Ruling Out Acute Deep Vein Thrombosis by ELISA Plasma D-Dimer Assay Versus Ultrasound

Ruling Out Acute Deep Vein Thrombosis by ELISA Plasma D-Dimer Assay Versus Ultrasound in Inpatients More Than 70 Years Old Alain F. Le Blanche, MD* Virginie Siguret, PhD Catherine Settegrana, MS Sorina Bohus, MD* Elisabeth Le Masne de Chermont, MD, PhD* Jean-Paul Andreux, PhD and Pascale Gaussem, PhD IVRY-SUR-SEINE, FRANCE ABSTRACT In geriatric care, deep-vein thrombosis (DVT) is mostly diagnosed by noninvasive techniques. The objectives of this prospective study were: (1) to evaluate the power of ELISA plasma D-dimer assay versus ultrasound (US) in ruling out acute DVT of the lower limbs in symptomatic geriatric inpatients, and (2) to determine the most effective D-dimer cutoff value over the age of 70 years. Over a 10-month period, inpatients with suspected lower limb DVT simultaneously underwent US examination and ELISA plasma D-dimer assay. Noninclusion criteria were comorbid conditions able to modify the D-dimer level. Data were processed by receiver operating characteristic (ROC) curve analysis. In total, 150 patients (125 women, 25 men), average age 86.3 years (range 70-101) were included. A diagnosis of lower limb DVT was established in 53 patients (35.3%). With a 500 ng/ml D-dimer cutoff conventional value, DVT was ruled out in only five patients (3.3%), whereas a 750 ng/ml value ruled out DVT in 19 patients (12.7%) with a sensitivity of 98.1%, and a negative predictive value of 95.0%. The only false negative corresponded to a patient with a 15-mm thrombus in the distal calf. In inpatients above 70, an ELISA plasma D-dimer value smaller than 750 ng/ml is a rapid reliable noninvasive means to rule out lower limb DVT, if color flow Doppler ultrasound is not available on site. From the *Department of Radiology and Medical Imaging, the Laboratory of Haematology and - Haemostasis, Charles Foix-Jean Rostand University Hospital Group, Ivry-sur-Seine, France. 1999 Westminster Publications, Inc., 708 Glen Cove Avenue, Glen Head, NY 11545, U.S.A. 873

Introduction Deep-vein thrombosis (DVT) of the lower limbs represents a public health challenge in Western countries, and the prevalence of this disease is higher in the geriatric population. 1>2 In fact, DVT represents a major cause of morbidity and mortality in the growing population of elderly patients.3 Since the annual incidence rate of DVT increases from 2.25 per 1,000 for inpatients aged 70-74 years to 3.15 per 1,000 for inpatients aged 85-89 years,4 it appears to be cost-effective to propose a diagnostic noninvasive first screening technique that is able to exclude DVT in primary health care or in chronic care institutions, in which imaging equipment is frequently poor or absent. In geriatric institutions, DVT represents 3% of mortality causes and has a 21% 1-year mortality rate.4 The gold standard diagnostic technique has been contrast venography (CV) for a long time.5 Although it remains useful for patients with a high body mass index, CV mostly requires transfer of the patient from a midterm center or chronic care facility to an acute hospital care unit. Owing to the nosocomial infections or confusion risk induced by transfer, there is some reluctance from physicians daily involved in geriatrics to transfer the patient and perform CV. Moreover, the injection of iodinated contrast media is not devoid of risks regarding hemodynamic intolerance reactions or extravasation. CV or inde- is also known to provide false-negative terminate results in the case of complete occlusion, extrinsic compression, or technical failure. 6,7 In a recent study, even muscular branch DVTs of the legs were detected more accurately by ultrasound (US) than by CV.6 US has now become a highly accurate and reliable noninvasive technique in the diagnosis of lower limb DVT,7-9 even in the subpopliteal area,6~lo and this low-cost examination technique is the most easily available in geriatric centers. Recently, the determination of plasma D-dimer levels, specific degradation products of crossed-linked fibrin, has become a routine diagnostic tool to exclude thromboembolic disease, by use of enzyme-linked immunosorbent assay (ELISA) techniques.11,12 In middle-aged patients, an ELISA plasma D-dimer cutoff value of 500 ng/ml has been previously reported to be efficient in the rule out of venous thromboembolism, when associated with negative compression ultrasonography.ll-14 The management of geriatric patients with suspicion of DVT implies minimally invasive diagnostic procedures and should avoid unnecessary exposure to anticoagulant therapy, for the resulting risk of hemorrhage is enhanced over the age of 72 years. 15 Rule out of DVT therefore requires rapid assessment in geriatric medicine. Aging has been proved to raise the D-dimer leve1,16,1~ and the 500 ng/ml cutoff value, effective in middle-aged patients, is assumed to be modified over the age of 70 years. It may also be increased by various comorbid conditions, including arterial thrombosis, 18 cancer, 19,20 cirrhosis,21 surgery within the last 4 weeks,22-24 and infection.ll Under these conditions, a strategy of diagnosis exclusion still needs to be established in the elderly, taking into account that geriatrics basically involves the workup of patients with multiple pathologic status. The objectives of this study were: (1) to evaluate the accuracy indexes of ELISA plasma D-dimer assay versus ultrasound out acute DVT of the lower limbs in (US) in ruling symptomatic geriatric inpatients and (2) to determine the most effective D-dimer cutoff value over the age of 70 years. Methods Patients and Study Design Over a 10-month period from July 1997 to April 1998, inpatients with clinically suspected first episode of lower limb acute DVT underwent plasma D-dimer assay and US examination on the same day. The clinical suspicion of lower limb DVT included edema, erythema, tenderness, and pain. Informed consent was required. Data were prospectively collected regarding the presence and site of DVT. The ineligibility criteria were absence of consent, age below 70 years, outpatients, clinical suspicion of pulmonary embolism, referral for DVT follow-up of a recent episode of acute DVT, ongoing anticoagulant therapy, anticoagulant therapy stopped for less than 1 week, and comorbid conditions able to raise the D-dimer level-ie, arterial thrombosis, cancer, cirrhosis, surgery within the past 4 weeks, and systemic infection. Because the D-dimer level may also be decreased by the time elapsed from onset of symptoms, 14,25 the patients were included only if both D-dimer assay and US were performed within 2 days after onset of symptoms. 874

D-Dimer Assay: Blood Samples Venous blood samples (3 ml) were collected from a forearm vein in Vacutainer tubes (Becton Dickinson, San Jose, CA) containing sodium citrate (0.129 M, 1/10 blood volume). Plasma was obtained by centrifugation at 2,500 g for 15 minutes at + 15 C and D-dimer assay was performed -, immediately. Assay Technique D-Dimer levels were quantified by use of an ELISA assay on the VIDAS immunoanalyser (biomerieux, Marcy 1 Etoile, France). VIDAS D-dimer is a fully automated quantitative assay for the specific measurement of cross-linked fibrin degradation products in human plasma. The assay principle is a two-step combination of a sandwich-type ELISA initial method and an enzyme-linked fluorescent assay (ELFA) final detection. Single-dose reagents are ready to use and the assay provides a quantitative result within 35 minutes. D-Dimer levels were routinely measured on a 1/5 dilution of citrated plasma. Under these conditions, the detection limit was 45 ng/ml and the linearity limit was 5,000 ng/ml. Reference Ultrasound Diagnosis: Examination Procedure Staff radiologists daily involved in evaluating patients for lower limb DVT performed the US examinations, being blinded to the results of D-dimer assay. All studies were performed with commercially available US equipment-128 X P4 model (Acuson, Mountain View, CA) -according to the depth of the vessel examined. Common and external iliac veins were examined with a 3.5-MHz curvilinear display transducer, and veins below the inguinal ligament were examined with a multifrequency 5.0-7.0-MHz linear display transducer. Configuration, color sensitivity, and gain setting were chosen to optimize imaging of the deep veins in each patient. Sonography was performed by a standard method, optimized by examination of the calf.2,6,10 All venous segments were visualized transversely and longitudinally. Each examination was performed bilaterally, evaluating the common and external iliac vein, common femoral vein, superficial femoral vein, popliteal vein, tibioperoneal trunk, posterior tibial veins, peroneal veins, anterior tibial veins, internal and external saphenous veins, and muscular branches, such as the gastrocnemius and soleal veins. Iliac veins were evaluated by color or power Doppler. Below the inguinal ligament, the suprapopliteal veins were examined by compression and color flow imaging in the supine position. Color or power Doppler was used for flow imaging in the pelvis, thigh, and popliteal fossa and for vein mapping in the leg. The popliteal vein and subpopliteal segments were examined by compression, under color flow vein mapping, while the patient was sitting. Spectral evaluation, phasic venous flow, and flow augmentation maneuvers by calf compression were routinely added when spontaneous flow was not adequate to completely fill the lumen with color. Assessment of Acute DVT The diagnosis of recent DVT was based on one or more of the following criteria: above the knee, visualization of the clot, noncompressibility of the vein in the absence of wall thickening with associated vein distension, and filling defect of color or power Doppler signal in the thigh or in axes of the popliteal fossa.26 In the calf, the diagnostic criterion was noncompressibility of the veins, with or without associated pain, in the absence of superficial collaterals. Because it has a 35% risk of superficial clot extension into the deep system,27 thrombosis of the upper one third of the internal saphenous vein was considered to be US positive. A muscular branch thrombosis with extension to the tibioperoneal trunk was classified as US positive. Rule out of Acute DVT When compression of the vein was subtotal but completely painless and associated with thickening or irregularity of the walls, narrowing or irregular lumen, or an adjacent venous collateral network, the findings were considered consistent with chronic DVT.26 Thrombosis of the internal saphenous vein without extension to its proximal one third or to the common femoral vein was classified as US negative. A muscular branch thrombosis without extension to the tibioperoneal trunk was also classified as US negative. Patients were considered free of DVT when US examination was negative on the day of inclusion. In case of persistent clinical high suspicion of DVT, the negative initial US diagnosis had 875

to be confirmed by negative US follow-up, to remain classified as negative. The average time interval between two US studies was 1 week. Data Processing A data form was completed for each patient at the time of the US study and included analysis of findings in the previously mentioned segments. The time elapsed from onset of symptoms and the presence of comorbid conditions potentially influencing D-dimer levels were recorded. The rate of informative cases was defined as the percentage of patients without DVT and D-dimer level below the cutoff value (rule-out rate). For D-dimer assay, a patient was classified as negative when below the cutoff value or positive when above the cutoff value. Prevalence and accuracy indexes-ie, sensitivity, specificity, negative predictive value (NPV), positive predictive value (PPV), and accuracy-were calculated according to the presence or absence of DVT on US examination for D-dimer cutoff values ranging from 500 to 1,000 ng/ml. A nonparametric receiver operating characteristic (ROC) curve was constructed 28 by plotting sensitivity (ordinate) versus loss of specificity (abscissa) for each cutoff value. Ninety-five percent confidence intervals were calculated by use of the binomial distribution. Results Of 426 patients referred to our Department for US examination over the study period, 276 were not eligible according to one or more of the noninclusion criteria: 68 had received anticoagulant therapy, 50 were below the age of 70 years, 48 had a clinical suspicion of pulmonary embolism, 47 suffered from systemic infections, 40 were outpatients, 39 had arterial thrombosis in various sites, 34 had cancer, and 27 had recent surgery. Twenty-six patients did not have D-dimer assay and US validation within 24 hours. Six patients failed to give their consent. Three patients suffered from cirrhosis. Finally, 150 geriatric inpatients (125 women and 25 men), aged 86.3 ± 5.0 years (mean ± sd, range 70-101) were included in the study. A diagnosis of DVT of the lower limbs was established in 53 patients according to the US examination, which represents a 35.3% prevalence (Figure 1). At the critical 500 ng/ml D-dimer conventional cutoff value, there was one false negative and five true negatives, whereas a 750 ng/ml D-dimer cutoff value ruled out the diagnosis of DVT in 19 patients, on the basis of the accepted standard US examination (Table I). Under these conditions, the sensitivities were comparable (98.1%, confidence interval: CI 94.3%-100%) at both D-dimer cutoff values, but the NPV was 95% Figure 1. Distribution of the D-dimer values in hospitalized geriatric patients. The 750 ng/ml cutoff value is indicated by a dashed line. The arrow indicates the single D-dimer false-negative patient. 876

Table I Diagnostic Performances for Various Thresholds of the D-Dimer Assay Number of patients: 150 (125 women, 25 men), all inpatients, with an average age of 86.3 years (70-101). Table II. Accuracy Indexes (%) of the Plasma D-Dimer Assay at Various Cutoff Values Prevalence = 35.3%; NPV: negative predictive value; PPV: positive predictive value. (CI 85.4%-100%) at 750 ng/ml, versus 83.3% (CI 53.4%-100%) at 500 ng/ml. Moreover, the rule-out rate was 12.7% at 750 ng/ml, compared to 3.3% at 500 ng/ml (Table II). All specificity values remained below 39%. Among the patients with DVT, one 75-yearold woman had a D-dimer level of 359 ng/mlie, below both 500 and 750 cutoff values. She suffered from a recent thrombus in the distal one third of her right peroneal vein. At US examination, the length of this symptomatic thrombus was equal to 15 mm (Figure 2). The US examination was independently repeated by two staff radiologists for control of the DVT assessment, and consensus was obtained. Subsequent followup of the patient during heparin treatment pro- 877

Figure 2. US examination of the false-negative case. In the DVT group, the only patient with a D-dimer level below both 500 and 750 cutoff values-359 ng/ml-had a recent thrombus in the distal one third of the right peroneal vein (arrowhead), and the size of this thrombus was 15 mm.., vided evidence of clot disappearance within 2 weeks. No sequelae of DVT were visible on 3- month US control, probably because of the limited initial thrombosis size. This patient with DVT was the only false-negative patient of our series, corresponding to sensitivity and NPV values less than 100%. ROC curve analysis (Figure 3) optimized the choice of a clinically efficient D-dimer cutoff value, and above age 70 years, a 750 ng/ml value appeared to reasonably rule out DVT with the strongest discriminant power, based on the highest sensitivity and NPV. Discussion Lower limb DVT is a frequently encountered severe disease of the elderly. In previous studies including middle-aged patients with suspected acute DVT, the prevalence of the disease represented about one half of the referred population.13,29 In our experience, this prevalence is 35.3%, close to that of another study. 14 Indeed, in the elderly, the physical signs of DVT are predominantly torpid, and the clinical suspicion may have been overestimated by the physicians of our geriatric units, as the elderly patient s complaint is known to be rarely targeted, may emerge or be interpreted with delay, owing to a loss of autonomy or to language disorders. The ultimate purpose of the present study was obviously not to eliminate the use of US by D-dimers but to evaluate the ability of plasma D-dimer assay to rule out lower limb acute DVT in symptomatic geriatric inpatients, compared to US. The other objective was to define an effective D-dimer cutoff value for patients over the age of 70 years. A rapidly available standard D-dimer had to be chosen. Because of assay technique all>13 their poor sensitivity, we rejected bedside latex 878

Figure 3. ROC curve for the determination of the D-dimer cutoff value of 750 ng/ml in the rule out of DVT above age 70 years. The curve was constructed by plotting the true-positive fraction (sensitivity, y-axis) versus the false-positive fraction (100 specificity, x-axis), for each cutoff value. techniques despite the fact that they can provide results in 5 minutes.3o,31 We chose the technique that is assumed to offer an acceptable rapidity (35 minutes) for a reliable emergency rule out.l2-14 In a previous comparative study including four ELISA techniques with low intraassay coefficients of variation, the VIDASO D-dimer assay used in the present study demonstrated high performance characteristics 13 for a cutoff value of 500 ng/ml.ll-14 The normal D-dimer level is higher in healthy elderly subjects than in middle-aged adults. 16,17 As expected, the 750 ng/ml D-dimer cutoff value reported in the present study is higher than that of studies including middle-aged adults. 1-14 One of the major ineligibility criteria was the existence of concomitant anticoagulant therapy known to decrease the D-dimer level, which is a cause of false-negative results.32 Based on US validation, our results demonstrate the efficiency of this 750 ng/ml ELISA D-dimer cutoff value with a 98.1% sensitivity and 95.0% NPV in the rule out of lower limb acute DVT in hospitalized patients over the age of 70. It is noteworthy that only one false-negative case with a D-dimer level of 359 ng/ml was evidenced, correlated. with a 15-mm clot in the distal right calf, visible on US. Similarly, in a previous study including 171 patients, the ELISA D-dimer assay was reported to be less effective for patients with isolated infrapopliteal distal thrombi smaller than 2 cm.l3 Therefore, the lack of sensitivity of the D-dimer assay in the presence of small clots may explain the false-negative case of our study. Unfortunately, the reported sensitivity, specificity, NPV, and PPV did not take into account these clots, and it is likely that these small distal thrombi were considered by the authors to induce a lower risk for pulmonary embolism, as previously reported,7,33 this latter point remaining uncertain.29 Although the usual source of pulmonary embolism is a DVT, the exclusion of patients with clinical suspicion of pulmonary embolism does 879

not seem inappropriate, regarding our first objective to specifically test the ELISA D-dimer assay in lower limb DVT. The true negative fraction of our study, in other terms the 12.7% rule-out rate, remains coherent with the results of other previous studies, involving D-dimer for the rule out of thromboembolic disease.11,34 Despite our strict noninclusion and PPV of D-dimer re- criteria, the specificity mained poor, as in other studies, which confirms that the ELISA D-dimer assay is a nondiagnostic test.13,14 This may be explained by the numerous associated diseases, for example, clinically silent tumors, frequently encountered in a geriatric population, and therefore missed in this recruitment. Moreover, it is emphasized that the hypercoagulability status of the elderly may elevate the D-dimer level. 17 Although some studies report the time saving of unilateral examination,35 or limited compression,36 we chose to examine our patients bilaterally from the ankle to the pelvis. However, there is a consensus to recognize that a few asymptomatic thromboses may be present in the contralateral limb. Because calf DVT may extend above the knee and therefore provide a high risk for pulmonary embolism in approximately 20% of cases,37 our survey protocol included investigation of calf DVT. In fact, US examination was the reference method of our noninvasive workup, and this implied investigation of the patients as completely as possible, including follow-up during anticoagulation.1,8 Under these conditions, ultrasonography has been considered to be a reference noninvasive technique-when performed according lower limb DVT in our geriatric patients. to strict examination rules-to assess D-Dimer assay is therefore useful only to rule out DVT and has no role in the assessment of DVT. Furthermore, the rule-out rate in inpatients is assumed to be further increased in outpatients, which seems to be of public health relevance when used in nursing homes or in primary health care. to a discrete cutoff value in a progressive condition such as DVT, inasmuch as the ELISA D-dimer assay represents an attractive low-cost rule-out technique in nursing homes or chronic care units not yet equipped with US or sometimes distant from the nearest US diagnostic facility. Therefore, from a geriatric point of view, this approach adds neither expense nor delay in some circumstances, frequently encountered in the elderly, such as isolated life conditions or poor access to an acute medical environment. Finally, in the absence of anticoagulants, this study demonstrates that the plasma D-dimer value below 750 ng/ml reason-, ably excludes DVT of the lower limbs in elderly inpatients over the age of 70 years, compared to the accepted noninvasive US standard. However, further studies, based on US diagnosis, are now needed in elderly outpatients to evaluate ELISA D-dimer assay and to determine its screening power for the noninvasive rule out of DVT, according to a cost-effectiveness approach. Acknowledgment The authors thank Suzette Pereira-Freira for secretarial expertise; Jacqueline Saindrenan, Danielle Germany, and Pervenche Guillemot for file follow-up; Liliane Lavoillotte for technical assistance; and Anthony Saul, MD, for manuscript revision. Alain F. Le Blanche, MD Department of Radiology and Medical Imaging Charles Foix-Jean Rostand University Hospital Group 7 avenue de la République F-94205 Ivry-sur-Seine France E-mail: alain.le-blanche@cfx.ap-hop-paris.fr Conclusion. Even though the most severe criticism of the ELISA D-dimer assay relies on its extremely low specificity, making this test of no clinical positive value, it does not appear to be unwise to adhere 880

References 1. Keogan MT, Paulson EK, Hertzberg BS, et al: Bilateral lower extremity evaluation of deep venous thrombosis with color flow and compression sonography. J Ultrasound Med 13:115-118, 1994. 2. Cronan JJ: Venous thromboembolic disease: The role of US. Radiology 186:619-630, 1993. 3. Anderson FA, Wheeler HB, Goldberg RJ, et al: A population-based perspective of the hospital incidence and case-fatality rates of deep vein thrombosis and pulmonary embolism. Arch Int Med 151:933-938, 1991. 4. Kniffin WD, Baron JA, Barrett J, et al: The epidemiology of diagnosed pulmonary embolism and deep venous thrombosis in the elderly. Arch Med Int 154:861-866, 1994. 5. Rabinov K, Paulin S: Roentgen diagnosis of venous thrombosis in the leg. Arch Surg 104:134-144, 1972. 6. Atri M, Herba MJ, Reinhold C, et al: Accuracy of sonography in the evaluation of calf deep vein thrombosis in both postoperative surveillance and symptomatic patients. Am J Roentgenol 166:1361-1367, 1996. 7. Lensing AWA, Prandoni P, Brandjes D, et al: Detection of deep-vein thrombosis by real-time B-mode ultrasonography. N Engl J Med 320:342-345, 1989. 8. Heijboer H, Bôller HR, Lensing AWA, et al: A comparison of real-time compression ultrasonography with impedance plethysmography for the diagnosis of deep-vein thrombosis in symptomatic outpatients. N Engl J Med 329:1365-1369, 1993. 9. Weinmann EE, Salzmann EW: Deep-vein thrombosis. N Engl J Med 331:1630-1641, 1994. 10. Polak JF, Culter SS, O Leary DH: Deep veins of the calf: Assessment with color Doppler flow imaging. Radiology 171:481-485, 1989. 11. Bounameaux H, DeMoerloose P, Perrier A, et al: Plasma measurement of D-dimer as diagnostic aid in suspected venous thromboembolism. Thromb Haemost 71:1-6, 1994. 12. Borg JY, Lévesque H, Cailleux N, et al: Rapid quantitative D-dimer assay and clinical evaluation for the diagnosis of clinically suspected deep vein thrombosis (letter to the editor). Thromb Haemost 77:602-603, 1997. 13. Elias A, Aptel I, Huc B, et al: D-dimer test and diagnosis of deep vein thrombosis: A comparative study of 7 assays. Thromb Haemost 76:518-522, 1996. 14. D Angelo A, D Alessandro G, Tomassini L, et al: Evaluation of a new rapid quantitative D-dimer assay in patients with clinically suspected deep vein thrombosis. Thromb Haemost 75:412-416, 1996. 15. Campbell NRC, Hull RD, Brant R, et al: Aging and heparin-related bleeding. Arch Intern Med 156:857-860, 1996. 16. Kario K, Matsuo T, Kobayashi H: Which factors affect D-dimer levels in the elderly? Thromb Res 62:501-508, 1992. 17. Mari D, Mannucci PM, Coppola R, et al: Hypercoagulability in centenarians: The paradox of successful aging. Blood 85:3144-3149, 1995. 18. Peltonen S, Lassila R, Rossi P, et al: Blood coagulation and fibrinolysis activation during sudden arterial occlusion of lower extremities: An association with ischemia and patient outcome. Thromb Haemost 74:1442-1446, 1995. 19. Von Tempelhoff GF, Dietrich M, Niemann F, et al: Blood coagulation and thrombosis in patients with ovarian malignancy. Thromb Haemost 77:456-461, 1997. 20. Edoute Y, Haim N, Rinkevich D, et al: Cardiac valvular vegetations in cancer patients: A prospective echocardiographic study of 200 patients. Am J Med 102:252-258, 1997. 21. Violi F, Ferro D, Basili S, et al: Ongoing prothrombotic state in the portal circulation of cirrhotic patients. Thromb Haemost 77:44-47, 1997. 22. Crippa L, Ravasi F, D Angelo SV, et al: Diagnostic value of compression ultrasonography and fibrinogen-related parameters for the detection of postoperative deep vein thrombosis following elective hip replacement: A pilot study. Thromb Haemost 74:1235-1239, 1995. 23. Böhler K, Hinterhuber G, Binder M, et al: Systemic activation of coagulation and fibrinolysis during varicose vein stripping. Dermatol Surg 23:46-50, 1997. 24. Cofrancesco E, Cortellaro M, Corradi A, et al: Coagulation activation markers in the prediction of venous thrombosis after elective hip surgery. Thromb Haemost 77:267-269, 1997. 25. Raimondi P, Bongard O, DeMoerloose P, et al: D- dimer plasma concentration in various clinical conditions : Implication for the use of this test in the diagnostic approach of venous thromboembolism. Thromb Res 69:125-130, 1993. 26. Lewis BD, James EM, Welch TJ, et al: Diagnosis of acute deep venous thrombosis of the lower extremities : Prospective evaluation of color Doppler flow imaging versus venography. Radiology 192:651-655, 1994. 27. Lutter KS, Kerr TM, Roedersheimer LR, et al: Superficial thrombophlebitis diagnosed by duplex scanning. Surgery 110:42-46, 1991. 881

28. Metz CE: Basic principles of ROC analysis. Semin Nucl Med 8:283-298, 1978. 29. Salzman EW: Venous thrombosis made easy. N Engl J Med 314:847-848, 1986. 30. Wells PS, Brill-Edwards P, Stevens P, et al: A novel and rapid whole-blood assay for D-dimer in patients with clinically suspected deep vein thrombosis. Circulation 91:2184-2187, 1995. 31. Turkstra F, Van Beek EJ, Wouter ten Cate J, et al: Reliable rapid blood test for the exclusion of venous thromboembolism in symptomatic outpatients. Thromb Haemost 76:9-11, 1996. 32. The DVTENOX study group. Markers of hemostatic system activation in acute deep venous thrombosis evolution during the first days of heparin treatment. Thromb Haemost 74:1442-1446, 1995. 33. Moser KM, LeMoine JR: Is embolic risk conditioned by location of deep Med 94:439-444, 1981. venous thrombosis? Ann Intern 34. Quinn RJ, Nour R, Butler SP, et al: Pulmonary embolism in patients with intermediate probability lung scans: Diagnosis with Doppler venous US and D-dimer measurement. Radiology 190:509-511, 1994. 35. Sheiman RG, McArdle CR: Bilateral lower extremity US in the patient with unilateral symptoms of deep venous thrombosis: Assessment of need. Radiology 194:171-173, 1995. 36. Pezzullo JA, Perkins AB, Cronan JJ: Symptomatic deep vein thrombosis: Diagnosis with limited compression US. Radiology 198:67-70, 1996. 37. Rose SC, Zwiebel WJ, Nelson BD, et al: Symptomatic lower extremity deep venous thrombosis: Accuracy, limitations, and role of color duplex flow imaging in diagnosis. Radiology 175:639-644, 1990. 882