P6459Comparison of the new BLLADS score with other scores for the prediction of major bleeding in patients with acute pulmonary embolism

Abstract

Abstract Background Hemorrhagic complications are a major obstacle for aggressive antithrombotic therapy in patients with acute pulmonary embolism (PE). Objectives We aimed to develop a simple risk score for predicting major bleeding (MB) in patients with acute PE using medical history and laboratory data at admission, including the potential influence of thrombolytic therapy, and to compare its predictive power to bleeding risk scores previously developed for patients with atrial fibrillation or venous thromboembolism. Methods A total of 630 consecutive patients treated for PE in six Serbian University hospitals were followed up for the occurrence of MB over a 90-day period after admission. A 6-component bleeding risk score was developed after Cox regression analysis of possible variables presented at admission. The use of thrombolytic therapy was also tested as a risk factor for bleeding and was integrated into the score. The ATRIA, HAS BLED, RIETE and VTE-BLEED scores were calculated for each patient at baseline and the predictive performances were compared with new score using c-statistics. Results MB occurred in 61 (9.7%) patients during the 90-day follow-up, with no increased risk of all-cause mortality (p=0.108). Six independent factors associated with MB were included in the final model (previous bleeding, leukocyte count ≥14x109/L, receipt of thrombolytic therapy, anemia, drugs associated with bleeding, and recent surgery; BLLADS). For the six- and five-variable models (without points for thrombolysis), C-indices were 0.774 (95% confidence interval [CI], 0.713–0.835, p<0.001) and 0.713 (95% CI, 0.639–0.788, p<0.001), respectively. The predictive power of the BLLADS score was found to be superior in comparison with other four scores: c-index 0.779 (95% CI 0.716–0.841, p<0.001), 0,614 (95% CI 0.535–0.692, p=0.005), 0.591 (95% CI 0.518–0.664, p=0.025), 0.589 (95% CI 0.518–0.659, p=0.029), 0.586 (95% CI 0.508–0.664, p=0.035), for continuous BLLADS, RIETE, VTE-BLEED, ATRIA and HAS BLED scores, respectively. Conclusion A simple six-variable score including the use of thrombolysis was developed with sufficient discriminative capacity comparing to current available scores for the prediction of 90-day MB for non-selected PE patients.