DrugInteractions

ADVERSE INTERACTIONS BETWEEN WARFARIN AND

NONSTEROIDAL ANTIINFLAMMATORY DRUGS: MECHANISMS,
CLINICAL SIGNIFICANCE, AND AVOIDANCE
Thomas YK Chan

WARFARIN IS COMMONLY USED in the treatment of thrombo-

embolic disease. Optimal therapy with this drug must aim
for a fine balance between prevention of intravascular
thrombosis and the production of unwanted bleeding. Sev-
OBJEL'TIVE: To review the mechanisms and clinical significance of eral factors may affect an individual's response to warfarin,
adverse interactions between warfarin and nonsteroidal anti-
inflammatory drugs(NSAIDs) and discuss how theseinteractions including age, diet, disease states, and drug interactions.
can be avoided. Numerous drugs have been reported to interact with war-
farin.!" These adverse interactions may result in loss of an-
DATA SOURCFS: Previous studies of interactions between warfarin
and NSAlDsor reports of adverse interactions wereidentified from
ticoagulant control and serious consequences. Extra care
a MEDLINE search(1976to present) and fromthe reference listsof should, therefore, be taken with the introduction of any
pertinent articles. new drug after stabilization of the warfarin dosage.
STIJDY SELECTION ANDDATA EXTRACTION: All articles wereconsidered
Despite extensive clinical experience with both warfarin
for inclusion in the review. Pertinent information was selected for and aspirin and other nonsteroidal antiinflammatory drugs
discussion. (NSAIDs), physicians are uncertain as to the safety of us-
ing them in combination! Although patients receiving war-
DATA SYNTIIESIS: All NSAIDs can prolong bleeding time by
inhibiting platelet function. High-dose aspirin has a direct farin are routinely advised to avoid NSAIDs, the risk of
hypoprothrombinemic effect. Phenylbutazone and its analogs bleeding during combination therapy and whether certain
enhance the hypoprothrombinemic effectof warfarin through a individuals are at particular risk have not been well de-
pharmacokinetic interaction by inhibiting the hepatic metabolism of fined. In this article, previous reports of interactions be-
warfarin. Mefenamic acidalsoenhances the anticoagulant effectof tween warfarin and NSAIDs and the possible mechanisms
warfarin, but the mechanism is not known. The clinical relevance of for these interactions are reviewed and some practical guid-
protein binding displacement in the interaction between warfarin ance on prescribing is given.
and NSAlDshas beenoverstated. although a significant one may be
morelikelyin the presence of highconcentrations of NSAIDs in
patients withslowelimination of warfarin (e.g.,thosewithsevere ClinicalPharmacology of Warfarin
heartfailure or impaired liverfunction). NSAlDscan induce
gastrointestinal bleeding, which is likely to be moresevereif Warfarin sodium is a racemic mixture of 2 optical iso-
warfarin is alsogiven. mers, R- and S-warfarin. These enantiomers are metabo-
CONCLUSIONS: The combined use of warfarin and NSAIDs is lized through different pathways and have different half-
generally discouraged because of the increased risk of bleeding in lives and potency; S-warfarin is 5 times more potent,' but
thesepatients. In patients receiving warfarin who also require is eliminated more rapidly than R-warfarin.7
NSAIDs, phenylbutazone and its analogs, high-dose aspirin. Warfarin is essentially completely absorbed after oral
mefenamic acid,excessive use of topical methyl salicylate. and administration, with the peak concentration occurring 2-6
NSAlDsthat are associated witha higherriskof bleeding peptic hours after ingestion." Warfarin has a relatively small vol-
ulcers shouldbe avoided. Patients shouldbe closely monitored for ume of distribution of approximately 10 U70 kg. The high
anticoagulant control and bleeding complications duringthe
combined use of warfarin and NSAlDs.
degree of binding of racemic warfarin to plasma albumin,
more than 99% at therapeutic concentrations, can explain
Ann Pharmacother 1995;29:1274-83. its clinical pharmacology: the prolonged half-life and bio-
logic effect.' At any given time, only the unbound portion
Thomas YK Chan MBChB FRCP (Edin), Associate Professor and Physician, De- is active and available to be metabolized by the liver. The
partment of Clinical Pharmacology, The Chinese University of Hong Kong, average biologic half-life in humans is about 35 hours,
Prince of Wales Hospital, Shatin, New Territories, Hong Kong, FAX 852/
26373929 which means that it takes about a week to reach steady-
Reprints: Thomas YK Chan MBChR FRCP (F..din) state,"

1274 • TheAnnalsofPharmacotherapy • /995 December. Volume 29

Downloaded from aop.sagepub.com at CORNELL UNIV on July 27, 2016
 Warfarin reversibly binds and inhibits vitamin K epox- been found to displace warfarin from plasma albumin
ide reductase and vitamin K quinonereductase.'These en- bindingsites in vitro,22 this interaction has only been dem-
zymes are responsible for converting the biologically inac- onstratedin vivo with some NSAIDs.19 Moreover, in vivo
tive vitamin K epoxide to active vitamin K hydroquinone. studies have shown that this displacementdoes not neces-
Thus, warfarin inhibits the final activation step in the syn- sarily increase the anticoagulant action of warfarin. An in-
thesis of vitamin K-dependent clotting factors (factors II, crease in the plasmaconcentration of unbound (pharmaco-
VII, IX, and X and protein C and protein S). As warfarin logically active) warfarinfollowing the introduction of the
competes with vitamin K for receptors, vitamin K content displacing NSAlD will be transient because of the con-
in the diet can be an importantdeterminant of anticoagula- comitantincreasein the clearance of unboundwarfarin un-
tion. til the previous concentration is reached." Thus, displace-
The anticoagulant effect is measured with the interna- ment may be important only for intermittent warfarin use
tional normalized ratio (INR), which is a standardized pro- and on introduction and withdrawal of the displacing
thrombin time (PI'). Different therapeutic rangesfor differ- NSAIDs.3 In such situations, the effect again is transient
ent clinical indications have been used." The target range and steady-state shouldbe reachedafter 7-10 days. A clin-
for the INR is critical becausethe level of anticoagulant ef- ically significant interaction also may be more likelyin the
fect is the major determinant of bleeding. The riskof bleed- presence of high concentrations of NSAIDs in patients
ing is increased in the elderly" and is highestat the start of with slow elimination of warfarin (e.g., those with severe
therapy." The risk for major bleeding during the first heart failure or impairedliver function)."
month of therapy is approximately 10 times the risk after Phenylbutazone substantially inhibits the metabolism of
the first year of therapy. The aim of treatment is to achieve S-warfarin while increasing the hepatic clearance of R-
the minimum anticoagulant effect required to prevent warfarin." The net result is increased anticoagulation with-
thrombosis or embolism. out a change in the overall clearance of racemic warfarin.
Patients vary widely in the dosage of warfarin required It is likely that analogs of phenylbutazone (oxyphenbuta-
to maintainthe INR at any given level." Variation of up to zone, apazone,and sulfinpyrazone) will interactwith war-
20-fold has been described in some cases,although 90% of farin in the same manner. Similar data for other NSAlDs
white patients require a daily dosage of 2-10 mg to main- are not available, as plasma concentrations of total war-
tain an INR of 1.8-2.5.J3Variation appears to be approxi- farin rather than its isomers have been measuredin studies
matelyequally the result of differences in the pharmacoki- of interaction with warfarin (Table 1).
netics (abouta 6-fold variation in totalclearance) and phar-
macodynamics (about a 6.5-fold variation in plasma free PHARMACODYNAMIC MECHANISMS
concentration-effectrelationship) of the drug." The mean
maintenance daily dosageis considerably less in Chinese(3 Twopharmacodynamic interactions havebeendescribed:
mg) than in white patients (4-6 mg)." In both ethnic dir~ct effects on platelets and inductionof upper gastroin-
groups, the age of the patientrather than body weightis an testinal (01) bleeding'? Aspirin irreversibly acetylates plate-
importantdeterminant of warfarin requirement.P" let cyclooxygenase at low or high doses and produces irre-
versible effects on platelet function that persist for the life
Interactions Between Warfarin and NSAIDs:Possible of the aspirin-treated platelet." Platelet aggregation thus
Mechanisms may be inhibitedand the bleeding time prolonged. Similar,
but usually lesser, effectson plateletfunction may be seen
There are several mechanisms by which NSAIDs may with nonaspirin NSAIDs, but unlike aspirin, these effects
increase the likelihoodand severityof bleeding in patients are lost when the drug or any active metabolites are re-
receiving warfarin. I -5,16-19 moved from the circulation.' Thus, the effects are usually
reversible in 24 hours for short-acting NSAlDs, but may
DIRECf HYPOPROTIlROMBINEMlC EFFECT OF ASPIRIN continue for many days with longer-acting NSAIDS.Ui Be-
cause of the adverse effects on hemostasis, NSAlDs in-
Aspirin and other salicylates may have a hypoprothrom- crease the risk of bleeding from the GI tract" or other
binemic effect by depressing the vitamin K-dependent sites:l8 during concomitant use.
synthesis of clottingfactors Vll, IX, and X. However, such Several studies have shown that a person exposed to
an action only becomes significant if the equivalent of NSAIDs has 3-4 times the risk of upper 01 bleeding, per-
more than 6 g (for a 70-kg man) of aspirin per day has foration, or both than a personwho has not been exposed."
been ingested." The PT is rarely prolonged unless the A wide range of mechanisms could explainthe occurrence
serum salicylate concentrations are more than 300 of NSAlD-induced gastroduodenal mucosal damage, in-
mg/ml.," concentrations at which salicylate toxicity also is cluding the inhibition of bicarbonate secretion, effects on
more likely. mucus formation, and vascular actions." The individual
risks are low, of the order of 1 episode for every 10000
PHARMACOKINETIC MECHANISMS NSAID prescriptions issued to people aged 60 years and
older in the UK,31 Becausemany such prescriptions are is-
Two major types of pharmacokinetic interactions have sued, there are many episodes. In a study of patients with
been described: displacementof warfarin from plasma al- rheumatoid arthritis, the incidence of hospitalization be-
bumin and inhibition of the metabolism of warfarin by cause of upper 01 bleedingin patientstaking NSAIDs was
NSAIDs,l9 The role of protein bindingdisplacementin in- 1.58% per year compared with 0,3% in those not taking
teractions with warfarinhas been overstated},18,19 Although the agents." In additionto old age, other clinical variables
most NSAIDs, which are also highly protein-bound, have of predictive value included use of prednisone, disability,

The Annals ofPharmacotherapy • 1995 December. Volume 29 • 1275

Downloaded from aop.sagepub.com at CORNELL UNIV on July 27, 2016
 Table I. Studies of Interactions Between Warfarin and Nonsteroidal Antiinflammatory Drugs
NSAID REF. SUBJECTS STUDY DESIGN MAJOR FINDINGS

Aspirin Holmes 12 healthy men all given warfarin (plasma prothrombin further decreases in plasma prothrombin concentrations
(1966)33 concentrations 1Q-.-20% of normal), after aspirin (2.75%) or mefenamic acid (3.49%);
then placebo, mefenamic acid 500 mg microscopic hematuria in 3 subjects
qid, placebo, aspirin 650 mg qid, or,
mefenamic acid, placebo, aspirin, pla-
cebo, each given for I wk
O'Reilly et al. 13 healthy men, 13 men taking stabilized dose of warfarin a decrease in prothrombin activity in 3 men taking low-
(1971)34 aged 21-28 y for ~2 wk given aspirin 0.65 g tid for 7 d dose aspirin and all 4 men taking high-dose aspirin;
=
(n 9) or 0.975 g tid for '5.7 d (n 4) =
bleeding in 2 men taking low-dose aspirin and in all 4
men taking high-dose aspirin
6 men taking warfarin (prothrombin bleeding time did not change with warfarin alone, but
activity <35% of normal for 7 d) given was prolonged from 4.0 to 10.3 min after aspirin added
aspirin 0.65 g tid for 3 d
Diflunisal Serlin et al. 5 healthy men, weeks 1-2,5 men receiving subthera- prothrombin complex activity fell (74% to 43%) in
(1980)35 aged 20--40 y peutic doses of warfarin (PT prolonged weeks 1-2, did not change in weeks 3-4, increased to
for a few seconds); weeks 3-4, warfarin >70% in weeks 5-6; plasma total warfarin concentra-
plus diflunisal 500 mg bid; weeks 5--6, tion fell by 39% and plasma free warfarin concentration
warfarin alone increased (1.02% to 1.34%) in weeks 3-4; plasma free
warfarin concentration returned to basal levels within a
few days of stopping diflunisal and plasma total warfarin
concentration returned to basal levels after 2 wk; difluni-
sal alone did not affect clotting factor activity
Etodolac Ermeret al. 18 healthy men, warfarin (20 mg on day I, 10 mg on during concomitant etodolac use, plasma total warfarin
(1994)36 aged 22-36 y days 2 and 3), etodolac 200 mg 12 concentration fell by 19%, total clearance increased by
hourly for 5 doses 13%, no change in peak PT response or area under the
curve of change in PT, when compared with warfarin
alone; most but not all subjects attained steady plasma
warfarin concentration by day 3, PT response varied
considerably among the group
Fenbufen Savitsky et al. 16 healthy days 1-10, warfarin 7.5 mg/d; days 11- 6 subjects withdrawn because PT >30 s; 5 subjects in
(1980)31 subjects 17, warfarin plus fenbufen 400 mg bid each group completed the study; days 11-17, fenbufen
= =
(n 8) or placebo (n 8); days 18-24, compared with placebo groups: PT (+1.9 s, NS), aPTT
=
fenbufen (n 8) or placebo (n 8) =
(+2 s, NS), warfarin concentration (-14%, p 0.03), =
similar warfarin disappearance rate and rate of normal-
ization of PT and aPTT
Ibuprofen Goncalves 50 patients factors II, VII, and X and aPTT assayed no significant changes in factors II, VII, and X concentra-
(1973)38 taking warfarin before and 7 d after ibuprofen 600- tions and aPTT before and after ibuprofen therapy
1200mg/d
Penner and 36 healthy men days 1-14, each group of 6 men given no changes in PT, aPTT, or plasma factors II, VII, IX,
Abbrecht aged 21--60 y placebo, ibuprofen 300 mg qid or 600 and X during or after 14 d of placebo or ibuprofen; sim-
(1975)39 mg qid, warfarin 7.5 rng/d, or warfarin ilar increases in PT and aPTT and similar decreases in
7.5 mg plus ibuprofen 300 mg qid or plasma factors II, VII, IX, and X during or after 14 d of
600 mgqid warfarin with or without ibuprofen; similar plasma war-
farin concentrations during or after 14 d of warfarin or
warfarin plus ibuprofen
Schulman and 19 patients aged patients receiving warfarin for 0.25-84 no change in PT after ibuprofen was added for I wk;
Henriksson 35-80 Ywith mo given ibuprofen 600 mg tid for I prolonged bleeding time on day 8 in 3 patients, 2 of
(1989)40 pulmonary wk whom also had microscopic hematuria or subcutaneous
thromboembol- hematomas; 3 of 16 patients with normal bleeding
ism and rheu- times developed subcutaneous hematomas
matic or non-
rheumatic
disorders
Indomethacin Vesell et al. 16 healthy men all subjects stabilized on warfarin thera- no differences between indomethacin and placebo with
(1975)41 aged 19-28 y py (PT 1.5-2.5 times control value) regard to changes in PT from baseline and the rate of
given indomethacin 100 mg/d or return to the pretreatment PT during recovery period
placebo for 5 d
19 healthy men all given warfarin 0.75 mglkg on days I no differences between indomethacin and placebo with
aged 22-27 y and 31; indomethacin 100 mgld or regard to effects on warfarin-induced hypoprothrom-
placebo on days 21 and 31 binemia or plasma warfarin half-life
Ketoprofen Mieszczak and 13 healthy men successive 7-d periods: run-in/keto- during ketoprofen use, no change in factors II, VII,liM
Winther aged 26-45 y profen!wash-outlplacebo or run-in! X, thrombin time and aPTT; massive lowering of plate-
(1993)° placebo/washoutlketoprofen; warfarin let aggregability (a 38-fold difference) by ADP
dosage adjusted during run-in period lasting for 24 h
(factors II, VII, and X lowered by 60%),
then fixed; ketoprofen 100 mg bid
given

= = =
ADP adenosine diphosphate; aPTT activated partial thromboplastin time; INR international normalized ratio; NS =not significant; NSAID =nonsteroidal
= =
antiinflammatory drug; OA osteoarthritis; PT prothrombin time; RA = rheumatoid arthritis.

/276 • The AnnalsofPharmacotherapy • 1995 December, Volume 29

Downloaded from aop.sagepub.com at CORNELL UNIV on July 27, 2016
 Warfarin-NSAID Interactions

Table 1. Studies ofInteractions Between Warfarin and Nonsteroidal Antiinflammatory Drugs (continued)
NSAID REF. SUBJECTS STUDY DESIGN MAJOR FINDINGS

Ketorolac Toon et a\. 10 healthy men all received ketorolac 10 mg qid or ketorolac did not significantly affect the magnitude of
(1990)43 aged 20-32 y placebo for 12 d and warfarin 25 mg warfarin-induced changes in PT and clotting time, or the
on day 6; plasma warfarin enantiomers pharmacokinetics of warfarin enantiomers; ketorolac
measured at 0-168 h after administra- increased bleeding time by a factor of 1.35 compared
tion of warfarin; crossover after ;:::14 d with placebo
Lornoxicam Ravic eta\. 12 healthy men days 1-5, lornoxicam 4 mg/d; days 9-24, significant changes in PT (23.6 vs. 19.5 s), INR (1.48 vs.
(1990)44 aged 22-41 y lornoxicam plus warfarin to achieve an 1.23) and total serum warfarin concentration (1.02 vs.
INR of 1.4-1.6 for;:::6 consecutive 0.77 ug/rnl.) at the end of phase 2 compared with phase
days; days 25-33, warfarin alone at 3; the significant decreases in PT, INR, and serum total
same dosage warfarin concentrations from phase 2 to phase 3 sug-
gested a significant interaction
Mefenamic acid Holmes see aspirin
(1966)33
Nabumetone Fitzgerald" 12 healthy men all subjects stabilized on warfarin for 3 no significant differences between subjects treated with
wk (mean INR 2.6-3.8), then also given nabumetone or placebo with respect to INR, IT, or
nabumetone 2000 mg/d or placebo for bleeding time
2wk
Hilleman et a\. 58 patients tak- all patients receiving chronic warfarin no significant differences in INR before and after
(1993)46 ing warfarin therapy also given nabumetone 1000 nabumetone for 6 wk
with OA or RA or 2000 mg/d for 6 wk
Naproxen Slattery et al. 10 healthy men warfarin 50 mg on days I and 29; during concomitant use with naproxen, serum free war-
(1979)47 aged 24-31 y naproxen 375 mg bid on days 19-35 farin concentration increased by 13%, the total clear-
ance and intrinsic clearance of warfarin did not change,
the area above the prothrombin complex activity vs.
time curve did not change
Jain et a\. 10 healthy sub- days 1-25 warfarin in a dose to prolong in the 5 subjects who received a constant daily dose of
(1979)48 jects aged PT to 1.5-2.0 times the normal value; warfarin, no significant difference in the serum free or
25-65 y days 11-20, also received naproxen total warfarin concentrations and PT before, during, and
375 mg bid after naproxen; the others required frequent dose adjust-
ment prior to and during naproxen therapy and 2 had
hematuria (day 18) or bruise aftertrauma (day 12)
Nimesulide Auteri et al. 10 patients all received warfarin 5 mg/d and then no significant change in PT, aPTT, and bleeding time
(1991)49 nimesulide 100 mg bid for 7 d before and after nimesulide
Phenylbutazone Banfield et al. 3 healthy warfarin 1.5 mg/kg before and 4 d during concomitant use of phenylbutazone, elimination
(1983)24 subjects into a 12-d regimen of phenylbuta- half-life of S-warfarin increased (25 to 46 h) and R-war-
zone 100 mg tid farin decreased (37 to 25 h); peak unbound concentra-
tions of both enantiomers increased, unbound clearance
of S-warfarin decreased by 4-fold; unbound clearance
of R-warfarin was almost unchanged
Sulindac Loftin and 19 healthy men all subjects stabilized on warfarin no significant difference in PT or plasma warfarin con-
Vesell aged 21-31 y therapy (PT 1.5-2.5 times the control centrations between the 2 groups in the 2nd phase;
(1979)50 value), then received sulindac 200 mg after warfarin was stopped, PT and aPTT were higher
bid (n = 10) or placebo (n = 9) for 10 and returned more slowly to normal in the sulindac
d; warfarin was stopped after 7 group
d of coadministration
Tenoxicam Eichler et al. 16 healthy 8 subjects given warfarin 0.75 mg/kg on no significant difference in plasma warfarin-time profile
(1992)5\ subjects aged 2 occasions, 4 wk apart, with or without or PT between treatment cycles
19-35 y tenoxicam 20 mg/d in preceding 2 wk
8 subjects given tenoxicam 20 mg/d no effects on PT, aPTT, IT, activity of factors II, V, VII,
for 14 d IX, and X, and bleeding time
8 subjects stabilized on warfarin therapy no significant difference between treatment periods with
(INR 1.47-1.81) given tenoxicam 20 respect to INR and warfarin requirements
mg/d for 2 wk: tenoxicam then stopped
while warfarin continued for 4 more wk

= = =
ADP adenosine diphosphate; aPTT activated partial thromboplastin time; INR international normalized ratio; NS =not significant; NSAID =nonsteroidal
= = =
antiinflammatory drug; OA osteoarthritis; PT prothrombin time; RA rheumatoid arthritis; TT thrombin time. =

the doseof NSAIDs, and previous problems withNSAIDs. can be determined at least in partby considering studies in
Upper GI bleeding is expectedto be more pronounced in healthy subjects or patients and reports of adverse interac-
patients takingwarfarin. tionswith or without bleeding. Giventhat concurrent war-
farin therapy is expectedto increase the severity of bleed-
Determining the Risk ofAdverse Interactions Between ing from peptic ulcers, the risk of pepticulceration associ-
Warfarin and NSAIDs ated with the use of individual NSAIDs also should be
considered. It is expected that adverse interactions between
Whetherall NSAIDs interactwith warfarin to a similar these 2 drugs are more likelyto be of clinical significance
extent and the safety in using these drugs in combination in subjects who are particularly sensitive to warfarin.

The AnnalsofPharmacotherapy • 1995 December, Volume 29 • 1277

Downloaded from aop.sagepub.com at CORNELL UNIV on July 27, 2016
Clinical Studies complications, ibuprofen could well have been the offend-
ing agent.
The study design and major findings from the previous The hazard from the excessive use of topical analgesic
studies of interactions between warfarin and aspirin,JJ.34 preparations containing methyl salicylate is not well recog-
diflunisal," etodolac," fenbufen," ibuprofen,J8-40 indo- nized. Three patients aged 41-65 years taking warfarin af-
methacin," ketoprofen," ketorolac," lornoxicam," mefe- ter placementof a prosthetic mitral valve had a markedly
namic acid,JJ nabumetone.v-" naproxen.t'-" nimesulide," prolonged INR (4.2-5.5), increasedblood salicylate con-
phenylbutazone." sulindacj" and tenoxicam" are shown in centrations(0.4-0.8 mmollL), and bleeding (bruisesin 2
Table l. 14,33-51 and GI hemorrhage in I) after excessiveuse of such oint-
ments for myalgiaor osteoarthritis pain." An 85-year old
CAUTIONS woman with congestive heart failure and atrial fibrillation
received warfarin for pulmonary embolism." She present-
Aside from findings concerning phenylbutazone and its ed with vaginal bleeding, ecchymosis over the left cheek,
analogs, findings fromstudies of interactions between war- the right flank, and the perineum, a decrease in hemoglo-
farin and other NSAIDs may not be extrapolated to all bin concentration from 117to 80 gIL, and a prolongation
clinical situations. First, individual differences with respect of PT from 18.5 to 55.0 seconds. Duringthe week priorto
to warfarin and NSAIDs do occur.! Second, patients re- admission, she had applied large quantities of topical
ceiving warfarin and NSAIDs in combination also may methyl salicylate to the skin over her joints because of
have liverdysfunction, cardiac failure, or other serious ill- arthralgias. She was given intravenous vitamin K and a
nesses that require other medications.In many of these bloodtransfusion. Methyl salicylate in Chinese proprietary
studies, subjects weregiven warfarin in subtherapeutic dos- medicines may interactwith warfarin in a similarmanner.
es. INRs were seldom reported and therefore the intensity A 66-year-oldman receiving warfarin for rheumatic heart
of anticoagulation could not be judged. Apart from the disease and cerebral embolism had an INR of more than
study concerning ketorolac or phenylbutazone, plasma 5.5 and GI bleeding from undiagnosed gastric carcinoma
concentrations of racemic warfarin rather than its enan- after he had applied Kwan Loong Medicated Oil (15%
tiomers were measured. It is most important to remember methyl salicylate) to his entire chest wall becauseof chest
that, even in the absenceof pharmacokinetic interactions, pain." He also took danshen (the root of Salvia miltior-
bleedingstill can occur because of pharmacodynamic in- rhiza Bge), which has been shownto affect the elimination
teractions. of warfarin in rats."
Concomitant use of warfarin and low-doseaspirin is as-
BleedingComplications sociated with an increased risk of bleeding. In 186patients
with mechanical heart valves or with tissue valves plus
There have been numerous reportsof bleeding compli- atrial fibrillation or thromboembolism," adding aspirin
cations in patients receiving long-term warfarin therapy 100 mg qd to warfarin treatment (targeted INR 3.0-4.5)
who also were taking NSAIDs. Minor or major bleeding was associated with a 55% increase in bleeding and a 27%
also was seen in healthy volunteers during the studies of increase in major bleeding episodes after an average fol-
interactions between warfarin and aspirin.P-" mefenamic low-up periodof 2.5 years compared with controls treated
acid," and naproxen" (Table I). with warfarinonly. However, the risk of the combination
Phenylbutazone and its analogs potentiate the anticoag- therapy was more than offset by the major reduction in
ulant effect of warfarin and increase the risk of hemor- systemic embolism or deaths from vascular causes.
rhage in all patients.Z4.SZ-6Z It is expected thatall phenylbuta- The incidence of bleeding complications was compared
zone analogs also will substantially inhibitthe metabolism in 2026 patients treated with warfarin alone (targeted INR
of the more potent S-warfarin while enhancingthe elimi- 2.5-4.2) and 1140patients treated with warfarin plus as-
nation of R-warfarin.14•6z,6J pirin 150mg qd (targeted INR 2.2-2.8).18 The mainindica-
Because phenylbutazone and its analogs are now rarely tionsfor warfarin therapy in the warfarin only and warfarin
used in clinical therapeutics and potentiation of warfarin- plus aspirin groups were as follows: acute myocardial in-
induced anticoagulant effect appears to be universal, previ- farction (27.6% vs. 61.9%), unstable angina (5.7% vs.
ous reports of adverse interactions between these drugs 21.0%), coronary artery bypass grafts (2.8% vs. 10.1 %),
and warfarinare not considered further. Reports concern- prosthetic heart valve (9.5% vs. 2.0%),thromboembolism
ing otherNSAIDs are summarized in Table 2.40,64-70 (38.6% vs. 7.0%), and atrial fibrillation (7.8% vs. 1.6%).
Bleedingcomplications following the.oral administra- After a total observation time of 4420 treatment-years, the
tion of ibuprofen,40,64 indomethacin,6S,66 ketoprofen," sulin- incidence of minor bleeding episodeswas significantly (p
dac,68,69 and tolmetin" have been reported. The first report < 0.(03) lower in the warfarin group than in the combined
of an adverse interaction between indomethacin and war- therapy group (1.4% vs. 2.9%). However, the incidence of
farincame from a prospective study of 500 warfarin-treat- serious(3.2% vs. 2.8%, not significant) or fatal (0.6% vs.
ed patients, but the clinical details were not included." In 1 0.4%, not significant) bleeding episodesdid not differ be-
reportconcerningibuprofen," a patient with normal liver tween the 2 groups. In those with bleeding complications,
function took an unknown numberof 200-mg tabletsover 39% in the warfarin groupand 35% in the combined thera-
4 days and about 30 tabletsof Darvocet-N (propoxyphene py grouphad an INR abovethe targeted range.
napsylate 100mg/acetaminophen 650 mg) over 3 days pri- Concurrent use of warfarin and NSAIDs not containing
or to the onset of bleeding. Although the authors believed aspirin is associated with a higherincidence of hemorrhag-
that an adverse interaction between Darvocet-N and war- ic peptic ulcers. In a retrospective cohort study of Ten-
farin was responsible for the change in PT and bleeding nesseeMedicaid enrollees aged 65 yearsor older," the in-

/278 • The AnnalsofPharmacotherapy • /995 December, Volume 29

Downloaded from aop.sagepub.com at CORNELL UNIV on July 27, 2016
 Warfarin-NSAID Interactions

Table2. Case Reportsof Bleeding Complications DuringCombinedUse of

Warfarin and Nonsteroidal Antiinflammatory Drugs"
DIAGNOSIS INDICATION
PATIENT (duration of (duration of CHANGES IN
NSAID REF. sex/age(y) warfarin therapy) NSAID therapy) PTRATIO COMPLICATIONS TREATMENT

Ibuprofen Schulman and 5 rheumatic PTE (0.3-48 mol rheumatic disorders subcutaneous hema-
Henriksson patients (600 mg tid for I wk) toma (n '" 4), mi-
(1989)40 croscopic hematuria
(see Table I) (n'" I)
Justice and M/24 loin pain, hematuria postconcussion pain PT 13.6 to skin bruising, oozing vitamin K
Kline syndrome (200 mg for> I d, also >100 s from shaving
(1988)64 took Darvocet-N) nicks
Indomethacin Koch-Weser clinical details not available
(1973)65
Chan et al. Ml57 DVT, hypertension, gout (~25 mg/d for INR <2.5 to 3.4 skin bruising, FFP
(1994)66 gout (6 d) ~lOd) hematuria
Ketoprofen Flessner and Ml62 dilated cardiomyopathy. pain (25 mg tid for 6 d) PT 18 t041 s bleeding duodenal transfusion,
Knight COPD, gout, carcin- and gastric ulcer, FFP
(1988)67 oma of posterior cri- esophagitis
coid region
Sulindac Ross and F/53 PTE, was taking flurbi- (100 mg bid for 3 INR 2.0 to 4.6 fresh blood per
Beeley profen 50 mg qid doses) rectum
(1979)68 (I mol
Thatcher and F/56 DVT. PTE (3 mol rheumatoid arthritis PT 17 to 26.6 s retropharyngeal vitamin K.
George hematoma FFP
(1987)69
Tolmetin Koren etal. Ml64 DVT, IODM. CRF, shoulder pain (400 mg PT 15-22 to epistaxis, occult vitamin K.
(1987)'0 IHD(4mo) bid for 3 doses) 70.2 s gastrointestinal FFP
bleeding

COPD '" chronic obstructive pulmonary disease; CRF '" chronic renal failure; DVT '" deep-vein thrombosis; FFP '" fresh frozen plasma; IODM '" insulin-de-
pendent diabetes mellitus; IHD '" ischemic heart disease; INR '" international normalized ratio; NSAID '" nonsteroidal antiinflammatory drug; PT", pro-
thrombin time; PTE", pulmonary thromboembolism.
"Refer to Table I for minor or major bleedings during studies of interactions between warfarin and aspirin,33.'"mefenamic acid," and naproxen" in healthy
subjects. and to text for reports involving topical methyl salicylate therapy.

cidenceof hospitalizations for hemorrhagic pepticulcersin plied topical trolarnine salicylate to his neck and shoulders
patients taking an anticoagulant was 10.2per 1000person- on severaloccasions becauseof pain."
years, more than 3 times that of those not taking an antico-
agulant. The incidenceof hospitalizations for hemorrhagic
peptic ulcers in patientstaking both warfarinand NSAIDs Risk of Upper Gastrointestinal Bleeding with an NSAJD
not containing aspirin was 26.3 per 1000 person-years, In a population-based, retrospective, case-control study
about 13 times that of those not taking either drug. The of 1457 adults aged 25-77 years with upper GI bleeding
prevalence of NSAIDuse among those takingan anticoag- and perforation in the UK between 1990 and 1993,80 the
ulant was 13.5%, the same as in those who were not taking relative risk associated with current use of NSAIDs not
anticoagulants. containing aspirin was 4.7. Previous upper GI bleeding
was the single most important predictorof the disease(rel-
LossofAnticoagulant Control Without Bleeding ative risk 13.5). For all NSAIDs together, the risk was
Complications greater for high doses than for low doses (relativerisk 7.0
vs. 2.6).Those taking apazone(23.4) and piroxicam (18.0)
An increase or a decrease in anticoagulanteffect when had the highest riskof upperGI bleeding. All otherNSAIDs
an NSAID is given to or withdrawn from patients receiv- with sufficient data for individual analysis (i.e., ibuprofen,
ing warfarin therapy also suggests an adverse interaction naproxen, diclofenac, ketoprofen, indomethacin) had rela-
between the 2 drugs. There have been reports concerning tive risks similarto that for overallNSAID use.
indomethacin," piroxicam,":" and sulindacso,67.79 given In a prospective, case-controlstudy, 1144patients aged
orally(Table 3). 60 years or older were admitted to hospitals in the UK
The INR of 8 patientsaged 15-64 years receiving war- with bleeding peptic ulcers between 1987 and 199J.31 In
farin for prosthetic heart valves or mitral stenosis was this study, peptic ulcer bleeding was strongly associated
markedly increased from 1.9-2.6 to 3.3-5,2 after using with the use of NSAIDs other than aspirin during the 3
topicalmethyl salicylate ointmentfor osteoarthritis or my- months before admission (relative risk 4.5). The relative
algia." They all had increasedblood salicylateconcentra- risk was lowest for ibuprofen (2.0) and diclofenac (4.2),
tions (0.2-0.6 mmollL) on admission. A 68-year-old man and intermediatefor indomethacin (11.3), naproxen (9.1),
taking warfarin for atrialfibrillation experienced prolonga- and piroxicam (11.3). Apazone (31.5)and ketoprofen (23.7)
tion of INR from 1.3-1.5 to 2.5 after he had liberally ap- had the highest risks. Risks also increased with drug dose

The Annals ofPharmacotherapy • 1995 December, Volume 29 • 1279

Downloaded from aop.sagepub.com at CORNELL UNIV on July 27, 2016
(low dose 2.5, intermediate dose 4.5, high dose 8.6) for all Other risk factors for warfarin-induced bleeding include
NSAIDs combined. a history of GI bleeding, cerebrovascular disease, heart dis-
ease, renal insufficiency, hypertension, atrial fibrillation,
cancer, and severeanemia."
Groups at Risk ofBleeding
Some individuals are expected to be particularly at risk Summary
of more severe bleeding as a consequence of adverse inter-
actions between warfarin and NSAIDs. These include pa- The following sections summarize the adverse drug re-
tientswho are more sensitive to warfarin" or proneto war- actions betweenwarfarin and NSAIDs.
farin-related bleeding, \I and those who are at higherrisk of
NSAID-induced GI bleeding." SALICYLATES
There is an age-relateddecline in the warfarin dose re-
quired to produce the same anticoagulant effect.' 3,'5,81,82 • Aspirin has a direct hypoprothrombinemic effect only
The decrease with age in the free plasma warfarin concen- if given in high doses (for a 70-kg man, the equivalent
tration required to produce a particular anticoagulant ef- of >6 g/d of aspirin given orally) or at serum salicylate
feet" and the greater reduction in the clotting factor syn- concentrations of morethan 300 mg/mLwithevidence
thesis in older subjects" suggest that the reduced dosage of acutetoxicity.
requirements are the result of increased target organ sensi- • In healthysubjects, aspirinin largerdosages (2-4 g/d)
tivity. Old age is also an independent risk factor for war- potentiates the anticoagulant effect of warfarin and in-
farin-related bleeding.\1,85 In those aged 65 years or older, creases the risk of bleedingin a dose-dependent man-
concurrent use of NSAIDs and oral anticoagulants is asso- ner.
ciated with a l3-fold increase in the risk of developing • In patients with prosthetic heart valves, concomitant
hemorrhagic peptic ulcers." The elderly are also more use of warfarin (INR 3.0-4.5) and low-dose aspirin
proneto NSAID-induced pepticulcers,JO,32 (100 mg/d) increases the risk of both minor and major
In general, liverdiseases are associated withan enhanced bleeding episodes.
anticoagulant response to warfarin." This increased sensi- • The risk of minor but not major or fatal bleeding
tivity is causedmore by impaired hepatic synthesis of vita- episodes was higher in patients treated with warfarin
min K-dependent clotting factors than by a reduced rate of plus aspirin 150 mg/d (lNR 2.2-2.8) than in those re-
warfarin metabolism. Studies of patients with acute viral ceivingwarfarin alone (INR 2.5-4.2).
hepatitis showed an increased anticoagulant response in • Excessive useof topical medicaments containing methyl
20% of the subjects, withoutany change in warfarin phar- salicylate in patients receiving warfarin may result in
macokinetics." systemic absorption and bleeding episodes.
In addition to smaller dosage requirements for warfar-
in," the incidence of upper GI hemorrhage is common PHENYLBUTAZONE AND ITS ANALOGS
among Chinesesubjects." NSAIDs are itnplicated in 11 %
of Chinese patients admitted to acute medical wards with • In all healthy subjects and patients, phenylbutazone
this complication." and its analogs are expected to potentiate the anticoag-

Table 3. Case Reports of Anticoagulant Control in Warfarin-Treated Patients During

Addition or Withdrawal of Nonsteroidal Antiinflammatory Drugs"
DIAGNOSIS INDICATION
PATIENT (duration of (duration of CHANGES IN PT ANDIOR
NSAID REF. sex/age (y) warfarin therapy) NSAID therapy) WARFARIN DOSAGE

Indomethacin Self et al. (1975)" F/82 PTE, CHF, hypertension, gout (75-100 mg/d 21.7 to 42.2 s
CRF(5 d) for 5 d)
Piroxicam Rhodes et al. M/60 DVT, PTE, angina osteoarthritis of knee 1.6-1.7 to 1.3-1.4 s when NSAID stopped; 1.3-1.4
(1985f' (long-term) (20 mgld for 1-2 wk) to 1.6-1.7 s when NSAID added
Mallet and Cooper F/87 atrial fibrillation, stroke arm pain (20 mg/d) 16.5-18.1 to 24.9 s after 4 d and warfarin dose from
(1991)71 (long-term) 5 to 3.75 mg/d while taking NSAID; 22.7-24.1 to
13.5 s when taking warfarin 2.5 mg/d and NSAID
stopped
Sulindac Loftin and Vesell M/55 PTE, CRF (3 mo) Barter's syndrome PT ratio 2 to 2.5, a direct relationship between the
(1979)50 (200-400 mg bid for NSAID dose and the PT; warfarin dosage reduced
several days) accordingly
Carter (1979)" MI72 AVR, ankylosing arthritis (200 mg bid 20.3 to 26.0 s after 3 wk, warfarin dosage gradually
spondylosis (8 y) for 3-5 wk) reduced from 7.5 to 4.25 mg/d over 5 wk
Ross and Beeley M/56 AVR (5 y) rheumatic pains (100 INR 3.2 to 10.0
(1979)68 bid for 5 d)

AVR = aortic valve replacement; CHF = congestive heart failure; CRF = chronic renal failure; DVT = deep-vein thrombosis; INR = international normal-
ized ratio; NSAID = nonsteroidal antiinflammatory drug; PT = prothrombin time; PTE = pulmonary thromboembolism.
"Refer to text for reports involving topical methyl salicylate therapy.

1280 • The AnnalsofPharmacotherapy • 1995 December, Volume 29

Downloaded from aop.sagepub.com at CORNELL UNIV on July 27, 2016
 Warfarin-NSAID Interactions

ulant effect of warfarin by substantially inhibiting S- sive use of topical methyl salicylate and mefenamic
warfarin while enhancing the elimination of R-war- acid should be avoided. Phenylbutazone and its ana-
farin. logs have already been removed from use in many
• Apazone has the greatest risk of inducing bleeding countries because of their other toxicities.
peptic ulcers. • In patients receiving warfarin, NSAIDs that are known
to be associated with a higher risk of bleeding peptic
OTHER NSAIDS ulcers should be avoided.
• In patients receiving warfarin, NSAIDs, if needed,
• All NSAIDs have antiplatelet effects that result in a should be given in lower dosages where feasible and
prolonged bleeding time. These effects are less pro- introduced slowly. PT should be monitored closely, es-
nounced in NSAIDs not containing aspirin than in as- pecially during the first 2 weeks of therapy. Patients
pirin. also should be monitored closely for any bleeding
• In therapeutic doses, NSAIDs not containing aspirin complications. When NSAIDs are stopped, be aware
alone do not have an effect on clotting factor activity. of the potential for loss of anticoagulant control. ~
• In healthy subjects, the following NSAIDs have not
been shown to enhance the anticoagulant effect of References
warfarin during combined use: diflunisal, etodolac, fen-
bufen, ibuprofen, indomethacin, ketorolac, lomoxicam, 1. Serlin Ml, Breckenridge AM. Drug interactions with warfarin. Drugs
1983;25:61Q..20.
nabumetone, naproxen, sulindac, and tenoxicam.
2. O'Reilly RA. Warfarin metabolismand drug-drug interactions. Adv Exp
• In healthy subjects, mefenamic acid has been shown to Med BioI1987;214:205-12.
enhance the anticoagulant effect of warfarin. 3. Buckley NA, Dawson AH. Drug interactionswith warfarin.Med 1 Aust
• In healthy subjects, following a period of coadminis- 1992;57:479-83.
tration with warfarin, a loss of anticoagulant control is 4. Freedman MD, Olatidoye AG. Clinically significant drug interactions
seen after the following NSAIDs are withdrawn: diflu- with the oral anticoagulants. Drug Saf 1994;I0:381-94.
5. O'Callaghan lW, Thompson RN, RussellAS. Combining NSAIDs with
nisal and lomoxicam. anticoagulants; yes and no. Can Med Assoc 11984;131:857-9.
• In patients receiving long-term warfarin therapy, the 6. Eble IN, West BD, Link KP. A comparison of the isomers of warfarin.
following NSAIDs have not been shown to enhance BiochemPharmacoI1966;15:1003-6.
the anticoagulant effect of warfarin during combined 7. Hewick DS, McEwan 1. Plasma half-lives,plasma metabolitesand anti-
use: ibuprofen, nabumetone, and nimesulide. coagulant efficacies of the enantiomers of warfarin in man. 1 Pharm
• During studies of interactions between NSAIDs and PharmacoI1973;25:458-65.
warfarin, bleeding complications have been reported 8. Holford NHG. Clinical pharmacokinetics and pharmacodynamics of
warfarin. Understanding the dose-effect relationship. Clin Pharma-
in healthy subjects (mefenamic acid, naproxen) or in cokinet 1986;11:483-504.
patients (ibuprofen) even in the absence of a change in 9. Saour 1, Gallus A. Warfarin: is it time to reduce target ranges again?
PT. Aust N Z 1 Moo 1993;23:692-6.
• There are isolated reports of bleeding complications 10. Landefeld CS, Goldman L. Major bleeding in outpatients treated with
and prolongation of PT when the following NSAIDs warfarin:incidence and predictionby factors known at the start of out-
patienttherapy.Am 1 Med 1989;87:144-52.
are given to patients taking long-term warfarin thera-
11. Landefeld CS, Beyth RJ. Anticoagulant-related bleeding: clinical epi-
py: indomethacin, ketoprofen, sulindac, and tolmetin. demiology,prediction,and prevention. Am 1 Med 1993;95:315-28.
• There are isolated reports of loss of anticoagulant con- 12. Routledge PA. Clinical pharmacology and the art of bespoke prescrib-
trol when the following NSAIDs are given to or with- ing. Br 1 Clin PharmacoI1988;26:339-45.
drawn from patients taking long-term warfarin thera- 13. Routledge PA, Chapman PH, Davies DM, Rawlins MD. Factors affect-
py: ibuprofen, indomethacin, piroxicam, and sulindac. ing warfarin requirements. A prospective population study. Eur 1 Clin
PharmacoI1979;15:319-22.
• The incidence of hospitalizations for hemorrhagic pep-
14. BjornssonTO. Anticoagulants. In: Williams RL, Brater DC, Mordenti1,
tic ulcers in patients aged 65 years or older taking both eds. Rational therapeutics. A clinical pharmacologic guide for the
warfarin and NSAIDs other than aspirin was 26.3 per health professional. New York: Marcel Dekker, 1990:541-77.
!Ooo person-years, about 13 times that of patients not 15. Chan TYK, Tsoi WC, CritchleylAlli. The determinantsof warfarinre-
taking either drug. quirements in Chinese patients. PharmacoepidemiolDrug Saf 1992;1:
• The use of the following NSAIDs is associated with a 281-2.
16. Pullar T, Capell HA. Interaction between oral anti-coagulantdrugs and
higher risk of bleeding peptic ulcers: piroxicam in non-steroidal anti-inflammatory agents: a review.Scott Med 11983;28:
adults, and ketoprofen in subjects aged 60 years or 42-7.
older. 17. Tonkin AL, Wing LMH. Interactions of non-steroidal anti-inflammatory
drugs. BaillieresClin Rheumatol 1988;2:455-83.
18. Johnson AG, SeidemanP, Day RO. Adversedrug interactions with non-
Recommendations steroidalanti-inflammatory drugs (NSAIDs): recognition, management
and avoidance. Drug Saf 1993;8:99-127.
Based on the data reviewed here, the following recom- 19. Johnson AG, Seidemann P, Day RO. NSAID-related adverse drug inter-
mendations can be made. actions with clinical relevance. An update. Int 1 Clin Pharmacol Ther
• NSAIDs should be avoided, as much as possible, in 1994;32:509-32.
patients taking warfarin. This is particularly important 20. Quick AJ, Clesceri L. Influence of acetylsalicylicacid and salicylamide
in those who are expected to be more susceptible to on the coagulationof blood.1 PharmacolExp Ther 1960;128:95-8.
the interactions between the 2 drugs and the adverse 21. Rothschild BM. Hematologic perturbations associated with salicylate.
Clin PharmacolTher 1979;26:145-52.
sequelae. 22. Diana Fl, Veronich K, Kapoor AL. Bindingof non-steroidal anti-inflam-
• In patients receiving warfarin, phenylbutazone and its matory agents and their effect on binding of racemic warfarin and its
analogs, or high dosages of aspirin (2-4 g/d), exces- enantiomersto human serum albumin.1 Pharm Sci 1989;78: 195-9.

The Annals ofPhannacotherapy • 1995 December. Volume 29 • 1281

Downloaded from aop.sagepub.com at CORNELL UNIV on July 27, 2016
23. Mackichan 11.Protein binding drug displacement interactions-fact or Clinical study on pharmacological interactionbetween nimesulide and
fiction?Clin Phannacokinet 1989;16:65-73. warfarin.Intl Clin Phannacol Res 1991;11:267-70.
24. BanfieldC, O'Reilly R, Chan E, Rowland M. Phenylbutazone-warfarin 50. Loftin IP, Vesell ES. Interaction between sulindac and warfarin: differ-
interaction in man: further stereochemical and metabolic considera- ent results in normal subjects and in an unusual patient with a potassi-
tions. Br 1 Clin PhannacoI1983;16:669-75. um-losingrenal tubulardefect.1 Clin Phannacol 1979;19:733-42.
25., Roth Gl, Majerus PW. The mechanism of the effects of aspirin on hu- 51. Eichler HG,lung M, Kyrle PA, Rotter M, Korn A. Absence of interac-
man platelets. 1. Acetylation of a particular fraction protein.1 Clin In- tion between tenoxicam and warfarin. Eur 1 Clin Pharmacol 1992;42:
vest 1975;56:624-32. 227-9.
26. McQueen EG, Facoory B, Faed 1M. Non-steroidal anti-inflammatory 52. Koch-Weser1, Sellers EM. Drug interactionswith coumarin anticoagu-
drugs and plateletfunction. N Z Med 1 1986;99:358-60. lants. N EnglJ Med 1971;285:547-58.
27. Shorr RI, Ray WA, Daugherty lR, Griffin MR. Concurrent use of non- 53. Fox SL. Potentiating of anticoagulants caused by pyrazole compounds.
steroidalanti-inflammatory drugs and oral anticoagulants placeselderly lAMA 1964;188:320-1.
persons at high risk for hemorrhagic peptic ulcer disease. Arch Intern 54. Eisen M1. Combined effect of sodium warfarin and phenylbutazone.
Med 1993;153:1665-70. lAMA 1964;189;64-5.
28. Hurlen M, Erikssen 1, Smith P, Arnesen H, Rollag A. Comparison of 55. Aggeler PM, O'Reilly RA, Leong L, Kowitz PE. Potentiationof antico-
bleeding complications of warfarin and warfarin plus acetylsalicylic agulanteffect of warfarinby phenylbutazone. N Eng11Med 1967;276:
acid: a study in 3166 out-patients. 1 Intern Med 1994;236:299-304. 496-501.
29. Hawkey C1. Non-steroidal anti-inflammatory drugs and peptic ulcers: 56. BuIll, MacKinnon 1. Phenylbutazoneand anticoagulantcontrol. Practi-
facts and figuresmultiply,but do they add up? BMJI990;300:278-84. tioner 1975;215:767-9.
30. Langman MJS, Brooks P, Hawkey Cl, Silverstein F, Yeomans N. Non- 57. Hobbs CB, Miller AL, Thornley lH. Potentiationof anticoagulantthera-
steroidalanti-inflammatory drug associatedulcer:epidemiology,causa- py by oxyphenylbutazone: a probable case. Postgrad Med 1 1965;41:
tion and treatment.1 Gastroenterol Hepatol 1991;6:442-9. 563-5.
31. LangmanMJS, Weill, WainwrightP, Lawson DH, RawlinsMD, Logan 58. Powell-lackson PRo Interactionbetween azapropazone and warfarin. Br
RFA, et al. Risks of bleeding peptic ulcer associated with individual Med 11977;1:1193-4.
non-steroidal anti-inflammatory drugs. Lancet 1994;343: 1075-8. 59. Green AE, Hort IF, Korn HET, Leach H. Potentiation of warfarin by
32. FriesIF, WilliamsCA, Bloch DA, Michel BA. Nonsteroidal anti-inflam- azapropazone (letter).Br Med j 1977;I:1532.
matory drug-associatedgastropathy: incidenceand risk factor models. 60. Win N, Mitchell DC,lones PAE, French EA. Azapropazone and war-
AmlMed 1991;91:213-22. farin. BMl 1991 ;302:969-70.
33. Holmes EL. Pharmacologyof the fenamates: IV. Toleration by normal 61. Thompson PL, Serjeant C. Potentially serious interaction of warfarin
human subjects.Ann Phys Med 1966;8(suppl):36-49. with sulfinpyrazone (letter).Med 1 AustI981;1:41.
34. O'Reilly RA, Sahud MA, Aggeler PM. Impact of aspirin and chlorthali- 62. O'Reilly RA. Stereoselectiveinteractionof sulfinpyrazone with racemic
done on the pharmacodynamics of oral anticoagulant drugs in man. warfarin and its separated enantiomorphs in mart, Circulation 1982;
Ann N Y Acad Sci 1971;1'79: 173-86. 65:202-7.
35. Serlin MJ, Mossman S, Sibeon RG, Tempero KF, Breckenridge AM. In- 63. BeeleyL. Azapropazoneand warfarin(letter).BMJI991;302:1341.
teraction between diflunisal and warfarin. Clin Pharmacol Ther 1980;
64. lust ice lL, Kline SS. Analgesics and warfarin. A case that brings up
28:493-8.
questionsand cautions. PostgradMed 1988;83:217-8,220.
36. Ermer lC, Hicks DR, Wheeler SC, Kraml M, lusko W1. Concomitant
65. Koch-Weser1. Hemorrhagicreactions and drug interactionsin 500 war-
etodolac affects neither the unbound clearance nor the pharmacologic
farin-treated patients(abstract). Clin Phannacol Ther 1973;14:139.
effect of warfarin. Clin Pharrnacol Ther 1994;55:305-16.
66. Chan TYK, Lui SF, Chung SY, Luk S, CritchleylA1H. Adverse interac-
37. SavitskyIP, TerzakisT, Bina P, ChiccarelliF, Haynes1. Fenbufen-war-
tion between warfarinand indomethacin. Drug Saf 1994;10:267-9.
farin interaction in healthy volunteers (abstract). Clin Pharmacol Ther
1980;27:284. 67. Flessner MF, Knight H. Prolongation of prothrombin time and severe
gastrointestinal bleeding associatedwith combined use of warfarinand
38. Goncalves L. Influence of ibuprofen on haemostasis in patients on anti-
ketoprofen(letter).lAMA 1988;259:353.
coagulanttherapy.1 Int Med Res 1973;1:180-3.
68. Ross lRY, Beeley L. Sulindac, prothrombin time, and anticoagulants
39. Penner lA, Abbrecht PH. Lack of interaction between ibuprofen and
(letter).Lancet 1979;2: 1075.
warfarin.Curr Ther Res 1975;18:862-71.
40. SchulmanS, Henriksson K Interaction of ibuprofenand warfarinon pri- 69. Thatcher 1, George D. Retropharyngea1 hematoma as a new cause of
acute upper airway obstruction in rheumatoid arthritis. 1 Rheumatol
mary haemostasis. Br 1 RheumatoI1989;28:46-9.
1987;14:1172-3.
41. Vesell ES, Passananti GT,lohnson AO. Failure of indomethacin and
warfarin to interact in normal human volunteers. 1 Clin Pharmacol 70. Koren IF, Cochran DL,lanes RL. Tolmetin-warfarin interaction (letter).
1975;15:486-95. Am 1 Med 1987;82: 1278-9.
42. Mieszczak C, Winther K. Lack of interaction of ketoprofen with war- 71. Yip ASB, Chow WH, Tai YT, Cheung KL. Adverse effect of topical
farin. Eur1 Clin Phannacoll993;44:205-6. methylsalicy1ateointment on warfarin anticoagulation: an unrecog-
nized potentialhazard. PostgradMed 11990;66:367-9.
43. Toon S, Holt BL, Mullins FGP, Bullingham R, Aarons L, Rowland M.
Investigations into the potential effects of multiple dose ketorolac on 72. Littleton F lr. Warfarin and topical salicy1ates (letter).lAMA 1990;263:
the pharmacokinetics and pharmacodynamics of racemic warfarin.Br 1 2888.
Clin Pharmacoll990;30:743-50. 73. Tam LS, Chan TYK, Leung WK, CritchleyJAlH. Warfarin interactions
44. Ravic M, Johnson A, Turner P, Ferber HP. A study of the interaction be- with Chinese traditional medicines: danshen and methyl salicylate
tween lornoxicamand warfarin in healthy volunteers. Hum Exp Toxi- medicatedoil. Aust N Z J Med 1995;25:258.
col 1990;9:413-4. 74. Lo ACT, Chan K, YeunglHK, Woo KS. The effects of danshen (Salvia
45. FitzgeraldDE. Double-blindstudyto establishwhetherthere is any inter- miltiorrhiza) on pharmacokinetics and pharmacodynamicsof warfarin
action between nabumetone and warfarin in healthy male volunteers. in rats. Eur J Drug Metab Phannacokinetl992;I7:257-62.
In: Panayi GS, Price Jl), Rotman H, eds. Nabumetone,a novel anti-in- 75. Turpie AG, Gent M, LaupacisA, Latour Y, GunstensenJ, Basile F, et al.
flammatory. Vol 69. London:Royal Societyof Medicine, 1985:47-52. A comparison of aspirin and placebo in patients treated with warfarin
46. HillemanDE, MohiuddinSM, LucasBD Jr. Nonsteroidal antiinflamma- after heart-valvereplacement. N Engl J Med 1993;329:524-9.
tory drug use in patients receiving warfarin:emphasis on nabumetone. 76. Self TH, Evans WE, FergusonT. Drug enhancementof warfarinactivity
Am 1 Med 1993;95(suppI2A):S30-4. (letter).Lancet 1975;2:557-8.
47. SlatteryIT, Levy G,lain A, McMahonFG. Effectofnaproxen on the ki- 77. Rhodes RS, Rhodes PJ, Klein C, Sintek CD. A warfarin-piroxicam in-
neticsof eliminationand anticoagulantactivityof a single dose of war- teraction. Drug IntellClin Phann 1985;19:556-8.
farin.Clin Phannacol Ther 1979;25:51-60. 78. Mallet L, Cooper JW. Prolongationof prothrombintime with the use of
48. lain A, McMahon FG, Slattery IT, Levy G. Effect of naproxen on the piroxicamand warfarin. Can J Hosp Pharm 1991 ;44:93-4.
steady-stateserum concentration and anticoagulant activityof warfarin. 79. Carter SA. Potentialeffect of sulindac on response of prothrombin-time
Clin Phannacol Ther 1979;25:61-6. to oral anticoagulants (letter).Lancet 1979;2:698-9.
49. Auteri A, Bruni F, Blardi P, Di Renzo M, Pasqui AL, Saletti M, et al. 80. RodriguezLAG, Jick H. Risk of upper gastrointestinal bleedingand per-

1282 • The AnnalsofPharmacotherapy • /995 December. Volume 29

Downloaded from aop.sagepub.com at CORNELL UNIV on July 27, 2016
 Warfarin-NSAID Interactions
foration associated with individualnon-steroidalanti-inflammatory REVUE DEL1'I'TERATURE: Les etudes traitant des interactionsentre la
drugs. Lancet1994;343:769-72. warfarine et les AINS ou les rapports d'interactions ont ete identifies a
81. RedwoodM, Taylor C, Bain Bl, Matthews IH. The association of age partir de la banque de donnees MEDLINE (1976 ace jour) ou apartir
withdosage requirement for warfarin. Age Ageing 1991;20:217-20. des listes de referencescitees dans les articlespertinents.
82. lames AH, Britt RP, RaskinoCL, Thompson sa. Factorsaffecting the SELEcnON DESETUDES: L'auteur a considers tous les articles pour les
maintenance doseof warfarin. 1 ClinPathoI1992;45:704-6. inclure acette revue.
83. Routledge PA,Chapman PH, Davies DM,Rawlins MD.Phannacokinet-
ics and pharmacodynamics of warfarin at steady state.Br 1 Clin Phar- REsUME: Tous les AINS peuvent prolonger Ie temps de saignement en
macoI1979;8:243-7. inhibant la fonction plaquettaire. L'aspirine, adose elevee, possede un
84. Shepherd AMM, Hewick DS,Moreland TA, Stevenson IH. Ageas a de- effet bypoprothrombinemiant direct. La phenylbutazoneet les
terminant of sensitivity to warfarin. Br 1 Clin PhannacoI1977;4:315- substancesanalogues augmentent I'action de la warfarine par
20. !'inhibition du metabolisme hepatique de celle-ci. L'acidemefenamique
85. HylekEM,SingerDE. Riskfactors for intracranial hemorrhage in outpa- augmente aussi l'effet anticoagulantde la warfarine mais par un
tientstakingwarfarin. AnnInternMed 1994;120:897-902. mecanismeinconnu. La significationclinique du deplacementde la
86. Chan TYK. Recognizing increased sensitivity to warfarinin liver dys- warfarine de ses sites de liaison aux proteines plasmatiquespar les AINS
function secondary to congestive heart failure. Pharmacoepidemiol a ere exageree, Par contre, une interactionsignificativeest susceptiblede
DrugSaf 1995 (in press). se produire en presence de concentrations elevees d' AINS chez des
87. Williams RL, ScharyWL, Blaschke TF, MeffinPI, MelmonKL, Row- patientsqui ont un ralentissementde l'elirnination de la warfarine (e.g.,
landM. Influence of acuteviralhepatitis on disposition and pharmaco- patients atteints d'insuffisance cardiaque severe ou dont la fonction
logiceffectof warfarin. ClinPhannacol Ther 1976;20:90-7. hepatiqueest alteree), Finalement, les AINS peuvent causer un
88. ChanTYK,Critchley lAlli, ChanlCN, Tomlinson B, ChanMTV.Dis- saig'nement gastrointestinalqui peut etre plus severe si la warfarine est
easeprofiles in Chinese in HongKong: an analysis of theprimary diag- administreeen concomitance.
nosesin 561 acutehospital medical admissions. Southeast Asian 1Trop CONCLUSIONS: L'administration concomitantede la warfarine et des
Med Public Health 1993;24:766-8. AINS devrait generalementetre evitee acause du risque accru de
89. ChanTYK,ChanlCN, Tomlinson B, Critchley lAlli. Adverse reactions saignementchez ces patients. Si un AINS est requis chez un patient qui
to drugsas a causeof admissions to a general teaching hospital in Hong recoit de la warfarine, les medicaments suivantsdevraient etre evites: la
Kong. DrugSaf 1992;7:235-40. phenylbutazone(et substancesanalogues), I'aspirine adose elevee,
Pacide mefenamique,l'utilisation topique excessive de salicylatede
methyle et les AINS qui sont associes aun risque plus elevee de
saignementen presence d'ulceres peptiques(piroxicamchez l'adulte et
REsUME ketoprofenechez la personne agee). Enfin, une surveillanceetroite de
l' anticoagulotherapie et des complications hemorragiques devrait etre
OBJECTIF: Passer en revue les mecanismes et la signification cliniquede effectuee dans de tels cas.
I'interaction medicamenteuseimpliquantla warfarineet les
antiinflammatoires non steroidiens(AINS) et dieter une ligne de MICHELE PLANlE
conduite afin d' eviter cette interactionautant que possible.

The AnnalsofPharmacotherapy • 1995 December, Volume 29 • 1283

Downloaded from aop.sagepub.com at CORNELL UNIV on July 27, 2016