D-Dimer thresholds for diagnosis of pulmonary embolism based on a single question: is it the most likely diagnosis? A prospective, multicentre, open-label, single-arm interventional study

Abstract

Background: Validated diagnostic strategies for pulmonary embolism allow raising the D-dimer threshold to 1000 ng/mL in selected patients with a low probability of pulmonary embolism, but adherence to this strategy in routine practice remains poor. Retrospective studies suggest that these rules could be replaced by a single clinical question as to whether pulmonary embolism is the most likely diagnosis. We aimed to assess the safety of a simplified approach that applies a 1000 ng/mL threshold when pulmonary embolism is unlikely and an age-adjusted threshold otherwise, including patients at a high probability for pulmonary embolism.

Methods: In this prospective interventional study in 13 French hospital emergency departments, patients aged ≥18 years with a clinical suspicion of pulmonary embolism, who were not on full-dose anticoagulant therapy and had not had a thromboembolic event in the past 6 months, were included. The intervention consisted of ruling out pulmonary embolism without chest imaging in patients for whom pulmonary embolism was not considered the most likely diagnosis and had a D-dimer value <1000 ng/mL. The age-adjusted D-dimer threshold (500 ng/mL if aged <50 years or age × 10 ng/mL if aged ≥50 years) was used for the other patients. The primary outcome was the diagnostic failure rate (ie, occurrence of a thromboembolic event at 3 months of follow-up) among patients for whom pulmonary embolism was initially ruled out. Safety was established if the upper bound of the two-sided 95% CI for the diagnostic failure rate was less than 1·85% in patients for whom pulmonary embolism was ruled out at initial testing and in a subgroup of patients for whom it was ruled out without chest imaging. This study is registered with ClinicalTrials.gov (NCT06190392) and is complete.

Findings: Between Jan 16 and Sept 5, 2024, 1365 patients were screened for eligibility and 1221 were included, of whom 80 (7%) were diagnosed with pulmonary embolism at initial testing. Pulmonary embolism was not considered the most likely diagnosis for 997 patients. The diagnostic failure rate was 0·00% (95% CI 0·00-0·34) and was 0·12% (0·01-0·55) after multiple imputation for 33 patients with no available follow-up, both within the safety threshold. In the prespecified subgroup of 796 patients for whom pulmonary embolism was ruled out without chest imaging, the diagnostic failure rate was 0·00% (0·00-0·46), also within the safety threshold. The strategy had a 32% chest imaging rate (384 of 1217 patients), an absolute reduction of 19% (16-21) compared with the fixed 500 ng/mL D-dimer threshold strategy (609 [50%] of 1215). There were 16 deaths during the 3-month follow-up.

Interpretation: A global simplified strategy using a D-dimer threshold of 1000 ng/mL in patients for whom pulmonary embolism was not the most likely diagnosis, and an age-adjusted threshold for other patients, safely excluded the diagnosis. These findings indicate that this simplified strategy could safely reduce the use of chest imaging in the emergency department.

Funding: Assistance Publique-Hôpitaux de Paris (Délégation à la Recherche Clinique et à l'Innovation).