Abstract and Introduction
The literature suggesting a relevant interaction between acetaminophen and warfarin is inconsistent. Considering the ubiquitous use of acetaminophen, a review of the effects on international normalized ratio (INR) in patients taking warfarin was necessary. Thus, we performed a search of the PubMed (1966–November 2010) and International Pharmaceutical Abstracts (1970–November 2010) databases to review the available literature addressing an acetaminophen-warfarin interaction and its possible mechanisms. A sample of case reports, in addition to all English-language studies were evaluated, and relevant references were examined for additional articles. Reports of nonwarfarin coumarin anticoagulants were excluded. Published documentation reporting an interaction between acetaminophen and warfarin is limited. Small prospective studies of various designs and case studies describe aberrant INR results in patients using acetaminophen while receiving warfarin. These INR elevations typically involved acetaminophen ingestion of at least 2 g/day for several consecutive days. In several small prospective studies, INR results were elevated to a statistically significant extent that would require a change in warfarin dosing and monitoring in clinical practice. The mechanism for this interaction remains to be elucidated yet is suggested to occur through alterations in hepatic metabolism. The use of moderate-to-high doses of acetaminophen while receiving warfarin results in supratherapeutic INRs in some patients. The characteristics that may predispose a patient to this interaction are unclear, yet the widespread use of acetaminophen calls for enhanced clinician awareness and reinforcement of patient education about this interaction.
Acetaminophen is a widely used analgesic and antipyretic agent. Warfarin, an oral anticoagulant, is also frequently used for various indications. In patients receiving warfarin, acetaminophen is the preferred analgesic, as nonsteroidal antiinflammatory drugs interact with warfarin and increase a patient's bleeding risk. However, an interaction between acetaminophen and warfarin has been documented in the medical literature and is recognized in the United States prescribing information and United Kingdom summary of product characteristics.[2, 3] The potentiation of warfarin therapy by coingestion of acetaminophen is a recognized phenomenon; however, its clinical relevance to individual patients and the mechanism by which it occurs are unclear.
A 2004 study identified the frequency of warfarin use with potentially interacting drugs. Of the 134,833 patients prescribed long-term warfarin in the retrospective study, 109,998 (81.6%) were identified to have received at least one prescription drug with potential to interact with warfarin; of these, 22.7% (or 18.5% of the total number of patients studied) received a prescription for an acetaminophen-containing product. This study did not evaluate the concomitant use of over-the-counter agents, so it is likely that the use of acetaminophen was higher than what was uncovered by the study. A recent European study found that 16.1% of patients receiving warfarin also received an acetaminophen-containing product. Among the patients receiving this combination of drugs, the incidence of fatal bleeding events was 4.6 times higher than that in patients receiving only acetaminophen and 2.7 times higher than that in patients receiving only warfarin.
In 2009, over 23 million prescriptions for warfarin were filled in the United States. Given the large number of patients receiving warfarin therapy and the potential for adverse outcomes from the use of the two drugs together, the presence of an interaction between acetaminophen and warfarin has important implications for patient care.
Published Literature on an Acetaminophen-warfarin Interaction
To better understand the implications of an acetaminophen-warfarin interaction and its possible mechanisms, we conducted a literature search using the PubMed (1966–November 2010) and International Pharmaceutical Abstracts (1970–November 2010) databases. The search terms used were acetaminophen, paracetamol, APAP, and warfarin. A sample of case reports in addition to all English-language studies were analyzed, and relevant references were examined for additional articles. References that focused on anticoagulants other than warfarin, such as acenocoumarol or phenprocoumon, were excluded from this review.
Dating back to 1991, before prothrombin times were standardized into the international normalized ratio (INR), a case report identified a potential interaction between acetaminophen and warfarin. A 66-year-old woman was receiving maintenance warfarin therapy at an average dose of 3 mg/day for secondary prevention of stroke; her care was provided at an anticoagulation clinic. Her prothrombin time was maintained at 15–19 seconds, with a maximum of 23 seconds over 4 years of therapy. After consuming 48 tablets of an acetaminophen-codeine combination over 10 days for recurring back pain, the patient was admitted after 3 days of hematuria and gingival bleeding. Her prothrombin time was 96 seconds. The patient denied alcohol intake or liver disease. After receiving 3 units of fresh frozen plasma, her prothrombin time returned to normal.
In a second case report, a 63-year-old woman who was receiving warfarin 4–4.5 mg/day after aortic valve replacement experienced an elevated INR on two separate occasions. The patient's anticoagulation had been managed by a hospital anticoagulation clinic, and her INR had been stable for several months. Her first elevated INR value was 9.6, at which point warfarin was withheld. The next day, the patient's INR rose to 12.0 and contact bleeding from the gums and spontaneous bruising in the groin area were noted. Nine days later, her INR was 5.4, despite receiving no warfarin for the previous 10 days. The following day, her INR was 2.6 and warfarin was restarted at 3 mg/day. Before this elevation, the patient ingested acetaminophen 2 g/day for 7 days as part of a combination drug for pain that included dihydrocodeine, under the direction of her general practitioner. About 2 weeks after restarting warfarin therapy, she was again given a combination drug containing acetaminophen 1.5 g/day and codeine for 8 days. On the patient's last day of taking this combination drug, her INR was 8.5 and warfarin again was withheld. Six days later, her INR was within therapeutic range, and warfarin was restarted. No other potentially interacting drugs were identified, and no vitamin K was given at either occurrence. This patient experienced two clinically relevant episodes of supratherapeutic INR after acetaminophen use.
Another case report described a 62-year-old woman receiving an average warfarin dose of 4.7 mg/day after mitral valve replacement, with a stable INR for several months, who came to the emergency department with a retroperitoneal hematoma. Her INR at this time was 7.5, significantly higher than her previous INR of 2.5, 1 month earlier. After consultation with the patient, it was discovered that she had taken 8–10 500-mg tablets of acetaminophen over 4 days for knee pain secondary to a fall. All of her drug therapy, including warfarin, had been unchanged. The patient's INR decreased to 1.9 after administration of 2 units of fresh frozen plasma and 1 mg of intravenous vitamin K.
A 74-year-old man was the subject of another case report. The patient was receiving warfarin 5 mg/day for 4 years for the treatment of atrial fibrillation, with a mean INR of 2.7. His INR rose to 4.0 when he reportedly took acetaminophen 2 g/day for 3 days for pain relief. Warfarin was stopped for 1 day and then restarted at the same dose. One week later, during which time no acetaminophen use was reported, the INR decreased to 2.2. After this event, the patient consented to receiving acetaminophen 1 g 4 times/day for 3 days, during which time laboratory tests were conducted. The INR rose from the preacetaminophen value of 2.3 to 6.4 on the day after acetaminophen was discontinued. Liver function tests were found to be within the normal range. Activities of factor VII and factor X decreased from 29.4% to 15.5% and 27% to 20.2%, respectively, representing decreased clotting activity after acetaminophen was administered.
The last case report that we reviewed discussed an 81-year-old woman receiving warfarin 4–4.5 mg/day after placement of a prosthetic aortic valve. The patient's INR had remained stable at approximately 3.0 for the previous 6 months. The patient took acetaminophen 1 g every 6 hours for 3 days, then 1 g every 6 hours as needed for 4 more days for fever. The patient then came to the hospital for evaluation of possible endocarditis; her INR value was 16.39 and she experienced gastrointestinal bleeding, which was her first noted bleeding event after starting warfarin therapy. After vitamin K and fresh frozen plasma were administered, the patient's INR decreased to less than 2.0; acetaminophen was discontinued, and warfarin was restarted and titrated to a therapeutic INR. The patient denied the use of any other prescription, over-the-counter, or herbal agents.
A prospective case-control study involving 93 cases and 196 controls assessed the effect of various risk factors, such as diet, concurrent illness, alcohol intake, acetaminophen ingestion, or other drugs, on INR. Case patients had an INR value greater than 6.0, were required to have been taking warfarin for over 1 month, and were followed for approximately 11 months. Eligible control patients were required to have an INR value of 1.7–3.3 with no change in warfarin dose. Target INR was 2.0–3.0 for all cases and controls.
The results showed that case patients were more likely to have ingested acetaminophen compared with controls (56% vs 36%, p=0.001) and were more likely to have ingested larger doses of acetaminophen (6756 vs 2938 mg/wk, p<0.001). The authors commented that patients who ingested a minimum of four regular-strength (325-mg) acetaminophen tablets per day for over 1 week had a 10-fold increase in the odds of developing an INR greater than 6.0 compared with those who took no acetaminophen at all (p=0.001). The INR increase was found to be dose dependent. A suggested mechanism for this increase in INR is a pharmacokinetic interaction at the cytochrome P450 (CYP) level. Other variables that were also found to be risk factors for an increased INR included advanced malignancy, changes in vitamin K and alcohol consumption, and acute diarrheal illness. However, the authors stated that these independent variables had no significant 2-way interactions.
The authors did not indicate how warfarin dosing was managed in the patients. An editorial on this study suggested that patients receiving warfarin who must also receive acetaminophen should have INR measurements performed 1–2 times/week and that the INR should not exceed 4.0. It should be noted, however, that several subsequent letters to the editor cautioned the authors of the study against making statements of causality based on a case-control study, whereas others noted that other factors may have contributed to increases in INR other than acetaminophen ingestion. Another important point is that hemorrhagic events were not reported in this study.
A prospective, longitudinal study of 171 patients receiving warfarin for anticoagulation used diaries to report variables and outcomes of interest. The diary surveyed warfarin use, bleeding events, exposure to factors reported to increase the risk of bleeding, and supratherapeutic anticoagulation. The patients were also asked about the use of complementary and alternative agents, disease states and symptoms, nonprescription drugs, alcohol intake, and vitamin K–rich foods. Bleeding events, including minor bleeds and bruises, were very common and occurred in 87 (51%) of 171 patients throughout the study. Ninety-five participants (56%) reported using acetaminophen during the 16-week study period. Acetaminophen was a statistically significant risk factor for bleeding events (p=0.02), but not significant for supratherapeutic INR results (p=0.13).
In a prospective, double-blind, crossover study, 11 patients received stable warfarin therapy for more than 1 month, with a target INR of 2.0–3.0. In one phase, subjects received 14 days of acetaminophen 1 g 4 times/day; in the other phase, subjects received placebo 4 times/day for 15 days. Subjects were randomly assigned to receive either the acetaminophen or the placebo phase first, followed by the other phase, with a 14-day washout period between the two phases. The mean ± SD maximum INR was 3.47 ± 0.75 in the acetaminophen phase and 2.61 ± 0.40 in the placebo phase (p=0.001). The mean ± SD maximum increase in INR was 1.04 ± 0.55 in the acetaminophen phase and 0.20 ± 0.32 in the placebo phase (p=0.003). The authors concluded that acetaminophen 4 g/day results in a clinically important interaction with warfarin.
In a continuation of the previous study, an additional nine patients were enrolled. Data from the total 20 patients were published separately. The primary end point was the area under the INR change–time curve (AUC-INR), a measure of the change in INR over time. In the 18 evaluable patients, the AUC-INR was higher in the acetaminophen phase than during the placebo phase (mean ± SD 6.64 ± 5.28 vs −1.02 ± 3.22, p<0.001). The change in INR was also higher during the acetaminophen phase than in the placebo phase (0.63 ± 0.65 vs −0.24 ± 0.55, p<0.001). Fourteen patients required no changes in warfarin dose, two required warfarin dose decreases after the acetaminophen phase, and three required increased warfarin doses after the placebo phase. The activities of factors II, VII, IX, and X were also significantly decreased during the acetaminophen phase. No bleeding episodes were observed during either phase. The authors concluded that acetaminophen 4 g/day increases INR during warfarin therapy.
In a prospective, double-blind, randomized, placebo-controlled study, the effect of acetaminophen on INR was investigated. Eligible patients were required to have had two consecutive stable INR results within the therapeutic range at least 3 weeks apart before the study, to have no signs or symptoms of bleeding, and to not have undergone any addition, deletion, or changes in agents known to interact with warfarin in the preceding 2 weeks before possible enrollment. Exclusion criteria included a history of alcohol abuse within the past year (or consumption exceeding 2 oz/day), history of noncompliance, history of hepatic disease, antiphospholipid syndrome, or elevated serum transaminase levels greater than 1.5 times the upper limit of normal. In the event a patient had ingested acetaminophen in the preceding 2 weeks, a 2-week washout period and subsequent INR assessment (required to be in the therapeutic range) was mandatory before inclusion in the study.
The primary end point was the difference in mean INR between groups at weekly intervals. Secondary end points were the calculated percentages of patients in each group with supratherapeutic and subtherapeutic INR values, as well as the difference in mean alanine aminotranferase (ALT) and aspartate aminotransferase levels between the two groups at biweekly intervals. The duration of the study was 4 weeks, during which patients were randomly assigned to receive acetaminophen 1 g 4 times/day, matching placebo tablets 4 times/day, or acetaminophen 1 g twice/day with matching placebo. Baseline serum hepatic transaminase level measurements were taken at the time of inclusion in the study, and patients were required to record daily nutritional intake including alcohol consumption. Thereafter, INR values were measured weekly, on the same day of the week that each patient was enrolled, and hepatic tests were conducted at 2 and 4 weeks. Patients were interviewed at each visit regarding their changes in daily drugs, vitamin K intake, compliance, and adverse events. Significant INR deviation (defined as ≤ 1.8 or ≥ 3.3 for those with a target INR of 2.0–3.0, or ≤ 2.2 or ≥ 3.8 for those with target INR of 2.5–3.5) resulted in withdrawal of the patient from the trial and subsequent corrective action as needed. Patients observed to have transaminase levels of greater than 2 times the upper limit of normal were also withdrawn from the study. Laboratory analysis included INR hepatic function evaluation.
Of 43 patients enrolled, 36 were assigned to a treatment group. Thirteen (54%) of 24 in the acetaminophen groups and 2 (17%) of 12 in the placebo group experienced a supratherapeutic INR exceeding the upper limit by 0.3 or more, which resulted in early termination of the trial. At week 2, a statistically significant increase in mean INR of 0.7 was observed between the acetaminophen 2-g/day group and the placebo group (95% confidence interval [CI] 0.27–1.2, p=0.01). The acetaminophen 4-g/day group experienced statistically significant increases in mean INR compared with placebo at week 1 (0.5, 95% CI 0.08–0.9, p=0.04), week 2 (0.6, 95% CI 0.21–1, p=0.01), and week 3 (1.0, 95% CI 0.48–1.5, p=0.01).
As mentioned above, more than half the patients assigned to an acetaminophen group developed a supratherapeutic INR by the fourth week of the study (INR increase ranged from 0.6–2.8). Of the 13 patients receiving acetamino-phen who experienced an elevated INR, 12 (92%) continued the same warfarin regimen after acetaminophen was discontinued, and in 83% (10/12) of those patients the INR returned to within the therapeutic range.
Among the secondary end points, one patient in each group had a subtherapeutic INR. The ALT levels were statistically significantly increased in the acetaminophen 4-g/day group by week 2, but were not significantly different by week 4. Of note, of the five patients who experienced ALT level elevations greater than the upper limit of normal, four were receiving acetaminophen 4 g/day, and all five had elevated INR results. One episode of minor bleeding and one minor bruising event were reported; both patients had therapeutic INR levels.
The authors concluded that a clinically relevant interaction occurs in patients receiving a warfarin regimen and begin daily ingestion of moderateto- high doses of acetaminophen. Although compliance with warfarin therapy during the study was determined to be 100%, this was purely on the basis of patient recall, which is often unreliable. Further limitations of the study include small sample size and early termination, as early withdrawal of patients with an INR of 0.3 or more above or below the limits of their therapeutic range possibly limited the observation of the magnitude of the interaction, both clinically and statistically. The authors did not explore the mechanism behind the interaction.
Proposed Mechanism of the Acetaminophen-warfarin Interaction
Warfarin is available as a racemic mixture of R- and S-enantiomers. Both R- and S-warfarin are metabolized by various hepatic enzymes to inactive metabolites. Interactions that increase S-warfarin, which has 3-fold higher potency, are known to increase the anticoagulant effect, whereas the resulting effects of interactions that solely increase R-warfarin are controversial. S-warfarin is predominantly metabolized by CYP2C9 and minimally by CYP3A4, and R-warfarin is metabolized by CYP3A4 and CYP1A2, and minimally by CYP2C19. Acetaminophen undergoes metabolism primarily by glucuronidation and sulfation with up to 15% metabolized by CYP1A2, CYP2E1, and CYP3A4. Acetaminophen does not interact directly with warfarin metabolism at CYP2C9 but has been observed to inhibit CYP3A4 in vitro.[19, 20]
It has been proposed that genetic polymorphism of CYP2C9, which is present in as many as 25% of Caucasians, may play a role in the variability of anticoagulant response to warfarin. Genetic polymorphism would cause this minority of patients to have a slower metabolic clearance of S-warfarin and require smaller doses of warfarin to maintain a therapeutic INR. This polymorphism could possibly shift the metabolism of S-warfarin to CYP3A4, increasing utilization of this usually minor pathway. With more substrate shunted to CYP3A4, the significance of drug interactions by this pathway is increased. This is a suggested mechanism of the interaction with acetaminophen, as acetaminophen interacts with both enantiomers at CYP3A4 and the R-warfarin at CYP1A2.
Additional components of the interaction include the observation that R-warfarin has been seen to inhibit CYP2C9 in vitro, potentially increasing the concentration of S-warfarin. Since R-warfarin is metabolized by either CYP1A2 or CYP3A4, for a significant drug interaction to occur with a coadministered agent, the interfering agent must compete for both isoforms simultaneously. Increased utilization of these enzymes for acetaminophen metabolism may occur in various settings where glucuronidation, sulfation, and CYP2E1 are decreased. The process of aging and hypoxic tissue states are suggested to reduce metabolism by these pathways, increasing acetaminophen as a substrate for CYP1A2 and CYP3A4. These mechanisms suggest a biologically plausible mechanism for increased anticoagulation with coadministration of acetaminophen with warfarin.
Some studies have investigated the effect of acetaminophen on INR and specific clotting factors, with conflicting results. One prospective study examined 30 cases of acetaminophen overdose versus eight controls with psychotropic overdose. The investigators reported that functional factor VII was significantly lower and inversely proportional to the INR in the acetaminophen group. The same investigators conducted a retrospective analysis of patients with acetaminophen overdose but without evidence of hepatic injury. The INR values of 143 patients were analyzed, and a correlation was found between acetaminophen dose ingested and an elevated INR. Contrary to these studies, a prospective, randomized, crossover study of 20 patients examined the pharmacokinetics of a single dose of the R- and S-enantiomers of warfarin and demonstrated no significant difference in activated factor VII concentrations. Measurements were made after receiving a single high dose of warfarin or placebo in patients receiving either no acetaminophen or acetaminophen 4 g/day.
Management of Drugs Interacting With Anticoagulant Therapy
The management of anticoagulant therapy is an important component of the treatment of various disease states. Thromboembolic prevention, treatment, and hypercoagulable states often require prolonged use of warfarin as the cornerstone of therapy. Maintaining the narrow therapeutic range required for the safe and effective use of this drug is essential to avoid suboptimal dosing and adverse events. Effectively monitoring the INR requires clinical knowledge, careful attention, and patient communication skills.
Knowledge of the mechanism of action of warfarin, mechanisms of drug interactions, and interpretation of laboratory values are essential for an effective clinician. Documentation of the severity of suspected drug interactions and recommendations for their management is imperfect, and there is a paucity of information in many cases. The severity of interactions with warfarin varies greatly and dictates very different recommendations for management and monitoring. In the most insignificant interactions, no change in dosage or monitoring is necessary, whereas some interactions require a significant empiric reduction in warfarin dosage and close monitoring of INR.
Reports of the interaction between acetaminophen and warfarin are sparse and generally have small sample sizes; however, an appreciable change in INR may occur in some patients taking both drugs. Most case reports include patients receiving warfarin with a previously stable INR who then took medium-to-maximum recommended doses of acetaminophen for several days with a resultant elevated INR. Prospective studies[12, 16–18] have associated acetaminophen with elevated INR values in patients receiving warfarin with previously stable INR values. In these patients, elevations in INR generally occurred with the administration of acetaminophen 2–4 g/day for several days and returned to the therapeutic range within days of discontinuing the acetaminophen. Although there is a lack of specific outcome data, such as bleeding or thromboembolic events, in many prospective trials, the relationship between elevated INR and hemorrhagic adverse events is well established and cannot be ignored.
Clinician Awareness and Patient Education
One troublesome aspect of the interaction between acetaminophen and warfarin is the wide availability and use of acetaminophen. As mentioned earlier, one study identified that 18.5% of a large number of patients taking warfarin used acetaminophen-containing products. Acetaminophen- containing combination prescription opioids and over-the-counter preparations are frequently the source of acetaminophen. The presence of acetaminophen in a patient's drug regimen may be neglected knowingly or unintentionally. With this in mind, it is understandable how patients may take a combination of these acetaminophen-containing products and achieve a high daily dose or even a dose that exceeds the currently accepted maximum daily dose of 4 g.
Thus, patients must receive proper education about reporting the use of any new over-the-counter or prescription drug, including acetaminophen, to the clinician managing their anticoagulation. If acetaminophen is necessary at doses near or greater than 2 g/day for more than 1 day, an extra INR measurement may be appropriate. Extra care should be taken in patients susceptible to fluctuations in INR, such as geriatric patients or those at high risk for bleeding or who have a history of significant bleeding. Based on existing observations, deviations from a therapeutic INR should be handled in accordance with guideline-based care. If patients experience an elevated INR and are found to have been using acetaminophen, discontinuation of acetaminophen, withholding of warfarin dose(s), and close INR monitoring should be performed as needed. Once a therapeutic INR or a predictable trend toward therapeutic INR is achieved, the warfarin should be resumed at the usual dosage.
An interaction between warfarin and the widely available acetaminophen may result in significant elevations of INR and put patients at increased risk for hemorrhagic complications. Patient-specific risk factors for this interaction are not clear and may have genetic and metabolic components. Although the mechanism of the interaction between acetaminophen and warfarin is not entirely established, it may be related to genetic polymorphism or a shunting of the metabolism of the R- or S-enantiomers due to acetaminophen interaction. A biologically plausible mechanism for the interaction does exist, and there are many variables in the hepatic clearance of warfarin and acetaminophen at the enzymatic level. Regardless, preemptive clinical acknowledgment of the interaction is necessary to reduce the risk of hemorrhagic adverse events by recommending avoidance of the combination or, when acetaminophen is necessary, performing adequate monitoring and dosage adjustment.