An evaluation of D-dimer in the diagnosis of pulmonary embolism: A randomized trial

Background: It may be safe to omit additional diagnostic testing in selected patients with suspected pulmonary embolism (PE) who have a negative D-dimer test, but this approach has never been evaluated in a randomized, controlled trial. Objective: To determine if additional diagnostic testing can be safely withheld in patients with suspected PE who have negative erythrocyte agglutination D-dimer test results. Design: Randomized comparisons in 2 subgroups of a prospective multicenter study. Setting: 7 university hospitals. Patients: 1126 outpatients or inpatients with suspected PE; of these, 456 patients with negative erythrocyte agglutination D-dimer test results were randomly assigned to the intervention groups. Patients were classified into 2 clinical probability groups: those with a low clinical probability of PE (low-probability group) and those with a moderate or high clinical probability of PE, a nondiagnostic ventilation-perfusion lung scan, and no evidence of proximal deep venous thrombosis on bilateral ultrasonography (moderate- or high-probability group). Interventions: The experimental intervention for both probability groups was no further diagnostic testing for PE. The control intervention for the low-probability group was a ventilation-perfusion lung scan followed by ultrasonography of the proximal deep veins of the legs on the same day. If the lung scan was nondiagnostic, ultrasonography of the legs was repeated 7 and 14 days later. The control intervention for the moderate- or high-probability group was ultrasonography of the proximal deep veins of the legs after 7 and 14 days. In the control and experimental groups, anticoagulation was withheld or withdrawn if PE was not diagnosed. Measurements: Symptomatic venous thromboembolism (VTE) during 6 months of follow-up. Results: Prevalence of VTE was 15.2% in the 1126 enrolled patients. In the low-probability group, VTE occurred during follow-up in 0 of 182 patients who had no additional diagnostic testing and in 1 of 185 patients who had additional testing (difference, -0.5 percentage point [95% CI, -3.0 to 1.6 percentage points]). In the moderate- or high-probability group, VTE occurred during follow-up in 1 of 41 patients who had no additional diagnostic testing and in 0 of 41 patients who had additional testing (difference, 2.4 percentage points [CI, -6.4 to 12.6 percentage points]). Limitations: The authors could not enroll 2000 patients as originally planned; 3 randomly assigned patients did not receive the allocated intervention, and 7 received inadequate follow-up. Personnel who performed follow-up evaluations were not blinded to the results of diagnostic testing at enrollment or to allocation group assignments. Conclusion: In patients with a low probability of PE who have negative D-dimer results, additional diagnostic testing can be withheld without increasing the frequency of VTE during follow-up. Low clinical probability and negative D-dimer results occur in 50% of outpatients and in 20% of inpatients with suspected PE.