Importance of Pretest Probability Score and D-Dimer Assay Before Sonography for Lower Limb Deep Venous Thrombosis

Transcription

1 Subramania m et al. Sonography of Lower Limb Deep Venous Thrombosis Vascular Imaging Original Research A C M E D E N T U R I C A L I M A G I N G AJR 2006; 186: X/06/ American Roentgen Ray Society Y Rathan M. Subramaniam 1,2,3 Tina Chou 1 Rebekah Heath 1 Robin Allen 4 O F Keywords: deep venous thrombosis, Hamilton score, Simplify D-dimer assay, sonography, veins DOI: /AJR Received September 2, 2004; accepted after revision January 4, Academic and Research Division, Department of Radiology, Waikato Hospital, Hamilton, New Zealand. 2 Department of Radiology, Waikato Clinical School, University of Auckland, Hamilton, New Zealand. 3 Present address: Department of Medical Imaging, The Canberra Hospital and the Australian National University, Canberra, Australia. Address correspondence to R. M. Subramaniam (rathan67@hotmail.com). 4 Department of Haematology, Waikato Hospital, Hamilton, New Zealand. Importance of Pretest Probability Score and D-Dimer Assay Before Sonography for Lower Limb Deep Venous Thrombosis OBJECTIVE. The purpose of our study was to develop and validate a clinical score (the Hamilton score) for the assessment of lower limb deep venous thrombosis (DVT) and to determine the usefulness of this score and a D-dimer assay before a complete lower limb sonographic examination. SUBJECTS AND METHODS. Five hundred forty-two consecutive ambulatory patients presenting to the emergency department were prospectively recruited, of whom 16 patients were excluded from the study. Eighteen history and examination variables were collected by the emergency department physicians. The Simplify D-dimer assay and a complete, single lower limb sonographic examination were performed in all patients. All patients with a negative sonographic examination for DVT were followed up for 3 months, and all those with a positive sonographic examination were given anticoagulation therapy. The Hamilton score was developed using the data from the first 214 patients and was prospectively validated in the next 312 patients. RESULTS. The most significant factors associated with a diagnosis of DVT were immobilization of the lower limb, active malignancy, and a strong clinical suspicion of DVT without other diagnostic possibilities by the emergency department physicians. Other factors were bed rest or recent surgery, male sex, calf circumference difference greater than 3 cm, and erythema. The Hamilton score was developed with the following weights: immobilization of the lower limb (2 points), active malignancy (2 points), strong clinical suspicion of DVT without other diagnostic possibilities by the emergency physicians (2 points), bed rest or recent surgery (1 point), male sex (1 point), calf circumference difference greater than 3 cm (1 point), and erythema (1 point). A score of 3 or greater indicates a likely probability for DVT, and a score of 2 or less represents an unlikely probability for DVT. Of the 103 patients with an unlikely probability Hamilton score and a negative D-dimer assay, only one patient had isolated calf DVT. A combined diagnostic strategy of unlikely-probability Hamilton score and a negative D- dimer would have a negative predictive value of 99% (95% confidence interval, %). CONCLUSION. An unlikely-probability Hamilton score and a negative Simplify D- dimer assay effectively exclude lower limb DVT, and a sonographic examination is unnecessary in this group of ambulatory emergency department patients. eep venous thrombosis (DVT) affects about 84 persons per D 100,000 each year [1]. Objective testing for DVT is essential because clinical assessment alone is unreliable [2 5]. Untreated DVT is associated with a high risk of pulmonary embolism [6], and false diagnosis of DVT results in unnecessary anticoagulant therapy, which is associated with a risk of bleeding [6 8]. Accurate diagnosis of DVT and prompt therapy [9] are essential to reduce the risk of thromboembolic complications. DVT also predisposes patients to postthrombotic or postphlebitic syndrome in 40 75% cases [10, 11]. Pretest probability score models, based on history and examination, for predicting the probability of DVT help clinicians improve the accuracy of their diagnosis of DVT. These models include the nine-component Wells score [5], the six-component St. Andre hospital score [12], the four-component Kahn score [13], and the six-component ambulatory score [14]. Recently, the Wells score has been modified to include 10 components [15]. The Wells score is a better predictor of DVT than are the Kahn and St. Andre s scores. The Wells score and the ambulatory score have similar operating characteristics [14]. 206 AJR:186, January 2006

2 Sonography of Lower Limb Deep Venous Thrombosis However, the modified Wells score has limitations in discriminating patients likely to have DVT and those unlikely to have DVT. In a study involving 1,096 ambulatory outpatients, 601 patients (54.8%) were identified as unlikely to have DVT and 495 patients (45.2%) as likely to have DVT using the modified Wells score [15]. This is despite the ambulatory population, which is expected to have a lower risk for DVT than hospital inpatients. The modified Wells score has overlapping redundant features, such as lower limb enlargement, calf enlargement, and pitting edema, that render the score less accurate in stratification and more cumbersome to calculate. Important risk factors such as prior history of DVT or pulmonary embolism, pregnancy, and the use of oral contraceptives were not considered when the Wells score was developed. Therefore, a pretest probability score needs to be developed that considers the important risk factors in the derivation population, stratifies patients more accurately into those having an unlikely probability and those having a likely probability, consists of no overlapping redundant features, and is easy to calculate. The objective diagnosis of DVT of the lower limbs now relies mainly on the use of sonography. In symptomatic patients, sonography has shown to be highly specific and sensitive compared with venography for both proximal [16, 17] and distal [18, 19] DVT. It is safe to withhold anticoagulant therapy in patients with clinically suspected DVT after a negative complete lower limb (including calf veins) sonographic [20 23] examination. The modified Wells score was based on a diagnosis of DVT using serial above-knee sonographic examinations rather than complete lower limb sonographic examinations [15]. In this study, we aimed to develop and validate a new pretest clinical probability score (the Hamilton score) with the objective of diagnosing DVT based on a single complete sonographic examination of a symptomatic lower limb and a 3-month clinical follow-up, and to determine the usefulness of an unlikely-probability Hamilton score and a negative Simplify D-dimer assay (Agen Biochemical) before a sonographic examination. Subjects and Methods Patients Five hundred forty-two consecutive patients who presented to the emergency department of a tertiary center for suspected lower limb DVT between October 2001 and May 2003 were prospectively recruited for the study. All patients were ambulatory outpatients and included both those referred to the tertiary center emergency department by primary care physicians in the community and those self-referred because of symptoms. The tertiary center has about 650 inpatient beds and departments of all specialties, including oncology and vascular surgery departments. Institutional ethics committee approval was obtained and all recruited patients signed a consent form. Emergency department physicians filled in a questionnaire comprising details of history and physical examination. Sixteen patients were excluded (current anticoagulation therapy, n = 7; failure to perform a D-dimer blood test before sonographic examination, n = 5; and technical inability to perform an adequate complete compression sonographic examination, n = 4). Thus the study population was 526 patients. D-Dimer Assay A latex agglutination immunochromatographic D-dimer assay called Simplify was performed in all 526 patients. This assay gives a semiquantitative result in 10 min. This immunochromatography test is performed on whole blood or plasma and is based on two D-dimer-specific antibodies: a murine monoclonal antibody, DD3B6/22, specific for D- dimer conjugated to colloidal gold particles, and a second D-dimer-specific murine monoclonal antibody. The antibody gold conjugate binds specifically to D-dimer-containing molecules in the patient sample to form a complex that migrates through a membrane in the aqueous phase until it is captured and concentrated on a zone to which the second D-dimer-specific murine monoclonal antibody has been bound [24]. The concentration of the complexes in this area causes a pink or purple line to appear on the membrane. Diagnosis of DVT Sonography Diagnosis of DVT was made by duplex compression (Acuson Sequoia 512 sonographic imaging system). The sonography was performed by experienced sonographers and radiology residents (third- and fourth-year) under the supervision of consultant radiologists. The results were interpreted by seven consultant radiologists who had experience of 2 10 years interpreting vascular sonographic studies as consultant radiologists, reflecting the day-to-day clinical practice. High-resolution, linear array transducers with variable frequency (6 8 MHz) probes were used (Acuson Sequoia 512 sonographic imaging system) in all studies. Patients were lying in the supine position with the symptomatic leg externally rotated and slightly flexed at the knee. Legs were examined from the level of the inguinal ligament to the medial malleolus. The common femoral vein, superficial femoral vein, popliteal vein, and trifurcation, and all three deep calf vein sets were examined. Compressibility of these veins was assessed at 2- to 3-cm intervals in the transverse plane. Noncompressibility of a segment of the veins was the sole criterion for diagnosis of DVT. Doppler examination of these veins was performed as supplemental information and was used only as a road map, but played no role in deciding the result (presence or absence of DVT) of the sonographic examination. Clinical follow-up and outcome All patients were followed up for 3 months from the date of a negative complete lower limb sonographic study. Patients were contacted by telephone at the end of the 3-month follow-up period; for those patients unable to be reached by telephone, their general practitioners were contacted to identify any subsequent venous thromboembolic events. All patients medical admission notes and imaging records were systematically reviewed to identify any subsequent venous thromboembolic events. For each abnormal event, data collected included clinical findings and results of tests, anticoagulation and duration, and cause and date of death from the death certificate. Each study had one of three possible end points: Patient was alive and had no subsequent thromboembolic event, patient was alive and had a subsequent thromboembolic event, or patient died during the follow-up period. Statistical Analysis Three types of statistical analysis were performed. The first approach was to conduct a decision tree analysis using a modified CHAID method (chi-square automatic interaction detection) (SPSS version 12.0, Statistical Package for the Social Sciences). The second approach was logistic regression (SAS, version 9.1.2, SAS Institute) with the outcome of whether DVT was diagnosed and 18 variables. The third approach was the Fisher s exact test two-tailed analysis of 18 variables with the diagnosis of DVT. Results The first 214 patients recruited for the study were 144 women and 70 men (age range, years). Forty-six patients (21.5%) were diagnosed with DVT. The upper limit of the thrombus was at or above the popliteal vein (proximal DVT) in 22 patients and below the popliteal vein (calf DVT) in 24 patients. The 46 patients were 25 men and 21 women with an average age of 53.5 years (±16.8 years [SD]) diagnosed with DVT. History and examination features in those with and those without DVT are summarized in Table 1. Of the 168 patients with an initial negative sonographic examination (negative cohort), 22 patients were readmitted to the emergency AJR:186, January

3 Subramaniam et al. TABLE 1: Characteristics of Patients With and Without Deep Venous Thrombosis (DVT) in the Derivation Sample (n = 214) Characteristic DVT Present (n = 46) No. (%) DVT Absent (n = 168) No. (%) History of DVT 8 (17.4) 19 (11.3) History of pulmonary embolism 3 (6.5) 6 (3.6) Male sex 25 (54.3) 45 (26.8) Age > 59 yr 17 (37.0) 71 (42.3) Active malignancy 6 (13.0) 5 (3.0) Recent plaster immobilization of lower limb 15 (32.6) 22 (13.1) (within 4 weeks) Recent trauma to lower limb 8 (17.4) 26 (15.5) Calf pain 40 (87.0) 149 (88.7) Family history 4 (8.7) 11 (6.5) Pregnancy 1 (2.2) 4 (2.4) Oral contraceptive use 6 (13.0) 24 (14.3) Bed rest > 3 days or recent surgery (within 12 (26.1) 23 (13.7) 4 weeks) Localized tenderness along deep veins 41 (89.1) 131 (78.0) Calf circumference comparison > 3 cm 16 (34.8) 33 (19.6) Pitting edema 17 (37.0) 40 (23.8) Erythema 14 (30.4) 31 (18.5) Swelling of entire leg 18 (39.1) 48 (28.6) Other possible diagnosis 12 (26.1) 89 (53.0) department with various presenting complaints during the 3-month follow-up period. Ten of those readmitted patients underwent lower limb sonography and one underwent CT pulmonary angiography (CTPA). The patient who underwent CTPA was diagnosed with pulmonary embolism on day 4 after the initial sonography. This patient was diagnosed with superficial thrombophlebitis on the initial sonographic scan. Of the 10 patients who had a second lower limb sonographic examination, none had any evidence of DVT. Of the remaining 146 patients in the negative cohort who never presented to the hospital during the follow-up period, none had any clinical evidence of a suspected DVT or pulmonary embolism or died on follow-up. In the decision tree analysis, the first split occurred using plaster immobilization, with 41% of those with immobilization having DVT (15/46 of those with DVT, but also 22/168 of those without DVT i.e., false-positive). Those without immobilization were split on active malignancy, with 54% of those without immobilization but with active malignancy having DVT (6/46, misclassifying an additional five patients). These two splits identify 21 of 46 of those with DVT, misclassifying 27 of 168 of those without that is, sensitivity of 46%, specificity of 87%. If those without active malignancy were split on strong clinical suspicion, 22% (20/46) of those with strong clinical suspicion by the emergency department physicians with no other diagnosis had DVT. Adding this further split would result in identifying 41 of the 46 with DVT but misclassifying 97 of the 168 without as false-positive that is, sensitivity of 89%, specificity of 42%. In the logistic regression analysis, the variables that were associated with DVT were plaster immobilization (p = 0.009) and strong clinical suspicion without other possible diagnosis by emergency physicians (p = 0.02), with a possible association of active malignancy (p = 0.07) and erythema (p = 0.08). Fisher s exact test two-tailed analysis identified plaster immobilization (p = 0.024), recent surgery (p = 0.004), male sex (p = ), calf circumference (p = 0.045), and erythema (p = 0.045) as being associated with diagnosis of DVT. On the basis of this analysis, the Hamilton score was developed (Appendix 1). A Hamilton score of 2 or less represents an unlikely probability of DVT, and a score of 3 and above represents a likely probability of DVT. The Hamilton score was validated and compared with the modified Wells score us- ing the next 312 patients. These patients were 117 men and 195 women with an average age of 55.8 years (± 20.3 years). Among these patients, 40 had previous DVT or pulmonary embolism, 20 had active malignancy, and one had both. A total of 67 patients (21.5%), with an average age of 55.3 years (± 15.8 years), were diagnosed as having DVT. Forty (59.7%) of these patients had isolated calf DVT and the other 27 (40.3%) had proximal DVT. Thirty-four of these patients were men and 33 were women. No subsequent thromboembolic event (DVT or pulmonary embolism) or death was noted among the 245 patients with a negative sonographic study at the end of the 3-month follow-up period. There were 197 (63.1%) patients in the Hamilton score unlikely category and 115 (36.9%) patients in the Hamilton score likely category, compared with 151 (48.4%) patients in the modified Wells score unlikely category and 161 (51.6%) patients in the modified Wells score likely category. There were 11.2% (22/197) of patients diagnosed with DVT in the Hamilton score unlikely population, and 11.3% (17/151) of patients diagnosed with DVT in the modified Wells score unlikely population. Forty-four (38.3%) of 115 patients were diagnosed with DVT in the Hamilton score likely population and 50 (31%) of 161 in the modified Wells score likely population. The modified Wells score is described in Appendix 2. One hundred forty-four patients (46.2%) had a negative D-dimer assay and 168 patients (53.8%) had positive D-dimer assay. Of the 144 patients with a negative Simplify D-dimer assay, 136 had negative sonographic examinations, one had a proximal DVT, and seven had an isolated calf DVT. Among the 168 patients with a positive D-dimer assay, 26 had a proximal DVT, 33 had an isolated calf DVT, and 109 had negative sonographic examinations. Of the 103 patients with an unlikely pretest probability by Hamilton score and a negative D-dimer assay, one patient had isolated calf DVT (Fig. 1). Of the 81 patients with an unlikely pretest probability by the modified Wells score and a negative D-dimer assay, one patient had isolated calf DVT. One hundred three (33%) of 312 patients had an unlikelyprobability Hamilton score and a negative D- dimer, compared with 26% (81/312) of patients with an unlikely-probability modified Wells score and a negative D-dimer. Sensitivity, specificity, positive predictive value, and negative predictive value for Hamilton score, Simplify D-dimer, and Hamilton score plus Simplify D-dimer are summarized in Table AJR:186, January 2006

4 Sonography of Lower Limb Deep Venous Thrombosis Score of 2 and negative D-dimer = 103 DVT = 1 No DVT = 102 Ambulatory emergency department patients recruited prospectively = 312 Hamilton score calculation, Simplify D-dimer assay performed = 312 Complete lower limb sonography of symptomatic leg and 3-month follow-up = 312 Score of 2 and positive D-dimer = 94 DVT = 21 No DVT = 73 Score of 3 and positive D-dimer = 74 DVT = 38 No DVT = 36 Score of 3 or negative D-dimer = 41 DVT = 7 No DVT = 34 All patients with DVT, including those with DVT in isolated calf vein, were treated with anticoagulation according to local treatment protocol. Fig. 1 Flow chart shows study outcome of validation phase. Of 103 patients with unlikely pretest probability by Hamilton score and negative Simplify (Agen Biochemical) D-dimer assay, one patient had isolated calf deep venous thrombosis (DVT). Numbers refer to numbers of patients. Discussion Our study showed the most significant variables associated with DVT are plaster immobilization of the affected limb, active malignancy, and strong clinical suspicion of DVT without another possible diagnosis by the emergency department physicians. Immobilization and malignancy are shown to be associated with DVT in other studies [5, 12]. Although other studies reported that the risk of DVT increases with age [1, 25], we found no association of age greater than 59 years and DVT. The studies of Wells et al. [26] and Kahn et al. [13] also found no independent association of age and DVT. Male sex was a predictor of DVT in our study and in some other studies [1, 13]. Kahn et al. [13] believed that the association of male sex and DVT was a self-referral bias resulting from women visiting general practitioners more often and for less lifethreatening diseases. A prior history of DVT or pulmonary embolism, pregnancy, and the use of oral contraceptives were not statistically significant in our study to include in the Hamilton score. This may be a result of the overreferral of patients with a prior history of DVT or pulmonary embolism by primary care physicians who fear missing a DVT. Although prior history of DVT was not considered in the derivation population, it was added to the recent modification of the Wells score [15]. The Hamilton score has seven nonoverlapping components: six objective components and a clearly defined subjective component (the clinician s intuitive judgment of a strong likelihood of DVT with no possibility of another diagnosis). This is in contrast to the 10-component modified Wells score with overlapping features and a less-well-defined subjective component of an alternative diagnosis at least as likely as DVT. The Hamilton score clearly identified more patients in the unlikely category than did the modified Wells score without increasing the false-negative rate, and it was more predictive of DVT in the likely category than the modified Wells score. The more patients identified appropriately in the unlikely category, the more patients will fall into the group of unlikely probability and a negative D-dimer. D-dimer is formed when cross-linked fibrin is lysed by plasmin, and elevated levels usually occur with thromboembolism. However, this is not specific for DVT or pulmonary embolism. Other causes, such as neoplasm, sepsis, postoperative state, and pregnancy, can cause elevated levels. The potential value of the D-dimer test, therefore, lies in its ability to exclude DVT. A necessity for a successful implementation of the D-dimer assay in clinical practice is a sufficiently high sensitivity and negative predictive value. Clinically useful D-dimer assays can be divided into three groups: In the first group, rapid enzyme-linked immunosorbent assays (ELISAs) have high sensitivity and negative predictive value, %, but low specificity [27 30], and are not suitable for individual patient testing [31]. In the second group, the latex agglutination tests tend to have some- TABLE 2: Performance Characteristics of the Hamilton Score, Simplify a D-Dimer, and Hamilton Score Plus Simplify D-Dimer Using Complete Compression Sonography as Reference Standard Assessment Tool Sensitivity (%) Specificity (%) Positive Predictive Value (%) Negative Predictive Value (%) Hamilton score ( ) ( ) ( ) ( ) Simplify D-dimer ( ) ( ) ( ) ( ) Hamilton score and Simplify D-dimer ( ) ( ) ( ) ( ) Note Numbers in parentheses are 95% confidence intervals. a Agen Biochemical. AJR:186, January

5 Subramaniam et al. Fig. 2 Flow chart shows diagnostic strategy for suspected lower limb deep venous thrombosis (DVT). Clinically excluding DVT by combining unlikely probability and negative D-dimer result can lead to a potential saving in health expenditures and allows timely discharge of patients. Hamilton score of 2 and negative D-dimer DVT excluded clinically what lower sensitivities but are more specific. Recently, highly sensitive latex agglutination tests have been developed, but they have lower specificities and their performance characteristics resemble those of ELISA assays. In the third group, SimpliRED (AGEN Biochemical Ltd.) assay, a whole-blood test that can be performed at the bedside, has a pooled sensitivity of 87.5% (95% confidence interval [CI], %) and specificity of 76.9% (95% CI, %) [32]. Only two studies have tested the accuracy of Simplify D-dimer assay in ambulatory emergency department patients for the diagnosis of lower limb DVT [24, 33]. In the first study, involving 120 consecutive outpatients, the sensitivity and negative predictive value were both 100% [24]. Neale et al. [33] recruited 187 ambulatory patients, and the sensitivity and negative predictive value of Simplify D-dimer assay were 90.2% and 92.2%, which are comparable to our results. Venography was used as the reference standard in that study, compared with complete compression sonography in our study for the diagnosis of DVT. Our study showed that a negative Simplify D-dimer assay and an unlikely-probability Hamilton score would effectively exclude lower limb DVT safely with no sonographic examination in patients presenting with suspected DVT to the emergency department. Several other studies have suggested that a Suspected lower limb DVT Calculate Hamilton score, perform D-dimer blood test Hamilton score of 3 or positive D-dimer Perform complete lower limb sonography, including calf veins high negative predictive value of the D-dimer test can be used as part of a diagnostic algorithm [28, 34, 35]. The value of combining the clinical pretest probability and D-dimer test results has been confirmed in other studies [36, 37]. In our study, patients with an unlikelyprobability Hamilton score and negative D- dimer results represented about 33% of the study population. In this group of patients, sonography could have been avoided to effectively exclude lower limb DVT on the basis of clinical probability and D-dimer results. Using the Hamilton score identified 27.2% (22/81) more patients in the unlikely probability and negative D-dimer group than did the modified Wells score. This diagnostic strategy (Fig. 2) of clinically excluding DVT by combining unlikely probability and negative D-dimer results can lead to a potential saving in health expenditures on DVT sonographic examinations and allows timely discharge of patients from emergency departments and rural hospitals, where sonography may not be readily available, without compromising patient safety. We used a complete, single lower limb sonographic examination to diagnose DVT rather than above-knee serial sonographic examinations suggested by other studies [5, 15, 28]. This is another important difference in derivation and validation of the Hamilton score compared with the modified Wells score, which used serial above-knee sonographic examinations. In our study, the sonographic study was repeated only if the patient had progressive or new symptoms despite the initial negative study and re-presented to the emergency department. In our study, DVT in the isolated calf vein represented about 60% of all those diagnosed with DVT. This figure is higher than those results obtained from studies of venography [38], with more calf DVT detected on sonography. This finding may be attributed to the changing spectrum of the disease as more patients are referred at an earlier stage because sonography is a more noninvasive examination than venography. This change has led to the higher proportion of isolated calf DVT among all DVT. In our study, all patients who had an isolated calf DVT were treated with anticoagulation according to local protocol. The sixth American College of Chest Physicians consensus conference on antithrombotic therapy [39] also recommended symptomatic isolated calf DVT should be treated with anticoagulant medications. Thus, it is important to diagnose isolated calf DVT by a complete lower limb sonographic examination rather than by simply diagnosing proximal extension of calf DVT using serial above-knee sonographic examinations. Our results are applicable to adult ambulatory outpatients in whom a technically adequate complete lower limb sonographic examination could be performed. Our study was performed in a single institution with experienced senior radiology residents (third- and fourth-year) and qualified sonographers performing the sonographic examinations under the supervision of consultant radiologists, and thus it is applicable in similar institutions. Our study may be limited in that we examined only the symptomatic leg rather than both legs on sonography, reflecting our local clinical practice. It has been debated whether bilateral lower limb sonography should be performed in patients with suspected unilateral DVT [40]. However, few clinical data support such a practice [41]. Although we may have missed a small silent thrombus in the asymptomatic leg, this is considered less likely because our study population consisted of ambulatory outpatients who presented to the emergency department. Another limitation is the possibility of a small thrombus in the symptomatic leg that was missed on sonography and remained clinically silent at the 3-month follow-up. In conclusion, the Hamilton score more accurately stratifies ambulatory outpatients into those having an unlikely pretest probability of DVT and those having a likely pretest proba- 210 AJR:186, January 2006

6 Sonography of Lower Limb Deep Venous Thrombosis bility, compared with the modified Wells score. An unlikely-probability Hamilton score and a negative Simplify D-dimer assay effectively exclude lower limb DVT, and sonography is unnecessary in this group of ambulatory emergency department patients. References 1. Anderson FA Jr, 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. The Worcester DVT Study. Arch Intern Med 1991; 151: McLachlin J, Richards T, Paterson J. An evaluation of clinical signs in the diagnosis of venous thrombosis. Arch Surg 1962; 85: Nicolaides AN, Kakkar VV, Field ES, Renney JT. The origin of deep vein thrombosis: a venographic study. Br J Radiol 1971; 44: Sandler DA, Martin JF, Duncan JS, et al. Diagnosis of deep-vein thrombosis: comparison of clinical evaluation, ultrasound, plethysmography, and venoscan with X-ray venogram. Lancet 1984; 2: Wells PS, Hirsh J, Anderson DR, et al. Accuracy of clinical assessment of deep-vein thrombosis. Lancet 1995; 345: ; erratum in Lancet 1995; 346: Ginsberg J. Management of venous thromboembolism. N Engl J Med 1996; 335: Caranasos G, Stewart R, Cluff L. Drug-induced illness leading to hospitalization. JAMA 1974; 228: Hull R, Hirsh J, Jay R, et al. Different intensities of oral anticoagulation therapy in the treatment of proximal-vein thrombosis. N Engl J Med 1982; 307: Bates SM, Ginsberg JS. Clinical practice: treatment of deep-vein thrombosis. N Engl J Med 2004; 351: Strandness DE Jr, Langlois Y, Cramer M, Randlett A, Thiele BL. Long-term sequelae of acute venous thrombosis. JAMA 1983; 250: Mudge M, Hughes LE. The long term sequelae of deep vein thrombosis. Br J Surg 1978; 65: Constans J, Nelzy ML, Salmi LR, et al. Clinical prediction of lower limb deep vein thrombosis in symptomatic hospitalized patients. Thromb Haemost 2001; 86: Kahn SR, Joseph L, Abenhaim L, Leclerc JR. Clinical prediction of deep vein thrombosis in patients with leg symptoms. Thromb Haemost 1999; 81: Constans J, Boutinet C, Salmi LR, et al. Comparison of four clinical prediction scores for the diagnosis of lower limb deep venous thrombosis in outpatients. Am J Med 2003; 115: Wells PS, Anderson DR, Rodger M, et al. Evaluation of D-dimer in the diagnosis of suspected deep-vein thrombosis. N Engl J Med 2003; 349: Becker DM, Philbrick JT, Abbitt PL. Real-time ultrasonography for the diagnosis of lower extremity deep venous thrombosis: the wave of the future? Arch Intern Med 1989; 149: Kearon C, Ginsberg JS, Hirsh J. The role of venous ultrasonography in the diagnosis of suspected deep venous thrombosis and pulmonary embolism. Ann Intern Med 1998; 129: Gottlieb R, Widjaja J, Tian L, Rubens D, Voci S. Calf sonography for detecting deep venous thrombosis in symptomatic patients: experience and review of the literature. J Clin Ultrasound 1999; 27: Atri M, Herba MJ, Reinhold C, et al. Accuracy of sonography in the evaluation of deep vein thrombosis in both postoperative surveillance and symptomatic patients. AJR 1996; 166: Elias A, Mallard L, Elias M, et al. A single complete ultrasound investigation of the venous network for the diagnostic management of patients with a clinically suspected first episode of deep venous thrombosis of the lower limbs. Thromb Haemost 2003; 89: Schellong SM, Schwarz T, Halbritter K, et al. Complete compression ultrasonography of the leg veins as a single test for the diagnosis of deep vein thrombosis. Thromb Haemost 2003; 89: Stevens SM, Elliott CG, Chan KJ, Egger MJ, Ahmed KM. Withholding anticoagulation after a negative result on duplex ultrasonography for suspected symptomatic deep venous thrombosis. Ann Intern Med 2004; 140: Subramaniam R, Heath R, Chou T, Cox K, Davis G, Swarbrick M. Deep venous thrombosis: withholding anticoagulation therapy after negative complete lower limb US findings. Radiology 2005; 237: Cini M, Legnani C, Cavallaroni K, Bettini F, Palareti G. A new rapid bedside assay for D-dimer measurement (Simplify D-dimer) in the diagnostic work-up for deep vein thrombosis. J Thromb Haemost 2003; 1: Nordstrom M, Lindblad B, Bergqvist D, Kjellstrom T. A prospective study of the incidence of deep-vein thrombosis within a defined urban population. J Intern Med 1992; 232: Wells PS, Anderson DR, Bormanis J, et al. Application of a diagnostic clinical model for the management of hospitalized patients with suspected deep-vein thrombosis. Thromb Haemost 1999; 81: Briscoe ME. Why do people go to the doctor? Sex differences in the correlates of GP consultation. Soc Sci Med 1987; 25: Perrier A, Desmarais S, Miron MJ, et al. Non-invasive diagnosis of venous thromboembolism in outpatients. Lancet 1999; 353: Perrier A, Howarth N, Didier D, et al. Performance of helical computed tomography in unselected outpatients with suspected pulmonary embolism. Ann Intern Med 2001; 135: Kruip MJ, Slob MJ, Schijen JH, van der Heul C, Buller HR. Use of clinical decision rule in combination with D-dimer concentration in diagnostic workup of suspected pulmonary embolism: a prospective management study. Arch Intern Med 2002; 162: Lee AY, Ginsberg JS. The role of D-dimer in the diagnosis of venous thromboembolism. Curr Opin Pulm Med 1997; 3: Fancher TL, White RH, Kravitz RL. Combined use of rapid D-dimer testing and estimation of clinical probability in the diagnosis of deep vein thrombosis: systematic review. BMJ 2004; 329:821; Epub 2004 Sep 21; erratum in BMJ 2004; 329: Neale D, Tovey C, Vali A, et al. Evaluation of the Simplify D-dimer assay as a screening test for the diagnosis of deep vein thrombosis in an emergency department. Emerg Med J 2004; 21: Bates SM, Grand Maison A, Johnston M, Naguit I, Kovacs MJ, Ginsberg JS. A latex D-dimer reliably excludes venous thromboembolism. Arch Intern Med 2001; 161: Kraaijenhagen RA, Piovella F, Bernardi E, et al. Simplification of the diagnostic management of suspected deep vein thrombosis. Arch Intern Med 2002; 162: Anderson DR, Wells PS, Stiell I, et al. Management of patients with suspected deep vein thrombosis in the emergency department: combining use of a clinical diagnosis model with D-dimer testing. J Emerg Med 2000; 19: Lennox AF, Delis KT, Serunkuma S, Zarka ZA, Daskalopoulou SE, Nicolaides AN. Combination of a clinical risk assessment score and rapid whole blood D-dimer testing in the diagnosis of deep vein thrombosis in symptomatic patients. J Vasc Surg 1999; 30: Cogo A, Lensing AW, Prandoni P, Hirsh J. Distribution of thrombosis in patients with symptomatic deep vein thrombosis: implications for simplifying the diagnostic process with compression ultrasound. Arch Intern Med 1993; 153: Hyers T, Agnelli G, Hull R, Morris T, Samama M, Tapson V. Antithrombotic therapy for venous thromboembolic disease. Chest 2001; 119[suppl 1]:176S 193S 40. Cronan J. Deep venous thrombosis: one leg or both legs? Radiology 1996; 200: Sheiman R, McArdle C. Bilateral lower extremity US in the patient with unilateral symptoms of deep venous thrombosis: assessment of need. Radiology 1995; 194: Appendixes 1 and 2 appear on the next page AJR:186, January

7 Subramaniam et al. APPENDIX 1: Hamilton Score Characteristics Score Plaster immobilization of lower limb 2 Active malignancy (within 6 months or current) 2 Strong clinical suspicion of DVT by emergency department physicians and no other diagnostic possibilities 2 Bed rest (> 3 days) or recent surgery (within 4 weeks) 1 Male sex 1 Calf circumference > 3 cm on affected side (measured 10 cm below tibial tuberosity) 1 Erythema 1 Note A score of 2 represents unlikely probability for deep venous thrombosis (DVT); a score of 3 represents likely probability for DVT. APPENDIX 2: Modified Wells Score [15] Clinical Characteristic Score Active cancer (patient receiving treatment for cancer within previous 6 months or currently receiving palliative treatment) 1 Paralysis, paresis, or recent plaster immobilization of lower extremities 1 Recently bedridden for 3 days or more, or major surgery within previous 12 weeks requiring general or regional anesthesia 1 Localized tenderness along distribution of deep venous system 1 Entire leg swollen 1 Calf swelling at least 3 cm larger than that on asymptomatic side (measured 10 cm below tibial tuberosity) 1 Pitting edema confined to symptomatic leg 1 Collateral superficial veins (nonvaricose) 1 Previously documented DVT 1 Alternative diagnosis at least as likely as DVT 2 Note A score of 2 indicates that probability of deep venous thrombosis (DVT) is likely; a score of < 2 indicates that probability of DVT is unlikely. 212 AJR:186, January 2006