External validation of the ICH score

Neurocrit Care. 2004;1(1):53-60. doi: 10.1385/NCC:1:1:53.


Introduction: The ICH score is a clinical grading scale that is composed of five components related to outcome after nontraumatic intracerebral hemorrhage (ICH): Glasgow Coma Scale score, ICH volume, presence of intraventricular hemorrhage, infratentorial origin, and age. The ICH score accurately risk-stratifies patients in the cohort from which it was developed, but it has not yet been fully externally validated. The purpose of this study was to determine whether the ICH score accurately risk-stratifies patients in an independent cohort.

Methods: Records of all patients with acute ICH presenting to Stanford Medical Center and Santa Clara Valley Medical Center from 1998 through 2000 were reviewed. Outcome was assessed as mortality at 30 days.

Results: A total of 175 patients formed the cohort for analysis. Thirty-day mortality for the entire cohort was 40%. ICH scores ranged from 0 to 5, and each increase in the ICH score was associated with an increase in 30-day mortality (p<0.01 for trend); 4 of 41 patients with an ICH score of 0 died, whereas all 5 patients with a score of 5 died. Receiver operating characteristic curves for the ICH score were similar for this cohort and for the initial ICH score cohort.

Conclusion: The ICH score accurately stratifies outcome in an external patient cohort. Thus, the ICH score is a validated clinical grading scale that can be easily and rapidly applied at ICH presentation. Ascale such as the ICH score could be used to standardize clinical treatment protocols or clinical studies.

Publication types

  • Research Support, N.I.H., Extramural
  • Validation Study

MeSH terms

  • Adult
  • Age Factors
  • Aged
  • Aged, 80 and over
  • Cerebellum / pathology
  • Cerebral Hemorrhage / mortality*
  • Cerebral Hemorrhage / pathology*
  • Cerebral Ventricles / pathology
  • Cohort Studies
  • Female
  • Glasgow Coma Scale*
  • Humans
  • Male
  • Middle Aged
  • Prognosis
  • Reproducibility of Results
  • Risk Adjustment*
  • Sensitivity and Specificity