Short-term performance of bleeding risk scores in anticoagulated older patients with acute pulmonary embolism

Background

Older patients with pulmonary embolism (PE) are at increased risk of bleeding, which is associated with adverse outcomes, making early risk assessment particularly important. This study evaluated the short-term performance of commonly used bleeding risk scores in patients aged ≥65 years.

Methods

An analysis was conducted in a multicenter, prospective study of patients aged ≥65 years with PE receiving anticoagulation (NCT02943343, 2016). The performance of the PE-SARD, Kuijer, RIETE, BACS, and ATRIA scores was evaluated using the time-dependent area under the curve (AUC) analysis. Calibration was assessed with calibration plots. Sensitivity, specificity, and positive predictive values were calculated. The primary outcomes were major bleeding (MB) at 7 and 14 days.

Results

Among 3661 anticoagulated patients aged ≥65 years with acute PE, MB occurred in 18 (0.5%) and 36 (1.0%) patients at 7 and 14 days, respectively, and clinically relevant non-major bleeding (CRNMB) in 69 (1.9%) and 105 (2.9%) patients. Patients with bleeding events had higher mortality than those without bleeding. Discrimination for short-term MB was low to moderate across scores, with RIETE (AUC 0.76, 95% CI 0.65–0.87) and PE-SARD (AUC 0.74, 95% CI 0.63–0.85) showing the most consistent performance at 7 days. Overall discrimination declined at 14 days. All scores demonstrated limited ability to predict CRNMB, and no model showed a consistent net clinical benefit.

Conclusions

In older patients with PE, bleeding risk scores showed low-to-moderate discrimination for short-term bleeding, with moderate early performance observed only in selected scores. None of the scores reliably identified CRNMB, highlighting the limitations of baseline risk assessment in acute care settings.