Abstract
BACKGROUND: Prognostic assessments, which are crucial for decision-making in critical illnesses, have shown unsatisfactory reliability. We compared the accuracy of a widely used prognostic score against a model derived from clinical data obtained 5 days after admission for patients with intracerebral hemorrhage (ICH), a condition for which prognostication has proven notoriously challenging and prone to bias. METHODS: Patients enrolled in a prospective observational cohort study of spontaneous ICH underwent hourly Glasgow Coma Scale (GCS) assessment. Outcome was measured at 3 months using the modified Rankin Scale (mRS). We analyzed the change in correlation between GCS and 3-month mRS scores from admission through day 5, and compared the performance of a parsimonious set of day 5 clinical variables against the ICH score. RESULTS: Data was collected on 254 subjects. The ICH score and day 5 GCS score were both correlated with 3-month mRS score (p < 0.001), but the correlation was stronger with day 5 GCS score (p < 0.05 by Fisher z-transformation). Premorbid mRS score, intraventricular hemorrhage and day 5 GCS score were independent predictors of outcome (all p < 0.05 in ordinal regression model). While ICH score correctly classified good (mRS 0-3) vs. poor (mRS 4-6) outcome in 73% of cases, the day 5 model correctly classified 83% of cases. CONCLUSIONS: A simple reassessment after 5 days of care significantly improves the accuracy of prognosticating outcome in patients with ICH. These data confirm the feasibility and potential utility of early reassessments in refining prognosis for patients who survive early stabilization of a severe neurologic injury.