4.2 Reversal of Anticoagulation After Acute Hemorrhage
Key Takeaways
- Warfarin-associated ICH: give 4-factor PCC dosed to INR and weight (label: 25/35/50 units/kg for INR 2–<4 / 4–6 / >6, weight capped at 100 kg) plus vitamin K 10 mg IV; FFP is slower and volume-heavy.
- Dabigatran reversal is idarucizumab 5 g IV (two 2.5 g vials); FXa inhibitors use andexanet alfa (low 400 mg bolus plus 480 mg infusion, or high 800 mg plus 960 mg).
- ANNEXA-I (NEJM 2024) showed better hemostatic control versus usual care (67.0% vs 53.1%) but more thrombosis (10.3% vs 5.6%), including ischemic stroke (6.5% vs 1.5%), without a mortality win.
- Heparin reverses with protamine; PATCH found routine platelet transfusion for antiplatelet-associated ICH worsened outcome, so do not transfuse platelets for that indication alone.
- Hold the anticoagulant immediately; restart timing is a later, indication-and-scan judgment, often weeks after ICH versus days after a large ischemic infarct without hematoma.
When the scan shows intracranial hemorrhage in a patient on an anticoagulant, the first moves are the same regardless of brand: stop the drug, control blood pressure, reverse what you can, and call the people who might evacuate a hematoma. This independent OpenExamPrep section is the reversal playbook for warfarin, dabigatran, factor Xa (FXa) inhibitors, heparin, and antiplatelet-associated ICH. Section 4.1 listed pointers; here are the doses, trial numbers, and the hold-versus-restart logic.
Warfarin: 4F-PCC plus vitamin K
Warfarin depletes vitamin K–dependent factors II, VII, IX, and X. Time since the last pill does not rescue you—the INR tells you the current hole in the hemostatic bucket. For major bleeding, four-factor prothrombin complex concentrate (4F-PCC) (U.S. labeled product: Kcentra) plus vitamin K 10 mg IV is the standard pair. Vitamin K rebuilds factors over hours; PCC replaces them in minutes. Giving vitamin K without PCC leaves a bleeding brain waiting. Giving PCC without vitamin K lets the INR rebound when the concentrate wears off.
Label dosing of 4F-PCC is by pretreatment INR and actual body weight, with weight credited only up to 100 kg:
| Pretreatment INR | 4F-PCC dose (Factor IX units/kg) | Maximum units |
|---|---|---|
| 2 to <4 | 25 | 2500 |
| 4–6 | 35 | 3500 |
| >6 | 50 | 5000 |
Worked example: 80 kg, INR 5.0 → 35 × 80 = 2800 units. A 120 kg patient with INR 5.0 is dosed as 100 kg → 3500 units, not 4200. AHA/ASA 2022 ICH guidance supports 25–50 IU/kg when INR is ≥2, and a lower 10–20 IU/kg range is sometimes discussed for INR 1.3–1.9; do not delay PCC while arguing about a 1.8 versus 2.1 in an expanding cerebellar hematoma if local protocol allows treatment.
Infuse vitamin K slowly (anaphylactoid reactions with rapid IV push are the classic warning). Recheck INR shortly after PCC (often within 15–60 minutes) and again as vitamin K takes effect. Fresh frozen plasma (FFP) at about 10–15 mL/kg can supply factors when PCC is unavailable, but it is slower, ABO-constrained, and a volume load that a 90-year-old with diastolic heart failure may not tolerate. FFP is the backup, not the preferred first syringe. Repeat PCC dosing is not well established and adds thrombosis risk; fix vitamin K and surgical bleeding instead of stacking concentrates by habit.
3-factor PCC lacks factor VII and is inferior for warfarin reversal when 4-factor product is available. Recombinant factor VIIa is not a routine warfarin-ICH strategy.
Dabigatran: idarucizumab 5 g
Dabigatran is a direct thrombin inhibitor. The specific reversal agent is idarucizumab, a monoclonal antibody fragment. The labeled dose is 5 g IV, given as two consecutive 2.5 g vials. Do not wait for a thrombin time if the history is clear and the patient is herniating; send a thrombin time or dilute thrombin time if you need to confirm residual effect, remembering that a normal thrombin time argues against clinically important dabigatran. Activated charcoal can be considered if the last ingestion was within about 2–4 hours. Because dabigatran is renally cleared, rebound anticoagulation occurs in kidney failure; a second 5 g dose is sometimes given if bleeding resumes and labs still show effect. If idarucizumab is unavailable, 4F-PCC or aPCC is a weaker, nonspecific fallback. Hemodialysis can remove dabigatran; it does not remove the FXa DOACs in a similarly useful way.
Idarucizumab does nothing for apixaban. Naming the drug class before you name the antidote prevents that error.
Factor Xa inhibitors: andexanet alfa and ANNEXA-I
Andexanet alfa is a decoy FXa that binds apixaban and rivaroxaban (labeled U.S. use). It is given as a bolus then a 120-minute infusion:
| Regimen | Bolus | Infusion |
|---|---|---|
| Low dose | 400 mg (~30 mg/min) | 4 mg/min × 120 min (480 mg) |
| High dose | 800 mg (~30 mg/min) | 8 mg/min × 120 min (960 mg) |
High dose is selected when the last apixaban dose was >5 mg or last rivaroxaban dose was >10 mg (or the dose is unknown) and the last dose was <8 hours ago or unknown. If the last FXa dose was ≥8 hours ago, low dose is used even for high tablet strengths. Edoxaban is often managed with 4F-PCC off-label because andexanet labeling is narrower; ANNEXA-I still informs the class discussion.
ANNEXA-I (Connolly et al., NEJM 2024) randomized patients with FXa-inhibitor–associated ICH to andexanet versus usual care (about 85.5% of usual-care patients received PCC). Hemostatic efficacy was 67.0% versus 53.1% (adjusted difference 13.4 percentage points). Anti-FXa activity fell far more with andexanet. Thrombotic events were 10.3% versus 5.6%; ischemic stroke was 6.5% versus 1.5%. Thirty-day death (27.8% vs 25.5%) and good modified Rankin scores were not appreciably better. Teach the trade: better hematoma-expansion control, more thrombosis including ischemic stroke, no proven mortality benefit. After andexanet, heparin may not anticoagulate reliably for a time—relevant if you had planned an urgent endovascular procedure requiring UFH.
4F-PCC remains a widely used off-label alternative when andexanet is unavailable, delayed, or judged not worth the thrombotic risk. ANNEXA-I usual care was mostly PCC, and PCC lost the hemostatic endpoint, so “PCC is identical” is a false comfort. It is still what many hospitals can give in 15 minutes.
Heparin, antiplatelets, hold, and restart
UFH: stop the infusion. Protamine ~1 mg per 100 units of heparin given in the prior 2–3 hours, typical maximum 50 mg per dose, infused slowly to limit hypotension and anaphylactoid reactions. Heparin’s half-life is about 30–90 minutes, so you reverse remaining drug, not the entire 24-hour bag. Recheck aPTT. Enoxaparin reversal is incomplete (~60% of anti-Xa activity): 1 mg protamine per 1 mg enoxaparin if the dose was within 8 hours; 0.5 mg/mg if later. Do not use FFP to reverse heparin. Fondaparinux is not meaningfully reversed by protamine.
Antiplatelet-associated ICH: PATCH (2016) randomized platelet transfusion versus standard care in patients on antiplatelet therapy with spontaneous ICH. Transfusion worsened death or dependence (adjusted common odds ratio about 2.05). AHA/ASA 2022 treats routine platelet transfusion as harmful when emergency neurosurgery is not planned. Transfusion may still be considered if thrombocytopenia is severe or if the neurosurgeon needs platelets for an operation—mostly an aspirin-era, surgery-linked discussion. Desmopressin (DDAVP) 0.3–0.4 mcg/kg IV is sometimes given to improve platelet function; outcome evidence is uncertain. DDAVP is not a license to ignore PATCH.
Hold versus restart: hold the anticoagulant the minute ICH is recognized. Restart is not a 4 a.m. default. After ICH, many teams wait weeks (AHA/ASA 2022 discusses on the order of 7–8 weeks as a consideration for anticoagulation in selected patients, Class 2b), balancing AF stroke risk against hematoma location (lobar cerebral amyloid angiopathy is more hostile to restart than a deep hypertensive putaminal bleed). After ischemic stroke without hematoma, delay is measured in days and tracks infarct size (section 4.1, ELAN). After andexanet or idarucizumab, the original indication (valve, AF, VTE) is still there—plan a restart when hemostasis is secure, not never and not in the same hour you reversed a cerebellar hematoma.
Putting the reversal tray in order
Work the history while the blood pressure cuff cycles. Last warfarin INR. Last DOAC name, dose, and hour taken. Creatinine (dabigatran and LMWH linger in kidney failure). Whether the patient received thrombolysis (no PCC indication for that alone). Whether emergency surgery is coming (changes the platelet discussion).
| Clinical picture | First reversal move | Common error |
|---|---|---|
| INR 4.8, warfarin, ICH | 4F-PCC by INR/weight + vitamin K 10 mg IV | FFP 2 units as the only therapy |
| Dabigatran this morning, ICH | Idarucizumab 5 g | Andexanet, which does not bind dabigatran |
| Apixaban 5 mg 3 hours ago, ICH | Andexanet (dose by label timing/strength) or 4F-PCC if andexanet is not available | Waiting for an anti-Xa that your lab cannot run at 02:00 |
| Heparin infusion, ICH | Stop UFH + protamine | FFP “to dilute heparin” |
| Aspirin only, ICH, no surgery | Medical ICH care ± consider DDAVP; no routine platelets | PATCH-style platelet transfusion |
Exam traps
Vitamin K 10 mg PO as the sole warfarin-ICH treatment. PCC dosed on 140 kg actual weight without the 100 kg cap. Idarucizumab 2.5 g as a complete dose. Andexanet for dabigatran. Quoting ANNEXA-I as a mortality trial. Ignoring the ischemic-stroke price of andexanet. Platelet transfusion for every clopidogrel-associated ICH. Restarting rivaroxaban the same shift as a reversed lobar ICH “because AF CHA2DS2-VASc is 5.” Using FFP to reverse heparin. Giving high-dose andexanet 12 hours after a 2.5 mg apixaban tablet solely because the hematoma looks scary—timing and last-dose strength choose the regimen, not fear alone.
If two anticoagulants are on the med rec (warfarin overlapping a DOAC during a messy transition), reverse the agents that are actually active. An INR of 1.1 with a dabigatran bottle in the pocket is a thrombin-inhibitor problem. An INR of 4.9 three days after dabigatran was stopped is still a warfarin problem.
An 80 kg patient on warfarin has an expanding ICH and INR 5.0. What is the preferred reversal pair?
ANNEXA-I compared andexanet with usual care (mostly PCC) in FXa-inhibitor ICH. Which statement is accurate?
A patient on aspirin (no anticoagulant, no planned emergency craniotomy) has spontaneous ICH. Which hemostatic plan matches PATCH and AHA/ASA 2022 teaching?