Comparative analysis of prognostic scores for functional outcome after ischemic stroke

Abstract

Background: Comparative data on the predictive performance of stroke prognostic scores in a real-world setting are sparse.

Objective: We aimed to compare the performance of existing scores for acute stroke outcomes in an observational cohort.

Methods: Using data from 12,486 patients with acute ischemic stroke (mean [SD] age, 72.5 [12.6] years; male, 59.4 %) prospectively registered in Fukuoka, Japan, between 2007 and 2017, we evaluated the predictive performance of six stroke prognostic scores, namely ASTRAL, iScore, PLAN, HIAT, SPAN-100, and THRIVE. The discriminative power of the scores was evaluated by the area under the receiver operating characteristic curve (AUROC). Calibration was evaluated using calibration plots. Overall performance, incorporating both discrimination and calibration, was assessed using Brier score.

Results: In comparative analyses using un identical study population, AUROCs for predicting 3-month poor functional outcome were 0.87 for ASTRAL, 0.88 for iScore, and 0.89 for PLAN among the scores for all patients, and 0.74 for HIAT, 0.81 for SPAN-100, and 0.78 for THRIVE among the scores for patients receiving reperfusion therapy. The calibration plots showed fair agreement between the outcome predictions and the observed outcomes in all scores, and no substantial difference was found among the scores. The analysis of overall performance indicated that PLAN was better than ASTRAL, whereas no significant difference was found among HIAT, SPAN-100, and THRIVE.

Conclusions: The predictive performance of all six scores was good, even in our observational cohort, reflecting the real-world setting. The prognostic scores could provide useful information for the management of acute stroke patients.

Keywords: Comparison; Outcome; Prediction; Prognostic score; Stroke.