| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Vascular and Interventional Radiology |
1 From the Departments of Radiology (S.O.T., J.K.F., M.S.J., H.S.) and Medicine (W.T.A., P.H.K.), Indiana University School of Medicine, University Hospital, Rm 0279, 550 N University Blvd, Indianapolis, IN 46202-5253. Received November 17, 1999; revision requested December 30; revision received January 24, 2000; accepted February 22. Supported by a grant from Bard Access Systems. Address correspondence to S.O.T. (e-mail: streroto@iupui.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: A retrospective analysis was performed of 774 catheters placed. Only patients with complete follow-up were included, which yielded a population of 279 catheters in 238 patients (166 in the SCV, 113 in the IJV; total of 26,242 catheter days). All catheters were placed by interventional radiologists with ultrasonographic (in IJV) or venographic (in SCV) guidance.
RESULTS: Initial complications were limited to one pneumothorax in the SCV group and one episode of oversedation in the IJV group. There was no difference in infection rates between the two sites (SVC vs IJV: 0.25 vs 0.32 per 100 catheter days; P > .99). The mean dwell time was slightly longer for SCV catheters (103 days) than for IJV catheters (79 days) (P = .04). Venous thrombosis developed in 13% of patients (0.12 per 100 catheter days) with an SVC catheter placed as compared with in 3% (0.04 per 100 catheter days) with an IJV catheter (P = .018). This difference persisted after adjustment for catheter size and side of placement (P = .025). The mean time to thrombosis was 36 days for SCV catheters and 142 days for IJV catheters.
CONCLUSION: The IJV is the preferred site for tunneled infusion catheter placement because of the lower incidence of symptomatic venous thrombosis.
Index terms: Catheters and catheterization, complications, 907.442, 9462.442 • Veins, access, 907.1269, 9462.1269 • Veins, interventional procedures, 907.1269, 9462.1269 • Veins, thrombosis, 907.442, 9462.442
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The association of subclavian venous (SCV) thrombosis (and stenosis) with SCV catheterization has been well established (1,2,4–10). Whether thrombosis precedes stenosis or vice versa is unclear. It has been demonstrated in patients undergoing dialysis that thrombosis and stenosis can be reduced when the right internal jugular vein (IJV) approach is used for hemodialysis catheters (9,10). There is a paucity of literature, however, in which the preferred route of nonhemodialysis tunneled indwelling catheters is discussed, and the limited information that is available is controversial. Henriques et al (11) found that axillary venous thrombosis was significantly more prevalent in catheters placed surgically via the SCV than in those placed via the IJV. In contradistinction, Nazarian et al (12) reported results from a randomized study showing that left IJV catheters were associated with an increased incidence of venous thrombosis compared with left SCV catheters. To date, no definitive conclusions can be drawn from the literature about the preferred access site for tunneled infusion catheters. The purpose of this study was to compare the incidence of symptomatic central venous thrombosis after catheterization of the SCV or IJV performed with imaging-guided percutaneous techniques.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
SCV catheters were placed as described previously (13), with fluoroscopic guidance in all cases except two. The puncture site was the axillosubclavian junction, and a 21-gauge needle system was used (Micropuncture; Cook, Bloomington, Ind). Ionic contrast material (diatrizoate meglumine, Hypaque 50; Nycomed, Princeton, NJ) was used for venographic guidance unless there was a contraindication to its use, in which case nonionic contrast material was used (iohexol, Omnipaque 300; Nycomed). Catheter tips were always placed at or just below the caval-atrial junction, as determined after the procedure by means of an erect chest radiograph obtained during full inspiration. IJV catheters were placed with ultrasonographic (US) guidance, as previously described (14). The procedure was otherwise identical to SCV catheter placement. All procedures were performed or supervised by one of several staff interventional radiologists (including S.O.T., M.S.J., H.S.) with extensive experience (2–6 years) in tunneled catheter placement.
The catheters were silicone rubber infusion catheters in single- (9.6-F), dual- (10–12-F) and triple- (12.5-F) lumen configurations (Hickman catheters; Bard Access Systems, Salt Lake City, Utah). Catheter care was provided by the oncology nursing service and did not differ between the SCV and IJV access periods. The standard catheter-flushing regimen was 10 U/mL heparin, 2.5 mL per lumen, twice weekly and after each use. No systematic anticoagulation protocol (eg, "minidose" warfarin) was in place at any time during the study. We did not specifically determine whether individual patients were being treated with anticoagulation therapy. Low-dose urokinase (Open Cath; Abbott, North Chicago, Ill) was used as needed to maintain catheter function. Data were not collected on the use of low-dose urokinase. The policy for this agent (which has clear-cut indications for use) did not change during the study.
Patients suspected of having venous thrombosis were evaluated with US, venography, or both, as clinically indicated (eg, by arm swelling or pain). US was performed by members of the abdominal radiology section, who used the standard clinical protocol to evaluate the upper arm vein, axillary vein, and SCV as far centrally as possible. Venography was performed by interventional radiologists (including S.O.T., M.S.J., H.S.), with manual injection of contrast material through peripheral intravenous sites. Venographic images were recorded with digital subtraction over the upper arm and chest to include the upper arm veins, axillosubclavian veins, and superior vena cava. When negative US findings were obtained, venography was always performed. The results of these examinations were collected prospectively as part of our quality assurance program and were tabulated in our database. Management of symptomatic venous thrombosis varied to some extent with the referring physician, but whenever possible the catheter was left in place and anticoagulation was begun. If the patient became asymptomatic, the catheter was left in place and anticoagulation was continued. If symptoms persisted or if the referring physician insisted on catheter removal at initial diagnosis of symptomatic thrombosis, the catheter was removed and anticoagulation was continued.
Patients who left our system were contacted by telephone to determine whether symptomatic venous thrombosis had occurred at any time during catheterization. Only patients in whom complete follow-up was obtained were included in the study. Complete follow-up was obtained for 279 catheters in 238 patients. That we were able to obtain complete follow-up in only 36% of cases (279 of 774) reflects the fact that ours is a tertiary-care institution.
The proportion of patients or catheters with infection or thrombosis were compared by using the Fisher exact test, and the odds ratios were calculated as a summary of the differences. The Fisher exact test was also used to compare thrombosis rates by sex and by indications for catheter placement. The Cochran-Mantel-Haenszel test and logistic regression were used to adjust for side of placement and size of catheter. Catheter dwell times were compared by using the Wilcoxon rank sum test, and catheter survival curves were compared by using the Wilcoxon log rank test. Kaplan-Meier estimates of the survival curves were plotted.
Two analyses were performed to determine whether inclusion of patients with multiple catheters might bias the result. In the first analysis, only the initial catheter placed in each of the 238 patients was evaluated. In the second analysis, all 279 catheters were evaluated and assumed to be independent, regardless of whether patients had multiple catheters at one or more sites. Statistical analysis was performed with the SAS software package (version 6.12; SAS Institute, Cary, NC).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
|
There were 21 patients with thromboses: seven (7.5%) of 93 with a catheter on the left side and 14 (9.7%) of 145 with a catheter on the right side (P = .645). Thus, there was no evidence that symptomatic thromboses differed by side. The mean time to thrombosis was 36 days for SCV and 142 days for IJV catheters. Eight (5.8%) of 137 subjects with a small catheter (9.6 or 10 F) had thromboses, compared with 13 (12.9%) of 101 with larger catheters. Although the larger catheters were associated with a higher incidence of thrombosis, this difference did not reach statistical significance in this analysis (P = .067, odds ratio = 2.382). There was however, a significant difference in thrombosis rate between SCV and IJV catheters. Eighteen (12.6%) of 143 subjects with an SCV catheter had thrombosis, compared with only three (3.2%) of 95 with IJV catheters (Fisher exact test, P = .018; odds ratio = 0.226). There were 0.12 thromboses per 100 catheter days in patients with an SCV catheter and 0.04 thromboses per 100 catheter days in those with an IJV catheter. With multiple logistic regression, the odds of developing thrombosis were estimated to be 20.8% as large for IJV as for SCV placement (P = .025). There was no sex difference in the thrombosis rate (9.8% for men [11 of 112] and 7.9% for women [10 of 126], P = .65).
Second Analysis
When patients with multiple catheters were included, there were 166 SCV catheters and 113 IJV catheters. The mean dwell time for the SCV catheters was 100.9 days, compared with 84.9 days for the IJV catheters (P = .067). Not unexpectedly, these results were similar to those of the above analysis.
There were 72 infections, in 27 (25.2%) of 107 catheters on the left and 45 (26.2%) of 172 of those on the right. Infection was present in 42 (26.1%) of 161 cases of 9.6- or 10-F catheter placement and in 30 (25.4%) of 118 12- or 12.5-F cases of catheter placement. There was no significant difference in the incidence of infection by side (P = .889) or by catheter size (P > .99). Infection was present in 39 (23.5%) of 166 cases of SCV catheter placement and in 33 (29.2%) of 113 cases of IJV catheter placement (P = .33). There were 0.23 infections per 100 catheter days for SCV catheters and 0.34 per 100 catheter days for IJV catheters. Again, with the Cochran-Mantel-Haenszel test used to adjust for side (right or left) and size of catheter (10 or 12 F), there was no evidence of a difference in infection rates between catheters placed in the SCV and IJV (P = .165).
Thrombosis was present in seven (6.5%) of 107 cases of left-sided catheter and in 16 (9.3%) of 172 cases of right-sided catheters (P = .505). The incidence of thrombosis was more closely related to catheter size in this second analysis. Eight (5.0%) of 161 patients with 9.6- or 10-F catheters experienced symptomatic thrombosis, compared with 15 (12.7%) of 118 with 12- or 12.5-F catheters (P = .027, odds ratio = 2.785). Twenty (12.1%) of the 166 patients with SCV catheters had thrombosis, compared with three (2.7%) of 113 with IJV catheters (P = .007, odds ratio = 0.199). There were 0.12 thromboses per 100 catheter days for SCV catheters and 0.03 per 100 catheter days for IJV catheters. With multiple logistic regression to adjust for side of placement and size of catheter, the odds of developing thrombosis were estimated to be 18.9% as large for IJV as for SCV placement (P = .015). The indication for catheter placement did not affect the incidence of thrombosis (P = .29; antibiotics, 2.3% [one of 44]; malignancy, 9.8% [20 of 204]; total parenteral nutrition, 0% [zero of 12]; and miscellaneous, 10.5% [two of 19]).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
More attention has been paid to catheter insertion site in the nephrology literature than in the oncology literature, possibly because the sequelae of catheter-induced venous thrombosis in that population are more dramatic. In patients with an ipsilateral functioning hemodialysis graft or fistula, central venous thrombosis and stenosis may result in severe arm swelling. Conversely, in the oncology population, venous thrombosis is occult in the majority of patients and, even when symptomatic, usually responds to anticoagulant therapy and/or catheter removal (1). In the hemodialysis literature, the IJV route is well established as the preferred access site for hemodialysis (18). This practice resulted from many different studies and reports implicating SCV access in central venous thrombosis and stenosis, but two studies in particular are cited most commonly. Both demonstrated a decreased incidence of venous stenosis with IJV vein approach. Cimochowski et al (9) reported a 50% incidence of SCV stenosis at venographic follow-up after placement of a temporary dialysis catheter. No detectable venous stenoses were noted after placement via the IJV (9). Similarly, Schillinger et al (10) demonstrated a 42% incidence of stenosis in patients with a catheter placed via the SCV, with only a 10% incidence in the IJV cohort.
Cimochowski et al (9) theorized that the increased incidence of stenosis after SCV catheterization was multifactorial. They hypothesized that stricture formation was encouraged by the site of SCV puncture coupled with the torquing of the stiff catheter over the first portion of the SCV as it crosses the first rib, with aggravation by the to-and-fro motion of the cardiac cycle. In contradistinction, catheters placed via the right IJV route assume a rather straight course to the heart from the site of insertion and, as such, do not subject the vein to trauma while in place.
Although our results do not show as dramatic a difference as those of Schillinger et al and Cimochowski et al, we examined only symptomatic thrombosis. Had we considered asymptomatic thrombosis, the differences might be have been much greater, because asymptomatic thrombosis is more common than symptomatic thrombosis in those studies in which it has been evaluated. Haire et al (3), in a venographic study of SCV catheters, found that 22 (63%) of 35 patients had thrombosis, but only three (9%) were symptomatic. Similarly, DeCicco et al (2) found that 63 (66%) of 95 patients with an SCV catheter had asymptomatic thrombosis, whereas only four (4%) were symptomatic.
Although there is growing understanding that the SCV route is commonly associated with venous thrombosis in the oncology population, there appears to be little acknowledgment of the superiority of the IJV route in these patients. Although Henriques et al (11) reported more than 6 years ago that the incidence of symptomatic venous thrombosis was lower with surgically placed external jugular vein (2.3%) or IJV (0%) catheters than with SCV (10.3%) catheters, the SCV route continues to be used commonly, particularly in the surgical community. This is evidenced not only by results from recent series (2,19) in which the SCV route was used but by local practices as well: At our institution, the SCV route continues to be the preferred access for surgeons, whether for nontunneled or for tunneled catheters. In fact, there is an interesting disparity between the growing body of literature describing radiologically placed catheters (most placed via the IJV) and the surgical access literature (the majority placed via the SCV).
Despite the trend away from the SCV route in the radiologic community, to our knowledge there are no published reports in which the percutaneous IJV and SCV routes were compared in the nonhemodialysis population. Henriques et al (11) compared the percutaneous SCV approach with IJV or external jugular vein cutdown. Not surprisingly, in addition to lower thrombosis rates, the jugular vein routes were associated with fewer pneumothoraces (0% for jugular vs 5.6% for SCV , P < .01). Watt et al (20), in an abstract presented in 1999, reported a significantly lower rate of symptomatic venous thrombosis with IJV versus SCV catheters in a nonrandomized study; these catheters were all placed percutaneously with imaging guidance.
Conversely, in 1997, Nazarian et al (12) described in an abstract a small randomized study in which left SCV catheter placement was compared with left IJV catheter placement. Their results showed an increase in the incidence of venous thrombosis detected with US (presumably clinically occult) in catheters placed via the left IJV. Their results also demonstrated poorer patient tolerance of left-sided IJV catheters and a greater incidence of infection. The results of our study support the contention of Henriques et al (11) and Watt et al (20) that the IJV is superior to the SCV as the access route in the nonhemodialysis population. It is unclear why Nazarian et al reached a different conclusion. One could argue that the left SCV route is less prone to thrombosis than the right because of the smooth arcing course toward the right atrium, while the left IJV route is more prone to thrombosis than the right because of the two turns needed to reach the right atrium. This could conceivably have biased the study in favor of the SCV route. On the other hand, in surgical series (2,17,21) in which the left and right SCVs have been examined separately, the thrombosis rate has been higher on the left than the right, although this was not found in our study.
The increased infection risk with the IJV reported by Nazarian et al (12) was not found in our study either. We did not study patient tolerance specifically, but we have not encountered complaints or problems reported from patients concerning catheter location. Since Nazarian et al have not published their study in a research article, we can only speculate as to what was meant by "patient tolerance." Although the catheter crosses the clavicle with the IJV approach, we have not encountered a patient in whom the approach could not be used due to cachexia, nor do we experience more difficult catheter placement in such patients. Although this factor was not specifically studied, another advantage to the IJV route is that in obese patients there is considerably less catheter retraction when the patient stands than occurs with the SCV route.
Our study did not address asymptomatic thrombosis of the access vein, and it is possible that the incidences of access-site thrombosis with IJV and SCV catheterization are similar but that IJV venous thrombosis is better tolerated by the patient and less likely to manifest clinically. This contention is not supported by the literature, however; Agraharkar et al (22) studied IJV thrombosis following dialysis catheter placement and found the incidence to be only 2%. It could be higher, however, in an oncology population.
Other limitations of our study are inherent in its retrospective nature. A prospective randomized trial might yield more precise data, but, given our results, we do not believe that such a study would be ethical. We reviewed only a small percentage of the catheters placed at our institution because of the limited follow-up available. We believe this reflects the tertiary nature of our practice. It is possible, however, that the incomplete follow-up harbors a selection bias toward higher rates of thrombosis. Even so, the relative rates of thrombosis should be unaffected by limited follow-up; moreover, the symptomatic thrombosis rates for SCV catheters are similar to those reported in other series.
Another limitation to our study was the difference in length of follow-up between the two groups. We obtained longer follow-up for patients with SCV catheters simply because this was the route typically used earlier in our series. Some may attribute the higher rates of thrombosis in the SCV group to the longer follow-up. In our cohort, however, the mean time to manifestation of symptomatic thrombosis in the SCV group was 36 days, which indicates that most SCV thromboses occurred early and that the slightly longer follow-up likely did not influence the observed difference. Therefore, we do not believe that differences in length of follow-up introduced a substantial bias into the results. As noted earlier, nearly all catheters placed via the SCV route required the administration of contrast material to guide venous puncture. It may be theorized that the use of contrast could induce a low-grade phlebitis, which may contribute to venous thrombosis. The causal relationship between contrast material and thrombophlebitis in the upper extremity is unknown, to our knowledge, and the symptomatic thrombosis rate seen was similar to that in surgical series (2,17,21) in which no contrast material was used. It is nonetheless possible that US-guided SCV puncture might yield a lower thrombosis rate. The heterogeneous nature of the patient population may limit the conclusions that can be drawn with regard to individual subpopulations, such as patients in whom catheters were placed for total parenteral nutrition. While we acknowledge this limitation, we believe the overall results are compelling enough that we choose the IJV as our primary route in all patients, regardless of indication.
The IJV route of catheterization with US guidance has many advantages. The lack of intravenous contrast material is an advantage in itself. First, there is no risk of contrast material reaction. Second, the potential role of contrast material–induced phlebitis is eliminated. Third, the use of contrast material to opacify the venous system necessitates the use of fluoroscopy, thereby increasing ionizing radiation to both the patient and the operator.
In our experience the risks of pneumothorax and infection did not differ significantly between sites of insertion, because these rates are so low that detecting differences in them would require much larger numbers of patients. Nonetheless, there was one pneumothorax in the SCV group and none in the IJV group. Whether in radiologic or surgical series, the incidence of pneumothorax from SCV puncture is consistently reported to be in the range of 2% or higher (11,13,19,23). Conversely, pneumothorax is rare in radiologic or surgical series when the IJV approach is used (11,14). In addition to these established advantages, we believe we have provided compelling evidence of the decreased incidence of venous thrombosis when IJV access is used. Therefore, we conclude that the IJV route is superior to the SCV route for placement of a tunneled central venous infusion catheter because of the decreased incidence of symptomatic venous thrombosis. Indeed, the IJV is now our access site of first choice, and we resort to SCV puncture only when neither IJV is suitable for catheterization.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Abbreviations: IJV = internal jugular vein, SCV = subclavian vein
Author contributions: Guarantor of integrity of entire study, S.O.T.; study concepts, S.O.T., J.K.F.; study design, S.O.T., J.K.F., W.T.A.; definition of intellectual content, S.O.T., J.K.F.; literature research, J.K.F., S.O.T.; data acquisition, S.O.T., J.K.F., P.H.K., H.S., M.S.J.; data analysis, J.K.F., S.O.T., W.T.A.; statistical analysis, W.T.A.; manuscript preparation, J.K.F., S.O.T.; manuscript editing and review, J.K.F., S.O.T., M.S.J., H.S., P.H.K.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  V. Kakzanov, P. Monagle, and A. K. C. Chan Thromboembolism in Infants and Children With Gastrointestinal Failure Receiving Long-Term Parenteral Nutrition JPEN J Parenter Enteral Nutr, January 1, 2008; 32(1): 88 - 93. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Dubois MD MSc, F. Rypens MD, L. Garel MD, M. David MD, J. Lacroix MD, and F. Gauvin MD MSc Incidence of deep vein thrombosis related to peripherally inserted central catheters in children and adolescents Can. Med. Assoc. J., November 6, 2007; 177(10): 1185 - 1190. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. J. Sands Vascular Access in the Adult Home Infusion Patient JPEN J Parenter Enteral Nutr, January 1, 2006; 30(1_suppl): S57 - S64. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Verso and G. Agnelli Venous Thromboembolism Associated With Long-Term Use of Central Venous Catheters in Cancer Patients J. Clin. Oncol., October 1, 2003; 21(19): 3665 - 3675. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. A. Berlin, L. S. Perrin, G. D. Soltes, M. H. Barth, J. O. Roehm, K. Ouriel, P. Dewachter, C. Mouton-Faivre, P.-M. Mertes, F. A. Beer, et al. Preventing Complications of Central Venous Catheterization N. Engl. J. Med., June 26, 2003; 348(26): 2684 - 2686. [Full Text] [PDF]
|
|
|
|
 |
|
 C. Male, P. Chait, M. Andrew, K. Hanna, J. Julian, and L. Mitchell Central venous line-related thrombosis in children: association with central venous line location and insertion technique Blood, June 1, 2003; 101(11): 4273 - 4278. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. P. O'Grady, M. Alexander, E. P. Dellinger, J. L. Gerberding, S. O. Heard, D. G. Maki, H. Masur, R. D. McCormick, L. A. Mermel, M. L. Pearson, et al. Guidelines for the Prevention of Intravascular Catheter-Related Infections Pediatrics, November 1, 2002; 110(5): e51 - 51. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Funaki Central Venous Access: A Primer for the Diagnostic Radiologist Am. J. Roentgenol., August 1, 2002; 179(2): 309 - 318. [Full Text] [PDF]
|
|
|
|
 |
|
 M. Hata, A. Rosalion, S. Seevanayagam, K. Kohch, and B. F. Buxton Upper extremity deep vein thrombosis after radial artery harvesting Ann. Thorac. Surg., April 1, 2002; 73(4): 1316 - 1317. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
