Illustration: A translucent brain in a ghosted head profile, a copper pool expanding deep in one hemisphere.
← EMA Cited
EMA Cited · Neurology · Stroke

Intracerebral hemorrhage

Updated Oct 24, 2026Adults · ED management13 sources · 12 checked in PubMed3 figuresESO/EANS 2025

Bottom line

  • Hematoma growth drives outcome. Most growth happens between 30 minutes and 3 hours after onset.1 Each 10% of growth raised the hazard of death (HR 1.05).2
  • For hematomas under 30 mL, lower systolic pressure below 140 mm Hg within 6 hours, ideally within 2. Don't drop it more than 70 mm Hg, or actively below 110.1 This is expert consensus; the trials were mixed.3,4
  • Reverse anticoagulants immediately. Warfarin: 4-factor PCC plus vitamin K 10 mg IV. Dabigatran: idarucizumab 5 g.1 Factor Xa inhibitors in the US: PCC, since andexanet left the US market in December 2025.5
  • Don't transfuse platelets for an antiplatelet-associated bleed. In PATCH, it raised the odds of death or dependence (OR 2.05).6
  • Use a care bundle. Targets for blood pressure, glucose, fever and anticoagulant reversal within an hour improved functional outcome (OR 0.86 for a worse outcome).7

First hour

  1. Check for anticoagulants and reverse them now (table below). For warfarin, aim for an INR under 1.3 within an hour.1
  2. Lower blood pressure to a systolic under 140 mm Hg if the hematoma is under 30 mL. Be cautious above 220 mm Hg, with hematomas over 30 mL, or before planned evacuation.1
  3. Treat glucose and fever: glucose 110–141 mg/dL (6.1–7.8 mmol/L) without diabetes, 141–180 mg/dL (7.8–10 mmol/L) with diabetes, and temperature under 37.5 °C. Avoid hypoglycemia.1
  4. Look for a cause. Use CT or MR angiography to select patients for catheter angiography, and consider a structured tool such as DIAGRAM.1
  5. Call neurosurgery early for large bleeds, intraventricular blood, or a cerebellar hemorrhage.1
  6. Admit to a stroke unit, or to intensive care if needed (strong recommendation).1
  7. Don't limit care by default in the first 24 hours. The guideline advises against general do-not-resuscitate or treatment-limiting policies this early.1
Axial non-contrast head CT showing a large bright right deep hematoma with a dark rim of edema compressing the right lateral ventricle
FIG 1Deep hematoma with edema. A large right-sided deep hemorrhage, ringed by darker edema, compresses the right lateral ventricle.James Heilman, MD · Wikimedia Commons · CC BY-SA 4.0
Axial head CT photographed from a monitor showing bright blood filling both frontal horns of the lateral ventricles and the left occipital horn
FIG 2Intraventricular extension. Blood fills both frontal horns and the left occipital horn. Intraventricular blood raised the hazard of death in pooled trial data (HR 2.23).2Glitzy queen00 at English Wikipedia · Wikimedia Commons · Public domain

Severity: the ICH score

FindingPoints
GCS 3–42
GCS 5–121
Age ≥ 801
Infratentorial origin1
Volume ≥ 30 mL1
Intraventricular blood1

In the derivation cohort, all 26 patients scoring 0 survived and all 6 scoring 5 died at 30 days.8 Use the score to share prognosis, not as the main basis for decisions, because early pessimism can become a self-fulfilling prophecy.1

Mortality exceeds 90% with volumes of 60 mL or more, and runs about 30–55% at 30 mL or more.1

Blood pressure

  • Who: hematomas under 30 mL. For systolic 150–220 mm Hg overall, the guideline calls the balance of benefit and harm uncertain.1
  • When: at least within 6 hours, and as fast as possible. In the guideline's meta-analysis of 9 trials, the benefit grew the earlier treatment started.1
  • How far: under 140 mm Hg. Avoid a fall of more than 70 mm Hg, and don't actively push below 110. Keep it under 140 for up to 7 days, and minimize swings.1
  • Agent: the guideline names no specific drug. Use a titratable IV agent per your protocol.1
FIG 3The blood pressure window. Drawn for this page from the ESO/EANS consensus statement for hematomas under 30 mL. Time zero is symptom onset. Most hematoma growth happens between 30 minutes and 3 hours.1

What the trials show

  • INTERACT2 (n = 2,839): target under 140 within 1 hour vs under 180. Death or major disability was 52.0% vs 55.6%, which missed significance (P = 0.06). The ordinal analysis favored intensive lowering (OR 0.87).3
  • ATACH-2 (n = 1,000; mean starting systolic 201): 110–139 vs 140–179 with nicardipine. Death or disability was 38.7% vs 37.7%, and kidney adverse events doubled (9.0% vs 4.0%).4
  • INTERACT3 (n = 7,036, stepped-wedge cluster trial): a bundle of early blood pressure lowering, glucose and fever control, and warfarin reversal improved function at 6 months (common OR 0.86) and cut serious adverse events (16.0% vs 20.1%).7
Sources disagree

How low to go: ATACH-2's lower target (110–139) brought no benefit and more kidney harm.4 That is why the guideline sets a floor of 110 and limits the drop to 70 mm Hg.1 The warfarin reversal target also differs: an INR under 1.3 in the guideline, under 1.5 in INTERACT3.1,7

Reversal and hemostatic drugs

SituationAdult doseNotes
Anticoagulants
Warfarin, INR ≥ 2.04-factor PCC 30–50 IU/kg + vitamin K 10 mg IVExpert consensus.1
Warfarin, INR 1.3–1.94-factor PCC 10 IU/kg + vitamin K 10 mg IVExpert consensus.1
DabigatranIdarucizumab 2 × 2.5 g IVGive immediately.1
Apixaban, rivaroxabanPCC, per institutional protocolBenefit unproven; weigh thrombosis risk.1 Standard in the US.5
Antiplatelets
Platelet transfusionNot recommendedWeak recommendation against, moderate evidence.1
DesmopressinNo recommendationUncertain; trials ongoing.1
No antithrombotic drug
Recombinant factor VIIaNot routinelyWeak recommendation against.1
Tranexamic acidMay be consideredIf trial enrollment isn't possible.1

Unreviewed doses. Verify against the current label and your institution's protocol before giving anything.

  • INCH (n = 50): PCC 30 IU/kg brought the INR to 1.2 or less within 3 hours in 67% of patients, vs 9% with plasma 20 mL/kg. Five plasma-treated patients died of hematoma expansion; none in the PCC group did.9
  • ANNEXA-I (n = 530, factor Xa inhibitor within 15 hours): andexanet improved hemostatic efficacy (67.0% vs 53.1%), but thrombotic events (10.3% vs 5.6%) and ischemic strokes (6.5% vs 1.5%) rose. Function and death at 30 days did not differ.10
  • PATCH (n = 190): platelet transfusion raised the odds of death or dependence at 3 months (OR 2.05), and more patients died in hospital (24% vs 17%).6
  • TICH-2 (n = 2,325): tranexamic acid within 8 hours did not improve function at 90 days (OR 0.88). Deaths by day 7 were fewer (9% vs 11%), but not by day 90.11
Sources disagree

The European guideline suggests considering andexanet within 15 hours of the last apixaban or rivaroxaban dose and 12 hours of onset.1 In the US, andexanet was withdrawn from the market effective December 22, 2025, over safety concerns, and 4-factor PCC is the standard.5

Surgery

  • Lobar supratentorial: the guideline suggests minimally invasive evacuation within 24 hours. The effect in deep bleeds is uncertain.1
  • ENRICH (n = 300; lobar or anterior basal ganglia, 30–80 mL, within 24 hours): surgery improved utility-weighted function (0.458 vs 0.374) and 30-day mortality (9.3% vs 18.0%). The benefit came from lobar bleeds.12
  • MIND (n = 236; 20–80 mL, within 72 hours, mostly deep): no benefit (OR 1.03). It stopped early.13
  • Cerebellar hemorrhage over 15 mL: surgical evacuation is suggested to improve survival.1
  • Intraventricular blood with hydrocephalus and reduced consciousness: an external ventricular drain is recommended.1
  • Severe deep bleeds (age 18–75, GCS 8–13, 30–100 mL, within 72 hours): consider decompressive craniectomy without evacuation.1

Other early decisions

  • No prophylactic anti-seizure drugs (weak recommendation against, moderate evidence).1
  • No steroids or other anti-inflammatory drugs outside trials (strong recommendation against).1
  • Intermittent pneumatic compression for immobile patients, to prevent proximal DVT.1

Pitfalls

  • Dropping a systolic of 220 straight to 110. The guideline caps the fall at 70 mm Hg.1
  • Ordering platelets for a patient on aspirin or clopidogrel.6,1
  • Reaching for plasma when PCC is available for warfarin reversal.9,1
  • Ordering andexanet in a US hospital. It is no longer sold there.5
  • Quoting a high ICH score as a reason to withhold care on day one.1
  • Missing a cerebellar bleed that needs surgery.1

How strong is the evidence?

The guideline found strong evidence only for stroke unit care and long-term blood pressure control. Most acute recommendations are weak or expert consensus.1 The blood pressure trials missed their primary outcomes.3,4 The care bundle trial was positive, but it ran in mostly low- and middle-income countries.7 The 2022 American Heart Association guideline could not be accessed for this page.

On-shift teaching

Questions to ask a learner at the bedside. Tap a question to see the answer. Every answer uses facts cited above.

Q1Name the five components of the ICH score.
GCS (3–4 = 2 points, 5–12 = 1), age ≥ 80, infratentorial origin, volume ≥ 30 mL, and intraventricular blood.8
Q2A 15 mL bleed with a systolic of 210. What's your target, and what are the limits?
Under 140 mm Hg, started as early as possible, within 6 hours. Don't drop it more than 70 mm Hg, and don't actively go below 110.1
Q3Why not aim for 110–139?
In ATACH-2, that target didn't reduce death or disability (38.7% vs 37.7%) and doubled kidney adverse events (9.0% vs 4.0%).4
Q4How do you reverse warfarin with an INR of 3.1?
4-factor PCC 30–50 IU/kg plus vitamin K 10 mg IV.1 In INCH, PCC normalized the INR within 3 hours in 67% vs 9% with plasma.9
Q5The patient takes clopidogrel. Should you transfuse platelets?
No. In PATCH, platelet transfusion raised the odds of death or dependence (OR 2.05).6
Q6What did ANNEXA-I show, and why does it matter less in the US now?
Andexanet improved hemostatic efficacy (67.0% vs 53.1%) but increased thrombotic events (10.3% vs 5.6%).10 It was withdrawn from the US market in December 2025.5

References

Evidence base

What this page is built on

13 sources, cited 68 times. Dots are sources, filled by how far each was read; stronger color means a larger share of the citations. Tap a tier to see its sources.

Read in full · 2Abstract read · 11Record only · 0Link checked · 0

Checked by the writer, not a physician. The dots show how each source was read, not that anyone has confirmed the claims.

How this page was made
  • Written from scratch. No source text was copied; facts are restated and cited.
  • Anchored on the 2025 ESO/EANS guideline, read in full on PubMed Central, including its synopsis of recommendations and consensus statements. Strengths and grades are quoted from it. The 2022 AHA/ASA guideline (PMID 35579034) could not be accessed, so none of its recommendations are stated here.
  • Each PubMed citation was resolved through NCBI E-utilities and each number checked against that paper's abstract. Doses come from the guideline text; no blood pressure drug doses are given because the guideline names none.
  • Adults with spontaneous (non-traumatic) hemorrhage. Traumatic and pediatric hemorrhage are not covered.
  • Figure 3 was drawn for this page from the guideline. Figures 1 and 2 are openly licensed or public-domain images from Wikimedia Commons, credited under each one, and they keep their own licenses.
  • Not for clinical use.
Not for clinical use

AI-generated draft. Not reviewed by a physician. This page is for education only and may contain errors, including in doses. Do not use it to diagnose, treat or make decisions about any patient. Rely on your own clinical judgment, your institution's protocols and current drug labeling. Reading this page does not create a physician-patient relationship. Full terms