Accuracy of available scoring systems for prediction of in-hospital major bleeding in acute pulmonary embolism patients

Background Bleeding prediction scores may help guide acute management of patients with pulmonary embolism (PE). However existing scoring systems have not been validated for in-hospital assessment. We compared 6 available bleeding scores for the prediction of major bleeding in the in-hospital phase in a real-life cohort. Methods 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. The VTE-BLEED, RIETE, ORBIT, HEMORR2HAGES, ATRIA, and HAS-BLED scores were assessed at baseline. ISTH-defined bleeding events were independently adjudicated. The accuracy of the overall, original 3-level and newly defined optimal 2-level outcome of the scores were evaluated and compared. Results We observed 82 first in-hospital major bleeding events [3.0% (95% CI, 2.4–3.7)]. Overall, the predictive power of bleeding scores was poor, with a Harrel's C index ranging from 0.57 to 0.69. The Riete score had the numerically highest model fit and discrimination capacity but without reaching statistical significance versus the ORBIT, HEMORR2HAGES, and ATRIA scores. The VTE-BLEED and HAS-BLED scores had significantly lower C indices, integrated discrimination improvement, and net reclassification improvement compared to the others. The rate of observed in-hospital major bleeding in score-defined low-risk patients was high, at between 15% to 34%. Conclusion Current available scoring systems are not sufficiently accurate for the prediction of in-hospital major bleeding in patients with acute PE. There is a need to develop risk scores specific to acute PE to optimally target bleeding-prevention strategies.