Background: There is no agreement for the etiologic classification of patients with intracerebral hemorrhage (ICH). In a series of patients with ICH, we performed a randomized head-to-head comparison between the two recently proposed etiologic classification systems.
Methods: We evaluated patients registered in a prospective database of consecutive patients. A simplified H-ATOMIC classification defines 8 categories: hypertension, amyloid, tumor, oral anticoagulants, malformation, infrequent, cryptogenic, and combination. SMASH-U also defines 8 categories: structural, medication, amyloid, systemic, hypertension, and undetermined, and nonstroke and stroke-non-ICH. Experienced stroke neurologists applied both classification systems to a randomly assigned list of patients. The concordances between the 2 systems were analyzed. In a subset of patients, the percent of agreement and the inter-rater reliability (kappa coefficient) were calculated.
Results: A total of 156 patients (age 72.3 ± 13.5 years) were evaluated, and 54 of these patients were evaluated by 2 neurologists. Concordance (a patient classified in equivalent categories for both systems) was 63%. The percentage of interobserver agreement was 85.5% for SMASH-U and 87.6% for H-ATOMIC. Inter-rater reliability was similar for SMASH-U (kappa .82) and H-ATOMIC (kappa .76). The range of reliability among neurologists was .66-.93 for SMASH-U and .66-.94 for H-ATOMIC.
Conclusions: The percentage agreement among investigators is remarkably high for both classification systems, and the inter-rater reliability is substantial to almost perfect for both systems. However, discrepancies between the 2 systems are frequent (in about one third of the patients) due to different categories and definitions.
Keywords: Intracerebral hemorrhage; classification; diagnosis; etiology.
Copyright © 2018 National Stroke Association. Published by Elsevier Inc. All rights reserved.