- Email: [email protected]
Characterization of central venous catheter–associated deep venous thrombosis in infants
Journal of Pediatric Surgery (2012) 47, 1159–1166
www.elsevier.com/locate/jpedsurg
Characterization of central venous catheter–associated deep venous thrombosis in infants Brian W. Gray a , Raquel Gonzalez b , Kavita S. Warrier c , Lauren A. Stephens a , Robert A. Drongowski a , Steven W. Pipe c , George B. Mychaliska a,⁎ a
Section of Pediatric Surgery, C.S. Mott Children's Hospital, University of Michigan Health System, Ann Arbor, MI Division of Pediatric Surgery, Children's Hospital of Michigan, Detroit, MI c Department of Pediatrics, C.S. Mott Children's Hospital, University of Michigan Health System, Ann Arbor, MI b
Received 25 February 2012; accepted 6 March 2012
Key words: DVT; Deep venous thrombosis; Central venous catheter; Infant
Abstract Purpose: Deep venous thrombosis (DVT) is a frequent complication in infants with central venous catheters (CVCs). We performed this study to identify risk factors and risk-reduction strategies of CVCassociated DVT in infants. Methods: Infants younger than 1 year who had a CVC placed at our center from 2005 to 2009 were reviewed. Patients with ultrasonically diagnosed DVT were compared to those without radiographic evidence. Results: Of 333 patients, 47% (155/333) had femoral, 33% (111/333) had jugular, and 19% (64/333) had subclavian CVCs. Deep venous thromboses occurred in 18% (60/333) of patients. Sixty percent (36/60) of DVTs were in femoral veins. Femoral CVCs were associated with greater DVT rates (27%; 42/155) than jugular (11%; 12/111) or subclavian CVCs (9%; 6/64; P b .01). There was a 16% DVT rate in those with saphenofemoral Broviac CVCs vs 83% (20/24) in those with percutaneous femoral lines (P b .01). Multilumen CVCs had higher DVT rates than did single-lumen CVCs (54% vs 6%, P b .01), and mean catheter days before DVT diagnosis was shorter for percutaneous lines than Broviacs (13 ± 17 days vs 30 ± 37 days, P = .02). Patients with +DVT had longer length of stay (86 ± 88 days vs 48 ± 48 days, P b .01) and higher percentage of intensive care unit admission (82% vs 70%, P = .02). Conclusions: Deep venous thrombosis reduction strategies in infants with CVCs include avoiding percutaneous femoral and multilumen CVCs, screening percutaneous lines, and early catheter removal. © 2012 Elsevier Inc. All rights reserved.
The incidence of deep venous thrombosis (DVT) is on the rise in pediatric hospitals. Raffini and colleagues [1] reported that the annual rate of DVT in a large cohort of children's ⁎ Corresponding author. C.S. Mott Children's Hospital, Section of Pediatric Surgery, 1500 E Medical Center Dr, Mott F3970, Ann Arbor, MI 48109. Tel.: +1 734 764 4151, fax: +1 734 936 9784. E-mail address: [email protected] (G.B. Mychaliska). 0022-3468/$ – see front matter © 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.jpedsurg.2012.03.043
hospitals increased by 70% from 2003 to 2007. One of the greatest risk factors for DVT in children is the presence of a central venous catheter (CVC) [2-4]. Infants, in particular, have been found to be at greater risk for CVC-related DVTs [5,6]. Proposed risk factors for DVT in infants with a CVC include sepsis, small vessel size, genetic thrombophilia, acquired thrombophilia from conditions such as congenital heart disease and renal failure, duration of CVC placement,
1160 long-term infusion of TPN, major surgery, and malignancy, but a standardized risk stratification protocol has never been developed for infants [7-9]. In addition to potential complications such as pain, edema, loss of catheter patency, pulmonary embolus, and postthrombotic syndrome, infants are at increased bleeding risk from the anticoagulants used to prevent and treat DVTs [9-11]. We performed this study to characterize CVC-associated DVT in infants younger than 1 year at our institution and to propose DVT riskreduction strategies.
1. Methods We performed a retrospective review of all infants younger than 1 year who had a CVC placed at the University of Michigan C.S. Mott Children's Hospital between 2005 and 2009. This study was conducted under approved institutional review board protocol no. 28442. Central venous catheters were inserted in the operating room or the intensive care unit (ICU) by pediatric surgery, pediatric cardiac surgery, pediatric intensive care, neonatal intensive care, or peripherally inserted central catheter (PICC) teams. Catheters inserted into external jugular, internal jugular, or facial veins were defined as “jugular” CVCs. Catheters placed into saphenous or femoral veins were defined as “femoral” CVCs, and subclavian catheters were defined as “subclavian.” In the femoral location, percutaneous lines were placed directly into the femoral vein, whereas tunneled Broviacs were saphenofemoral. No patients were given prophylactic anticoagulation while CVCs were in place, and infants being treated with therapeutic anticoagulation at the time of CVC placement were excluded from the study. Umbilical vein and lumbar vein catheters were also excluded. Patients were divided into those with ultrasonically diagnosed DVT (+DVT) and those without radiographic evidence of DVT (−DVT). Ultrasounds (USs) were not performed on all patients prospectively. Diagnostic USs were performed only in patients who exhibited signs and symptoms of DVT, such as limb edema or erythema, or in cases where possible DVTs were incidentally noted on other imaging modalities [12]. Patients who were not studied with US were assumed to not have a DVT. In −DVT patients, the control CVC was defined as the catheter placed nearest to 1 year of age. Statistical analyses were performed using the t test for equality of means, analysis of variance, the Pearson χ2 test, and logistic regression using the SPSS software package (IBM, Chicago, IL). Significance was defined as P b .05. All values are reported as the mean ± SD, unless otherwise indicated.
2. Results There were a total of 333 patients included in our study, and 1-year survival was 78%. Most infants were admitted to
B.W. Gray et al. an ICU (Table 1). Nearly 40% of the patients were born prematurely, and 60% had at least 1 congenital malformation including congenital lung or diaphragmatic abnormalities (20%) and congenital heart defects (22%). In total, this study encompassed more than 29,000 catheter days. Overall, 613 CVCs were placed, with 65% being tunneled Broviac CVCs, 13% PICC lines, and 17% percutaneously inserted CVCs. Data on the type of CVC were missing in 5% of lines. Patients had as few as 1 and as many as 6 CVCs. One CVC was counted in each patient as a DVT or control catheter. Of the 333 study catheters, 155 (47%) were placed in a femoral vein, 111 (33%) in a jugular vein, and 64 (19%) in a subclavian vein. Diagnostic US was performed in 35% (117/333) of patients. Signs that led to diagnostic US were limb edema in 77% of cases, erythema in 2%, edema + erythema in 8%, and possible DVT found on other imaging in 11%. Deep venous thromboses were diagnosed in 60 (51%) of 117 patients who received an US, leading to an 18% (60/333) overall DVT incidence. Deep venous thromboses were found in 10% of patients with Broviacs, 28% of those with PICC lines, and 60% of infants with percutaneous CVCs. This difference was significant on univariate testing (P b .01). In total, nontunneled percutaneous CVCs were in place in 47% (28/60) of the +DVT patients. Lastly, the 2 groups had similar numbers of catheters per patients: 1.78 +DVT and 1.73 −DVT.
Table 1
Patient characteristics
Total no. of patients +DVT −DVT Survival at 1 y of age M/F Race White Black Hispanic Asian Admission location ICU Floor Common conditions Prematurity Congenital malformations Sepsis Surgery with CVC in place Gastrointestinal diagnosis Thoracic diagnosis Total catheters Broviac PICC Percutaneous CVC Total catheter days
n
%
333 60 273 260 186/147
18 82 78 56/44
235 59 10 5
70 18 3 2
224 85
67 25
129 199 67 194 175 144 613 398 80 102 29,000
39 60 20 58 52 43 65 13 17
Characterization of CVC–associated DVT in infants In both patients with +DVT and patients with −DVT, 1year survival was 78% (Table 2B). However, infants who developed a DVT had a significantly longer mean length of stay (85.8 days vs 48.4 days) on both univariate and multivariate analyses (P b .01; Table 2A). In addition, the +DVT group had a significantly higher rate of ICU admission (85% vs 70%, P = .02) and higher incidence of medically induced paralysis (17% vs 0.4%, P b .01; Table 2B). The mean and median total catheter days, including all catheters, were significantly greater in the −DVT group on both univariate and multivariate analyses: mean, 50.9 ± 56.7 days vs 80.1 ± 103.6 days; median, 29 days vs 47 days (P = .036 on univariate, P b .01 on logistic regression). The 2 groups had similar ventilator days and mean number of operations per patient (Table 2A). In terms of comorbid conditions, +DVT had higher incidence of meningitis (5% vs 0%, P b .01), but the overall sepsis/infection rate was similar (32% +DVT, 25% −DVT; P = .33). In addition, 5% of the +DVT group had perinatal asphyxia compared with 0.4% of the −DVT group (P = .02). Femoral DVTs encompassed 60% of all diagnosed DVTs, and jugular DVTs accounted for the other 40% (Table 3; P = .04). There were no subclavian DVTs found. Femoral CVCs were associated with higher DVT rates than other catheter sites (27% femoral, 11% jugular, 9% subclavian; P b .01; Fig.). Subclavian CVCs were in place in 4 patients who
Table 2
Comparison of +DVT and −DVT groups +DVT (n = 60) −DVT (n = 273) P
A. Length of stay (d) Vent days EGA at birth (wk) Birth weight (kg) Total catheter days (all catheters) Total operations B. Survival to 1 y, n (%) ICU admission, n (%) Medically induced paralysis, n (%)
85.8 ± 87.6
48.4 ± 48.5
b.01 ⁎, ⁎⁎
28.5 ± 46.5 35.4 ± 5.0
23.1 ± 34.8 33.6 ± 5.2
.32 .015 ⁎
2.5 ± 0.9
2.3 ± 1.1
50.9 ± 56.7
80.1 ± 103.6
2.5 ± 1.7
2.1 ± 1.6
47 (78)
212 (78)
49 (85)
174 (70)
.38 .036 ⁎, ⁎⁎
.14
1161 Table 3
CVC site vs DVT location
DVT location
CVC site
Total
Jugular
Femoral
Subclavian
Jugular Femoral Subclavian Total
10 2 0 12 (20%)
10 32 0 42 (70%)
4 2 0 6 (10%)
24 (40%) 36 (60%) ⁎ 0 60
Significantly more DVTs were found in femoral veins (36/60) than jugular veins. No DVTs were found in subclavian veins, although 6 subclavian CVCs were associated with DVTs in other sites. Femoral CVCs were associated with 70% of DVTs. ⁎ P = .04.
developed jugular DVTs and 2 patients who had femoral thrombi (Table 3). In total, 30% (18/60) of DVTs occurred in locations other than the concurrent CVC site. In 3 of the 18 cases, patients had a history of CVC placement in the eventual DVT location. In 9 cases, the DVT did not correlate with any previous CVCs, and in 6 cases, there were missing data. Overall, femoral CVC location was associated with 70% of the DVTs in our patients (Table 3). When analyzing just the femoral location, patients with saphenofemoral Broviac lines had a 16% DVT rate, and those with nontunneled percutaneous femoral CVCs had an 83% (20/24) DVT rate (P b .01). The number of catheter lumens also correlated with incidence of DVT. Single-lumen CVCs had a 6% DVT rate, while multiple-lumen CVCs had a significantly higher incidence at 54% (P b .01, Table 4). Within the multiplelumen group, double-lumen CVCs had a 48% DVT rate and triple-lumen catheters had an 82% incidence (9/11). Singlelumen CVCs had a mean size of 3.8 ± 0.9 French (Fr), while multilumen CVCs were a mean of 4.9 ± 1.4 Fr (P b .01). In addition, all multilumen CVCs were placed percutaneously. The mean number of catheter days from CVC placement to DVT diagnosis was 20.8 ± 28.5 days for the entire group. Days to DVT did not significantly differ between the CVC placement sites, with 19.8 ± 22.5 for jugular, 20.5 ± 31.7 for femoral, and 25.2 ± 15.0 for subclavian catheters. However, those with percutaneous lines had a significantly shorter time to DVT diagnosis than those with Broviacs: 12.9 ± 16.6 days vs 30.3 ± 36.9 days (P = .02).
1.0 .02 ⁎
Table 4
CVC lumen number and DVT incidence N
10 (17%)
1 (0.4%)
b.01 ⁎
Values for +DVT and −DVT groups are presented as mean ± SD, unless otherwise indicated. Abbreviation: EGA, estimated gestational age. ⁎ Significance (P b .05) on univariate analysis. ⁎⁎ Significance (P b .05) on logistic regression.
Single lumen +DVT −DVT Multiple lumens +DVT −DVT
180 10 170 65 35 30
Multilumen CVCs had a significantly higher incidence of DVT. ⁎ P b .01.
% 6 95 54 ⁎ 46
1162
Fig. Central venous catheter site in 333 patients and incidence of CVC-associated DVT. Femoral CVCs had a significantly higher DVT association (⁎42/155, P b .01) than jugular (12/111) and subclavian (6/64) CVCs.
Deep venous thrombosis treatment consisted of catheter removal in all patients, low-molecular-weight heparin in 67% of cases, and unfractionated heparin in 12%. No pharmacologic treatment was given in 21% because of elevated bleeding risk. The mean duration of treatment was 8.3 ± 8.5 weeks. No patients received thrombolytics. Five patients had a negative hypercoagulability workup, and 1 patient with +DVT had a familial history of hypercoagulability. Long-term complications consisted of recurrent thrombus in 2 patients and persistent limb edema in 1 infant.
3. Discussion The increasing incidence of CVC-related DVT remains an unsolved clinical problem. The aim of this study was to characterize CVC-associated DVT in infants younger than 1 year at a single institution by identifying risk factors for CVC-related DVT, describing DVT management and outcomes, and proposing DVT risk-reduction strategies. In our cohort, 333 infants younger than 1 year had an 18% incidence of DVT overall, with most of these related to femoral CVCs. The presence of DVT also correlated with a longer hospital length of stay, ICU admission, medically induced paralysis, percutaneous CVCs, and the use of multilumen CVCs. The DVT group had similar 1-year survival to controls and a low incidence of long-term morbidity. Central venous catheters are commonly placed in hospitalized infants. The procedural complications include bleeding, arterial puncture, pneumothorax, and malposition. Long-term complications include infection, migration, mechanical malfunction, catheter thrombosis, and DVT
B.W. Gray et al. [8,14]. In particular, the incidence of CVC-related DVT in infants ranges from 5% to 50% [13,15,16]. Furthermore, evidence suggests that a substantial amount of patients with CVCs may have a DVT but remain asymptomatic [17,18]. Therefore, the true incidence of CVC-related DVT may be higher than most studies report. Several studies have investigated risk factors for infants and children receiving CVCs. The Canadian Childhood Thrombophilia Registry was able to identify conditions associated with DVT in the presence of a CVC [4,9]. Raffini et al [1] found that most children who developed DVTs had at least 1 complex comorbid medical condition. In another study, Revel-Vilk and Ergaz [7] suggested that risk factors for central line–related thrombosis can be categorized into those related to the catheter, the individual, and the underlying disease and its treatment. Although each of these investigators described the risks associated with DVT in children and infants in their cohort, there has never been a standardized risk stratification protocol developed for DVT in these patients. Similar to prior studies, most of the children in our study were critically ill. Perhaps, because patients requiring central access were particularly ill at baseline, there were few characteristics distinguishing the +DVT group from the −DVT group. Infants with +DVT had a later estimated gestational age and similar birth weight, ventilator days, and operations per patient. In addition, they had similar rates of malignancy, sepsis/infection, congenital malformations, and congenital heart disease. However, there were several notable distinctions between the groups. The +DVT group had a significantly longer mean length of hospital stay by almost 40 days, a higher percentage of ICU patients, and greater rates of medical paralysis. This is an indication that sicker, more sedentary patients may be more prone to DVT. Sandoval and colleagues [16] and Spyropoulos et al [20] found that children who developed DVTs while inpatients had long ICU stays, and the IMPROVE study group identified immobility of greater than 7 days as an independent risk factor for DVT in adults. Although relatively rare, we also observed higher rates of meningitis and perinatal asphyxia in the +DVT group. These factors were identified by van Ommen and colleagues [3] as top risk factors after CVC use for DVT in neonates. In support of these observations, there is a growing volume of literature relating systemic and local inflammation with formation and exacerbation of venous thrombosis [21,22]. Interestingly, an increasing number of catheter days did not correlate with an increased risk for DVT. We found that the +DVT group had significantly less mean and median catheter days. Several patients with −DVT had indwelling catheters for quite long durations because of conditions such as short bowel syndrome. On the other hand, DVTassociated CVCs were removed directly after DVT diagnosis. Our observations show that complications like DVT are an impetus to readdress the need for venous access, a
Characterization of CVC–associated DVT in infants conversation often delayed for those patients who never have problems with their CVCs. Our most substantial finding was the preponderance of DVTs associated with femoral CVCs compared with jugular and subclavian vein CVCs. Interestingly, there are multiple reviews in the pediatric and adult literature citing an increased DVT rate with femoral catheters [6,23-30]. In studies that separate patients by age, smaller patients (b1 year old) are found to have a higher incidence of femoral CVC-associated DVTs than any other group [5,6]. With smaller patients, there is a greater catheter to vessel size ratio, particularly in the smaller femoral vessels of infants. This creates a state of low and turbulent venous flow around the femoral catheter that can compound upon the endothelial damage, inflammatory response, and thrombogenic response caused by the catheter [25,31,32]. In addition, femoral CVCs are preferentially inserted in urgent and emergent situations, in which patients are sicker and likely immobilized, exacerbating the low-flow state. The femoral CVCs placed in the most ill patients are percutaneous lines, and we found those to have a much higher DVT rate compared with saphenofemoral Broviacs (83% vs 16%). Even with the use of US, placement of percutaneous lines often requires multiple venipunctures to access the vein and difficulty threading the wire, which may result in more vessel trauma than the cutdown technique used to insert saphenofemoral Broviac catheters. This increased endothelial damage may increase the risk of thrombus formation. Although there were no subclavian DVTs diagnosed by US in our study, this does not indicate that subclavian CVCs are safer than all other CVCs. It is possible that there were false-negative subclavian US studies in our cohort. Several investigators have demonstrated the low sensitivity of US for evaluating the subclavian veins for DVT because of the decreased ability to compress the vessel under the clavicle [17,28,33]. In these cases, venography can be used as the “gold standard” for DVT diagnosis [34]. Surprisingly, there were 18 patients in our study who developed DVTs at a location other than the concurrent CVC site. Three of these DVTs were found to be where prior CVCs had been inserted, but there were 9 DVTs that occurred where no CVC had been placed. It is possible that these DVTs were influenced by underlying thrombophilia in these infants, such as the increasing number of one of the genetic factors described in the literature [35,36]. In addition, in comparison with older children, neonates have increased factor VII concentrations and differences in whole blood measures of hemostasis. They also have been found to have a decreased amount of several anticoagulant and fibrinolytic proteins [2,36]. Another important observation from this study is the timing of DVT diagnosis in relation to CVC presence. Patients in our study were diagnosed with DVT on a mean catheter day 21. In separate prospective studies of CVC-associated DVT in children, Beck et al [5] and Talbott et al [6] both found that DVTs started to develop within the first week and as early as
1163 the first day of after CVC insertion. However, many of these DVTs were subclinical when first diagnosed. Most patients in our study were only surveyed for DVT after symptoms presented, implying that DVTs may have started to form as early as the first day of CVC presence. Interestingly, we found that patients with percutaneous lines had DVTs diagnosed almost 2 weeks before those with Broviacs (13 days vs 30 days). We have not found similar observations in the literature, but we can assume that this is caused by similar factors related to percutaneous lines discussed previously. In addition to CVC location, we found that multilumen catheters had a significantly higher association with DVT. This finding is not substantiated in the literature, but it may be explained by the fact that multilumen catheters were significantly larger than single-lumen CVCs. In addition, multilumen CVCs placed in infants are percutaneous lines that are associated with higher DVT rates, as discussed previously. None of the patients in this study were given chemoprophylaxis against DVT. There are no standardized and validated DVT prophylaxis protocols, and in a survey of 24 pediatric ICUs, Braga and Young [11] found only 1 with a written DVT prophylaxis protocol [37]. In one institutional protocol, Raffini and colleagues use early mobility and mechanical prophylaxis measures first, saving chemoprophylaxis for the children at highest risk for DVT. The American College of Chest Physicians goes as far to recommend no chemoprophylaxis for children with CVCs because of the paucity of supporting evidence for chemoprophylaxis and the significant potential bleeding risk to patients [19]. Thus, it would seem that the aggressive use of chemoprophylaxis is unwarranted in these patients until better evidence is collected. This study has several weaknesses. As a retrospective study, we did not gather information on all CVCs and complications prospectively, hindering our ability to perform a comprehensive data analysis. In addition, because we only imaged infants with clinical signs of DVT or evidence of DVT from other studies, we likely missed several subclinical DVTs in our patient cohort. In addition, we may have underreported the incidence of subclavian DVTs because we did not verify our relatively insensitive US findings with venography. Thus, the 18% incidence we reported may not be the true incidence of CVC-related DVT in these infants. Another weakness is that this study was not designed to accurately capture all treatment-related complications such as bleeding. Finally, we did not incorporate any long-term follow-up into our analysis, other than clinic notes from postdischarge hematology visits. Therefore, although we found very low morbidity in patients with DVT, we cannot remark on any longer-term complications.
4. Conclusions This study revealed that there is a significant risk of DVT among infants with a CVC. Although infants who develop
1164 DVTs are more likely to be critically ill, there are few distinguishing characteristics compared with patients with CVCs who do not develop DVTs. Thus, a reliable risk stratification protocol would be very difficult to create for this group. On the other hand, the placement location and management of CVCs can impact DVT incidence. We found substantially higher DVT rates for femoral CVCs, percutaneous lines, and multilumen catheters. Because the routine use of prophylactic anticoagulation may not be the best option to prevent DVTs, we suggest that one may have to accept the inherent risk of CVCs while looking for other avenues to reduce the risk of DVT. These include avoiding the use of femoral percutaneous lines except out of necessity, screening percutaneous lines for DVT, switching femoral lines to other sites as soon as possible, avoiding the use of larger multilumen catheters, and reducing overall CVC days.
References [1] Raffini L, Huang YS, Witmer C, et al. Dramatic increase in venous thromboembolism in children's hospitals in the United States from 2001 to 2007. Pediatrics 2009;124:1001-8. [2] Sutor AH, Uhl M. Diagnosis of thromboembolic disease during infancy and childhood. Semin Thromb Hemost 1997;23:237-46. [3] van Ommen CH, Heijboer H, Buller HR, et al. Venous thromboembolism in childhood: a prospective two-year registry in the Netherlands. J Pediatr 2001;139:676-81. [4] Monagle P, Adams M, Mahoney M, et al. Outcome of pediatric thromboembolic disease: a report from the Canadian Childhood Thrombophilia Registry. Pediatr Res 2000;47:763-6. [5] Beck C, Dubois J, Grignon A, et al. Incidence and risk factors of catheter-related deep vein thrombosis in a pediatric intensive care unit: a prospective study. J Pediatr 1998;133:237-41. [6] Talbott GA, Winters WD, Bratton SL, et al. A prospective study of femoral catheter-related thrombosis in children. Arch Pediatr Adolesc Med 1995;149:288-91. [7] Revel-Vilk S, Ergaz Z. Diagnosis and management of central-line– associated thrombosis in newborns and infants. Semin Fetal Neonatal Med. [8] Journeycake JM, Buchanan GR. Thrombotic complications of central venous catheters in children. Curr Opin Hematol 2003;10:369-74. [9] Massicotte MP, Dix D, Monagle P, et al. Central venous catheter related thrombosis in children: analysis of the Canadian Registry of Venous Thromboembolic Complications. J Pediatr 1998;133:770-6. [10] Journeycake JM, Buchanan GR. Catheter-related deep venous thrombosis and other catheter complications in children with cancer. J Clin Oncol 2006;24:4575-80. [11] Raffini L, Trimarchi T, Beliveau J, et al. Thromboprophylaxis in a pediatric hospital: a patient-safety and quality-improvement initiative. Pediatrics 127:e1326-32. [12] Kearon C, Julian JA, Newman TE, et al. Noninvasive diagnosis of deep venous thrombosis. McMaster Diagnostic Imaging Practice Guidelines Initiative. Ann Intern Med 1998;128:663-77. [13] Salonvaara M, Riikonen P, Kekomaki R, et al. Clinically symptomatic central venous catheter-related deep venous thrombosis in newborns. Acta Paediatr 1999;88:642-6. [14] Rey C, Alvarez F, De La Rua V, et al. Mechanical complications during central venous cannulations in pediatric patients. Intensive Care Med 2009;35:1438-43. [15] Revel-Vilk S. Central venous line–related thrombosis in children. Acta Haematol 2006;115:201-6.
B.W. Gray et al. [16] Sandoval JA, Sheehan MP, Stonerock CE, et al. Incidence, risk factors, and treatment patterns for deep venous thrombosis in hospitalized children: an increasing population at risk. J Vasc Surg 2008;47: 837-43. [17] Male C, Kuhle S, Mitchell L. Diagnosis of venous thromboembolism in children. Semin Thromb Hemost 2003;29:377-90. [18] Tousovska K, Zapletal O, Skotakova J, et al. Treatment of deep venous thrombosis with low molecular weight heparin in pediatric cancer patients: safety and efficacy. Blood Coagul Fibrinolysis 2009;20: 583-9. [19] Monagle P, Chalmers E, Chan A, et al. Antithrombotic therapy in neonates and children: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines (8th Edition). Chest 2008;133:887S-968S. [20] Spyropoulos AC, Anderson Jr FA, Fitzgerald G, et al. Predictive and associative models to identify hospitalized medical patients at risk for venous thromboembolism. Chest. [21] Libby P, Simon DI. Inflammation and thrombosis: the clot thickens. Circulation 2001;103:1718-20. [22] Wakefield TW, Henke PK. The role of inflammation in early and late venous thrombosis: are there clinical implications? Semin Vasc Surg 2005;18:118-29. [23] Venkataraman ST, Thompson AE, Orr RA. Femoral vascular catheterization in critically ill infants and children. Clin Pediatr (Phila) 1997;36:311-9. [24] DeAngelis GA, McIlhenny J, Willson DF, et al. Prevalence of deep venous thrombosis in the lower extremities of children in the intensive care unit. Pediatr Radiol 1996;26:821-4. [25] Gutierrez JA, Bagatell R, Samson MP, et al. Femoral central venous catheter-associated deep venous thrombosis in children with diabetic ketoacidosis. Crit Care Med 2003;31:80-3. [26] Krafte-Jacobs B, Sivit CJ, Mejia R, et al. Catheter-related thrombosis in critically ill children: comparison of catheters with and without heparin bonding. J Pediatr 1995;126:50-4. [27] Joynt GM, Kew J, Gomersall CD, et al. Deep venous thrombosis caused by femoral venous catheters in critically ill adult patients. Chest 2000;117:178-83. [28] Male C, Julian JA, Massicotte P, et al. Significant association with location of central venous line placement and risk of venous thrombosis in children. Thromb Haemost 2005;94:516-21. [29] Mian NZ, Bayly R, Schreck DM, et al. Incidence of deep venous thrombosis associated with femoral venous catheterization. Acad Emerg Med 1997;4:1118-21. [30] Worly JM, Fortenberry JD, Hansen I, et al. Deep venous thrombosis in children with diabetic ketoacidosis and femoral central venous catheters. Pediatrics 2004;113:e57-60. [31] Chidi CC, King DR, Boles Jr ET. An ultrastructural study of the intimal injury induced by an indwelling umbilical artery catheter. J Pediatr Surg 1983;18:109-15. [32] Pottecher T, Forrler M, Picardat P, et al. Thrombogenicity of central venous catheters: prospective study of polyethylene, silicone and polyurethane catheters with phlebography or post-mortem examination. Eur J Anaesthesiol 1984;1:361-5. [33] Baarslag HJ, van Beek EJ, Koopman MM, et al. Prospective study of color duplex ultrasonography compared with contrast venography in patients suspected of having deep venous thrombosis of the upper extremities. Ann Intern Med 2002;136:865-72. [34] Baarslag HJ, Koopman MM, Reekers JA, et al. Diagnosis and management of deep vein thrombosis of the upper extremity: a review. Eur Radiol 2004;14:1263-74. [35] Wermes C, von Depka Prondzinski M, Lichtinghagen R, et al. Clinical relevance of genetic risk factors for thrombosis in paediatric oncology patients with central venous catheters. Eur J Pediatr 1999;158(Suppl 3): S143-6. [36] Saxonhouse MA, Burchfield DJ. The evaluation and management of postnatal thromboses. J Perinatol 2009;29:467-78.
Characterization of CVC–associated DVT in infants [37] Braga AJ, Young AE. Preventing venous thrombosis in critically ill children: what is the right approach? Paediatr Anaesth 21: 435-40.
Discussion Discussant: Dr Douglas Barnhart (Salt Lake City, UT): I have a question about your study design. It seemed from your abstract, it doesn't sound as if you routinely screened everyone, so my question is, who are the 333 patients? And how does that impact the way you interpret these data. I understand it's really the 333 patients that got ultrasounds. Is it you're just more likely to be suspicious of a DVT in the femoral region and therefore get an ultrasound, and find a DVT, or is it possible that femoral DVTs are more likely to be symptomatic and/or detected on physical examination and therefore you get the US? Do you have any idea of what the rate of silent DVT in the subclavian and jugular vein catheters are because if it's only driven by (if I'm right), that your study population are only those that got ultrasound, so you really don't know what the real rate of DVT is. My second question is, how do you explain the fact that your DVTs aren't where the catheters are and that's not a small number on them—and how do you account for that? Response: Dr. Gray: Those are excellent questions. To answer your first question, the 333 patients were all the patients who got central venous catheters and the patients who were studied with ultrasound were those who were symptomatic. This is a retrospective review, so if a patient was found to have edema, or erythema, or some other symptom in the ICU, they were studied. It is likely we missed some patients who had a DVT but weren't studied. Additionally, patients who had a femoral catheter did risk a higher incidence of edema because the catheters in the femoral veins have a substantially higher catheter to vein size ratio. I agree that there is some bias here that we only studied patients who had some kind of symptom or a possible DVT found on another imaging modality. You asked about patients who we looked at who had subclavian or jugular catheters. The first problem with imaging the subclavian catheter by ultrasound is that there is a higher rate of false negatives because you can't get proper compression of the vessel using an ultrasound probe under the clavicle. Hence, we probably missed some DVTs in the subclavian veins even when we looked for them, and so I think that definitely is a skewed population here. Regarding your second question, 18 of our 60 patients who had DVTs had the thrombus in a place other
1165 than at the concurrent catheter site. We looked at our data and found that 3 of the 18 actually had a previous catheter at that same exact site, so there may have been some endothelial damage there that contributed to it. However, the rest of the patients had no evidence of a catheter at the thrombus site. A possible explanation is that patients in this population may have systemic inflammation that could predispose one to DVT. Unfortunately, we did not evaluate this in our study. Discussant: Dr Michael Tirabasi (Madison, WI): I'd like to congratulate you on an excellent paper about a common problem that we all face. I had a question about your femoral group where it looked like you grouped both catheters inserted into the femoral vein via the saphenous and those directly placed in the femoral vein together. I was curious if you made any attempt to separate that data as instinctively you might imagine there may be more direct femoral vein injury with catheters that weren't fed through the saphenous vein. Response: Dr Gray: That's an interesting question. We actually initially tried to evaluate that, but when looking at the medical record, we had some difficulty distinguishing if lines were specifically placed in the saphenous or femoral vein. Hence, we had to assume that all percutaneous lines were femoral and Broviacs were saphenofemoral. I think that this is probably a weakness of the study. Discussant: Dr Steven Fishman (Boston, MA): I recognize I have a biased patient population, but I question the comment about long-term morbidity. You said there was very little long-term morbidity in those that had DVTs. How many of those were followed with subsequent ultrasounds, and how long was the follow-up or did you even contact patients later on to see if they had morbidity because sometimes these kids come back a lot longer with very significant morbidity. Response: Dr Gray: Every patient that had a DVT was followed by our hematology service, and all those patients that were discharged from the hospital were seen back in the hematology clinic. In the data that I presented, the 3% who had a recurrent DVT and 2% who had chronic edema were all followed in this clinic. The mean treatment course we had was 8.3 weeks in those patients. Discussant: Dr Cynthia Downard (Louisville, KY): This is a very interesting clinical study, and I think we all deal with this problem. Was there any separation as far as temporary versus tunnelled catheters and the catheter size and the type of placement whether it was a cut down or percutaneous placement? I think the most important question is, is a temporary multilumen femoral catheter a
1166 surrogate for a DVT in a patient who is really sick in the intensive care unit who is going to have other hypercoagulable issues? Response: Dr Gray: To answer your first question, we looked at the percentage of patients who had a percutaneous temporary catheter, and we found that 60% of the patients
B.W. Gray et al. who had a percutaneous catheter placed in the ICU, probably under unstable conditions, developed DVTs at that catheter site, which was significant. However, we couldn't draw a lot of conclusions from it because of your second point, that the patients who had these nontunnelled catheters were likely ill patients who had many other risk factors for DVT.
www.elsevier.com/locate/jpedsurg
Characterization of central venous catheter–associated deep venous thrombosis in infants Brian W. Gray a , Raquel Gonzalez b , Kavita S. Warrier c , Lauren A. Stephens a , Robert A. Drongowski a , Steven W. Pipe c , George B. Mychaliska a,⁎ a
Section of Pediatric Surgery, C.S. Mott Children's Hospital, University of Michigan Health System, Ann Arbor, MI Division of Pediatric Surgery, Children's Hospital of Michigan, Detroit, MI c Department of Pediatrics, C.S. Mott Children's Hospital, University of Michigan Health System, Ann Arbor, MI b
Received 25 February 2012; accepted 6 March 2012
Key words: DVT; Deep venous thrombosis; Central venous catheter; Infant
Abstract Purpose: Deep venous thrombosis (DVT) is a frequent complication in infants with central venous catheters (CVCs). We performed this study to identify risk factors and risk-reduction strategies of CVCassociated DVT in infants. Methods: Infants younger than 1 year who had a CVC placed at our center from 2005 to 2009 were reviewed. Patients with ultrasonically diagnosed DVT were compared to those without radiographic evidence. Results: Of 333 patients, 47% (155/333) had femoral, 33% (111/333) had jugular, and 19% (64/333) had subclavian CVCs. Deep venous thromboses occurred in 18% (60/333) of patients. Sixty percent (36/60) of DVTs were in femoral veins. Femoral CVCs were associated with greater DVT rates (27%; 42/155) than jugular (11%; 12/111) or subclavian CVCs (9%; 6/64; P b .01). There was a 16% DVT rate in those with saphenofemoral Broviac CVCs vs 83% (20/24) in those with percutaneous femoral lines (P b .01). Multilumen CVCs had higher DVT rates than did single-lumen CVCs (54% vs 6%, P b .01), and mean catheter days before DVT diagnosis was shorter for percutaneous lines than Broviacs (13 ± 17 days vs 30 ± 37 days, P = .02). Patients with +DVT had longer length of stay (86 ± 88 days vs 48 ± 48 days, P b .01) and higher percentage of intensive care unit admission (82% vs 70%, P = .02). Conclusions: Deep venous thrombosis reduction strategies in infants with CVCs include avoiding percutaneous femoral and multilumen CVCs, screening percutaneous lines, and early catheter removal. © 2012 Elsevier Inc. All rights reserved.
The incidence of deep venous thrombosis (DVT) is on the rise in pediatric hospitals. Raffini and colleagues [1] reported that the annual rate of DVT in a large cohort of children's ⁎ Corresponding author. C.S. Mott Children's Hospital, Section of Pediatric Surgery, 1500 E Medical Center Dr, Mott F3970, Ann Arbor, MI 48109. Tel.: +1 734 764 4151, fax: +1 734 936 9784. E-mail address: [email protected] (G.B. Mychaliska). 0022-3468/$ – see front matter © 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.jpedsurg.2012.03.043
hospitals increased by 70% from 2003 to 2007. One of the greatest risk factors for DVT in children is the presence of a central venous catheter (CVC) [2-4]. Infants, in particular, have been found to be at greater risk for CVC-related DVTs [5,6]. Proposed risk factors for DVT in infants with a CVC include sepsis, small vessel size, genetic thrombophilia, acquired thrombophilia from conditions such as congenital heart disease and renal failure, duration of CVC placement,
1160 long-term infusion of TPN, major surgery, and malignancy, but a standardized risk stratification protocol has never been developed for infants [7-9]. In addition to potential complications such as pain, edema, loss of catheter patency, pulmonary embolus, and postthrombotic syndrome, infants are at increased bleeding risk from the anticoagulants used to prevent and treat DVTs [9-11]. We performed this study to characterize CVC-associated DVT in infants younger than 1 year at our institution and to propose DVT riskreduction strategies.
1. Methods We performed a retrospective review of all infants younger than 1 year who had a CVC placed at the University of Michigan C.S. Mott Children's Hospital between 2005 and 2009. This study was conducted under approved institutional review board protocol no. 28442. Central venous catheters were inserted in the operating room or the intensive care unit (ICU) by pediatric surgery, pediatric cardiac surgery, pediatric intensive care, neonatal intensive care, or peripherally inserted central catheter (PICC) teams. Catheters inserted into external jugular, internal jugular, or facial veins were defined as “jugular” CVCs. Catheters placed into saphenous or femoral veins were defined as “femoral” CVCs, and subclavian catheters were defined as “subclavian.” In the femoral location, percutaneous lines were placed directly into the femoral vein, whereas tunneled Broviacs were saphenofemoral. No patients were given prophylactic anticoagulation while CVCs were in place, and infants being treated with therapeutic anticoagulation at the time of CVC placement were excluded from the study. Umbilical vein and lumbar vein catheters were also excluded. Patients were divided into those with ultrasonically diagnosed DVT (+DVT) and those without radiographic evidence of DVT (−DVT). Ultrasounds (USs) were not performed on all patients prospectively. Diagnostic USs were performed only in patients who exhibited signs and symptoms of DVT, such as limb edema or erythema, or in cases where possible DVTs were incidentally noted on other imaging modalities [12]. Patients who were not studied with US were assumed to not have a DVT. In −DVT patients, the control CVC was defined as the catheter placed nearest to 1 year of age. Statistical analyses were performed using the t test for equality of means, analysis of variance, the Pearson χ2 test, and logistic regression using the SPSS software package (IBM, Chicago, IL). Significance was defined as P b .05. All values are reported as the mean ± SD, unless otherwise indicated.
2. Results There were a total of 333 patients included in our study, and 1-year survival was 78%. Most infants were admitted to
B.W. Gray et al. an ICU (Table 1). Nearly 40% of the patients were born prematurely, and 60% had at least 1 congenital malformation including congenital lung or diaphragmatic abnormalities (20%) and congenital heart defects (22%). In total, this study encompassed more than 29,000 catheter days. Overall, 613 CVCs were placed, with 65% being tunneled Broviac CVCs, 13% PICC lines, and 17% percutaneously inserted CVCs. Data on the type of CVC were missing in 5% of lines. Patients had as few as 1 and as many as 6 CVCs. One CVC was counted in each patient as a DVT or control catheter. Of the 333 study catheters, 155 (47%) were placed in a femoral vein, 111 (33%) in a jugular vein, and 64 (19%) in a subclavian vein. Diagnostic US was performed in 35% (117/333) of patients. Signs that led to diagnostic US were limb edema in 77% of cases, erythema in 2%, edema + erythema in 8%, and possible DVT found on other imaging in 11%. Deep venous thromboses were diagnosed in 60 (51%) of 117 patients who received an US, leading to an 18% (60/333) overall DVT incidence. Deep venous thromboses were found in 10% of patients with Broviacs, 28% of those with PICC lines, and 60% of infants with percutaneous CVCs. This difference was significant on univariate testing (P b .01). In total, nontunneled percutaneous CVCs were in place in 47% (28/60) of the +DVT patients. Lastly, the 2 groups had similar numbers of catheters per patients: 1.78 +DVT and 1.73 −DVT.
Table 1
Patient characteristics
Total no. of patients +DVT −DVT Survival at 1 y of age M/F Race White Black Hispanic Asian Admission location ICU Floor Common conditions Prematurity Congenital malformations Sepsis Surgery with CVC in place Gastrointestinal diagnosis Thoracic diagnosis Total catheters Broviac PICC Percutaneous CVC Total catheter days
n
%
333 60 273 260 186/147
18 82 78 56/44
235 59 10 5
70 18 3 2
224 85
67 25
129 199 67 194 175 144 613 398 80 102 29,000
39 60 20 58 52 43 65 13 17
Characterization of CVC–associated DVT in infants In both patients with +DVT and patients with −DVT, 1year survival was 78% (Table 2B). However, infants who developed a DVT had a significantly longer mean length of stay (85.8 days vs 48.4 days) on both univariate and multivariate analyses (P b .01; Table 2A). In addition, the +DVT group had a significantly higher rate of ICU admission (85% vs 70%, P = .02) and higher incidence of medically induced paralysis (17% vs 0.4%, P b .01; Table 2B). The mean and median total catheter days, including all catheters, were significantly greater in the −DVT group on both univariate and multivariate analyses: mean, 50.9 ± 56.7 days vs 80.1 ± 103.6 days; median, 29 days vs 47 days (P = .036 on univariate, P b .01 on logistic regression). The 2 groups had similar ventilator days and mean number of operations per patient (Table 2A). In terms of comorbid conditions, +DVT had higher incidence of meningitis (5% vs 0%, P b .01), but the overall sepsis/infection rate was similar (32% +DVT, 25% −DVT; P = .33). In addition, 5% of the +DVT group had perinatal asphyxia compared with 0.4% of the −DVT group (P = .02). Femoral DVTs encompassed 60% of all diagnosed DVTs, and jugular DVTs accounted for the other 40% (Table 3; P = .04). There were no subclavian DVTs found. Femoral CVCs were associated with higher DVT rates than other catheter sites (27% femoral, 11% jugular, 9% subclavian; P b .01; Fig.). Subclavian CVCs were in place in 4 patients who
Table 2
Comparison of +DVT and −DVT groups +DVT (n = 60) −DVT (n = 273) P
A. Length of stay (d) Vent days EGA at birth (wk) Birth weight (kg) Total catheter days (all catheters) Total operations B. Survival to 1 y, n (%) ICU admission, n (%) Medically induced paralysis, n (%)
85.8 ± 87.6
48.4 ± 48.5
b.01 ⁎, ⁎⁎
28.5 ± 46.5 35.4 ± 5.0
23.1 ± 34.8 33.6 ± 5.2
.32 .015 ⁎
2.5 ± 0.9
2.3 ± 1.1
50.9 ± 56.7
80.1 ± 103.6
2.5 ± 1.7
2.1 ± 1.6
47 (78)
212 (78)
49 (85)
174 (70)
.38 .036 ⁎, ⁎⁎
.14
1161 Table 3
CVC site vs DVT location
DVT location
CVC site
Total
Jugular
Femoral
Subclavian
Jugular Femoral Subclavian Total
10 2 0 12 (20%)
10 32 0 42 (70%)
4 2 0 6 (10%)
24 (40%) 36 (60%) ⁎ 0 60
Significantly more DVTs were found in femoral veins (36/60) than jugular veins. No DVTs were found in subclavian veins, although 6 subclavian CVCs were associated with DVTs in other sites. Femoral CVCs were associated with 70% of DVTs. ⁎ P = .04.
developed jugular DVTs and 2 patients who had femoral thrombi (Table 3). In total, 30% (18/60) of DVTs occurred in locations other than the concurrent CVC site. In 3 of the 18 cases, patients had a history of CVC placement in the eventual DVT location. In 9 cases, the DVT did not correlate with any previous CVCs, and in 6 cases, there were missing data. Overall, femoral CVC location was associated with 70% of the DVTs in our patients (Table 3). When analyzing just the femoral location, patients with saphenofemoral Broviac lines had a 16% DVT rate, and those with nontunneled percutaneous femoral CVCs had an 83% (20/24) DVT rate (P b .01). The number of catheter lumens also correlated with incidence of DVT. Single-lumen CVCs had a 6% DVT rate, while multiple-lumen CVCs had a significantly higher incidence at 54% (P b .01, Table 4). Within the multiplelumen group, double-lumen CVCs had a 48% DVT rate and triple-lumen catheters had an 82% incidence (9/11). Singlelumen CVCs had a mean size of 3.8 ± 0.9 French (Fr), while multilumen CVCs were a mean of 4.9 ± 1.4 Fr (P b .01). In addition, all multilumen CVCs were placed percutaneously. The mean number of catheter days from CVC placement to DVT diagnosis was 20.8 ± 28.5 days for the entire group. Days to DVT did not significantly differ between the CVC placement sites, with 19.8 ± 22.5 for jugular, 20.5 ± 31.7 for femoral, and 25.2 ± 15.0 for subclavian catheters. However, those with percutaneous lines had a significantly shorter time to DVT diagnosis than those with Broviacs: 12.9 ± 16.6 days vs 30.3 ± 36.9 days (P = .02).
1.0 .02 ⁎
Table 4
CVC lumen number and DVT incidence N
10 (17%)
1 (0.4%)
b.01 ⁎
Values for +DVT and −DVT groups are presented as mean ± SD, unless otherwise indicated. Abbreviation: EGA, estimated gestational age. ⁎ Significance (P b .05) on univariate analysis. ⁎⁎ Significance (P b .05) on logistic regression.
Single lumen +DVT −DVT Multiple lumens +DVT −DVT
180 10 170 65 35 30
Multilumen CVCs had a significantly higher incidence of DVT. ⁎ P b .01.
% 6 95 54 ⁎ 46
1162
Fig. Central venous catheter site in 333 patients and incidence of CVC-associated DVT. Femoral CVCs had a significantly higher DVT association (⁎42/155, P b .01) than jugular (12/111) and subclavian (6/64) CVCs.
Deep venous thrombosis treatment consisted of catheter removal in all patients, low-molecular-weight heparin in 67% of cases, and unfractionated heparin in 12%. No pharmacologic treatment was given in 21% because of elevated bleeding risk. The mean duration of treatment was 8.3 ± 8.5 weeks. No patients received thrombolytics. Five patients had a negative hypercoagulability workup, and 1 patient with +DVT had a familial history of hypercoagulability. Long-term complications consisted of recurrent thrombus in 2 patients and persistent limb edema in 1 infant.
3. Discussion The increasing incidence of CVC-related DVT remains an unsolved clinical problem. The aim of this study was to characterize CVC-associated DVT in infants younger than 1 year at a single institution by identifying risk factors for CVC-related DVT, describing DVT management and outcomes, and proposing DVT risk-reduction strategies. In our cohort, 333 infants younger than 1 year had an 18% incidence of DVT overall, with most of these related to femoral CVCs. The presence of DVT also correlated with a longer hospital length of stay, ICU admission, medically induced paralysis, percutaneous CVCs, and the use of multilumen CVCs. The DVT group had similar 1-year survival to controls and a low incidence of long-term morbidity. Central venous catheters are commonly placed in hospitalized infants. The procedural complications include bleeding, arterial puncture, pneumothorax, and malposition. Long-term complications include infection, migration, mechanical malfunction, catheter thrombosis, and DVT
B.W. Gray et al. [8,14]. In particular, the incidence of CVC-related DVT in infants ranges from 5% to 50% [13,15,16]. Furthermore, evidence suggests that a substantial amount of patients with CVCs may have a DVT but remain asymptomatic [17,18]. Therefore, the true incidence of CVC-related DVT may be higher than most studies report. Several studies have investigated risk factors for infants and children receiving CVCs. The Canadian Childhood Thrombophilia Registry was able to identify conditions associated with DVT in the presence of a CVC [4,9]. Raffini et al [1] found that most children who developed DVTs had at least 1 complex comorbid medical condition. In another study, Revel-Vilk and Ergaz [7] suggested that risk factors for central line–related thrombosis can be categorized into those related to the catheter, the individual, and the underlying disease and its treatment. Although each of these investigators described the risks associated with DVT in children and infants in their cohort, there has never been a standardized risk stratification protocol developed for DVT in these patients. Similar to prior studies, most of the children in our study were critically ill. Perhaps, because patients requiring central access were particularly ill at baseline, there were few characteristics distinguishing the +DVT group from the −DVT group. Infants with +DVT had a later estimated gestational age and similar birth weight, ventilator days, and operations per patient. In addition, they had similar rates of malignancy, sepsis/infection, congenital malformations, and congenital heart disease. However, there were several notable distinctions between the groups. The +DVT group had a significantly longer mean length of hospital stay by almost 40 days, a higher percentage of ICU patients, and greater rates of medical paralysis. This is an indication that sicker, more sedentary patients may be more prone to DVT. Sandoval and colleagues [16] and Spyropoulos et al [20] found that children who developed DVTs while inpatients had long ICU stays, and the IMPROVE study group identified immobility of greater than 7 days as an independent risk factor for DVT in adults. Although relatively rare, we also observed higher rates of meningitis and perinatal asphyxia in the +DVT group. These factors were identified by van Ommen and colleagues [3] as top risk factors after CVC use for DVT in neonates. In support of these observations, there is a growing volume of literature relating systemic and local inflammation with formation and exacerbation of venous thrombosis [21,22]. Interestingly, an increasing number of catheter days did not correlate with an increased risk for DVT. We found that the +DVT group had significantly less mean and median catheter days. Several patients with −DVT had indwelling catheters for quite long durations because of conditions such as short bowel syndrome. On the other hand, DVTassociated CVCs were removed directly after DVT diagnosis. Our observations show that complications like DVT are an impetus to readdress the need for venous access, a
Characterization of CVC–associated DVT in infants conversation often delayed for those patients who never have problems with their CVCs. Our most substantial finding was the preponderance of DVTs associated with femoral CVCs compared with jugular and subclavian vein CVCs. Interestingly, there are multiple reviews in the pediatric and adult literature citing an increased DVT rate with femoral catheters [6,23-30]. In studies that separate patients by age, smaller patients (b1 year old) are found to have a higher incidence of femoral CVC-associated DVTs than any other group [5,6]. With smaller patients, there is a greater catheter to vessel size ratio, particularly in the smaller femoral vessels of infants. This creates a state of low and turbulent venous flow around the femoral catheter that can compound upon the endothelial damage, inflammatory response, and thrombogenic response caused by the catheter [25,31,32]. In addition, femoral CVCs are preferentially inserted in urgent and emergent situations, in which patients are sicker and likely immobilized, exacerbating the low-flow state. The femoral CVCs placed in the most ill patients are percutaneous lines, and we found those to have a much higher DVT rate compared with saphenofemoral Broviacs (83% vs 16%). Even with the use of US, placement of percutaneous lines often requires multiple venipunctures to access the vein and difficulty threading the wire, which may result in more vessel trauma than the cutdown technique used to insert saphenofemoral Broviac catheters. This increased endothelial damage may increase the risk of thrombus formation. Although there were no subclavian DVTs diagnosed by US in our study, this does not indicate that subclavian CVCs are safer than all other CVCs. It is possible that there were false-negative subclavian US studies in our cohort. Several investigators have demonstrated the low sensitivity of US for evaluating the subclavian veins for DVT because of the decreased ability to compress the vessel under the clavicle [17,28,33]. In these cases, venography can be used as the “gold standard” for DVT diagnosis [34]. Surprisingly, there were 18 patients in our study who developed DVTs at a location other than the concurrent CVC site. Three of these DVTs were found to be where prior CVCs had been inserted, but there were 9 DVTs that occurred where no CVC had been placed. It is possible that these DVTs were influenced by underlying thrombophilia in these infants, such as the increasing number of one of the genetic factors described in the literature [35,36]. In addition, in comparison with older children, neonates have increased factor VII concentrations and differences in whole blood measures of hemostasis. They also have been found to have a decreased amount of several anticoagulant and fibrinolytic proteins [2,36]. Another important observation from this study is the timing of DVT diagnosis in relation to CVC presence. Patients in our study were diagnosed with DVT on a mean catheter day 21. In separate prospective studies of CVC-associated DVT in children, Beck et al [5] and Talbott et al [6] both found that DVTs started to develop within the first week and as early as
1163 the first day of after CVC insertion. However, many of these DVTs were subclinical when first diagnosed. Most patients in our study were only surveyed for DVT after symptoms presented, implying that DVTs may have started to form as early as the first day of CVC presence. Interestingly, we found that patients with percutaneous lines had DVTs diagnosed almost 2 weeks before those with Broviacs (13 days vs 30 days). We have not found similar observations in the literature, but we can assume that this is caused by similar factors related to percutaneous lines discussed previously. In addition to CVC location, we found that multilumen catheters had a significantly higher association with DVT. This finding is not substantiated in the literature, but it may be explained by the fact that multilumen catheters were significantly larger than single-lumen CVCs. In addition, multilumen CVCs placed in infants are percutaneous lines that are associated with higher DVT rates, as discussed previously. None of the patients in this study were given chemoprophylaxis against DVT. There are no standardized and validated DVT prophylaxis protocols, and in a survey of 24 pediatric ICUs, Braga and Young [11] found only 1 with a written DVT prophylaxis protocol [37]. In one institutional protocol, Raffini and colleagues use early mobility and mechanical prophylaxis measures first, saving chemoprophylaxis for the children at highest risk for DVT. The American College of Chest Physicians goes as far to recommend no chemoprophylaxis for children with CVCs because of the paucity of supporting evidence for chemoprophylaxis and the significant potential bleeding risk to patients [19]. Thus, it would seem that the aggressive use of chemoprophylaxis is unwarranted in these patients until better evidence is collected. This study has several weaknesses. As a retrospective study, we did not gather information on all CVCs and complications prospectively, hindering our ability to perform a comprehensive data analysis. In addition, because we only imaged infants with clinical signs of DVT or evidence of DVT from other studies, we likely missed several subclinical DVTs in our patient cohort. In addition, we may have underreported the incidence of subclavian DVTs because we did not verify our relatively insensitive US findings with venography. Thus, the 18% incidence we reported may not be the true incidence of CVC-related DVT in these infants. Another weakness is that this study was not designed to accurately capture all treatment-related complications such as bleeding. Finally, we did not incorporate any long-term follow-up into our analysis, other than clinic notes from postdischarge hematology visits. Therefore, although we found very low morbidity in patients with DVT, we cannot remark on any longer-term complications.
4. Conclusions This study revealed that there is a significant risk of DVT among infants with a CVC. Although infants who develop
1164 DVTs are more likely to be critically ill, there are few distinguishing characteristics compared with patients with CVCs who do not develop DVTs. Thus, a reliable risk stratification protocol would be very difficult to create for this group. On the other hand, the placement location and management of CVCs can impact DVT incidence. We found substantially higher DVT rates for femoral CVCs, percutaneous lines, and multilumen catheters. Because the routine use of prophylactic anticoagulation may not be the best option to prevent DVTs, we suggest that one may have to accept the inherent risk of CVCs while looking for other avenues to reduce the risk of DVT. These include avoiding the use of femoral percutaneous lines except out of necessity, screening percutaneous lines for DVT, switching femoral lines to other sites as soon as possible, avoiding the use of larger multilumen catheters, and reducing overall CVC days.
References [1] Raffini L, Huang YS, Witmer C, et al. Dramatic increase in venous thromboembolism in children's hospitals in the United States from 2001 to 2007. Pediatrics 2009;124:1001-8. [2] Sutor AH, Uhl M. Diagnosis of thromboembolic disease during infancy and childhood. Semin Thromb Hemost 1997;23:237-46. [3] van Ommen CH, Heijboer H, Buller HR, et al. Venous thromboembolism in childhood: a prospective two-year registry in the Netherlands. J Pediatr 2001;139:676-81. [4] Monagle P, Adams M, Mahoney M, et al. Outcome of pediatric thromboembolic disease: a report from the Canadian Childhood Thrombophilia Registry. Pediatr Res 2000;47:763-6. [5] Beck C, Dubois J, Grignon A, et al. Incidence and risk factors of catheter-related deep vein thrombosis in a pediatric intensive care unit: a prospective study. J Pediatr 1998;133:237-41. [6] Talbott GA, Winters WD, Bratton SL, et al. A prospective study of femoral catheter-related thrombosis in children. Arch Pediatr Adolesc Med 1995;149:288-91. [7] Revel-Vilk S, Ergaz Z. Diagnosis and management of central-line– associated thrombosis in newborns and infants. Semin Fetal Neonatal Med. [8] Journeycake JM, Buchanan GR. Thrombotic complications of central venous catheters in children. Curr Opin Hematol 2003;10:369-74. [9] Massicotte MP, Dix D, Monagle P, et al. Central venous catheter related thrombosis in children: analysis of the Canadian Registry of Venous Thromboembolic Complications. J Pediatr 1998;133:770-6. [10] Journeycake JM, Buchanan GR. Catheter-related deep venous thrombosis and other catheter complications in children with cancer. J Clin Oncol 2006;24:4575-80. [11] Raffini L, Trimarchi T, Beliveau J, et al. Thromboprophylaxis in a pediatric hospital: a patient-safety and quality-improvement initiative. Pediatrics 127:e1326-32. [12] Kearon C, Julian JA, Newman TE, et al. Noninvasive diagnosis of deep venous thrombosis. McMaster Diagnostic Imaging Practice Guidelines Initiative. Ann Intern Med 1998;128:663-77. [13] Salonvaara M, Riikonen P, Kekomaki R, et al. Clinically symptomatic central venous catheter-related deep venous thrombosis in newborns. Acta Paediatr 1999;88:642-6. [14] Rey C, Alvarez F, De La Rua V, et al. Mechanical complications during central venous cannulations in pediatric patients. Intensive Care Med 2009;35:1438-43. [15] Revel-Vilk S. Central venous line–related thrombosis in children. Acta Haematol 2006;115:201-6.
B.W. Gray et al. [16] Sandoval JA, Sheehan MP, Stonerock CE, et al. Incidence, risk factors, and treatment patterns for deep venous thrombosis in hospitalized children: an increasing population at risk. J Vasc Surg 2008;47: 837-43. [17] Male C, Kuhle S, Mitchell L. Diagnosis of venous thromboembolism in children. Semin Thromb Hemost 2003;29:377-90. [18] Tousovska K, Zapletal O, Skotakova J, et al. Treatment of deep venous thrombosis with low molecular weight heparin in pediatric cancer patients: safety and efficacy. Blood Coagul Fibrinolysis 2009;20: 583-9. [19] Monagle P, Chalmers E, Chan A, et al. Antithrombotic therapy in neonates and children: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines (8th Edition). Chest 2008;133:887S-968S. [20] Spyropoulos AC, Anderson Jr FA, Fitzgerald G, et al. Predictive and associative models to identify hospitalized medical patients at risk for venous thromboembolism. Chest. [21] Libby P, Simon DI. Inflammation and thrombosis: the clot thickens. Circulation 2001;103:1718-20. [22] Wakefield TW, Henke PK. The role of inflammation in early and late venous thrombosis: are there clinical implications? Semin Vasc Surg 2005;18:118-29. [23] Venkataraman ST, Thompson AE, Orr RA. Femoral vascular catheterization in critically ill infants and children. Clin Pediatr (Phila) 1997;36:311-9. [24] DeAngelis GA, McIlhenny J, Willson DF, et al. Prevalence of deep venous thrombosis in the lower extremities of children in the intensive care unit. Pediatr Radiol 1996;26:821-4. [25] Gutierrez JA, Bagatell R, Samson MP, et al. Femoral central venous catheter-associated deep venous thrombosis in children with diabetic ketoacidosis. Crit Care Med 2003;31:80-3. [26] Krafte-Jacobs B, Sivit CJ, Mejia R, et al. Catheter-related thrombosis in critically ill children: comparison of catheters with and without heparin bonding. J Pediatr 1995;126:50-4. [27] Joynt GM, Kew J, Gomersall CD, et al. Deep venous thrombosis caused by femoral venous catheters in critically ill adult patients. Chest 2000;117:178-83. [28] Male C, Julian JA, Massicotte P, et al. Significant association with location of central venous line placement and risk of venous thrombosis in children. Thromb Haemost 2005;94:516-21. [29] Mian NZ, Bayly R, Schreck DM, et al. Incidence of deep venous thrombosis associated with femoral venous catheterization. Acad Emerg Med 1997;4:1118-21. [30] Worly JM, Fortenberry JD, Hansen I, et al. Deep venous thrombosis in children with diabetic ketoacidosis and femoral central venous catheters. Pediatrics 2004;113:e57-60. [31] Chidi CC, King DR, Boles Jr ET. An ultrastructural study of the intimal injury induced by an indwelling umbilical artery catheter. J Pediatr Surg 1983;18:109-15. [32] Pottecher T, Forrler M, Picardat P, et al. Thrombogenicity of central venous catheters: prospective study of polyethylene, silicone and polyurethane catheters with phlebography or post-mortem examination. Eur J Anaesthesiol 1984;1:361-5. [33] Baarslag HJ, van Beek EJ, Koopman MM, et al. Prospective study of color duplex ultrasonography compared with contrast venography in patients suspected of having deep venous thrombosis of the upper extremities. Ann Intern Med 2002;136:865-72. [34] Baarslag HJ, Koopman MM, Reekers JA, et al. Diagnosis and management of deep vein thrombosis of the upper extremity: a review. Eur Radiol 2004;14:1263-74. [35] Wermes C, von Depka Prondzinski M, Lichtinghagen R, et al. Clinical relevance of genetic risk factors for thrombosis in paediatric oncology patients with central venous catheters. Eur J Pediatr 1999;158(Suppl 3): S143-6. [36] Saxonhouse MA, Burchfield DJ. The evaluation and management of postnatal thromboses. J Perinatol 2009;29:467-78.
Characterization of CVC–associated DVT in infants [37] Braga AJ, Young AE. Preventing venous thrombosis in critically ill children: what is the right approach? Paediatr Anaesth 21: 435-40.
Discussion Discussant: Dr Douglas Barnhart (Salt Lake City, UT): I have a question about your study design. It seemed from your abstract, it doesn't sound as if you routinely screened everyone, so my question is, who are the 333 patients? And how does that impact the way you interpret these data. I understand it's really the 333 patients that got ultrasounds. Is it you're just more likely to be suspicious of a DVT in the femoral region and therefore get an ultrasound, and find a DVT, or is it possible that femoral DVTs are more likely to be symptomatic and/or detected on physical examination and therefore you get the US? Do you have any idea of what the rate of silent DVT in the subclavian and jugular vein catheters are because if it's only driven by (if I'm right), that your study population are only those that got ultrasound, so you really don't know what the real rate of DVT is. My second question is, how do you explain the fact that your DVTs aren't where the catheters are and that's not a small number on them—and how do you account for that? Response: Dr. Gray: Those are excellent questions. To answer your first question, the 333 patients were all the patients who got central venous catheters and the patients who were studied with ultrasound were those who were symptomatic. This is a retrospective review, so if a patient was found to have edema, or erythema, or some other symptom in the ICU, they were studied. It is likely we missed some patients who had a DVT but weren't studied. Additionally, patients who had a femoral catheter did risk a higher incidence of edema because the catheters in the femoral veins have a substantially higher catheter to vein size ratio. I agree that there is some bias here that we only studied patients who had some kind of symptom or a possible DVT found on another imaging modality. You asked about patients who we looked at who had subclavian or jugular catheters. The first problem with imaging the subclavian catheter by ultrasound is that there is a higher rate of false negatives because you can't get proper compression of the vessel using an ultrasound probe under the clavicle. Hence, we probably missed some DVTs in the subclavian veins even when we looked for them, and so I think that definitely is a skewed population here. Regarding your second question, 18 of our 60 patients who had DVTs had the thrombus in a place other
1165 than at the concurrent catheter site. We looked at our data and found that 3 of the 18 actually had a previous catheter at that same exact site, so there may have been some endothelial damage there that contributed to it. However, the rest of the patients had no evidence of a catheter at the thrombus site. A possible explanation is that patients in this population may have systemic inflammation that could predispose one to DVT. Unfortunately, we did not evaluate this in our study. Discussant: Dr Michael Tirabasi (Madison, WI): I'd like to congratulate you on an excellent paper about a common problem that we all face. I had a question about your femoral group where it looked like you grouped both catheters inserted into the femoral vein via the saphenous and those directly placed in the femoral vein together. I was curious if you made any attempt to separate that data as instinctively you might imagine there may be more direct femoral vein injury with catheters that weren't fed through the saphenous vein. Response: Dr Gray: That's an interesting question. We actually initially tried to evaluate that, but when looking at the medical record, we had some difficulty distinguishing if lines were specifically placed in the saphenous or femoral vein. Hence, we had to assume that all percutaneous lines were femoral and Broviacs were saphenofemoral. I think that this is probably a weakness of the study. Discussant: Dr Steven Fishman (Boston, MA): I recognize I have a biased patient population, but I question the comment about long-term morbidity. You said there was very little long-term morbidity in those that had DVTs. How many of those were followed with subsequent ultrasounds, and how long was the follow-up or did you even contact patients later on to see if they had morbidity because sometimes these kids come back a lot longer with very significant morbidity. Response: Dr Gray: Every patient that had a DVT was followed by our hematology service, and all those patients that were discharged from the hospital were seen back in the hematology clinic. In the data that I presented, the 3% who had a recurrent DVT and 2% who had chronic edema were all followed in this clinic. The mean treatment course we had was 8.3 weeks in those patients. Discussant: Dr Cynthia Downard (Louisville, KY): This is a very interesting clinical study, and I think we all deal with this problem. Was there any separation as far as temporary versus tunnelled catheters and the catheter size and the type of placement whether it was a cut down or percutaneous placement? I think the most important question is, is a temporary multilumen femoral catheter a
1166 surrogate for a DVT in a patient who is really sick in the intensive care unit who is going to have other hypercoagulable issues? Response: Dr Gray: To answer your first question, we looked at the percentage of patients who had a percutaneous temporary catheter, and we found that 60% of the patients
B.W. Gray et al. who had a percutaneous catheter placed in the ICU, probably under unstable conditions, developed DVTs at that catheter site, which was significant. However, we couldn't draw a lot of conclusions from it because of your second point, that the patients who had these nontunnelled catheters were likely ill patients who had many other risk factors for DVT.