A new prognostic model and score to predict short term outcome after intracerebral hemorrhage

Spontaneous intracerebral hemorrhage (ICH) accounts for the highest mortality of all strokes. An early and reliable prognostic indication in ICH patients is potentially useful for initiating individual treatment and for informing patients and families. The published predictive models are largely based on neurological features; however, a part of them need to be judged with the subjective viewpoint of physician. Recently, we reviewed the data of 716 patients of ICH within 30 days. By univariate analysis and receiver operating characteristic (ROC) curves, four independent predictors for death outcome were identified from 11 laboratorial items and patient’s basic information, such as blood glucose (Glu), white blood cell count (WBC), lactate dehydrogenase (LDH) and age. With the AUC of 0.769, the prognostic model was developed by logistic regression. When the threshold of prediction index was set to 0.106, the sensitivity (88.9%) and specificity (55.6%) for death outcome were obtained by ROC curves. Furthermore, a simple predictive score was developed, whose AUC was 0.745. On a range from 0 - 4, the score of ≥3 could predict death outcome with good sensitivity (71.4%) and specificity (72.3%). The objective regression model and score had perfect discriminatory power for the prognosis of ICH within 30 days without the subjective judgment of physicians. It should be easily used by both neurological specialists and community doctors.   Key words: Intracerebral hemorrhage, prognosis, regression model, predictive score.