The Use of a D-dimer Assay in Patients Undergoing CT Pulmonary Angiography for Suspected Pulmonary Embolus

Clinical Radiology (2002) 57: 41±46 doi:10.1053/crad.2001.0740, available online at http://www.idealibrary.com on

The Use of a D-dimer Assay in Patients Undergoing CT Pulmonary Angiography for Suspected Pulmonary Embolus G U Y J. C . B U R K I L L *, J A M E S R . G . B E L L *, RO G E R J . S . C H I N N *, J E R E M I A H C . HE A LY *, C H R I S T I N E CO S T E L LO {, LY N AC TO N {, S IM O N P. G . PA DL E Y * *Department of Diagnostic Radiology, Chelsea and Westminster Hospital, London, U.K. and {Department of Haematology, Chelsea and Westminster Hospital, London, U.K. Received: 17 October 2000

Revised: 1 February 2001 Accepted: 7 February 2001

PURPOSE: To assess the ability of a semi-quantitative latex agglutination D-dimer test Accuclot2 with bedside measurements of arterial oxygen saturation, respiratory and cardiac rates to exclude pulmonary embolism (PE) on computed tomographic pulmonary angiography (CTPA). MATERIALS AND METHODS: All patients referred to our CT unit for investigation of suspected acute pulmonary embolism were enrolled. Pulse oximetery, respiratory rate, heart rate and blood sampling for D-dimer testing were carried out just before CT. A high resolution CT (HRCT) of the chest was followed by a CT pulmonary angiogram (CTPA). The images were independently interpreted at a workstation with cine-paging and 2D reformation facilities by three consultant radiologists blinded to the clinical and laboratory data. If positive, the level of the most proximal embolus was recorded. Discordant imaging results were re-read collectively and consensus achieved. RESULTS: A total of 101 patients were enrolled. The CTPA was positive for PE in 28/101 (28%). The D-dimer was positive in 65/101 (65%). Twenty-six patients had a positive CT and positive D-dimer, two a positive CT but negative D-dimer, 39 a negative CT and positive D-dimer, and 34 a negative CT and negative D-dimer. The negative predictive value of the Accuclot2 D-dimer test for excluding a pulmonary embolus on spiral CT was 0.94. Combining the D-dimer result with pulse oximetry (normal SaO2 5 90%) improved the negative predictive value to 0.97. CONCLUSION: A negative Accuclot2 D-dimer assay proved highly predictive for a negative CT pulmonary angiogram in suspected acute pulmonary embolus. If this D-dimer assay were included in the diagnostic algorithm of these patients a negative D-dimer would have unnecessary CTPA rendered in 36% of patients. Burkill, G. J. C. et al. (2002). Clinical Radiology 57, 41±46. # 2002 The Royal College of Radiologists Key words: pulmonary arteries, thrombosis, D-dimer, CT, embolism, pulmonary.

Despite extensive discussion in the medical and radiological literature, there remains controversy as to the most e€ective investigation for acute pulmonary thromboembolic disease. The shortcomings of lung scintigraphy have been appreciated for over a decade [1], though in the U.K. it continues to be the ®rst line, and often the only specialist investigation performed in these patients. While conventional pulmonary angiography is the purported gold standard for the diagnosis of acute pulmonary embolism, it is fast becoming a lost art [2±4] and, in practice, a signi®cant proportion of patients are managed speculatively [4] despite recommendations to the contrary by the British Thoracic Society [5]. It has been suggested that computed tomographic pulmonary angiography (CTPA) is the imaging investigation Author for correspondence and guarantor of study: Dr S. P. G. Padley, Department of Radiology, Chelsea and Westminster Hospital, 369 Fulham Road, London, SW10 9NH, U.K. Fax: ‡ 44 (0) 020 8746 8588; E-mail: [email protected] 0009-9260/02/010041+06 $35.00/0

best suited to meet the shortfall in diagnostic certainty left by the decline of pulmonary angiography [6,7]. The workload generated by thromboembolic disease is signi®cant, with pulmonary embolism (PE) accounting for an estimated 0.23±1.0% of acute hospital admissions [8,9], and approximately 45 specialist imaging investigations per hospital per month being undertaken for PE diagnosis [3]. These ®gures indicate that the potential demand for CTPA, with its inherent cost and ionizing radiation implications, could be considerable. There is a need to regulate this demand. D-dimer is a speci®c ®brin degradation product (FDP) elevated in venous thromboembolic (VTE) disease as well as in a number of other clinical states such as disseminated intravascular coagulation (DIC), sickle cell crisis, renal failure, malignancy and pre-eclampsia [10]. Until recently the quanti®cation of D-dimer levels relied on highly sensitive but timeconsuming enzyme-linked immunosorbent assays (ELISA) which, although validated in the detection of acute # 2002 The Royal College of Radiologists

42

CLINICAL RADIOLOGY

thromboembolic disease, have not proved practical as screening tests [11±13]. Newer latex agglutination and rapid ELISA assays now available are coming into clinical use [10,14,15]. In this study, we evaluated the Accuclot2 (Sigma Diagnostics, St Louis, MO, U.S.A.) D-dimer test, a rapid latex agglutination assay, against CT pulmonary angiography in patients with suspected acute pulmonary embolus. MATERIALS AND METHODS

Following ethics committee approval, consecutive patients referred to the CT unit with suspected acute pulmonary embolism were enrolled in the study. Patients with a prior history of thromboembolic disease or contraindication to intravenous contrast medium were excluded. Arterial oxygen saturation (SaO2) (Hewlett Packard M1275A, U.S.A.) and cardiac and respiratory rates were recorded. The ability of the patient to breathhold was assessed. A 20 G cannula was sited in a medial antecubital fossa vein through which 4.5 ml blood was drawn into a 0.9% sodium citrate sample bottle for the D-dimer assay. High resolution CT images were obtained (1.5 mm collimation, 10 mm interval, high spatial frequency algorithm). All CT studies were undertaken on a Siemens Somatom Plus 4 system (Siemens, Erlangen, Germany) with a 0.75 second gantry rotation time. This was immediately followed by a CT pulmonary angiogram. 150 ml Omnipaque 300 (Nycomed, Amersham, U.K.) contrast medium was power injected intravenously at a ¯ow rate of 4 ml/s. The volume of contrast medium had been increased from 100 ml early in the study following the initial report of the European multi centre (ESTIPEP) study, which recommended the use of the larger volume [16]. The initial cohort of patients was excluded from the ®nal study group. Craniocaudal data acquisition was commenced from the aortic arch with a Z-axis coverage of 12 cm. If the patient was judged able to breathhold for at least 22 seconds, a slice thickness of 2 mm ( pitch ˆ 2.0) was employed. If not, the slice thickness was broadened to 3 mm ( pitch ˆ 2.0), which required a breathhold of just 15 seconds. The delay after initiation of injection was adjusted by the supervising radiologist according to clinical factors such as age, cardiothoracic ratio and evidence of cardiac failure. Typical injection delays ranged between 15 and 25 seconds. All CT images were reviewed independently by three consultant radiologists (R.C., J.H. and S.P.) on a commercial workstation (Siemens MV VA 31, Erlangen, Germany). They were scored for image quality on a scale of 1±5 (1 ˆ excellent) and diagnostic con®dence (low, medium or high). The position of the most central thrombus was recorded for all positive examinations. If there was disagreement, the images were re-read collectively and consensus achieved. Blood samples were centrifuged and plasma obtained for immediate D-dimer assay during normal working hours or frozen for analysis on the next working day using the visually read semi-quantitative Accuclot2 test. A D-dimer

value of 0.25 mg/l or greater was recorded as a positive result. Serial dilutions were made to give a semi-quantitative estimate of the level of D-dimer present.

RESULTS

A total of 149 patients were enrolled. As the ®rst 29 patients were administered 100 ml of contrast medium they were excluded from further analysis. D-dimer results were not available for 19 of the 120 patients undergoing CTPA with 150 ml contrast medium owing to inadequate sample volume or delayed sample analysis. The remaining 101 patients made up the study population. There were 46 men and 55 women with a mean age of 58 years. Seventy-three CTs were determined to be negative and 28 positive for pulmonary embolus. The location of the most centrally sited embolus is given in Table 1. Five studies were discordant on the ®rst reading for the presence of PE. Consensus was reached on these, three being considered to be positive and two negative. The average or predominant diagnostic con®dence level for all three readers was high in 77 cases, intermediate in 22, and low in just two. The mode of the image quality score for all readers for all CT examinations was 2. The D-dimer assay was positive in 65 patients and negative in 36. The combined results of the CTPA and D-dimer assay are given in Table 2. Two of the 28 patients with positive CTPA had a negative D-dimer assay. In both cases the proximal extent of the thrombus was at a segmental level. Pulse oximetry data was available in 90 of the 101 patients studied. The oxygen saturation was 590% in 78 patients, of which 58 had a negative CTPA. Heart rate was recorded in 90 patients. A rate of 4100/min was found in 74, of whom 56 had a negative CTPA. The respiratory rate was recorded Table 1 ± Proximal extent of pulmonary emboli in the 28 PE positive patients Location of most central PE

Number of patients

Main Lobar Segmental Subsegmental

9 10 8 1

PE ˆ pulmonary embolus.

Table 2 ± CTPA results and D-dimer assays in the 101 patients with completed protocols

Positive D-dimer Negative D-dimer Total

Positive CTPA

Negative CTPA

Total

26 2 28

39 34 73

65 36 101

CTPA ˆ computed tomographic pulmonary angiography.

THE USE OF A D-DIMER ASSAY IN CT ANGIOGRAPHY FOR SUSPECTED PULMONARY EMBOLUS

43

Table 3 ± Predictive value of a negative D-dimer, SaO2 590%, respiratory rate 420/min and a heart rate 4100/min alone and in combination for excluding PE diagnosed on CTPA Observation

Correct exclusion

Negative predictive value

Negative D-dimer SaO2 5 90% Respiratory rate 420 breaths/min Heart rate 4100 beats/min Negative D-dimer and SaO2 5 90% Negative D-dimer and respiratory rate 420 breaths/min Negative D-dimer and heart rate 4100 beats/min Negative D-dimer and SaO2 5 90% and respiratory rate 420 breaths/min and heart rate 4100 beats/min

34/36 58/78 46/63 56/74 28/29 20/22 24/25 18/19

0.94 0.74 0.68 0.73 0.97 0.91 0.96 0.95

PE ˆ pulmonary embolus. CTPA ˆ computed tomographic pulmonary angiography.

in 86 patients. A rate of 420 breaths/min was recorded in 63, of whom 46 had a negative CTPA. The negative predictive value of the D-dimer assay and cardiorespiratory data alone and in combination is given in Table 3. Thirty-one of the 36 patients with a negative D-dimer had complete cardiorespiratory observations. Of the two patients with a negative D-dimer but positive CTPA, the ®rst had a normal SaO2 (93%), respiratory rate (20/min) and heart rate (84/min). The second was mildly hypoxic (SaO2 ˆ 88%), and tachycardic (heart rate ˆ 117/min) with a normal respiratory rate (16/min). This data suggests that only one of the patients with a positive CTPA had no indication of embolus from the D-dimer or cardiorespiratory observations made. DISCUSSION

In the present study the D-dimer assay was more sensitive than the three cardiorespiratory observations (singly or combined) in the prediction of a negative CT pulmonary angiogram. If a negative Accuclot2 D-dimer assay were accepted as evidence for the absence of PE, 36 of 101 CTs (36%) would have been avoided, at the cost of missing two cases of PE. If the exclusion criteria were a combination of a negative D-dimer and an SaO2 5 90%, then 29 of 96 (30%) CTs would have been avoided, with one false negative. In our sample, adding the respiratory rate and heart rate observations to the D-dimer and pulse oximetry data would further reduce speci®city with no improvement in sensitivity. The ®nding of normal cardiorespiratory observations in patients who are otherwise shown to have an acute pulmonary embolus is well described [5]. However, the majority of our patients with PE had surprisingly normal pulse oximetry and respiratory rate measures (eight of 26 were receiving oxygen support) despite the majority of thrombi residing at the main pulmonary artery or lobar level. The explanation for these observations may in part be due to delays in presentation to the CT unit, allowing time for cardiorespiratory compensation. Since the introduction of rapid D-dimer assays, interest has focused on the validation against more accurate, but relatively cumbersome, standard ELISA techniques. Initial

reports were disappointing. Data pooled from 11 deep vein thrombosis (DVT) and nine PE studies gave standard ELISA assays an average sensitivity of 96.8% for DVT and PE and speci®cities of 35% and 45% respectively. By comparison, the latex assays had a sensitivity of 83% for DVT (speci®city 68%). A more encouraging sensitivity of 92% (speci®city 54%) for PE was tempered by a low concordance between various latex assays (kappa coecient ˆ 0.43) [11]. The comparatively high speci®city of latex assays compared to ELISAs is well established. An analysis of six latex assays against two standard ELISAs gave 100% speci®city for all the latex tests, but sensitivities ranged between 81.3% and 56.3% for ®ve of the tests. The Accuclot2 test was the only rapid assay to achieve an adequate sensitivity (93.8%) in line with the manufacturer's claims [10]. Similar results for the Accuclot2 test (sensitivity 92%, speci®city 53%) were achieved by Freyburger et al., out-performing the SimpliRED2 assay (sensitivity 79%, speci®city 67%), which has already been clinically evaluated and promoted as suitable for screening patients when used in conjunction with the arterial oxygen tension [17]. The VIDAS2 assay is a quantitative fully automated rapid ELISA. It was the only rapid test to achieve 100% sensitivity and a high correlation coecient when compared to the reference methods. This was at the expense of a fall in speci®city to 38% [14]. The authors concluded that of the seven rapid tests examined which included the Accuclot2 assay, only the VIDAS2 test was suciently sensitive to exclude acute thromboembolic disease. These results are mirrored in a recent study comparing 10 novel D-dimer methods in suspected DVT established on venography. The VIDAS2 assay was once again 100% sensitive, a level only matched by the Tinaquant assay (Boehringer Ingelheim, Ingelheim-am-Rhein, Germany); speci®cities were 41% and 39% respectively [18]. A case against accepting a test sensitivity of less than 100% was o€ered by van Beek et al. They estimated that 0.1% of evaluated patients with clinically suspected pulmonary embolism would die for every 2% decrease in sensitivity [19]. Although an admirable aim for any medical investigation, a sensitivity of 100% is not realistically achievable, and increasing sensitivity is usually only

44

CLINICAL RADIOLOGY

SUSPECTED ACUTE PE

*PE unlikely

*PE likely

D-dimer

Normal

Elevated

CXR

STOP

Normal

Abnormal

V/Q

Normal High probability

STOP

Non-diagnostic

TREAT

CTPA

+ve

-ve

TREAT

STOP or consider compression US or pulmonary angiography

Fig. 1 ± Algorithm for the diagnosis of pulmonary embolism. *Based on Wells et al clinical scoring system [21].

achieved at the cost of reduced speci®city, as demonstrated in the case of D-dimer assays. Methods for improving the utility of the less sensitive but more speci®c rapid assays have been applied. These include using the test in conjunction with a pre-test probability score. Despite an overall sensitivity of just 85% and speci®city of 68%, the performance of the SimpliRED2 assay was enhanced by categorizing patients as having a low, moderate or high risk of PE according to a largely objective pre-test probability score. A 99% negative predictive value was achieved in patients with a negative D-dimer and low pre-test clinical probability score which occurred in 44% of the study population [20]. A potential limitation of this study was the reliance on lung scintigraphy, compression ultrasound to the calf trifurcation and 3-month follow-up rather than a more de®nitive imaging test [20]. The same authors have subsequently re®ned their clinical prediction model into a two-category system of PE

likely or unlikely [21]. An alternative technique for improving test performance is combining the result with simple bedside observations and arterial oxygen measurements, as was the case in this and a previous study [17]. A number of studies have demonstrated the inverse relationship between D-dimer speci®city and advancing age, no doubt related to the concomitant rise in comorbidity [22±24]. Therefore the performance of the Ddimer assay could be enhanced by excluding older patients. In suspected pulmonary embolism, a highly sensitive rapid assay was negative in 40% of patients below 50 years of age but just 8% in the over 70s [24]. Twenty-nine of our patients were over 70 years of age; 15 (52%) of these patients had a positive D-dimer but no evidence of PE on CT. Conversely, 24 of the 72 patients (33%) aged 70 or less had a positive D-dimer but negative CT. The cost of a D-dimer assay varies from approximately £3 to £5 (excluding labour costs). Thus it is not surprising

THE USE OF A D-DIMER ASSAY IN CT ANGIOGRAPHY FOR SUSPECTED PULMONARY EMBOLUS

that even if these tests spare only a minority of patients from further investigation, they will still o€er a signi®cant overall cost saving. Indeed, these assays remain economical even in patients with co-morbid conditions when the speci®city is known to be very low [14]. The authors recognize a number of limitations of this study. First we would justify the exclusion of the 29 preliminary studies by recognition of the imaging bene®ts of a greater volume of contrast medium, and the application of a standard radiographic technique. A further limitation includes the sole reliance on spiral CT pulmonary angiography for the ®nal diagnosis of pulmonary embolism. In our unit, as in the U.K. generally, pulmonary angiography is rarely performed and is no longer a practical option for establishing the diagnosis of PE [3]. In addition, the preliminary ®ndings of the largest prospective study to date determined the sensitivity of spiral CT to be high at 95% with a speci®city of 97% [16]. The gold standard of pulmonary angiography has been demonstrated to be less than perfect [25], and recently Baile et al. found no signi®cant di€erences in the ability of spiral CT and conventional angiography to detect methacrylate beads in the subsegmental pulmonary vasculature of anaesthetized pigs as de®ned at post mortem [26]. The diagnostic capability of CT falls from the central to the peripheral pulmonary arteries. The clinical signi®cance and diagnostic importance of isolated subsegmental emboli remains unresolved. Isolated subsegmental emboli constitute 6±36% of cases of pulmonary embolic disease [27±30]. Given the detection of just one patient (3.6%) with isolated subsegmental pulmonary emboli in our study it is likely that other cases were missed. Follow-up studies of untreated patients with initially negative test results found PE in similar proportions of patients regardless of the initial method of investigation with a frequency of 1.0±5.4% for CT, 0% for normal and 3.1% for low probability lung scintigrams and 0.6±4.2% for pulmonary angiography [31±38]. Although these ®gures suggest false negative diagnoses are uncommon, occurring with similar frequency in conventional and CT pulmonary angiography, they highlight the limitations of all the currently available imaging investigations for the detection of isolated subsegmental pulmonary emboli. It is therefore reasonable to consider further investigation or even empirical treatment despite a negative CTPA when the pre-test probability is high, particularly for those patients with limited cardiorespiratory reserve (Fig. 1). An additional limitation of this study was the potential delay in presentation to the CT unit with some patients receiving anticoagulation before investigation. However, we believe our practice re¯ects that found in the majority of UK hospital radiology departments. Furthermore D-dimer has been shown to remain elevated above the reference range for at least 7 days following the diagnosis of PE [13]. The impact of the delay is likely to a€ect the two tests (which were carried out together) in a similar, though not necessarily identical, way. There is an undoubted role for D-dimer testing in suspected acute venous thromboembolic disease. The

45

results of this and other studies demonstrate the value as well as the shortcomings of currently available D-dimer tests in the exclusion of pulmonary embolus. A D-dimer assay can be most usefully employed in the avoidance of further investigation when the clinical likelihood of PE is deemed to be low. However, when the pre-test probability of a PE is high, based on history, examination and cardiorespiratory measurements, a negative D-dimer should not prevent further investigation. The algorithm currently in use in our institution is shown in Figure 1. REFERENCES 1 PIOPED Investigators. The value of the ventilation/perfusion scan in acute pulmonary embolism: results of the prospective investigation of pulmonary embolism diagnosis (PIOPED). JAMA 1990;263:2753±2759. 2 Cooper TJ, Hayward MWJ, Hartog M. Survey on the use of pulmonary scintigraphy and angiography for suspected pulmonary thromboembolism in the UK. Clin Radiol 1991;43:243±245. 3 Burkill GJC, Bell JRG, Padley SPG. Survey on the use of pulmonary scintigraphy, spiral CT and conventional pulmonary angiography for suspected pulmonary embolism in the British Isles. Clin Radiol 1999;54:807±810. 4 Murchison JT, Gavan DR, Reid JH. Clinical utilisation of the nondiagnostic lung scintigram. Clin Radiol 1997;52:295±298. 5 BTS Working Party. Suspected acute pulmonary embolism: a practical approach. Thorax 1997;52(Suppl.4): S1±24. 6 Goodman LR, Lipchik RJ. Diagnosis of acute pulmonary embolism: time for a new approach. Radiology 1996;199:25±27. 7 Hansell DM. Spiral computed tomography and pulmonary embolism: current state. Clin Radiol 1997;52:575±581. 8 Stein PD, Huang H, Afzal A, Noor HA. Incidence of acute pulmonary embolism in a general hospital. Chest 1999;116: 909±913. 9 Stein PD, Henry JW. Prevalence of acute pulmonary embolism among patients in a general hospital and at autopsy. Chest 1995;108:978±981. 10 Charles LA, Edwards T, Macik BG. Evaluation of sensitivity and speci®city of six D-dimer latex assays. Arch Pathol Lab Med 1994;118:1102±1105. 11 Bounameaux H, de Moerloose P, Perrier A, Reber G. Plasma measurement of D-dimer as a diagnostic aid in suspected venous thromboembolism: an overview. Throm Haemostas 1994;71:1±6. 12 Bounameaux H, Schneider PA, Reber G, de Moerloose P, Krahenbuhl B. Measurement of plasma D-dimer for diagnosis of deep venous thrombosis. Am J Clin Pathol 1989;91:82±85. 13 Bounameaux H, Cira®ci P, de Moerloose P, et al. Measurement of D-dimer in plasma as diagnostic aid in suspected pulmonary embolism. Lancet 1991;337:196±200. 14 Freyburger G, Trillaud H, Labrouche S, et al. D-dimer strategy in thrombosis exclusion. Thromb Haemost 1998;79:32±37. 15 Bernardi E, Prandoni P, Lensing AWA, et al. D-dimer testing as an adjunct to ultrasonography in patients with clinically suspected deep vein thrombosis: prospective cohort study. BMJ 1998;317: 1037±1040. 16 Herold Ch J, Lamm C, Remy-Jardin M, et al. Prospective evaluation of pulmonary embolism: initial results of the European multicentre trial (ESTIPEP). ECR Scienti®c Programme and Abstracts 1999;Suppl. 1(9):226±227. 17 Egermayer P, Town GI, Turner JG, Heaton DC, Mee AL, Beard MEJ. Usefulness of D-dimer, blood gas, and respiratory rate measurements for excluding pulmonary embolism. Thorax 1998;53: 830±834. 18 van der Graaf F, van der Borne H, van der Kolk M, de Wild PJ, Janssen GWT, van Uum SHM. Exclusion of deep venous thrombosis with D-dimer testing: comparison of 13 D-dimer methods in 99 outpatients suspected of deep venous thrombosis using venography as reference standard. Thromb Haemost 2000;83: 191±198.

46

CLINICAL RADIOLOGY

19 van Beek EJR, Schenk BE, Michael BC, et al. The role of plasma D-dimer concentration in the exclusion of pulmonary embolism. Br J Haematol 1996;92:725±732. 20 Ginsberg JS, Wells PS, Kearon C, et al. Sensitivity and speci®city of a rapid whole blood assay for D-dimer in the diagnosis of pulmonary embolism. Ann Intern Med 1998;129:1006±1011. 21 Wells PS, Anderson DR, Rodger M, et al. Derivation of a simple clinical model to categorize patient probability of pulmonary embolism: increasing the model's utility with the SimpliRED D-dimer. Thromb Haemost 2000;83:416±420. 22 Perrier A, Desmarais S, Goerhing C, et al. D-dimer testing for suspected pulmonary embolus in outpatients. Am J Respir Crit Care Med 1997;156:492±496. 23 Tardy B, Tardy-Poncet B, Viallon A, et al. Evaluation of D-dimer ELISA test in elderly patients with suspected pulmonary embolism. Thromb Haemost 1998;79:38±41. 24 Mottier D, Couturaud F, Oger E, Leroyer C. Clinical usefulness of D-dimer tests in excluding pulmonary embolism is highly dependent upon age. Thromb Haemost 1998;80:527. 25 Stein PD, Henry JW, Gottshalk A. Reassessment of pulmonary angiography for the diagnosis of pulmonary embolism: relation of interpreter agreement to the order of the involved pulmonary arterial branch. Radiology 1999;210:689±691. 26 Baile EM, King GG, Muller NL, et al. Spiral computed tomography is comparable to angiography for the diagnosis of pulmonary embolism. Am J Respir Crit Care Med 2000;161: 1010±1015. 27 ACCP Consensus Committee on Pulmonary EmbolimOpinions regarding the diagnosis and management of venous thromboembolic disease. Chest 1998;113:499±504. 28 Gurney JW. No fooling around: direct visualisation of pulmonary embolism. Radiology 1993;188:618±619. 29 Stein PD, Henry JW, Relyea B. Untreated patients with pulmonary embolism: outcome, clinical, and laboratory assessment. Chest 1995;107:931±935.

30 Hull RD, Raskob GE, Pineo GF, Brant RF. The low probability lung scan: a need for change in nomenclature. Arch Intern Med 1995;155:1845±1851. 31 Goodman LR, Lipchik RJ, Kuzo RS, Liu Y, McAuli€e TL, O'Brien DJ. Subsequent pulmonary embolism: risk after a negative helical CT pulmonary angiogram ± prospective comparison with scintigraphy. Radiology 2000;215:535±542. 32 Garg K, Sieler H, Welsh CH, Johnston RJ, Russ PD. Clinical validity of helical CT being interpreted as negative for pulmonary embolism: implications for patient treatment. AJR Am J Roentgenol 1999;172:1627±1631. 33 Ferretti GR, Bosson JL, Bu€az PD, et al. Acute pulmonary embolism: role of helical CT in 164 patients with intermediate probability at ventilation-perfusion scintigraphy and normal results at duplex US of the legs. Radiology 1997;205:453±458. 34 Henry JW, Relyea B, Stein PD. Continuing risk of thromboemboli among patients with normal pulmonary angiograms. Chest 1995;107:1375±1378. 35 Stein PD, Athanasoulis C, Alavi A, et al. Complications and validity of pulmonary angiography in acute pulmonary embolism. Circulation 1992;85:462±468. 36 Hull RD, Hirsh J, Carter CJ, et al. Pulmonary angiography, ventilation lung scanning, and venography for clinically suspected pulmonary embolism with abnormal perfusion lung scan. Ann Intern Med 1983;98:891±899. 37 Cheeley R, McCartney WH, Perry JR, et al. The role of noninvasive tests versus pulmonary embolism. Am J Med 1981;70: 17±22. 38 Novelline RA, Baltarnovich OH, Athanasoulis CA, Waltman AC, Green®eld AJ, McKusick KA. The clinical course of patients with suspected pulmonary embolism and a negative pulmonary arteriogram. Radiology 1978;126:561±567.