An original risk scheme to predict in-hospital major bleeding in acute pulmonary embolism: The Syncope, Troponin, Anemia, Anti-platelet therapy, Renal dysfunction (STAR) bleeding score

Abstract

Background and objectives Bleeding-prediction score may help guiding management of patients with acute pulmonary embolism (PE). We aimed to develop the first risk stratification scheme to predict in-hospital major bleeding in PE. Method We recorded in-hospital characteristics of 2754 PE patients included in a prospective observational multicenter cohort study contributing 18,028 person-days of follow-up. Bleeding events were independently adjudicated using medical record review. We used the best subsets logistic regression to select candidates included in the in-hospital major bleeding score. The final model was internally validated by bootstrapping (500 replicates) and compared to the PE-derived long-term bleeding RIETE and VTE-BLEED scores. Results We observed 82 first in-hospital major bleeding [3.0% (95% CI, 2.4–3.7]. Five independent variables were included in the final model and weighted by regression coefficients: anemia (hemoglobin\xa0 \xa03 points). The C-index for the continuous risk score was 0.78 and remained good (i.e., 0.75) in the bootstrapped internal validation. The global fit of the STAR bleeding score was significantly higher than those of the RIETE and VTE-BLED scores allowing net reclassification improvement of major bleeding predicted probabilities (37% and 41%, respectively). Conclusion A simple 5-variable risk score was effective in quantifying the risk of PE-related in-hospital major bleeding in a large community-based cohort of patients. External validations are warranted.