ICH Score
Intracerebral Hemorrhage Prognosis
Mortality prediction reflects Hemphill et al. (2001) validation.
Clinical Intelligence
Select patient metrics to execute high-fidelity mortality prediction based on the Original Hemphill ICH Score.
Ready For Calculation
Guidelines & Evidence
Verified
Last Review: 2026-07-17
When to Use
Primary Indications
Predicting 30-day mortality in patients with spontaneous (non-traumatic) intracerebral hemorrhage – the most widely validated prognostic scale for ICH
Risk stratification on presentation – guides intensity of monitoring (ICU vs ward), family discussions about prognosis, and decisions regarding withdrawal of care
Standardizing clinical research – used as entry criterion, stratification variable, and outcome adjustment in ICH trials (e.g., CLEAR III, MISTIE III, FAST trials)
Quality benchmarking – comparing observed vs expected mortality across stroke centers (risk-adjusted outcomes)
Early communication with families – provides a data-driven estimate of survival probability, though MUST be accompanied by caveats (see clinical edge)
Guiding transfer decisions – patients with ICH score ≥3 have high mortality (≥40-50%) and may warrant transfer to tertiary centers with neurosurgery and intensive care
Contraindications / Limitations
DO NOT use for traumatic ICH (e.g., hemorrhagic contusions, epidural/subdural hematoma) – derived and validated only for spontaneous, non-traumatic ICH
DO NOT use for hemorrhagic conversion of ischemic stroke – different pathophysiology (reperfusion injury, not primary vessel rupture)
DO NOT use to justify early withdrawal of care – high scores (4-5) have high mortality, but AHA/ASA guidelines recommend aggressive full care for at least the first 24 hours to avoid self-fulfilling prophecy
Limited utility in low-resource settings without CT – requires CT for ICH volume, location, and intraventricular hemorrhage assessment
Poor discrimination at extremes – Score 0 (17-23% mortality in some studies, not 0% across all populations), Score 5 (variable mortality depending on withdrawal of care practices)
Does NOT predict functional outcome – only 30-day mortality. A surviving patient with high ICH score may have severe disability (modified Rankin Scale 4-5)
Does NOT account for hematoma expansion – the "spot sign" on CTA predicts expansion and may worsen prognosis beyond baseline score
Performance Metrics (30-Day Mortality)
| ICH Score | Original Cohort (Hemphill 2001, n=152) | Uganda Cohort (Abdallah 2018, n=73) | Interpretation |
|---|---|---|---|
| 0 | 0% (0/26) | 17% (2/12) | Low risk – excellent prognosis in high-income settings; higher mortality (17%) in Uganda suggests limited ICU access affects even "good prognosis" patients |
| 1 | 13% (5/38) | 23% (3/13) | Low-intermediate risk – mortality increases modestly |
| 2 | 26% (11/42) | 47% (8/17) | Intermediate risk – significant mortality (26-47%); ICU monitoring recommended |
| 3 | 72% (13/18) | 48% (11/23) | High risk – Uganda cohort had lower mortality (48% vs 72%) possibly due to younger age (mean 60 vs 70+), HIV+ status, or different withdrawal practices |
| 4 | 97% (1/1) | 100% (6/6) | Very high risk – near-uniform mortality in both studies irrespective of setting |
| 5 | 100% (6/6) | 100% (2/2) | Extremely high risk – universal mortality at 30 days |
| 6 | (Not present in original) | (Not present) | Theoretical max score (GCS 3-4=2pts + age≥80=1 + infratentorial=1 + volume≥30=1 + IVH=1 = 6) but rarely seen |
Additional Predictors Identified in Uganda Validation Study
In the 2018 Uganda validation study (n=73, mean age 60, 45% female, 14% HIV-positive), two additional factors independently predicted 30-day mortality beyond the standard ICH score:
Female sex – adjusted relative risk (aRR) 2.17 (95% CI 1.32-3.59). Women had higher case fatality at lower ICH scores compared to men. Mechanism unclear; prior studies have shown conflicting results (some show higher mortality in men, others no difference).
HIV infection – aRR 1.92 (95% CI 1.07-3.43). HIV has been previously associated with poor stroke outcomes in sub-Saharan Africa, particularly with advanced disease and immunosuppression.
Clinical implication: In settings with high HIV prevalence and/or where sex differences are observed, the ICH score may moderately underperform. The addition of 1 point for female sex and 1 point for HIV seropositivity improved the AUC from 0.73 to 0.81 (modest improvement). Further validation needed before modifying the standard score.
How it Works
ICH Score Components (Hemphill 2001)
| Factor | 0 Points | 1 Point | 2 Points | Points Awarded |
|---|---|---|---|---|
| Glasgow Coma Scale (GCS) | 13–15 | 5–12 | 3–4 | ______ |
| Age | <80 years | ≥80 years | — | ______ |
| ICH Location | Supratentorial | Infratentorial (cerebellum, brainstem) | — | ______ |
| ICH Volume (ABC/2 method) | <30 cm³ | ≥30 cm³ | — | ______ |
| Intraventricular Hemorrhage (IVH) | Absent | Present (any amount) | — | ______ |
| TOTAL SCORE | ______ (0-6) |
Step-by-Step Calculation
ABC/2 Volume Measurement Technique
ABC/2 Method (Kothari et al. 1996):
A = Greatest hemorrhage diameter on the slice with the largest area of hemorrhage (measured in cm).
B = Greatest diameter perpendicular to A on that same slice (measured in cm).
C = Number of CT slices with visible hemorrhage × slice thickness (usually 0.5 cm or 1 cm).
Volume (cm³) = (A × B × C) / 2
Example: A = 4 cm, B = 3 cm, C = 8 slices × 0.5 cm = 4 cm. Volume = (4 × 3 × 4) / 2 = 48 / 2 = 24 cm³ (<30 cm³, no point).
Accuracy: Good correlation with planimetry (r=0.85-0.95) for ellipsoid-shaped hemorrhages. Less accurate for irregular or multilobulated hemorrhages (underestimates volume by 10-30%). Still accepted for clinical scoring.
Risk Stratification Categories
| ICH Score | Risk Category | Predicted 30-Day Mortality (Range Across Studies) | Suggested Management |
|---|---|---|---|
| 0-1 | Low risk | 0-23% | ICU or step-down monitoring. Aggressive medical management (BP control, reversal of coagulopathy, seizure prophylaxis). Good prognosis for survival, but may still have significant disability. |
| 2 | Intermediate risk | 26-47% | ICU monitoring strongly recommended. Neurosurgical consultation if cerebellar ICH >3 cm or lobar ICH with mass effect. Prepare family for possible prolonged course. |
| 3 | High risk | 48-72% | ICU mandatory. Aggressive medical management but with realistic prognosis discussion. Evaluate for early transfer to tertiary center if not already there. Consider DNR status discussion after 24 hours of full care. |
| 4-6 | Very high risk | 97-100% | ICU care, but high likelihood of mortality regardless of intervention. After 24 hours of full care and repeat CT showing no improvement, family discussion about goals of care (comfort measures) appropriate. However, avoid early withdrawal (<24 hours) due to risk of self-fulfilling prophecy. |
Clinical Pearls
Critical Pearl #1: Avoid the Self-Fulfilling Prophecy
The single most dangerous misuse of the ICH score is using a high score (e.g., 4 or 5) to justify early withdrawal of life-sustaining treatment. AHA/ASA guidelines (2015, reaffirmed 2022) state: "In patients with ICH, regardless of ICH score, aggressive full care is recommended for at least the first 24 hours after ICH onset." The original derivation cohort had high mortality at high scores, but this was in part due to withdrawal of care (limitation of life support) based on perceived poor prognosis. Several subsequent studies have shown that early aggressive care (including intubation, ICP monitoring, surgical evacuation when appropriate) can improve outcomes even in high-score patients, particularly younger ones. Clinical approach: (1) Provide full ICU care for first 24 hours. (2) Obtain repeat CT at 24 hours to assess for hematoma expansion (which worsens prognosis) or stability. (3) Only then discuss prognosis with family, using the ICH score as ONE data point, not a definitive prediction. (4) Document that the score was derived from populations that may not reflect your patient (e.g., younger, HIV+, different withdrawal practices).
Critical Pearl #2: The "Spot Sign" Overrides the ICH Score
The ICH score is calculated based on baseline CT findings (at presentation). However, if a CTA (CT angiogram) is performed and shows the "spot sign" (contrast extravasation within the hematoma indicating active bleeding), this predicts ongoing hematoma expansion and independently increases mortality risk, even in patients with low ICH scores. Management: (1) Spot sign positive patients require more aggressive BP control (target SBP <140, ideally <130), (2) Consider reversing any coagulopathy emergently, (3) Repeat CT in 6 hours to assess expansion, (4) Lower threshold for surgical evacuation if expanding and accessible. Do NOT be falsely reassured by a low ICH score if spot sign present.
Critical Pearl #3: External Validation in Low-Resource Settings (Uganda Study)
The 2018 Uganda validation study (Abdallah et al.) is the first to validate the ICH score in sub-Saharan Africa. Key findings: (1) The ICH score remained a strong predictor of 30-day mortality (aRR 1.48 per 1-point increase). (2) However, mortality at low scores (score 0: 17%; score 1: 23%) was higher than in the original US cohort (score 0: 0%; score 1: 13%). This likely reflects limited access to intensive care, neurosurgery, and neurocritical care in Uganda. (3) Patients were younger (mean age 60 vs 70+ in US studies), suggesting ICH occurs at earlier ages in Africa (possibly due to hypertension, HIV, or genetic factors). (4) HIV infection (14% of cohort) and female sex were independent predictors of mortality beyond the ICH score. Clinical implication: In low-resource settings, the ICH score still stratifies risk but absolute mortality is higher at every score level. Consider adding 1 point for HIV-positive or female patients in these settings (research ongoing).
Common Pitfalls in Using the ICH Score
Using for traumatic ICH – The score was derived for spontaneous (non-traumatic) ICH. Traumatic hemorrhagic contusions have different pathophysiology (shear injury, not hypertensive vessel rupture) and different prognosis. Do NOT apply.
Mis-measuring ICH volume – Common errors: (1) Including perihematomal edema in volume measurement (only measure hyperdense blood, not surrounding hypodensity), (2) Using wrong slice thickness (ensure C = slices × thickness in cm, not number of slices alone), (3) Measuring in mm instead of cm (volume will be off by factor of 1000). Standard: measure in cm, then volume in cm³ (mL).
Ignoring intraventricular hemorrhage (IVH) – IVH is an independent predictor of mortality. Even a small amount of blood in the occipital horn of the lateral ventricle or third ventricle adds 1 point. Do NOT require large IVH to assign the point.
Using GCS after sedation or paralytics – If patient was intubated and sedated before GCS assessment, use the last reliable GCS before sedation (often from EMS report or emergency department triage). If no reliable GCS available, treat as missing; score cannot be calculated.
Failing to account for coagulopathy – The ICH score does NOT include INR or antiplatelet use, yet coagulopathy dramatically worsens prognosis (higher expansion risk, higher mortality). A patient with ICH score 1 (e.g., GCS 15, age <80, supratentorial, volume 25 cm³, no IVH) but INR 4.0 on warfarin has far worse prognosis than score suggests. Do NOT be falsely reassured; reverse coagulopathy emergently.
Using the score to predict functional outcome – The ICH score predicts 30-day MORTALITY, not functional outcome. A surviving patient with score 3 may have severe disability (modified Rankin Scale 5). For long-term functional outcome prediction, other scales (e.g., FUNC score) may be more appropriate.
Over-reliance on the score in young patients – A 40-year-old with ICH score 3 (e.g., GCS 10, age <80, infratentorial, volume 35 cm³, IVH present) has much better prognosis than the score suggests (original cohort was elderly, mean age >70). The age cutoff of ≥80 was chosen because few patients <80 died in the derivation cohort; however, younger patients with high scores may survive with aggressive care.
ICH Score vs Other Intracerebral Hemorrhage Prognostic Scales
| Scale | Components | Outcome Predicted | Strengths | Weaknesses |
|---|---|---|---|---|
| ICH Score (Hemphill 2001) | GCS, age, infratentorial, volume≥30, IVH | 30-day mortality | Simple, widely validated, integer-based, used in trials | Does not include coagulopathy, does not predict functional outcome, less accurate in young patients |
| FUNC Score (Rost 2008) | Age, GCS, ICH volume, IVH, pre-ICH cognitive impairment | 90-day functional independence (mRS 0-3) | Predicts functional outcome (not just mortality), useful for rehabilitation planning | More complex, less widely validated, requires pre-ICH cognitive status (often unknown) |
| MAX-ICH Score (Ji 2018) | GCS, age, IVH, anticoagulation use | 30-day mortality | Includes coagulopathy (warfarin use), validated in modern cohort | Newer, less widely adopted, does not include ICH volume or location |
| ICH-GS (Cheung 2014) | GCS, age, ICH volume, IVH, glucose, systolic BP, creatinine | 30-day mortality | Includes metabolic parameters (glucose, renal function), more precise risk estimation | Complex (requires lab values), not as widely validated |
Next Steps
Step-by-Step Management Based on ICH Score
Sample Clinical Documentation for ICH Scoring
Example Note (ED Admission): "Patient is a 75-year-old female with history of hypertension and atrial fibrillation on warfarin (INR 2.8 at presentation). Presenting GCS 9 (E2, V2, M5). Head CT shows left basal ganglia ICH volume 45 cm³ (A=5cm, B=4cm, C= (5 slices × 0.5cm)=2.5cm → (5×4×2.5)/2 = 25cm³? Correction: Actually calculated as (5×4×2.5)/2 = 50/2? Let me recalc properly: A=5, B=4, C=9 slices × 0.5cm = 4.5cm → (5×4×4.5)/2 = 90/2 = 45 cm³). Intraventricular hemorrhage present (blood in occipital horn of left lateral ventricle). No infratentorial involvement. ICH Score = GCS 5-12 (1 point) + age ≥80? No (0) + infratentorial? No (0) + volume ≥30 (1 point) + IVH present (1 point) = TOTAL SCORE 3 out of 6 (predicted 30-day mortality ~48-72% based on validation studies). Plan: Admit to neuro-ICU. Reverse warfarin with IV vitamin K 10mg + 4-factor PCC 2000 units. BP goal SBP <140 mmHg with nicardipine drip. Neurosurgery consulted for possible EVD if hydrocephalus develops. Avoid early withdrawal of care; reassess at 24 hours with repeat CT."
Recommended Family Discussion Script (High ICH Score, e.g., 4)
"Your family member has had a severe bleeding stroke in the brain. We use a scoring system called the ICH score to help us think about prognosis. With a score of 4 out of 6, studies show that about 97-100% of similar patients do not survive to 30 days. However, those studies included many older patients and some who had care withdrawn early. Your family member is [younger/middle-aged] and we will provide aggressive full care for the next 24 hours to give the best possible chance. After 24 hours, we will repeat the CT scan to see if the bleeding has expanded or stabilized. At that point, we will re-evaluate and discuss whether continued aggressive care is likely to achieve a meaningful recovery. For now, the priority is controlling blood pressure, reversing the blood thinner, and monitoring for brain swelling. It is important that you know we will not give up in the first 24 hours. We will update you in person tomorrow morning after the repeat CT. In the meantime, do you have any questions about what we are doing right now?"
The Evidence
Original Derivation Study (2001)
The ICH score: a simple, reliable grading scale for intracerebral hemorrhage
Hemphill JC 3rd et al. • Stroke. 2001;32(4):891-897. doi: 10.1161/01.str.32.4.891. PMID: 11283388.
View SourceExternal Validation in Uganda (2018)
Validation of the Intracerebral Hemorrhage Score in Uganda: A prospective cohort study
Abdallah A et al. • Stroke. 2018;49(12):3063-3066. doi: 10.1161/STROKEAHA.118.022057. PMID: 30571425; PMCID: PMC6309793.
View SourceABC/2 Volume Measurement Validation
The ABCs of measuring intracerebral hemorrhage volumes
Kothari RU et al. • Stroke. 1996;27(8):1304-1305. doi: 10.1161/01.str.27.8.1304. PMID: 8711791.
View SourceAHA/ASA Guidelines for ICH Management
Guidelines for the Management of Spontaneous Intracerebral Hemorrhage: A Guideline for Healthcare Professionals From the American Heart Association/American Stroke Association
Hemphill JC 3rd et al. • Stroke. 2015;46(7):2032-2060. doi: 10.1161/STR.0000000000000069. Epub 2015 May 28. PMID: 26022637.
View SourceOrigins & History
J. Claude Hemphill III and the UCSF ICH Score
The ICH Score was developed by Dr. J. Claude Hemphill III (born 1965), an American neurointensivist and neurologist at the University of California, San Francisco (UCSF). At the time (late 1990s), traumatic brain injury had the Glasgow Coma Scale, subarachnoid hemorrhage had the Hunt-Hess and WFNS scales, and ischemic stroke had the NIH Stroke Scale. But for intracerebral hemorrhage (the deadliest stroke subtype) no standardized prognostic scale existed. Hemphill and colleagues (David C. Bonovich, Louis Besmertis, Geoffrey T. Manley, S. Claiborne Johnston) retrospectively reviewed 152 consecutive ICH patients presenting to UCSF from 1997-1998. Using logistic regression, they identified five independent predictors of 30-day mortality. They deliberately weighted the predictors to create an integer-based scale (0-6) that could be calculated easily at the bedside. The score was published in *Stroke* in April 2001 and quickly became the most widely used ICH prognostic scale worldwide, cited over 3,000 times. Subsequent studies validated it in Asian, European, and African populations (including the 2018 Uganda study by Abdallah et al.). Hemphill later served as Chair of the AHA/ASA ICH Guideline Writing Committee (2015) and as President of the Neurocritical Care Society (2010-2012). The ICH score remains a core component of stroke education and clinical practice.
Key Contributors and Timeline
| Year | Contributor(s) | Institution | Contribution |
|---|---|---|---|
| 1996 | Kothari RU, Brott T, Broderick JP | University of Cincinnati | ABC/2 method for measuring ICH volume – enabled rapid, reliable volume assessment without planimetry. Essential component of ICH score. |
| 2001 | Hemphill JC, Bonovich DC, Besmertis L, Manley GT, Johnston SC | University of California, San Francisco (UCSF) | Original derivation of ICH Score (n=152). Published in Stroke. Proposed 5-component scale (GCS, age, infratentorial, volume, IVH) with 0-6 range. |
| 2004-2013 | Multiple validation studies (Clarke J, Fernandes H, Godoy DA, Jamora RD, Parry-Jones AR, etc.) | Global (US, UK, Philippines, Argentina, Brazil, China, Japan, Korea) | External validation of ICH score across diverse populations. Confirmed stepwise increase in mortality with score. Noted higher mortality in community vs academic settings. |
| 2015 | Hemphill JC (chair), Greenberg SM, Anderson CS, et al. (AHA/ASA) | American Heart Association / American Stroke Association | AHA/ASA ICH Guidelines officially endorse ICH score for risk stratification (Class I, Level of Evidence B). Emphasize avoidance of early withdrawal of care based on score. |
| 2018 | Abdallah A, Chang JL, O'Carroll CB, Okello S, et al. | Mbarara University of Science and Technology (Uganda) / Harvard / Mayo Clinic | First validation of ICH score in sub-Saharan Africa (n=73). Demonstrated higher mortality at low scores (0:17%, 1:23%) due to limited ICU access. Identified HIV and female sex as additional risk factors. |
| 2022 | AHA/ASA ICH Guideline Update (Greenberg SM, Ziai WC, et al.) | American Heart Association | Reaffirms ICH score utility but stresses limitations in younger populations and coagulopathic patients. Includes updated BP targets and reversal strategies. |
Limitations of Original Derivation Study
Single-center, retrospective – Derivation from 152 patients at UCSF (tertiary academic center). May not generalize to community hospitals or low-resource settings (as Uganda 2018 study confirmed – higher mortality at low scores).
Temporal bias – Data from 1997-1998 antedated modern ICH management (including aggressive BP lowering per ATACH-2/INTERACT2, 4-factor PCC for warfarin reversal, minimally invasive surgery). Current outcomes may be better at all score levels.
Withdrawal of care bias – In the derivation cohort, 6 of 6 patients with ICH score 5 died, but this was partly due to withdrawal of life support based on perceived poor prognosis. In subsequent studies with less aggressive withdrawal, occasional survival at score 5 has been reported (though still rare).
Age threshold (≥80 years) – Original study had too few patients <80 with high scores to determine whether age interacts with other predictors. Younger patients (e.g., 40-60 years) with high scores may have better outcomes than score suggests.
Not validated for functional outcome – Only 30-day mortality. A surviving patient with high score may have profound disability (mRS 4-5). Clinicians may overestimate quality of life in survivors.
Does not include coagulopathy or antiplatelet use – Original cohort had only 7% on antithrombotics. Modern cohorts have higher rates (20-40% on anticoagulants or antiplatelets). Coagulopathy independently worsens prognosis and should be considered separately.
Last Comprehensive Review: 2026-07-17