Functional Outcome Prediction in Japanese Patients with Nonsurgical Intracerebral Hemorrhage: The FSR ICH Score.

Introduction: There has been limited research on predicting the functional prognosis of patients with nonsurgical intracerebral hemorrhage (ICH) from the acute stage. The aim of this study was to develop a risk prediction model for the natural course in patients with nonsurgical ICH and to evaluate its performance using a multicenter hospital-based prospective study of stroke patients in Japan.

Methods: We consecutively registered a total of 1,017 patients with acute ICH (mean age, 68 years) who underwent conservative treatment and followed them up for 3 months. The study outcome was a poor functional outcome (modified Rankin Scale score, 4-6) at 3 months after ICH onset. To develop the risk prediction model for natural course in patients with nonsurgical ICH, we included the following clinical common factors assessed on admission in daily clinical practice for ICH: age, sex, medical history (hypertension, diabetes mellitus, dyslipidemia, pre-stroke dementia, previous stroke, coronary artery disease, smoking status, alcohol drinking status, oral anticoagulation, and antiplatelet medication), admission status (time from onset to admission, systolic blood pressure, diastolic blood pressure, pulse pressure, plasma glucose levels, severity of the stroke), and neuroradiologic data (ICH location, intraventricular hemorrhage, and hematoma volume). The risk prediction model for poor functional outcome was developed using logistic regression analysis. In addition, the risk prediction model was translated into a point-based simple risk score (FSR ICH score) using the approach in the Framingham Heart Study.

Results: At 3 months after the ICH onset, 323 (31.8%) patients developed a poor functional outcome. Age, diabetes mellitus, pre-stroke dementia, NIHSS score on admission, intraventricular hemorrhage, and hematoma volume were included in the risk prediction model. This model demonstrated excellent discrimination (C statistic = 0.884 [95% confidence interval, 0.863-0.905]; optimism-corrected C statistic based on 200 bootstrap samples = 0.877) and calibration (Hosmer-Lemeshow goodness-of-fit test: p = 0.72). The FSR ICH score, a point-based simple risk score, also showed excellent discrimination, with a C statistic of 0.882 (95% CI: 0.861-0.903).

Conclusions: We developed a new risk prediction model for 3-month poor functional outcome in patients with nonsurgical ICH using a multicenter hospital-based prospective study in Japan. The current risk prediction model has the potential to be a useful tool for estimating the natural course in patients with nonsurgical ICH, aiding in making treatment decisions, including surgical options, early formulation of rehabilitation plans, and efficient utilization of medical resources.