Early and Late Mortality Prognostication in Acute Ischemic Stroke

Accurate prognostication after acute ischemic stroke (AIS) aids in long-term management. We previously assessed ASTRAL, DRAGON, and THRIVE in predicting good vs. poor outcomes at 90-days and found they do not reliably predict poor function (mRS >3). In this study, we evaluate their ability to predict early and late mortality post-AIS, testing whether distinguishing survival vs. mortality improves accuracy. We also examined several clinical factors for an association with mortality post-AIS to enhance prognostication and identify strong predictors for developing a superior stroke scale in the future.

A retrospective analysis of 233 ischemic stroke cases from 2014-2023 at Sanford Medical Center was conducted. Prognostic scores were calculated at 7 and 90 days. ROC curve analysis was used to assess discriminative ability. We also compared late, 90-day mortality with pre-stroke mRS, Trial of ORG 10172 in Acute Stroke Treatment (TOAST) stroke etiologies, altered mental status on admission, and ipsilateral extracranial artery stenosis on neck computed tomography angiography (CTA neck). This analysis was generated using GraphPad Prism version 10.0 (GraphPad Software, Inc., San Diego, CA).

ASTRAL showed "excellent" prognostic ability for 7-day mortality (AUROC 0.92) and "considerable" accuracy at 90 days (AUROC 0.83). DRAGON and THRIVE performed "considerable" and "fair", respectively. Pre-stroke mRS, cardioembolic etiology, and altered mental status were strongly associated with 90-day mortality (p < 0.0001). Ipsilateral extracranial stenosis, acute glucose, and hemoglobin A1c were not predictive.

Among the three scoring tools, ASTRAL exhibited the strongest ability to predict early mortality, but its accuracy decreased for late mortality. The discrepancy in ASTRAL, THRIVE, and DRAGON’s performance may be attributed to the different variables each tool utilizes for scoring. Certain clinical factors may hold greater prognostic value than others. Our preliminary analysis suggests pre-stroke mRS, TOAST stroke etiology, and altered mental status at presentation may have good prognostic value.