Edibles and Blood Thinners: Warfarin, Eliquis, Plavix, and the Cannabinoid That Actually Moves Your INR

Every cannabis drug interaction on the internet comes with a warning. This one comes with a case file. Warfarin is the only drug for which cannabis has published, patient-level evidence of an interaction: seven case reports, INR readings as high as 11.55, and weekly warfarin doses cut by 22 to 31 percent to compensate. The cannabinoid with the cleanest trail through that file is CBD, the one sold at the pharmacy checkout as the safe half of the plant. In the only two cases where anyone knew the cannabinoid dose, both patients were taking Epidiolex, and in both the INR climbed during titration and settled only after the warfarin came down.

THC has cases too, and they are messier: smoked marijuana, recreational use, an edible of unknown strength. So THC matters less here than the marketing fear suggests, CBD matters more than the wellness marketing admits, and the practical point for anyone on warfarin is that the product label is no help at all. The gummy says 20mg. The Epidiolex patient whose INR hit 6.86 was taking 1,056mg a day. The real question is dose, the real answer is an INR check, and only the clinic that manages the warfarin can order one.

For apixaban, rivaroxaban and clopidogrel there is pharmacology and no case file, and this page says so rather than dressing the inference up as evidence.

Can I take CBD gummies with warfarin?

Not without the clinic that manages your INR knowing about it. Two published cases document a rising INR when pharmaceutical CBD was added to stable warfarin, with the warfarin dose cut by roughly 30 and 20 percent to compensate. Both patients were taking hundreds of milligrams a day. Nobody has measured a gummy-sized dose against warfarin.

The first case came out of the University of Alabama at Birmingham's compassionate-use CBD program. Grayson and colleagues described a 44-year-old man with a mechanical mitral valve, post-stroke epilepsy and a warfarin dose of 7.5mg a day that had held his INR between 2.0 and 2.6 for at least six months. Epidiolex started at 5 mg/kg a day and rose by 5 mg/kg every two weeks. Two weeks in, at 530mg a day, his INR was 2.96. Two weeks after that, at 1,056mg a day, it was 6.86. Over the following months his warfarin came down to 5.36mg, about 30 percent less, and the INR settled between 2 and 3 with no bleeding.

The second came from the same hospital's specialty pharmacy: a 46-year-old on warfarin for a deep vein thrombosis who started Epidiolex for intractable epilepsy with his INR watched through titration, and who needed a cut of nearly 20 percent to stay in range. That is the whole of the measured-dose evidence for CBD and warfarin, and it is more than most cannabis interactions have.

In a patient.info thread that opened in November 2017 with a British warfarin patient asking about CBD oil for anxiety, a retired ER nurse on lifelong warfarin answered that CBD blocks the enzyme that clears warfarin and that a high INR means a real chance of internal bleeding. Another poster, nine years on warfarin, wrote that her primary care doctor had prescribed CBD oil for her arthritis without knowing it interacted, while her hematologist handled the clots, and that the two sometimes forget the other exists. Nobody in the thread had a dose in milligrams, which is the detail this page exists to add.

How much does CBD raise INR?

In the Epidiolex cases, one patient's INR went from 2.22 to 6.86 over four weeks of titration and the other needed a nearly 20 percent dose cut to stay in range. Across all seven published warfarin cases, the 2023 systematic review found INR rises of 0.4 to 9.61 points and weekly warfarin cuts of 22 to 31 percent.

Smythe, Wu and Garwood searched six databases through May 2023 and found seven case reports of a warfarin interaction. Cannabis doses were either extremely high, more than 260mg of THC or more than 600mg of CBD a day, or unknown. One of the seven patients bled. None clotted. The authors graded the set as very low-quality evidence and asked for a formal interaction study, which as of today nobody has run.

The table sets out every drug class on this page, the cannabinoid involved, the pathway the sources name, how strong the evidence is, and what was observed in a patient.

Does THC raise INR too?

Yes, in case reports, with far less precision. Smoked marijuana was blamed for INRs of 10.41 and 11.55 with a gastrointestinal bleed, recreational cannabis for an INR of 4.6, a month of edibles plus smoking for 7.2, and a dispensary patient's 7.35mg THC increase for 5.2. In most of them nobody knew the THC dose.

The earliest and ugliest case is Yamreudeewong's from 2009: a 56-year-old man eleven years into warfarin for a mechanical valve, admitted with an upper GI bleed and an INR of 10.41, then readmitted two weeks after discharge with a nosebleed that would not stop and an INR of 11.55. He had been smoking more marijuana because he was depressed. Damkier's group in Denmark added a 27-year-old with a mechanical valve at 4.6 after recreational use. Hsu and Painter reported the one edible case: a 35-year-old stable on 10mg of warfarin a day since 2010 whose INR jumped to 7.2 after a month of edible cannabis and smoking. Brown and colleagues described a 67-year-old chronic pain patient in Buffalo whose home meter read 5.2 after he raised his THC by 7.35mg.

Set against all that is one patient who did not move. Thomas and colleagues followed an 85-year-old VA patient on warfarin who started an oromucosal cannabis product at 0.3mg of THC and 5.3mg of CBD a day, with 0.625mg of each as needed, and his INR barely fluctuated from October 2018 to September 2019. The authors concluded that route and dose decide whether the interaction shows up. Five milligrams of CBD a day is about a quarter of one Wyld Strawberry gummy.

A 2025 Ontario cohort counted actual bleeds, which no case report can do. Bérété and colleagues (Age and Ageing) followed patients 66 and over with an authorised cannabis prescription: among 378 also on warfarin, the risk of bleeding against 1,646 controls was 1.19 (95% CI 0.71 to 1.98). That interval includes no effect and also allows a near doubling, and the authors had no data on which cannabinoids anyone took or by what route. Its clearer signal sat elsewhere: across narrow-margin drugs, the risk of drug-related intoxication was 2.61 (95% CI 1.42 to 4.79). Our older adults guide has that study in context.

Which liver enzyme is behind the warfarin interaction?

Nobody has proven one. The case authors blame CYP2C9, the enzyme that clears S-warfarin, the potent half of the drug. The Epidiolex label's own CYP2C9 test found no change in tolbutamide levels at 7.5 mg/kg twice daily. What the label does document is moderate CYP2C19 inhibition, and the warfarin label lists CYP2C19 among the enzymes that clear warfarin.

The CPIC guideline puts S-warfarin at three to five times the potency of R-warfarin and says it is cleared predominantly by CYP2C9. The current warfarin label (Upsher-Smith's Jantoven, effective October 28, 2026) lists CYP2C9, 2C19, 2C8, 2C18, 1A2 and 3A4 as the enzymes involved and calls CYP2C9 the principal one. Its dosing table shows what a weak CYP2C9 looks like: 5 to 7mg a day for a patient with two normal copies of the gene and a normal VKORC1, 0.5 to 2mg for a patient with two CYP2C9*3 copies, and CPIC puts the *2 and *3 alleles at roughly 30 to 40 and 80 to 90 percent less S-warfarin metabolism. A drug that behaved like one of those alleles would push a patient a bracket down that table.

Here the label and the case reports part ways. Jazz ran the standard CYP2C9 probe, a single 500mg dose of tolbutamide, alongside Epidiolex at 7.5 mg/kg twice daily, and tolbutamide exposure did not change. Grayson's patient hit 6.86 at 5 mg/kg twice daily, below the probe dose. The label is specific about what CBD does. It is a moderate inhibitor of CYP2C19, a weak inhibitor of CYP1A2 and an inhibitor of UGT1A9; it left midazolam, the CYP3A4 test drug, unchanged at 750mg twice daily; CYP2D6 does not appear on it. CYP2C19 and CYP1A2 both sit on the warfarin label's enzyme list, mostly on the R-warfarin side. The interaction is documented in patients and the mechanism is inferred.

Can I take edibles with Eliquis or Xarelto?

No case report exists for either. A 2022 review found no published interaction between CBD and any direct oral anticoagulant, and the 2023 systematic review called the evidence outside warfarin non-existent. Both labels warn only about drugs that block P-gp and strongly inhibit CYP3A4 together, and the Epidiolex label shows CBD doing the first and not the second.

The Eliquis label says apixaban is a substrate of both CYP3A4 and P-glycoprotein, and tells prescribers to halve the dose, or avoid the combination at 2.5mg twice daily, when a patient takes a combined P-gp and strong CYP3A4 inhibitor. The Xarelto label says to avoid the same combined inhibitors outright. Neither gives guidance for a drug that affects P-gp while leaving CYP3A4 alone, and that is the profile the Epidiolex label describes: midazolam unchanged at 750mg twice daily, and a 2.5-fold rise in everolimus, a P-gp substrate, at 12.5 mg/kg twice daily, with a request for drug-level monitoring of other P-gp substrates taken by mouth. Apixaban and rivaroxaban are not named. Tacrolimus complicates that profile: a drug on the same two pathways roughly tripled in a 2025 trial at an average 826mg of CBD a day, by a route its authors could not pin down, and our immunosuppressants guide has the numbers. Nobody has run that test on apixaban or rivaroxaban.

Paduch and Thomason found no published article on CBD with any direct oral anticoagulant through May 2020, and Smythe's 2023 search found the same absence three years later. An apixaban user in the patient.info thread put it in four words: the jury is out. The P-gp lever exists at Epidiolex doses, it is untested at gummy doses, and nobody has published an apixaban or rivaroxaban patient whose levels moved. Surgical patients are the group most often put on one of these drugs with no warning, and our post-surgical pain page covers that window.

Does CBD interact with clopidogrel or aspirin?

Clopidogrel, on paper, yes, and in the opposite direction from warfarin. The Epidiolex label names clopidogrel: CBD moderately inhibits CYP2C19, the enzyme that activates clopidogrel, so the active metabolite falls and the label says to consider a higher clopidogrel dose. The Plavix label carries a boxed warning about weak CYP2C19 metabolizers. Aspirin has no enzyme handle and no case report.

Clopidogrel is a prodrug. The Plavix label's boxed warning says its effectiveness depends on conversion to an active metabolite, principally by CYP2C19, and section 5.1 tells prescribers to avoid omeprazole and esomeprazole because both inhibit that enzyme and cut the antiplatelet effect. The Epidiolex label makes the same connection by name: for CYP2C19 substrates such as clopidogrel whose efficacy comes from the active metabolite, concomitant Epidiolex is expected to lower the metabolite and reduce efficacy, and the label says to consider a dosage increase. Greger's 2020 review of anticoagulant and antiplatelet interactions reached the same conclusion. With warfarin the worry is too much drug. With clopidogrel it is too little, and too little clopidogrel does not announce itself with a nosebleed.

The only patient on record is a 76-year-old in a geriatric pharmacy case series, on clopidogrel and aspirin after a stroke, who had been using CBD oil three times a day for his knee. The pharmacists advised him off it and gave him acetaminophen. Nobody measured his platelet function, so the case documents a decision rather than a measured interaction. The label finding behind it comes from Epidiolex dosing, which starts at 2.5 mg/kg twice daily, 350mg a day for a 70kg adult; at 10 or 20mg of CBD a day the clopidogrel question has never been studied.

Aspirin is simpler and emptier. It does not depend on a CYP enzyme to work or to clear, so a cannabinoid has no pharmacokinetic handle on it, and a PubMed search on November 7, 2026 for cannabis, aspirin and platelet interaction returned nothing. Whether an edible adds to the bleeding risk aspirin already carries has not been measured in anyone.

How many gummies would it take to reach the doses in the case reports?

Dozens a day. The Grayson patient's INR first moved at 530mg of CBD a day and reached 6.86 at 1,056mg. Wyld's Strawberry 20:1, the most CBD-heavy gummy in our rankings, carries 20mg per piece, so the first threshold is 27 gummies and the second is 53, about $85 of gummies a day at $16 per ten.

On November 7, 2026, Purple Star MD in San Francisco listed Wyld Strawberry 20:1 on Leafly at $16 for ten gummies, 20mg of CBD and 1mg of THC each by Wyld's own spec. The same menu had Wyld Pomegranate 1:1 at $16, 10mg of THC and 10mg of CBD per gummy, and Camino's Sours Orchard Peach 1:1 at $19, also 10mg of each per piece. A whole container of the Strawberry, 200mg of CBD in one sitting, lands under the first dose step in the Grayson protocol. Reaching 530mg through the Pomegranate would deliver 530mg of THC alongside it, which would end the experiment for other reasons.

There is a certain comedy in a market that prints a blood-thinner warning on a tin whose entire contents fall short of the lowest dose anyone has documented an interaction at. The warning is still correct, for a reason the tin does not explain: no study has looked between 20mg and 530mg. The margin where most gummy users live is the margin where the data stops, and the Thomas patient at 5mg a day is the only published reading inside it.

CBD oil by the dropper is the exception: a few hundred milligrams a day, which the hemp market encourages for pain and sleep, sits inside the range where Smythe's review found the warfarin cases cluster.

What the anticoagulation clinic can use

Vague disclosure gives a warfarin nurse nothing to adjust. The clinic needs the product, the milligrams of CBD and THC a day, the date it started or changed, and how the INR is being checked. With those four details the case reports turn into a monitoring plan instead of a warning.

1. The product and the CBD in milligrams per day. Twenty milligrams and 500mg are different conversations.
2. The THC in milligrams per day, since the THC cases exist even where their doses do not.
3. The date it started or last changed. The Grayson INR moved within two weeks of each dose step.
4. How the INR is checked now. One patient.info poster runs a home meter every three or four days, a cadence that would have caught every rise in the case file early.

How to bring cannabis up with a prescriber is covered in our guide to using edibles for medical purposes, and the same CYP2C19 finding comes back, with a sedation problem attached, in our guide to edibles and antidepressants.