A comparative evaluation of existing grading scales in intracerebral hemorrhage

Abstract

Background: In recent years, a multitude of clinical grading scales have been created to help identify patients at greater risk of poor outcome following ICH. We sought to validate and compare eight of the most frequently used ICH grading scales in a prospective cohort.

Methods: Eight grading scales were calculated for 67 patients with non-traumatic ICH enrolled in the prospective intracerebral hemorrhage outcomes project (ICHOP) database. Receiver operating characteristic (ROC) analysis, including area under the curve (AUC) and maximum Youden Index were used to assess the ability of each score to predict in-hospital mortality, long-term (3 months) mortality, and functional outcome at 3 months (mRS ≥ 3).

Results: All scales demonstrated excellent to outstanding discrimination for in-hospital and long-term mortality, with no significant differences between them after controlling for the false discovery rate. All scales demonstrated acceptable to outstanding discrimination for functional outcome at 3 months, with the new ICH score demonstrating significantly lower AUC than 6 of the 8 scores. Essen ICH score was the only score to demonstrate outstanding discrimination for each outcome measure.

Conclusion: Though significant differences were minimal in our cohort, we showed the existing selection of ICH grading scales to be useful in stratifying patients according to risk of mortality and poor functional outcome. Continued validation and comparison in large prospective cohorts will bring the goal of a singular prognostic model for ICH closer to fruition.