVT was associated with a moderate risk of PE. A survival benefit for anticoagulation could not be documented.
Originally described in the late nineteenth century by Paget and von Schroetter, 1,2 thrombosis of the axillary and subclavian veins was thought to be a rare occurrence. In fact, prior to 1970, the incidence of upper extremity deep venous thrombosis (DVT) was reported to be less than 2% and felt to have little risk for pulmonary embolism (PE). 3–6 However, as the care of patients, particularly the hospitalized patient, required more frequent use of central venous access for cardiac monitoring, infusion of vasoactive drugs, dialysis access, and implantable cardiac devices, secondary thromboses of the axillary, subclavian, or internal jugular veins (upper torso deep veins) began to be “recognized as being more common than previously reported.” 7
As axillary, subclavian, or internal jugular DVT (upper torso DVT [UTDVT]) became more common in the hospitalized patient, venous duplex scanning became more widely available and replaced venography as the usual method of upper torso vein evaluation. Venous duplex scanning has the advantage over venography in that it is noninvasive, can be performed at the bedside, and in skilled hands has been proven to be highly sensitive and specific. 8 A number of studies have examined the relationship between axillary, subclavian, and internal jugular DVT and PE (Table 1). However, many of these studies include both primary and secondary DVTs in their analysis. 7,9–16 Furthermore, small sample size and PE rates based mostly on ventilation-perfusion (V/Q) scanning further limit their interpretation.
Upper Torso Deep Venous Thrombosis and Pulmonary Embolism
PE = pulmonary embolism; UTDVT = upper extremity deep venous thrombosis.
N = number of patients reported.
With that background, we report a 2-year retrospective study of hospitalized patients subjected to upper torso duplex scanning. The intent of this study was to investigate the frequency of acute UTDVT in this patient cohort, the risk factors associated with its occurrence, and the importance of anticoagulation in the incidence of PE, hospital length of stay (LOS), and survival.
Methods
The charts of inpatients who underwent venous duplex scanning of the upper torso veins from January 2002 to December 2003 at the University of Southern California University Hospital, an urban tertiary care hospital, were retrospectively reviewed. All duplex scans were performed by a registered vascular technologist following the Society for Vascular Ultrasound guidelines. The scan included the cervical portion of the jugular vein to the subclavian confluence and the upper extremity axillary and the subclavian vein segment in its entirety except for the subclavian vein lying posterior to the clavicle. The diagnosis of UTDVT was based on absence of flow as noted on color flow Doppler ultrasonography. Pulsed-wave Doppler ultrasonography was used to examine for respiratory variation, and B-mode ultrasonography was used to estimate the age of any thrombus based on its echogenicity. Hypoechoic thrombi were categorized as “acute” (< 2 weeks old). Hyperechoic thrombi were considered older than 2 weeks and categorized as “chronic.” When thrombi contained both hyper- and hypoechoic properties, they were considered acute for the purposes of subsequent analyses. Compressibility was felt to be contraindicated in all veins except the brachial vein. Patients with symptoms suggesting a possible PE underwent concurrent lower extremity venous scans as well. All studies were reviewed, and the diagnosis was confirmed by a board-certified vascular surgeon or board-certified radiologist.
Demographics, underlying comorbidities (renal failure requiring hemodialysis, diabetes mellitus, current or past malignancy, confirmed hypercoaguable state), presence of a central venous catheter (current or within the prior 2 weeks), intensive care unit (ICU) stay, and surgical procedure were recorded. The indication for upper torso venous duplex scanning was determined. Use of therapeutic anticoagulation either with systemic unfractionated heparin, subcutaneous low-molecular-weight heparin, and/or warfarin was noted. The use of anticoagulant therapy for a positive scan was at the discretion of the attending physician.
For those patients who had multiple scans, only the first scan was used for analysis. Mean hospital LOS, mean length of follow up, 30-day mortality rate, and 12-month survival were calculated. A PE rate was calculated based on a high probability V/Q or computed tomographic angiographic-pulmonary arteriographic (CT-PA) scan. Comparisons between patients with and without a UTDVT were made using the Student t-test for differences between continuous variables and the Fisher exact test for differences between nominal variables. All mean values are reported with the standard error of the mean. Logistic regression with stepwise selection was used to assess the influence of age > 65 years, the presence of a central line (current or within the last 2 weeks), hospitalization in the ICU, malignancy (current or former), hypercoagulability, postoperative state, and renal failure on the development of a UTDVT. Survival data were obtained from hospital chart and outpatient records, calculated using the Kaplan-Meier methods, and analyzed with the log-rank test. Survival curves shown have a standard error of less than 10% unless otherwise indicated. A p value of < .05 was taken as the level of significance for all statistical analyses.
Results
In 189 patients, 1,195 veins were studied by duplex scan. Patients' ages ranged from 18 to 94 years (mean age 58 years), and 49% were male. The indications for scanning included arm swelling (106; 56%), rule-out PE-tachycardia, dyspnea, hypoxia (68; 36%), and fever (15; 8%). The incidences of various comorbidities and risk factors are divided between patients with a UTDVT and those without and are listed in Table 2.
Characteristics of Patients with and without Upper Torso Deep Venous Thrombosis
ICU = intensive care unit; LOS = length of stay; UTDVT = upper torso deep venous thrombosis.
*Fisher exact test.
†Logistic regression.
Scanning of bilateral upper torso veins was performed in 103 patients; only the right upper torso veins were evaluated in 33 patients, and 53 patients underwent scanning of the left upper torso veins only. One hundred six UTDVTs were discovered in 63 patients (33%). The internal jugular vein was the most common site of a UTDVT (Figure 1), with 28 patients having an isolated internal jugular DVT. The remaining 126 patients and scans were negative for a UTDVT.
Number and site of 106 internal jugular (IJ), subclavian, and axillary deep venous thromboses (UTDVT) in 63 patients.
Those patients found to have a UTDVT and those who did not were similar in most clinical variables (see Table 2). Patients with an acute UTDVT had a longer hospital LOS and were more often in the ICU, but this difference was not statistically significant. The only factor found on univariate and multivariate analysis to be a risk factor for the development of a UTDVT was the presence of a central venous line (current or within the last 2 weeks). There was no statistically significant difference in absolute 30-day mortality or late 12-month survival (Figure 2) between patients with and without a UTDVT.
Survival curve of patients with and without upper torso deep venous thrombosis (UTDVT).
Thirty-eight (60%) of the 63 patients in the UTDVT group were treated with anticoagulation therapy. Nineteen patients received unfractionated heparin, 16 low-molecular-weight heparin, and 3 warfarin only. Twenty-five or 40% of patients with a UTDVT received no anticoagulation. No patient had a superior vena cava filter placed.
Thirty-day mortality for patients not anticoagulated was 20% versus 13% for those anticoagulated, but this was not a significant difference (p = .7). Of the 38 patients initially treated, 24 or 38% had anticoagulation therapy continued after discharge from the hospital. At 12 months, no survival benefit was documented in patients with a UTDVT who were anticoagulated (p = .10) (Figure 3).
Percent survival of patients with upper torso deep venous thrombosis treated or not treated with anticoagulation.
Sixty-eight patients underwent upper torso venous duplex scanning specifically because of a suspected PE. Twenty (29%) patients were found to have a UTDVT, and 16 underwent further evaluation for PE using either CT-PA (n = 14) or V/Q scanning (n = 2). Four patients had no further evaluation for PE. Five patients had a CT-PA scan that was positive for a PE. Both V/Q scans were negative. In these five patients, all had internal jugular vein(s) thrombosis and one patient had an associated axillary-subclavian vein thrombosis as well. All patients had an indwelling central venous line, and all were in the ICU at the time of the study. Three were postoperative patients. One patient had a synchronous lower extremity DVT and four additional patients had a lower extremity DVT in the absence of a UTDVT. All patients were treated with anticoagulation therapy. Anticoagulation therapy was directed by the individual practitioner and consisted of either unfractionated heparin infusion, low-molecular-weight heparin, or warfarin alone. Three of five patients had anticoagulation therapy continued after discharge (two with warfarin, one with low-molecular-weight heparin). One patient died in the hospital, although the death was not attributed to the PE.
Discussion
Prior to 1970, upper extremity DVT was regarded as little risk for PE. 3–6 During the past 25 years, upper extremity DVT has gained much attention, generating many reports on the subject. The exact implication of this finding, particularly in the hospitalized patient, with regard to PE risk remains unclear. Because most previous studies on the subject include both primary and secondary DVT, may or may not include internal jugular thromboses, and focus solely on PE risk, only a few studies in the recent literature report a true incidence of UTDVT in a hospitalized cohort in which the DVT has a high likelihood for being secondary to central venous instrumentation. Prandoni and colleagues reported on 13 inpatients (22%) found to have a DVT based on duplex ultrasonography and venography performed for evaluation of a symptom. 12 Another study by Schmittling and colleagues reported on 174 inpatients who underwent scanning for evaluation of a symptom and found 40 patients (23%) to have a DVT. 15 The 33% DVT rate found in this inpatient population, although slightly higher, is consistent with these previous reports. The presence of a central venous catheter (current or within in the previous 14 days) was the only factor found to be influential on the development of UTDVT in the current study. The influence of central venous catheters on the development of DVT has been demonstrated in many studies. 12,15,16 Ascher and colleagues recently documented that the risk of morbidity and mortality from a UTDVT was equivalent for subclavian, axillary, and internal jugular thromboses. 17
Most studies and review articles examining UTDVT recommend anticoagulation therapy and consider it the “cornerstone of therapy.” 18–20 This recommendation is largely based on a concern for the possibility of fatal PE. The other, perhaps better, recommended indication for anticoagulation therapy is to reduce the incidence of postthrombotic syndrome. 18,19 Although most patients with an upper extremity DVT will recover without sequelae, 20 a small percentage can have disabling long-term symptoms. Hingorani and colleagues reported on 170 patients with upper extremity DVT and found that 7 (4%) had persistent significant swelling of the arm, 13 and Prandoni and colleagues reported that 1 patient (2%) developed severe postthrombotic sequelae. 12 The current study had follow-up to only 12 months; consequently, the true incidence of postthrombotic syndrome cannot be determined.
Despite the general recommendations to treat with anticoagulation, 40% of the patients in the current study found to have an acute UTDVT were not treated with anticoagulation therapy. Despite the infrequent use of anticoagulation therapy observed in the current study, a difference in survival between those anticoagulated and those observed was not demonstrated. There was no difference in 30-day mortality or the 12-month survival estimate between those treated and those observed. Furthermore, the overall rate of 6% confirmed, symptomatic PE rate is modest and did not result in mortality. However, when the indication of rule-out PE is considered, the positive UTDVT rate was 29%, and when positive for a UTDVT, the documented PE rate was 25%.
Much of the current literature again reports wide-ranging PE rates owing to both primary and secondary UTDVT. 7,9–14,16 Mustafa and colleagues reported on 65 patients with a UTDVT, and none developed a PE. 14 The largest prospective registry of axillary-sublcavain thromboses, compiled by Joffe and colleagues, documented a PE rate less than 5%, although up to 20% of patients did not have a pulmonary imaging study. 21 A large UTDVT study by Hingorani and colleagues found that 5% of 546 patients with confirmed UTDVT were found to have a PE. 16 At the other end of the spectrum, Harley and colleagues reported on 12 patients with an upper extremity DVT and found 4 (33%) to have suffered a PE. 9 The 6% incidence of PE documented in our study is similar to the findings of Hingorani and colleauges, although it may understate the overall incidence of all PE, both symptomatic and asymptomatic.
Several problems limit interpretation of PE in previous studies. First, all previous studies on UTDVT, including the current one, are underpowered with regard to PE rate. If one considers the data with regard to PE from a lower extremity DVT, 5,000 confirmed DVTs would be required for any study to be properly powered. 22 Second, the current literature reports PE rates based on V/Q lung scanning-a technology with marginal sensitivity and specificity compared with CT-PA. 23–25 In fact, no previous study examining the incidence of PE and UTDVT reported using CT-PA. 7–14,16,21 To truly document the overall PE incidence requires a prospective study that relies on routine use of CT-PA rather than V/Q scanning. The costs, radiation exposure, logistics, and risks of intravenous contrast will make such a study difficult to perform. Lastly, some of the previous studies examining the incidence of PE essentially screened asymptomatic patients using V/Q scans, and those with a high-probability scan were included in the PE rate calculation. 11,12 The relevance of a high-probability V/Q scan in an asymptomatic patient is debatable and may overestimate the clinically significant PE rate.
In the current study, the symptoms of a possible PE and a high index of suspicion in all five patients led to venous duplex scanning and the detection of the UTDVT. Anticoagulation was then instituted after the PE was diagnosed. In no patient did a documented PE occur after a UTDVT was diagnosed and anticoagulation was withheld. The observation that symptoms of PE often lead to the discovery of UTDVT and the uncertain role of anticoagulation in reducing PE has been noted previously. 12,22 In a review article by Becker and colleagues, it was noted that PE occurred with similar frequency in both anticoagulated and non anticoagulated groups and most pulmonary emboli occurred before treatment was initiated. 18
An interesting finding in our report is that patients with a documented PE had a consistent risk profile that included ICU patient, indwelling central line, and internal jugular thrombosis. This clinical triad may be of use in identifying patients with a UTDVT who are at an increased risk for a PE. All patients with a PE except one had an isolated internal jugular DVT. Given that many previous studies on upper extremity DVT do not include the internal jugular vein, the potential contribution of internal jugular thrombosis to the development of a PE may be underrecognized. The increased use of the internal jugular vein as a central venous access site and the findings of this report suggest that routine interrogation of the internal jugular vein is advisable in patients where the source of a PE is being investigated.
In conclusion, internal jugular, subclavian, and axillary DVT appears to be a relatively common finding in a hospitalized cohort of symptomatic patients when subjected to venous duplex scanning. Those patients with a current or recent central venous catheter are at highest risk. The benefits of anticoagulation remain elusive. Improved 30-day and 12-month survival could not be demonstrated for patients receiving anticoagulation. Internal jugular, subclavian, and axillary DVT can clearly lead to a PE; however, the risk of PE as documented by CT-PA in this study is modest, and no PE was fatal. Patients in the ICU with an indwelling central line and internal jugular thrombosis appear to be at highest risk for this complication. The documentation of a PE in the presence of an isolated internal jugular vein thrombosis attests to its importance as a source for PE in the hospitalized patient.