Testing the ICH Score
To the Editor:
Hemphill et al1 present an analysis of 161 patients carried out to determine a reliable grading score for the prediction of 30-day mortality in patients following a spontaneous intracerebral hemorrhage (ICH). Factors independently associated with 30-day mortality were Glasgow Coma Score, age >80 years, ICH volume, ICH of infratentorial position, and presence of intraventricular hemorrhage. A score based on these variables was assigned to each patient. All patients within their dataset with an ICH score of 0 survived, and all patients with a score of 5 (highest score assigned) died.
Hemphill et al restricted the testing of the scoring system to the data that produced it. We were interested in whether this scoring system could be of similar predictive value in patients treated in our unit. From 1994 to date, all patients admitted following a spontaneous supratentorial ICH have been recorded on a prospective database and followed up to 6 months after ictus. Although we do not have specific mortality at 30 days, we have recorded outcome at neurosurgical discharge, which was on average 2 to 4 weeks after ictus. Up to August 1999, 440 patients had been entered. Data were missing on 47 patients, not allowing a score to be calculated, but the remaining 393 patients were scored as shown in Figure 1.
We had only 1 patient scoring either of the highest 2 scores, as only 5% of patients in our database had an infratentorial ICH. A single patient scored 6 (<1%), 17 scored 4 (4%), 74 scored 3 (19%), 116 scored 2 (30%), 97 scored 1 (25%), and 88 patients scored 0 (22%). We also found 100% mortality in patients scoring 5 or 6 (although this is only 1 patient) but a 5% mortality with a score of 0 or 1, as shown in Figure 2.
We were also interested to see whether this scoring system could predict unfavorable outcome (severe disability, death, or vegetative state) at neurosurgical discharge. From Figure 3 it is apparent that for all patients scoring above 2, the rate of unfavorable outcome approaches 100%. In fact, 38% of patients with a score of 0 and 70% with a score of 1 are not independent at neurosurgical discharge.
It therefore appears that this ICH scoring system is generally applicable inasmuch as the mortality is low in patients with an ICH score of 0 or 1. Thirty-day mortality rises steeply with a score of 2 or above. Unfavorable outcomes, however, are common in patients with a low ICH score, and this rises to almost 100% with scores of 2 and above. We feel that some sort of system to predict those capable of making an independent recovery from their ICH would be more helpful and is not provided by this simple ICH grading score.
- ↵Hemphill JC, Bonovich DC, Besmertis L, Manley GT, Johnston SC. The ICH Score: a simple, reliable grading scale for intracerebral hemorrhage. Stroke. 2001; 32: 891–897.
The ICH Score is a clinical grading scale designed for risk stratification of patients after acute nontraumatic intracerebral hemorrhage (ICH).1 The outcome measure chosen for scale development was mortality at 30 days, which is a commonly used measure in other ICH prediction models. The ICH Score was demonstrated accurate in the population from which it was developed. However, in order for a scale such as the ICH Score to be useful, it must demonstrate validity in other patient populations. Additionally, in studies of acute neurologic catastrophes such as stroke or traumatic brain injury, functional outcome, and not just mortality, may be an important outcome measure as well.
Fernandes et al bring out these important points in their retrospective testing of the ICH Score on their patient population. As demonstrated by their Figure 2, the ICH Score accurately risk stratifies patients with regard to mortality across the range of scores, even though the distribution of patients with various ICH Scores differs somewhat from the UCSF ICH Cohort from which the ICH Score was developed, and “neurosurgical discharge,” not the uniform time point of 30 days, is used for outcome assessment. However, they also attempt to use the ICH Score to stratify functional outcome at this same early time point of “neurosurgical discharge” and found that the ICH Score was weighted heavily toward poor outcome across the entire scale.
In both cohorts of ICH patients, early death and disability were too frequent to distinguish nondisability from other outcomes by hospital discharge or 30 days after ICH. Perhaps a longer duration of follow-up would provide more useful information about recovery and long-term functional outcome. Most acute stroke and traumatic brain injury studies, including the Surgical Trial in Intracerebral Hemorrhage (STICH), use functional assessment at more clinically meaningful time points such as 3 to 6 months after the event.2–4⇓⇓ Furthermore, at least in the United States, functional status may not be the only determinant in the timing of hospital discharge. Early transfer to rehabilitation may occur in centers without comprehensive facilities, health insurers may demand transfer of patients from an initial acute care hospital to a covered facility regardless of the functional status of the patient at the time, and patients likely to die may be transferred to hospice. We believe, therefore, that an assessment of the utility of a scale such as the ICH Score in stratifying surviving patients with regard to long-term functional outcome should use a standardized outcome measure at a uniform time point that is clinically meaningful regarding ICH recovery and not prone to local variation in hospital discharge practices. Such a scale would likely have utility in stratifying patients for clinical research and clinical care in ICH.