Variable D-dimer thresholds for diagnosis of clinically suspected acute pulmonary embolism

Computed tomography pulmonary angiography (CTPA) is frequently requested using diagnostic algorithms for suspected pulmonary embolism (PE). For suspected deep vein thrombosis, it was recently shown that doubling the D-dimer threshold in patients with low pretest probability safely decreased the number of ultrasonograms. We evaluated the safety and efficiency of a similar strategy in patients with suspected PE. We performed a post-hoc analysis of 2213 consecutive patients of two cohort studies with suspected PE who were managed according to current standards: PE ruled out in case of unlikely probability (Wells rule ≤ 4 points) and a D-dimer level < 0.5 μg mL(-1) . CTPA was performed in all other cases. All patients were followed for 3 months. We calculated 3-month venous thromboembolism (VTE) incidence and the number of required CTPAs for selective D-dimer thresholds in patients with low clinical probability ( < 2 points, D-dimer threshold < 1.0 μg mL(-1) ) and intermediate probability (2-6 points, D-dimer threshold < 0.5 μg mL(-1) ). Using standard management, PE could be excluded without CTPA in 26% of patients, with a 3-month VTE incidence of 0.88% (95% confidence interval [CI] 0.29-2.1%). Using selective D-dimer thresholds, PE could be excluded without CTPA in 36% of patients, with a 3-month VTE incidence of 2.1% (95% CI 1.2-3.4%) in patients managed without CTPA, an increase of 1.2 percentage points (95% CI -0.3 to 2.2). Applying selective D-dimer thresholds reduces the need for CTPA by 11 percentage points but is associated with an increased failure rate. Prospective studies should evaluate the safety and net clinical benefit of this approach