OBJECTIVES:To validate the safety profile of a rapid ELISA D-dimer as the first diagnostic step in the clinical suspicion of pulmonary embolism (PE) in outpatients admitted to an emergency department (ED), and to retrospectively evaluate the appropriateness of the physician's prescription.
DESIGN AND SETTING:An observational study of all patients admitted to the ED of an urban university teaching hospital with signs and symptoms justifying the prescription of a rapid ELISA D-dimer measurement (Vidas; Biomerieux; France) as the first line diagnostic test for PE. Acute PE was established or excluded according to an appropriate combination of the D-dimer concentration, the lung scintigraphy, the spiral computerized tomography (spiral CT), the venous ultrasonography, and the arteriography in case of uncertain results. All patients with D-dimer values under the cut-off point of 500 ng/ml were followed up after 6 months.
RESULTS:395 patients were studied. A normal D-dimer concentration < 500 ng/ml was found in 179 patients (45% of the cohort). The retrospective analysis showed that none of these patients were found to have a high pre-test clinical probability. None of these 179 patients received anticoagulation nor displayed a PE event during a 6-month period (negative predictive value 100%; 95% CI, 98.0 to 100%; sensitivity 100%; 95% CI, 90.3 to 100%). Among the 216 patients (55%) with D-dimer values above 500 ng/ml, PE was confirmed in 32 cases, for a prevalence of the disease of 8.1%. Eighty-six patients (22%) had no additional testing in spite of positive D dimer values > 500 ng/ml, pointing out a 22% rate of inappropriate use of the D-dimer measurement.
CONCLUSION:This observational study confirms that a normal rapid ELISA D-dimer value (< 500 ng/ml) used as a first diagnostic step in ruling out the diagnosis of PE is a safe clinical practice when the pre-test clinical probability is low or intermediate. Nevertheless, the low prevalence rate of the disease (8.1%) suggests a potential overused and inappropriate prescription.