RESUMO
INTRODUCTION: Bleeding is a major complication of anticoagulation in acute pulmonary embolism (APE) while estimating individual bleeding risk remains challenging. Elevated D-dimer levels (DD) have been shown to predict bleeding events. OBJECTIVES: (1) direct comparison of the capacity of bleeding risk prediction scores (VTE-BLEED, RIETE, HAS-BLED, HEMORR2HAGES) to prognosticate in-hospital bleeding events in the acute phase of APE in a real-life population of APE patients;(2) augmentation of the discriminative capacity of fore mentioned scores with DD. MATERIALS: Post-hoc analysis of a prospective observational study. DD levels were measured using the VIDAS D-dimer Exclusion test. Receiver operating characteristic curves, areas under the curve (AUC) for bleeding prediction were calculated for scores and DD. Bleeding scores+DD were compared using an established index quantifying the reclassification of patients (net reclassification index, NRI). RESULTS: 310 APE patients were included. 35(11.3%) bleeding events occurred (hematomas, GI, urinary tract, retroperitoneal, uterine, CNS, respiratory tract): 17 major (MB) and 18 clinically-relevant non-major bleedings (CRNMB), none were fatal. All scores had satisfactory AUCs (0.754-0.767), except HAS-BLED (AUCâ¯=â¯0.512; 0.455-0.569). DD were higher in patients with bleeding events (29,911â¯ng/ml vs. 4805â¯ng/ml, pâ¯=â¯.031), AUC 0.621(0.520-0.721), pâ¯=â¯.02. DDâ¯=â¯5750â¯ng/ml was characterized by ORâ¯=â¯2.3(95%CI 1.05-5.0) for all bleeding events. Adding DD improved the discriminatory capacity of tested scores in the non-high risk of bleeding category, NRI 0.07-03. CONCLUSIONS: Of the tested scores RIETE, HEMORR2HAGES, VTE-BLEED performed best at identifying APE patients at risk of in-hospital bleeding complications. DD levels may predict in-hospital bleeding events and may improve identifying patients classified as non-high risk who experience bleeding complications.