Insensitivity of Color Doppler Flow Imaging for Detection of Acute Calf Deep Venous Thrombosis in Asymptomatic Postoperative Patients

i/ascular Diagnosis

Insensitivity of Color Doppler Flow Imaging for Detection of Acute Calf Deex, Venous Thrombosis in ~ s G ~ t o m a tPostoperative ic Patients1 Steven C. Rose, M D William J. Zwiebel, M D Louis E. Murdock, M D Aaron A. Hofmann, M D Derek L. Priest, RVT Rhonda A. Knighton, RVT Teriesa M . Swindell, RVT Peter F. Lawrence, M D Franklin J. Miller, M D

Index terms: Extremities, thrombosis, 93.751 Extremities, US, 44.12984, 45.12984 Thrombosis, US, 93.12984 Thrombosis, venous, 93.751 Veins, US studies, 93.12984

JVIR 1993; 4:111-117 Abbreviations: CDFI = color Doppler flow imaging, DVT = deep venous thrombosis

From the Departments of Radiology (S.C.R., W.J.Z., F.J.M.), Orthopedics (L.E.M., A.A.H.), and Surgery (P.F.L.), and t h e Non-Invasive Laboratory (D.L.P., R.A.K., T.M.S.), University of Utah Medical Center, 50 N Medical Dr, Salt Lake City, UT 84132, and the Veterans Administration Medical Center, Salt Lake City. Received April 1, 1992; revision requested J u n e 11; revision received July 27; accepted August 17. Address reprint requests to W.J.Z. Current address: Department of Radiological Sciences, UCLA Medical Center, Los Angeles. ' SCVIR, 1993

PURPOSE: Although color Doppler flow imaging (CDFI)has been shown to accurately depict calf vein thrombosis in symptomatic patients, this technique has not been proved accurate for detection of calf vein thrombosis in a population restricted to asymptomatic postoperative patients. PATIENTS AND METHODS: To evaluate the accuracy of CDFI in asymptomatic postoperative patients, both CDFI and contrast venography were performed on 78 limbs of 76 patients without symptoms of deep venous thrombosis (DVT)who had undergone either hip or knee replacement. CDFI and venographic examination were interpreted blindly with respect to the results of the other modality or clinical findings. Venography was the standard for comparison of results. RESULTS: Fifty-sixpercent of CDFI examinations of the calf vein were technically adequate. The remaining studies were compromised technically by limb swelling and/or obesity. For the technically adequate CDFI studies, calf vein thrombosis was detected in eight of 10 patients. Calculated sensitivity in this cohort was 80%,and specificitywas 97%.The sensitivity of CDFI for acute calf DVT in all patients, regardless of image quality, was 42%. CONCLUSION: These observations suggest that state-of-the-artCDFI is not an accurate examination for acute calf vein DVT in asymptomatic postoperative patients. CDFI is associated with a high rate of technically compromised studies and relatively low sensitivity in studies that are deemed technically satisfactory. These observations do not preclude the use of CDFI in postoperative patients for detection of thrombus extension into the popliteal vein or for detecting thrombosis of more proximal lower extremity veins.

L O W E R extremity deep venous thrombosis (DVT) and pulmonary embolism remain major causes of postoperative morbidity and mortality. Without perioperative prophylaxis, postoperative DVT occurs in approximately 40%-60% of patients who undergo elective total hip replacements (1-3) and 72%-84% of patients who undergo total knee arthroplasty (4,5). Fatal pulmonary embolism occurs in 2%-3% of such patients (2). The more effective prophylaxis regimens may reduce the risk of postoperative DVT by approximately one-half (1-3). Since most patients who develop postoperative DVT are asymptomatic, a sensitive and specific surveillance technique is required to iden-

tify DVT (2,6). The importance of DVT detection from a clinical standpoint is to determine which patients require a full course of anticoagulation. Venography has been accepted as the definitive means to diagnose DVT, but this technique is associated with patient discomfort and a low but defined incidence of complications (skin slough, contrast materialrelated reactions, nephrotoxicity, or phlebitis). In general, noninvasive imaging tests (eg, Doppler ultrasound [US] and plethysmography) are not sufficiently sensitive for detecting DVT in asymptomatic postoperative patients. This lack of sensitivity has been attributed to a high incidence of relatively small, nonocclusive thrombi in this population (2,6-8).

112

Journal of Vascular and Interventional Radiology

January-February 1993

Color Doppler flow imaging (CDFI) has been shown in prior studies to accurately enable diagnosis of femoropopliteal DVT in patients with clinical symptoms suggestive of DVT (9). The accuracy of CDFI for diagnosis of acute symptomatic calf vein DVT also has been documented but is dependent on obtaining technically adequate studies (9). To our knowledge, the accuracy of CDFI has not been documented for acute calf vein thrombosis in a population restricted to asymptomatic patients. The assessment of CDFI accuracy for calf DVT in this population is important, since the calf veins are thought to be the site of origin of DVT in most patients. Thus, the above stated goal of early detection of venous thrombosis in postoperative patients is contingent upon detecting calf vein involvement. Because of the lack of information on the accuracy of calf vein CDFI in asymptomatic postoperative patients, we prospectively evaluated CDFI for the detection of calf DVT in patients following total hip or total knee arthroplasty.

PATIENTS AND METHODS Study Population Between October 3, 1989, and January 18, 1991, 110 patients (112 limbs) who had undergone unilateral hip or knee replacement were enrolled in a protocol to compare competing antithrombotic prophylactic regimens3 for the prevention of DVT. This investigation was approved by the institutional review board. and informed consent was obtained from each patient. Each patient was randomly assigned to one of the antithrombotic prophylactic regimens, which was begun on the day of the operation. A venogram was of the affected extremity was obtained on approximately the 6th postoperative day, just prior to discharge. Venogra-

"ubcutaneous reconstituted depolymerized heparin and orally administered warfarin.

phy was the diagnostic standard against which CDFI was compared. Because of well-documented difficulty with radiographic differentiation between acute and chronic venous thrombosis, patients with a history of prior DVT or pulmonary embolism were not included in the study population. Patients also were excluded r to intravebecause of ~ r i o reaction nous contrast media, renal insufficiency, or childbearing concerns. Following these exclusions from the drug trial, a total of 112 limbs were available for study, but venography was not performed in 33 of these limbs for the following reasons: patient refusal (n = 131, failure to establish intravenous access (n = l l ) , equipment malfunction (n = l ) , and not specified (n = 8). Eleven of the 112 limbs in the study were not studied with CDFI for the following reasons: patient refusal (n = 8), equipment malfunction (n = I),and not specified (n = 2). Of note, in 10 cases neither venography nor CDFI was performed. Following these exclusions from the drug trial, the study population consisted of 76 patients in whom 78 lower extremities were examined with both venography and CDFI. The mean patient age was 64 years (range, 18-86 years). Fifty-five patients were men; 21 were women. Seventeen patients had undergone total hip replacement and 59, total knee arthroplasty (two bilateral). In all cases, bilateral lower extremity CDFI was performed within 24 hours

Venographic Technique Ascending venography was performed on a tilt table (reverse Trendelenburg position, approximately 30"-45") with absence of weight bearing on the lower extremity being studied. Intravenous access was obtained by using a 20- or 22-gauge sheathed needle (Insyte; Deseret Medical, Sandy, Utah) placed in a vein on the dorsum of the foot. Fluoroscopic monitoring was employed to ensure proper limb position and adequate venous opacification during

manual injection of contrast media (Conray 43; Mallinckrodt, St Louis). Tight tourniquets were applied both at the ankle and immediately below the knee. Three overhead radiographs of the opacified calf veins were obtained from lateral, anteroposterior. and in 15" internal rotation. The tourniquets were released, and the three radiographic views of the calf were repeated. Fluoroscopically monitored anteroposterior radiographs were obtained over the knee, thigh, and hip. In patients with a knee prosthesis, a supplementary lateral image of the knee was taken to evaluate that portion of the popliteal vein that is obscured by the prosthesis on the anteroposterior view. Finally, an anteroposterior image centered over the pelvis was obtained, with the tilt table in a horizontal position, the lower limb elevated by the radiologist, and the patient performing a Valsalva maneuver. Approximately 150 mL of contrast media was used. After the venographic study was finished, the intravenous access route was flushed with 250 mL of saline mixed with 1000 U of heparin.

Venographic Interpretation Venograms were independently double-read by two experienced angiographers blinded to both the clinical symptoms and interpretation of the CDFI studies. Discordant venographic interpretations were resolved by consensus. A venogram was considered positive only when acute thrombosis was diagnosed. Thrombosis was diagnosed venographically when a consistent intraluminal filling defect or segmental venous occlusion not attributable to venographic technique was present. Thrombosis was deemed acute either when the thrombus was outlined by contrast material ("tram-track" sign) or when veins that provided collateral blood flow around an occluded segment were of small caliber (10). Chronic thrombosis was diagnosed venographically when an occluded deep venous segment was accompanied by tortuous, large-caliber collateral veins. Although an attempt was made to ex-

Rose et a1

113

Volume 4 Number 1

clude patients with a history of DVT prior to the orthopedic procedure, three patients included in the study population had unsuspected chronic DVT seen on the venogram. None of these three had findings indicative of acute DVT; therefore, they remained in the study and their results were tabulated as negative for acute DVT.

CDFI Technique The technique that we used for CDFI evaluation of lower extremity veins has been published previously in detail (11). Either of two CDFI units (Angiodynograph or Quantum 2000; Quantum Medical Systems, Issaquah, Wash) was used with a 3.0-, 5.0-, or 7.5-MHz transducer. Although this report concerns only the calf veins, it is pertinent to describe our venous duplex technique in its entirety to document that the CDFI studies were performed carefully and thoroughly. With the hip externally rotated, the patient was positioned supine on an examination table tilted to approximately 30" reverse Trendelenburg. The examination was begun with longitudinal CDFI assessment of the distal external iliac and common femoral veins. The proximal portion of the long saphenous vein was then examined in the longitudinal plane. Doppler signals were next evaluated (for spontaneity, pulsatility, and Valsalva response) at a point just proximal to the junction of the deep and superficial femoral veins. CDFI examination was not routinely attempted in the proximal external iliac and common iliac vein if the femoral venous Doppler signals were normal. After the Doppler evaluation, the deep and superficial femoral veins were examined with CDFI in longitudinal planes from the groin distally to the adductor canal. The goal of CDFI in all areas was to visualize the vein wall and document that blood flow extended to the wall. When the longitudinal examination of the femoral segment was completed, the commoi and superficial femoral systems were reexam-

ined in transverse image planes to assess the response of the veins to compression. With completion of this portion of the examination, the extremity was externally rotated (if possible) for examination of the popliteal and calf veins. If it was possible to rotate the leg, these veins were examined from a posteromedial approach. Alternatively, if the extremity could not be rotated, the popliteal and calf vein examinations were conducted ~ r i n c i pally from a lateral approach. ~ i come pression examination was limited in such cases. The popliteal vein examination included longitudinal CDFI and transverse compression, beginning as high as possible in the adductor canal and continuing inferiorly as far as possible into the popliteal branches. The calf examination either proceeded from the knee to the ankle or vice versa, depending on which aproach best documented the course of ihe deep veins. The objective, in all cases, was to trace the anterior tibial, posterior tibial, and peroneal systems in their entirety. The primary mode of examination was longitudinal CDFI imaging. The goal of longitudinal imaging was B-mode demonstration of the vein wall coupled with CDFI verification that flow extended to the wall. If the vein wall could not be seen clearly, the goal was to demonstrate a solid, uniform "column" of flow with CDFI. Transverse compression examination also was used whenever possible. Attention was given to the large muscular branches in the soleus and gastrocnemius muscles, when these were visible.

Definition of CDFI Technical Adequacy In the iliofemoral-popliteal region, an adequate examination consisted of the following: (a) B-mode definition of the vein wall and documentation of flow to the wall and (b) documentation of response to compression. For the calf veins, an examination was considered adequate if a uniform, continuous column of flow could be seen in all segments of the posterior tibial and peroneal veins. B-mode vi-

sualization of the vein wall and assessment of compression were not possible in many of the calf veins, due to limited visualization.

Prior Documentation of the Sonographic Technique The accuracy of the CDFI technique used in this study has been confirmed previously by us in a series of symptomatic patients with suspected DVT (9). For calf vein examination, the sensitivity, specificity, positive predictive value, and negative predictive value of CDFI exceeded 95% in patients in whom the examination was technically adequate. The definition of technical adequacy used in the preceding study was the same as that described herein. CDFI Interpretation and Tabulation CDFI studies were examined by one of three experienced vascular US interpreters, each of whom were blinded to the clinical information, the side and type of operation (both lower extremities were examined with CDFI in all cases), and the results of the venogram. The diagnosis of acute thrombosis determined with CDFI was based on the following: (a)the presence of low-echogenicity material within a distended vein lumen, (b) the absence of normal vein compressibility, ( c ) a void within the color-encoded blood flow image, or ( d )the absence of visible flow within a vessel segment. Doppler flow abnormalities also were considered. In all cases of thrombosis, an attempt was made to differentiate between recently formed thrombus (above criteria) and the chronic residua of prior episodes of thrombosis (thickened, irregular, strongly echogenic walls and strongly echogenic filling defects in small or nondistended veins). Hospital charts were reviewed for demographic data, type of orthopedic surgery, and the presence of lower extremity symptoms suggestive for DVT. No patients had lower extremity symptoms suggestive for DVT beyond the usual postoperative pain and swelling.

114

Journal of Vascular and Interventional Radiology

January-February 1993

Figure. False-positiveCDFI study of the peroneal vein. (a)Non-augmented longitudi-

nal CDFI view of the proximal calf demonstrated negligible flow in either the posterior tibial (arrowheads)or peroneal (arrows)veins. (b) CDFI view following augmentation of flow by means of manual compression of the distal calf. Brisk flow was present within the posterior tibial veins but not the peroneal veins. In retrospect, the absence of flow augmentation was due to the fact that the peroneal vein courses between the noncompressible tibia and fibula. (c)Anteroposterior venographic view of the calf proved that the entire deep venous system was free of thrombus. C.

RESULTS Distribution of Thrombus Acute lower extremity DVT was present at venography in 24 (31%)of the 78 limbs studied. The calf was the sole site of acute DVT in 23 of 24 cases. In the one remaining case, an additional small non-occlusive thrombus was attached to a femoral vein valve cusp. Eleven limbs had solitary thrombi of the calf conduit veins, six had solitary thrombi of the muscular veins (soleal sinuses or gastrocnemius veins), the remaining seven had from two to four discrete areas of thrombosis in both the conduit and muscular branches. All Calf Examinations For all 78 limbs included in the study, CDFI was 42% sensitive and 98% specific for acute calf vein DVT at any location, regardless of CDFI

examination quality. CDFI had a positive predictive value of 91% and a negative predictive value of 79% (Table 1).

TechnicallyAdequate Examinations Forty-four (56%)of CDFI examinations met the criteria for technical adequacy. For these extremities, CDFI was 80% sensitive and 97% specific for calf DVT at any location; positive predictive value was 89% and negative predictive value was 94% (Table 1).CDFI errors in this subset included a missed, isolated, 9-cm occlusive thrombus located within one of the paired peroneal branches; a missed 5.5-cm, non-obstructing soleal sinus thrombus in a second patient; and absence of flow augmentation within the peroneal veins that was falsely interpreted as acute thrombosis (the venogram demonstrated these veins to be patent) (Figure).

Technically Inadequate Examinations For the 34 extremities in which the CDFI examination was deemed technically inadequate, CDFI was 14% sensitive and 100% specific for calf DVT at any location; positive predictive value was loo%, and negative predictive value was 63%. Factors contributing to technical compromise were extensive postoperative edema in 11extremities, obesity in seven, both edema and obesity in six, extensive collateral veins in two, lack of cooperation in two, obscuration of the sonographic image by extensive arterial calcification in one, and no indicated cause in five. Effects of Anatomic Location Twenty-one noncontiguous thrombi were identified venographi-

Rose e t a1

115

Volume 4 Number 1

Table 1 Comparison of CDFI with Venography for Acute Calf Vein DVT CDFI Studies Technically Technically All Adequate Inadequate (n = 78) (n = 44) (n = 34) Parameter 10 8 2 True-positive 53 33 20 True-negative 1 1 0 False-positive 14 2 12 False-negative 42% 80% 14% Sensitivity (TP/[TP + FNI) 98% 97% 100% Specificity (TN/[TN + FPI) 91% 89% 100% Positive predictive value (TP/[TP + FPI) Negative predictive value (TNI[TN + FNI) 98% 94% 63% 81% 93% 65% Accuracy ([TP + TNl /all cases) Note.-FN = false-negative, FP = false-positive, TN = true-negative, TP = truepositive.

Table 2 Relationship of CDFI Detection to Thrombus Location, Size, and Occlusivity Detected Variable at CDFI Location Conduit calf veins 9/21 (43) Soleal veins 3/12 (25) Gastrocnemius veins 111(100) Thrombus size > 10 cm 417 (57) 3-10 cm 5/12 (42) < 3 cm 012 (0) Occlusivity Occlusive 5/11 (45) Near-occlusive 518 (63) Non-occlusive (small) 012 (0) Note.-Twenty-one noncontiguous acute thrombi are tabulated separately.

cally in calf conduit veins (posterior tibial, peroneal, or anterior tibial veins). The relationship between CDFI detection of these thrombi and anatomic location is listed in Table 2. CDFI correctly identified nine of 21 acute thrombi located in the infrapopliteal conduit channels (sensitivity, 43%), three of 12 thrombi located in the soleal sinuses (sensitivity, 25%), and one thrombus located within a gastrocnemius vein.

Effects of Calf Thrombus Characteristics The length of calf vein thrombi and the degree of obstruction of blood flow were correlated with the CDFI interpretation as summarized in Table 2. CDFI allowed correct diagnosis of four of seven calf conduit vein thrombi longer than 10 cm and five of 12 thrombi between 3 and 10 cm long; it failed to depict two thrombi shorter than 3 cm. CDFI depicted five of 11occlusive calf conduit vein thrombi and five of eight near-occlusive thrombi but failed to depict two small. non-obstructive thrombi. DISCUSSION A sensitive surveillance technique for detection of asymptomatic DVT is desirable, given the importance of prevention of postoperative DVT and the fact that many patients with DVT are asymptomatic unless a pulmonary embolus occurs. Overall, the sensitivitv of CDFI for calf vein thrombolis in this study was poor (42%, regardless of image quality). If only the technically adequate examinations were considered, the sensitivity was improved (80%) but was still less than desirable for an "early warning'' screening examination for thrombosis that may either propagate into more proximal veins or

cause chronic postthrombotic sequelae. The combination of a low sensitivity level and a low rate of technically adequate examinations calls into question the value of CDFI for assessment of asymptomatic postoperative patients. Multiple authors have reported a similarlypoor sensitivity (26'3241%) with use of compression US for detedion of calf vein DVT in asymptomatic patients who have undergone hip or knee surgery (12-14). Another author reported that CDFI missed all thrombi (sensitivity,0%)in 15 asymptomatic patients with calf vein DVT after knee arthroplasty (15). The disappointing results of CDFI for detection of calf vein DVT do not entirely preclude the use of this modality for surveillance of patients who have undergone hip or knee replacement and are receiving antithrombotic prophylaxis. If the goal is to detect thrombus in the popliteal vein or more proximally, then CDFI probably is an acceptable technique. Several published studies suggest that compression US is accurate for detecting thrombus at the popliteal vein level or more proximally in high-risk asymptomatic patients (12,16-20). The exceptions are the studies reported by Borris et al (12,16) in which small, non-occlusive thrombi were difficult to detect in the femoropopliteal region. It should be noted that approximately 15%-32% of patients with calf DVT experience proximal extension of thrombosis to involve the femoral popliteal venous segments (21,22). This thrombus, which may have originated de novo or may have propagated from the calf, represents a significant risk for pulmonary embolus or the post-phlebitic syndrome. When discovered, such thrombus requires a full course of anticoagulant therapy (2). If CDFI is used for surveillance in patients after hip and knee replacements, it appears that continued antithrombotic prophylaxis may be indicated even if DVT is not detected, since CDFI does not reliably detect calf DVT. The alternative of obtaining a venogram to exclude DVT at any location may be more reasonable.

116

Journal of Vascular and Interventional Radiology

January-February 1993

The findings of this study do not necessarily preclude the use of CDFI for detection of calf vein thrombosis in other asymptomatic patients who are easier to examine than postoperative hip or knee replacement patients. These results also do not preclude the use of CDFI for examination of calf veins in symptomatic patients. Several studies of standard com~ressionUS or CDFI have demonskated acceptable sensitivity (880/0-95%) and specificity (92%-100%) for calf vein examination in symptomatic individuals (9.23-26). Review of the data irovides insight regarding the cause of the disappointingly low sensitivity of CDFI in this series of patients. The quality of the calf vein examination was the principal variable affecting CDFI sensitivity. In the 56% of patients with technically adequate examinations, the sensitivity was 80%. In comparison, CDFI sensitivity was only 14% in the 44% of patients with technically compromised studies. The primary causes of limited visualization were obesity (commonly encountered in patients with degenerative arthropathy) and edema (presumably incited by the operative procedure). The other factor limiting image quality was lack of patient mobility during the postoperative period. Our ability to position the extremity in proper image planes and to apply compressive maneuvers was hampered considerably by patient immobility. There seemed to be no increase in the likelihood of a compromised calf vein study in patients who underwent total knee rather than total hip replacement, a finding which suggests that all three of the above mentioned factors contributed to technical difficulties. The relatively poor results of this study might be taken as an indictment of the quality of the CDFI examinations included in this study or of the equipment employed. Such arguments are countered by the fact that we have previously reported excellent results for CDFI calf vein examination in symptomatic patients using the same technique and the same US instruments (9).

The location of thrombus is a second factor that had an identifiable effect on sonographic accuracy. The sensitivity of CDFI for detection of conduit vein thrombi (43%) was superior to that for soleal vein thrombi (25%).This difference is important, since the soleal veins are a common site of origin of calf DVT, ostensibly due to stasis of blood in these muscular veins (6,27). The size of individual thrombi also appears to have affected the sensitivity of CDFI for calf vein DVT. CDFI sensitivity for thrombi longer than 10 cm (57%)was slightly greater than that for thrombi between 3 and 10 cm (42%).CDFI failed to depict two calf thrombi shorter than 3 cm. It is not surprising that smaller thrombi were more difficult to detect than longer thrombi in this patient population. A logical relationship was not established between the degree of impediment to blood flow and the detectability of acute calf vein thrombi (sensitivity was 45% for occlusive thrombi, and 63% for nearly occlusive thrombi). One might reasonably expect that occlusive thrombi might be more easily detected with CDFI, given the diagnostic criteria used in this study. These two classes of thrombi, however, although distinct venographically, are very similar from a sonographic perspective. This factor probably accounted for the seemingly contradictory sonographic results. A more important observation may be the failure of CDFI to detect two small non-occlusive thrombi. It is noteworthy that Borris et al(12,16) reported similar difficulties with detection of small aboveknee thrombi as well as thrombi located in the calf veins with use of gray-scale (compressive) US. Acknowledgments: We express our

gratitude to Shari Rosner, Becca Wesselman, and Susie Dennis for their assistance in preparation of this manuscript. We also thank C. Gregory Elliott, MD, for his review of the manuscript. References

1. Hull RD, Raskob GE. Current concepts review: prophylaxis of venous

thromboembolic disease following hip and knee surgery. J Bone Joint Surg [Am] 1986; 68:146-150. Haake DA, Berkman SA. Venous thromboembolic disease after hip surgery: risk factors, prophylaxis, and diagnosis. Clin Orthop Re1 Res 1989; 242:212-231. Hull RD, Raskob GE, Gent M, et al. Effectiveness of intermittent pneumatic leg compression for prevention of deep vein thrombosis after total hip replacement. JAMA 1990; 263:2313-2317. Lotke PA, Ecker ML, Alavi A, Berkowitz H. Indications for the treatment of deep venous thrombosis following total knee replacement. J Bone Joint Surg [Am] 1984; 66: 202-208. Stulberg BN, Insall JN, Williams GW, Ghelman B. Deep-vein thrombosis following total knee replacement: an analysis of six hundred and thirty-eight arthroplasties. J Bone Joint Surg [Am] 1984; 66:194-201. Kalebo P, Anthmyr BA, Eriksson BI, Zachrisson BE. Phlebographic findings in venous thrombosis following total hip replacement. Acta Radiol 1990; 31:259-263. Paiement G, Wessinger SJ, Waltman AC, Harris WH. Surveillance of deep vein thrombosis in asymptomatic total hip replacement patients: impedance phlebography and fibrinogen scanning versus roentgenographic phlebography. Am J Surg 1988; 155:400-404. Comerota AJ, Katz ML, Grossi RJ, et al. The comparative value of non-invasivetesting for diagnosis and surveillance of deep vein thrombosis. J Vasc Surg 1988; 7:40-49. Rose SC, Zwiebel WJ, Nelson BD, et al. Symptomaticlower extremity deep venous thrombosis: accuracy, limitations, and role of color duplex flow imaging in diagnosis. Radiology 1990; 175:639-644. Zachrisson BE, Jansen HJ. Phlebographic signs in fresh post-operative venous thrombosis of the lower extremity.Acta Radiol Diagn 1973; 14232-96. Knighton RA, Priest DL, Zwiebel WJ, Lawrence PF, Miller FJ, Rose SC. Techniques for color-flow sonography of the lower extremity. RadioGraphics 1990; 10:775-786. Borris LC, Christiansen HM, Lassen MR, Olsen AD, Schott P (The Venous Thrombosis Group). Compar-

Roseet a1

117

Volume 4 Number 1

13.

14.

15.

16.

ison of real-time B-mode ultrasonography and bilateral ascending phlebography for detection of postoperative deep vein thrombosis following elective hip surgery. Thromb Haemost 1989; 61:363-365. Ginsberg JS, Caco CC, Brill-Edwards PA, et al. Venous thrombosis in patients who have undergone major hip or knee surgery: detection with compression ultrasound and impedance plethysmography. Radiology 1991; 181:651-654. Elliott CG, Suchyta M, Rose SC, et al. Duplex ultrasonography for the detection of deep vein thrombi after total hip or knee arthroplasty. Angiology (in press). Sumner DS, Londrey GL, Spadone DP, et al. Study of deep venous thrombosis in high-risk patients using color flow Doppler. In: Bergan J J , Yao JST, eds. Venous disorders. Philadelphia: Saunders, 1991; 6376. The Venous Thrombosis Group: Borris LC, Christiansen HM, Lassen MR, Olsen AD, Schott P. Real-time B-mode ultrasonography in the diagnosis of post-operative deep vein thrombosis in non-symptomatic high-risk patients. Eur J Vasc Surg 1990; 4:473-475.

17. Barnes RW, Nix ML, Barnes CL, et al. Perioperative asymptomatic venous thrombosis: role of duplex scanning versus venography. J Vasc Surg 1989; 9:251-260. 18. Woolson ST, McCrory DW, Walter JF, Maloney WJ, Watt JM, Cahill PD. B-mode ultrasound scanning in the detection of proximal venous thrombosis after total hip replacement. J Bone Joint Surg [Am] 1990; 72:983-987. 19. Comerota AJ, Katz ML, Greenwald LL, Leefmans E, Czeredarczuk M, White JV. Venous duplex imaging: should it replace hemodynamic tests for deep venous thrombosis? J Vasc Surg 1990; 1153-61. 20. Dorfman GS, Froehlich JA, Cronan J J , Urbanek PJ, Herndon JH. Lower-extremity venous thrombosis in patients with acute hip fractures: determination of anatomic location and time of onset with compression sonography. AJR 1990; 154:851855. 21. Kakkar W, Howe CT, Flanc C, Clarke MB. Natural history of post-operative deep-vein thrombosis. Lancet 1969; 2:230-232. 22. Huisman MV, Buller HR, Ten Cate JW, Vreeken J. Serial impedance plethysmography for suspected deep

23.

24.

25.

26.

27.

venous thrombosis in outpatients: the Amsterdam General Practitioner study. N Engl J Med 1986; 314:823826. Yucel EK, Fisher JS, Egglin TK, Geller SC, Waltman AC. Isolated calf venous thrombosis: diagnosis with compression US. Radiology 1991; 179:443-446. Elias A, LeCorff G, Bouvier JL, Benichou M, Serradimigni A. Value of real time B-mode ultrasound imaging in the diagnosis of deep vein thrombosis of the lower limbs. Int Angiol1987; 6:175-182. Habscheid W, Hohmann M, Wilhelm T, Epping J . Real-time ultrasound in the diagnosis of acute deep venous thrombosis of the lower extremity. Angiology 1990; 41599-608. Cavaye D, Kelly AT, Graham JC, Appleberg M, Briggs GM. Duplex ultrasound diagnosis of lower limb deep venous thrombosis. Aust N Z J Surg 1990; 60:283-288. Nillius AS, Nylander G. Deep vein thrombosis after total hip replacement: a clinical and phlebographic study. Br J Surg 1979; 66:324-326.