Background/Purpose: The ICH score (ICHS) is commonly used to predict 30-day mortality in spontaneous ICH. Several investigators have evaluated additional factors in an attempt to refine this score, though none of these modifications have been incorporated into routine practice. We sought to determine whether incorporating additional radiographic features, specifically herniation and/or the degree of midline shift (MLS), improved the performance of the ICH score.
Methods: We retrospectively reviewed 180 consecutive ICH patients admitted to a single comprehensive stroke center between September 2014 and November 2015. The outcome of interest was poor prognosis, defined as mRS of 4-6, at time of discharge. Admission head CTs were evaluated by a vascular neurologist to determine whether uncal, tonsillar or subfalcine herniation were present, and to quantify the degree of midline shift. Midline shift was treated as a categorical variable determined by measuring the shift of the septum pellucidum from midline at the level of the basal ganglia.
Results: Patients with radiographic herniation present on admission had a much higher odds of poor outcome; however, this measure was specific (98%) but not sensitive (23%). Degree of midline shift performed similarly, with a specificity of 86% and a sensitivity of 36%. In our population, the ICH score was a better predictor of poor outcome than any of the combinations shown in the Table. For each unit increase in ICH score, the odds of a poor outcome increased by a factor of 2.6.
Conclusion: Although clinicians often view radiographic evidence of herniation and midline shift as indicators of dire prognosis, incorporation of these features into a modification of the ICH score failed to improve upon the original scale.