Warfarin and Anticoagulation Reversal

The pharmacology, and why the INR lags

Warfarin inhibits vitamin K epoxide reductase, so vitamin K cannot be recycled, so the liver cannot carboxylate the vitamin K dependent factors: II, VII, IX and X, along with the natural anticoagulants protein C and protein S. It does nothing to the factors already circulating. You are waiting for them to decay.

Their half-lives are what create every clinically important quirk of this drug. Factor VII is roughly 6 hours. Protein C is about 8. Factor IX is around 24, factor X around 40, and factor II is roughly 60 hours or more.

Three consequences you need to hold in your head:

• The INR rises before the patient is anticoagulated. The prothrombin time is most sensitive to factor VII, the fastest to fall. An INR of 2.3 on day two of warfarin reflects a depleted factor VII and a nearly intact factor II. That patient is not protected. This is why you overlap with a parenteral agent for at least 5 days and until the INR is therapeutic on two occasions at least 24 hours apart.
• Early warfarin is transiently prothrombotic. Protein C falls as fast as factor VII, faster than the procoagulant factors it opposes. That imbalance is the mechanism behind warfarin induced skin necrosis and behind venous limb gangrene when warfarin is started in acute HIT.
• The INR falls before the patient is safe. After vitamin K or PCC, a corrected INR reflects restored factor VII. It does not guarantee restored factor II. Do not read a normalized INR as proof that hemostasis is normal.

The supratherapeutic INR, by band

The first question is always the same, and it is not about the number: is the patient bleeding? Everything downstream depends on the answer. Illustrative approach:

Vitamin K: route and dose logic

Vitamin K is the only part of warfarin reversal that is durable. Everything else is a temporary transfusion of factors that will decay.

• Oral is well absorbed, predictable, and the right route for the patient who is not bleeding. Expect a meaningful INR fall by about 24 hours.
• Intravenous is faster, with the INR beginning to move at roughly 4 to 6 hours and substantially corrected by 12 to 24. Reserve it for serious bleeding and urgent procedures. There is a small risk of an anaphylactoid reaction, so it should be diluted and infused slowly, over 20 to 30 minutes, and not pushed.
• Subcutaneous absorption is erratic and unpredictable. It has largely been abandoned. Do not choose it because you are nervous about the IV route.
• Intramuscular injection into an anticoagulated patient is a hematoma. Do not.

The dose logic: use the smallest dose that achieves what you need. Big doses of vitamin K make the patient resistant to warfarin for a week or more, which is a real problem in someone who needs to be anticoagulated again. That is precisely why 10 mg is for the bleeding patient and 2.5 mg is for the high number.

4-factor PCC versus fresh frozen plasma

Where 4-factor PCC is available, it is the preferred product for serious warfarin associated bleeding, and the reasons are practical rather than subtle.

PCC is dosed by body weight and presenting INR, commonly in the range of 25 to 50 units/kg with a cap, and your protocol will have the table. It works within minutes. FFP requires roughly 15 mL/kg, which is about four units in an average adult, and in an 80 year old with an intracranial hemorrhage and a stiff ventricle you will put them into pulmonary edema before you correct the INR.

Reversing the direct oral anticoagulants

First, find out what they actually take and when they last took it. Half the time the answer changes the plan entirely, because a DOAC taken 30 hours ago in a patient with normal renal function is largely gone.

Know before the emergency which of these your hospital actually stocks and how long it takes to get. Finding out at the bedside costs you forty minutes.

Reversal quick reference

Periprocedural management

Two questions: how much bleeding does this procedure cause, and how badly does this patient clot if you stop the drug.

Plenty of procedures need nothing at all. Cataract surgery, most dental work, many skin procedures and diagnostic endoscopy without biopsy are routinely done on a therapeutic INR. Stopping anticoagulation for them creates risk and prevents nothing.

For a procedure that does require interruption, warfarin is typically held about 5 days beforehand, with an INR checked on the day before or the morning of. If it is still elevated, low dose oral vitamin K is a reasonable rescue rather than canceling.

Bridging

Bridging with therapeutic heparin or enoxaparin during the interruption used to be routine and mostly should not be. In atrial fibrillation, trial evidence found that bridging did not reduce stroke and did substantially increase major bleeding. The population that still generally gets bridged is the genuinely high thrombotic risk group: mechanical mitral valves, older caged ball or tilting disc valves, a stroke or systemic embolus within the last 3 months, VTE within the last 3 months, and severe thrombophilia.

DOACs are not bridged. Their onset and offset are fast enough that you simply hold them, typically for 1 to 2 days before the procedure depending on renal function and the bleeding risk of the operation, and restart when hemostasis is secure.

Interactions that actually matter

The interaction list for warfarin runs to hundreds of entries and is useless at the bedside. These are the ones that cause real events on a medical ward:

• Raise the INR, sometimes dramatically: trimethoprim-sulfamethoxazole, metronidazole, fluconazole and the other azoles, amiodarone, fluoroquinolones and macrolides. Amiodarone is the sneaky one because the effect builds over weeks and the patient is discharged before anyone sees it.
• Lower the INR: rifampin, carbamazepine, phenytoin, phenobarbital, St John's wort. Rifampin is the classic, and the effect is large.
• Raise bleeding risk without touching the INR: aspirin, clopidogrel, NSAIDs, SSRIs. A normal INR in a patient on warfarin plus ibuprofen is not reassurance.
• Other: sustained high dose acetaminophen raises the INR. Acute alcohol binges raise it, chronic use with liver induction can lower it. Enteral tube feeds reduce warfarin absorption and cause a stubbornly subtherapeutic INR that resolves when the feeds are held around the dose.
• Diet: the goal is consistency, not avoidance. Telling a patient to stop eating green vegetables is bad advice. Telling them to eat about the same amount every week is good advice.

Practical rule: any time you start an antibiotic in a patient on warfarin, decide when you will recheck the INR before you sign the order, and write it in the plan.

Restarting anticoagulation after a bleed

This is genuinely difficult and I want to be honest that there is no clean rule. You have a patient with a real indication for anticoagulation who has just demonstrated a real capacity to bleed on it, and both risks are ongoing.

What actually helps:

• Fix the lesion. A bleeding duodenal ulcer that has been clipped and treated is a different risk profile from an unexplained melena. Most gastrointestinal bleeding sources can be treated, and once treated, restarting is often reasonable within days rather than weeks. Observational data has generally suggested that not restarting carries a higher mortality than restarting in patients with a strong indication.
• Intracranial hemorrhage is the hard case. Restarting is typically deferred by weeks, and the timing depends on the type of bleed, the indication, and whether the underlying cause, amyloid angiopathy for example, makes recurrence likely. This is a neurology and neurosurgery decision, not an overnight one.
• Reconsider the indication itself. Sometimes the honest answer is that the anticoagulation was never that strongly indicated, or that a left atrial appendage occlusion referral is the better path, or that the antiplatelet the patient was also taking can simply be stopped.
• Involve the patient. This is exactly the kind of decision where the patient's own weighting of a stroke against a bleed should be part of the record.

Related: common night calls for the bleeding patient at 2 a.m., and the discharge summary, where the anticoagulation plan and its follow-up INR date have to be written down or the plan does not exist.