Potential Interactions with Warfarin Treatment

We have been following Agatha, an 85 year-old woman with atrial fibrillation who had elected to receive warfarin treatment. After 1 week of “traditional dosing” (5 mg/day with no loading dose), she had an international normalized ratio (INR) of 1.8 (goal 2.0 to 3.0) with no adverse effects. Acknowledging that subtherapeutic warfarin dosing provides little, if any, risk reduction, her total weekly dose (TWD) was increased by approximately 15% by administering 7.5 mg on Mondays and Thursdays and 5 mg on all other days (TWD of 40 mg). One week later, her INR is now 2.4 (goal range 2.0 to 3.0), and therefore her warfarin dose will be maintained at this level.

New Presentation

One month later, Agatha presents to the clinic with signs and symptoms suggestive of a lower urinary tract infection; urinalysis shows numerous white blood cells and positive nitrite. The physician would like to prescribe sulfamethoxazole-trimethoprim (SMX-TMP) 800 mg/160 mg, 1 tablet by mouth twice daily for 3 days. The physician is concerned about the potential of SMX-TMP to raise the INR. Should Agatha’s warfarin dose be adjusted? If so, how much should the dose be adjusted and when should the INR be monitored again?

Potential Interaction to Consider

The list of medications, foods, and herbal substances with the potential to interact with warfarin—either enhancing or diminishing its anticoagulant effect—is exhaustive and continues to expand. The majority of drug interactions lead to elevated INR values, which increase the risk for bleeding; interacting drugs that lower INR values potentially increase the risk for thromboembolic events. Pathways by which these interactions occur most commonly involve changes in warfarin metabolism, displacement of warfarin from its protein-bound state, or interruption of vitamin K synthesis or absorption.¹ 

Unfortunately, even amongst drugs with known potential interactions, whether an individual patient will experience the interaction, the degree of intensity of the interaction, and the exact timing of the interaction (when will the impact upon the INR begin and cease) is quite unpredictable. 

Fortunately, most of the actual interactions happen in only a minority of circumstances and to a sufficiently modest degree that coagulation status is not meaningfully altered. Because the impact can sometimes be dramatic, it is wise to be vigilant for interactions and increase the frequency of INR monitoring.

Warfarin is a racemic mixture of R- and S-warfarin isomers, each of which has independent anticoagulant activity (Table). As both isomers are metabolized by different hepatic cytochrome P450 (CYP) enzymes, a potential interacting medication may predominantly impact the metabolism of only 1 isomer. For example, a medication, food, or herbal inhibiting CYP 3A4 (eg, macrolide antibiotics, grapefruit juice) is most likely to increase the levels of only R-warfarin; as this isomer is significantly less potent than the S-isomer, increases in R-warfarin levels are unlikely to lead to a significant deviation in the INR. On the contrary, medications selectively impacting the CYP 2C9 mediated metabolism of S-warfarin (eg, amiodarone, fluconazole) are much more likely to lead to INR deviations, often requiring warfarin dose adjustments. 

Perhaps, the most common group of medications associated with warfarin drug interactions is antibiotics. Most antibiotics do not lead to significant deviations in the INR; indeed, the presence of infection can lead to increased INR values alone, regardless of whether or not antibiotics are prescribed.¹ Antibiotics commonly prescribed for upper respiratory infections, sinusitis, and otitis (eg, azithromycin, cephalosporins, and beta-lactam antibiotics), despite the potential for interaction, do not commonly manifest a meaningful impact upon the INR. 

On the other hand, metronidazole and SMX-TMP are fairly often associated with INR deviations when used concomitantly with warfarin; because these antibiotics inhibit the metabolism of the more potent warfarin isomer, S-warfarin (via CYP 2C9), they can be predicted to more often result in meaningful INR alterations. Due to the relative potency of the S-isomer, small increases in levels can lead to significant increases in the INR. 

Indeed it is not uncommon for practitioners to empirically lower the dose of warfarin by 35% to 50% when these antibiotics (ie, metronidazole, SMX-TMP) are initiated, although there is insufficient evidence to determine whether this strategy limits the risk for adverse outcomes. Given the large number of available antibiotics, consideration should be given to utilize agents with a lower chance of interacting with warfarin. However, avoidance of such medications may not always be possible in certain scenarios. Product labeling for antibiotics lacks specific advice about warfarin monitoring or dose adjustment, hence clinicians must derive their own preferred monitoring methods, some of which will be discussed below.

Drug-Warfarin Interactions 

Because the frequency of meaningful INR perturbations with most antibiotics is modest, even with metronidazole and SMX-TMP, in most cases we prefer to increase the frequency of INR monitoring during antibiotic administration rather than changing the dose of warfarin empirically. Changes in INR typically do not manifest for at least 48 hours after the first dose of an antibiotic, so we advise monitoring on days 3 and 5 to 7 if possible, to detect any initial INR impact (day 3) and maximum INR impact (day 5-7). A supratherapeutic INR on either of these days likely merits warfarin dose adjustment. Antibiotic impact upon INR is generally lost as early as 3 days and most often within 5 days of discontinuation of the antibiotic, so unless marked deviation of INR exists beyond this point, further INR monitoring to survey antibiotic impact in unnecessary beyond 5 days post-discontinuation.

Consideration of the need for warfarin modulation depends upon the duration of intended antibiotic treatment (or any potentially interacting medication). Short-term antibiotics (5 days or less) are probably not sufficiently long to merit INR monitoring because by the time initial impact of antibiotic treatment is discernible (day 3) and maximum antibiotic effect has occurred (day 5), the interacting drug will already be discontinued and the interaction will automatically disappear. 

On the other hand, antibiotic therapy of 1 week or longer provides an opportunity for meaningful monitoring to detect interaction. During long-term antibiotics, 1 of 3 scenarios will commonly emerge. We will suggest using our above-mentioned days 3 and days 5 to 7 “benchmark” to sort out preferred management in a situation where an antibiotic will be continued for at least more than 5 days.

Day 3: Meaningfully Supratherapeutic INR 

If supratherapeutic INR is ≥0.5 above intended therapeutic range, it is likely that warfarin dosing will need to be reduced as the INR has already begun to rise. Omit the next warfarin dose (today, if possible) and reduce TWD by halving the dose on 1 additional day.

Example: A patient stable on 5 mg/d began a 14-day course of SMX-TMP for cellulitis suspicious for MRSA on Monday. (Note: We acknowledge the availability of alternative antibiotic options and in those receiving warfarin, encourage investigating alternative options. For the purposes of this example, we will assume this represents the most appropriate treatment option). The INR was checked on Thursday (day 3) and noted at 3.7. Since the patient takes their warfarin in the evening (5 mg/d) omit today’s dose (a 14% dose reduction of 5 mg out of the weekly 35 mg) and also take half the usual 5 mg dose tomorrow (Friday)—resulting in a TWD reduction of 21% (7.5 mg/35 mg). The patient should return within 2 to 4 days (days 5-7 of interacting medication) to see both the maximum effects of the interacting medication and gauge the early results of the warfarin dose adjustment. On Monday (1 week after initiating antibiotics or day 7), if the INR is therapeutic, we will continue same TWD of  27.5 mg. 

As the interacting medication has been in place for 7 days, it is likely at steady state exerting its full effects; therefore, it is unlikely that the INR will precipitously increase or decrease after this point. Moving forward the warfarin dose can be adjusted as needed (10% to 20%) in order to raise or lower the INR in response to a subtherapuetic or supratherapuetic INR, respectively, as previously discussed.

Day 3: Therapeutic or Minimally Supratherapetuic INR 

If INR is therapeutic or supratherapeutic but ≤0.5 above goal, the CHEST guidelines specifically mention that during long-term warfarin monitoring INR deviations of  ≤0.5 above goal need not require warfarin dose adjustment in those on previously stable warfarin regimens. Clinical trials addressing this issue have found that well-intended, fine-tuning of warfarin dosing that attempts to bring down such modest INR elevations are more likely to lead to a subtherapeutic level than be beneficial. 

For a patient who is beginning a regimen that might induce meaningful INR alteration, an “early warning” sign of that impact can be discerned by measuring the INR on day 3 after initiation. In this circumstance, even an INR of 0.1 to 0.5 above goal may be a harbinger for progressively more intense deviations. Hence, a patient with even such “minor” INR elevations early on merits closer scrutiny. We will determine whether INR perturbations from such interaction require further intervention by also checking the INR on days 5 to 7.

Outcome of Our Case

On days 5-7, if the INR is not further elevated beyond what was noted on day 3, one can anticipate that further meaningful interaction is not likely. If on day 5 the INR is ≥3.5 (0.5 points above goal) it should be treated as any other elevated INR, noting that we are at this point seeing the maximum (or near maximum) interaction effect of the interacting drug. 

For Agatha, although an antibiotic-warfarin interaction is possible, the brief duration of antibiotic treatment (3 days) would suggest that we do not need to perform any intensification of monitoring. If there is an impact on INR, it would be brief and the antibiotic will have been discontinued by the time we would note any impact, thereby alleviating any impact of the interaction. Were Agatha to have received treatment for a longer period (eg, 10 days or longer), intensified monitoring would have been appropriate.

Amidodorane and Fluconazole

The abovementioned scenarios are the most common and potentially meaningful interactions when coadministering agents with antibiotics. On the other hand, even though some drug-warfarin interactions are not as commonly encountered, the predictably higher likelihood of problematic INR elevations with some drugs (eg, amidodarone, fluconazole) is sufficiently great that it is worth addressing some specific examples. 

Both amiodarone and fluconazole inhibit the metabolism of S-warfarin (the more active isomer) via inhibition of CYP 2C9. Both are also considered “strong” 2C9 inhibitors. If amiodarone is started, the package insert recommends an empiric 35% to 50% reduction in the dose of warfarin since elevations in the INR are almost a certainty. Frequent INR monitoring (ie, weekly to biweekly until stable) will be required to ensure a therapeutic INR is maintained. 

Considerations for increasing the frequency of INR monitoring or dose modulation also depend on duration of drug administration. As amiodarone is generally chronically administered, warfarin dose adjustment will almost always be required. However, as fluconazole is often given in single dose (or several single separated doses), the limited duration of treatment may not necessitate empiric warfarin dose adjustment. Hence, although amiodarone and fluconazole share mechanistic similarities, the short-term dosing of fluconazole obviates the need for alteration of monitoring or dose modulation.

What’s The “Take Home”?

Numerous medications, foods, and herbal substances can interact with warfarin. Depending on the mechanism of interaction and which warfarin isomer is affected determines the likelihood of the interaction being clinically relevant. Drug interactions that affect S-warfarin are most clinically relevant.

Most interacting medications that can lead to INR deviations will evidence this effect within 3 to 5 days of initiation. Most often, the interaction dissipates within 3 to 5 days following drug discontinuation.

The INR can be assessed at days 3 to 5 to determine if an interaction has occurred and adjust the dose of warfarin as necessary.

If a potentially interacting medication is intended to be used for ≥5 days, even if an INR impact were to occur, its duration would likely to be so brief that manipulation of warfarin dosing is as likely to cause misadventure as it is to be beneficial. Hence, unless INR excursions ≥0.5 above INR goal, no warfarin dose adjustment need be instituted offset of interaction. ■

Eric A. Dietrich, PharmD, BCPS, graduated from UF College of Pharmacy in 2011 and completed a 2-year fellowship in family medicine where he was in charge of a coumadin clinic. He now works for the UF Colleges of Pharmacy and Medicine. 

Louis Kurtizky, MD, is a family physician affiliated with the University of Florida Family Medicine Residency Program, where he commonly co-manages warfarin cases with his colleagues.

ReferenceS:

1.  Clark NP, Delate T, Riggs CS, et al. Warfarin interactions with antibiotics in the ambulatory care setting. JAMA Intern Med. 2014;174:409-416.

2.Antithrombotic Therapy and Prevention of Thrombosis, 9th ed: CHEST Evidence-Based Clinical Practice Guidelines. 2012;141(2 suppl).