Abstract TP054: Clinical Utility of Prognostic Scores in Acute Ischemic Stroke: A Comparative Analysis

Background: Prognostic scores are widely used to estimate functional outcomes after acute ischemic stroke, yet comparative performance of these scales remains limited. We assessed the predictive accuracy of ASTRAL, DRAGON, and THRIVE scores for 90-day outcomes. Methods: We retrospectively reviewed 233 ischemic stroke cases (2014-2023) at a Level 1 stroke center. ASTRAL, DRAGON, and THRIVE scores were calculated for each case, and 90-day modified Rankin Scale (mRS) outcomes were documented or derived. Receiver operating characteristic (ROC) curves were generated to assess discriminative ability, with accuracy, sensitivity, specificity, and predictive values calculated. Analyses were performed with GraphPad Prism 10. Results: ASTRAL demonstrated the highest AUROC (0.80), followed by DRAGON (0.79) and THRIVE (0.76); confidence intervals overlapped. At optimal cutoffs (ASTRAL 24, DRAGON 5, THRIVE 3), accuracy was 76.2%, 77.2%, and 64.4%, respectively. ASTRAL and DRAGON significantly outperformed THRIVE (p < 0.005). THRIVE showed high sensitivity (92%) but low specificity (55%). Across all scores, negative predictive values (90.7%) exceeded positive predictive values (48.4%). Conclusions: ASTRAL and DRAGON scores demonstrated superior overall accuracy compared to THRIVE in predicting 90-day outcomes after ischemic stroke. While THRIVE’s high sensitivity makes it useful for identifying patients likely to have favorable recovery, its poor specificity limits clinical reliability. These findings highlight the need for development of more balanced prognostic tools.