The main hypothesis for this study is that the diagnostic approach of pulmonary embolism (PE) by evaluation of clinical probability, D-dimer test dosage and multi-detector helical computed tomography (hCT) is as safe as the classical "approach" using clinical probability, D-dimer test, lower limb compression ultrasonography and multi-detector helical computed tomography.

The second hypothesis involves evaluating the role of searching distal, i.e. infrapopliteal, deep venous thromboses (DVTs) in the diagnostic approach of pulmonary embolism.

Suspected pulmonary embolism [PE] is a frequent clinical problem and remains a diagnostic challenge. The diagnostic approach of PE relies on sequential diagnostic tests, such as plasma D-dimer measurement, lower limb compression ultrasonography, ventilation-perfusion lung scan or helical computed tomography [hCT] and pulmonary angiography. In addition, the diagnostic workup is usually stratified according to the clinical probability of pulmonary embolism.

First-generation hCTs were based on a single-detector technique and had a limited 70% sensitivity. Moreover, in two recent outcome studies, a proximal deep venous thrombosis diagnosed by lower limb compression ultrasonography was found in 15% of patients with a normal single-detector hCT. However, in management studies, the association of a normal lower limb compression ultrasonography and a normal single-detector hCT has been proved safe to rule out PE in patients with a non-high clinical probability, with a less than 2% rate of thromboembolic events during a 3-month follow-up in patients left untreated based on that combination. Hence, lower limb compression ultrasonography must be combined with a single-detector CT to safely rule out pulmonary embolism.

More recently, the multi-detector hCT has become widely available. These new scanners allow one to improve image definition and to diminish slice thickness without increasing acquisition time and, hence, to better visualize the segmental and sub-segmental pulmonary arteries. In a recent study [CTEPm], the investigators evaluated the performances of a multi-detector hCT. The rate of false negative results, assessed by the rate of patients with proximal DVT on ultrasonography and a negative hCT, was much lower using the multi-detector CT [0.9%, 95% confidence interval: 0.3-3.3%] than with the single-detector technique [15%]. This raises the possibility that hCT might be used as a single test to rule out PE in patients with an elevated D-dimer level. Indeed, ELISA plasma D-dimer measurement in patients with a non-high clinical probability of PE remains a logical first step since it safely rules out PE in approximately 30% of patients. Such a strategy should now be validated in a prospective randomised outcome study to compare its safety and cost-effectiveness with that of a previously validated strategy combining D-dimer, lower limb compression ultrasonography and hCT.

Therefore, the investigators plan to compare the efficacy and cost-effectiveness of two distinct diagnostic strategies for suspected pulmonary embolism in emergency center patents in a non-inferiority trial. The strategies are: 1) an algorithm based on clinical assessment, D-dimer and multi-detector hCT (DD-hCT); and 2) a previously validated strategy that includes lower limb compression (DD-US-hCT). The main outcome will be the rate of thromboembolic events during a formal 3-month follow-up in patients not anticoagulated on the basis of a PE ruled out by the diagnostic criteria in each study arm. A secondary outcome will be the costs of diagnosis in each study arm, including the costs of suspected and/or confirmed thromboembolic events during follow-up.

In addition, the investigators aim to prospectively and pragmatically validate the revised Geneva score for assessing the clinical probability of PE by using it in the sequential diagnostic work-up for PE. Classification performances of this score will be assessed by checking its capacity to discriminate between patients having a low, intermediate, or high probability of PE.

Finally, in a nested study, the investigators aim to compare the clinical usefulness of a complete lower limb compression ultrasonography versus a classical "4-points ultrasonography" (ultrasonography of the proximal veins) in the patients from the ultrasound arm of the study (DD-US-hCT). Indeed, preliminary data indicate that complete ultrasound detects a DVT in a higher proportion of patients with PE than proximal ultrasound (US) and, hence, has the potential for further limiting the requirement for chest imaging in patients with suspected PE. Because finding a distal DVT is of uncertain clinical significance, such a finding will not be reported to the clinicians in charge of the patient and those patients will undergo a helical CT and be treated (or not) according to the CT result. Outcome measurements will be the number of patients in whom hCT could have been avoided by ruling in the diagnosis of PE in the presence of a distal DVT (proportion of patients with both distal DVT on ultrasonography and pulmonary embolism on multi-detector hCT) and the number of patients who would have been unduly anticoagulated if considering the presence of a distal DVT in the complete compression ultrasonography (proportion of patients with distal DVT on ultrasonography, but without pulmonary embolism on multi-detector hCT, and without a thromboembolic event during the 3 month follow-up).

Inclusion Criteria:

• All outpatients admitted to the emergency ward for suspected pulmonary embolism, defined as acute onset of new or worsening shortness of breath or chest pain without another obvious etiology, will be included in the study, provided they meet all other criteria and they have signed an informed consent form.

Exclusion Criteria:

• PE suspicion raised more than 24 hours after admission to the hospital
• Absence of informed consent
• Life expectancy of less than 3 months
• Geographic inaccessibility for follow-up
• Pregnancy
• Patients anticoagulated for a disease other than venous thromboembolism (for instance, atrial fibrillation)
• Patients allergic to contrast medium
• Impaired renal function (creatinine clearance less than 30 ml/min, as calculated by the Cockroft formula)