The diagnostic performance of the YEARS and Pulmonary Embolism Graduated D-dimer algorithms in patients with prior venous thrombosis suspected of pulmonary embolism.

Background: Evidence on how diagnostic algorithms perform in patients with prior venous thromboembolism (VTE) is scarce. Objectives: This study determined the diagnostic miss rate during 3-month follow-up and the proportion of avoided imaging using the YEARS and Pulmonary Embolism Graduated D-dimer (PEGeD) algorithms. Methods: This preplanned analysis was part of the international prospective multicenter PREDICTORS study that included outpatients suspected of recurrent VTE managed with validated diagnostic algorithms and D-dimer assays per local practices. Clinical variables were collected to compute the original Wells, modified Wells, and revised Geneva scores. Patients without VTE were followed up for 3 months. Events were independently adjudicated. For this analysis, YEARS and PEGeD scores were computed post hoc for patients with suspected pulmonary embolism (PE). Results: In total, 188 patients were included. Overall, PE incidence at baseline was 31% (58/188), and 47% (47/99) in those without anticoagulation. Applying YEARS, 87 patients (46%) would have had PE ruled out without imaging. The overall diagnostic miss rate would have been 2 of 83 (2.4%; 95% CI, 0.7%-8.4%) and 2 of 29 (6.9%; 95% CI, 1.9%-22%) in those without anticoagulation: both patients had no YEARS items and D-dimer of <1000 ng/mL but had imaging at baseline showing PE. For PEGeD, these numbers were 80 (43%) for efficiency, with an associated diagnostic miss rate of 2 of 77 (2.6%; 95% CI, 0.7%-9.0%) and 2 of 26 (7.7%; 95% CI, 2.1%-24%) in those without anticoagulation: both patients had a Wells score of <4.5 and D-dimer of <1000 ng/mL but had imaging at baseline showing PE. Conclusion: Both diagnostic algorithms showed comparable safety and efficiency in patients with prior VTE. Notably, the diagnostic miss rate was higher than expected and exceeded the safety margin.