European Radiology

, Volume 14, Issue 1, pp 93–98

Value of negative spiral CT angiography in patients with suspected acute PE: analysis of PE occurrence and outcome


    • Department of RadiologyUniversity Hospital of Vienna-AKH
  • N. Klein
    • Department of RadiologyUniversity Hospital of Vienna-AKH
  • D. Fleischmann
    • Department of RadiologyUniversity Hospital of Vienna-AKH
  • A. Kaneider
    • Department of RadiologyUniversity Hospital of Vienna-AKH
  • C. Novotny
    • Department of Nuclear MedicineUniversity Hospital of Vienna-AKH
  • S. Kreuzer
    • Department of RadiologyUniversity Hospital of Vienna-AKH
  • C. Riedl
    • Department of RadiologyUniversity Hospital of Vienna-AKH
  • E. Minar
    • Division of Angiology, Department of Internal MedicineUniversity Hospital of Vienna-AKH
  • K. Janata
    • Department of Emergency MedicineUniversity Hospital of Vienna-AKH
  • C. J. Herold
    • Department of RadiologyUniversity Hospital of Vienna-AKH

DOI: 10.1007/s00330-003-2016-3

Cite this article as:
Krestan, C.R., Klein, N., Fleischmann, D. et al. Eur Radiol (2004) 14: 93. doi:10.1007/s00330-003-2016-3


The aim of this study was to analyze pulmonary embolism (PE) occurrence and retrospective clinical outcome in patients with clinically suspected acute PE and a negative spiral CT angiography (SCTA) of the pulmonary arteries. Within a 35-month period, 485 consecutive patients with clinical symptoms of acute PE underwent SCTA of the pulmonary arteries. Patients with a negative SCTA and without anticoagulation treatment were followed-up and formed the study group. Patient outcome and recurrence of PE was evaluated retrospectively during a period of 6 months after the initial SCTA, and included a review of computerized patient records, and interviews with physicians and patients. Patients were asked to fill out a questionnaire concerning all relevant questions about their medical history and clinical course during the follow-up period. Special attention was focused on symptoms indicating recurrent PE, as well as later confirmation and therapy of PE. Of the 485 patients, 325 patients (67%) had a negative scan, 134 (27.6%) had radiological signs of PE, and 26 (5.4%) had an indeterminant result. Of 325 patients with a negative scan, 269 (83%) were available for follow-up. The main reasons for loss to follow-up were change of address, name, or phone number, or non-resident patients who left abroad. Of 269 patients available for follow-up, 49 patients (18.2% of 269) received anticoagulant treatment because of prior or recent deep venous thrombosis (32.6%) or a history of PE (34.7%), cardiovascular disease (18.4%), high clinical probability (8.2%), positive ventilation–perfusion scan (4.2%), and elevated D-dimer test (2%). The remaining 220 patients, who did not receive anticoagulant medication, formed the study group. Of this study group, 1 patient died from myocardial infarction 6 weeks after the initial SCTA, and the postmortem examination also detected multiple peripheral emboli in both lungs (p=0.45%; 0.01–2.5, 95% confidence interval). The PE did not occur in any other patient. In patients with suspected PE and negative SCTA without anticoagulant therapy, the risk of recurrent PE in this study was less than 1% and similar to that in patients after a negative pulmonary angiogram. Therefore, we conclude that patients can be managed safely without anticoagulation therapy; however, this approach may not be appropriate for critically ill patients and those with persistent high clinical suspicion of acute PE.


Pulmonary embolismHelical computed tomographyClinical effectiveness


Pulmonary embolism is a common, potentially fatal disease [1]. In the majority cases, blood clots originate from the deep venous system of the lower extremities; thus, PE and deep venous thrombosis (DVT) can be considered manifestations of the same process, namely, venous thromboembolism (VTE). The diagnosis of DVT or acute PE generally results in immediate treatment with local or systemic thrombolysis and secondary prophylaxis with oral Vit-K-antagonists over a time period of 3–6 months. Since this treatment is associated with a considerable risk of bleeding complications, therapy should be strictly reserved for patients with an accurate diagnosis of VTE. On the other hand, withholding of therapy in patients with undiagnosed PE results in a significantly increased risk of a potentially fatal recurrent episode of PE [2]; therefore, any diagnostic method for PE should demonstrate a sensitivity, specificity, and positive and negative predictive values that are high enough to keep over- and undertreatment to a minimum.

The diagnostic approach in patients with suspected PE has changed over the past several years. With the introduction of spiral CT angiography (SCTA), the role of ventilation–perfusion (V/Q) scanning as a primary diagnostic tool has been increasingly questioned and, in most institutions, SCTA of the pulmonary arteries has replaced V/Q scanning. It could also be argued that SCTA has reduced the need for pulmonary angiography, which was used to resolve questions in cases of non-diagnostic VQ scans [3]. The advantages of SCTA are its non-invasiveness, cost-effectiveness, its ability to provide information about alternative problems that can mimic PE [4], and, most importantly, the fact that SCTA provides a definitive diagnosis in a significantly larger proportion of patients than V/Q scanning (92 vs 72%) [5]. The most important limitation is SCTA's limited accuracy in the detection of clots in subsegmental peripheral pulmonary vessels; however, the clinical relevance of this limitation remains unclear. It has been reported that isolated subsegmental emboli may be the precursor of a second larger embolic event, and that their pathophysiologic impact is larger in patients with preexisting cardiorespiratory disease [6]. In prospective studies, the number of patients with isolated subsegmental PE was small and ranged between 17 and 22% of patients with a positive SCTA [7, 8]. Pulmonary angiography as a reference method also shows limited accuracy in the evaluation of peripheral vessels, and consequently, its role as the gold standard has been questioned [9]. As a result, the sensitivity of SCTA in the identification of peripheral PE and, as a consequence, the value of SCTA in excluding PE, is difficult to determine.

An alternative approach to determine the clinical value of an imaging method is the assessment of patient outcome. Outcome analysis does not compare the accuracy of different diagnostic methods, but provides information on the follow-up of a given patient with a certain test result, thus providing indirect clues about the accuracy of the primary test result. Outcome analysis in patients with clinically suspected PE and a negative CTA result provides information about the accuracy of the primary negative test result. If a patient with a negative test result remains free of symptoms suggestive of PE, or if PE can be excluded through repeated examinations or alternative methods, then this would validate the result of the primary test, at least from the clinical-outcome standpoint. The most appropriate way to assess the risk of recurrent PE is to monitor a large cohort of patients during a defined follow-up period. Follow-up studies have shown that the risk of subsegmental thromboemboli in patients with a normal pulmonary angiogram or a normal V/Q scan is low and has been reported to be between 0 and 1.6% [10, 11]. Previous outcome studies of patients with a negative SCTA and clinically suspected PE found a 0.5–5% incidence of PE within a 6- to 12-month period [12, 13].

In this study, we set out to investigate the occurrence of PE in patients with an initially negative SCTA, and thereby to obtain medical information on the validity of the negative test result.

Materials and methods

Study group

Within a 35-month period, 485 patients (mean age 53.5 years; 245 women, 240 men) underwent SCTA of the pulmonary arteries for clinically suspected acute PE at our institution. There were 316 patients (65.2%) referred from internal medicine and surgical wards and 169 patients (34.8%) referred from the emergency department and other outpatient centers. Eight (2.5%) inpatients were critically ill and referred from the ICU, and they all had a negative SCTA; 42 had a history of malignant disease and 143 had a history of cardiovascular disease.

In addition to SCTA, patients underwent further or additional diagnostic work-up as follows: Of all 485 patients, 165 patients underwent V/Q scanning and SCTA within 7 days or less. In 106 patients, V/Q scanning had been used as the first-line examination; of these, 20 had a normal, 13 a nearly normal, 32 a low-probability, 18 an intermediate-probability, and 23 a high-probability V/Q scan. Fifty-nine patients had the V/Q scan after the SCTA. In 4 patients a pulmonary angiogram was performed and was classified as normal. In the patient group with a negative SCTA (269 patients), 13 patients underwent venography of the lower leg veins, and 16 patients underwent duplex sonography. No venous examinations were performed in patients without a previous history or clinical symptoms of DVT.

Of the total number of 485 SCTA exams (Fig. 1), 325 were classified as negative, 134 ( 28%) as positive, and 26 (5%) as indeterminant, respectively. In 269 of the 325 patients (83%) with a negative SCTA result, detailed information pertinent to the questionnaire, and thus to the determination of clinical outcome, could be sampled retrospectively. Fifty-six patients were lost to follow-up. The main reasons for loss to follow-up were change of address, name, or phone number, or non-resident patients who left abroad. The distribution of age, gender, and patient status of patients with and without follow-up is given in Table 1.
Fig. 1.

Flow chart to illustrate patients and subgroups from the initial total number of 485 SCTAs to the core study sample of 220 untreated patients with follow-up data

Table 1.

Differences in patient characteristics with negative SCTA±follow-up (absolute values; t test, χ square) n.s. not significant

269 patients with follow-up (years)

56 patients lost to follow-up (years)

Age (p<0.05)



Male (n.s.)

141 (52)

26 (46)

Female (n.s.)

128 (48)

30 (54)

Inpatient (n.s.)

181 (67)

29 (52)

Outpatient (n.s.)

82 (31)

25 (46)

Critically ill (n.s.)

6 (2)

2 (2)

Numbers in parentheses are percentages

Of 269 patients, 49 (18.2%) were anticoagulated despite a negative SCTA and were subsequently excluded from data analysis. Sixteen percent (n=8) of these 49 patients had a history of prior DVT and 16% (n=8) patients had recent symptomatic DVT, which was proven in 7 patients by venography, and in one patient by positive duplex sonography. Other reasons were history of PE in 35% (n=17), cardiovascular disease in 18% (n=9), a persistent high clinical probability of PE in 8% (n=4), a high-probability V/Q scan in 4% (n=2), and an elevated D-dimer in 2% (1 patient).

Our study group consisted of 220 patients with a negative SCTA result for PE, who were not treated with anticoagulation and for whom clinical follow-up data were available over a period of at least 6 months (Fig. 1). Of this core study group of 220 patients, 59 (26.8%) had a V/Q scan. Distribution of the V/Q scan results were: 19 (8.6%) with a normal V/Q scan; 9 (4%) with a nearly normal V/Q scan; 21 (9.5%) with a low-probability; 6 (2.7%) with an intermediate probability; and 4 (1.8%) with a high-probability V/Q scan. Four patients had a pulmonary angiogram, which was reported negative. In the untreated patient group with a negative SCTA exam, 3 patients underwent duplex sonography of the lower legs and 11 patients underwent a venography.

CT evaluation

All patients underwent an SCTA scan of the pulmonary arteries. The SCTA scans were obtained on a helical CT scanner (Philips Tomoscan SR 7000, Philips, Best, The Netherlands). The following scanning parameters were used: 5-mm collimation; pitch=1; and 3-mm reconstruction increment. In all patients, 120 ml (300 mg iodine/ml) of contrast agent were injected by a power injector at a rate of 2.0–2.7 ml/s through a 20-G needle into the antecubital vein. A standard delay of 20 s was used until scanning was started. The scanning direction was caudal to cranial, reaching from the lung base to the top of the aortic arch. Images were reconstructed with a standard algorithm, a soft tissue window (50/300 HU), and a lung window (–500/1500 HU).

Analysis of CT data

All examinations were read on hard copies by observers with a minimum of 5 years experience with body CT and 3 years of experience reading SCTA of the pulmonary arteries. The SCTA was classified as positive, if a partial or complete filling defect or a mural defect was present in at least one pulmonary artery. The SCTA scans were rated negative if optimally contrasted pulmonary arteries were visible without CT signs of acute pulmonary embolism; SCTA scans were judged indeterminate if the technical quality of the scans was insufficient or the ambiguity of findings (partial-volume defects, irregular arterial walls, abrupt cutoff of segmental artery branches) did not allow for a definitive diagnosis [14].

Computed tomography demonstrated additional findings or an alternative diagnosis in 143 (65%) of 325 patients with a negative SCTA: infiltrates in 36; atelectasis in 16; pleural effusion in 47; pleural effusion and atelectasis in 18; intrapulmonary lesions in 10; bronchial carcinoma in 1; sequestration in 1; hamartoma in 1; and an abnormal diameter of the main pulmonary arteries in 13 patients, suggestive for pulmonary arterial hypertension.

Clinical follow-up

Patients with a negative SCTA, and without anticoagulation treatment, were evaluated by retrospective clinical follow-up, which lasted 6 months after the last patient had been enrolled in the study. The questionnaire (Table 2) was designed to gather the best possible information on the patient's clinical course, PE recurrence, and outcome. The follow-up consisted of computerized searches in medical records at our institution to obtain data regarding VT or DVT and concomitant diseases. Data from the National Bureau of Statistics were obtained concerning death certificates. Autopsy reports were reviewed with regard to the diagnosis of PE. In addition, patients themselves or the responsible family physician were interviewed by phone using a standardized questionnaire. The questionnaire was mailed to patients who could not be reached by telephone. Anticoagulant medication during the follow-up period was also recorded. In 269 patients follow-up data were available, but 56 (17%) of the 325 patients with a negative SCTA exam were lost to follow-up analysis.
Table 2.

Standard patient questionnaire with seven questions regarding clinical information about deep venous thrombosis or pulmonary embolism and other concomitant diseases as well as anticoagulant therapy

Pulmonary embolism patient questionnaire


Did you have symptoms of pulmonary embolism (unexplained dyspnea, chest pain, coughing of blood, increased breathing frequency, dilatation of neck veins, fever) or a proven pulmonary embolism (verified by your physician) during the first 6 months after the initial CT scan of your chest or prior to it?


Did you have a history of deep venous thrombosis (leg swelling, pain) within 6 months after your CT scan or prior to it?


Did you have specific medical tests such as a lung scan, pulmonary angiography, or ultrasound/contrast study of your legs within 6 months after your CT scan? If yes, what were the results of these tests?


Following the CT scan of your chest, did you receive anticoagulant medication?


Did you have recent surgery prior to your CT scan (<4 weeks)?


Did you have other severe diseases (malignant disease)?


Are you suffering from cardiac diseases?


All data were entered into a database and descriptive statistics of the study population were performed. The incidence of subsequent PE and DVT was calculated from the study data. The probability for developing PE or DVT was estimated by calculating the percentage of patients with subsequent venous thromboembolism including the 95% confidence interval for this estimate.


There were 220 patients included for final analysis (=study group; Fig. 1) The mean age was 55.2 years (range 17–96 years), 118 (54%) were men and 102 (46%) were women. Sixty-nine patients died during the follow-up period. Their mean age was 63.5±14 years (43 men and 26 women). The cause of death as specified on the death certificate was cardiovascular disease in 44 (63.8%), malignant disease in 16 (23.2%), septicemia in 6 (8.7%), liver failure in 2 (2.9%), and chronic obstructive pulmonary disorder in 1 (1.4%) patient. Autopsies were performed in 20 patients (29%). In 1 patient autopsy revealed clots in multiple pulmonary arteries and DVT in the lower extremities. The patient was a 50-year-old women who suffered from coronary heart disease. She underwent coronary bypass surgery and died from acute myocardial infarction. Six weeks before her death, she had undergone an SCTA examination that was also retrospectively classified as normal. This was the only patient who could be identified with recurrent PE. No other cases of venous thromboembolism were diagnosed, neither in the group of the 151 surviving patients nor in the group of patients who died within the follow-up period. This amounted to a PE rate of 0.45% (95% confidence interval: 0.01–2.5%).


The aim of our study was to determine whether it is safe to withhold anticoagulation therapy in patients with clinical suspicion of PE but negative SCTA examinations. The final study group evaluated consisted of 220 patients who had a negative SCTA and did not undergo treatment. Forty-two of these 220 patients had an additional normal or near normal V/Q scan or a negative venous study, and 4 patients had undergone a negative pulmonary angiogram.

By assessing the clinical outcome of these patients over a time period of at least 6 months, we wanted to determine the incidence of recurrent PE that would have been avoided by appropriate treatment. The ultimate goal was to achieve a definitive answer about how reliably SCTA could, using a single-detector technique, exclude clinically relevant PE. A low incidence of PE would allow the clinician to refuse anticoagulant medication after a negative SCTA examination.

The analysis of clinical outcome and incidence of recurrent PE in follow-up periods has received increasing attention during the past several years. Initially, such analyses were applied to validate the clinical use of diagnostic methods such as V/Q scanning and pulmonary angiography. For V/Q scanning, it has been demonstrated that a negative test result has the same value in ruling out embolism as does a normal pulmonary angiogram. The conclusion was that the diagnostic work-up for suspected PE does not need to be extended beyond a normal perfusion scan. Anticoagulant therapy can be discontinued, except in the presence of documented venous thrombosis or persistent high clinical suspicion of PE and/or in critically ill patients [15]. This conclusion is based on a series of investigations: Henry et al. found that, for patients with suspected acute PE and nearly normal V/Q scanning in the PIOPED trial, no patients in the outcome group had clinically evident PE. The authors, however, alluded to the fact that the study could not determine whether the finding of the absence of PE was due to a low prevalence of DVT from the outset, or due to the fact that some patients did well with an undiagnosed PE and without treatment [16]. Garg et al. [17] reported no PE or DVT occurrence after a negative normal V/Q scan and a 3.8% recurrence after a very-low/low-probability V/Q scan in their 6-month follow-up study. Van Beek et al. [18] demonstrated that in patients with clinical suspicion of acute PE and a negative perfusion lung scan, anticoagulation therapy could be safely withheld, with no recurrent PE during a 6-month follow-up period.

A number of studies have targeted the outcome after a negative pulmonary angiogram. These trials showed that the subsequent risk of thromboemboli among patients with a normal pulmonary angiogram appears to be very low. In 54 patients followed over 12 months after a negative pulmonary angiogram and with a high clinical suspicion of acute PE, no documented PE was found [19]. Another series showed an incidence of 1.6% of PE recurrence within 12 months in a population of 380 patients [10].

In our retrospective analysis, we attempted to determine the value of a negative SCTA examination. In our cohort of patients, the incidence of recurrent symptomatic PE or DVT was less than 1%: only 1 of 220 patients followed after a negative SCTA and without further anticoagulant treatment showed evidence of PE as proven by autopsy. In that respect, our results are in good agreement with previously published results that report an incidence of PE ranging from below 1 to 5.4%. Some of these studies included a multiple of our study population and suffered from only low exclusion rates. Four of five studies also had only a limited number of patients who also underwent an examination of the lower leg veins.

Goodman et al. [13] analyzed the outcome of 198 patients with a negative SCTA, and found a 1% risk of PE after a negative SCTA in untreated patients. The reasons for anticoagulation in 63 patients were not given in the paper. Similar to our study group, conclusions about the value of a negative SCTA are limited by the fact that as many as 42% of patients had ultrasonography of the lower limbs. The percentage of patients lost to follow-up after inclusion was 9.5%. Gottsater et al. [20] followed 215 patients after negative SCTA and found an incidence of 1.4% recurrent PE during a period of 3 months. This study was performed in a retrospective setting; the authors lost 1.2% of patients to follow-up and the number of patients with venous studies was also small (10 of 215 patients) [20]. The most revealing study found a recurrence rate of less than 1% in 512 patients with prospective follow-up. The authors concluded that the addition of compression ultrasonography to SCTA would be of limited value [21].

Swensen et al. [22] followed 993 patients and found a cumulative incidence of DVT in 0.5% of patients. All SCTA scans were performed with an electron beam scanner and 3-mm collimation. They found DVT in only 8 patients. The use of ultrasonography of the lower leg veins was not mentioned in this study. Only 1.9% of patients in this cohort were lost to follow-up during this 3-month period. Ferretti et al. [12] performed the only study in which all patients underwent additional examination of the leg veins. In their prospective study, they found an incidence of 5.4% subsequent DVT or PE during 3 months of follow-up; however, this study group excluded the analysis of subsegmental emboli. Of 112 patients, three had positive findings at repeated duplex ultrasound of the calf after negative SCTA. During 3 months of follow-up, 2 patients had a high-probability lung scan and 1 patient had a positive Doppler US of the legs. The major difference between Ferretti et al.'s [12] study and ours is that their study included only patients with intermediate probability (inconclusive) V/Q scans and negative duplex ultrasound of the lower extremities. In addition, prospective follow-up was performed with US and a repeated V/Q scan at 3 months. In our study, we included all patients with negative SCTA and clinical suspicion of PE; however, duplex studies were performed only in a limited number of patients and follow-up was retrospective.

Our study suffers from several limitations: Firstly, the follow-up was performed in a retrospective setting and a relatively high number of patients, namely 17%, had to be excluded because of unavailability of follow-up information. Secondly, only 29% of the 69 patients who died in the follow-up period underwent an autopsy. In this subgroup of patients, PE could have been missed in an unknown proportion of patients. This subgroup makes up 18% of the core study population with a negative SCTA and is larger than in Goodman et al's. [13] study, but comparable to the study by Gottsater et al. [20]. Thirdly, only a minority of patients underwent examinations of the lower leg veins. The incidence of DVT in patients with negative SCTA, and clinical suspicion of PE but no direct symptoms of DVT, has been reported to be up to 6% [23]. Fourthly, the SCTA scans in our study were performed with a large collimation of 5 mm and a single detector technique. It is well known that thinner collimation and multidetector CT significantly improves the analysis of peripheral arteries, but this might result in diagnosing very small clots that do not need treatment, so one might argue that in fact a larger collimation is an advantage rather than a limitation in our study, although it is also noted that the clinical value of subsegmental PE remains undetermined and the indication for anticoagulation in this subgroup has not yet been determined [6].

Despite these limitations, our results appear to have validity, as they are in line with those of Goodman et al. [13], Swensen et al. [22], Garg et al. [17], and Gottsater et al. [20].


In conclusion, we found a very low incidence of PE after negative SCTA for suspected acute PE in patients who did not receive anticoagulant treatment after the initial SCTA during a follow-up period of 6 months. Our results may add more weight to the body of evidence that suggests that it is safe to withhold patients from antithrombotic treatment after a negative helical CT scan because the risk of clinically evident PE occurrence after a negative CT scan is minimal and similar to the risk of PE after a negative or a nearly normal V/Q scan or a negative pulmonary angiography. Exceptions may be patients with unresolved high clinical suspicion of PE or critically ill patients, in whom withholding of anticoagulants has not yet been studied in larger cohorts of patients. The SCTA seems to be similarly effective in excluding clinically significant pulmonary emboli compared with pulmonary angiography and can prevent unnecessary further treatment and direct safe patient management.

Copyright information

© Springer-Verlag 2004