Derivation and validation of the BLEEPE score for risk stratification of major bleeding in patients with acute pulmonary embolism
M C Vedovati, L A Cimini, V Valeriano, I Montenovo, A P Murgia, L Pierpaoli, G Tratar, S Zaccaroni, A Bongarzoni, A Vinci, A Mavri, L Martinelli, G Pepe, S Cozzio, C BecattiniAbstract
Background
Major bleeding (MB) is the most feared complication of anticoagulant treatment in patients with acute pulmonary embolism (PE). Existing bleeding risk scores showed a modest performance, especially for predicting short-term MB.
Purpose
The aim of this study in patients with acute PE was to develop and validate a risk score for MB at 30-day during anticoagulation: the BLEEPE score.
Methods
Data from prospective cohorts of patients with acute PE were merged in a collaborative database. Potential predictors (univariate p-value ≤ 0.1) for MB within 30 days were evaluated in a multivariable Cox regression model. Candidate predictors were identified and scored based on clinical relevance and β-coefficient and used to derive the BLEEPE score. The score was internally validated by bootstrap and externally validated in the COPE study population (NCT03631810). The performance of the BLEEPE score was assessed by discrimination (c-statistics), sensitivity, specificity, positive and negative predictive value (NPV).
Results
Age>65 years, prior MB, syncope, and hemoglobin<11 g/dl were identified as independent predictors of MB in the derivation cohort: incidences of MB in the high and low-risk groups were 7.9% and 1.1%, respectively (HR 7.59, 3.29-17.51) (Table). Incidences in the validation cohort were 6.9% and 1.6%, respectively (HR 4.48, 95% CI 3.39-6.46). C-statistics in derivation and validation cohorts were 0.735 and 0.680, respectively. The NPV was 99% in the derivation and 98% in the validation cohort.
Conclusions
The BLEEPE score showed a good accuracy and discrimination for 30-day MB in patients with acute PE. The simple parameters included in the BLEEPE score could help to implement strategies for personalized management in this population.For image description, please refer to the figure legend and surrounding text.