The ICH score is a simple clinical grading scale that estimates 30-day mortality after spontaneous intracerebral hemorrhage, the deadliest form of stroke. Developed by J. Claude Hemphill III and colleagues in 2001, it assigns points based on five factors found on a patient’s initial assessment and brain scan, producing a total between 0 and 6. Higher scores correspond to higher risk of death, with the relationship between score and mortality following a steep and consistent gradient across multiple validation studies. The score’s simplicity has made it the most widely used bedside tool for communicating the severity of a brain bleed, but that very influence has triggered a serious ethical debate about whether the score inadvertently hastens the deaths it claims to predict.
The Five Components
Each of the ICH score’s five components was independently associated with 30-day mortality in the original study of 152 patients. Points are assigned as follows:
- Glasgow Coma Scale (GCS): A GCS of 3 to 4 earns 2 points, 5 to 12 earns 1 point, and 13 to 15 earns 0 points. This is the most heavily weighted factor because the patient’s level of consciousness at arrival is the strongest single predictor of outcome.
- Hematoma volume: A bleed measuring 30 cubic centimeters or larger earns 1 point; smaller bleeds earn 0.
- Intraventricular hemorrhage: If blood has extended into the brain’s ventricles (the fluid-filled chambers at its center), 1 point is added.
- Infratentorial origin: If the bleeding started below the tentorium (in the brainstem or cerebellum rather than the larger cerebral hemispheres above), 1 point is added.
- Age: Patients aged 80 or older receive 1 point; younger patients receive 0.
The maximum possible score is 6, though this combination is rare because it requires a very large bleed in the brainstem of an elderly patient who is deeply comatose with blood also in the ventricles. In practice, most patients cluster between 0 and 4.1PubMed. The ICH score: a simple, reliable grading scale for intracerebral hemorrhage
How Mortality Tracks with Score
The score’s value lies in how cleanly each additional point corresponds to a jump in 30-day death rates. In an external validation study, every one-point increase was associated with a statistically significant rise in mortality, and all five patients who scored 5 died within 30 days.2PubMed. External validation of the ICH score A larger analysis across Florida hospitals found that mortality was about 7% for scores of 0 to 2, roughly 42% for scores of 3 to 4, and 66% for scores of 5 to 6.3PubMed Central. Association of the ICH Score With Withdrawal of Life-Sustaining Treatment Over a 10-Year Period
That gradient makes the score useful for giving families a realistic picture of severity within hours of a brain bleed. But it is crucial to understand what the score does not do: it does not tell you that any individual patient will or will not survive. A score of 3 means roughly four in ten patients with similar features died in the studies, which also means roughly six in ten survived. The score describes populations, not people.
Why Consciousness Matters So Much
The GCS component is the only factor worth up to 2 points, and that extra weight is earned. In one study comparing predictors head to head, GCS alone had an area under the curve of 0.91 for predicting 30-day death, slightly outperforming the full ICH score. Patients with a GCS of 3 to 4 had a 30-day mortality of about 68%, while those with a GCS of 13 to 15 had zero deaths in the same cohort.4PLoS ONE. Hematoma Shape, Hematoma Size, Glasgow Coma Scale Score and ICH Score: Which Predicts the 30-Day Mortality Better for Intracerebral Hematoma? Other single-center analyses have similarly confirmed that a GCS below 13 is one of the strongest independent predictors of both death and poor functional recovery.5PubMed Central. Mortality and morbidity in patients with spontaneous intracerebral hemorrhage: A single-center experience
This makes intuitive sense: if a brain bleed has already damaged enough tissue to render someone deeply unconscious, the injury is extensive. But it also introduces a timing problem. GCS can fluctuate in the first hours after a hemorrhage, especially if the bleed is still expanding or if the patient has been sedated for intubation. A score calculated on a sedated patient may overstate how bad things are, which is one reason clinicians sometimes re-assess after sedation has been minimized.
Measuring the Bleed
The ICH score draws a binary line at 30 cubic centimeters of blood. That threshold needs a reliable measurement method, and the standard one at the bedside is the ABC/2 formula. On a CT scan, you identify the slice with the largest area of hemorrhage, measure the longest diameter (A), the widest perpendicular diameter (B), and the number of slices with blood multiplied by slice thickness (C), then multiply and divide by two. The method was originally validated as a fast approximation that correlates well with more precise computer-traced measurements and takes under a minute.6PubMed. The ABCs of measuring intracerebral hemorrhage volumes
The formula works best for bleeds that are roughly oval-shaped. Irregularly shaped hematomas, which are common, tend to get overestimated by ABC/2. A systematic analysis of thousands of scans found that the method overestimated volume in six out of eight prior studies, though the overestimation was modest enough that categorical agreement (small, medium, or large bleed) was still reasonable.7PubMed Central. Accuracy of the ABC/2 Score for Intracerebral Hemorrhage: Systematic Review and Analysis of MISTIE, CLEAR-IVH, and CLEAR III For most clinical decisions that hinge on whether a bleed crosses the 30-cubic-centimeter line, ABC/2 is good enough. Where precision matters more, such as in research trials tracking whether a hemorrhage is growing, computer-based volumetric methods are increasingly used instead.
Where the Bleeding Starts
The infratentorial component of the ICH score treats all bleeds below the tentorium as a single category, but the reality is more nuanced. Cerebellar hemorrhages and brainstem hemorrhages behave quite differently. Brainstem bleeds carry dramatically higher mortality; one study found an odds ratio for death of about 37 compared with other infratentorial locations.8PubMed. Infratentorial Intracerebral Hemorrhage Patients with cerebellar hemorrhage, by contrast, were several times more likely to regain functional independence at three months than those with brainstem bleeds.9PubMed Central. Natural History of Infratentorial Intracerebral Hemorrhages: Two Subgroups with Distinct Presentations and Outcomes
Research comparing supratentorial and infratentorial hemorrhages has emphasized that the clinical and imaging features predicting death differ between the two groups, arguing they should not be treated as one entity.10PubMed. Spontaneous Intracerebral Hemorrhage The ICH score’s single binary point for infratentorial origin does not capture this distinction. A cerebellar bleed and a brainstem bleed receive the same additional point, even though their prognoses can diverge widely. This is one of the recognized simplifications baked into a tool designed for fast bedside use.
Blood in the Ventricles
When a brain bleed ruptures into the ventricles, the prognosis worsens. The ICH score assigns one point for any intraventricular hemorrhage, regardless of how much blood is present. But the actual relationship is dose-dependent: larger volumes of ventricular blood cause more injury, both immediately and over time, by obstructing the flow of cerebrospinal fluid and increasing pressure inside the skull.11PubMed Central. Intraventricular hemorrhage: severity factor and treatment target in spontaneous intracerebral hemorrhage The ICH score’s binary yes-or-no approach to ventricular blood is another area where simplicity trades away precision. A patient with a thin layer of blood settling in one ventricle gets the same point as one whose ventricles are completely filled.
The Self-Fulfilling Prophecy Problem
This is arguably the most important issue surrounding the ICH score, and it is not about the math. Because the score is so widely used to communicate prognosis, it can influence whether families and clinicians decide to continue aggressive treatment. If a high score leads to early withdrawal of life-sustaining treatment, the patient dies, and that death then reinforces the score’s predictive accuracy in future studies. The concern is circular: the score predicts death partly because it causes death.
The data supporting this worry are substantial. In one analysis, withdrawal of care was the proximate cause of death in about 57% of all patients who died after intracerebral hemorrhage, and the rate of withdrawal increased as the ICH score rose.12PubMed. The Intracerebral Hemorrhage Score: A Self-Fulfilling Prophecy? A decade-long study of the Florida Stroke Registry found that patients with scores of 5 to 6 were about 19 times more likely to have life-sustaining treatment withdrawn than those with scores of 0 to 2, and early withdrawal within 48 hours of admission was roughly 10 times more likely in the highest-score group.13PubMed Central. Association of the ICH Score With Withdrawal of Life-Sustaining Treatment Over a 10-Year Period Those associations held steady over the entire ten-year study period, suggesting that awareness of the self-fulfilling prophecy problem has not yet changed clinical behavior at scale.
This does not mean the score is useless or that withdrawal decisions are always wrong. Some patients with very high scores have injuries that are genuinely unsurvivable. But the evidence suggests that for patients in the middle range, especially scores of 3 or 4, the decision to withdraw treatment early may be premature. The American Heart Association’s guidelines now advise against issuing new “do not resuscitate” orders in the first 24 hours after an intracerebral hemorrhage in part because of these concerns.
Beyond 30-Day Mortality
The original ICH score was designed to predict death within 30 days. But families also want to know whether a survivor will be able to function. A prospective validation study of 243 patients found that the ICH score also stratified 12-month functional outcomes, with higher scores corresponding to greater disability across several cutpoints on the modified Rankin Scale.14PubMed Central. Prospective validation of the ICH Score for 12-month functional outcome Roughly 39% of patients in that cohort died during the initial hospitalization, but among survivors, meaningful recovery was possible. A separate study of patients with severe initial disability from intracerebral and intraventricular hemorrhage found that about 43% recovered to a good functional outcome at one year, and roughly a third achieved a two-point or greater improvement on the disability scale.15JAMA Neurology. One-Year Outcome Trajectories and Factors Associated with Functional Recovery Among Survivors of Intracerebral and Intraventricular Hemorrhage With Initial Severe Disability
These findings underscore a point that gets lost in the urgency of the acute setting: the ICH score captures a snapshot at admission, but recovery from brain hemorrhage unfolds over months. A score that predicts early death reasonably well does not necessarily predict what life looks like a year later for those who survive.
Modified Scores and Delayed Calculation
Researchers have tried to improve on the original ICH score by adding variables or refining how existing ones are weighted. The max-ICH score, for instance, incorporates additional clinical detail and has shown good ability to predict long-term unfavorable functional outcomes, with an area under the curve of about 0.83.16PubMed Central. The max-intracerebral hemorrhage score predicts long-term outcome of intracerebral hemorrhage However, when the original ICH score and the max-ICH score were compared head to head in an external validation cohort, the two performed similarly for three-month mortality and functional outcomes, with no significant difference in accuracy even in the subgroup that received maximal treatment without any care limitations.17PubMed Central. Assessment and comparison of the max-ICH score and ICH score by external validation
A different approach involves simply recalculating prognostic scores a few days after admission rather than relying solely on the initial assessment. The original ICH score, along with several modified versions, performed significantly better at predicting six-month mortality when calculated four days after the bleed rather than at baseline.18Stroke and Vascular Neurology. Predicting long-term outcomes in acute intracerebral haemorrhage using delayed prognostication scores This makes sense because by day four, the hematoma has usually reached its final size, early swelling has evolved, and the clinical picture has stabilized. The improvement in accuracy from delayed calculation is a practical reminder that the admission score is a preliminary estimate, not a final verdict.
Anticoagulant-Related Hemorrhages
Patients who develop brain bleeds while on blood thinners are a distinct clinical group because the anticoagulant can make the hemorrhage larger and harder to stop. A natural question is whether the ICH score works as well for these patients. A multicenter study found that the score performed well in anticoagulant-related bleeds, with a threshold of 4 points being the most useful cutoff for predicting 30-day death. The score was actually more accurate in hemorrhages related to newer anticoagulants (area under the curve of 0.92) than in those related to older agents like warfarin (area under the curve of 0.77).19PubMed. The prognostic utility of ICH-score in anticoagulant related intracerebral hemorrhage
Some researchers have wondered whether adding anticoagulant use as a sixth variable would sharpen predictions. While being on blood thinners does independently increase the odds of 30-day death, studies testing a modified score with anticoagulant use added have found that the extra variable does not measurably improve the score’s overall accuracy.20PubMed Central. Predicting Prognosis of Intracerebral Hemorrhage (ICH): Performance of ICH Score Is Not Improved by Adding Oral Anticoagulant Use The five original components already capture enough of the damage that adding one more variable does not move the needle.
How Well It Travels Across Populations
The ICH score was derived from a cohort at a single US medical center. Whether it performs equally well in different countries, ethnicities, and healthcare settings matters for a tool used globally. A Taiwanese study comparing locally derived scores with the original Hemphill score found no significant difference in predictive power, supporting the idea that the score generalizes across ethnic groups.21European Neurology. Risk Factors of In-Hospital Mortality of Intracerebral Hemorrhage and Comparison of ICH Scores in a Taiwanese Population A Peruvian validation study, however, found only moderate accuracy in a cohort of surgically treated patients, with areas under the curve in the 0.69 to 0.71 range for various time points.22PubMed. Validation of the ICH score and ICH-GS in a Peruvian surgical cohort: a retrospective study That lower accuracy is likely explained by the surgical cohort itself: the score was designed to predict outcomes in a general hemorrhage population, and patients selected for surgery are already a subset whose expected trajectory has been altered by the intervention.
Imaging Advances Beyond the Initial CT
The ICH score is calculated from a plain CT scan obtained at admission, but additional imaging can reveal something the score does not capture: whether the bleed is still actively growing. Hematoma expansion in the hours after a brain hemorrhage is one of the strongest drivers of deterioration and death. A marker called the “spot sign” on CT angiography, which looks like a small bright dot of contrast leaking within the hematoma, flags patients whose bleeds are likely to enlarge. The presence of a spot sign increased the odds of significant hematoma expansion roughly sixfold in one analysis, and remained a strong predictor even in patients who arrived at the hospital several hours after symptom onset.23PubMed Central. CTA Spot Sign Predicts Hematoma Expansion in Patients with Delayed Presentation After Intracerebral Hemorrhage
When characterized more systematically, the spot sign’s positive predictive value for expansion reached about 69% overall and climbed to 85% in patients scanned within three hours.24PubMed Central. Systematic Characterization of the Computed Tomography Angiography Spot Sign in Primary Intracerebral Hemorrhage Identifies Patients at Highest Risk for Hematoma Expansion The Spot Sign Score The spot sign is not part of the ICH score itself, but it provides a complementary piece of information: while the ICH score describes how bad things are right now, the spot sign estimates whether they are about to get worse. Some clinicians use both together when deciding how aggressively to pursue blood-pressure lowering or surgical intervention in the first hours.
Automated Hemorrhage Detection
AI tools are beginning to change how brain bleeds are identified and measured, which could eventually streamline ICH score calculation. An AI-based CT hemorrhage detection system validated across a global dataset detected bleeding with about 92% sensitivity and 96% specificity, slightly outperforming general radiologists in sensitivity while matching them closely in specificity.25PubMed Central. Real world validation of an AI-based CT hemorrhage detection tool These tools are not replacing the clinician’s judgment about what to do with the information, but they may speed up the step of recognizing that a hemorrhage exists and roughly how large it is, especially in settings without around-the-clock neuroradiology coverage. Automated volumetric measurements of hematoma size could also reduce the variability inherent in the ABC/2 method, potentially making the ICH score’s volume component more consistent across hospitals.