Abstract WP374: Validation of a Novel Claims-Based Stroke Severity Index in Patients Hospitalized with Intracerebral Hemorrhage
Introduction: Stroke severity, an important outcome predictor after intracerebral hemorrhage (ICH), is usually unavailable in administrative databases.
Hypothesis: A claims-based stroke severity index (SSI) with seven predictors is valid for patients with ICH.
Methods: Consecutive 1,360 ICH patients of three stroke centers with stroke registry data linked to the National Health Insurance Research Database constituted the study cohort. The admission National Institutes of Health Stroke Scale (NIHSS) and the 3-, 6-, and 12-month modified Rankin Scale (mRS) were the reference standards of concurrent and predictive validity, respectively, and were compared with SSI using Pearson’s correlation coefficient. We fitted base logistic regression models (age, sex, and Charlson’s Comorbidity index) to predict mortality at 3, 6, 12 months, added the NIHSS or the SSI separately to the base models, and compared the model discrimination with the base model.
Results: The SSI correlated with the admission NIHSS (r = 0.731, 95% CI: 0.705-0.755) and mRS at 3, 6, and 12 months (r = 0.696, 95% CI: 0.665-0.724; r = 0.685, 95% CI: 0.653-0.715; and r = 0.664, 95% CI: 0.622-0.702, respectively). Adding the SSI or the NIHSS to the base models increased model discrimination. The areas under the curve (AUCs) of the base models for 3-, 6-, and 12-month mortality were 0.694, 0.689, 0.706; they improved after adding either NIHSS (AUCs: 0.905, 0.893, and 0.884) or SSI (AUCs: 0.843, 0.831, and 0.840) to the base model.
Conclusions: The SSI might be a valid NIHSS proxy and enable adjustment for severity in ICH outcome studies using an administrative claims database.
Author Disclosures: C. Hsieh: None. S. Sung: None. L. Hung: None.
- © 2016 by American Heart Association, Inc.