Accuracy of HAS-BLED and BleedMAP Scores in Predicting Postoperative Bleeding in Anticoagulated Patients Undergoing non-Cardiac Surgery

Background There are limited data on the performance of bleeding risk scores in predicting postoperative bleeding in anticoagulated patients undergoing elective non-cardiac surgery. Objectives To evaluate the predictive accuracy of the HAS-BLED and BleedMAP scores for postoperative bleeding risk in this population. Methods This retrospective study enrolled anticoagulated patients undergoing elective surgery referred to the Preoperative Medical Consultation clinic. The C-statistic was used to evaluate the discriminatory performance of the HAS-BLED and BleedMAP scores in predicting overall bleeding following non-cardiac surgery. The predictive performance of the two scores was compared using DeLong's method. Results Among 591 patients who had previously received anticoagulation therapy, 84.9% received warfarin, 9.5% enoxaparin, and 5.6% direct oral anticoagulants. Indications included atrial fibrillation (67.9%), venous thromboembolism (18.1%), and mechanical valve replacement (13.5%). Procedures were classified as minor (16.6%), low to moderate risk (46.4%), or high bleeding risk (37.1%). The mean HAS-BLED and BleedMAP scores were 2.35 ± 1.36 and 0.94 ± 0.81, respectively. At 1-month postoperative follow-up, the overall bleeding rate was 40.4%, with 2.5% classified as major bleeding events. The HAS-BLED score had a C-statistic of 0.512 (95% confidence intervals (CI): 0.434–0.591) for predicting overall bleeding, while the BleedMAP score showed moderate predictive ability (C-statistic: 0.581; 95% CI: 0.531-0.631). The two scores had no statistically significant difference ( P = .13). Conclusions The HAS-BLED score showed limited discriminatory ability for postoperative bleeding in anticoagulated patients, though its negative predictive value was high. No significant difference in predictive performance was observed between the HAS-BLED and BleedMAP scores.