Abstract
BACKGROUND: The morbidity, mortality, and monetary cost associated with intracerebral hemorrhage (ICH) is devastatingly high. Several scoring systems have been proposed to prognosticate outcomes after ICH, although the original ICH Score is still the most widely used. However, recent research suggests that systemic physiologic factors, such as those included in the Acute Physiology and Chronic Health Evaluation II score, may also influence outcome. In addition, no scoring systems to date have included premorbid functional status. Therefore, we propose a scoring system that incorporates these factors to prognosticate 3-month and 12-month functional outcomes. METHODS: We used the Random Forest machine-learning technique to identify factors from a dataset of more than 200 data points per patient that were most strongly affiliated with functional outcome. We then used linear regression to create an initial model based on these factors and modified weightings to improve accuracy. Our scoring system was compared with the ICH Score for prognosticating functional outcomes. RESULTS: Two separate scoring systems (Intracerebral Hemorrhage Outcomes Project 3 [ICHOP(3)] and ICHOP(12)) were developed for 3-month and 12-month functional outcomes using Glasgow Coma Scale, National Institutes of Health Stroke Scale, Acute Physiology and Chronic Health Evaluation II, premorbid modified Rankin Scale (mRS), and hematoma volume (3-month only). Patient outcomes were dichotomized into good (mRS score, 0-3) and poor (mRS score, 4-6) categories based on functional status. Areas under the curve in the derivation cohort for predicting mRS score were 0.89 (3-month) and 0.87 (12-month); both were significantly more discriminatory than the original ICH Score. CONCLUSIONS: The ICHOP scores may provide more comprehensive evaluation of a patient's long-term functional prognosis by taking into account systemic physiologic factors as well as premorbid functional status.