Diagnostic accuracy of HAT, SEDAN, and HALP scores for predicting hemorrhagic transformation after intravenous thrombolysis in acute ischemic stroke: a retrospective single-center study

Objective: Hemorrhagic transformation (HT) is a clinically significant complication of intravenous thrombolysis (IVT) in acute ischemic stroke, carrying substantial morbidity in its parenchymal subtypes. We evaluated and compared the diagnostic accuracy of the HAT, SEDAN, and HALP scores for predicting post-IVT HT. Materials and Methods: This retrospective single-center diagnostic accuracy study (STARD 2015) enrolled 317 consecutive IVT-treated acute ischemic stroke patients (January 2023–January 2026). The reference standard was any HT on follow-up non-contrast CT. ROC-derived AUC with 95% confidence intervals (DeLong method) was the primary accuracy measure; optimal cut-offs were determined by the Youden index; pairwise comparisons used the DeLong method. Results: Fifty-seven patients (18.0%) developed HT. The SEDAN score achieved the highest AUC (0.745; 95% CI 0.675–0.810), followed by HAT (0.721; 95% CI 0.653–0.790); the HALP score showed no meaningful discriminatory value (AUC 0.527; 95% CI 0.440–0.617). At the Youden-optimal cut-off of ≥ 2, the SEDAN score attained a sensitivity of 80.4%, a specificity of 60.8%, and a negative predictive value (NPV) of 93.4%. No significant AUC difference was detected between HAT and SEDAN (ΔAUC = –0.022; P = .826). Conclusions: The SEDAN score demonstrated the highest diagnostic accuracy for post-IVT HT prediction, with statistically indistinguishable performance versus HAT (95% CI for ΔAUC: –0.222 to +0.177); formal equivalence was not established. The HALP score provided no clinically meaningful discrimination. A SEDAN score below 2 may identify low-risk patients at the bedside. The high mechanical thrombectomy rate in HT-positive patients (71.9%) warrants cautious interpretation. Prospective multicenter validation is needed.