Warfarin carries one of the longest and most clinically consequential drug interaction lists of any medicine in routine use. It is estimated that drug interactions contribute to more than 20% of warfarin-related adverse events requiring hospital admission [1]. Yet warfarin is not a drug that can simply be swapped out: it remains the anticoagulant of choice for mechanical heart valves, rheumatic mitral stenosis, and antiphospholipid syndrome — populations in whom the interaction risk cannot be avoided, only managed.
This guide provides a complete, practitioner-level account of warfarin drug interactions: the pharmacology that underlies them, the individual drugs and drug classes that matter most, the role of dietary vitamin K, and a practical framework for safe co-prescribing.
Why Warfarin Interacts with So Many Drugs
Understanding the pharmacology is not academic — it directly predicts which interactions will be severe and how quickly they will develop.
The racemic mixture and CYP metabolism
Warfarin is dispensed as a racemic mixture of two enantiomers. S-warfarin is approximately four times more pharmacologically active than R-warfarin and is principally metabolised by CYP2C9. R-warfarin is metabolised primarily by CYP1A2 and CYP3A4 [3]. Any drug that inhibits or induces CYP2C9 therefore has a disproportionate effect on the active component of the mixture.
Mechanism of action and the vitamin K connection
Warfarin inhibits vitamin K epoxide reductase (VKOR), preventing the recycling of vitamin K. This reduces the activation of clotting factors II, VII, IX, and X, as well as the anticoagulant proteins C and S. The narrow therapeutic window — typically a target INR of 2.0–3.0 for most indications, or 2.5–3.5 for mechanical mitral valves — means even a 20% change in warfarin bioavailability or sensitivity can push the INR into dangerous territory [5].
Genetic variability and the interaction baseline
Genetic polymorphisms in CYP2C9 (affecting metabolism) and VKORC1 (affecting target sensitivity) explain much of the inter-individual variation in warfarin dose requirements [3]. A patient who is a CYP2C9 poor metaboliser may already be on a very low dose; adding a CYP2C9 inhibitor in this setting can cause severe anticoagulation with only a modest dose. Pharmacogenomic testing is not yet routine in most UK practice, but awareness of this variability is important when interactions behave unpredictably.
Pharmacokinetic Interactions: CYP Enzyme Effects
CYP2C9 Inhibitors — drugs that increase INR
These drugs reduce the metabolism of S-warfarin, raising plasma concentrations and pushing the INR upward. The magnitude and speed of onset depend on the potency of inhibition and the inhibitor's own pharmacokinetics.
Fluconazole is one of the most potent CYP2C9 inhibitors in clinical use [6]. A standard five-day course of oral fluconazole for vaginal candidiasis can double or triple the INR within 48–72 hours. This is a common scenario on general wards — a patient on warfarin receives a fluconazole prescription and is discharged without an INR review plan. INR should be checked two to three days after starting fluconazole and again two to three days after completing the course. If fluconazole is essential, reduce the warfarin dose preemptively by 25–50% and monitor closely. Topical or intravaginal preparations carry negligible systemic absorption and are generally safe. Amiodarone inhibits both CYP2C9 and CYP1A2, affecting metabolism of both warfarin enantiomers [6]. What makes the amiodarone interaction distinctively dangerous is its time course: amiodarone has a half-life of 40–55 days, meaning the inhibitory effect develops slowly over weeks to months and persists for an equally prolonged period after the drug is stopped. Many patients are loaded with amiodarone in hospital for new atrial fibrillation, then discharged on their usual warfarin dose — and the INR rises insidiously over the subsequent weeks. The interaction typically requires a 30–50% warfarin dose reduction, but titration must be guided by frequent INR checks over several months. The same vigilance applies when amiodarone is eventually withdrawn. Metronidazole (and its prodrug ornidazole) is a potent CYP2C9 inhibitor that commonly affects warfarin in real-world practice [1]. It is frequently prescribed for dental infections, Helicobacter pylori eradication, Clostridioides difficile, and surgical prophylaxis — all scenarios where the prescriber may not immediately consider the anticoagulation implications. An INR check three to four days into metronidazole treatment and after completing the course is mandatory. Some guidelines recommend an empirical 25% warfarin dose reduction for the duration of the course. Co-trimoxazole (trimethoprim-sulfamethoxazole) potentiates warfarin through two independent mechanisms: CYP2C9 inhibition by the sulfamethoxazole component, and stereoselective displacement of S-warfarin from plasma albumin binding sites [1]. This dual mechanism makes the co-trimoxazole–warfarin interaction particularly reliable — it almost always causes a clinically significant INR rise. A 25–50% warfarin dose reduction during the course is commonly required. Trimethoprim alone also inhibits CYP2C9 and carries the same warning. Fluoxetine and fluvoxamine are CYP2C9 inhibitors within the SSRI class [6]. The interaction is important because depression and anxiety are common comorbidities in patients with chronic atrial fibrillation and other conditions requiring long-term anticoagulation. Sertraline and citalopram have substantially weaker CYP2C9 inhibitory effects, though they increase bleeding risk through antiplatelet mechanisms (see pharmacodynamic interactions below). When initiating an SSRI in a patient on warfarin, sertraline or citalopram are generally preferable to fluoxetine or fluvoxamine. Ciprofloxacin inhibits CYP1A2, affecting R-warfarin metabolism and producing a moderate but clinically meaningful INR rise in most patients [6]. This interaction is often underappreciated because ciprofloxacin does not affect CYP2C9 directly — prescribers focused on the S-warfarin pathway may overlook it. INR should be checked within three to five days of starting any quinolone in a patient on warfarin. Clarithromycin and erythromycin inhibit CYP3A4, affecting R-warfarin and producing a moderate interaction [6]. Azithromycin has minimal CYP inhibitory effect and is generally considered the safer macrolide choice when treating a patient on warfarin, though antibiotic-mediated reduction in gut flora synthesis of vitamin K (see below) means some INR monitoring is still prudent.CYP2C9 Inducers — drugs that decrease INR
Enzyme inducers accelerate warfarin metabolism, reducing plasma concentrations and the anticoagulant effect. Under-anticoagulation is equally dangerous — thrombosis in a patient with a mechanical valve can be rapidly fatal.
Rifampicin is the most potent CYP enzyme inducer in clinical use, capable of reducing warfarin plasma concentrations by up to 90% [7]. The interaction develops within days of starting rifampicin and reverses with equal speed on stopping. Patients on warfarin who require rifampicin — typically for tuberculosis or staphylococcal prosthetic valve endocarditis — may need warfarin dose increases of 200–400% to maintain therapeutic INR. This requires daily INR monitoring at initiation and again on cessation. In practice, INR-guided dose adjustments must be systematic, because the interaction is so potent that fixed-dose rules are unreliable. When rifampicin is unavoidable, daily or alternate-day INR checks during titration are not excessive. Carbamazepine and phenytoin are hepatic enzyme inducers that substantially increase warfarin clearance [6]. Patients with epilepsy and concurrent anticoagulation needs (e.g. atrial fibrillation following stroke) represent a challenging clinical group. Phenytoin has the additional complication of displacing warfarin from protein binding sites initially, causing a transient INR rise before the inductive effect takes hold — so the interaction can be biphasic. Levetiracetam and lamotrigine do not induce CYP enzymes and are preferable anticonvulsants when warfarin co-prescribing is required. St John's Wort (Hypericum perforatum) is a potent inducer of CYP3A4 and P-glycoprotein, reducing warfarin bioavailability by a clinically significant margin [7]. The MHRA has issued specific guidance warning against co-use. Critically, St John's Wort is an over-the-counter herbal supplement, and patients do not routinely volunteer its use. All patients starting warfarin must receive explicit written advice to avoid St John's Wort and to disclose all herbal and over-the-counter preparations at every consultation. Chronic alcohol use induces CYP2E1 and increases warfarin clearance, lowering INR in regular heavy drinkers. Acute alcohol ingestion, by contrast, inhibits CYP2C9 and transiently raises INR. This explains why patients who drink episodically have highly variable INR readings that are difficult to interpret outside the clinical context.Pharmacodynamic Interactions
Pharmacodynamic interactions change the anticoagulant effect without altering warfarin pharmacokinetics. They are mediated through additive or antagonistic effects on haemostasis.
Aspirin and NSAIDs pose substantial bleeding risk through two mechanisms: irreversible inhibition of platelet function (aspirin) or reversible COX-1 inhibition (NSAIDs), combined with GI mucosal damage and prostaglandin depletion [2]. The combination of therapeutic anticoagulation with aspirin or NSAIDs markedly increases the risk of major GI haemorrhage. Low-dose aspirin 75 mg is appropriate alongside warfarin only in specific high-risk cardiovascular settings with an accepted increased bleeding risk — not as incidental analgesia. Ibuprofen and diclofenac should be avoided. Paracetamol is the analgesic of choice in patients on warfarin, though high doses (more than 2 g per day regularly) can mildly potentiate anticoagulation, likely through effects on vitamin K metabolism. SSRIs and SNRIs inhibit platelet serotonin uptake, reducing platelet aggregation and increasing bleeding risk independently of any effect on warfarin pharmacokinetics [6]. This pharmacodynamic interaction adds to any pharmacokinetic effect from CYP2C9 inhibition. The combination of warfarin and an SSRI requires an INR check after initiation and clinician awareness of increased upper GI bleeding risk; proton pump inhibitor gastroprotection should be considered. Broad-spectrum antibiotics reduce the gut flora's contribution to endogenous vitamin K synthesis (menaquinones). The magnitude of this effect varies widely between individuals and antibiotic regimens, but in patients with already borderline dietary vitamin K intake, even standard antibiotic courses can push the INR upward [1]. This is an often-overlooked mechanism: it is not CYP-mediated but dietary, acting at the target rather than during drug metabolism. An INR check midway through or shortly after any prolonged antibiotic course is reasonable practice. Heparin, DOACs, and antiplatelet drugs carry obviously additive anticoagulant or antiplatelet effects. These combinations are occasionally intentional (e.g. bridging therapy) but must be time-limited and carefully monitored. DOACs and warfarin should not be co-prescribed without specialist guidance. Thyroid status deserves separate mention as a pharmacodynamic modifier. Hyperthyroidism increases the catabolism of vitamin K-dependent clotting factors, potentiating warfarin. Hypothyroidism has the opposite effect. Any significant change in thyroid function — including initiation or dose adjustment of levothyroxine — requires INR review, as it alters the patient's intrinsic sensitivity to warfarin independently of any drug interaction.Dietary Vitamin K: The Non-Drug Interaction
Vitamin K directly competes with warfarin's mechanism of action. Foods rich in vitamin K — predominantly dark green leafy vegetables such as kale, spinach, broccoli, Brussels sprouts, and spring greens — can significantly antagonise anticoagulation [5].
The practical message for patients is consistency, not avoidance. Instructing a patient to eat no green vegetables is nutritionally counterproductive and leads to binge-eating patterns that create dramatic INR swings. A patient who eats a consistent, moderate amount of vitamin K-containing foods week to week is far more predictable to manage than one whose intake fluctuates with the seasons or social occasions.
Cranberry juice merits specific mention: multiple case reports and MHRA communications have linked cranberry juice consumption with significant INR elevation, likely through CYP2C9 inhibition by flavonoid constituents [7]. Patients on warfarin should be advised to avoid regular large quantities of cranberry juice or cranberry supplements.
Grapefruit juice inhibits CYP3A4 in the gut wall and can increase bioavailability of several drugs, though its interaction with warfarin is relatively modest compared to its effect on statins or ciclosporin.
INR Monitoring Strategy When Interactions Are Unavoidable
When a potentially interacting drug cannot be avoided, a structured monitoring approach converts an uncontrolled risk into a manageable one:
Before prescribing: Check a validated drug interaction resource (the BNF Appendix 1, MedNext Formulary's interaction checker, or Stockley's Drug Interactions). Identify the direction and likely magnitude of the expected INR change. Document this assessment. At initiation: Check INR at baseline, then again at two to five days depending on the potency of the interaction. For high-risk interactions (fluconazole, metronidazole, rifampicin), check at 48–72 hours. During treatment: Frequency depends on the drug and duration. A five-day antibiotic course requires fewer checks than amiodarone loading. As a practical rule, any drug course lasting more than seven days in a patient on warfarin warrants at least two INR checks during it. At completion: INR rebound — or re-bounce — is equally important. When an inhibitor is stopped, the INR may fall as the inhibitory effect dissipates; when an inducer is stopped, INR may rise. Check INR within five to seven days of completing an interacting drug course, and again after a further week if the INR is not back in range. Dose adjustment: For most pharmacokinetic inhibitors (excluding amiodarone), empirical dose reductions of 25–50% at initiation are reasonable while awaiting INR results. Dosing can be titrated back up as the INR normalises. Documenting the reasoning for dose adjustments is essential for continuity. Communication: Ensure the patient's anticoagulation clinic or GP is informed whenever an interacting drug is prescribed, especially during out-of-hours prescribing, emergency admissions, or when patients see multiple clinicians.A Practical Management Framework
The following framework provides a simple, reproducible approach to co-prescribing in warfarin patients:
Tier 1 — Check always: Use the BNF or MedNext Formulary interaction checker before prescribing any new drug to a patient on warfarin. This takes under 30 seconds and is non-negotiable. Tier 2 — Alert the team: If an interacting drug is being prescribed, flag this to the patient, the ward pharmacist, and the anticoagulation clinic or GP. Do not assume the anticoagulation team will know. Tier 3 — Monitor intensively for high-risk combinations: Fluconazole, amiodarone, metronidazole, co-trimoxazole, rifampicin — these require INR checks within 48–72 hours of initiation and specific dose adjustment plans. Tier 4 — Educate the patient: Patients on long-term warfarin who understand their drug's interaction risks are their own best safety net. Advise on St John's Wort, herbal supplements, cranberry juice, and the importance of disclosing all new prescriptions and over-the-counter purchases. Tier 5 — Consider alternatives: Where the choice is clinician-driven (e.g. choice of SSRI, choice of antifungal, choice of antiepileptic), select the agent with the most favourable interaction profile. Fluconazole can often be substituted with topical antifungals. Fluoxetine can often be substituted with sertraline. Levetiracetam can substitute for carbamazepine in many epilepsy settings.When to Consider Switching to a DOAC
Although this guide focuses on warfarin management, it is appropriate to consider whether a patient would be better served by switching to a direct oral anticoagulant (DOAC). DOACs have fewer drug interactions, no dietary interaction with vitamin K, and do not require INR monitoring. They are now preferred for non-valvular atrial fibrillation in most UK guidelines [5].
Warfarin remains essential for:
- Mechanical prosthetic heart valves
- Mitral stenosis of rheumatic origin
- Antiphospholipid syndrome (DOACs have inferior evidence in this indication)
- Patients with established stable INR control who prefer to remain on warfarin
For patients in these categories, mastery of warfarin drug interactions is not optional — it is core clinical competency.
References
- Holbrook AM, Pereira JA, Labiris R, et al. Systematic overview of warfarin and its drug and food interactions. Arch Intern Med. 2005;165(10):1095-1106.
- Ageno W, Gallus AS, Wittkowsky A, et al. Oral anticoagulant therapy: Antithrombotic Therapy and Prevention of Thrombosis, 9th ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines. Chest. 2012;141(2 Suppl):e44S-e88S.
- Limdi NA, Veenstra DL. Warfarin pharmacogenetics. Pharmacotherapy. 2008;28(9):1084-1097.
- Daly AK. Pharmacogenetics of the cytochromes P450. Curr Top Med Chem. 2004;4(16):1733-1744.
- National Institute for Health and Care Excellence. Anticoagulation - oral. NICE Clinical Knowledge Summary. nice.org.uk. 2023.
- Baxter K, Preston CL (eds). Stockley's Drug Interactions. 12th ed. Pharmaceutical Press; 2024.
- Medicines and Healthcare products Regulatory Agency. Warfarin: reminder of factors that can increase or decrease the anticoagulant effect. Drug Safety Update. 2019.