Warfarin-Associated Hypoprothrombinemia: An Unusual Presentation

Dorian Williams, Charles D. Ponte


Am J Health Syst Pharm. 2003;60(3) 

In This Article


Few published case reports document the interaction between azithromycin and warfarin. In one report, a 41-year-old woman with end-stage diabetic renal nephropathy was receiving warfarin to maintain patency of a permanent left internal jugular catheter for hemodialysis treatments.[1] Her INR had remained stable for approximately six months before receiving a five-day course of oral azithromycin for bronchitis. Three days after completing her treatment with azithromycin, her INR had increased to 4.88 without evidence of bleeding. The warfarin was discontinued, and her INR returned to normal four days later. The authors concluded that the hypoprothrombinemia was caused by an interaction between warfarin and azithromycin and was not due to biochemical and metabolic causes.

In another case, a 53-year-old Native American man came to the hospital with pulmonary infiltrates and hemoptysis.[2] Significant in his medical history were systemic lupus erythematosus, hypertension, seizures, and valvular heart disease. Warfarin was prescribed after mitral valve repair with stable INRs reported for approximately eight months before admission. Seven days before admission, the patient received a five-day course of oral azithromycin for an upper respiratory tract infection. Upon admission, his INR was too high to quantify, and warfarin therapy was stopped. The patient received freshly frozen plasma and four daily doses of intramuscular phytonadione. His INR peaked at 7.7 on day 3 and by day 4 it had decreased to 2.9, and continued to decrease thereafter. The patient's clinical status deteriorated during the hospital stay, and he died on day 12. The cause of death was determined to be multi-organ system failure, which led to a cardiac arrest.

In another report, a 71-year-old black woman went to a local emergency room complaining of chest pain on her right side one day after beginning a five-day course of oral azithromycin for an upper respiratory tract infection.[3] She had a history of severe ischemic and valvular heart disease. She had been taking warfarin for about five years after prosthetic mitral valve replacement. Her INR and warfarin dosage had remained stable for much of this time. Twelve days before her emergency room visit, her INR was 2.48 on a stable warfarin dose of 7.5 mg/day. During the initial emergency room visit, she was given a prescription for 10 tablets of acetaminophen with codeine (300 mg of acetaminophen and 30 mg of codeine phosphate) and discharged. After completing the azithromycin course, she returned to the emergency room complaining of chest pain and approximately one week of right upper quadrant abdominal pain. Her INR was 11.5. She was told to stop taking the warfarin and see her primary care provider. All laboratory test values were normal. The following day, she returned to the emergency room with chest pain and two-pillow orthopnea. Upon admission, she stated that the pain had been present since a fall two weeks earlier. Her INR was 15.6. The warfarin was withheld, and after one dose of subcutaneous phytonadione, her INR decreased to 6.4 and steadily declined during her hospital stay. A CT scan of her abdomen revealed a mass in the right chest wall, pressing on the liver. Another CT scan revealed changes consistent with a right upper quadrant hematoma. When her INR reached 2.10, warfarin was reinitiated, and both the dosage and INR remained stable. The authors ruled out pathophysiological, nutritional, pharmacokinetic, and other potential drug-drug interactions. They concluded that, despite the lack of causation, azithromycin was probably linked to the exaggerated hypoprothrombinemia.

In another recent publication, two elderly men developed warfarinassociated hypoprothrombinemia after receiving oral azithromycin for pneumonia.[4] In the first case, an 80-year-old man was admitted to a hospital for probable pneumonia. He had a history of type 2 diabetes mellitus, valvular heart disease with aortic and mitral valve replacement, and congestive heart failure. His INR had been stable (3.6-4.3) for at least two months before this hospitalization. Intravenous ampicillin was empirically started on day 1; on the next day, a three-day course of oral azithromycin 500 mg was added. On day 3, gentamicin was initiated for endocarditis. On hospital day 6, the patient was switched to oral amoxicillin-clavulanate potassium as monotherapy. His INR was 7.2 on day 8, two days after discontinuing oral azithromycin. His INR continued to increase, and a single dose of i.v. phytonadione was administered on day 10. The patient had also received a total of 12 doses of acetaminophen, which may have contributed to the excessive hypoprothrombinemia.

In the second case, an 85-year-old man was admitted to a local hospital with urinary sepsis. His past medical history was significant for chronic obstructive pulmonary disease, anemia, gastroesophageal reflux disease, glaucoma, a right-sided cerebrovascular accident, and prostate cancer. His INR was 1.7, so his daily 2-mg dose of warfarin was increased to 3 mg. On day 1, pneumonia was diagnosed, and azithromycin 500 mg p.o. was started. By day 4, his INR had risen to 4.6. The decision was made to stop azithromycin. In addition, his liver transaminase levels (i.e., ALP, AST, ALT, and GGT) were elevated. This increase in liver function test (LFT) results was thought to be azithromycin induced. The patient was discharged on day 10. His INR was 1.6 on a warfarin dose of 2.5 mg, but his LFT values remained elevated. Neither of these two patients experienced bleeding complications as a result of their elevated INRs. The authors concluded that azithromycin was the most likely cause of both patients' elevated INRs due to the temporal relationship and return to the same or similar doses of warfarin after resolution of the increased anticoagulation.

In each case, causality was questioned because of the variety of potentially confounding variables, including concomitant hypoalbuminemia, fever, hepatic dysfunction, infection, concomitant antibiotics, and poor appetite.[4,5] A total of 41 spontaneous reports of possible interactions between warfarin and azithromycin have been reported to Pfizer Inc. and DuPont Pharmaceuticals.[3] Recent phone discussions with drug information specialists at both companies failed to elucidate an updated number of spontaneous reports of adverse events involving this drug combination. Reasons cited include company policy restrictions following the recent Pfizer merger and an apparent lack of retrievable data capability by DuPont Pharmaceuticals (Ponte CD, West Virginia University, personal communication, 2002 Aug). In addition, a lack of an interaction between azithromycin and warfarin was reported in a retrospective case-control study.[5] Thus, it is unknown whether any significant interaction exists between these two drugs.

When our patient was admitted to the hospital for pneumonia, her INR was within the therapeutic range at 3.1. Three days before hospital admission, her INR was 2.1. Whether this change represented an upward trend or reflected normal fluctuation typical for many patients is indeterminate. Her INR for the previous five months had ranged from 1.4 to 2.5. Within 72 hours of admission, her INR had sharply risen to 14.8, leading to a spontaneous hematoma on the anterior abdominal wall and peritoneum and a splenic hemorrhage. She had no history of falls or other traumas that could have caused these complications. During her hospital stay, the patient had been started on several new medications, including azithromycin, prednisone, ibuprofen, and acetaminophen, that may have contributed to her pronounced hypoprothrombinemic response. The remainder of her inpatient medications are not known to interact with warfarin.

Ibuprofen is a highly protein-bound nonsteroidal antiinflammatory drug (NSAID) that is expected to compete with warfarin for protein-binding sites. However, there are no published pharmacokinetic data to support this contention. Our patient had received a total of three 400-mg doses of ibuprofen between hospital days 2 through 4 and one 15-mg i.v. dose of ketorolac tromethamine on hospital day 2. It is doubtful that these NSAIDs contributed to her bleeding propensity. Although most NSAIDs have the ability to prolong bleeding time via their antiplatelet activity, the effect is small and probably not significant. Thus, it appears unlikely that decreased platelet aggregation contributed to the development of the hematoma. Corticosteroids have also been implicated in exaggerating the response to warfarin, although the nature of this interaction and its significance are poorly documented.[6] Whether the vascular effects from corticosteroids (increased vascular permeability) can contribute to an increased risk of bleeding from anticoagulants is also indeterminate. Thus, it is unlikely that the oral prednisone she received contributed to her bleeding.

The effect of acetaminophen on warfarin response is unclear. Although there are conflicting reports from the 1960s and 1970s, recent evidence suggests that acetaminophen can potentiate the effects of warfarin and that the effect may be dose-related.[7,8,9,10] In a well-known case-control study from 1998, investigators concluded that acetaminophen was a dose-related independent risk factor for persons with an INR greater than 6.0.[11] The highest risk (odds ratio [OR] = 10 [95% CI, 2.6-37.9]) was noted in persons consuming at least 9100 mg of acetaminophen per week. Those ingesting 325-2267 mg per week were not at an increased risk (OR = 1.1 [95%CI, 0.5-2.5]) of having an INR greater than 6.0. Our patient took a total of 975 mg of acetaminophen on hospital day 3 and 325 mg on hospital day 4. It is doubtful that acetaminophen contributed to her elevated INR.

In the case of our patient, azithromycin is the most likely offending agent. Concomitant medications, patient nonadherence, laboratory error, and other recognized influences on INR (e.g., hepatic dysfunction, congestive heart failure, fever, pneumonia) had been ruled out or were not considered contributing factors during her hospital stay. Decreased appetite, fever, and infection (e.g., pneumonia) can contribute to an exaggerated response from warfarin.[3] Our patient did complain of a reduced appetite for several days before admission. It is unknown if this resulted in a reduced intake of dietary vitamin K and significant hypoprothrombinemia, although she did not appear to be clinically malnourished when admitted to the hospital. Her albumin level at admission was unknown, but was reported to be normal (4.0 g/dL) approximately eight months earlier. Her liver function tests (e.g., AST, ALT, ALP, and GGT) remained normal throughout her hospital stay. Whether azithromycin conclusively led to the precipitous elevation in her INR and the resultant abdominal bleeding is unclear since rechallenge was not attempted due to ethical considerations. However, it is important to remember that her INR was within the therapeutic range at admission. The changes in her INR resulted from manipulation that occurred in a controlled hospital setting.

Our patient developed abdominal symptoms soon after beginning azithromycin (day 2). These symptoms were most likely associated with her developing intraabdominal hemorrhage. An elevated INR was noticed on day 4. It is difficult to corroborate the rapidity of the rise in our patient's INR and the onset of her complaints with the findings from the published literature. In two of the four published case reports of suspected azithromycin-warfarin interactions, the elevated INRs were only noted two to three days following completion of the five-day courses of azithromycin.[1,2] Therefore, one cannot determine the rate of rise in the INR for these patients.

In another case, a patient came to the emergency room complaining of right upper quadrant pain five days after a five-day course of azithromycin.[3] Her symptoms had been present for about a week. Her INR at admission was 11.15. Again, it is unknown how quickly her INR had risen and when she developed a right upper quadrant hematoma. In addition, the patient reported a fall two weeks before her initial visit to the emergency room, although its relationship to the hematoma was not addressed.

In the remaining two cases, the INR of an elderly man increased from 3.9 to 7.2 within 48 hours of stopping a three-day course of oral azithromycin. The INR of the second elderly man peaked at 4.6 after four doses of oral azithromycin.[4] However, uncertainty about when the INR was obtained in relation to the antimicrobial doses (following the third or fourth dose) and the concomitant increase in the warfarin dose makes interpretation of the data from the second patient problematic.